Page 1


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

Managing Director Antonio Majocchi (Pavia)

Associate Editors Hélène Cavé (Paris), Pavan Reddy (Ann Arbor), Andreas Rosenwald (Wuerzburg), Monika Engelhardt (Freiburg), 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 Omar I. Abdel-Wahab (New York); 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); Simon Mendez-Ferrer (Madrid); 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 Neri (VT)


haematologica calendar of events

Journal of the European Hematology Association Published by the Ferrata Storti Foundation EHA-SWG Scientific Meeting on Integrated Diagnosis Strategies in Oncohematology for the management of cytopenias and leukocytosis Chair: MC Béné February 8-10, 2018 Barcelona, Spain EHA-ISHBT Hematology Tutorial on Lymphoproliferative and Plasma Cell Disorders February 16-18, 2018 Lucknow, India EuroClonality Workshop: Clonality assessment in Pathology European Scientific foundation for Laboratory Hemato Oncology (ESLHO) Chairs: PJTA Groenen, F Fend, AW Langerak February 19-21, 2018 Nijmegen, The Netherlands

EHA Hematology Tutorial on Thalassemia May 3-4, 2018 Shiraz, Iran 23rd Congress of EHA June 14-17, 2018 Stockholm, Sweden EHA-SAH Hematology Tutorial on lymphoid Malignancies and Plasma Cell Dyscrasias September 13-14, 2018 Buenos Aires, Argentina EHA-SWG Scientific Meeting on Aging and Hematology Chair: D Bron October 12-14, 2018 Location TBC

EHA-SWG Scientific Meeting on New Molecular Insights and Innovative Management Approaches for Acute Lymphoblastic Leukemia Chair: N Gökbuget April 12-14, 2018 Barcelona, Spain EHA-TSH Hematology Tutorial on Acute Leukemias April 28-29, 2018 Istanbul, Turkey

Calendar of Events updated on December 11, 2017


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

Table of Contents Volume 103, Issue 1: January 2018 Cover Figure Image generated by www.somersault1824.com.

Editorials 1

The role of SIRT6 in tumors Vanessa Desantis et al.

4

Remission is good - relapse is bad Paul S. Gaynon

6

High-throughput sequencing for rapid diagnosis of inherited platelet disorders: a case for a European consensus Alan T Nurden and Paquita Nurden

Review Article 9

The prothrombotic state in paroxysmal nocturnal hemoglobinuria: a multifaceted source Barnaby Peacock-Young

Articles Hematopoiesis

18

Protein arginine methyltransferase 6 controls erythroid gene expression and differentiation of human CD34+ progenitor cells Stefanie C. Herkt et al.

Bone Marrow Failure

30

Cancer in the National Cancer Institute inherited bone marrow failure syndrome cohort after fifteen years of follow-up Blanche P. Alter et al.

Myeloproliferative Disorders

40

Expansion of EPOR-negative macrophages besides erythroblasts by elevated EPOR signaling in erythrocytosis mouse models Jieyu Wang et al.

51

Treatment of essential thrombocythemia in Europe: a prospective long-term observational study of 3649 high-risk patients in the Evaluation of Anagrelide Efficacy and Long-term Safety study Gunnar BirgegĂĽrd et al.

Myelodysplastic Syndromes

61

Erythropoietin inhibits osteoblast function in myelodysplastic syndromes via the canonical Wnt pathway Ekaterina Balaian et al.

69

Labile plasma iron levels predict survival in patients with lower-risk myelodysplastic syndromes Louise de Swart et al.

Acute Myeloid Leukemia

80

Depletion of SIRT6 enzymatic activity increases acute myeloid leukemia cells' vulnerability to DNA-damaging agents Antonia Cagnetta et al.

Haematologica 2018; vol. 103 no. 1 - January 2018 http://www.haematologica.org/


haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation 91

Variable outcome and methylation status according to CEBPA mutant type in double-mutated acute myeloid leukemia patients and the possible implications for treatment Dima El-Sharkawi et al.

101

Randomized study of continuous high-dose lenalidomide, sequential azacitidine and lenalidomide, or azacitidine in persons 65 years and over with newly-diagnosed acute myeloid leukemia Bruno C. Medeiros et al.

Acute Lymphoblastic Leukemia

107

Predictive value of minimal residual disease in Philadelphia-chromosome-positive acute lymphoblastic leukemia treated with imatinib in the European intergroup study of post-induction treatment of Philadelphia-chromosome-positive acute lymphoblastic leukemia, based on immunoglobulin/T-cell receptor and BCR/ABL1 methodologies Giovanni Cazzaniga et al.

Non-Hodgkin Lymphoma

116

Inhibition of focal adhesion kinase overcomes resistance of mantle cell lymphoma to ibrutinib in the bone marrow microenvironment Martina Rudelius et al.

126

Histone deacetylase inhibitors downregulate CCR4 expression and decrease mogamulizumab efficacy in CCR4-positive mature T-cell lymphomas Akihiro Kitadate et al.

Chronic Lymphocytic Leukemia

136

Microenvironmental stromal cells abrogate NF-kB inhibitor-induced apoptosis in chronic lymphocytic leukemia Carl Philipp Simon-Gabriel et al.

Platelet Biology & Its Disorders

148

Introducing high-throughput sequencing into mainstream genetic diagnosis practice in inherited platelet disorders JosĂŠ M. Bastida et al.

163

Comparison of up-front treatments for newly diagnosed immune thrombocytopenia - a systematic review and network meta-analysis Yasuyuki Arai et al.

Coagulation & Its Disorders

172

Comparative profiling of HLA-DR and HLA-DQ associated factor VIII peptides presented by monocyte-derived dendritic cells Ivan Peyron et al.

179

Analyses of the FranceCoag cohort support differences in immunogenicity among one plasma-derived and two recombinant factor VIII brands in boys with severe hemophilia A Thierry Calvez et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/103/1.toc

e1

Long-term safety of deferiprone treatment in children from the Mediterranean region with beta-thalassemia major: the DEEP-3 multi-center observational safety study Sebastian Botzenhardt et al. http://www.haematologica.org/content/103/1/e1

Haematologica 2018; vol. 103 no. 1 - January 2018 http://www.haematologica.org/


haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation e5

Hsp90 inhibition disrupts JAK-STAT signaling and leads to reductions in splenomegaly in patients with myeloproliferative neoplasms Gabriela S. Hobbs et al. http://www.haematologica.org/content/103/1/e5

e10

BCR-ABL1 compound mutants display differential and dose-dependent responses to ponatinib Konstantin Byrgazov et al. http://www.haematologica.org/content/103/1/e10

e13

Single-molecule DNA sequencing of acute myeloid leukemia and myelodysplastic syndromes with multiple TP53 alterations Laurence LodĂŠ et al. http://www.haematologica.org/content/103/1/e13

e17

Chromothripsis is linked to TP53 alteration, cell cycle impairment, and dismal outcome in acute myeloid leukemia with complex karyotype Frank G. RĂźcker et al. http://www.haematologica.org/content/103/1/e17

e21

Small bone marrow adipocytes predict poor prognosis in acute myeloid leukemia Wei Lu et al. http://www.haematologica.org/content/103/1/e21

e25

Panobinostat monotherapy and combination therapy in patients with acute myeloid leukemia: results from two clinical trials Richard F. Schlenk et al. http://www.haematologica.org/content/103/1/e25

e29

A germ-line deletion of APOBEC3B does not contribute to subtype-specific childhood acute lymphoblastic leukemia etiology Amelia D. Wallace et al. http://www.haematologica.org/content/103/1/e29

e32

Treat or palliate: outcomes of very elderly myeloma patients Fotios Panitsas et al.. http://www.haematologica.org/content/103/1/e32

Case Reports Case Reports are available online only at www.haematologica.org/content/103/1.toc

e35

Clinical severity in adult warm autoimmune hemolytic anemia and its relationship to antibody specificity Philippe Chadebech et al. http://www.haematologica.org/content/103/1/35

e39

Response to MEK inhibition with trametinib and tyrosine kinase inhibition with imatinib in multifocal histiocytic sarcoma Sophie Voruz et al. http://www.haematologica.org/content/103/1/39

Haematologica 2018; vol. 103 no. 1 - January 2018 http://www.haematologica.org/


EDITORIALS The role of SIRT6 in tumors Vanessa Desantis, Aurelia Lamanuzzi, Angelo Vacca Department of Biomedical Sciences and Human Oncology, Unit of Internal Medicine “G. Baccelli”, University of Bari Medical School, Policlinico, Bari, Italy E-mail: angelo.vacca@uniba.it doi:10.3324/haematol.2017.182675

D

ysfunctional DNA-damage response and consequent genomic instability play a pivotal role in the initiation and progression of both solid and hematologic tumors. Preservation of DNA integrity is, in fact, a key cellular function, hence several mechanisms that repair the damaged DNA need to be studied. Recent studies have focused on key players that are able to improve the DNA repair and thus may act as targets for new therapeutic approaches. Several data have been obtained on overexpression and hyperactivity of Sirtuins (SIRTs), a family of proteins with deacylase or mono-adenosine diphosphate (ADP)-ribosyltransferase activities that degrade nicotinamide adenine dinucleotide (NAD+) enzymes to enable their biological processes1 and promote longevity.2 In mammalian cells, the Sirtuin family is composed of seven members that show different subcellular localization and functions (transcription, metabolism, fat mobilization, DNA repair, stress responses, apoptosis, tumorigenesis and aging),3,4 and conserve the catalytic domain and the NAD+ binding site.5 In cancer and agingassociated pathways, SIRT6 is crucial since it prevents genomic instability, maintains telomere integrity, and regulates metabolic homeostasis and DNA repair.6 SIRT6 can be considered a double-edged sword in cancer because of its dual role of both tumor suppressor and oncogene (Table 1). In healthy conditions, SIRT6 either acts as a gatekeeper of DNA repair mechanisms or regulates cell survival and proliferation. Following the DNA damage, SIRT6 triggers the apoptotic process, hence it is down-regulated in several cancers. However, in other cancers, it is up-regulated, corroborating the idea that it can also act as oncogene.

SIRT6 as a tumor suppressor Studies in colorectal, breast, ovarian, hepatocellular, lung, and other tumors correlate the reduction of SIRT6 expression with tumor progression and poor clinical outcome. In the presence of DNA-damage, SIRT6 promotes apoptotic cell death, ensuring damaged cells do not proliferate. Sebastian et al.7 demonstrated in vivo that SIRT6 deficiency favors tumor growth and invasiveness. They also showed that SIRT6 is involved in the Warburg effect, a glycolytic metabolic shift important for supporting rapid tumor growth. SIRT6 promotes both in vitro and in vivo tumor suppression through repression of hypoxia-inducible factor 1alpha (HIF-1α) that inhibits glycolytic metabolism in cancer cells.7 Interestingly, in mouse and human pancreatic ductal adenocarcinoma (PDAC), the SIRT6 knockdown is due to repression of Myc-target oncofetal protein Lin28b that negatively regulates the let-7 family of miRNAs.8 In detail, loss of SIRT6 triggers activation of Lin28 promoter, Myc recruitment, and consequent activation of Lin28b, the downstream let-7 target genes (HMGA2, IGF2BP1) and IGF2BP3 that accelerate the PDAC progression and metastasis.8 In human haematologica | 2017; 102(12)

colon cancer, Lin et al.9 discovered the crosstalk between UPS10 and SIRT6 that regulates cell-cycle progression and proliferation, and showed that the dysregulated USP10 function promotes tumorigenesis through SIRT6 degradation. Lin et al. also showed an important reduction in USP10 (a deubiquitinase protein) and SIRT6 expression. Indeed, the downregulation of USP10 triggers SIRT6 instability and negatively controls the transcriptional activity of the c-Myc oncogene that inhibits cell-cycle progression, cancer cell growth, and tumor formation.9 In liver cancer, the SIRT6 suppression is regulated by the c-Jun/c-Fos pathway:10 c-Fos induces SIRT6 transcription and represses survivin by reducing histone H3K9 acetylation and NF-κB activation. The increase in SIRT6 impairs cancer development by targeting the anti-apoptotic activity of survivin. Min et al.10 identified in human dysplastic liver nodules a specific expression pattern characterized by increased c-Jun-survivin and reduced cFos-SIRT6 level. In hepatocellular carcinoma (HCC), Bhardwaj et al.11 found that SIRT6 acts as a tumor suppressor because it deacetylates nuclear pyruvate kinase M2 (PKM2) inhibiting cell proliferation and tumorigenesis via PKM2. In ovarian cancer, Zhang et al.12 showed that SIRT6 is downregulated at mRNA and protein levels in tumor cells compared to normal cells. Moreover, SIRT6 reduces the expression of neurogenic locus notch homolog protein 3 (Notch3) while the Notch3 overexpression antagonists SIRT6 exert an effect on the ovarian cell proliferation; SIRT6 thus inhibits the proliferation of ovarian tumor cells through regulation of Notch3.12 In breast cancer, the repression of SIRT6, mediated by runt-related transcription factor 2 (RUNX2), regulates metabolic pathways and promotes tumor development.13 More specifically, Choe et al.13 showed that RUNX2 downregulates the SIRT6 expression at both mRNA and protein levels, and that endogenous SIRT6 expression is lower in the tumor breast tissue and cell lines expressing high levels of RUNX2 regulating the metabolic pathways. In addition, Han et al.14 demonstrated in non-small cell lung cancer (NSCLC) that SIRT6 inhibits Twist1 expression. Twist1 is a member of basic helix-loop-helix transcription factor family that promotes tumor proliferation and malignant transformation. Thus, SIRT6 is able to inhibit tumor cell proliferation through Twist1 suppression. Finally, the overexpression of E2F transcription factor 1 (E2F-1) in bladder and prostate cancer induces the downregulation of SIRT6 that closely correlates with cancer progression and poor prognosis.15

SIRT6 as a tumor promoter In contrast to these studies, several papers show that overexpression of SIRT6 in solid and in hematologic tumors can promote oncogenic activity. Ming et al.16 demonstrated the oncogenic role of up-regulated SIRT6 in human skin squamous cell carcinoma (SCC): in skin keratinocytes, SIRT6 is 1


Editorials

increased upon exposure to ultraviolet B (UVB) light through the activation of the AKT pathway and promotes the cyclooxygenase 2 (COX-2) expression that represses the AMP-activated protein kinase (AMPK) signaling and increases proliferation and cell survival.16 Zhang et al.17 demonstrated that SIRT6 overexpression in HCC suppresses tumor growth by blocking extracellular signalregulated kinases (ERK) 1/2 signaling pathway. In addition, Feng et al.18 and Ran et al.19 showed that SIRT6 plays an oncogenic role in HCC. In particular, the overexpression of SIRT6 is required for induction of transforming growth factor (TGF)-β1 and H2O2/HOCl reactive oxygen species (ROS) that mediate tumorigenesis. TGF-β1 upregulates the SIRT6 expression inducing the activation of ERK and Smad pathways, and altering the effect of these

proteins on cellular senescence.18 Ran et al.19 demonstrated an oncogenic effect of SIRT6 via chromatin remodeling. At molecular level, SIRT6 induces deacetylation of H3K9 that blocks Bcl-2-associated X protein (Bax) transcription. As a consequence, it enhances p53 and E2F-1 chromatin accessibility thus inhibiting apoptosis. Elhanati et al.20 and Lefor et al.21 correlated SIRT6 regulation to two microRNAs (miR-) in two different cancers. At basal conditions, SIRT6 and miR-122 negatively regulate each other in HCC. SIRT6 down-regulates miR-122 by deacetylating H3K56 in the promoter region. The miR122 binds SIRT6 3′ UTR and reduces its levels, while the loss of the negative correlation between SIRT6 and miR122 expression is significantly associated with better prognosis.20 In addition, miR-34a plays a key role during

Table 1. SIRT6 expression and its role in cancer.

↓ ↓

SIRT6 status

Study

Cancer type 7

Sebastian Min10

Colorectal Liver

Lin9

Colon

Han14

Non-small cell lung

Choe13

Breast

Wu15

Bladder and prostate

Zhang12

Ovarian

Bhardwaj11

Hepatocellular carcinoma

Kugel8

Pancreatic ductal adenocarcinomas Squamous cell carcinoma Squamous cell carcinoma Hepatocellular carcinoma Hepatocellular carcinoma Hepatocellular carcinoma Hepatocellular carcinoma

Lefort21

Zhang17

Elhanati20

Ran19

Cea22

Ming16

Feng18

Role of SIRT6

Pathways and regulators

Warburg effect and tumor growth Tumorigenic effect

Suppression of Myc-regulated genes (HIF-1 ) Increased levels of c-Jun-survivin and reduced levels of c-Fos-SIRT6 Reduction of deubiquitinase protein USP10 antagonizes the transcriptional activity of the c-Myc oncogene Inhibition of Twist1 expression

Cancer cell growth; tumorigenic effect

Inhibition of tumor proliferation and malignant transformation Reduction of mitochondrial oxygen consumption rates or respiration E2F1b suppresses SIRT6 transcription Inhibition of proliferation Inhibition of cell proliferation and tumorigenesis via nuclear pyruvate kinase M2 (PKM2) Loss of SIRT6 accelerates tumor progression and metastasis miR-34a target Proliferation and tumor cell survival by means of UVB Tumor growth suppression Tumorigenesis miR-122 target and poor prognosis Block of apoptosis

Multiple myeloma

Genome instability and poor prognosis

Cagnetta24

Acute myeloid leukemia

Genome instability and poor prognosis

Regulation of RUNX2-mediated metabolic changes increases pAkt, HK2, and PDHK1 glycolytic protein level E2F1 binds to SIRT6 promoter and suppresses its activity Reduction of Notch 3 expression Deacetylation of PKM2

Activation of Lin28 and the downstream let-7 target genes (HMGA2, IGF2BP1, and IGF2BP3) Expression of miR-34a induces cell differentiation through DNA methylation or p53 Activation of AKT pathway and COX-2 expression repressing AMPK signaling ERK1/2 signaling pathway TGF-β1/H2O2/HOCl ROS up-regulate SIRT6 expression by ERK and Smad pathways Deregulated miR-122 binds to three sites on the SIRT6 3' UTR and reduces its level Deacetylation of H3K9 blocking Bax transcription and enhancing p53 and E2F-1 chromatin accessibility Increased ELK1 and ERK signaling-related gene In CD34+ blasts, SIRT6 expression is associated to ongoing DNA damage and intense replicative stress

↓ ↓ ↓

UVB: ultraviolet B cancer.

2

haematologica | 2018; 103(1)


Editorials

Figure 1. SIRT6 in cancer acts as tumor suppressor and tumor promoter in different cellular pathways.

the differentiation process of HCC and SIRT6 represents one of its targets. SIRT6 downregulation induces differentiation effects mediated by miR-34a.21 The role of SIRT6 is not well known in hematologic malignancies. In multiple myeloma (MM), SIRT6 is highly expressed as adaptive response to genomic stability, and its overexpression is associated to proliferation and poor prognosis.22 Cea et al.22 demonstrated in vitro and in a human MM xenograft model that SIRT6 down-regulates the expression of ERK signaling-related genes and suppresses the activity of ETS-domain transcription factor (ELK1), increasing DNA repair level via Chk1 (a critical messenger of the genome integrity checkpoints involved in the evolution of human cancer23), and conferring resistance to DNA-damaging agents. In this scenario, the paper by Cagnetta et al.24 studies the biological relevance and the genomic instability and poor prognosis associated with the mRNA upregulation of SIRT6 in the acute myeloid leukemia (AML) cells compared with low SIRT6 levels detected in normal CD34+ hematopoietic progenitors. SIRT6 participates in DNA double-strand break repair by deacetylation of C-terminal binding protein (CtBP), interacting protein (CtIP), poli ADP-ribosio polimerase-1 (PARP-1) and DNA-protein kinase (PK) complex. Indeed, AML cells are able to recruit SIRT6 in DNA-damaged sites and to promote deacetylation by means of DNA-PKs and CtIP. On the contrary, downregulation of SIRT6 expression both in vitro and in a murine xenograft model of human AML promotes genomic instability that sensitizes AML cells to daunorubicin (DNR) haematologica | 2018; 103(1)

and cytarabine (ARA-C). Importantly, the results from Cagnetta et al. suggest an innovative chemotherapy that may selectively target AML cells enhancing their sensitivity to DNA-damage agents (DDAs).24 In conclusion, SIRT6 fulfills a controversial role in the pathogenesis of several cancers (Figure 1). It is clear that SIRT6 plays a crucial role in the regulation of tumorigenesis through its implication in different biological pathways where it can act as tumor suppressor or oncogene. The pleiotropism of SIRT6 means that studies directed toward understanding the cellular mechanisms through which the Sirtuin impacts cancer are difficult to carry forward but tremendously exciting. As Cagnetta et al. suggest,24 it is important that SIRT6 be included in the prospective clinical trials as a novel strategy of anti-tumor therapy.

References 1. Gertler AA, Cohen HY. SIRT6, a protein with many faces. Biogerontology. 2013;14(6):629-639. 2. Longo VD, Kennedy BK. Sirtuins in aging and age-related disease. Cell. 2006;126(2):257-268. 3. Rajendran R, Garva R, Krstic-Demonacos M, Demonacos C. Sirtuins: molecular traffic lights in the crossroad of oxidative stress, chromatin remodeling, and transcription. J Biomed Biotechnol. 2011;2011: 368276. 4. Haigis MC, Sinclair DA. Mammalian sirtuins: biological insights and disease relevance. Annu Rev Pathol. 2010;5:253-295. 5. D’Onofrio N, Vitiello M, Casale R, Servillo L, Giovane A, Balestrieri ML. Sirtuins in vascular diseases: emerging roles and therapeutic potential. Biochim Biophys Acta. 2015;1852(7):1311-1322. 6. Lerrer B, Gertler AA, Cohen HY. The complex role of SIRT6 in car-

3


Editorials cinogenesis. Carcinogenesis. 2016;37(2):108-118. 7. Sebastian C, Zwaans BMM, Silberman DM, et al. The histone deacetylase SIRT6 is a tumor suppressor that controls cancer metabolism. Cell. 2012;151(6):1185-1199. 8. Kugel S, Sebastián C, Fitamant J, et al. SIRT6 suppresses pancreatic cancer through control of Lin28b. Cell. 2016;165(6):1401-1415. 9. Lin Z, Yang H, Tan C, et al. USP10 antagonizes c-Myc transcriptional activation through SIRT6 stabilization to suppress tumor formation. Cell Rep. 2013;5(6):1639-1649. 10. Min L, Ji Y, Bakiri L, et al. Liver cancer initiation is controlled by AP1 through SIRT6-dependent inhibition of survivin. Nat Cell Biol. 2012;14(11):1203-1211. 11. Bhardwaj A, Das S. SIRT6 deacetylates PKM2 to suppress its nuclear localization and oncogenic functions. Proc Natl Acad Sci USA. 2016;113(5):E538-547. 12. Zhang J, Yin XJ, Xu CJ, et al. The histone deacetylase SIRT6 inhibits ovarian cancer cell proliferation via down-regulation of Notch 3 expression. Eur Rev Med Pharmacol Sci. 2015;19(5):818-824. 13. Choe M, Brusgard JL, Chumsri S, et al. The RUNX2 Transcription Factor Negatively Regulates SIRT6 Expression to Alter Glucose Metabolism in Breast Cancer Cells. J Cell Biochem. 2015;116(10): 2210-2226. 14. Han Z, Liu L, Liu Y, Li S. Sirtuin SIRT6 suppresses cell proliferation through inhibition of Twist1 expression in non-small cell lung cancer. Int J Clin Exp Pathol. 2014;7(8):4774-4781. 15. Wu M, Seto E, Zhang J. E2F1 enhances glycolysis through suppressing Sirt6 transcription in cancer cells. Oncotarget. 2015;6(13):1125211263.

16. Ming M, Han W, Zhao B, et al. SIRT6 promotes COX-2 expression and acts as an oncogene in skin cancer. Cancer Res. 2014;74(20):5925-5933. 17. Zhang ZG, Qin CY. Sirt6 suppresses hepatocellular carcinoma cell growth via inhibiting the extracellular signal-regulated kinase signaling pathway. Mol Med Rep. 2014;9(3):882-888. 18. Feng XX, Luo J, Liu M, et al. Sirtuin 6 promotes transforming growth factor-β1/H2O2/HOCl-mediated enhancement of hepatocellular carcinoma cell tumorigenicity by suppressing cellular senescence. Cancer Sci. 2015;106(5)559-566. 19. Ran LK, Chen Y, Zhang ZZ, et al. SIRT6 overexpression potentiates apoptosis evasion in hepatocellular carcinoma via BCL2-associated X protein-dependent apoptotic pathway. Clin Cancer Res. 2016;22(13):3372-3382. 20. Elhanati S, Ben-Hamo R, Kanfi Y, et al. Reciprocal regulation between SIRT6 and miR-122 controls liver metabolism and predicts hepatocarcinoma prognosis. Cell Rep. 2016;14(2):234-242. 21. Lefort K, Brooks Y, Ostano P, et al. A miR-34a-SIRT6 axis in the squamous cell differentiation network. EMBO J. 2013;32(16):22482263. 22. Cea M, Cagnetta A, Adamia S, et al. Evidence for a role of the histone deacetylase SIRT6 in DNA damage response of multiple myeloma cells. Blood. 2016;127(9):1138-1150. 23. Bartek J, Lukas J. Chk1 and Chk2 kinases in checkpoint control and cancer. Cancer Cell. 2003;3(5):421-429. 24. Cagnetta A, Soncini D, Orecchioni S, et al. Depletion of SIRT6 enzymatic activity increases acute myeloid leukemia cells vulnerability to DNA-damaging agents. Haematologica. 2018;103(1):80-90.

Remission is good - relapse is bad Paul S. Gaynon Children's Hospital Los Angeles, University of Southern California, Los Angeles, CA, USA E-mail: pgaynon@chla.usc.edu doi:10.3324/haematol.2017.182667

T

he prognostic significance of minimal residual disease (MRD), or perhaps 'measurable' residual disease,1 is well-established acute and chronic leukemia.2,3 The vast effort of European investigators in standardizing MRD assessment by polymerase chain reaction (PCR) and flow cytometry merits recognition and credit.4,5 At present, we have several independent quantitative monitoring strategies, namely, PCR on DNA targets, reverse transcription (RT)-PCR on abnormal ribonucleic acid (RNA) transcribed from fusion genes or overexpression of normal messenger (m)RNA, and flow cytometry. Their relative implications remain under investigation. MRD results, whatever the target, depend on specimen quality. Marrow aspirates represent a variable mixture of marrow and peripheral blood. Sensitivity depends on the number of cells or amount of nucleic acid interrogated. Leukemia may present with uniform marrow replacement and remit homogeneously across the marrow. Early relapse, however, may be patchy or perhaps anatomically localized with only later dissemination. Peripheral blood may be of use, despite a consistently lower and not always predictable presence of leukemic blasts in the peripheral blood relative to the bone marrow.6 The comparison of quantitative MRD strategies based on DNA and RNA is complex. The DNA target may persist from residual dying cells or in cells lacking leukemogenic potential, vis-à-vis the persistence of DNMT3A mutations in acute myeloid leukemia (AML),7 represent4

ing clonal hematopoiesis and not always associated with relapse. While one or two copies of DNA targets are present per cell, the expression of both the target RNA and the housekeeping genes employed as denominators can vary from patient to patient, and from cell to cell for individual patients. Interventions may affect gene expression as well as cell number. The RNA target may also be present in cells lacking leukemogenic potential. RNA is more labile than DNA. In this issue of Haematologica, Cazzaniga et al. compare MRD monitoring by RQ-PCR of DNA-based rearranged immunoglobulin/ T-cell receptor gene rearrangements (IG/TR), and of RNA-based BCR/ABL1 fusion transcript in 90 young people with Philadelphia chromosome-positive acute lymphoblastic leukemia (PH+ ALL) who were allocated to imatinib on the European intergroup study of post-induction treatment of PH+ ALL (EsPhALL; EudraCT 2004-0014647-30; clinicaltrials.gov Identifier: 00287105). Of the 57 patients characterized, about 90% had the p190 transcript and 10% the p210 transcript.8 Imatinib treatment was initiated after the first time point (tp1), at the completion of Induction IA at 5-7 weeks from diagnosis, and continued intermittently. Contemporary protocols for PH+ ALL begin tyrosine kinase inhibitors earlier and continue them without interruption. None of the nine patients with undetectable MRD by PCR targeting IG /TR after one month of therapy (end induction IA) relapsed. MRD positive patients had a similar ~35% relapse rate, whether MRD was quantifiable (≥ haematologica | 2018; 103(1)


Editorials

5x10-4) or positive below the quantifiable range (< 5x10-4). Imatinib began with Induction 1B. MRD by IG /TR at the end of Induction IB (time point 2, tp2) was again prognostic. Fourteen of 64 patients first became negative at tp2 and had a 14% relapse rate. MRD was monitored with each subsequent high-risk (HR) Block. Eleven of 37 and 7 of 21 patients first became negative after HR Block 1 and HR Block 2, respectively. Attaining negativity after tp2 carried no apparent benefit. One might attribute this revelation to the vagaries of small numbers. Alternatively, one might ask whether the persistence of excessive disease for too long a period of time provided an opportunity for mutation and the eventual emergence of resistant clones, despite the eventual eradication of the clones detectable from diagnosis. Of interest, MRD response correlated well with conventional age and white blood cell count-based risk classification. In addition, while 7/10 patients with positive but unquantifiable MRD at tp1 prior to treatment with imatinib became negative at tp2 after initiating imatinib therapy, only 7/54 quantifiable MRD positive patients became negative at tp2, despite the imatinib regimen (P<0.01, chisquared test). The response to the initial conventional cytotoxic chemotherapy and the response to subsequent therapy, including imatinib, appear to be linked. BCR/ABL1 negativity at tp1 and tp2, like IG/TR negativity, carried a favorable prognosis. BCR/ABL1 and IG/TR estimates of MRD were concordant for 69% of paired samples, although numerical values for BCR/ABL1 were higher at tp1 and tp2, where sample numbers were sufficient to make a useful comparison. Curiously, when MRD is assessed by flow cytometry, outcomes worsen stepwise with increasing values.9 With PCR-based assays, results which are positive but below the quantifiable range still carry a high risk of relapse, both in PH+ ALL and in other patients with B-cell ALL (B-ALL). The Berlin-Frankfurt-Münster risk assignment algorithm is based on the persistence of MRD, more than the absolute MRD level.10 Any positivity at tp1 or tp2, quantifiable or non-quantifiable, excludes patients from the standard-risk group. The persistence of MRD ≥ 10-3 at tp2 places patients in a higher risk group. RQ-PCR for BCR/ABL1 assesses fusion transcript. The marker is clonal, not sub-clonal, and perhaps even 'supraclonal'. Expression may not be limited to fully leukemogenic clones or even to lymphocytes. The authors cite Hovorkova et al. who found discordance in about 20% of cases with BCR/ABL1 positivity in T-lymphocytes, unlike chronic myelogenous leukemia (CML), but not in putative stem cells (CD4+, CD38-, CD133+).11 This was true both for patients with p190 transcripts associated with ALL and patients with p210 transcripts associated with CML. Similarly, in AML the persistence of DNMT3A mutations are common, representing clonal hematopoiesis and not always associated with relapse.7 Remission is good and relapse is bad. Therapy fails weeks or months before relapse is clinically apparent. Aggressive monitoring for submicroscopic relapse (molecular failure) has received little attention in pediatric B-ALL due to the generally low rates of relapse and prolonged years of risk.10 In the past, two-thirds of pediatric relapses

haematologica | 2018; 103(1)

occurred in the first 3 years after diagnosis. Masurekar et al. have now established that on the contrary, two-thirds of relapses now occur after 3 years.12 Early recognition of treatment failure has received more attention in adult ALL, where relapse is more common and the time to relapse is shorter.13 However, certain subsets of pediatric ALL, such as PH+ ALL, severe hypodiploid ALL, and infant KMT2a-rearranged ALL still have substantial early failure rates. New therapeutic modalities, such as blinatumomab, inotuzumab, and chimeric antigen receptor (CAR)-T cells,14 may place a new premium on prompt recognition of treatment failure. Our ability to detect MRD reliably will lead to new definitions of clinical treatment failure.

References 1. Hourigan CS, Gale RP, Gormley NJ, Ossenkoppele GJ, Walter RB. Measurable residual disease testing in acute myeloid leukaemia. Leukemia. 2017;31(7):1482-1490. 2. Pui C-H, Pei D, Raimondi SC, et al. Clinical impact of minimal residual disease in children with different subtypes of acute lymphoblastic leukemia treated with response-adapted therapy. Leukemia. 2017;31(2):333-339. 3. Nunes V, Cazzaniga G, Biondi A. An update on PCR use for minimal residual disease monitoring in acute lymphoblastic leukemia. Expert Rev Mol Diagn. 2017;17(11):953-963. 4. Lucio P, Gaipa G, van Lochem EG, et al. BIOMED-1 concerted action report: flow cytometric immunophenotyping of precursor B-ALL with standardized triple-stainings. Leukemia. 2001;15(8):1185-1192. 5. Pongers-Willemse MJ, Seriu T, Stolz F, et al. Primers and protocols for standardized detection of minimal residual disease in acute lymphoblastic leukemia using immunoglobulin and T cell receptor gene rearrangements and TAL1 deletions as PCR targets Report of the BIOMED-1 CONCERTED ACTION: Investigation of minimal residual disease in acute leukemia. Leukemia. 1999;13(1):110-118. 6. Zeijlemaker W, Kelder A, Oussoren-Brockhoff YJM, et al. Peripheral blood minimal residual disease may replace bone marrow minimal residual disease as an immunophenotypic biomarker for impending relapse in acute myeloid leukemia. Leukemia. 2016;30(3):708-715. 7. Debarri H, Lebon D, Roumier C, et al. IDH1/2 but not DNMT3A mutations are suitable targets for minimal residual disease monitoring in acute myeloid leukemia patients: a study by the Acute Leukemia French Association. Oncotarget. 2015;6(39):42345-42353. 8. Cazzaniga G, De Lorenzo P, Alten J, et al. Predictive value of minimal residual disease in Philadelphia-chromosome-positive acute lymphoblastic leukemia treated with imatinib in the European intergroup study of post-induction treatment of Philadelphia-chromosome-positive acute lymphoblastic leukemia, based on immunoglobulin/T-cell receptor and BCR/ABL1 methodologies. Haematologica. 2018;103 (1):107-115. 9. Borowitz MJ, Devidas M, Hunger SP, et al. Clinical significance of minimal residual disease in childhood acute lymphoblastic leukemia and its relationship to other prognostic factors: a Children's Oncology Group study. Blood. 2008;111(12):5477-5485. 10. van Dongen JJM, van der Velden VHJ, Brüggemann M, Orfao A. Minimal residual disease diagnostics in acute lymphoblastic leukemia: need for sensitive, fast, and standardized technologies. Blood. 2015;125(26):3996-4009. 11. Hovorkova L, Zaliova M, Venn NC, et al. Monitoring of childhood ALL using <em>BCR-ABL1</em> genomic breakpoints identifies a subgroup with CML-like biology. Blood. 2017;129(20):2771-2781. 12. Masurekar AN, Parker CA, Shanyinde M, et al. Outcome of central nervous system relapses in childhood acute lymphoblastic leukaemia - Prospective open cohort analyses of the ALLR3 trial. PLoS One. 2014;9(10):e108107. 13. Gokbuget N, Kneba M, Raff T, et al. Adult patients with acute lymphoblastic leukemia and molecular failure display a poor prognosis and are candidates for stem cell transplantation and targeted therapies. Blood. 2012;120(9):1868-1876. 14. Valecha GK, Ibrahim U, Ghanem S, Asti D, Atallah J-P, Terjanian T. Emerging role of immunotherapy in precursor B-cell acute lymphoblastic leukemia. Expert Rev Hematol. 2017;10(9):783-799.

5


Editorials

High-throughput sequencing for rapid diagnosis of inherited platelet disorders: a case for a European consensus Alan T. Nurden and Paquita Nurden Institut de Rythmologie et de Modélisation Cardiaque, Plateforme Technologique d’Innovation Biomédicale, Hôpital Xavier Arnozan, Pessac, France E-mail: nurdenat@gmail.com doi:10.3324/haematol.2017.182295

T

he diagnosis of bleeding disorders caused by inherited defects of platelet function or production (or combinations of both) forms an intrinsic part of the work of every hematological laboratory.1 The classic procedure for the work-up of inherited platelet disorders (IPDs) is based on an initial examination of the patient by a specialist in the field, followed by a series of often complex biological tests designed to identify molecular pathways likely to be affected in each case. Algorithms have been proposed to aid in this task while human phenotype ontology (HPO) annotation and cluster analysis has been recommended as part of phenotyping.2,3 However, in many laboratories the tests chosen are often limited by availability, while intensive HPO interrogation is time-consuming and often not performed in the context of an initial hospital consultation. Despite such restrictions, diagnosis advances sufficiently for many patients to justify DNA sequencing of selected genes; a situation usually reserved for patients with classic IPDs and a well-defined phenotype. Frequently the work-up is long and costly, and requires the patient to make multiple visits to his/her local hospital, and possibly to a specialized centre. Too often, the result ends in the lack of a clearly defined diagnosis. In the current edition of Haematologica, Bastida et al.4 present a high-quality pilot study introducing highthroughput gene sequencing into the mainstream of genetic diagnostic practice for IPDs. Working in the context of a national collaboration, these authors designed a novel platform to identify gene variants in 82 patients, the majority of whom come from the Iberian Peninsula, with IPDs of previously undiagnosed molecular causes. Their procedure consisted of using 1399 probes targeting 1106 regions of 72 genes known to be associated with IPDs and/or encoding proteins significant in platelet physiology. With step-by-step use of bioinformatics tools and informed interpretation of clinical and biological data, they were able to identify likely disease-causing genes in the majority of patients. First, they validated their platform by successfully confirming the mutations previously assigned by Sanger sequencing in an additional group of 10 patients with known IPDs. For 34 patients with a high suspicion of defined IPDs, they identified candidate variants in 30 patients (88.2%) (Figure 1). The success rate for a cohort of 48 patients of uncertain etiology was lower, with 26 variants identified in 16 genes in 26 cases. In 22 cases disease-causing genes were not identified. Pioneering studies worldwide in the application of next-generation sequencing projects to IPDs came from the BRIDGE-Bleeding and Platelet Disorders (BPD) consortium led by Professor Willem Ouwehand, Cambridge, 6

UK; a project in which we were early members, submitting DNA from 80 well-characterized patients in 2011. BRIDGE initially used whole exome sequencing (WES), and later whole genome sequencing (WGS) to identify gene variants, while also establishing the ThromboGenomics gene platform, which originally contained 63 genes.5,6 The BRIDGE studies represent the gold standard in the discipline, helped by the sequencing capacity of the Sanger Institute and their extensive backup bioinformatic analysis. In Europe, in parallel, a Birmingham-based UK group led by Professor Steve Watson looked specifically at UK patients with IPDs in the Genotyping and Phenotyping of Platelets (GAPP) study.7 The data obtained from these groups and others world-wide8,9 helped Bastida et al.4 to compose their gene platform. However, one disadvantage of selected gene platforms is that successful identification of disease-causing variants is limited to the genes tested, and this may explain the 22 cases where the causal gene defect was not found in the Spanish study. WES, and particularly WGS, offer the potential to identify new genes, especially when combined with selective approaches such as comparing patient phenotypes with those of knock-out mouse models or gene datasets; approaches that allowed us to identify TRPM7 (from an existing mouse model) and DIAPH1 (from a search of genes responsible for deafness) in patients belonging to the French cohort which we submitted to BRIDGE.10,11 Recent advances in bioinformatic analysis of next-generation sequencing data are enhancing causal gene identification in IPDs, helped by HPO questioning and the study of variant penetrance within large families.12 It would be interesting to now apply WGS to those cases in the Spanish cohort for whom disease-causing variants were not identified. It should be underlined that finding a novel gene variant is not in itself final proof that it is causing the disease, even with penetration within the family. Cosegregated but non-highlighted genetic variants may also be contributing to phenotype. Thus, structure/function studies on the protein are often necessary, and/or studies on mouse or zebrafish models as was performed for a recently described SRC variant;13 a situation that now applies to several of the gene variants identified by Bastida et al.4 Improving bioinformatic approaches may also be helpful in identifying gene variants which often display heterozygous expression and only cause bleeding in rare phenotypes when acting in combination. Haplogroups of genetic variants that are individually benign may be widespread, originating in part from now dispersed ethnic and religious minorities. In this regard, we were surhaematologica | 2018; 103(1)


Editorials

Figure 1. Genotype/phenotype in IPDs: an Iberian Peninsula Study. An illustration showing a platelet (center) surrounded by boxes naming the genes that are arbitrarily grouped into gene families constituting the platform composed by Bastida et al.4 The genes with mutations thought likely or probable to be pathogenic for IPDs within the Iberian Peninsula are highlighted in blue. In red are those genes recently highlighted in the literature to be disease-causing, and which are candidates to be added to the platform.

prised upon discovering that when comparing single nucleotide polymorphisms (SNPs) within selected domains of ITGA2B and ITGB3 in patients with Glanzmann thrombasthenia (GT), combinations of SNPs reoccurred in ostensibly non-related families.14 Furthermore, studies on GT have clearly shown that while defects in ITGA2B and ITGB3 confirm the disease, other genetic and epidemiological factors define bleeding risk and possibly the frequency of other associated pathologies.15 This suggests that next-generation sequencing data for IPDs should also be analyzed and gene platforms designed to identify gene variants that may influence bleeding independently of the disease-causing genotype and/or increased susceptibility for associated illnesses â&#x20AC;&#x201C; an example being the risk of hematological malignancy with genes such as RUNX1 or ETV6 or with defects in immunity, as has been recently performed for inhibitor development in hemophilia A.16,17 This would be of great benefit with regard to patient care. So, what is the best way to continue? While highlighting the advances in Europe, the Spanish-based study in the current issue of Haematologica by Bastida et al.,4 or the parallel Scandinavian study18 show the now obvious need for greater collaboration and standardization. In Europe, many countries already have national or regional sequencing centers specializing in rare diseases, while haematologica | 2018; 103(1)

public databases such as the Exome Aggregation Consortium (ExAC) provide continued analysis of rare gene variants through input from genome-wide association studies and large-scale sequencing, such as the 1000 genomes project. Although easier to manage, highthroughput gene screening platforms need to evolve continuously as disease-causing variants continue to be identified and the whole genome landscape of IPDs emerges.19,20 The ideal would be to establish a consensus for gene platforms within Europe, permitting local reference centers throughout the European Community to analyze IPD samples. For difficult cases, the next logical step would be to provide back-up links to specialized centers, as highlighted by BRIDGE, with facilities for WGS in order to research not only new disease-causing variants but also the effects of gene enhancers and pleiotropy. Ideally a result should be obtained within days for the bulk of patients, both for the cause of a disease and to predict the risk of bleeding. Thus, next-generation technology can form part of frontline hospital practice, with biological tests performed on the basis of gene testing rather than vice versa. It would also permit uniformity while conserving national identity. In the platelet field this would lead to cost-cutting, increased efficiency and much improved patient care. Advances in Europe would also provide a model for other parts of the world. 7


Editorials

References 1. Gresele P, Harrison P, Bury L, et al. Diagnosis of suspected inherited platelet function disorders: results of a worldwide survey. J Thromb Haemost. 2014; 12(9):1562-1569. 2. Noris P, Pecci A, Di Bari F, et al. Application of a diagnostic algorithm for inherited thrombocytopenias to 46 consecutive patients. Haematologica. 2004; 89(10):1219-1225. 3. Westbury SK, Turro E, Greene D, et al. Human phenotype ontology annotation and cluster analysis to unravel genetic defects in 707 cases with unexplained bleeding and platelet disorders. Genome Med. 2015; Apr 9;7(1):36. Doi: 10.1186/s13073-015-015-5. eCollection 2015. 4. Bastida JM, Lozano ML, Benito R, et al. Introducing high-throughput sequencing into the mainstream of genetic diagnosis practice in inherited platelet disorders. Haematologica. 2018;103(1):000-000. 5. Lentaigne C, Freson K, Laffan MA, Turro E, Ouwehand WH, on behalf of the BRIDGE-BPD Consortium and the ThromboGenomics Consortium. Inherited platelet disorders: towards DNA-based diagnosis. Blood 2016; 127(23):2815-2823. 6. Simeoni I, Stephens JC, Hu F., et al. A high-throughput sequencing test for diagnosing inherited bleeding, thrombotic, and platelet disorders. Blood 2016; 127(23):2791-2803. 7. Johnson B, Lowe GC, Futterer J, et al. Whole exome sequencing identifies genetic variants in inherited thrombocytopenia with secondary qualitative function defects. Haematologica 2016; 101(10):1170-1179. 8. Kahr WH, Pluthero FG, Elkadri A, et al. Loss of the Arp2/3 complex component ARPC1B causes platelet abnormalities and predisposes to inflammatory disease. Nat Commun 2017; Apr 3;8:14816. doi: 10.1038/ncomms14816. 9. Ferreira CR, Chen D, Abraham SM, et al. Combined alpha-delta platelet storage pool deficiency is associated with mutations in GFI1B. Mol Genet Metab 2017; 120(3):288-294. 10. Stritt S, Nurden P, Favier R, et al. Defects in TRPM7 channel function

8

11. 12. 13.

14.

15. 16. 17. 18.

19. 20.

deregulate thrombopoiesis through altered cellular Mg(2+) homeostasis and cytoskeletal architecture. Nat Commun 2016; Mar 29:711097. Doi: 10. 1038/ncomms 11097. Stritt S, Nurden P, Turro E, et al. A gain-of-function variant in DIAPH1 causes dominant macrothrombocytopenia and hearing loss. Blood 2016; 127(23):2903-2914. Greene D, NIHR BioResource, Richardson S, Turro E. A fast association test for identifying pathogenic variants involved in rare diseases. Am J Hum Genet 2017; 101(1):104-114. Turro E, Greene D, Wijgaerts A, et al. A dominant gain-of-function in universal tyrosine kinase SRC causes thrombocytopenia, myelofibrosis, bleeding, and bone pathologies. Sci Transl Med 2016; 8(328):328ra30. doi: 10.1126/scitranslmed.aad7666. Epub 2016 Mar 2. Pillois X, Nurden AT. Linkage disequilibrium amongst ITGA2B and ITGB3 gene variants in patients with Glanzmann thrombasthenia confirms that most disease-causing mutations are recent. Br J Haematol 2016; 175(4):686-695. Nurden AT. Should studies on Glanzmann thrombasthenia not be telling us more about cardiovascular disease and other major illnesses? Blood Rev 31(5):287-299. Sood R, Kamikubo Y, Liu P. Role of RUNX1 in haematological malignancies. Blood 2017; 129(15):2070-2082. Gorski MM, Blighe K, Lotta LA, et al. Whole-exome sequencing to identify genetic risk variants underlying inhibitor development in severe haemophilia A patients. Blood 2016; 127(23):2924-2933. Leinoe E, Zetterberg E, Kinalis S, et al. Application of whole-exome sequencing to direct the specific functional testing and diagnosis of rare inherited bleeding disorders in patients from the Oresund region, Scandinavia. Br J Haematol 2017; 178(2):308-322. Astle WJ, Elding H, Jiang T, et al. The allelic landscape of human blood cell trait variation and links to common complex disease. Cell 2016; 167(5):1415-1429. Chami N, Chen MH, Slater AY, et al. Exome genotyping identifies pleiotropic variants associated with red blood cell traits. Am J Hum Genet 2016; 99(1):8-21.

haematologica | 2018; 103(1)


REVIEW ARTICLE

The prothrombotic state in paroxysmal nocturnal hemoglobinuria: a multifaceted source

Ferrata Storti Foundation

Barnaby Peacock-Young,1 Fraser L. Macrae,1 Darren J. Newton,2 Anita Hill3* and Robert A S Ariëns1*

Thrombosis and Tissue Repair Group, Division of Cardiovascular and Diabetes Research, Leeds Institute of Cardiovascular and Metabolic Medicine, Multidisciplinary Cardiovascular Research Centre, University of Leeds; 2Section of Experimental Haematology, Leeds Institute of Cancer and Pathology, University of Leeds and 3 Department of Haematology, St James’s University Hospital, Leeds, UK 1

*AH and RA contributed equally to this work.

Haematologica 2018 Volume 103(1):9-17

ABSTRACT

P

aroxysmal nocturnal hemoglobinuria is a rare acquired hematological disorder, the most serious complication of which is thrombosis. The increased incidence of thrombosis in paroxysmal nocturnal hemoglobinuria is still poorly understood, but unlike many other thrombotic disorders, predominantly involves complement-mediated mechanisms. This review article discusses the different factors that contribute to the increased risk of thrombosis in paroxysmal nocturnal hemoglobinuria. Paroxysmal nocturnal hemoglobinuria leads to a complex and multifaceted prothrombotic state due to the pathological effects of platelet activation, intravascular hemolysis and neutrophil/monocyte activation. Platelet and endothelial microparticles as well as oxidative stress may play a role. Impaired fibrinolysis has also been observed and may be caused by several mechanisms involving interactions between complement activation, coagulation and fibrinolysis. While many factors may affect thrombosis in paroxysmal nocturnal hemoglobinuria, the relative contribution of each mechanism that has been implicated is difficult to quantify. Further studies, including novel in vivo and in vitro thrombosis models, are required in order to define the role of the individual mechanisms contributing to thrombosis, impaired fibrinolysis and clarify other complement-driven prothrombotic mechanisms in paroxysmal nocturnal hemoglobinuria.

Correspondence: r.a.s.ariens@leeds.ac.uk

Received: August 8, 2017. Accepted: November 23, 2017. Pre-published: December 15, 2017.

Introduction

doi:10.3324/haematol.2017.177618

Paroxysmal nocturnal hemoglobinuria (PNH) is a rare hematological disorder of multipotent hematopoietic stem cells. It is caused by an acquired mutation in the Xlinked phosphatidylinositol glycan class A gene (PIG-A), causing stem cell progeny (mature blood cells) to lack complement regulatory proteins and exposing them to complement attack.1 Thromboembolic events are the most common cause of morbidity and mortality in PNH and account for 40-67% of deaths; 40% of patients having suffered an event before diagnosis and 29-44% of patients suffering at least one event throughout the course of their disease.2 Despite such a large role in the burden of the disease, the mechanism behind the development of thrombosis is poorly understood, highlighting the importance of thrombosis management in PNH patients and elucidating more information regarding the nature of the thrombotic event.3 Multiple proposed mechanisms behind the increased incidence of thrombosis include a prothrombotic state in conjunction with platelet abnormalities and impaired fibrinolysis.3,4 The review herein will discuss changes to the hemostatic system in PNH, and highlight areas that require future research in the prothrombotic processes involved in PNH.

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/9 ©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.

Paroxysmal Nocturnal Hemoglobinuria Pathophysiology The incidence of PNH is estimated at 0.1–0.2/100 000 persons per year.5 PNH is haematologica | 2018; 103(1)

9


B. Peacock-Young et al.

caused by an acquired inactivating mutation of the PIG-A gene located on the X chromosome. The PIG-A gene codes for an enzyme involved in the formation of the N-acetylglucosaminyl phosphatidylinositol biosynthetic protein which is necessary for the first step in the biosynthesis of glycosylphosphatidyinositol (GPI) anchors.6 GPI, a glycolipid moiety, anchors numerous proteins to the cell surface, with >12 GPI-anchored proteins (GPI-APs) located on hemopoietic cells.1 Studies have shown the presence of a small number of GPI anchor deficient cells in the blood of healthy controls as well as in patients with PNH.7 This implies that the presence of the PIG-A mutation alone is not sufficient to allow the PNH clone to dominate. The process behind the clonal expansion of the PIG-A mutated stem cells in PNH patients is not fully understood. Two mutually cooperative hypotheses exist to explain the clonal expansion of PNH cells; one involves immune selection-mediated expansion and the second predicts that dominant PNH clones acquire a growth advantage.8 The UL16 binding protein 1 (ULBP1), a stress-induced ligand for the NKG2D receptor, is a GPI-linked glycoprotein thought to be lost on PNH stem cells. Loss of ULBP1 may prevent their destruction by NKG2D+ lymphocytes allowing for immune-escape from ULBP-NKG2D engagement in the bone marrow.9 PIG-A mutant cells have been demonstrated to be less sensitive to T lymphocytes and, along with leukemic cells with the same mutation, possess increased resistance to natural killer cells.9,10 The growth phenotype is thought to be due to the observed upregulation of the early growth response factor 1 gene (EGR-1) and ectopic expression of the high mobility transcription factor coding genes HMGA2 has been reported in a few cases.11 The two most significant GPI-APs thought to play a major role in the pathophysiology of PNH are complement regulatory proteins CD55 and CD59.12 CD55 interacts with C4b and C3b and interferes with their ability to analyzed the conversion of C2 and factor B to active C2a and Bb, thus preventing the formation of C4b2a and C3bBb (both forms of C3 convertase). CD59 interacts directly with the membrane attack complex (MAC), formed at the end of the immune complement cascade, preventing pore formation and cell lysis.13 The lack of these complement regulatory proteins allows for complement attack leading to erythrocyte lysis, platelet activation and loss of thrombotic modulators on granulocytes, causing many of the symptoms of PNH.14-16 Exacerbations, or ‘paroxysms’, are sudden increases in symptoms, most noticeably hemoglobinuria and anemia, caused by infection or other inflammatory stimuli.17 Eculizumab, a monoclonal antibody inhibiting C5 cleavage, has been shown to significantly reduce the symptoms of PNH as well as associated morbidity and mortality, and is currently the only licensed treatment for PNH patients.18

Clinical presentation Patients may present with ‘classical’ PNH characterized by clinical and laboratory evidence of intravascular hemolysis with no clinical evidence of an underlying bone marrow disorder. Others have evidence of hemolytic PNH as well as clinical evidence of bone marrow disorder, such as myelodysplasia or aplastic anemia. A further group may be defined as ’subclinical PNH’ where a small proportion of PNH cells are found but with no evidence of hemolysis or thrombosis.19 The presence of PNH cells is identified using flow cytometry, determining the proportion of GPI negative granulocytes, monocytes and erythrocytes.19 PNH red 10

blood cells can be labelled type I, II or III; type I cells have normal expression of GPI-APs, type II have partial deficiency and type III lack all GPI-APs.20 The clinical manifestations are variable; intravascular hemolysis, thrombosis and anemia are significant, however, other symptoms may be present.1 Schrezenmeier et al. analyzed 1610 patients and showed the proportion of symptoms, such as fatigue (80%), dyspnoea (64%), hemoglobinuria (62%), abdominal pain (44%), chest pain (33%) and impaired renal function (14%), with only 16% of patients having a history of thrombotic events.21 However, Hill et al. have shown that the presence of subclinical thrombosis is significantly underestimated.22 The ongoing effect of intravascular hemolysis, as previously described, is responsible for causing most of the symptoms in PNH (Figure 1). Intravascular hemolysis results in the release of free hemoglobin which is normally cleared by haptoglobin, hemopexin and the scavenger CD163. These clearing mechanisms are overwhelmed in PNH and lead to the accumulation of high levels of free hemoglobin in the plasma, resulting in the scavenging and depletion of nitric oxide (NO).23 The subsequent excess of hemoglobin leads to the visible hemoglobinuria, while the depletion of NO, a potent vasodilator, results in vasoconstriction, decreased regional blood flow and muscular contraction, causing chest and abdominal pain, amongst other symptoms.3 Moyo et al. reported significant differences in the proportion of PNH cells in patients with symptoms of abdominal pain, hemoglobinuria and esophageal spasm (causing chest pain and dysphagia). The mean proportion of PNH granulocytes in patients with these symptoms was at least twice that of patients who did not possess these clinical manifestations.24 As previously described, thrombosis is the most serious complication associated with PNH. Thrombotic events are reported to be of venous origin in 85% of cases, arterial in 15% of cases and involve more than one site at the same time in 20.5% of cases.5 Thrombosis can occur at any site, with deep vein thrombosis, pulmonary embolism, myocardial infarction or cerebral vascular attack all commonly observed complications.25 There appears to be an increased incidence of thrombosis at atypical sites, such as the hepatic vein resulting in Budd-Chiari syndrome, occurring in 4044% of PNH patients, in addition to thrombosis in the vasculature of the central nervous system, mesenteric, dermal veins and the cavernous sinus.26 The proportion of PNH cells and clone size has also been associated with thrombotic complications. Hall et al. demonstrated that in patients with a proportion of PNH granulocytes greater than 50%, the 10year risk of thrombosis was 44%, but in patients with a proportion of PNH granulocytes less than 50% the risk was 5.8%.27 Through logistical regression, Moyo et al. calculated the increase in odds ratio for thrombosis to be 1.64 for each 10% increase in the proportion of PNH cells.24 These findings correlate with other studies which have shown that the occurrence of thrombosis is noticeably elevated in PNH patients with a proportion of PNH cells as low as 10% when compared to normal population controls.28,29

Platelet activation The mechanisms behind thrombus formation in PNH are complex and subject to continued research. Interactions between the complement system, platelets and coagulation likely explain some of the increased risk of thrombosis. Due to the multifactorial and variable nature of the disease, it is haematologica | 2018; 103(1)


Multifaceted prothrombotic state in PNH

likely that a combination of several factors may contribute to the increased incidence of thrombus formation and associated mortality (Figure 2). Platelets have been reported to play a significant role in the formation of thrombus in PNH patients, by both contributing to a prothrombotic state and initiating clot formation.30 One would expect that due to the deficiency of CD55 and CD59, lysis of platelets occurs and contributes to thrombocytopenia. However, this is not the case, as the lifespan of platelets in PNH patients is normal.31 Rather than complement resulting in the lysis of platelets, an intrinsic mechanism of adaption and resistance to complement attack has been observed which subsequently contributes to the prothrombotic state.32 It has been shown that upon increased deposition of MAC (C5b-9) on the membrane of platelets, rather than lysis, complement accumulation on the platelet surface triggers morphological changes.15 The loss of platelet membrane phospholipid asymmetry through the action of an adenosine triphosphate (ATP)dependent enzyme, gelsolin, aminophospholipid translocase, lipid scramblase and calpain allow for cytoskeletal and phospholipid bilayer changes.33 The now activated platelet secretes α-granules, and in conjunction with membrane depolarization, α-granules fuse with the platelet membrane.34 This results in exocytosis of the vesiculated MAC complex and the production of prothrombotic plateletderived microparticles (PMPs).15

Platelet-Derived Microparticles The exact role of PMPs is not fully understood, however, they are considered to play a role in the generation of a prothrombotic state in PNH.15 Three key hypotheses for the

prothrombotic nature of PMPs have been identified.35 First, platelet microparticles are formed by platelets upon activation, therefore their membranes possess the same prothrombotic properties as the activated platelet membrane.36 Second, PMPs can bind clotting cascade components, such as activated factors V (Va) and VIII (VIIIa); furthermore, the densities of these protein binding sites on PMPs appear to exceed those on activated platelet membranes.37,38 Finally, when isolated PMPs are added back to platelet free pooled plasma without the addition of coagulation activators, they trigger thrombin generation, demonstrating that microparticles generated in vivo can stimulate coagulation.39 Sinauridze et al. estimated that PMP membranes have a 50to 100-fold higher specific procoagulant activity than activated platelets.35 PMPs have been shown to express many of the following membrane binding protein complexes which are normally observed on activated platelets: glycoprotein Ib (GPIb), which binds von Willebrand factor (VWF) initiating formation of the platelet plug;40 platelet endothelium adhesion molecule (PECAM-1), an immunoglobulin superfamily member involved in leukocyte transmigration, angiogenesis, and integrin activation;41 the integrin glycoprotein IIb/IIIa (GpIIb-IIIa) or αIIbβ3, a receptor for fibrinogen and VWF, further aiding platelet aggregation and plug formation;42 and Pselectin, a cellular adhesion molecule which exacerbates symptoms via a feedback loop through continued stimulation of the alternative pathway, initiating activation of the classical pathway as well as stimulating further platelet aggregation.43 The expression of membrane proteins involved in thrombus formation on platelet microparticles suggests an ability to contribute to the prothrombotic state.

Figure 1. Summary of the multiple factors thought to contribute to the prothrombotic state in PNH and interaction. Further detail is provided in the text. MAC: membrane attack complex; NET: neutrophil extracellular traps; TFPI: tissue factor pathway inhibitor; VWF: von Willebrand factor; WPB: Weibel-Palade bodies; NO: nitric oxide; ROS: reactive oxygen species; TF: tissue factor.

haematologica | 2018; 103(1)

11


B. Peacock-Young et al.

However, further studies are necessary to analyze and quantify their specific role in PNH-induced risk of thrombosis. Phosphatidylserine is expressed on the surface of plateletderived microparticles as a result of MAC binding-induced morphological changes.44 Phosphatidylserine is normally confined to the inner leaflet of the platelet membrane, however, it is translocated to the outer leaflet due to the action of the lipid scramblase enzyme and exposed as a result of platelet activation.33 Phosphatidylserine interacts via positively charged calcium ions with negatively charged γ-carboxyglutamic acid (GLA) domains in vitamin K-dependent clotting factors, e.g., VII (FVII), IX, X, and prothrombin.45 This catalyzes the formation of the procoagulant enzyme complexes prothrombinase (VaXa) and tenase (VIIIaIXa),37 and allows for the accelerated conversion of prothrombin to thrombin and stimulation of the coagulation cascade.44 Coagulopathy has been observed in patients with Castaman’s defect and Scott syndrome, and is thought to result from defects in the action of scramblases to translocate phosphatidylserine to the membrane surface.46,47

Residual Activated Platelets The prothrombotic properties of residual activated platelets (platelets post microparticle production) is still a matter of discussion. Activated platelets in PNH patients have been shown to possess greater than ten times the factor V binding sites compared with those from normal controls.16 Activated platelets have been shown to promote thrombus formation through neutrophil interaction, resulting in the release of serine proteases and nucleosomes, activating factor X.48 Like their exocytosed microparticles, activated platelets also express P-selectin, which is thought to further stimulate the complement pathway.49 Surprisingly, a study by Grünewald et al. found evidence of hyporeac-

tive platelets in PNH.50 It is possible that the failure of activated platelets to bind fibrinogen, VWF and to aggregate is due to receptor GpIIb-IIIa complexes in proximity to MAC pores becoming uncoupled from the intracellular transduction mechanisms normally involved in their activation.37 This observation was consistent with the findings of Gralnick et al., who reported variable amounts of platelet activation and further observed reduced VWF binding.30 Grünewald et al. hypothesized a mechanism of dual causality responsible for platelet hyporeactivity.50 One mechanism is hyperstimulation of platelets due to sustained complement attack.16,37,50 Chronic hyperstimulation of the coagulation system was hypothesized to further downregulate the activity of activated platelets.4,50 This highlights the complex variable nature of thrombosis in PNH, suggesting that platelets post activation possibly have a diminished role in PNH-induced thrombosis in comparison to other thrombotic diseases.

Hemolysis As well as contributing to a wide array of symptoms in PNH, hemolysis is thought to contribute to a prothrombotic state, but its role is becoming increasingly scrutinized.51 Erythrocytes have been reported to produce microparticles as a result of MAC-induced apoptosis.52 Some microparticles observed in PNH have indeed been confirmed to originate from erythrocytes; however, Hugel et al. reported that this was not ‘to a significant extent’ while ‘very high levels’ of microparticles of platelet origin were detected.15 Two studies have concluded that the level of erythrocyte microparticles produced in PNH patients was similar to healthy controls;15,53 hence, it is possible that the contribution of erythrocyte microparticles to the prothrombotic state in PNH is only minimal.

Figure 2. Hemostasis mechanisms in PNH patients are imbalanced towards thrombosis. NET: neutrophil extracellular traps; WPB: Weibel-Palade bodies; ROS: reactive oxygen species; NO: nitric oxide.

12

haematologica | 2018; 103(1)


Multifaceted prothrombotic state in PNH

Free Hemoglobin and Endothelial Dysfunction Excess free hemoglobin is a further possible mechanism that may underpin the prothrombotic state in PNH. Upon intravascular hemolysis, free hemoglobin is rapidly bound to the serum protein haptoglobin expressed on monocytes/macrophages forming a complex which is then endocytosed and degraded by CD163.24,54,55 The oxygen binding component of hemoglobin, ferrous heme, can be oxidised to ferric heme, resulting in rapid binding to hemopexin.56 The resulting reaction has vasodilatory, antiproliferative, anti-inflammatory, and antioxidant properties through the release of carbon monoxide, the biliverdin metabolite biliverdin reductase, and the uptake of anti-inflammatory interleukin-10 and heme oxygenase into circulating monocytes.54,57 The scavenging mechanisms described above can become saturated resulting in increased levels of free hemoglobin in the circulation, leading to a prothrombotic state in addition to other symptoms.23,24 There is increasing evidence supporting possible prothrombotic effects of free hemoglobin on platelets and the vascular endothelium.58 A recent study by Belcher et al. showed that heme rapidly stimulates the release of Weibel-Palade bodies (WPBs) from the vascular endothelium.59 Degranulation of WPBs releases VWF and P-selectin onto the surface of endothelial cells, stimulating coagulation and the complement cascade.59 In vivo studies have demonstrated that the infusion of crosslinked hemoglobin increased platelet aggregation and adhesion to the endothelium of an injured vessel wall.60 Free hemoglobin has been observed to directly bind to VWF exposed on the endothelium which increases its affinity for the glycoprotein Ib (GPIb) receptor on the surface of platelets.61 Conjointly, the addition of free hemoglobin to human serum causes inhibition of the VWF cleaving protease ADAMTS13, an enzyme critical in limiting platelet thrombus formation.62,63 Heme administration in healthy volunteers has been demonstrated to cause thrombophlebitis, vascular inflammation and obstruction.64 Patients with PNH have also been observed to possess increased levels of endothelial-derived microparticles (EMPs).40,65 GPI-deficient monocytes are thought to release microparticles rich in tissue factor (TF) upon complement damage.66 Uptake of monocyte-derived microparticles concomitantly increased endothelial TF expression while producing EMPs. Two separate studies have observed increased levels of EMPs in patients with PNH.40,67 The number of EMPs produced relative to PMPs is thought to be small, and as such its contribution to the prothrombotic state is minimal.53 Further procoagulant effects of endothelial cells result from prolonged exposure to free heme.68 Endothelial exposure to heme induces tissue factor expression, which can initiate coagulation, and also activates expression of intracellular adhesion molecule 1 (ICAM1), vascular cell adhesion molecule (VCAM1), and Eselectin.69 The now activated endothelial cells recruit inflammatory cells, promoting thrombus formation at the vessel wall.23 This can be further enhanced by pro-inflammatory cytokines and chemokines which have been observed as being over-expressed in other hemolytic disorders.70 It is still unknown whether bone marrow-derived endothelial cells in PNH patients harbour the GPI-AP deficiency.66 Tissue factor pathway inhibitor (TFPI) inhibits tissue factor and therefore coagulation.71 It is predominantly produced by the endothelium (85%), however, haematologica | 2018; 103(1)

platelets, monocytes and plasma are other sources.66,71 TFPI is expressed in two isoforms, TFPI-α and TFPI-β.72 It is still disputed which TFPI isoform is the most abundantly expressed isoform, however TFPI-β, a GPI-AP expressed on ECs, is thought to exert 80% of anticoagulant activity.72,73 Possible deficiency of the GPI anchor of TFPI-β in PNH may therefore reduce the anticoagulant properties of the endothelium in PNH and contribute to thrombus formation.74,75 Anti-thrombin is enhanced by binding to the heparan sulphate receptor, a GPI-linked protein expressed on endothelial cells and hypothesized to be lost in patients with PNH.76 As the heparin sulphate receptor is GPIlinked, its loss is thought to contribute to a prothrombotic state. No studies have fully investigated the significance of the heparan sulphate receptor, however, a compensatory mechanism has been suggested after there was no change in fibrin deposition in animal studies of heparan sulphate deficiency.77

Reactive Oxygen Species Free hemoglobin has also been demonstrated to produce reactive oxygen species (ROS) via two mechanisms:23 the amphipathic heme interacts with the phospholipid membrane, and via the Fenton reaction catalyzes the production of ROS,78 and extracellular hemoglobin autoxidizes to methemoglobin catalyzed by peroxidase enzymes which further generates ROS.79 A study by Amer et al. has also shown that PNH cells have a higher oxidative status when compared to normal cells.80 It is not clear if this is as a result of free hemoglobin, platelet hyperactivity (as previously discussed) or a pre-existing defect in PNH cells that exacerbates these pathologies, thus, further research into this area is necessary. The formation of ROS is well-documented to produce phospholipid disorganisation, induce cytotoxicity and promote inflammation,23,81 and studies have also shown that ROS can directly enhance platelet activation.82 The Fenton reaction also activates protein kinase C as well as ROS which have been demonstrated to activate platelets.82

Neutrophils & Monocytes Neutrophils have also been reported to contribute to the prothrombotic mechanisms in PNH.23,56 ROS produce neutrophil extracellular traps (NETs), exposed extracellular chromatin with peripherally attached enzymes.59 Extracellular chromatin has been observed as a structural component in deep vein thrombi as well as contributing to the pathogenesis involved in their formation.83 Studies have found that histones, which are also released from NETs, increase thrombin generation and platelet activation by impairing the activation of protein C.84 The impairment of protein C activation in turn prevents the inactivation of clotting FV and FVIII which, as examined previously, form the prothrombinase and tenase complexes, respectively, on platelets.85 Recent studies have demonstrated a specific link between NETs and the formation of venous thrombosis. This mechanism may explain the high prevalence of thrombi in veins at atypical sites and warrants further study in PNH.83,84 The membrane attacks complex formation on the surface of monocytes and neutrophils and is also thought to contribute to the procoagulant state in PNH.3,23 Complement-induced cell activation results in the expression of tissue factor as well as plasminogen activator inhibitor 1, contributing to thrombus 13


B. Peacock-Young et al.

formation while impairing fibrinolysis (discussed below).86 Proteinase 3 (PR3), an enzyme thought to reduce thrombin-induced platelet activation binds to the GPI-anchored cofactor NB1 (CD177) expressed on neutrophils.87 A deficiency of NB1 may therefore contribute to platelet activation and exacerbate the procoagulant state through proteolysis of the protein C receptor, degradation of TFPI and upregulation of TF expression on endothelial cells.88-90

Prothrombotic Feedback Mechanisms Thrombin, the generation of which is increased by many of the mechanisms described above, has been observed to independently activate complement proteins C3 and C5.91 Plasmin, an enzyme involved in fibrin clot degradation and stimulated by fibrin itself, has also recently been shown to cleave C5.92 This suggests a feedback mechanism in which thrombin generation, fibrin deposition and fibrinolysis may in turn activate the complement system, which reciprocally leads to more platelets and coagulation activation, exacerbating the thrombotic response (see Figure 1 and Figure 2 for a summary of prothrombotic mechanisms involved in PNH).93 Plasmin has been shown in vitro to initiate the synthesis of platelet activating factor (PAF) from endothelial cells,94 while a further in vivo study has shown a correlation between the concentration of the terminal complement complex (C5b-9) and PAF. This may begin to highlight a mechanism in which the MAC contributes to plasmin-induced synthesis of PAF in endothelial cells, however, it is unclear whether this would contribute to thrombosis or platelet hyporeactivity.95

Nitric Oxide Depletion Hemoglobin and nitric oxide (NO) bind in an irreversible reaction, the rate of which is suggested to increase by up to 500-600 times as a result of intravascular hemolysis and the loss of heme compartmentalisation.96 Intravascular hemolysis also releases arginase which breaks down L-arginine, the substrate for NO synthesis.97 NO depletion has been well-established as a potent regulator of smooth muscle tone, causing vascular constriction and contributing to a prothrombotic state.1,3,56 One study has found a 12-fold increase in the consumption of NO in PNH patients when compared to healthy volunteers in addition to an increase in pulmonary hypertension.98 As well as having a vasodilatory effect, NO also binds to platelets, causing signal transduction that downregulates the expression of the fibrinogen binding integrin glycoprotein IIb/IIIa, reduces levels of intracellular calcium and inhibits platelet activation.99,100 A reduction in circulating NO results in further dysregulation of platelets, and, combined with local vasoconstriction can contribute to intravascular thrombosis.56

Fibrin Clot Structure There is a growing body of evidence in the literature that individuals with an increased risk of thrombosis form fibrin clots with an altered three-dimensional structure.101 The altered clot structure is comprised of thinner but more tightly packed fibrin fibres, which, possibly combined with fewer binding sites for plasmin and tissue plasminogen activator (tPA), leads to impaired fibrinolysis. The dense clot structures are more resistant to fibrinolysis due to the increased number of fibres that need to be lysed and a reduced permeation of the lytic enzymes into the denser 14

clot structure.102 Moreover, clots with densely packed fibres are stiffer and more resistant to mechanical deformation.103 Due to these structural and functional changes, dense clots with smaller pores are associated with an increased risk of thrombosis. There have been no studies to date that have investigated clot structure in patients with PNH. Abnormal clot structure could be an additional mechanism by which the risk of thrombosis is increased in PNH, deserving of further study. Additionally, the link between PNH and clot structure may be of interest, since complement activation and factors have been associated with effects on fibrin clot structure and function. For example, alternative complement pathway activation has been associated with the production of denser, more tightly packed clots;104 furthermore, C3 has been shown to be incorporated into the fibrin clot, leading to thinner fibrin fibres and a stiffer clot with increased resistance to fibrinolysis.105 In addition, MASP-1 has been shown to influence clot formation and activate coagulation factor XIII, leading to an increased resistance of the clot to fibrinolysis.106 There may yet be other, unidentified mechanisms by which complement activation may regulate clot structure and function. High plasma levels of fibrinogen, the molecular precursor to fibrin, are known to affect clot structure.107 Seregina et al. measured fibrinogen levels in PNH patients pre- and post- treatment with eculizumab (n=3), and found no difference from normal controls.18 This suggests that clot structure is not being modulated by high levels of fibrinogen in PNH patients, but the small sample size means further study is necessary. Studies have shown that high concentrations of thrombin during fibrin clot formation results in prothrombotic fibrin structure and more stable clots.108 Surprisingly, one study found lower levels of thrombin generation in PNH patients. However, this may have been partly caused by the fact that thrombin generation in this study was measured in the absence of platelets and endothelial cells, which have been shown to be activated in PNH and enhance thrombin generation, as previously discussed.75 An alternative study found significantly elevated levels of thrombin generation on endothelial cells in patients with PNH.67 As examined previously, both activated platelets, PMPs and NETs increase thrombin generation and, therefore, may modulate clot structure through increased thrombin generation. Oxidised red blood cells integrated into fibrin clots have also been observed, enhancing clot stability.109 The oxidation of fibrin fibres has also been proposed to alter clot structure, however, this has been shown to both impair fibrin formation as well as produce thinner fibres and weaker clots.110 The role of microparticles in modulating clot structure is still a matter of dispute. Aleman et al. found that while platelet microparticles appeared to have no effect on clot structure, monocyte-derived microparticles supported faster fibrin deposition and a denser, more stable fibrin clot.111

Impaired Fibrinolysis Impaired fibrinolysis has been indicated in patients with PNH.3,4,66 Urokinase-type plasminogen activator receptor (uPAR) is a GPI-AP, and as such is absent from PNH monocytes and granulocytes.112 This results in increased plasma levels of free uPAR protein, which is thought to compete with the membrane bound uPA receptor, competitively haematologica | 2018; 103(1)


Multifaceted prothrombotic state in PNH

inhibiting cell-based plasmin generation and therefore contributing to a prothrombotic state in PNH.113,114 As discussed above, neutrophils, when activated, can express plasminogen activator inhibitor-1, impairing plasminogen and urokinase and therefore inhibiting fibrinolysis. Studies have also observed activated platelets releasing inhibitory proteases, plasminogen activator inhibitor-1 and α2-antiplasmin.115 More studies are needed to identify further causes of impaired fibrinolysis, whether it be resulting from GPI-AP loss or the hyperactivity of prothrombotic cells.

Animal models Mouse models have been used in the study of PNH, however it has proven difficult to replicate the thrombotic sequelae resulting from CD55 and CD59 deficiency. Complications arise due to the fact that there are two different CD55 and CD59 coding genes, a and b, respectively, and as a result, isolating their respective phenotypes following knockout has proven challenging.116,117 Mice also possess a unique transmembrane protein, complement-receptor 1-related gene protein (Crry), a functional homologue of human membrane cofactor protein which plays a critical role in protecting developing fetuses in mice from lethal complement attack. It has only been possible to produce Crry/C3 double knockout mice, which showed evidence of extravascular hemolysis, however, mice erythrocytes and platelets lacking Crry have been demonstrated to be more susceptible to hemolysis.118,119 Generating PIG-A mutations has proven difficult, as PIG-A deletion in embryonic stem cells is lethal.120 A PIG-A floxed mosaic mice model was generated with the coexistence of normal and mutated cells mimicking PNH patients, however, the PNH clone failed to clonally expand and produce PNH symptoms.121 Kellet et al. succeeded in creating a model with 100% of red blood cells being GPI-AP negative, however, no symptoms of PNH were observed. Limited knowledge of the clonal expansion mechanism has made it difficult to create animal models in which the thrombotic mechanisms in PNH can be analyzed. Future studies into clonal expan-

References 1. Brodsky RA. Paroxysmal nocturnal hemoglobinuria. Blood. 2015;124(18):2804–2812. 2. Socié G, Mary J, de Gramont A, Rio B, et al. Paroxysmal nocturnal haemoglobinuria: long-term follow-up and prognostic factors. Lancet. 1996;348(9027):573–577. 3. Hill A, Kelly RJ, Hillmen P. Thrombosis in paroxysmal nocturnal hemoglobinuria. Blood. 2013;121(25):4985–4996. 4. Grünewald M, Siegemund A, Grünewald A, et al. Plasmatic coagulation and fibrinolytic system alterations in PNH: relation to clone size. Blood Coagul Fibrinolysis. 2003;14(7): 685–695. 5. Devalet B, Mullier F, Chatelain B, Dogné JM, Chatelain C. Pathophysiology, diagnosis, and treatment of paroxysmal nocturnal hemoglobinuria: a review. Eur J Haematol. 2015;95(3):190–198. 6. Miyata T, Takeda J, Iida Y, et al. The cloning of PIG-A , a component in the early step of GPI-anchor biosynthesis. N Engl J Med. 1994;259(5099):1318–1320.

haematologica | 2018; 103(1)

sion mechanisms may lead to the development of more appropriate in vivo models, enabling the study of the mechanisms of thrombosis in PNH.

Summary and future perspectives Thrombosis risk is greatly increased in patients with PNH and is the greatest cause of morbidity and mortality in the disease. However, the mechanisms underpinning this are far from clear and are likely to be different from those of other thrombotic disorders. The prothrombotic state in PNH is extremely complex, with many different factors resulting from platelet activation, intravascular lysis and neutrophil/monocyte activation all thought to play a role. Further research is necessary in order to quantify how much each of these factors contribute to the prothrombotic state as well as to analyze their role in vivo. The newly hypothesized role of NETs especially warrants investigation, as this may explain the high and sustained incidence of atypical thrombosis in PNH. Impaired fibrinolysis and alterations to clot structure also appear to be hallmarks of thrombotic events, and it is important to determine the role of eculizumab in modifying these. Further identification of GPI-APs involved in clot structure and impaired fibrinolysis as well as clarification as to whether endothelial cells lack GPI-APs is necessary in order to understand the complex mechanism of thrombosis in PNH. With PNH research now in the post-eculizumab era, our understanding of the complement-mediated disease processes has improved dramatically, but those underpinning thrombotic complications are still insufficiently understood, despite the many hypothetical mechanisms which have been proposed. Future studies, including those involving animal models of PNH, may help to address this hiatus while simultaneously highlighting similar prothrombotic mechanisms in other, related hemolytic complement diseases. Identifying the factors that most significantly contribute to thrombus formation in PNH would allow for the application of more targeted therapies, potentially minimizing the disease burden and further improving patient outcomes.

7. Bessler M, Mason P, Hillmen P, Luzzatto L. Somatic mutations and cellular selection in paroxysmal nocturnal haemoglobinuria. Lancet. 1994;343:951–953. 8. Nakakuma H, Kawaguchi T. Pathogenesis of selective expansion of PNH clones. Int J Hematol. 2003;77(2):121–124. 9. Kawaguchi T, Nakakuma H. New insights into molecular pathogenesis of bone marrow failure in paroxysmal nocturnal hemoglobinuria. Int J Hematol. 2007;86(1):27–32. 10. Hanaoka N, Kawaguchi T, Horikawa K, Nagakura S, Mitsuya H, Nakakuma H. Immunoselection by natural killer cells of PIGA mutant cells missing stress-inducible ULBP. Blood. 2006;107(3):1184–1191. 11. Inoue N, Izui-Sarumaru T, Murakami Y, et al. Molecular basis of clonal expansion of hematopoiesis in 2 patients with paroxysmal nocturnal hemoglobinuria (PNH). Blood. 2006;108(13):4232–6 12. Schubert J, Roth A. Update on paroxysmal nocturnal haemoglobinuria: on the long way to understand the principles of the disease. Eur J Haematol. 2015;94(6):464–473. 13. Rollins SA, Sims PJ. The complement-

14.

15.

16.

17.

18.

inhibitory activity of CD59 resides in its capacity to block incorporation of C9 into membrane C5b-9. J Immunol. 1990;144(9): 3478–3483. Jankowska AM, Szpurka H, Calabro M, et al. Loss of expression of neutrophil proteinase3: A factor contributing to thrombotic risk in paroxysmal nocturnal hemoglobinuria. Haematologica. 2011;96(7):954–962. Hugel B, Socie G, Vu T, Toti F, et al. Elevated levels of circulating procoagulant microparticles in patients with paroxysmal nocturnal hemoglobinuria and aplastic anemia. Blood. 1999;93:3451–3456. Wiedmer T, Hall SE, Ortel TL, Kane WH, Rosse WF, Sims PJ. Complement-induced vesiculation and exposure of membrane prothrombinase sites in platelets of paroxysmal nocturnal hemoglobinuria. Blood. 1993;82 (4):1192–1196. Hillmen P, Lewis SM, Bessler M, Luzatto L, Dacie J V. Natural history of paroxysmal nocturnal hemoglobinuria. Blood. 1993;333 (19):1253–1258. Seregina EA, Tsvetaeva N V, Nikulina OF, et al. Eculizumab effect on the hemostatic state

15


B. Peacock-Young et al.

19.

20.

21.

22.

23. 24.

25. 26.

27.

28.

29.

30.

31.

32.

33.

34. 35.

36.

37.

16

in patients with paroxysmal nocturnal hemoglobinuria. Blood Cells Mol Dis. 2015;54(2):144–150. Parker C, Omine M, Richards S, et al. Diagnosis and management of paroxysmal nocturnal hemoglobinuria. Blood. 2005;106 (12):3699–3709. Hill A, Richards SJ, Hillmen P. Recent developments in the understanding and management of paroxysmal nocturnal haemoglobinuria. Br J Haematol. 2007;137(3):181–192. Schrezenmeier H, Muus P, Socié G, et al. Baseline characteristics and disease burden in patients in the international paroxysmal nocturnal hemoglobinuria registry. Haematologica. 2014;99(5):922–929. Hill A, Reid SA, Rother RP, et al. High definition contrast-enhanced MR imaging in paroxysmal nocturnal hemoglobinuria (PNH) suggests a high frequency of subclinical thrombosis. Blood. 2015;108(11):292. L’Acqua C, Hod E. New perspectives on the thrombotic complications of haemolysis. Br J Haematol. 2015;168(2):175–185. Moyo VM, Mukhina GL, Garrett ES, Brodsky RA. Natural history of paroxysmal nocturnal haemoglobinuria using modern diagnostic assays. Br J Haematol. 2004;126 (1):133–138. Hill A, Kelly RJ, Hillmen P. Thrombosis in paroxysmal nocturnal hemoglobinuria. Blood. 2013;121(25):4985–4996. Ziakas P, Poulou L, Rokas G, Bartzoudiss D, Voulgarelis M. Thrombosis in paroxysmal nocturnal hemoglobinuria: sites, risks, outcomes. An overview. J Thromb Haemost. 2007;5(3):642–645. Hall C, Richards S, Hillmen P. Primary prophylaxis with warfarin prevents thrombosis in paroxysmal nocturnal hemoglobinuria (PNH). Blood. 2003;102(10):3587–3591. Hoekstra J, Leebeek FWG, Plessier A, et al. Paroxysmal nocturnal hemoglobinuria in Budd-Chiari Syndrome: Findings from a cohort study. J Hepatol. 2009;51(4):696–706. Fowkes FJI, Price JF, Fowkes FGR. Incidence of diagnosed deep vein thrombosis in the general population: systematic review. Eur J Vasc Endovasc Surg. 2003;25(1):1–5. Gralnick HR McKeown LP, Merryman P, Wilson O, Chu I, Kimball J VM. Activated platelets in paroxysmal nocturnal hemoglobinuria. Br J Haematol. 1995;91:697–702. Devine D V, Rosse WF. Regulation of the activity of platelet-bound C3 convertase of the alternative pathway of complement by platelet factor H. Proc Natl Acad Sci USA . 1987;84(16):5873–5877. Wiedmer T, Esmon CT, Sims PJ. Complement proteins C5b-9 stimulate procoagulant activity through platelet prothombinase. Blood. 1986;123(23):3533–3535. Bevers EM, Comfurius P, Dekkers DWC, Zwaal RFA. Lipid translocation across the plasma membrane of mammalian cells. Biochim Biophys Acta - Mol Cell Biol Lipids. 1999;1439(3):317–330. Blair P, Flaumenhaft R. Platelet alpha–granules: basic biology and clinical correlates. Blood Rev. 2009;23(4):177–189. Sinauridze EI, Kireev DA, Popenko NY, et al. Plateletmicroparticle membranes have 50- to 100-fold higher specificprocoagulant activity than activated platelets. Thromb Haemost. 2007;97(6):425–434. Sims PJ, Wiedmer T, Esmon CT, Weiss HJ, Shattil SJ. Assembly of the platelet prothrombinase complex is linked to vesiculation of the platelet plasma membrane. 1989; Sims PJ, Faioni EM, Wiedmer T, Shattil SJ. Complement proteins C5b-9 cause release of membrane vesicles from the platelet surface

38.

39.

40.

41.

42.

43.

44.

45. 46.

47. 48.

49.

50.

51.

52.

53.

54.

55.

that are enriched in the membrane receptor for coagulation factor Va and express prothrombinase activity. J Biol Chem. 1989;264(29):17049-17057. Gilbert GE, Sims PJ, Wiedmer T, Furie B, Furie BC, Shattil SJ. Platelet-derived microparticles express high affinity receptors for factor VIII. J Biol Chem. 1991;266 (26):17261–17268. Nieuwland R, Berckmans RJ, RotteveelEijkman RC, et al. Cell-derived microparticles generated in patients during cardiopulmonary bypass are highly procoagulant. Circulation. 1997;96(10):3534-3541. Simak J, Holada K, Risitano AM, Zivny JH, Young NS, Vostal JG. Elevated circulating endothelial membrane microparticles in paroxysmal nocturnal haemoglobinuria. Br J Haematol. 2004;125(6):804–813. Losy J, Niezgoda A, Wender M. Increased serum levels of soluble PECAM-1 in multiple sclerosis patients with brain gadoliniumenhancing lesions. J Neuroimmunol. 1999;99(2):169–172. Schwarz M, Katagiri Y, Kotani M, et al. Reversibility versus persistence of GPIIb/IIIa blocker-induced conformational change of GPIIb/IIIa (alphaIIbbeta3, CD41/CD61). J Pharmacol Exp Ther. 2004;308(3):1002– 1011. Del Conde I, Crúz MA, Zhang H, López JA, Afshar-Kharghan V. Platelet activation leads to activation and propagation of the complement system. J Exp Med. 2005;201(6):871– 879. Lacroix R, Dignat-George F. Microparticles as a circulating source of procoagulant and fibrinolytic activities in the circulation. Thromb Res. 2012;129 Suppl 2:S27/29. Owens AP, Mackman N. Microparticles in hemostasis and thrombosis. Circ Res. 2011;108(10):1284–1297. Castaman G, Yu-Feng L, Battistin E, Rodeghiero F. Characterization of a novel bleeding disorder with isolated prolonged bleeding time and deficiency of platelet microvesicle generation. Br J Haematol. 1997;96(3):458–463. Weiss HJ, Lages B. Family studies in Scott syndrome. Blood 1997;90(1):475–476. Massberg S, Grahl L, von Bruehl ML, et al. Reciprocal coupling of coagulation and innate immunity via neutrophil serine proteases. Nat Med. 2010;16(8):887–896. Peerschke EI, Yin W, Grigg SE, Ghebrehiwet B. Blood platelets activate the classical pathway of human complement. J Thromb Haemost. 2006;4(9):2035–2042. Grünewald M, Grünewald A, Schmid A, et al. The platelet function defect of paroxysmal nocturnal haemoglobinuria. Platelets. 2004;15(3):145–154. Devalet B, Mullier F, Chatelain B, Dogné J-M, Chatelain C. The central role of extracellular vesicles in the mechanisms of thrombosis in paroxysmal nocturnal haemoglobinuria: a review. J Extracell vesicles. 2014;3:1–8. Kozuma Y, Sawahata Y, Takei Y, Chiba S, Ninomiya H. Procoagulant properties of microparticles released from red blood cells in paroxysmal nocturnal haemoglobinuria. Br J Haematol. 2011;152(5):631–639. Simak J, Gelderman MP. Cell membrane microparticles in blood and blood products: Potentially pathogenic agents and diagnostic markers. Transfus Med Rev. 2006;20(1):1–26. Kristiansen M, Graversen JH, Jacobsen C, et al. Identification of the haemoglobin scavenger receptor. Nature. 2001;409(6817):198– 201. Schaer C a, Vallelian F, Imhof A, Schoedon G, Schaer DJ. CD163-expressing monocytes

56.

57.

58.

59.

60.

61. 62.

63.

64.

65. 66.

67.

68.

69.

70.

71.

constitute an endotoxin-sensitive Hb clearance compartment within the vascular system. J Leukoc Biol. 2007;82(1):106–110. Rother RP, Bell L, Hillmen P. The clinical sequelae of intravascular hemolysis a novel mechanism of human disease. JAMA. 2005;293(13):1653-1662. Philippidis P, Mason JC, Evans BJ, et al. Hemoglobin scavenger receptor CD163 mediates interleukin-10 release and heme oxygenase-1 synthesis: antiinflammatory monocyte-macrophage responses in vitro, in resolving skin blisters in vivo, and after cardiopulmonary bypass surgery. Circ Res. 2004;94(1):119–126. Ballarín J, Arce Y, Torra Balcells R, et al. Acute renal failure associated to paroxysmal nocturnal haemoglobinuria leads to intratubular haemosiderin accumulation and CD163 expression. Nephrol Dial Transplant. 2011;26(10):3408–3411. Belcher JD, Chen C, Nguyen J, et al. Heme triggers TLR4 signaling leading to endothelial cell activation and vaso-occlusion in murine sickle cell disease. Blood.123(3):377–390. Olsen SB, Tang DB, Jackson MR, Gomez ER, Ayala B, Alving BM. Enhancement of platelet deposition by cross-linked hemoglobin in a rat carotid endarterectomy model. Circulation. 1996;93(2):327-332. Jacquemin M, Peerlinck K. Free hemoglobin: a boost to platelet thrombi. Blood. 2015;126(20):2261–2262. Studt JD, Kremer Hovinga JA, Antoine G, et al. Fatal congenital thrombotic thrombocytopenic purpura with apparent ADAMTS13 inhibitor: in vitro inhibition of ADAMTS13 activity by hemoglobin. Blood. 2005;105(2): 542–544. Zhou Z, Han H, Cruz MA, López JA, Dong JF, Guchhait P. Haemoglobin blocks von Willebrand factor proteolysis by ADAMTS13: A mechanism associated with sickle cell disease. Thromb Haemost. 2009;101(6): 1070–1077. Simionatto CS, Cabal R, Jones RL, Galbraith RA. Thrombophlebitis and disturbed hemostasis following administration of intravenous hematin in normal volunteers. Am J Med. 1988;85(4):538–540. Piccin A, Murphy WG, Smith OP. Circulating microparticles: pathophysiology and clinical implications. Blood Rev. 2007;21(3):157–171. van Bijnen STA, van Heerde WL, Muus P. Mechanisms and clinical implications of thrombosis in paroxysmal nocturnal hemoglobinuria. J Thromb Haemost. 2012;10 (1):1–10. Helley D, de Latour RP, Porcher R, et al. Evaluation of hemostasis and endothelial function in patients with paroxysmal nocturnal hemoglobinuria receiving eculizumab. Haematologica. 2010;95(4):574–581. Setty BNY, Betal SG, Zhang J, Stuart MJ. Heme induces endothelial tissue factor expression: potential role in hemostatic activation in patients with hemolytic anemia. J Thromb Haemost. 2008;6(12):2202–2209. Wagener FA, Feldman E, de Witte T, Abraham NG. Heme induces the expression of adhesion molecules ICAM-1, VCAM-1, and E selectin in vascular endothelial cells. Proc Soc Exp Biol Med. 1997;216(3):456-463. Qari MH, Dier U, Mousa SA. Biomarkers of inflammation, growth factor, and coagulation activation in patients with sickle cell disease. Clin Appl Thromb. 2012;18(2):195– 200. Maroney SA, Mast AE. Expression of tissue factor pathway inhibitor by endothelial cells and platelets. Transfus Apher Sci. 2008;38:9– 14.

haematologica | 2018; 103(1)


Multifaceted prothrombotic state in PNH 72. Broze GJ, Girard TJ, Girard TJ. Tissue factor pathway inhibitor: structure-function. Front Biosci (Landmark Ed). 2012;17:262-280. 73. Girard TJ, Tuley E, Broze GJ. TFPI Beta is the GPI-anchored TFPI isoform on human endothelial cells and placental microsomes. Blood. 2012;119(5):1256–1262. 74. Maroney SA, Cunningham AC, Ferrel J, et al. A GPI-anchored co-receptor for tissue factor pathway inhibitor controls its intracellular trafficking and cell surface expression. J Thromb Haemost. 2006;4(5):1114–1124. 75. Van Bijnen STA, Osterud B, Barteling W, et al. Alterations in markers of coagulation and fibrinolysis in patients with Paroxysmal Nocturnal Hemoglobinuria before and during treatment with eculizumab. Thromb Res. 2015;136(2):274–281. 76. Weitz JI. Heparan sulfate: Antithrombotic or not? J Clin Invest. 2003;111(7):952–954. 77. HajMohammadi S, Enjyoji K, Princivalle M, et al. Normal levels of anticoagulant heparan sulfate are not essential for normal hemostasis. J Clin Invest. 2003;111(7):989–999. 78. Ferreira V, Pangburn M, Cortés C. Complement control protein factor H: the good, the bad, and the inadequate. Mol Immunol. 2010;47(13):2187–2197. 79. Banerjee D, Mazumder S, Sinha AK. The role of inhibition of nitric oxide synthesis in the aggregation of platelets due to the stimulated production of thromboxaneA2. Blood Coagul Fibrinolysis. 2014;25(6):585-591. 80. Amer J, Zelig O, Fibach E. Oxidative status of red blood cells, neutrophils, and platelets in paroxysmal nocturnal hemoglobinuria. Exp Hematol. 2008;36(4):369–377. 81. Ataga KI, Cappellini MD, Rachmilewitz EA. Beta-thalassaemia and sickle cell anaemia as paradigms of hypercoagulability. Br J Haematol. 2007;139(1):3–13. 82. Iuliano L, Pedersen JZ, Praticò D, Rotilio G, Violi F. Role of hydroxyl radicals in the activation of human platelets. Eur J Biochem. 1994;221(2):695–704. 83. Fuchs TA, Brill A, Duerschmied D, et al. Extracellular DNA traps promote thrombosis. Proc Natl Acad Sci USA. 2010;107(36): 15880–15885. 84. Martinod K, Wagner DD. Thrombosis: tangled up in NETs. Blood. 2014;123(18):2768– 2776. 85. Nafa K, Bessler M, Mason P, et al. Factor V Leiden mutation investigated by Amplification Created Restriction Enzyme Site (ACRES) in PNH patients with and without thrombosis. Haematologica. 1996;81(6): 540–542. 86. Ritis K, Doumas M, Mastellos D, et al. A novel C5a receptor-tissue factor cross-talk in neutrophils links innate immunity to coagulation pathways. J Immunol. 2006;177(7): 4794–4802. 87. von Vietinghoff S, Tunnemann G, Eulenberg C, et al. NB1 mediates surface expression of the ANCA antigen proteinase 3 on human neutrophils. Blood. 2007;109(10):4487–4493. 88. Renesto P, Si-Tahar M, Moniatte M, et al. Specific inhibition of thrombin-induced cell activation by the neutrophil proteinases elastase, cathepsin G, and proteinase 3: evidence for distinct cleavage sites within the aminoterminal domain of the thrombin receptor. Blood. 1997;89(6):1944-1953. 89. Villegas-Mendez A, Montes R, Ambrose LR, Warrens AN, Laffan M, Lane DA. Proteolysis of the endothelial cell protein C receptor by neutrophil proteinase 3. J Thromb Haemost.

haematologica | 2018; 103(1)

2007;5(5):980–988. 90. Steppich BA, Seitz I, Busch G, Stein A, Ott I. Modulation of tissue factor and tissue factor pathway inhibitor-1 by neutrophil proteases. Thromb Haemost. 2008;100(6):1068–1075. 91. Huber-Lang M, Sarma JV, Zetoune FS, et al. Generation of C5a in the absence of C3: a new complement activation pathway. Nat Med. 2006;12(6):682–687. 92. Leung LL, Morser J. Plasmin as a complement C5 convertase. EBioMedicine. 2016;5:20–21. 93. Foley JH, Walton BL, Aleman MM, et al. Complement activation in arterial and venous thrombosis is mediated by plasmin. EBioMedicine. 2016;5:175–182. 94. Montrucchio G, Bergerone S, Bussolino F, et al. Streptokinase induces intravascular release of platelet-activating factor in patients with acute myocardial infarction and stimulates its synthesis by cultured human endothelial cells. Circulation. 1993;88:1476– 1483. 95. Lupia E, Del Sorbo L, Bergerone S, Emanuelli G, Camussi G, Montrucchio G. The membrane attack complex of complement contributes to plasmin-induced synthesis of platelet-activating factor by endothelial cells and neutrophils. Immunology. 2003;109(4): 557–563. 96. Liu X, Miller MJS, Joshi MS, SadowskaKrowicka H, Clark DA, Lancaster JR. Diffusion-limited reaction of nitric oxide with erythrocytes. J Biol Chem. 1998;273(30):18709–18713. 97. Schnog JJB, Jager EH, van der Dijs FPL, et al. Evidence for a metabolic shift of arginine metabolism in sickle cell disease. Ann Hematol. 2004;83(6):371–375. 98. Hill A, Sapsford RJ, Scally A, et al. Under-recognized complications in patients with paroxysmal nocturnal haemoglobinuria: Raised pulmonary pressure and reduced right ventricular function. Br J Haematol. 2012;158(3):409–414. 99. Freedman JE, Loscalzo J. Nitric oxide and its relationship to thrombotic disorders. J Thromb Haemost. 2003;1(6):1183–1188. 100. Morel O, Jesel L, Freyssinet JM, Toti F. Cellular mechanisms underlying the formation of circulating microparticles. Arterioscler Thromb Vasc Biol. 2011;31(1):15–26. 101. Bridge KI, Philippou H, Ariëns R. Clot properties and cardiovascular disease. Thromb Haemost. 2014;112(5):901–908. 102. Collet JP, Park D, Lesty C, et al. Influence of fibrin network conformation and fibrin fiber diameter on fibrinolysis speed: dynamic and structural approaches by confocal microscopy. Arterioscler Thromb Vasc Biol. 2000;20(5):1354–1361. 103. Ajjan R, Lim BC, Standeven KF, et al. Common variation in the C-terminal region of the fibrinogen β-chain: effects on fibrin structure, fibrinolysis and clot rigidity. Blood. 2008;111(2):643–650. 104. Shats-Tseytlina EA, Nair CH, Dhall DP. Complement activation: a new participant in the modulation of fibrin gel characteristics and the progression of atherosclerosis? Blood Coagul Fibrinolysis. 1994;5(4):529–535. 105. Howes JM, Richardson VR, Smith KA, et al. Complement C3 is a novel plasma clot component with anti-fibrinolytic properties. Diabetes Vasc Dis Res. 2012;9(3):216–25. 106. Hess K, Ajjan R, Phoenix F, Dobó J, Gál P, Schroeder V. Effects of MASP-1 of the complement system on activation of coagulation

factors and plasma clot formation. PLoS One. 2012;7(4):e35690. 107. Undas A. Fibrin clot properties and their modulation in thrombotic disorders. Thromb Haemost. 2014;112(1):32–42. 108. Wolberg AS, Monroe DM, Roberts HR, Hoffman M. Elevated prothrombin results in clots with an altered fiber structure: A possible mechanism of the increased thrombotic risk. Blood. 2003;101(8):3008–3013. 109. Lipinski B, Pretorius E, Oberholzer HM, van der Spuy WJ. Interaction of fibrin with red blood cells: the role of iron. Ultrastruct Pathol. 2012;36(2):79–84. 110. Martinez M, Weisel JW, Ischiropoulos H. Functional impact of oxidative post-translational modifications on fibrinogen and fibrin clots. Biophys Chem. 2005;257(5):2432– 2437. 111. Aleman MM, Gardiner C, Harrison P, Wolberg AS. Differential contributions of monocyte- and platelet-derived microparticles towards thrombin generation and fibrin formation and stability. J Thromb Haemost. 2011;9(11):2251–2261. 112. Ploug M, Plesner T, Ebbe R, et al. The receptor for urokinase-type plasminogen activator is deficient on peripheral blood leukocytes in patients with paroxysmal nocturnal hemoglobinuria. Blood. 1992;79(1):1447–1455. 113. Ninomiya H, Hasegawa Y, Nagasawa T, Abe T. Excess soluble urokinase-type plasminogen activator receptor in the plasma of patients with paroxysmal nocturnal hemoglobinuria inhibits cell-associated fibrinolytic activity. Int J Hematol . 1997;65(3):285–291. 114. Rønne E, Pappot H, Grøndahl-Hansen J, et al. The receptor for urokinase plasminogen activator is present in plasma from healthy donors and elevated in patients with paroxysmal nocturnal haemoglobinuria. Br J Haematol. 1995;89(3):576–581. 115. Rendu F, Brohard-Bohn B. The platelet release reaction: granules’ constituents, secretion and functions. Platelets. 2001;12(5): 261–273. 116. Baalasubramanian S, Harris CL, Donev RM, et al. CD59a Is the primary regulator of membrane attack complex assembly in the mouse. J Immunol. 2004;173(6):3684–3692. 117. Song WC, Deng C, Raszmann K, et al. Mouse decay-accelerating factor: selective and tissue-specific induction by estrogen of the gene encoding the glycosylphosphatidylinositol-anchored form. J Immunol. 1996;157(9):4166–4172. 118. Xu C, Mao D, Holers VM, Palanca B, Cheng AM, Molina H. A critical role for murine complement regulator crry in fetomaternal tolerance. Science. 2000;287(5452):498–501. 119. Kim DD, Miwa T, Song WC. Retrovirusmediated over-expression of decay-accelerating factor rescues Crry-deficient erythrocytes from acute alternative pathway complement attack. J Immunol. 2006;177(8): 5558–5566. 120. Rosti V, Tremml G, Soares V, Pandolfi PP, Luzzatto L, Bessler M. Murine embryonic stem cells without pig-a gene activity are competent for hematopoiesis with the PNH phenotype but not for clonal expansion. J Clin Invest. 1997;100(5):1028–1036. 121. Tremml G, Dominguez C, Rosti V, et al. Increased sensitivity to complement and a decreased red blood cell life span in mice mosaic for a nonfunctional Piga gene. Blood. 1999;94(9):2945–2954.

17


ARTICLE

Hematopoiesis

Ferrata Storti Foundation

Protein arginine methyltransferase 6 controls erythroid gene expression and differentiation of human CD34+ progenitor cells Stefanie C. Herkt,1 Olga N. Kuvardina,1 Julia Herglotz,1 Lucas Schneider,1 Annekarin Meyer,1 Claudia Pommerenke,2,3 Gabriela Salinas-Riester,2 Erhard Seifried,1 Halvard Bonig1 and Jörn Lausen1

Institute for Transfusion Medicine and Immunohematology, Goethe-University and German Red Cross Blood Service, Frankfurt am Main; 2Transcriptome Analysis Laboratory, MedicalUniversity Göttingen and 3Present address: Leibnitz-Institute DSMZ; Braunschweig, Germany

1

Haematologica 2018 Volume 103(1):18-29

ABSTRACT

H

Correspondence: j.lausen@blutspende.de

Received: June 13, 2017. Accepted: October 6, 2017. Pre-published: October 12, 2017.

ematopoietic differentiation is driven by transcription factors, which orchestrate a finely tuned transcriptional network. At bipotential branching points lineage decisions are made, where key transcription factors initiate cell type-specific gene expression programs. These programs are stabilized by the epigenetic activity of recruited chromatin-modifying cofactors. An example is the association of the transcription factor RUNX1 with protein arginine methyltransferase 6 (PRMT6) at the megakaryocytic/erythroid bifurcation. However, little is known about the specific influence of PRMT6 on this important branching point. Here, we show that PRMT6 inhibits erythroid gene expression during megakaryopoiesis of primary human CD34+ progenitor cells. PRMT6 is recruited to erythroid genes, such as glycophorin A. Consequently, a repressive histone modification pattern with high H3R2me2a and low H3K4me3 is established. Importantly, inhibition of PRMT6 by shRNA or small molecule inhibitors leads to upregulation of erythroid genes and promotes erythropoiesis. Our data reveal that PRMT6 plays a role in the control of erythroid/megakaryocytic differentiation and open up the possibility that manipulation of PRMT6 activity could facilitate enhanced erythropoiesis for therapeutic use.

Introduction doi:10.3324/haematol.2017.174516 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/18 ©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.

18

Hematopoietic lineage decisions are driven by transcription factors, which define cell type-specific gene expression and thus instruct lineage specification during terminal differentiation. A subset of transcription factors is important for hematopoietic stem cell emergence and also for later lineage-specific gene expression.1-5 For multilineage differentiation processes, such as hematopoiesis, epigenetic stabilization of gene expression programs is of central importance. In this process of epigenetic gene regulation, transcription factors recruit cofactors with enzymatic activity to target genes. These cofactors are able to change chromatin organization by modification.6,7 The most prominent epigenetic modifications are methylation of DNA on cytidines and a large number of different posttranslational modifications of histones. These interdependent modifications mostly take place at the histone tails and comprise a pattern which can encode distinct functions.6,8,9 The important function of transcriptional regulators in hematopoiesis is highlighted by the observation that alterations of transcriptional regulators can convert one cell type into another.10-15 At the megakaryocytic/erythroid lineage bifurcation transcription factors such as RUNX1, FLI1, KLF1, GATA1 and TAL1 play a decisive role in the establishment of the megakaryocytic or erythroid gene expression program, respectively.16-21 The transcription factor RUNX1 (also known as AML1: acute myeloid leukemia 1) plays a major role in hematopoietic stem cell emergence.22,23 Furthermore, RUNX1 is important for the establishment of the megakaryocytic gene expression program and the haematologica | 2018; 103(1)


Epigenetic influence of PRMT6 on hematopoiesis

repression of erythroid genes.24-28 Depending on the associated cofactors RUNX1 can act as an activator or repressor of gene expression. Importantly, RUNX1 cooperates with central epigenetic complexes such as the trithorax-(MLL)-complex and the polycomb-(PRC)-complex, which trigger the activating trimethylation of lysine 4 on histone 3 (H3K4me3) and the repressive trimethylation of lysine 27 on histone 3 (H3K27me3), respectively.29-32 RUNX1 also interacts with protein arginine methyltransferase 6 (PRMT6).33,34 PRMT6 is a member of the PRMT-family, which consists of enzymes that methylate arginine residues on proteins, including histones.35 PRMT6-mediated asymmetric dimethylation of histone 3 at the arginine at position 2 (H3R2me2a) counteracts the activating H3K4me3, thus PRMT6 acts predominantly as a repressor of gene expression.36-38 It has been demonstrated that PRMT6 has an influence on embryonic stem cell identity.39 In megakaryocytic/erythroid progenitors, PRMT6 is recruited by RUNX1 to target genes and acts as a repressor by setting the H3R2me2a mark. This way RUNX1/PRMT6 contribute to the establishment of bivalent chromatin marking at megakaryocytic differentiation genes, such as CD41, in progenitors.33 Despite the evident role of PRMT6 at megakaryocytic/erythroid branching, its cell type-specific function has not been studied in detail. In light of the notion that small molecule inhibition of epigenetic enzymes could influence in vitro differentiation it is instructive to study the biological processes mediated by PRMT6. We found that decreased PRMT6 activity in primary human CD34+ progenitor cells leads to increased in vitro erythroid differentiation, whereas overexpression of PRMT6 decreases erythroid differentiation. During megakaryocytic differentiation of progenitor cells PRMT6 contributes to the suppression of erythroid genes by establishment of a repressive chromatin environment. Interestingly, PRMT6 inhibition by a small molecule also enhances erythropoiesis. This opens up the possibility of using PRMT6 inhibitors for more effective in vitro differentiation of erythrocytes.

Methods Cell culture K562 (ATCC CCL-243) and HEK293T/17 (ATCC CRL-11268) cells were cultured in RPMI-1640 and DMEM medium, respectively. Growth media were supplemented with 10% fetal calf serum, 2 mM glutamine and 1% penicillin/streptomycin. For megakaryocytic differentiation K562 cells were treated with 30 nM 12-o-tetradecanylphorbol-13-acetate (TPA; Sigma, Darmstadt, Germany). The cells were harvested after 3 days and analyzed using flow cytometry. Samples of granulocyte colony-stimulating factor mobilized peripheral or bone marrow human primary CD34+ cells from healthy donors were used, with approval of the ethics committee (permit #329-10). CD34+ cells were immunomagnetically enriched according to the manufacturer’s instructions (Miltenyi, Bergisch Gladbach, Germany) and expanded under serum-free conditions using Stem Span (SFEMI, Stemcell Technologies, Vancouver, Canada) as described previously.27,28,33 The cells were then subjected to erythroid or megakaryocytic differentiation.33,40 After 6 days the differentiation status was determined by fluorescence activated cell sorting (FACS) and cells were used for mRNA analysis or chromatin immunoprecipitation (ChIP). For overexpression and knockdown experiments expanded cells were transduced with haematologica | 2018; 103(1)

lentiviral vectors. Transduced GFP+ cells were sorted and subsequently subjected to colony-forming unit (CFU) assay in methylcellulose, according to the manufacturer’s instructions (Miltenyi, Bergisch Gladbach, Germany). Colonies were counted 12 days after seeding. For erythroid-megakaryocytic differentiation in liquid culture, isolated bone marrow CD34+ cells were maintained in serum-free expansion medium SFEMII (Stemcell Technologies, Vancouver, Canada) supplemented with 100 ng/mL stem cell factor, 10 ng/mL interleukin-3, 10 ng/mL interleukin-6, 0.5 U/mL erythropoietin and 50 ng/mL thrombopoietin. Differentiation was verified by FACS and mRNA analysis. The PRMT6 inhibitor MS023 was obtained from Biomol (Hamburg, Germany).

Chromatin immunoprecipitation Cell lysates and the ChIP assay were performed according to the X-ChIP protocol from Abcam, with modifications. For immunoprecipitation 3-10 μg of specific antibody were used. ChIP DNA was purified using DNA purification columns ChIP DNA Clean and Concentrator (Zymo Research, Irvine, USA) and analyzed by quantitative polymerase chain reaction (PCR). DNA recovery was calculated as percentage of the input. Error bars represent the standard deviation from at least four determinations. Histone modification ChIP values were corrected for nucleosome density using ChIP values for histone 3 (H3). ChIP-ReChIP was performed as described previously.40 The sequences of primer pairs used for the ChIP-PCR analysis are available upon request. Antibodies used in this study are listed in the Online Supplementary Material.

Gene expression analysis Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany). cDNA was synthesized using Omniscript reverse transcriptase (Qiagen). Quantitative reverse transcriptase PCR was performed using SYBR Green PCR Mastermix (Eurogentec, Luettich, Belgium). Relative amounts of mRNA were normalized against glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression values. Primer sequences are available upon request. Knockdown constructs and vector information are given in the Online Supplementary Material. The gene expression array of shPRMT6 K562 cells was analyzed via the limma package of Bioconductor. Differentially expressed genes were filtered to a minimum of two-fold change and Benjamini-Hochberg corrected P-value <0.05 as previously described.40 Data were deposited in the GEO-Expression database, GSE92251. Further functional association of candidate genes was performed with the webtool DAVID using standard settings.41,42 Western blot analysis was performed as described in the Online Supplementary Material.

Statistics Array data were processed as stated above. ChIP and quantitative reverse transcriptase PCR data were analyzed using PRISM software. The error bars represent the standard deviation from the mean. P values were calculated using the Student t-test from at least four determinations. P values <0.05 were considered statistically significant (*P<0.05; **P<0.01; ***P<0.001).

Results PRMT6 inhibits erythropoiesis PRMT6 is associated with RUNX1 on megakaryocytic target genes in progenitor cells and present on erythroid genes upon megakaryocytic differentiation.28,33 This shows 19


S.C. Herkt et al.

that PRMT6 plays a role in gene expression control at megakaryocytic/erythroid branching and might, therefore, influence differentiation. To explore this possibility, we performed a CFU assay. Human primary CD34+ cells were transduced with shRNA knockdown vectors and PRMT6 overexpression vectors, respectively (Online Supplementary Figure S1). Transduced cells were sorted and subjected to a CFU assay under conditions which allow myeloid differentiation, including erythropoiesis (Figure 1A). The knockdown of PRMT6 resulted in a decrease of the relative number of monocytic colonies, whereas the overexpression of PRMT6 increased the number of monocytic colonies as well as granulocytic colonies (Figure 1B,C). Interestingly, the knockdown of PRMT6 doubled the number of erythroid colonies (Figure 1D) and resulted in a modest reduction of total colony number (Figure 1E). In contrast, PRMT6 overexpression reduced erythroid colony formation (Figure 1F) and had no influence on colony number (Figure 1G). Under the conditions employed for the CFU assay, megakaryocytic differentiation could not be monitored. However, in a previous study we found that PRMT6 inhibits megakaryocytic genes in progenitor cells but leaves these promoters upon megakaryocytic differentiation.33 Thus, we wondered whether PRMT6 would alter erythroid/megakaryocytic differentiation under condi-

tions that allow for both erythroid and megakaryocytic differentiation. To examine this, we transduced human CD34+ cells with a PRMT6 overexpression vector. Two days after transduction the cells were transferred to growth medium, which contained thrombopoietin and erythropoietin (Figure 2A).43 After 10 days of culture we measured the erythroid differentiation markers GYPA and CD71 by FACS. Furthermore, we determined the megakaryocytic differentiation markers CD41 and CD61. We found that the percentage of GYPA+ cells was about 40% in the control (Figure 2B, left) and 20% of all cells had high GYPA expression in the control (Figure 2B, right). Upon PRMT6 overexpression the number of GYPA+ cells was reduced. Moreover, the GYPAhi population was virtually absent upon PRMT6 overexpression (Figure 2B). Similarly, the number of CD71hi cells (another erythroid marker) was also reduced upon PRMT6 expression (Figure 2C). In contrast, the expression of the megakaryocytic markers, CD41 and CD61, was increased upon PRMT6 expression (Figure 2D,E). Corresponding flow cytometry data are shown in Online Supplementary Figure S2. Taken together, our data indicate that more mature erythroid cells with high GYPA and high CD71 expression are almost absent and the number of cells with megakaryocytic markers is increased upon PRMT6 overexpression. These data indicate that the cells

A

B

D

20

C

E

F

G

Figure 1. PRMT6 inhibits erythroid differentiation. (A) Schematic workflow of the colony-forming unit (CFU) assay. Human CD34+ cells were transduced with PRMT6 knockdown (shPRMT6), PRMT6 expression, or control vector. Transduced GFP+ cells were sorted by FACS and subjected to the CFU assay. Colonies were counted on day 10-14 after seeding. (B, C) CFU assay of CD34+ cells upon PRMT6 knockdown and overexpression. Human CD34+ cells were transduced with PRMT6 knockdown vector (shPRMT6), PRMT6 expression vector, or control vector. Transduced GFP+ cells were sorted by FACS and subjected to a CFU assay. Colonies were counted 10-14 days after seeding. (B) CFU assay upon knockdown of PRMT6 using two different shRNA. Unspecific shRNA was used as a control. (C) CFU assay upon PRMT6 overexpression. Empty vector serves as the control. CFUG colony-forming unit-granulocyte, CFUM colony-forming unit-monocyte, CFUGM colony-forming unit-granulocyte, monocyte, BFU-E burst forming unit-erythroid, CFU-E colony forming unit-erythroid. (D) Knockdown of PRMT6 using two different shRNA (shP6) enhances erythroid differentiation of CD34+ cells in the CFU assay. (E) The total number of colonies in the CFU assay after PRMT6 knockdown is shown. (F) The relative frequency of erythroid colonies (in percent) was decreased upon PRMT6 overexpression compared to the control. (G) The total number of colonies in the CFU assay after PRMT6 overexpression is shown. Error bars show the standard deviation calculated from at least four determinations. The P-values were calculated using the Student t-test. *P<0.05; **P< 0.01; ***P<0.001.

haematologica | 2018; 103(1)


Epigenetic influence of PRMT6 on hematopoiesis

have shifted from an erythroid to a megakaryocytic phenotype upon PRMT6 overexpression.

PRMT6 represses erythroid genes To investigate how PRMT6 influences erythroid differentiation, we analyzed gene expression downstream of PRMT6 upon knockdown of PRMT6. For this we used K562 erythroleukemia cells, which express GYPA and low levels of the erythroid master regulator KLF1. Gene expression analysis was studied by array analysis 5 days after transduction of shPRMT6 expression vectors (Figure 3A,B, Online Supplementary Figure S3). PRMT6 knockdown resulted in changed expression of more than 1,000 genes (Online Supplementary Figure S3). About half of the genes were upregulated and the other half downregulated upon PRMT6 knockdown (Online Supplementary Figure S3). Gene ontology analysis (GO-terms) using DAVID41,42 revealed that PRMT6 influences genes with distinct functions. The most significant GO-categories were “response to wounding” and “negative regulation of cell growth” (Online Supplementary Figure S3). The GO-category hematopoiesis

was also enriched (Figure 3C). Ten of the 20 genes involved in hematopoiesis have a known function in erythropoiesis. Erythroid-specific genes were mostly upregulated. This includes ALAS2 (delta-aminolevulinate synthase 2), which plays a role in heme biosynthesis, AHSP (alpha hemoglobin stabilizing protein) and the erythroid differentiation marker GYPA (glycophorin A) (Figure 3C). CEBPα and c-Kit were downregulated upon PRMT6 knockdown (Figure 3C). To further examine the influence of PRMT6 on erythroid gene expression we measured the expression of erythroid genes by quantitative reverse transcriptase PCR 7 days after transduction of K562 cells (Online Supplementary Figure S3). PRMT6 knockdown resulted in a marked increase of the erythroid markers GYPA, ALAS2 and AHSP (Figure 3D-F). Similar to the array data c-Kit expression was decreased upon PRMT6 knockdown (Figure 3G). The erythroid genes KLF1 and β-globin were significantly increased at this time point after knockdown (Figure 3H,I). Like KLF1, the erythroid transcription factors TAL1 and GATA1 were also influenced by the level of PRMT6 in K562 cells (Online Supplementary Figure S4). The

A

B

D

C

E

haematologica | 2018; 103(1)

Figure 2. PRMT6 overexpression alters erythroid/megakaryocytic differentiation (A) Schematic workflow of the differentiation analysis. + Human CD34 cells were isolated and cultured in SFEMI with the indicated supplements (expansion medium). At day 3 the cells were transduced with a PRMT6 expression vector or empty control vector, 2 days later the cells were transferred to erythroid/megakaryocytic differentiation medium (ery/mega medium). Ery/mega medium was SFEMII with the indicated supplements and 2 mM glutamine was also added. The transduced GFP+ cells were analyzed after 10 days in ery/mega medium by FACS. (B) The number of GYPA+ cells (left) and the number of GYPAhi cells was determined (right) with and without PRMT6 overexpression using a GYPA antibody (CD235a). (C) The number of CD71+ (left) and CD71hi (right) cells was determined. (D) The number of CD41+ cells is shown. (E) The number of CD61+ cells is shown. (B-E) Cell numbers are given in percent related to the total number of transduced GFP+ cells. Error bars give the standard deviation from four independent determinations. The P-values were calculated using the Student t-test. **P<0.01; ***P< 0001.

21


S.C. Herkt et al.

effect of PRMT6 knockdown on GYPA expression was also detectable at the level of the cell surface (Figure 3J, Online Supplementary Figure S5). Furthermore, expression of the early erythroid surface marker CD71 was increased (Figure 3K, Online Supplementary Figure S5). The K562 cells displayed a reddish color upon PRMT6 knockdown, indicating increased heme production (Online Supplementary Figure S5). Moreover, the knockdown of PRMT6 in human CD34+ cells influenced the expression of GYPA, ALAS2, AHSP, c-Kit, KLF1 and β-globin in the same direction as in K562 cells (Figure 3L). These data indicate that PRMT6 has a repressive influence on the expression of some erythroid genes, which is released upon PRMT6 knockdown.

A

B

D

E

H

I

L

22

GYPA is a direct target of PRMT6 For further analysis of PRMT6 function we focused on this proteinâ&#x20AC;&#x2122;s influence on the glycophorin A gene (GYPA). GYPA is a membrane protein and the main marker of erythroid differentiation. Expression of GYPA is directly controlled by the transcription factors TAL1 and GATA1.44 Our data show that PRMT6 inhibits erythroid differentiation and the expression of the erythroid differentiation gene GYPA. Examination of published ChIP-Seq data revealed that the promoter of GYPA also harbors functional RUNX1 binding sites in addition to TAL1 and GATA1 sites (Online Supplementary Figure S6). These transcription factors are known to be associated with PRMT6.28 By

C

F

G

J

K

Figure 3. Gene expression analysis upon PRMT6 knockdown. (A, B) K562 cells were transduced with two different shRNA constructs against PRMT6 (shP6) and the knockdown was evaluated by quatitative reverse transcriptase q-RTPCR and western blot. (C) Gene expression array analysis was performed with shPRMT6 K562 cells 5 days after transduction. Hematopoiesis-associated genes are shown. The numbers give the changes upon PRMT6 knockdown as logfold2. PRMT6 expression was reduced -2.82 logfold2 compared to the control, expressing a non-targeting shRNA. Genes marked in red have a described role in erythropoiesis. (D-I) A subset of genes from the array analysis was reanalyzed by quantitative real-time PCR 7 days after PRMT6 transduction. Error bars represent the standard deviation from at least four determinations and two independent knockdowns. (J,K) PRMT6 knockdown in K562 cells led to an increase of the GYPA (CD235a) and CD71 cell surface expression measured by FACS. The median fluorescence intensity (MFI) of GYPAAPC and CD71-APC staining in shcontrol (shctrl.) and shPRMT6 (shP6) cells is shown. (L) The expression of the genes was measured by qRT-PCR in CD34+ cells upon knockdown of PRMT6. Knockdown cells were sorted and maintained in ery/mega medium for 5 days. Gene expression was determined by quantitative reverse transcriptase PCR. The knockdown values (shPRMT6#1/2) represent the combined data from two different knockdown constructs. Error bars display the standard deviation calculated from at least four determinations. The P-values were calculated using the Student t-test. *P<0.05; **P<0.01; ***P<0.001.

haematologica | 2018; 103(1)


Epigenetic influence of PRMT6 on hematopoiesis

A

B

C

E

D

F

H

I

K

L

G

J

M

N

Figure 4. GYPA is a direct target of hematopoietic transcription factors and PRMT6. (A) Scheme of the GYPA promoter showing the position of the ChIP-primers. (B) ChIP with K562 cells indicates binding of PRMT6 to the promoter region of GYPA (P2) but not to an upstream region (P1) or an unrelated control region (chr.18). (C) RUNX1 binds to the promotor region (P2) of GYPA but not to an upstream region (P1) or an unrelated control region on chromosome 18 (chr.18). (D) Quantitative ChIPReChIP of RUNX1 and PRMT6 with the given antibody combinations shows co-occupancy of RUNX1 with PRMT6 at the GYPA promoter (left) but not at a control region (chr.18) in K562 cells. (E, F) ChIP assay after RUNX1 knockdown shows reduced RUNX1 and PRMT6 binding to the GYPA promoter. (G) H3R2me2a modification at the GYPA promoter is decreased upon RUNX1 knockdown. (H, I) RUNX1 overexpression decreased GYPA mRNA expression in K562 and CD34+ cells measured by quantitative reverse transcriptase PCR. (J) Knockdown of PRMT6 in CD34+ cells cultured in ery/mega medium results in increased GYPA mRNA expression with time. The GYPA expression level of the corresponding time point was set as one. (K-N) Changes at the GYPA promoter upon erythroid differentiation of CD34+ cells. (K) RUNX1 binding to the GYPA promoter remains unchanged upon erythroid differentiation. (L) TAL1 binding to the GYPA promoter is increased upon erythroid differentiation. (M) PRMT6 binding is reduced upon erythroid differentiation. (N) Upon erythroid differentiation the repressive H3R2me2a modification at the GYPA promoter is decreased. Note that in K-N the values for IgG are small, so that the bar for the IgG control is not visible. Quantitative ChIP-PCR values are shown as percentage input. Values gathered for histone modification H3R2me2a were normalized with a ChIP against unmodified histone H3. Error bars show the standard deviation from at least four independent evaluations. The P-values were calculated using the Student t-test. *P<0.05; **P<0.01; ***P<0.001.

haematologica | 2018; 103(1)

23


S.C. Herkt et al.

ChIP we detected PRMT6 and RUNX1 at the proximal promoter region of GYPA in K562 cells (Figure 4A-C, Online Supplementary Figure S6). As RUNX1 is able to recruit PRMT6 to target genes,33 we examined whether RUNX1 and PRMT6 co-occupy the GYPA promoter. Using ChIP-ReChIP we did in fact detect RUNX1 and PRMT6 together at this promoter, as indicated by the enrichment of GYPA promoter DNA in the RUNX1/PRMT6 ChIP-ReChIP, but not at a control locus (Figure 4D). The notion that RUNX1 is important for PRMT6 recruitment was further supported by a ChIPassay after RUNX1 knockdown. RUNX1 knockdown led to decreased RUNX1 binding at the GYPA promoter (Figure 4E). PRMT6 occupancy of the GYPA promoter was decreased (Figure 4F) and H3R2me2a was diminished, an effect mediated by PRMT6 (Figure 4G). When we overexpressed RUNX1, the GYPA levels were reduced in K562 and CD34+ cells (Figure 4H,I). Furthermore, RUNX1 repressed the GYPA promoter in a reporter gene assay (Online Supplementary Figure S6). Interestingly, the knockdown of PRMT6 in CD34+ cells under differentiation conditions, which on its own induced some GYPA expression, led to increased GYPA expression (Figure 4J). When we induced CD34+ cells towards the eythroid lineage, we found that RUNX1 binding remained unchanged on the promoter (Figure 4K), TAL1 binding increased (Figure 4L) and PRMT6 binding decreased (Figure 4M). Concomitantly, the PRMT6-mediated H3R2me2a histone mark was decreased (Figure 4N). Taken together, our data show that RUNX1 contributes to the binding of PRMT6 to the GYPA promoter and support the notion that PRMT6 is a repressor of GYPA expression.

Differentiation-associated epigenetic changes The erythroid gene GYPA is upregulated upon erythroid differentiation and downregulated during megakaryocytic differentiation of human CD34+ cells (Online Supplementary Figure S7). Furthermore, PRMT6 expression is increased during erythroid and megakaryocytic differentiation, whereas RUNX1 is only increased during megakaryocytic differentiation (Online Supplementary Figure S7). Our data show that PRMT6 is associated with repression of GYPA. Thus, we wondered whether PRMT6 is connected to the downregulation of GYPA expression during megakaryocytic differentiation of human primary progenitor cells (Online Supplementary Figures S7 and S8). We found that RUNX1 binding to the GYPA promoter increased during megakaryocytic differentiation of CD34+ cells (Figure 5A), whereas TAL1 binding remained unchanged and GATA1 binding decreased (Figure 5B,C). In line with a repressor function of PRMT6 we found that PRMT6 binding to the GYPA promoter increases upon megakaryocytic differentiation of hCD34+ cells (Figure 5D). It was suggested that PRMT6-mediated H3R2me2a negatively influences WDR5 binding and that the protein arginine deaminase PADI4 can counteract PRMT6 activity.34,40 Accordingly, WDR5 and PADI4 binding decreases at the GYPA promoter (Figure 5E,F). As a consequence, the activating H3K4me3 modification decreases and the repressive H3R2me2a and H3K27me3 methylation marks increase (Figure 5G-I). Concomitant to the increase of the repressive histone modification H3K27me3, the binding of EZH2, which mediates this modification, increases (Figure 5J). In line with the notion that a repressive chromatin environment is established, 24

binding of the repressive histone deacetylase 1 (HDAC1) is increased upon megakaryocytic differentiation (Figure 5K) and occupancy of RNA-polymerase II is decreased (Figure 5L). Similar changes can also be observed during megakaryocytic differentiation of K562 cells (Online Supplementary Figure S9). In summary, our data demonstrate that PRMT6 and associated repressors contribute to the repression of GYPA expression during megakaryocytic differentiation.

Pharmacological inhibition of PRMT6 increases erythroid gene expression We have shown that binding of the RUNX1-associated repressor PRMT6 is upregulated during megakaryopoiesis and decreased during erythropoiesis at the GYPA locus. Furthermore, knockdown of PRMT6 increases erythropoiesis. Thus, inhibition of PRMT6 enzymatic function might lead to a shift in differentiation. Recently, small molecule inhibitors of PRMT6, which decrease the repressive H3R2me2a methylation in cells, were introduced.45,46 Accordingly, treatment of K562 cells with the PRMT6 inhibitor MS023 for 3 days increased GYPA expression at the mRNA level already at a concentration of 0.05 μM and reached its plateau at 1 μM (Figure 6A). Induction of the erythroid surface marker GYPA was also detected by flow cytometry (Figure 6B,C). Furthermore, other erythroid genes such as AHSP, ALAS2 and β-globin were upregulated 3 days after treatment with inhibitor (Figure 6D-F), resembling the effect of knockdown of PRMT6. Furthermore, KLF1 mRNA and protein levels were increased upon inhibitor treatment (6G, H). Inhibitor treatment had no influence on the amount of PRMT6 protein (Figure 6I), but reduced H3R2 asymmetric methylation as expected (Figure 6J). H4R3me2a, which is mediated by PRMT other than PRMT6, remained unchanged by inhibitor treatment (Figure 6K). The increased expression of GYPA, AHSP and ALAS2 upon inhibitor treatment was inhibited by PRMT6 overexpression, but not in the case of β-globin and KLF1 (Online Supplementary Figure S10). The induction towards erythroid differentiation by PRMT6 inhibition was also seen in an increase of the reddish color of the cell pellet upon inhibitor treatment of K562 cells, indicating increased heme production (Online Supplementary Figure S5).

Inhibition of PRMT6 increases erythroid differentiation of CD34+ cells Our data indicate that PRMT6 inhibition might enhance erythroid differentiation. To investigate this notion directly, we treated primary human CD34+ cells with PRMT6 inhibitor in liquid culture under conditions which allow erythroid or megakaryocytic differentiation. This treatment shifted differentiation towards erythropoiesis, as indicated by the higher levels of GYPA and CD71 surface markers (Figure 7A,B and Online Supplementary Figure S11). Expression of other PRMT6-associated erythroid genes was also increased upon PRMT6 inhibition at the mRNA level (Figure 7C). Treatment of CD34+ cells with PRMT6 inhibitor led to decreased H3R2me2a and increased H3K4me3. H4R3me2a remained unchanged at the GYPA promoter (Figure 7D-F). Similarly, at the established PRMT6 target KLF1,28 H3R2me2a was reduced upon inhibitor treatment (Online Supplementary Figure S12). Subsequently, we analyzed human CD34+ cell differentiation upon PRMT6 inhibition in a CFU assay to examine haematologica | 2018; 103(1)


Epigenetic influence of PRMT6 on hematopoiesis

differentiation independently of cell surface markers. We detected an increase of erythroid colonies upon PRMT6 inhibition (Figure 7G) and a decrease of granulocytic colonies (Figure 7H). These alterations were accompanied by an almost 50% decrease of total colonies at high inhibitor concentration (Figure 7I). Taken together, these data indicate that inhibition of PRMT6 increases erythropoiesis during differentiation of progenitor cells.

Discussion The interplay between transcription factors and their epigenetic cofactors is decisive for the establishment and

A

B

C

D

E

F

G

H

I

J

K

L

haematologica | 2018; 103(1)

maintenance of a cell type-specific gene expression program. In this process, the chromatin environment at cell type-specific genes is adjusted according to cell fate decisions taken at key lineage fate bifurcations. Consequently, alterations in DNA and histone modification patterns activate one gene expression program at the expense of the other. In this study, we made some significant novel observations regarding gene expression control during megakaryopoietic/erythroid lineage differentiation. Our data demonstrate that PRMT6 inhibits erythroid gene expression during lineage differentiation. Under conditions that allow erythroid or megakaryocytic differentiation, the knockdown of PRMT6 enhances erythropoiesis, whereas

Figure 5. Occupancy of the GYPA promoter upon megakaryocytic differentiation of human CD34+ cells. Binding of transcription factors and histone modifications at the GYPA promoter were determined before and after differentiation megakaryocytic (CD34/CD34-M) by ChIP. (A) RUNX1 binding was increased upon megakaryocytic differentiation. (B) TAL1 binding remained similar after megakaryocytic differentiation. (C) GATA1 binding was decreased upon megakaryocytic differentiation. (D) PRMT6 binding was increased after megakaryocytic differentiation. (E) WDR5 binding was decreased upon megakaryocytic differentiation. (F) PADI4 binding was decreased after megakaryocytic differentiation. (G) H3K4me3 was decreased after megakaryocytic differentiation. (H) H3R2me2a was increased upon megakaryocytic differentiation. (I) H3K27me3 was increased upon megakaryocytic differentiation. (J) EZH2 binding increased after megakaryocytic differentiation. (K) HDAC1 binding increased upon megakaryocytic differentiation. (L) Binding of RNA-polymerase II to the GYPA promoter decreased upon megakaryocytic differentiation. Quantitative PCR values of ChIP experiments are shown as percentage input. Values gathered for histone H3 modifications were normalized with ChIP against unmodified histone H3. Error bars depict the standard deviation from at least four evaluations. The P-values were calculated using the Student ttest. *P<0.05; **P<0.01; ***P<0.001.

25


S.C. Herkt et al.

PRMT6 overexpression inhibits erythropoiesis in CFU assays. Furthermore, we showed that PRMT6 mediates the repressive H3R2me2a modification at erythroid genes such as GYPA and KLF1. We detected low levels of PRMT6 present on the GYPA promoter in progenitor cells, which increase upon megakaryopoiesis. Concomitantly, H3R2me2a is increased and this goes hand in hand with the establishment of a repressive histone modification pattern with reduced H3K4me3 at the promoter upon megakaryopoiesis. An analysis of PRMT6 function at the megakaryocytic/erythroid branching with a hematopoi-

A

D

G

I

etic knockout mouse model would be very attractive. Our data also indicate that the transcription factor RUNX1 contributes to PRMT6 recruitment to GYPA promoter as knockdown of RUNX1 reduces PRMT6 occupancy. Interestingly, PRMT6 is present on the promoter of megakaryocytic differentiation genes such as CD41 in progenitor cells. In this case, loss of PRMT6 leads to upregulation of CD41 in stem cell expansion medium.33 Moreover, upon megakaryopoiesis RUNX1 activates these megakaryocytic genes33 and in the same cells RUNX1 is present together with PRMT6 in a repressive

B

C

E

F

H

J

K

Figure 6. Inhibition of PRMT6 increases erythroid gene expression. (A) GYPA expression increases at the mRNA level after treatment of K562 cells with the indicated concentrations of the PRMT6 inhibitor MS023 for 3 days. The control was treated with solvent only (DMSO). Expression was measured by quantitative reverse transcriptase PCR. (B,C) GYPA expression at the cell surface upon treatment of K562 cells with PRMT6 inhibitor for 3 days was determined by flow cytometry using an anti-CD235a-APC antibody. GYPA positivity is given in percent according to the indicated gating. (D-F) Expression of the erythroid genes AHSP, ALAS2 and β-globin increased upon treatment of K562 cells with the indicated concentrations of PRMT6 inhibitor for 3 days. (G,H) Expression of KLF1 increased upon inhibitor treatment on the mRNA and protein level. Expression was measured by quantitative reverse transcriptase PCR and western blot analysis. (I) Western blot analysis of PRMT6 protein expression upon inhibitor treatment of K562 cells for 3 days. (J) Western blot analysis of histone 3 methylation (H3R2me2a) upon inhibitor treatment of K562 cells for 3 days. (K) Western blot analysis of histone 4 methylation (H4R3me2a) upon inhibitor treatment of K562 cells for 3 days. Error bars indicate the standard deviation from four independent determinations. The P values were calculated using the Student t-test. *P<0 .05; **P<0.01; ***P<0.001.

26

haematologica | 2018; 103(1)


Epigenetic influence of PRMT6 on hematopoiesis

complex on erythroid genes.28 In combination, our data indicate that there are two distinct facets of the RUNX1/PRMT6 complex, one associated with megakaryocytic genes in progenitors33 and the other with erythroid genes upon megakaryopoiesis.27,28 Furthermore, we detected genes, that are upregulated or downregulated upon PRMT6 knockdown. This hints towards repressive and activating roles of PRMT6 depending on the gene, as recently proposed.47 How the formation of distinct RUNX1 complexes is regulated is not known; however, different promoter contexts and the modification status of RUNX1 could have a regulatory influence.48-50 Furthermore, different isoforms of RUNX1 could convey altered protein:protein interactions of RUNX1 on distinct

A

B

D

G

promoters. It is also conceivable that RUNX1 itself is methylated by PRMT6 as it was described that PRMT1 is able to perform histone and non-histone methylation in conjunction with RUNX1.48 Recently, it has been shown that the expression of RUNX1 isoform differs between megakaryocytic cells and erythoid cells.51 Our observations hint towards an essential function of PRMT6 in the shutdown of the erythroid gene expression program during megakaryocytic differentiation. The notion that a RUNX1/PRMT6 complex mediates this repression is also supported by our observation that RUNX1 knockdown or PRMT6 knockdown similarly lead to increased GYPA and KLF1 expression (this study and 28). Given that PRMT6 cannot bind DNA directly, its recruitment is

E

H

C

F

I

Figure 7. Inhibition of PRMT6 increases erythroid differentiation of CD34+ cells. (A,B) GYPA expression increases upon treatment of primary human CD34+ cells with PRMT6 inhibitor as measured by flow cytometry using an anti-CD235a-APC antibody. (C) Expression of the erythroid genes AHSP, ALAS2, β-globin and KLF1 increases upon treatment of hCD34+ cells with the indicated concentration of PRMT6 inhibitor for 3 days. Expression was measured by quantitative reverse transcriptase PCR. (D-F) ChIP assay upon PRMT6 inhibitor treatment of CD34+ cells for 3 days. H3R2me2a was decreased upon inhibitor treatment and H3K4me3 was increased upon inhibitor treatment. H4R3me2a remained unchanged upon inhibitor treatment. (G) Treatment of human CD34+ cells with PRMT6 inhibitor MS023 enhances erythroid differentiation at the given inhibitor concentrations in a CFU assay. Error bars give the standard deviation from four independent inhibitor treatments. (H) Treatment of human CD34+ cells with PRMT6 inhibitor MS023 reduced granulocytic differentiation at the given inhibitor concentrations in a CFU assay. Error bars give the standard deviation from four independent inhibitor treatments. (I) The total number of colonies in the CFU assay with human CD34+ cells upon treatment with PRMT6 inhibitor is shown. Error bars indicate the standard deviation from four independent determinations. The P values were calculated using the Student t-test. *P<0.05; **P<0.01; ***P<0.001.

haematologica | 2018; 103(1)

27


S.C. Herkt et al.

dependent on a physical interaction with transcription factors such as RUNX1, as well as possibly with other transcription factors present at regulatory sites. Our recent observation that PRMT6 can also be associated with the important transcription factor, TAL1,40 supports the notion that PRMT6 can be present in distinct gene regulatory complexes, depending on the promoter and the cell type. Recently, PRMT6 has been found to interact with members of the polycomb complex (PRC) and to contribute to PRC-mediated repression.52 It currently remains unclear whether PRMT6 is stably associated with a larger transcriptional complex. It was, however, shown that PRMT6 regulates cell proliferation and senescence.53-56 Our data show that there is also a decrease in colony formation upon knockdown of PRMT6 in human CD34+ cells or when the cells are treated with the PRMT6 inhibitor. However, upon inhibitor treatment no major increase in cell death was observed. For a definite statement on the

References 1. Cantor AB, Orkin SH. Hematopoietic development: a balancing act. Curr Opin Genet Dev. 2001;11(5):513-519. 2. Cantor AB, Orkin SH. Transcriptional regulation of erythropoiesis: an affair involving multiple partners. Oncogene. 2002;21(21):3368-3376. 3. Orkin SH. Diversification of haematopoietic stem cells to specific lineages. Nat Rev Genet. 2000;1(1):57-64. 4. Orkin SH, Zon LI. Hematopoiesis: an evolving paradigm for stem cell biology. Cell. 2008;132(4):631-644. 5. Goode DK, Obier N, Vijayabaskar MS, et al. Dynamic gene regulatory networks drive hematopoietic specification and differentiation. Dev Cell. 2016;36(5):572-587. 6. Goldberg AD, Allis CD, Bernstein E. Epigenetics: a landscape takes shape. Cell. 2007;128(4):635-638. 7. Li E. Chromatin modification and epigenetic reprogramming in mammalian development. Nat Rev Genet. 2002;3(9):662-673. 8. Strahl BD, Allis CD. The language of covalent histone modifications. Nature. 2000;403(6765):41-45. 9. Jenuwein T, Allis CD. Translating the histone code. Science. 2001;293(5532):10741080. 10. Graf T, Enver T. Forcing cells to change lineages. Nature. 2009;462(7273):587-594. 11. Riddell J, Gazit R, Garrison BS, et al. Reprogramming committed murine blood cells to induced hematopoietic stem cells with defined factors. Cell. 2014;157(3):549564. 12. Xie H, Ye M, Feng R, Graf T. Stepwise reprogramming of B cells into macrophages. Cell. 2004;117(5):663-676. 13. Feng R, Desbordes SC, Xie H, et al. PU.1 and C/EBPalpha/beta convert fibroblasts into macrophage-like cells. Proc Natl Acad Sci USA. 2008;105(16):6057-6062. 14. Ness SA, Kowenz-Leutz E, Casini T, Graf T, Leutz A. Myb and NF-M: combinatorial activators of myeloid genes in heterologous cell types. Genes Dev. 1993;7(5):749-759.

28

effect of MS023 on cell growth a detailed analysis of apoptosis, senescence and cell cycle distribution would be essential. The anti-proliferative effect of the loss of PRMT6 merits further investigation in relevant mouse leukemia models. Knockdown of PRMT6 enhances erythropoiesis and pharmacological inhibition of PRMT6 also supports erythropoiesis of primary human CD34+ cells. Significant efforts are being made worldwide to develop efficient in vitro protocols for the production of therapeutic cells from hematopoietic or embryonic stem cells.57 Epigenetic compounds, which target epigenetic factors, such as PRMT6, could contribute to more effective in vitro differentiation in the future. Acknowledgments This work was supported by a grant from the Deutsche Forschungsgemeinschaft to JL (SPP1463, DFG JL1389 5-2). We would like to thank Helge Hussong for assistance.

15. Batta K, Florkowska M, Kouskoff V, Lacaud G. Direct reprogramming of murine fibroblasts to hematopoietic progenitor cells. Cell Rep. 2014;9(5):1871-1884. 16. Zang C, Luyten A, Chen J, Liu XS, Shivdasani RA. NF-E2, FLI1 and RUNX1 collaborate at areas of dynamic chromatin to activate transcription in mature mouse megakaryocytes. Sci Rep. 2016;6:30255. 17. Tijssen MR, Cvejic A, Joshi A, et al. Genome-wide analysis of simultaneous GATA1/2, RUNX1, FLI1, and SCL binding in megakaryocytes identifies hematopoietic regulators. Dev Cell. 2011;20(5):597-609. 18. Dore LC, Chlon TM, Brown CD, White KP, Crispino JD. Chromatin occupancy analysis reveals genome-wide GATA factor switching during hematopoiesis. Blood. 2012;119(16):3724-3733. 19. Ichikawa M, Asai T, Saito T, et al. AML-1 is required for megakaryocytic maturation and lymphocytic differentiation, but not for maintenance of hematopoietic stem cells in adult hematopoiesis. Nat Med. 2004;10(3): 299-304. 20. Moreau T, Evans AL, Vasquez L, et al. Largescale production of megakaryocytes from human pluripotent stem cells by chemically defined forward programming. Nat Commun. 2016;7:11208. 21. Pimanda JE, Ottersbach K, Knezevic K, et al. Gata2, Fli1, and Scl form a recursively wired gene-regulatory circuit during early hematopoietic development. Proc Natl Acad Sci USA. 2007;104(45):17692-17697. 22. Tober J, Yzaguirre AD, Piwarzyk E, Speck NA. Distinct temporal requirements for Runx1 in hematopoietic progenitors and stem cells. Development. 2013;140(18): 3765-3776. 23. Lancrin C, Sroczynska P, Stephenson C, Allen T, Kouskoff V, Lacaud G. The haemangioblast generates haematopoietic cells through a haemogenic endothelium stage. Nature. 2009;457(7231):892-895. 24. Dore LC, Crispino JD. Transcription factor networks in erythroid cell and megakaryocyte development. Blood. 2011;118(2):231239. 25. Elagib KE, Racke FK, Mogass M, Khetawat

26. 27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

R, Delehanty LL, Goldfarb AN. RUNX1 and GATA-1 coexpression and cooperation in megakaryocytic differentiation. Blood. 2003;101(11):4333-4341. Goldfarb AN. Transcriptional control of megakaryocyte development. Oncogene. 2007;26(47):6795-6802. Kohrs N, Kolodziej S, Kuvardina ON, et al. MiR144/451 Expression is repressed by RUNX1 during megakaryopoiesis and disturbed by RUNX1/ETO. PLoS Genet. 2016;12(3):e1005946. Kuvardina ON, Herglotz J, Kolodziej S, et al. RUNX1 represses the erythroid gene expression program during megakaryocytic differentiation. Blood. 2015;125(23):3570-3579. Huang G, Zhao X, Wang L, et al. The ability of MLL to bind RUNX1 and methylate H3K4 at PU.1 regulatory regions is impaired by MDS/AML-associated RUNX1/AML1 mutations. Blood. 2011;118 (25):6544-6552. Koh CP, Wang CQ, Ng CE, et al. RUNX1 meets MLL: epigenetic regulation of hematopoiesis by two leukemia genes. Leukemia. 2013;27(9):1793-1802. Ross K, Sedello AK, Todd GP, et al. Polycomb group ring finger 1 cooperates with Runx1 in regulating differentiation and self-renewal of hematopoietic cells. Blood. 2012;119(18):4152-4161. Yu M, Mazor T, Huang H, et al. Direct recruitment of polycomb repressive complex 1 to chromatin by core binding transcription factors. Mol Cell. 2012;45(3):330343. Herglotz J, Kuvardina ON, Kolodziej S, et al. Histone arginine methylation keeps RUNX1 target genes in an intermediate state. Oncogene. 2013;32(20):2565-2575. Lausen J. Contributions of the histone arginine methyltransferase PRMT6 to the epigenetic function of RUNX1. Crit Rev Eukaryot Gene Expr. 2013;23(3):265-274. Migliori V, Phalke S, Bezzi M, Guccione E. Arginine/lysine-methyl/methyl switches: biochemical role of histone arginine methylation in transcriptional regulation. Epigenomics. 2010;2(1):119-137. Guccione E, Bassi C, Casadio F, et al.

haematologica | 2018; 103(1)


Epigenetic influence of PRMT6 on hematopoiesis

37.

38.

39.

40.

41.

42.

43.

Methylation of histone H3R2 by PRMT6 and H3K4 by an MLL complex are mutually exclusive. Nature. 2007;449(7164):933-937. Hyllus D, Stein C, Schnabel K, et al. PRMT6mediated methylation of R2 in histone H3 antagonizes H3 K4 trimethylation. Genes Dev. 2007;21(24):3369-3380. Iberg AN, Espejo A, Cheng D, et al. Arginine methylation of the histone H3 tail impedes effector binding. J Biol Chem. 2008;283 (6):3006-3010. Lee YH, Ma H, Tan TZ, et al. Protein arginine methyltransferase 6 regulates embryonic stem cell identity. Stem Cells Dev. 2012;21(14):2613-2622. Kolodziej S, Kuvardina ON, Oellerich T, et al. PADI4 acts as a coactivator of Tal1 by counteracting repressive histone arginine methylation. Nat Commun. 2014 May 29;5:3995. doi: 10.1038/ncomms4995. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4(1):44-57. Huang da W, Sherman BT, Zheng X, et al. Extracting biological meaning from large gene lists with DAVID. Curr Protoc Bioinformatics. 2009;Chapter13:Unit 13.11. Ebert BL, Lee MM, Pretz JL, et al. An RNA interference model of RPS19 deficiency in Diamond-Blackfan anemia recapitulates defective hematopoiesis and rescue by dexamethasone: identification of dexamethasone-responsive genes by microarray. Blood. 2005;105(12):4620-4626.

haematologica | 2018; 103(1)

44. Lahlil R, Lecuyer E, Herblot S, Hoang T. SCL assembles a multifactorial complex that determines glycophorin A expression. Mol Cell Biol. 2004;24(4):1439-1452. 45. Eram MS, Shen Y, Szewczyk MM, et al. A potent, selective, and cell-active inhibitor of human type I protein arginine methyltransferases. ACS Chem Biol. 2016;11(3):772-781. 46. Mitchell LH, Drew AE, Ribich SA, et al. Aryl Pyrazoles as potent inhibitors of arginine methyltransferases: identification of the first PRMT6 tool compound. ACS Med Chem Lett. 2015;6(6):655-659. 47. Casadio F, Lu X, Pollock SB, et al. H3R42me2a is a histone modification with positive transcriptional effects. Proc Natl Acad Sci USA. 2013;110(37):14894-14899. 48. Zhao X, Jankovic V, Gural A, et al. Methylation of RUNX1 by PRMT1 abrogates SIN3A binding and potentiates its transcriptional activity. Genes Dev. 2008;22(5):640-653. 49. Huang H, Woo AJ, Waldon Z, et al. A Src family kinase-Shp2 axis controls RUNX1 activity in megakaryocyte and T-lymphocyte differentiation. Genes Dev. 2012;26(14):1587-1601. 50. Wang L, Huang G, Zhao X, et al. Post-translational modifications of Runx1 regulate its activity in the cell. Blood Cells Mol Dis. 2009;43(1):30-34. 51. Draper JE, Sroczynska P, Tsoulaki O, et al. RUNX1B expression is highly heterogeneous and distinguishes megakaryocytic and erythroid lineage fate in adult mouse

52.

53.

54.

55.

56.

57.

hematopoiesis. PLoS Genet. 2016;12(1): e1005814. Stein C, Notzold RR, Riedl S, Bouchard C, Bauer UM. The arginine methyltransferase PRMT6 cooperates with polycomb proteins in regulating HOXA gene expression. PLoS One. 2016;11(2):e0148892. Stein C, Riedl S, Ruthnick D, Notzold RR, Bauer UM. The arginine methyltransferase PRMT6 regulates cell proliferation and senescence through transcriptional repression of tumor suppressor genes. Nucleic Acids Res. 2012;40(19):9522-9533. Kleinschmidt MA, de Graaf P, van Teeffelen HA, Timmers HT. Cell cycle regulation by the PRMT6 arginine methyltransferase through repression of cyclin-dependent kinase inhibitors. PLoS One. 2012;7(8): e41446. Phalke S, Mzoughi S, Bezzi M, et al. p53Independent regulation of p21Waf1/Cip1 expression and senescence by PRMT6. Nucleic Acids Res. 2012;40(19):9534-9542. Wang X, Huang Y, Zhao J, Zhang Y, Lu J, Huang B. Suppression of PRMT6-mediated arginine methylation of p16 protein potentiates its ability to arrest A549 cell proliferation. Int J Biochem Cell Biol. 2012;44(12):2333-2341. Batta K, Menegatti S, Garcia-Alegria E, Florkowska M, Lacaud G, Kouskoff V. Concise review: recent advances in the in vitro derivation of blood cell populations. Stem Cells Transl Med. 2016;5(10):13301337.

29


ARTICLE

Bone Marrow Failure

Ferrata Storti Foundation

Cancer in the National Cancer Institute inherited bone marrow failure syndrome cohort after fifteen years of follow-up Blanche P. Alter,1 Neelam Giri, 1 Sharon A. Savage1 and Philip S Rosenberg 2 Clinical Genetics and 2Biostatistics Branches, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD, USA

1

Haematologica 2018 Volume 103(1):30-39

ABSTRACT

T

Correspondence: alterb@mail.nih.gov

Received: August 4, 2017. Accepted: October 13, 2017. Pre-published: October 19, 2017.

he National Cancer Institute Inherited Bone Marrow Failure Syndromes Cohort enrolls patients with the four major syndromes: Fanconi anemia, dyskeratosis congenita, DiamondBlackfan anemia, and Shwachman-Diamond syndrome, and follows them with a common comprehensive protocol. The current analysis includes more than double the numbers of patients and person-years since our first report, published in 2010. Patients with Fanconi anemia and dyskeratosis congenita developed head and neck and anogenital squamous cell carcinomas at rates that were hundreds-fold greater than those of the general population. In competing risk analyses the cumulative incidence of severe bone marrow failure, leading to stem cell transplantation or death, was more than 70% by age 60. Patients with Diamond-Blackfan anemia developed lung, colon, and cervical cancer at rates greater than those of the general population. The cumulative incidence of severe bone marrow failure in those with Diamond-Blackfan anemia was 50% by age 60. The smaller group, with ShwachmanDiamond syndrome, have not as yet developed a significant number of solid tumors, but 40% developed bone marrow failure by age 50. The risk of solid tumors following stem cell transplantation in Fanconi anemia and in dyskeratosis congenita was significantly higher than in nontransplanted patients. There was no clear association of genotype with cancer in any of the syndromes. Cancer was most common in Fanconi anemia, followed by dyskeratosis congenita; Diamond-Blackfan anemia and Shwachman-Diamond syndrome are less cancer-prone, but nonetheless all patients are at increased risks of bone marrow failure and specific cancers. clinicaltrials.gov Identifier: 00027274

doi:10.3324/haematol.2017.178111

Introduction Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/30 Š2017 NIH (National Institutes of Health)

30

The inherited bone marrow failure syndromes (IBMFS) include specific genetic disorders with varying degrees of hematopoietic failure, birth defects, high risks of cancer, and mutations in genes in unique molecular pathways.1 The most frequent syndrome is Fanconi anemia (FA), a primarily autosomal recessive disorder in which pathogenic variants have been recognized in at least 21 genes whose products collaborate in the FA/BRCA DNA repair pathway. Children with FA may have bone marrow failure (BMF) and characteristic physical findings, such as short stature, radial ray anomalies, microcephaly, abnormal kidneys, and others.2 There is notable clinical heterogeneity, with diagnosis of adults with normal physical appearances and normal hematopoiesis, who are only identified when they develop typical FA cancers or through family studies. The diagnosis of FA is confirmed by the detection of increased chromosome breaks after culture of cells with a DNA cross-linker, such as diepoxybutane or mitomycin C.3 Case reports of malignancies in patients with FA were first reviewed in 1996, and the types and risks of cancer have been described in retrospective and contemporary cohorts, including patients from North America, France, Germany, Israel, and elsewhere.4-9 We previously reported overall risks of leukemia and solid tumors in the National Cancer Institute (NCI) FA cohort of approximately 30- to 50-fold; the haematologica | 2018; 103(1)


Cancer in the NCI IBMFS cohort after 15 years

highest risks were acute myeloid leukemia (AML), head and neck squamous cell carcinoma (HNSCC), and vulvar SCC;8 risks were similar in the other FA cohorts. The risks and types of cancer are not as well studied in the other IBMFS. Patients with dyskeratosis congenita (DC) may present with marrow failure and features of the diagnostic triad: dysplastic nails, lacy reticular pigmentation, and oral leukoplakia.2 Others may have a variety of additional findings, such as pulmonary or hepatic fibrosis, avascular necrosis of the hips, or stenosis of lacrimal ducts, esophagus, and/or urethra.10 The diagnostic test for DC is the presence of very short telomeres in blood leukocytes.11 Pathogenic variants have been identified in more than 12 genes involved in telomere biology.1 The cancer pattern in DC is similar to that of FA, with lower risks in DC.8,12 Patients with Diamond-Blackfan anemia (DBA) also have an increased risk of cancer.13 Patients with DBA are usually diagnosed in infancy or childhood, with macrocytic anemia and elevated levels of red cell adenosine deaminase.2,14 More than 14 primarily autosomal dominant genes have been identified in the ribosome biogenesis pathway.15 The solid tumors in DBA are osteosarcoma, colon, and lung cancer, distinct from the other IBMFS. Shwachman-Diamond syndrome (SDS) is characterized by exocrine pancreatic insufficiency and neutropenia.2 Cancer information is limited to case reports and small series; cancer is primarily AML.16 Most patients with SDS have biallelic pathogenic variants in SBDS, which is involved in ribosome biogenesis; a few patients have mutations in DNAJC21 or EFL1.17,18 The potential impact of hematopoietic cell transplant (HCT) on cancer in IBMFS is a crucial question. We previously compared cancer in non-transplanted patients in the North American Survey with cancer in transplanted patients in the French FA cohort.5 We found that the risk increased by more than 4-fold, while the median age dropped by 16 years. The potential contributions of total body irradiation and chronic graft-versus-host disease (GvHD) were not statistically significant in that study. There are no comparable age-dependent analyses from other FA cohorts or other types of IBMFS, although data from Europe suggest an increase in cancer based on the interval from transplantation in FA.19 Herein, we report a new analysis of the cumulative cancer experience of individuals enrolled in the NCI IBMFS Cohort. Our first report included 196 IBMFS patients and 4302 person-years among the four IBMFS enrolled and analyzed for seven years, up to and including 2008.8 We now have more than double the number of affected individuals (530) and person-years (12607) after 15 years, with more precise estimates of hazard rates and relative risks. The current data permit stronger evidence-based counseling of patients with an IBMFS.

Methods The NCI IBMFS Cohort opened in January 2002, and continues to accrue patients (protocol 02-C-0052, clinicaltrials.gov Identifier: 00027274). The study was approved by the NCI Institutional Review Board, and participants or their proxies sign consent and medical release forms. Enrollment is voluntary, with information provided directly by family members. A family contact initiates a telephone interview, followed by completion of a Family History Questionnaire. Each family member fills in an Individual haematologica | 2018; 103(1)

Information Questionnaire, and biennial follow-up forms are sent. First-degree family members are included in the study evaluations, in order to identify undiagnosed cases, determine carrier status, and provide genetic counseling. All participants enroll in the “Field Cohort”, and a subset (the “Clinical Center Cohort”) is evaluated at the National Institutes of Health Warren G Magnuson Clinical Center in Bethesda MD, USA. Competing adverse events include severe BMF (sufficiently severe to lead to death or HCT);4 myelodysplastic syndrome (MDS; severe cytopenia with dyspoietic morphology of marrow cells, with or without a cytogenetic clone);20 acute leukemia, usually AML; or solid tumors or lymphomas.8 Patients in the current report enrolled from January 2002 through December 2015, with follow-up through May of 2016. Clinical diagnoses were validated by syndrome-specific tests (chromosome breakage for FA; telomere length by flow cytometry and fluorescent in situ hybridization for DC; red blood cell adenosine deaminase for DBA; and serum isoamylase and trypsinogen for SDS),3,14,21,22 and confirmed by genetic testing whenever possible. Cancer diagnoses were provided by self-report or by proxy, and confirmed in 60% by review of medical records. Lack of a report of cancer led to the conservative assumption that the patient had not had cancer. The classification of a patient as having MDS was done according to the medical records provided by the patient. We were unable to do central review, and thus MDS may have been overreported according to minor dyspoieses or cytogenetic reports. Analyses were done using Microsoft Excel Office 365 Proplus version 1609 (Microsoft, Redmond, WA, USA), Stata 14.2 (StataCorp, College Station, TX, USA), and MATLAB2017A (the MathWorks, Natick, MA, USA). Survival probabilities were calculated by the Kaplan-Meier method in the absence of competing risks with censoring at death.23 Cumulative incidence and causespecific hazards accounting for competing risks (BMF, leukemia, or solid tumors) were determined as described previously.4 The ratio of observed-to-expected cancers (O/E) was derived from general population incidence data from the Surveillance, Epidemiology, and End Results (SEER) Program, adjusting for age, sex, race, and birth cohort.24 Sex ratios were analyzed with the binomial test of comparison with a ratio of 1:1. Statistical tests were 2-sided, and P-values <0.05 were considered significant.

Results We enrolled 360 families with at least one member with DBA, DC, FA, or SDS (Table 1), including 530 affected individuals and 12607 affected person-years. There was an excess of males among those with DC compared with the other syndromes. The median age in May 2016 (or age at death) was higher in DC, associated with diagnosis in older adults, while the median survival age was higher in patients with DBA (67 years) compared with 51 years in DC and 39 years in FA, due to the lower risk of death from BMF or cancer in DBA. More than 80% of the affected participants survived to adulthood (≥18 years of age). Thirty and 40% of those with DC and FA had received an HCT, compared with 7 and 25 % of those with DBA or SDS. The crude death rates were more than 30% in DC and FA, and around 10% in DBA and SDS. Malignancies in patients who had not undergone HCT were more frequent in DC and FA (>10%) than in DBA and SDS (Table 2 and Table 3). Leukemia was reported in 3% of DC, FA, and SDS (and none of those with DBA), while solid tumors occurred in 5% of DBA, 7% of DC, 31


B.P. Alter et al.

12% of FA, and 3% of those with SDS. Two patients with DBA, four with DC, and five with FA had multiple malignancies (Table 2). The O/E ratio for any malignancy in non-transplanted patients was 2.5 for DBA, 4.2 for DC, 19 for FA, and 8.5 for SDS (Table 3). The significant cancers in DBA were

lung, colon, and cervix; in DC they were HNSCC (primarily tongue), AML, non-Hodgkin lymphoma (NHL), and anal SCC; in FA, HNSCC (primarily tongue), AML, vulva, esophagus, brain, and anal SCC were presented; and in SDS one ovarian cancer and one AML were statistically significant.

Table 1. Participants. Number of Families Number of Patients Person-Years Birth Years, Median (range)† Male:Female M:F ratio Number alive** Age alive, Median, years (range)** Number deceased** Age deceased, Median, years (range)** Age Median, years (range)** Median Survival, years (CI)*** % reached ≥18 years

DBA

DC

FA

SDS

Total

87 135 3458 1996 (1921-2014) 72:63 1.1:1 122 (90%) 19 (1.5-87) 13 (10%) 33 (0-70) 20 (0-87) 67 (57-70) 95

108 197 5655 1991 (1902-2014) 127:70 1.8:1* 127 (74%) 25 (4-79) 70 (36%) 29 (1.3-82) 26 (1.3-82) 51 (46-55) 90

130 163 2854 1996 (1945-2014) 72:91 0.79:1 111 (68%) 17 (2-64) 52 (32%) 20 (0.2-58) 17 (0.2-64) 39 (35-44) 80

35 35 640 1998 (1962-2012) 15:20 0.75:1 31 (89%) 17 (4-39) 4 (11%) 31 (13-53) 18 (4-53) 41 (32-50) 95

360 530 12607 1993 (1921-2014) 286:244 1.2:1 391 (74%) 20 (1.5-87) 139 (26%) 26 (0-82) 22 (0-87) -

For birth year, DC older than DBA, FA or SDS. *P<0.001 for excess of males. **Status and age in May 2016 or at death. DC older than DBA or FA, DBA or SDS. ***From KaplanMeier survivals (see Figure 3). CI, 95% confidence interval. DBA: Diamond-Blackfan anemia; DC: dyskeratosis congenita; FA: Fanconi anemia; SDS: Shwachman-Diamond syndrome; M: male; F: female.

Table 2. Complications in participants. No or pre-stem cell transplantation (HCT) Number N with malignancy (%)* N with leukemia (%) N with solid tumor (%) N with lymphoma (%) N of malignancies among N with malignancy N of leukemias N of solid tumors N of lymphomas N with multiple malignancies (%)** N with myelodysplastic syndrome (MDS) (%) N died from malignancy, no HCT (%) Post-stem cell transplantation (HCT) N who had HCT (% of total) N with malignancy after HCT (%)*** N died after HCT related to HCT N died from malignancy after HCT (%)

DBA

DC

FA

SDS

Total

135 7 (5) 0 7 (5) 0 9 0 9 0 2 (1.5) 3 (2) 3 (2)

197 20 (10) 6 (3) 14 (7) 2 (1) 27 7 17 3 4 (2) 21 (11) 8 (4)

163 21 (13) 5 (3.1) 20 (12) 0 26 5 21 0 5 (3) 26 (16) 15 (9)

35 2 (6) 1 (2.9) 1 (2.9) 0 2 1 1 0 0 6 (17) 2 (6)

530 50 (9.4) 12 (2.3) 37 (7) 2 (0.4) 55 13 49 3 11 (2) 56 (11) 28 (5)

10 (7.4) 1 (10) 2 (20) 1 (10)

60 (30.5) 3 (5) 17 (28) 1 (1.7)

63 (38.7) 7 (11) 16 (25) 4 (6)

3 (8.6) 0 -

136 (25.7) 11 (8) 35 (26) 6 (4)

* Percent of number with no or pre-HCT. **Multiple malignancies in non-transplanted patients: DBA: Cervix and lung; colon and liver. DC: Leukemia, cervix and thyroid; nonHodgkin lymphoma (NHL), lip, NHL; Acute lymphocytic leukemia (ALL), acute myeloid leukemia (AML) and soft tissue; mouth and pharynx. FA: mouth and mouth; tongue and esophagus; anus and vulva; cervix and vulva; tongue and lip. ***Percent of number who had HCT. DBA: Diamond-Blackfan anemia; DC: dyskeratosis congenita; FA: Fanconi anemia; SDS: Shwachman-Diamond syndrome; N: number; HCT: stem cell transplantation.

32

haematologica | 2018; 103(1)


Cancer in the NCI IBMFS cohort after 15 years

Following HCT, one of 10 patients with DBA developed post-transplant lymphoproliferative disease (PTLD); three of 60 patients with DC developed tongue SCC, Hodgkin disease, and PTLD, respectively; seven of 63 patients with FA developed a total of 12 malignancies, including HNSCC, pharyngeal SCC, vulvar SCC, thyroid cancer, and PTLD (Table 4). One brain tumor developed 10 months after HCT in a patient with FANCD1/BRCA2. The effect of HCT on increased cancer rates is reflected in the higher O/E ratios for the transplanted vs. the non-transplanted groups: DBA 81 vs. 2.5, DC 30 vs. 4.2, FA 55 vs. 19 (Table 3 and Table 4). MDS was analyzed separately (Table 3), since it was based on self-report or medical records, but the slides were not centrally reviewed. The O/E ratio for MDS in DBA was 42, in DC 578, in FA 5669; there were no MDS cases in SDS. Nonmelanoma skin cancer (Table 3) was rare in DBA and not reported in SDS, but quite frequent in DC and FA, with multiple cancers in a few patients, including both

A

B

basal cell carcinomas (BCC) and skin SCC. There was one SCC and six BCC in DBA, 14 SCC and six BCC in DC, and 23 SCC and 12 BCC in FA. Nine out of 197 patients with DC and 11 out of 163 with FA had one or more skin cancers. Two patients with DC had two each, two had three each, and one had five skin cancers; while six patients with FA had two skin cancers each, one had three, and one reported 17 separate skin cancers. The competing risks of adverse first events are demonstrated in Figure 1. The cumulative incidence of HCT for severe BMF or death without malignancy rose linearly in DC to more than 70% by age 70, due to an annual hazard rate of about 2% in childhood with a sharp rise at around age 50 that reached 10% by age 80 years. The cumulative incidence of severe BMF in FA leveled off at 70% by age 50, with a peak hazard rate of 4% at age 12-15, and a plateau of 5% at age 45. Severe BMF following the syndrome-specific red blood cell aplasia in DBA reached around 40% by age 70; the maximal level in SDS was 20%.

C

D

Figure 1. Complications in patients in the NCI inherited bone marrow failure syndromes cohort. (A-D) Cumulative incidence (top panels) and annual hazard rates (bottom panels) of competing adverse events by age in patients with (A) Diamond-Blackfan anemia (DBA), (B) Dyskeratosis congenita (DC), (C) Fanconi anemia (FA) and (D) Shwachman-Diamond syndrome (SDS). Adverse events are severe bone marrow failure leading to hematopoietic cell transplantation or death (HCT or death, blue), acute leukemia (red), or solid tumors (ST, black). Top panels, cumulative incidence of experiencing each event as the initial cause of failure (observed, stairstep lines) and 95% confidence intervals (shaded areas). Bottom panels, annual hazard rates (modeled, smooth curves), i.e., % per year who develop a given event type among subjects who are still susceptible, and 95% confidence intervals (shaded areas).

A

B

C

D

Figure 2. Cumulative incidence and annual hazard rates of myelodysplastic syndrome (MDS) by age. Patients with (A) Diamond-Blackfan anemia (DBA), (B) Dyskeratosis congenita (DC), (C) Fanconi anemia (FA) and (D) Shwachman-Diamond syndrome (SDS). Top panels show observed (stair-step lines) and modeled (smooth curves) cumulative incidence curves. Bottom panels show hazard rates (smooth curves) as described in Figure 1. MDS is not considered a competing risk.

haematologica | 2018; 103(1)

33


B.P. Alter et al.

The cumulative incidence of solid tumors was close to 20% by age 65 in both DC and FA (Figure 1). The contours of the hazard rates were different, essentially plateauing at 2% per year in DC, while the hazard rate increased more than exponentially in FA after age 30. Solid tumor cumu-

lative incidence in DBA was more than 50% by age 70, and was around 40% in SDS by age 45. Leukemia had a cumulative incidence of under 10% in DC, and under 5% in FA and SDS, by ages 70, 30, and 20, respectively, with low hazard rates; no cases of AML were reported in our

Table 3. Cancers in patients in the NCI IBMFS cohort who did not have HCT

Syndrome

Cancer

Ages (years) Median (range)

Observed N

Expected N

O/E

95% CI

DBA N=135 PY=3458

All sites* Solid Tumors* Lung* Colon* Cervix* Liver MDS* Skin 8 ca, N = 5 1 NMS, N = 1 1 SCC, N = 1 6 BCC, N = 4 All sites* Solid Tumors* HNSCC* Tongue* Leukemia* AML* NHL* Anus* Esophagus Rectum Cervix Thyroid MDS* Skin 22 ca, N = 9 2 NMS, N = 2 14 SCC, N = 7 6 BCC, N = 4 All sites* Solid Tumors* HNSCC* Tongue* Leukemia* AML* Vulva* Esophagus* Brain* Anus* Lung Cervix Breast MDS* Skin 35 ca, N = 11 23 SCC, N = 10 12 BCC, N = 4 All sites* Solid Tumors Ovary* Leukemia AML*

49 (17-70) 49 (17-70) 53 (47-70) 34, 55 17, 40 34 13 45 (39-63)

9 9 4 2 2 1 1 8 1 1 6 27 17 11 8 7 5 3 1 1 1 1 1 18 22 2 14 6 26 21 10 5 5 5 3 2 2 1 1 1 1 26 35 23 12 2 1 1 1 1

3.68 3.12 0.34 0.18 0.05 0.04 0.02

2.5 2.9 12 11 37 27 42

1.1-4.7 1.3-5.5 3.2-30 1.4-41 4.6-136 0.7-151 1.1-234

6.47 5.57 0.15 0.04 0.3 0.07 0.29 0.02 0.04 0.16 0.14 0.27 0.03

4.2 3.1 74 216 24 73 11 47 28 6.4 7.3 3.7 578

2.8-6.1 1. 8-4.9 37-133 94-427 9. 5-48 23-169 2.2-30 1.2-262 0.7-157 0.2-36 0.2-41 0.1-20 343-914

1.37 1.1 0.02 0 0.13 0.02 0.01 0 0.09 0 0.04 0.05 0.3 0

19 19 527 1054 40 213 582 1266 23 256 26 20 3.4 5669

12-28 12-29 253-970 342-2460 13-92 69-496 120-1702 153-576 2.8-84 6.5-1427 0.7-144 0.5-109 0.1-19 3703-8307

0.24 0.18 0.01 0.03 0

8.5 5.5 169 34 202

1.02-30 0.1-31 4.3-944 0.9-194 5-1126

DC N=197 PY=5655

FA N=163 PY=2854

SDS N=35 PY=640

38 (18-63) 38 (18-61) 38 (18-61) 33 (18-42) 40 (28-63) 40 (28-56) 57 (43-65) 34 38 37 37 37 31 (4-73) 35 (14-54)

30 (4-58) 34 (4-58) 37 (29-53) 37 (29-42) 17 (12-27) 17 (12-27) 29 (24-39) 34, 35 3.7, 4.9 33 58 22 30 13 (1-57) 33 (26-41)

31 (19-43) 43 43 19 19

*Bold = significant at P<0.05. Italicized = subset of category in the row above. Age is age at malignancy; age at first skin cancer if multiple skin cancers. Pt: patient; N: number of patients; O/E: observed-to-expected; CI: confidence interval; SCC: squamous cell carcinoma; BCC: basal cell carcinoma; HNSCC: head and neck squamous cell carcinoma; ca: cancer; NMS: nonmelanoma skin cancer; DBA: Diamond-Blackfan anemia; DC: dyskeratosis congenita; FA: Fanconi anemia; SDS: Shwachman-Diamond syndrome; PY: personyears; MDS: myelodysplastic syndrome; AML: acute myeloid leukemia; NHL: non-Hodgkin lymphoma.

34

haematologica | 2018; 103(1)


Cancer in the NCI IBMFS cohort after 15 years

DBA cohort. MDS, a non-competing risk, was analyzed separately (Figure 2). By age 50 in each syndrome, the respective cumulative incidences (and confidence intervals) were 50% (35-65%) in FA, 20% in DC (15-25%), 65% (25-100%) in SDS, and 5% (0-15%) in DBA. Thus the cumulative incidence was higher in FA than in DC, and both were higher than in DBA. The median overall survival was longest in DBA (67 years), followed by DC (51 years), SDS (41 years) and FA (39 years) [Figure 3 (arrows), Table 1]. These data include cases who were only identified as adults, such as patients with DC who had pulmonary fibrosis as older adults, patients with FA with solid tumors, or parents of affected children in dominant disorders such as DBA or some DC genotypes, who had the same pathogenic variant as their child. The most frequent causes of death in the non-transplanted patients were aplastic anemia and malignancies in FA and DC as well as pulmonary disease in DC, and iron overload and malignancies in DBA. Transplanted patients with FA developed solid tumors, pneumonia, GvHD, and renal disease, while those with DC developed solid tumors and pulmonary fibrosis. The effect of HCT on cancer incidence is shown in Figure 4. Patients with FA had a striking increase in cumulative

incidence of cancer and left shift to younger age (Figure 4A). These data are skewed by the medulloblastoma 10 months after HCT in the patient with FANCD1/BRCA2; the trend persisted after removal of that data point (Figure 4B).25 Cancer cumulative incidence was also higher and earlier post-HCT in DC (Figure 4C). We identified the genotype in 61% of the patients with DBA, 78% of those with DC, 73% of those with FA, and 82% of those with SDS (Table 5). Twenty-eight patients with DBA had pathogenic variants in RPS19, and 21 in RPS29. There were five cancers in four individuals from one DBA family among the 21 patients with mutations in RPS29. This significant association (P<0.001) is confounded by the familial component. The majority of the pathogenic variants in the patients with DC were in DKC1, TERT, TERC, RTEL1 and TINF2; the number of patients with cancer were 5, 3, 5, 0 and 1, respectively. The excess of cancers in those with mutations in DKC1, TERT, or TERC was significant (P=0.02); however, the patients with pathogenic variants in RTEL1 and TINF2 (and fewer cancers) were generally younger than in the first three genotype groups. The most frequent genotypes in the patients with FA were FANCA and FANCC, and there was no association of genotype with cancer; there were 11 cancers in 70 patients with

A

B

C

D

Figure 3. Survival curves for patients in the NCI IBMFS cohort, showing proportion alive by age in years. (A) Diamond-Blackfan anemia (DBA), (B) Dyskeratosis congenita (DC), (C) Fanconi anemia (FA) and (D) Shwachman-Diamond syndrome (SDS). Tick marks represent patients who were still alive at the time of analysis. Observed survival (stair-step lines) and smoothed survival (smooth curves) are shown; shaded areas show 95% confidence intervals for the smoothed curves.

haematologica | 2018; 103(1)

35


B.P. Alter et al.

FANCA, three in 19 patients with FANCC, and seven in 74 patients with other genotypes (global P=0.5).

Discussion The NCI IBMFS cohort includes more than 500 patients with one of the four major rare IBMFS, more than twice

A

the number discussed in our previous report.8 The current estimates are more precise, extend through older ages, and cover all four syndromes (versus mostly FA and DC in the prior report). These significant data are most stable (narrower confidence intervals) for FA and DC, due to the larger number of events than in the other syndromes. The median survival improved by 5-10 years. The data reflect the additional eight years of follow-up, as well as

B

C

Figure 4. Cumulative incidence of solid tumors by age in non-transplanted and transplanted patients. Non-transplanted, blue and transplanted, red. (A) Fanconi anemia (FA), including one patient with FANCD1/BRCA2 who developed a brain tumor at age 32 months, 21 months following a stem cell transplant. The rate ratio for transplanted/non-transplanted is 3.5 (95% CI 1.2 â&#x20AC;&#x201C; 9.2). (B) FA, excluding the patient with FANCD1/BRCA2. The rate ratio is 3 (95% CI 0.97 - 8.2). (C) Dyskeratosis congenita (DC). The rate ratio is 5.7 (95% CI 1.1 â&#x20AC;&#x201C; 20.3).

Table 4. Cancers in patients in the NCI IBMFS cohort, following transplant

Syndrome DBA N=10 PY=53 DC N=60 PY=248

FA N=63 PY=544

SDS N=3 PY=33

Cancer

Ages (years) Median (range)

Observed N

Expected N

O/E

95% CI

All sites, N = 1 Solid Tumors NHL* All sites, N = 3 Solid Tumors HNSCC Tongue Hodgkin NHL* Skin BCC Skin SCC All sites, N = 7 Solid Tumors HNSCC Lip Tongue Hypopharynx Larynx Vulva Brain Thyroid NHL* Skin BCC Skin SCC All sites Solid Tumors

10

1 0 1 3 1 1 1 1 1 1 3 12 11 5 1 2 2 1 2 1 2 1 9 12 0 0

0.01 0.01 0 0.1 0.08 0 0 0.01 0.01

81 0 983 30 13 432 1561 164 141

2.05-451 0-451 25-5474 6-87 0.3-73 11-2404 40-8699 4-914 4-786

0.22 0.16 0.01 0 0 0 0 0 0.02 0.02 0.02

55 67 933 4,662 1,451 16935 1383 6298 61 100 66

29-97 33-120 303-2178 118-25975 176-5244 2051-61174 35-7704 763-22752 1.5-339 12-360 1.7-369

0.01 0

0 0

0-579 0-908

10 19 (15-29) 18.8 18.8 18.8 14.9 29.1 50 35 (14-37) 31 (3-51) 33 (3-51) 41 (25-51) 41 25, 48 33, 51 42 25, 28 2.7 24, 34 30 32 31 (30-41)

Bold = significant at P<0.05. Italicized = subset of category in the row above. *NHL, probably post-transplant lymphoproliferative disease (PTLD). Age is age at malignancy; age at first skin cancer if multiple skin cancers. One patient with FA had 8 BCC and 9 SCC. Another with FA had 1 BCC and 1 SCC. Observed N: number of events; N: number of patients; O/E: observed-to-expected; CI: confidence interval; SCC: squamous cell carcinoma; BCC: basal cell carcinoma; HNSCC: head and neck squamous cell carcinoma; DBA: DiamondBlackfan anemia; DC: dyskeratosis congenita; FA: Fanconi anemia; SDS: Shwachman-Diamond syndrome; PY: person-years; NHL: non-Hodgkin lymphoma.

36

haematologica | 2018; 103(1)


Cancer in the NCI IBMFS cohort after 15 years

improvements in clinical management, combined with diagnosis of affected individuals at older ages. The older participants increasingly comprise parents and grandparents in autosomal dominant disorders (DBA, some DC) who carry the same pathogenic variant as the proband. In addition, a growing number of individuals may have been diagnosed with DC as older adults, after the development of recently recognized complications such as pulmonary fibrosis or hepatic disease, coupled with the expansion of genetic testing and appreciation of a broader clinical spectrum. The diagnosis of FA is also occurring in older individuals, such as a patient who presented with a solid tumor at age 30,26 or another identified as a potential HCT donor for an adult sibling with FA (Alter, unpublished). These two patients had hematopoietic somatic mosaicism which may have contributed to their survival free of BMF.3 More than one-third of the patients with DC or FA received an HCT. The cumulative incidence of receiving an HCT or dying from severe BMF by age 50 was 70% in FA and 50% in DC. The annual hazard rate for HCT or non-malignant death was 3-5% in FA by age 10, and 1-2% in children with DC, and rose after age 50 in DC. The hazard rate in FA appeared to level off in adults with FA, but kept increasing in adults with DC. HCT was reported in less than 10% of the participants with DBA or SDS, perhaps because some patients with DBA respond to corticosteroids or receive red blood cell transfusions and iron chelation. Those with SDS and neutropenia may have responded to treatment with granulocyte-colony stimulating factor (G-CSF), and did not develop pancytopenia. All of the syndromes had relatively low frequencies of leukemia, but high O/E ratios for AML. The O/E for AML in the NCI patients with DC was 73, and was above 200 in those with FA, consistent with the O/E of 300 in our earlier report, but lower than the O/E of 600-1000 in three other FA cohorts.4,6,7 High O/E ratios for AML reflect its development in patients with an IBMFS at a younger age

than in sporadic adult cases. The confidence intervals are wide due to the small numbers of cases in the cohorts. The lower values in the current analysis compared with earlier studies may indicate regression to the mean, or may reflect enrolment biases toward patients with certain cancers in the earlier phase of our cohort. Some participants in our cohort were only diagnosed during family studies after a sibling developed leukemia or a solid tumor; the affected sibling with a malignancy was also enrolled in our cohort. Nevertheless, BMF as well as MDS remain major complications in adults with each of the four major IBMFS. The major cancer sites in FA included HNSCC (primarily tongue), AML, vulva, esophagus, and brain, and are consistent in all the reported FA cohorts. Malignancies in DC were also HNSCC and AML as well as NHL. The unique cancer spectrum in DBA was lung, colon and cervix in the NCI DBA cohort, and lung, colon and osteosarcoma in the DBAR.13 Hence, our new analysis supports the conclusion that the rate of development of cancer is highest in FA, less in DC, and substantially lower and of different types in DBA. Our SDS cohort is small, and the single cases of leukemia and ovarian cancer are too few to indicate an increased rate of these malignancies. The pathophysiology of HNSCC in these syndromes is not clear; the possible role of human papillomavirus in HNSCC is controversial,27 but seems unlikely.28,29 Solid tumors following HCT were first quantified in our comparison of cancer incidence in patients with FA transplanted in Paris with non-transplanted patients in the North American Survey (NAS).4,5 We now confirm a similarly increased risk and earlier median age for cancer in transplanted patients. Others had suggested a transplantrelated increased risk of solid tumors in FA, although the cumulative incidence was not reported by age.19,30 We found a similar transplant effect on cancer in patients with DC. We do not have sufficient data and the transplant reg-

Table 5. Genotypes according to frequency, total group and those with cancer without HCT.

DBA Genes RPS19 RPS29* RPL26 RPL35A RPL11 RPS7 RPS24 RPS10 RPL5 RPS17 RPL15 Unknown** Total with Gene Total Overall

DC Genes

Total (CA)

DKC1 TERT TERC RTEL1 TINF2 PARN WRAP53 CTC1 ACD

35 (5) 34 (3) 29 (5) 23 22 (1) 4 3 (1) 3 1

Total (CA) 28 (1) 21 (5) 7 7 4 4 3 3 3 2 1 52 (1) 83 (6) 135 (7)

43 (5) 154 (15) 197 (20)

FA Genes

Total (CA)

FANCA FANCC FANCD2 FANCG FANCI FANCJ FANCD1* FANCB FANCF FANCR

70 (11) 19 (3) 7 7 (1) 4 4 3 (2) 2 2 1 44 (4) 119 (17) 163 (21)

SDS Genes

Total (CA)

SBDS2*** SBDS1***

26 (2) 1

8 27 (2) 33 (2)

All patients, not according to families, not age adjusted. Several were assigned a gene by pedigree. Columns indicate the number of patients with each genotype (number with cancer). *Significant association of cancer with genotype. Four people in one RPS29 family had three lung, one cervix, and one colon cancer. **Incomplete or truly unknown and unknowable (no DNA available) ***SBDS1 means only one mutated allele identified. SBDS2 means two mutated alleles. DBA: Diamond-Blackfan anemia; DC: dyskeratosis congenita; FA: Fanconi anemia; SDS: Shwachman-Diamond syndrome; CA: cancer

haematologica | 2018; 103(1)

37


B.P. Alter et al.

imens are too diverse to identify specific associations between cancer risk with the use of irradiation in the preparative regimens, or the development of GvHD. Future studies of radiation-free HCT, with careful documentation of GvHD, and focusing on the age of the patient rather than the interval following HCT, may shed light on this concern. The assumption must be that any patient with an IBMFS who has received an HCT will, at best, not be at less risk for a solid tumor, and thus cancer surveillance remains a high priority. Our cohort remains too small to lead to significant conclusions about the relation of genotype to cancer. The association with RPS29 in DBA is confounded by all five cancers occurring in a single family. The brain tumors in all three of the patients with FANCD1/BRCA2, two without HCT and one within 10 months following HCT (and thus probably not due to the HCT), are consistent with the known extremely high risk of this cancer in patients with that genotype.25 The NCI cohort is too small to have large numbers of patients with each genotype (except FANCA). It will be important to consider whether a pathogenic variant leads to a null or a hypomorphic phenotype, since the former might be expected to be more severe than the latter.31 Therefore, further follow-up and enrollment of additional subjects are necessary to determine additional genotype-phenotype associations in FA, DC, and DBA. Limitations of our study include biased enrolment because of possible specific interest in cancer risks. Our participants are predominantly North American; specific genotypes and cancer risks may differ in other populations. Mild cases with less severe BMF and/or fewer birth defects, or with mosaicism, may not have been identified. There may have been left truncation of very severe patients, who died without a diagnosis and were not entered into our study retrospectively. We may be skewed toward adult patients, since we are not specifically a Childrenâ&#x20AC;&#x2122;s hospital. Strengths of our study include comparison of patients with the major IBMFS in a single cohort, with presumably

References 1. Wegman-Ostrosky T, Savage SA. The genomics of inherited bone marrow failure: from mechanism to the clinic. Br J Haematol. 2017;177(4):526-542. 2. Shimamura A, Alter BP. Pathophysiology and management of inherited bone marrow failure syndromes. Blood Rev. 2010; 24(3):101-122. 3. Fargo JH, Rochowski A, Giri N, Savage SA, Olson SB, Alter BP. Comparison of chromosome breakage in non-mosaic and mosaic Fanconi anemia patients, relatives, and other inherited bone marrow failure syndrome patients. Cytogenet Genome Res. 2014;144(1):15-27. 4. Rosenberg PS, Greene MH, Alter BP. Cancer incidence in persons with Fanconi anemia. Blood. 2003;101(3):822-826. 5. Rosenberg PS, Socie G, Alter BP, Gluckman E. Risk of head and neck squamous cell cancer and death in patients with Fanconi Anemia who did and did not receive transplants. Blood. 2005;105(1):67-73. 6. Rosenberg PS, Alter BP, Ebell W. Cancer risks in Fanconi anemia: findings from the

38

7.

8.

9.

10.

11.

12.

similar participation biases. The magnitude of the risks of adverse events showed similar trends in other cohorts (e.g., the German FA Registry, Israel and NAS), and in our earlier report. Our participants are seen by the same medical consultants who now have more than 15 years of experience of dealing with these rare disorders, and studied in the same laboratories. In summary, the NCI cohort is now sufficiently mature to provide more stable estimates of the types and risks of cancer in the major rare IBMFS. Continued follow-up will help identification of effective methods for cancer prevention and surveillance in these high-risk populations, and effective case management. Since our data were locked-in as of May, 2016, we became aware of nine additional cancers in nine patients, eight who had no prior cancer. Three in DC, no HCT: tongue, thyroid, and skin. Three in DC, post-HCT: tongue, pharynx, and bladder. Two in FA, no HCT: AML, and skin. One in FA, post-HCT: bladder, in a patient with a prior tongue cancer. Acknowledgments The authors would like to thank all the families who participated, physicians who referred patients, our colleagues in the Clinical Genetics Branch of the NCI, and the subspecialty physicians at the National Institutes of Health for their evaluations of the participants. Study management and nursing and genetic counseling expertise were provided through contract HHSN261201100018C with Westat Inc (Rockville, MD, USA); and we thank Sara Glashofer, Lisa Leathwood, Maureen Risch, and Ann Carr. Observed-to-expected cancer ratios were provided by Jeremy Miller at Information Management Systems (Silver Spring, MD, USA) through NIH contract HHSN26120110000I. Funding This work was supported in part by the intramural program of the National Institutes of Health and the National Cancer Institute.

German Fanconi Anemia Registry. Haematologica. 2008;93(4):511-517. Tamary H, Nishri D, Yacobovich J, el al. Frequency and natural history of inherited bone marrow failure syndromes: the Israeli Inherited Bone Marrow Failure Registry. Haematologica. 2010;95(8):1300-1307. Alter BP, Giri N, Savage SA, et al. Malignancies and survival patterns in the National Cancer Institute inherited bone marrow failure syndromes cohort study. Br J Haematol. 2010;150(2):179-188. Kutler DI, Singh B, Satagopan J, et al. A 20year perspective on the International Fanconi Anemia Registry (IFAR). Blood. 2003;101(4):1249-1256. Vulliamy TJ, Marrone A, Knight SW, Walne A, Mason PJ, Dokal I. Mutations in dyskeratosis congenita: their impact on telomere length and the diversity of clinical presentation. Blood. 2006;107(7):2680-2685. Alter BP, Baerlocher GM, Savage SA, et al. Very short telomere length by flow FISH identifies patients with Dyskeratosis Congenita. Blood. 2007;110(5):1439-1447. Alter BP, Giri N, Savage SA, Rosenberg PS. Cancer in dyskeratosis congenita. Blood. 2009;113(26):6549-6557.

13. Vlachos A, Rosenberg PS, Atsidaftos E, Alter BP, Lipton JM. Incidence of neoplasia in Diamond Blackfan anemia: a report from the Diamond Blackfan Anemia Registry. Blood. 2012;119(16):3815-3819. 14. Fargo JH, Kratz CP, Giri N, et al. Erythrocyte adenosine deaminase: diagnostic value for Diamond-Blackfan anaemia. Br J Haematol. 2013;160(4):547-554. 15. Mirabello L, Khincha PP, Ellis SR, et al. Novel and known ribosomal causes of DiamondBlackfan anaemia identified through comprehensive genomic characterisation. J Med Genet. 2017;54(6):417-425. 16. Myers KC, Bolyard AA, Otto B, et al. Variable clinical presentation of Shwachman-Diamond syndrome: update from the North American ShwachmanDiamond Syndrome Registry. J Pediatr. 2014;164(4):866-870. 17. Dhanraj S, Matveev A, Li H, et al. Biallelic mutations in DNAJC21 cause ShwachmanDiamond syndrome. Blood. 2017; 129(11):1557-1562. 18. Stepensky P, Chacon-Flores M, Kim KH, et al. Mutations in EFL1, an SBDS partner, are associated with infantile pancytopenia, exocrine pancreatic insufficiency and skele-

haematologica | 2018; 103(1)


Cancer in the NCI IBMFS cohort after 15 years

19.

20.

21.

22.

23.

tal anomalies in a Shwachman-Diamond like syndrome. J Med Genet. 2017;54(8):558566. Peffault de Latour R, Porcher R, Dalle JH, et al. Allogeneic hematopoietic stem cell transplantation in Fanconi anemia: the European Group for Blood and Marrow Transplantation experience. Blood. 2013; 122(26):4279-4286. Alter BP, Caruso JP, Drachtman RA, Uchida T, Velagaleti GV, Elghetany MT. Fanconi anemia: Myelodysplasia as a predictor of outcome. Cancer Genet Cytogenet. 2000;1 17(2):125-131. Alter BP, Rosenberg PS, Giri N, Baerlocher GM, Lansdorp PM, Savage SA. Telomere length is associated with disease severity and declines with age in dyskeratosis congenita. Haematologica. 2012;97(3):353-359. Ip WF, Dupuis A, Ellis L, et al. Serum pancreatic enzymes define the pancreatic phenotype in patients with Shwachman-Diamond syndrome. J Pediatr. 2002; 141(2):259-265. Kaplan EL, Meier P. Nonparametric estima-

haematologica | 2018; 103(1)

24.

25.

26.

27.

tion from incomplete observations. J Am Stat Assoc. 1958;53(282):457-481. Surveillance, Epidemiology, and End Results (SEER) Program (HYPERLINK " h t t p : / / w w w. s e e r. c a n c e r. g o v " www.seer.cancer.gov) SEER*Stat Database: Incidence - SEER 9 Regs Research Data, Nov 2014 Sub (1973-2012) <Katrina/Rita Population Adjustment> - Linked To County Attributes - Total U.S., 1969-2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, released April 2015, based on the November 2014 submission. Alter BP, Rosenberg PS, Brody LC. Clinical and molecular features associated with biallelic mutations in FANCD1/BRCA2. J Med Genet. 2007;44(1):1-9. Alter BP, Joenje H, Oostra AB, Pals G. Fanconi anemia: adult head and neck cancer and hematopoietic mosaicism. Arch Otolaryngol Head Neck Surg. 2005;131(7):635-639. Kutler DI, Wreesmann VB, Goberdhan A, et

28.

29.

30. 31.

al. Human papillomavirus DNA and p53 polymorphisms in squamous cell carcinomas from Fanconi anemia patients. J Natl Cancer Inst. 2003;95(22):1718-1721. van Zeeburg HJ, Snijders PJ, Wu T, et al. Clinical and molecular characteristics of squamous cell carcinomas from Fanconi anemia patients. J Natl Cancer Inst. 2008;100(22):1649-1653. Alter BP, Giri N, Savage SA, Quint WG, de Koning MN, Schiffman M. Squamous cell carcinomas in patients with Fanconi anemia and dyskeratosis congenita: A search for human papillomavirus. Int J Cancer. 2013;133(6):1513-1515. Curtis RE, Rowlings PA, Deeg J, et al. Solid cancers after bone marrow transplantation. N Engl J Med. 1997;336(13):897-904. Faivre L, Guardiola P, Lewis C, et al. Association of complementation group and mutation type with clinical outcome in fanconi anemia. European Fanconi Anemia Research Group. Blood. 2000;96(13):40644070.

39


ARTICLE

Myeloproliferative Disorders

Ferrata Storti Foundation

Expansion of EPOR-negative macrophages besides erythroblasts by elevated EPOR signaling in erythrocytosis mouse models

Jieyu Wang,1,2,* Yoshihiro Hayashi,1,* Asumi Yokota,1 Zefeng Xu,1,3 Yue Zhang,1,3 Rui Huang,1 Xiaomei Yan,1 Hongyun Liu,2 Liping Ma,2 Mohammad Azam,1 James P. Bridges,4 Jose A. Cancelas,1 Theodosia A. Kalfa,5 Xiuli An,6 Zhijian Xiao,3,# and Gang Huang1,3,#

Haematologica 2018 Volume 103(1):40-50

Divisions of Pathology and Experimental Hematology and Cancer Biology, Cincinnati Children’s Hospital Medical Center, OH, USA; 2Yat-sen Memorial Hospital, Sun Yat-sen University, Guangzhou, China; 3State Key Laboratory of Experimental Hematology, Institute of Hematology and Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, China; 4Division of Pulmonary Biology, Cincinnati Children’s Hospital Medical Center, OH, USA; 5Cancer and Blood Diseases Institute, Cincinnati Children's Hospital Medical Center, OH, USA and 6 Laboratory of Membrane Biology, New York Blood Center, New York, NY, USA 1

*JW and YH contributed equally to this work; #ZJX and GH contributed equally to this study as joint senior authors

ABSTRACT

A

Correspondence: gang.huang@cchmc.org

Received: May 17, 2017. Accepted: October 10, 2017. Pre-published: October 19, 2017. doi:10.3324/haematol.2017.172775 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/40

ctivated erythropoietin (EPO) receptor (EPOR) signaling causes erythrocytosis. The important role of macrophages for the erythroid expansion and differentiation process has been reported, both in baseline and stress erythropoiesis. However, the significance of EPOR signaling for regulation of macrophages contributing to erythropoiesis has not been fully understood. Here we show that EPOR signaling activation quickly expands both erythrocytes and macrophages in vivo in mouse models of primary and secondary erythrocytosis. To mimic the chimeric condition and expansion of the disease clone in the polycythemia vera patients, we combined Cre-inducible Jak2V617F/+ allele with LysM-Cre allele which expresses in mature myeloid cells and some of the HSC/Ps (LysM-Cre;Jak2V617F/+ mice). We also generated inducible EPO-mediated secondary erythrocytosis models using Alb-Cre, Rosa26loxP-stop-loxP-rtTA, and doxycycline inducible EPAS1-double point mutant (DPM) alleles (Alb-Cre;DPM mice). Both models developed a similar degree of erythrocytosis. Macrophages were also increased in both models without increase of major inflammatory cytokines and chemokines. EPO administration also quickly induced these macrophages in wild-type mice before observable erythrocytosis. These findings suggest that EPOR signaling activation could induce not only erythroid cell expansion, but also macrophages. Surprisingly, an in vivo genetic approach indicated that most of those macrophages do not express EPOR, but erythroid cells and macrophages contacted tightly with each other. Given the importance of the central macrophages as a niche for erythropoiesis, further elucidation of the EPOR signaling mediated-regulatory mechanisms underlying macrophage induction might reveal a potential therapeutic target for erythrocytosis.

©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.

40

Introduction The glycoprotein hormone erythropoietin (EPO)1,2 and EPO receptor (EPOR)3,4 signaling tightly controls red cell mass (RCM). EPOR signaling activates several downstream pathways through Janus kinase 2 (JAK2).5,6 Activation of EPOR signaling leads to erythrocytosis, which is defined as an absolute increase in RCM.7 EPOR primarily expresses on colony forming unit-erythroid (CFU-E) to basophilic erythroblasts.8,9 It has also been reported that EPOR is expressed in non-erythroid cells, such as macrophages;2 and non-hematopoietic tissues, such as the brain, kidney, and heart.14 However, the technical limitations for EPOR detection and the possibility of false-positive results in some cases have also been reported.14,15 Erythrocytosis is defined as primary when developed by an EPO-independent, haematologica | 2018; 103(1)


Expansion of EPOR–macrophages by EPOR signaling

erythroid cell-intrinsic mechanism due to constitutively activated EPOR signaling by gain-of-function EPOR or JAK2 mutations.7,16,17 In contrast, secondary erythrocytosis is due to an erythroid cell-extrinsic, EPO-dependent mechanism.7,16 EPAS1 (also known as HIF2A) is a master regulator of EPO gene expression.18 In normoxia, EPAS1 protein is immediately degraded through a variety of posttranslational modifications. Hypoxia, or a defect in the oxygensensing signaling pathway due to tumors or mutations, leads to excess EPO production and secondary erythrocytosis.16 Several mouse models mimicking primary or secondary erythrocytosis have been reported.19-21 However, there is no head-to-head comparison of the pathobiology between primary and secondary erythrocytosis in these models. To understand the dynamics of EPOR signalingmediated erythrocytosis and to establish new therapeutic strategies, animal models which present faithful clinically relevant phenotypes are needed. It has been proposed over several decades that a subset of macrophages participate in erythroblastic island (EIs) formation and support erythropoiesis by providing iron, promoting proliferation of erythroblasts, and being involved in the enucleation process.22-25 Recently, the critical role of macrophages, not only in baseline (homeostatic) erythropoiesis but also in stress erythropoiesis, has depletion of been demonstrated.26,27 Specific macrophages in the JAK2V617F-mediated polycythemia vera (PV) mouse models was shown to rescue the PV erythrocytosis phenotype.26,27 However, how EPOR signaling influences these macrophages in the erythrocytosis context is still not fully understood. In this study, we generated novel erythrocytosis mouse models which faithfully recapitulate both primary and secondary erythrocytosis, and we utilized them to address the effect and significance of EPOR signaling activation on erythroid cells and other cell types in vivo. We found that EPOR signaling activation could induce not only erythroid cell expansion but also macrophages.

Methods Mice The mice with the genotype of Cre-inducible Jak2V617F/+ and EPOR-Cre have been described previously.28,29 LysM-Cre, Alb-Cre, and Rosa26-LSL-rtTA-GFP mice were purchased from Jackson Laboratory. C57BL/6 mice were obtained from the Cincinnati Children's Hospital Medical Center (CCHMC) / Cancer and Blood Diseases Institute mouse core. The EPAS1 (also known as HIF2A) double point mutant (DPM) constructs (Pro531Ala and Asn847Ala) were generated by PCR mutagenesis using human EPAS1 cDNA as a template. The IRES sequence was PCR amplified from the pIRES2-EGFP vector (Clontech, Mountain View, CA, USA) and cloned onto the multiple cloning site of the (tetO)7CMV-bGH-poly(A) vector. Next, a FLAG tagged human ARNT cDNA was cloned 3’ of the IRES sequence and the EPAS1DPM cDNA was cloned 5’ of the IRES sequence. The constructs were linearized and microinjected individually into the pronucleus of fertilized eggs from FVB/N mice as previously described.30 All mice were backcrossed to C57BL/6 strain mice at least eight times. All animals were housed in the animal barrier facility at CCHMC. All animal studies were conducted according to an approved Institutional Animal Care and Use Committee protocol and federal regulations. haematologica | 2018; 103(1)

Flow cytometric analysis Flow cytometric analyses and cell sorting were performed with FACSCanto II or FACSAria II instruments (BD, San Jose, CA, USA). Peripheral blood (PB), bone marrow (BM), spleen (SP), and liver cells were immunostained using the following antibodies: CD45 (30-F11), CD71 (C2), CD44 (IM7) (BD), CD11b (M1/70), CD34 (RAM34), CD16/32 (93) (eBioscience, San Diego, CA, USA), CD115 (AFS98), Ly6G (RB6-8C5), Gr-1 (RB6-8C5), F4/80 (BM8), Ter119 (Ter-119), Sca-1 (D7) (Biolegend, San Diego, CA, USA), and c-Kit (2B8) (BD). For the staining of hematopoietic stem cells and progenitors, lineage marker cocktail containing antimouse CD3e (145-2C11), CD4 (RM4-5), CD8a (53-6.7), B220 (RA3-6B2), Ter119 (Ter-119), CD11b (M1/70), and Gr-1 (RB6-8C5) antibodies (all from BD) was used. Macrophage populations were defined as Gr-1low CD115int F4/80+ SSClow. Phosphate buffered saline (PBS) with 2% fetal bovine serum (FBS) was used as a FACS buffer. For some experiments, 2-5 mM ethylenediaminetetraacetic acid (EDTA) was added to the FACS buffer. To obtain the hematopoietic cells, SP and liver were mashed through a 100 μm cell strainer and resuspended in the FACS buffer. Data were analyzed with the FlowJo software (Tree Star, Ashland, OR, USA).

Statistical analysis Statistical analyses were performed using Student t-test or one/two-way ANOVA with multiple comparisons correction. Survival of mice was analyzed using the log rank test. P<0.05 was considered statistically significant. Information concerning quantitative RT-PCR, multiplex immunoassay, ELISA, histology, Wright Giemsa staining, and colony forming assay is available in the Online Supplementary Methods.

Results Pathophysiological modeling of primary and secondary erythrocytosis in novel mouse models To explore in detail the role of EPOR signaling in erythrocytosis development in vivo, we generated two mouse models of primary and secondary erythrocytosis. PV is the most common disease characterized by primary erythrocytosis, arising from acquired somatic JAK2 mutations in hematopoietic stem cells (HSCs).31-34 Interestingly, not all of the HSCs and progenitors in the BM of PV patients have JAK2 mutations; normal or other clones besides JAK2 mutant clones may also be present. The PV phenotype arises from HSC/Ps clones carrying JAK2 mutations over time. Although Vav1-Cre- or EIIA-Cremediated physiological PV models using this same conditional allele have been reported, the caveat is that, in these models, all of the hematopoietic cells have Jak2 mutation before birth.27,28 Mx1-Cre-mediated conditional allele has also been reported;35 again, in these models, all of the hematopoietic cells have Jak2 mutation. To faithfully recapitulate this chimeric condition and the natural clonal expansion process happening in PV patients with the JAK2 mutation, we used LysM-Cre allele and combined it with Cre-inducible JAK2V617F mutant knock-in heterozygous allele (JAK2V617F/+),28 then generated LysMCre;Jak2V617F/+ mice (LysM-Cre;Jak2V617F mice) (Figure 1A). LysM-Cre allele expresses not only in mature myelomonocytic cells, but also in some subsets of HSCs and progenitor cells.36 To confirm the expression pattern of LysM-Cre, we used Rosa26-loxP-Stop-loxP (Rosa26LSL)41


J. Wang et al.

A

C

B

D

E

G

F

H

Figure 1. In vivo pathophysiological modeling of primary and secondary erythrocytosis. (A) Primary erythrocytosis model using LysM-Cre allele, and Cre-inducible Jak2V617F/+ knock-in allele. (B) PCR of DNA of sorted bone marrow (BM) or peripheral blood (PB) cells from Jak2V617F-mutant knock-in (KI) mice with or without LysM-Cre. (C and D) Inducible secondary erythrocytosis model using indicated alleles. Schematic for EPAS1 double point mutant (DPM) and doxycycline-inducible hepatocyte-specific induction of the DPM protein is shown in (D). (E) Appearance of hind paw of the indicated mice. (F) Spleen size in the indicated mice (n=8 each). (G) White blood cell (WBC) counts and platelet (Plts) counts in PB from the indicated mice at 2-3 months of age (Control: n=7; LysM-Cre;Jak2V617F mice; n=9) or 2 months after HIF-2Îą induction (n=9). Data are meanÂąStandard Deviation (s.d.). (H) Survival of LysM-Cre; Jak2V617F mice (n=9) and doxycycline-administrated AlbCre;DPM mice (n=9). *P<0.05; **P<0.01; ****P<0.0001; NS: not significant [one-way ANOVA with multiple comparisons correction (G) or two-way ANOVA with multiple comparisons correction (F), and log rank test (H)].

42

haematologica | 2018; 103(1)


Expansion of EPOR–macrophages by EPOR signaling

eGFP reporter allele (Online Supplementary Figure S1A). GFP expression was observed in the subsets of lineage marker- Sca-1+ c-Kit+ cells (LSKs), common-myeloid progenitors (CMPs), granulocyte-macrophage progenitors (GMPs), and megakaryocyte-erythrocyte progenitors (MEPs), as well as in mature myeloid cells (granulocytes and monocytes) (Online Supplementary Figure S1B). We did not find GFP expression in the mature lymphoid lineage (Online Supplementary Figure S1B). These results are consistent with previous reports.36 To confirm LysM-Cre mediated excision of wild-type exon and inversion of Jak2V617F/+ mutant allele in LysM-Cre;Jak2V617F mice, we sorted BM (LSKs, CMPs, GMPs, and MEPs) and PB (monocytes, neutrophils, and lymphocytes) cells from Jak2V617F/+ mice with or without LysM-Cre. We then performed PCR

A

E

G

B

using the primers previously described.28 Consistent with the GFP expression pattern in the cells from LysMCre;GFP reporter mice (Online Supplementary Figure S1B), we found both bands for floxed allele and inversion of Jak2V617F/+ mutant in LSKs, CMPs, GMPs, MEPs, monocytes, and neutrophils, but not in lymphocytes, from LysM-Cre;Jak2V617F/+ mice (Figure 1B). Interestingly, in MEPs from LysM-Cre;Jak2V617F mice, we found a weak band for the floxed allele and a strong band for Cre-mediated excision (Figure 1B), suggesting the expansion of the Jak2V617F cells within the MEPs. Indeed, we found increased GFP+ cells in MEPs along with LSKs, CMPs, and GMPs in LysM-Cre;GFP/Jak2V617F reporter PV mice (Online Supplementary Figure S1C and D). BFU-E and CFU-E were

C

D

F

H

Figure 2. The novel mouse models LysM-Cre;Jak2V617F and Alb-Cre;DPM demonstrate a similar degree of erythrocytosis. (A and B) EPO mRNA expression in hepatocytes (A) and kidney (B) from indicated mice. Data are mean±Standard Deviation (s.d.) from duplicate samples. Results are representative of two independent experiments. (C) Serum erythropoietin (EPO) levels in the indicated mice (n=3-6 each). (D) Red blood cell (RBC) counts, hematocrit (Ht), and hemoglobin (Hb) counts in peripheral blood (PB) from the indicated mice at 2-3 months of age (Control, n=7; LysM-Cre;Jak2V617F mice, n=9) or two months after HIF-2α induction (n=9). Data are mean±Standard Deviation (s.d.). (E-H) Flow cytometric analysis of erythroid lineage cells in bone marrow (BM) and spleen (SP) from the indicated mice. Representative plots of erythropoiesis profiles (CD44 and FSC) in BM and SP CD45-, Ter119dim/+ population (E), and absolute number of erythroblasts (F) and each fraction of erythroblasts/mature erythroid cells in BM and SP (G and H) from indicated mice are shown (n=4 each). I (Pro): proerythroblasts; II (Baso): basophilic erythroblasts; III (Poly): polychromatic erythroblasts; IV (Ortho): orthochromatic erythroblasts; V (Retic): reticulocytes; VI (RBC): red blood cells. Data are mean±s.d. *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001; NS: not significant (one-way ANOVA with multiple comparisons correction (D) or two-way ANOVA with multiple comparisons correction (F-H)).

haematologica | 2018; 103(1)

43


J. Wang et al.

also increased in LysM-Cre;Jak2V617F mice (Online Supplementary Figure S1E). Next, in order to model the EPO-induced secondary erythrocytosis, we generated a liver-specific inducible EPAS1 double-point-mutant (DPM) (P531A and N847A) transgenic mouse (Figure 1C and D). Besides germline mutations, other acquired conditions, such as hypoxia or tumors, could also cause excessive EPO-mediated secondary erythrocytosis at various ages. Thus, we decided to generate a mouse model in which the excessive EPO condition can be induced. Liver is a major source of EPO during embryonic and early postnatal development in mice.18 In adult erythropoiesis, although the kidneys become the major source of EPO, the hepatocytes still retain the capacity for EPO production.37 Various gain-of-function EPAS1 mutations, near or at the primary hydroxylation site proline-531 (P531), have been identified.16,18,38 The EPAS1-DPM allele has mutations in P531 site, which is an important site for the normal EPAS1 protein degradation in normoxia, and in the asparagine-847 (N847) site, which is a critical site for transcriptional inactivation in normoxia (Figure 2B). Using hepatocyte-specific Alb-Cre allele and Rosa26LSL reverse-tetracycline-controlled transactivator (rtTA) driver (Rosa26LSL-rtTA), stable and constitutively active EPAS1 protein was induced in the adult hepatocytes in a doxycycline-dependent manner (Figure 1D). We compared the results from these mice with the results from wild-type control mice. Both LysM-Cre;Jak2V617F and Alb-Cre;DPM mice presented observable redness in the mouth and front and hind paws (Figure 1E). LysM-Cre;Jak2V617F mice showed more severe splenomegaly than Alb-Cre;DPM mice (Figure 1F). We also determined red and white pulps of SP in these mice. The white pulp was diminished and the normal SP architecture was destroyed in both LysM-Cre;Jak2V617F and Alb-Cre;DPM mice in comparison with control mice (Online Supplementary Figure S2). White blood cell (WBC)

A

and platelet (Plts) counts in the PB from LysMCre;Jak2V617F mice were significantly higher than those in the PB from wild-type and Alb-Cre;DPM mice (Figure 1G). The median survival of the LysM-Cre;Jak2V617F mice was 123 days, while the Alb-Cre;DPM mice survived longer (Figure 1H). Taken together, both our mouse models could quickly develop erythrocytosis, recapitulating the pathophysiological condition of human primary and secondary erythrocytosis.

Both primary and secondary erythrocytosis mouse models develop similar degree of erythrocytosis Serum EPO level is a basic parameter used for differential diagnosis of erythrocytosis. Thus, we confirmed the EPO expression pattern in our erythrocytosis models. The expression levels of EPO mRNA in the hepatocytes from Alb-Cre;Rosa26LSLrtTA;EPAS1-DPM mice (Alb-Cre;DPM mice) were significantly up-regulated after doxycycline administration (Figure 2A) while the expression of EPO mRNA in the hepatocytes from LysM-Cre;Jak2V617F mice and wild-type control mice was undetectable (Figure 2A). The expression levels of EPO mRNA in the kidney cells from LysM-Cre;Jak2V617F and Alb-Cre;DPM mice were down-regulated compared to wild-type mice (Figure 2B). Serum EPO level in Alb-Cre;DPM mice was dramatically increased compared to that in wild-type mice (Figure 2C), while the EPO level in LysM-Cre;Jak2V617F mice was decreased (Figure 2C), consistent with observations in the patients with secondary or primary erythrocytosis, respectively. Both models quickly developed a similar degree of erythrocytosis within three months after birth or doxycycline administration, respectively (Figure 2D). We next determined the erythroid expansion and differentiation in our erythrocytosis models according to the Ter119/CD44/cellsize-based gating strategy39 (Figure 2E-H). Total erythroid population in BM and SP from both models was signifi-

B

Figure 3. Alteration of macrophage population in the erythrocytosis mice. (A and B) Flow cytometric analysis of bone marrow (BM)/spleen (SP) macrophage population in the indicated mice (n=5 each). The absolute number and frequency of total macrophages from the indicated mice are shown. Data are meanÂąStandard Deviation (s.d.). *P<0.05; **P<0.01; ***P<0.001; NS: not significant (two-way ANOVA with multiple comparisons correction).

44

haematologica | 2018; 103(1)


Expansion of EPOR–macrophages by EPOR signaling

cantly increased compared to the control (Figure 2E-H). In BM from both models, mature RBCs, but not erythroblasts, were increased compared to the control (Figure 2F and G). We found that splenomegaly in LysMCre;Jak2V617F mice was more obvious than that in AlbCre;DPM mice (Figure 1F). Concordant with splenomegaly, expansion of erythroblasts and reticulo-

A

cytes in SP from LysM-Cre;Jak2V617F mice was more prominent compared to Alb-Cre;DPM mice (Figure 2F and H). Extramedullary erythropoiesis could occur in liver under the stress conditions. Thus, we also determined erythropoiesis in liver (Online Supplementary Figure S3A). In control mice, we found only a small CD71+CD45– population, which could contain erythroblasts in the liver. On

B

C

Figure 4. Inflammatory cytokine and chemokine profile in erythrocytosis mice. (A and B) Multiplex immunoassay using plasma samples from the indicated mice (n=4 each). Heat map showing log2-fold changes in concentration of 32 cytokines/chemokines normalized by mean value of control samples. Gray box indicates undetectable. Concentration of individual cytokines/chemokines are shown in (B). (C) The expression of Il1b, Il6, and Il12 mRNA in sorted bone marrow (BM) macrophages from the indicated mice. Data are mean±Standard Deviation (s.d.). *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001; NS: not significant (two-way ANOVA with multiple comparisons correction).

haematologica | 2018; 103(1)

45


J. Wang et al.

the other hand, the CD71+CD45– population was significantly increased in Alb-Cre;DPM mice compared with control (Online Supplementary Figure S3B and C). This trend was also observed in LysM-Cre;Jak2V617F mice. However, there was no significant difference in the liver erythropoiesis between LysM-Cre;Jak2V617F and AlbCre;DPM mice (Online Supplementary Figure S3B and C). Taken together, these results suggest that the sensitivity of, and response to the EPO-EPOR signaling pathway in erythroid cells and the niche environment could differ depending on whether activation is cell-intrinsic or extrinsic.

Macrophages are increased in both erythrocytosis models Macrophages are an important erythroid cell-extrinsic component for baseline and stress erythropoiesis.24,26 Thus, we next analyzed the macrophage population (Gr1low CD115int F4/80+ SSClow )40 in our models. Although there was no change in the absolute number and frequency of total BM macrophages (Figure 3A), the absolute number and frequencies of total SP macrophages from both LysMCre;Jak2V617F mice and Alb-Cre;DPM mice were significantly increased compared to those in the wild-type mice

(Figure 3B). These results suggest that macrophages are increased in the pathogenesis of erythrocytosis.

Inflammatory cytokines and chemokines are not elevated in both erythrocytosis models Cytokines and chemokines play an important role in regulating dynamics of macrophages. To characterize the profile of cytokines and chemokines, we performed multiplex immunoassay using plasma samples from our erythrocytosis models (Figure 4A). Surprisingly, although the levels of some inflammatory cytokines/chemokines, such as Cxcl9 and Cxcl10, were increased in the plasma from LysMCre;Jak2V617F mice, the levels of most inflammatory cytokines/chemokines, such as Ccl2, Ccl5, Cxcl1, Cxcl5, Ccl2, Il1b, Il6, Il12, Tnf were not increased in the plasma from LysM-Cre;Jak2V617F mice and Alb-Cre;DPM mice compared to the control (Figure 4A and B). We also measured the gene expression levels of several inflammatory cytokines in sorted BM macrophages obtained from our erythrocytosis mice. The expression levels of Il1b, Il6, and Il12 mRNA were significantly decreased in the erythrocytosis mice (Figure 4C). These results indicate that it is unlikely that an increase in macrophages in our erythrocytosis models is the result of inflammatory signaling activation.

A

C

B

D E

46

Figure 5. The effect of erythropoietin (EPO) injection on macrophage populations. (A) Recombinant human EPO (rhEPO) or phosphate buffered saline (PBS) were daily injected intraperitoneally (ip) into wild-type mice for seven days (n=4 each). (B) White blood cell (WBC) counts, red blood cell (RBC) counts, hematocrit (Ht), hemoglobin (Hb), and platelet (Plts) counts in peripheral blood (PB) from the control mice (Vehicle) and rhEPO injected mice (rhEPO) on day 4 and day 8. Data are mean±Standard Deviation (s.d.). (C) Spleen weight in the indicated mice. Data are mean±s.d. (D and E) Flow cytometric analysis of bone marrow (BM)/spleen (SP) macrophages from the indicated mice. Data are mean±s.d. *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001; NS: not significant (Student t-test).

haematologica | 2018; 103(1)


Expansion of EPORâ&#x20AC;&#x201C;macrophages by EPOR signaling

BM/SP macrophages increase fast in number in response to EPO in an EPO-injection model of erythrocytosis The critical role of subset of macrophages has been reported in erythrocytosis.27 We found that macrophages are increased in our primary and secondary erythrocytosis mouse models. To confirm whether EPOR signaling could induce macrophages, we injected intraperitoneally recombinant human EPO (rhEPO) into wild-type mice daily for seven days (50 U/day) (Figure 5A). On day 4, erythrocytosis was not yet prominent (Figure 5B), while splenomegaly was significant (Figure 5C). On day 8, the rhEPO injected mice had developed observable erythrocy-

A

tosis (Figure 5B). On day 4, after three days of rhEPO injections, BM macrophages in rhEPO injected mice were also significantly more increased than those from vehicleinjected control mice (Figure 5D and E). The increase in SP macrophages in rhEPO injected mice was prominent on day 8 but not on day 4 (Figure 5F and G). These results suggest that BM/SP macrophages are immediately induced in response to EPO-EPOR signaling.

Macrophages may not express EPOR but tightly interact with erythroblasts We found that EPO injection quickly increased macrophages as well as erythroid cells. EPOR expression

B

C D

E

F

G

Figure 6. Most of the macrophages do not express EPOR. (A) GFP-reporter mice using EPOR-Cre knock-in allele and Rosa26/loxP-Stop-loxP (LSL)/eGFP allele. (B) Expression pattern of EPOR-Cre (GFP) in bone marrow (BM) single-cell-gated macrophage fraction. (C) CD71 and Ter119 expression pattern in the single-cell-gated BM macrophage population in (B). Expression pattern of EPOR-Cre (GFP) in the individual fractions are shown. (D) Wright Giemsa staining of single-cell-sorted macrophages in the indicated fraction defined in (C). Samples were prepared with normal FACS buffer. (E) Flow cytometric analysis of the same BM sample with different preparation (normal FACS buffer with or without EDTA). Representative plots are shown. (F) Wright Giemsa staining of single-cell-sorted macrophages in the indicated fraction defined in (E) with the FACS buffer containing 2 mM EDTA. (G) Summary of this study. EPOR signaling activation expands erythroid and macrophage populations.

haematologica | 2018; 103(1)

47


J. Wang et al.

is required for direct response to EPO. It has been reported that EPOR expresses not only on erythroid progenitors/precursors but also on macrophages.10-13 To confirm the distribution of EPOR expression in hematopoietic cells using a genetic approach, we used EPOR-Cre knock-in allele and Rosa26LSL-GFP reporter allele (EPOR-Cre;GFP reporter mice) (Figure 6A). As expected, in erythroid lineage we found GFP expression on the surface of CD71+ Ter119-/dim cells (pro-erythroblasts) and CD71+ Ter119+ FSChigh cells (basophilic erythroblasts) (Online Supplementary Figure S3). As previously suggested, we found GFP expression in a single-cell-gated macrophage fraction in EPOR-Cre;GFP reporter mice (Figure 6B). However, surprisingly, we found CD71+ and/or Ter119+ populations in this single-cell-gated macrophage fraction (Figure 6C), and a majority of GFP+ cells belong to this CD71+ and/or Ter119+ subset (Figure 6C). FACS sorting revealed that those GFP+ cells are not just macrophages, but macrophages with erythroblasts adhered to them (Figure 6D). On the other hand, most of the CD71– Ter119– macrophages were single macrophages, while some macrophages bind with a nucleus which is likely to be extruded nucleus ('pyrenocyte') (Figure 6D). We confirmed these findings by analyzing the same samples with a buffer containing EDTA. By adding 2-5 mM EDTA, we found reduced CD71+ Ter119-/+ population in the macrophage fraction while some macrophages still retained bound CD71+ Ter119+ erythroblasts (Figure 6E and F). These results suggest that macrophages may not present EPOR on the cell surface, but that they tightly connect with erythroblasts that express EPOR on their surface.

Discussion In this study, we generated novel mouse models of primary and secondary erythrocytosis. Using these, we demonstrated that activation of EPOR signaling could alter not only erythrocytes but also macrophages (Figure 6H). LysM-Cre;Jak2V617F mice can recapitulate the chimeric condition in BM and a clonal expansion of mutant clone with JAK2V617F. Since conventional stabilization of EPAS1 has been reported to result in short-life phenotype (only 6-8 weeks),21 our inducible EPO expression model (Alb-Cre;DPM mice) has an obvious advantage in that an excessive level of EPO can be induced in adult mice of various ages. Thus, these are by far the most physiologically relevant primary and secondary erythrocytosis models. Both erythrocytes and macrophages co-operate to achieve an absolute increase in RCM. Interestingly, Alb-Cre;DPM adult mice had longer survival, while LysM-Cre;Jak2V617F mice quickly developed a lethal phenotype of PV. High WBC and/or Plt counts in the PV patients have been considered risk factors for life-threatening thrombosis.41 Indeed, WBC and Plt counts in LysM-Cre;Jak2V617F mice were higher than those in Alb-Cre;DPM mice. It has also been suggested that the JAK2V617F mutation promotes adhesion of eythrocytes through an EPOR-independent pathway.42 This EPORindependent pre-thrombotic effect could also contribute to shorter survival in LysM-Cre;Jak2V617F mice. Major inflammatory cytokines and chemokines were not increased in Alb-Cre;DPM mice. Consistent with previous reports,11,13 this indicates that EPOR signaling activa48

tion by itself may not induce systemic inflammation. A JAK2V617F mutation may activate not only EPOR signaling in erythrocytes but also a variety of cytokine receptor signaling which also uses JAK2 for signaling in other hematopoietic cells.43,44 It has been reported that retroviral overexpression of JAK2-mutant in the bone marrow transplantation mouse models results in elevated Tnfa expression in the serum.45 TNFα has also been reported to be increased and to promote the expansion of JAK2-mutant clones in the myeloproliferative neoplasm.46 However, the expression levels of Tnfa in the plasma from LysMCre;Jak2V617F mice were decreased. The pathophysiological expression level of JAK2-mutant (but not retroviral overexpression) by itself might not induce a systemic inflammatory cytokinemia. Given that a variety of mutations, such as mutations in TET2, ASXL1, DNMT3A, and EZH2, have recently been identified in PV patients as well as JAK2 mutation.47 The combined effect of JAK2 mutation and those other mutations may create an inflammatory environment, and further accelerate the expansion of the disease clone and full blown disease progression in PV patients. Despite the fact that a similar degree of erythrocytosis was observed in both LysM-Cre;Jak2V617F and Alb-Cre;DPM mice, the number of erythroblasts in the SP of LysM-Cre;Jak2V617F was significantly higher than that in Alb-Cre;DPM mice. However, there was no significant difference in the BM and liver erythropoiesis between those two models. Since JAK2 mutation could activate a variety of cytokine receptor signaling, JAK2 activationmediated non-EPOR signaling in hematopoietic cells may intrinsically or extrinsically affect the dynamics of mature erythroid cell production, maintenance, or life-span. Indeed, some inflammatory chemokines, such as Cxcl9 and Cxcl10, were actually increased in the serum from LysM-Cre;Jak2V617F mice. By a genetic approach, using EPOR-Cre knock-in and GFP reporter allele, we found restricted EPOR expression on limited stages of erythroid cells (from CFU-E to asophilic-erythroblasts), which is consistent with previous findings.8,9 On the other hand, we did not identify EPOR expression on most of the macrophages, despite the fact that, in our study, the macrophage population was quickly expanded in response to EPO injection. Instead, we found a tight interaction between macrophages and EPORexpressing erythroid cells. This may give the mistaken impression that macrophages also express EPOR. However, we still cannot exclude the possibility that a subpopulation of macrophages, which are the central macrophages binding to erythroblasts, express EPOR on their cell surface. In rhEPO injected wild-type mice, expansion of macrophages first occurred in BM. Although several macrophage populations reside in SP, the expansion of SP macrophages in response to EPO was delayed. Since BM is the main site of erythropoiesis in steady state, most of the early erythroblasts which express EPOR are to be found in BM but not in SP. Thus, this delay of SP macrophage expansion after EPO injection may also suggest: 1) a lack of proper macrophages in SP at steady state; and/or 2) that the expansion of macrophages in SP in response to EPO needs EPOR expressing erythroblasts which first expand in BM and then migrate from BM to SP. EPOR signaling may alter the macrophages into erythropoiesis-supportive macrophages via inter-cellular signaling by erythroid cells. Interestingly, erythroblasts are known haematologica | 2018; 103(1)


Expansion of EPOR–macrophages by EPOR signaling

to secrete erythroid factors, including GDF15, GDF11, TWSG1, and ERFE, which suppress hepcidin and increase iron availability for hemoglobin synthesis.48 Studies so far have focused on their effect on hepcidin production in hepatocytes. Central macrophages in the erythroblastic islands are where the cells provide the erythroblasts with iron. Thus, these potential erythroid factors may influence macrophages via a paracrine mechanism. Future study is needed to understand the details of growth factors, cytokines, or chemokine signaling that may play an important role in both steady stage and stress erythropoiesis. Blocking this erythroblast-macrophage communication through direct cell-cell contact or through indirect interactions could be an alternative therapeutic target for both primary and secondary erythrocytosis. Our current study sheds light on the significance of EPOR signaling in erythroid cells and macrophages during erythrocytosis. Without expressing EPOR on their cell surface, macrophages expanded in response to EPO. Further elucidation of the mechanism by which EPO/EPOR signaling could expand macrophages and affect their function will: 1) provide us with a clue for controlling those macrophages which may form erythroblastic islands in vivo; and 2) lead to identification of novel therapeutic targets, beside JAK2 inhibition, for erythrocytosis. Our new mouse models are also useful in vivo tools to test the

References 1. Miyake T, Kung CK, Goldwasser E. Purification of human erythropoietin. J Biol Chem. 1977;252(15):5558-5564. 2. Jacobs K, Shoemaker C, Rudersdorf R, et al. Isolation and characterization of genomic and cDNA clones of human erythropoietin. Nature. 1985;313(6005):806-810. 3. D'Andrea AD, Lodish HF, Wong GG. Expression cloning of the murine erythropoietin receptor. Cell. 1989;57(2):277-285. 4. Constantinescu SN, Wu H, Liu X, Beyer W, Fallon A, Lodish HF. The anemic Friend virus gp55 envelope protein induces erythroid differentiation in fetal liver colonyforming units-erythroid. Blood. 1998; 91(4):1163-1172. 5. Witthuhn BA, Quelle FW, Silvennoinen O, et al. JAK2 associates with the erythropoietin receptor and is tyrosine phosphorylated and activated following stimulation with erythropoietin. Cell. 1993;74(2):227-236. 6. Livnah O, Johnson DL, Stura EA, et al. An antagonist peptide-EPO receptor complex suggests that receptor dimerization is not sufficient for activation. Nat Struct Biol. 1998;5(11):993-1004. 7. Patnaik MM, Tefferi A. The complete evaluation of erythrocytosis: congenital and acquired. Leukemia. 2009;23(5):834-844. 8. Gregory CJ. Erythropoietin sensitivity as a differentiation marker in the hemopoietic system: studies of three erythropoietic colony responses in culture. J Cell Physiol. 1976;89(2):289-301. 9. Broudy VC, Lin N, Brice M, Nakamoto B, Papayannopoulou T. Erythropoietin receptor characteristics on primary human erythroid cells. Blood. 1991;77(12):2583-2590.

haematologica | 2018; 103(1)

efficacy of new therapeutic options on cellular and molecular components involved in the pathogenesis of erythrocytosis. Acknowledgments The authors would like to thank G. Freudiger (Cincinnati Children’s Hospital Medical Center) for assisting us in the experiment. Funding This work was supported by the Cincinnati Children's Hospital Research Foundation (to GH), the Leukemia Research Foundation (to GH), the OCRA (to GH), National Institutes of Health (NIH) (R01DK105014 to GH), National Natural Science Funds of China (n. 81300392 to JW, n. 81370611 to ZFX, n. 81470338 to YZ, n. 81470297, n. 81770129 to GH, and n. 81530008, n. 81470295 to ZJX), Tianjin science and technology projects (13CYBJC42400 to YZ), CCHMC Research and Development Project through the Cystic Fibrosis Foundation (to JPB). We would like to acknowledge the assistance of the Research Flow Cytometry Core in the Division of Rheumatology at Cincinnati Children’s Hospital Medical Center for Luminex assay. All flow cytometric data were acquired using equipment maintained by the Research Flow Cytometry Core in the Division of Rheumatology at Cincinnati Children’s Hospital Medical Center.

10. Lifshitz L, Tabak G, Gassmann M, Mittelman M, Neumann D. Macrophages as novel target cells for erythropoietin. Haematologica. 2010;95(11):1823-1831. 11. Nairz M, Schroll A, Moschen AR, et al. Erythropoietin contrastingly affects bacterial infection and experimental colitis by inhibiting nuclear factor-kappaB-inducible immune pathways. Immunity. 2011;34(1):61-74. 12. Nairz M, Sonnweber T, Schroll A, Theurl I, Weiss G. The pleiotropic effects of erythropoietin in infection and inflammation. Microbes Infect. 2012;14(3):238-246. 13. Luo B, Gan W, Liu Z, et al. Erythropoeitin Signaling in Macrophages Promotes Dying Cell Clearance and Immune Tolerance. Immunity. 2016;44(2):287-302. 14. Elliott S, Busse L, McCaffery I, et al. Identification of a sensitive anti-erythropoietin receptor monoclonal antibody allows detection of low levels of EpoR in cells. J Immunol Methods. 2010;352(12):126-139. 15. Elliott S, Sinclair A, Collins H, Rice L, Jelkmann W. Progress in detecting cell-surface protein receptors: the erythropoietin receptor example. Ann Hematol. 2014; 93(2):181-192. 16. Lee FS, Percy MJ. The HIF pathway and erythrocytosis. Annu Rev Pathol. 2011; 6:165-192. 17. Feng G, Zhang T, Liu J, et al. MLF1IP promotes normal erythroid proliferation and is involved in the pathogenesis of polycythemia vera. FEBS Lett. 2017;591(5):760773. 18. Franke K, Gassmann M, Wielockx B. Erythrocytosis: the HIF pathway in control. Blood. 2013;122(7):1122-1128.

19. Li J, Kent DG, Chen E, Green AR. Mouse models of myeloproliferative neoplasms: JAK of all grades. Dis Model Mech. 2011;4(3):311-317. 20. Tan Q, Kerestes H, Percy MJ, et al. Erythrocytosis and pulmonary hypertension in a mouse model of human HIF2A gain of function mutation. J Biol Chem. 2013;288(24):17134-17144. 21. Kim WY, Safran M, Buckley MR, et al. Failure to prolyl hydroxylate hypoxiainducible factor alpha phenocopies VHL inactivation in vivo. EMBO J. 2006;25(19):4650-4662. 22. Chasis JA, Mohandas N. Erythroblastic islands: niches for erythropoiesis. Blood. 2008;112(3):470-478. 23. Rhodes MM, Kopsombut P, Bondurant MC, Price JO, Koury MJ. Adherence to macrophages in erythroblastic islands enhances erythroblast proliferation and increases erythrocyte production by a different mechanism than erythropoietin. Blood. 2008;111(3):1700-1708. 24. Hom J, Dulmovits BM, Mohandas N, Blanc L. The erythroblastic island as an emerging paradigm in the anemia of inflammation. Immunol Res. 2015;63(1-3):75-89. 25. Giger KM, Kalfa TA. Phylogenetic and Ontogenetic View of Erythroblastic Islands. Biomed Res Int. 2015;2015(873628. 26. Chow A, Huggins M, Ahmed J, et al. CD169(+) macrophages provide a niche promoting erythropoiesis under homeostasis and stress. Nat Med. 2013;19(4):429-436. 27. Ramos P, Casu C, Gardenghi S, et al. Macrophages support pathological erythropoiesis in polycythemia vera and beta-thalassemia. Nat Med. 2013;19(4):437-445. 28. Mullally A, Lane SW, Ball B, et al.

49


J. Wang et al.

29.

30.

31.

32.

33.

34.

50

Physiological Jak2V617F expression causes a lethal myeloproliferative neoplasm with differential effects on hematopoietic stem and progenitor cells. Cancer Cell. 2010;17(6):584-596. Heinrich AC, Pelanda R, Klingmuller U. A mouse model for visualization and conditional mutations in the erythroid lineage. Blood. 2004;104(3):659-666. Bridges JP, Lin S, Ikegami M, Shannon JM. Conditional hypoxia inducible factor1alpha induction in embryonic pulmonary epithelium impairs maturation and augments lymphangiogenesis. Dev Biol. 2012;362(1):24-41. Baxter EJ, Scott LM, Campbell PJ, et al. Acquired mutation of the tyrosine kinase JAK2 in human myeloproliferative disorders. Lancet. 2005;365(9464):1054-1061. James C, Ugo V, Le Couedic JP, et al. A unique clonal JAK2 mutation leading to constitutive signalling causes polycythaemia vera. Nature. 2005; 434(7037):1144-1148. Kralovics R, Passamonti F, Buser AS, et al. A gain-of-function mutation of JAK2 in myeloproliferative disorders. N Engl J Med. 2005;352(17):1779-1790. Levine RL, Wadleigh M, Cools J, et al. Activating mutation in the tyrosine kinase JAK2 in polycythemia vera, essential thrombocythemia, and myeloid metaplasia with myelofibrosis. Cancer Cell. 2005; 7(4):387-397.

35. Akada H, Yan D, Zou H, Fiering S, Hutchison RE, Mohi MG. Conditional expression of heterozygous or homozygous Jak2V617F from its endogenous promoter induces a polycythemia vera-like disease. Blood. 2010;115(17):3589-3597. 36. Ye M, Iwasaki H, Laiosa CV, et al. Hematopoietic stem cells expressing the myeloid lysozyme gene retain long-term, multilineage repopulation potential. Immunity. 2003;19(5):689-699. 37. Fandrey J. Oxygen-dependent and tissuespecific regulation of erythropoietin gene expression. Am J Physiol Regul Integr Comp Physiol. 2004;286(6):R977-988. 38. Comino-Mendez I, de Cubas AA, Bernal C, et al. Tumoral EPAS1 (HIF2A) mutations explain sporadic pheochromocytoma and paraganglioma in the absence of erythrocytosis. Hum Mol Genet. 2013;22(11):21692176. 39. Liu J, Zhang J, Ginzburg Y, et al. Quantitative analysis of murine terminal erythroid differentiation in vivo: novel method to study normal and disordered erythropoiesis. Blood. 2013;121(8):e43-49. 40. Chow A, Lucas D, Hidalgo A, et al. Bone marrow CD169+ macrophages promote the retention of hematopoietic stem and progenitor cells in the mesenchymal stem cell niche. J Exp Med. 2011;208(2):261-271. 41. Barbui T, Masciulli A, Marfisi MR, et al. White blood cell counts and thrombosis in polycythemia vera: a subanalysis of the

42.

43.

44.

45.

46.

47.

48.

CYTO-PV study. Blood. 2015;126(4):560561. De Grandis M, Cambot M, Wautier MP, et al. JAK2V617F activates Lu/BCAM-mediated red cell adhesion in polycythemia vera through an EpoR-independent Rap1/Akt pathway. Blood. 2013;121(4):658-665. Tyner JW, Bumm TG, Deininger J, et al. CYT387, a novel JAK2 inhibitor, induces hematologic responses and normalizes inflammatory cytokines in murine myeloproliferative neoplasms. Blood. 2010; 115(25):5232-5240. Quintas-Cardama A, Kantarjian H, Cortes J, Verstovsek S. Janus kinase inhibitors for the treatment of myeloproliferative neoplasias and beyond. Nat Rev Drug Discov. 2011;10(2):127-140. Bumm TG, Elsea C, Corbin AS, et al. Characterization of murine JAK2V617Fpositive myeloproliferative disease. Cancer Res. 2006;66(23):11156-11165. Fleischman AG, Aichberger KJ, Luty SB, et al. TNFalpha facilitates clonal expansion of JAK2V617F positive cells in myeloproliferative neoplasms. Blood. 2011;118(24):63926398. Lundberg P, Karow A, Nienhold R, et al. Clonal evolution and clinical correlates of somatic mutations in myeloproliferative neoplasms. Blood. 2014;123(14):2220-2228. Kim A, Nemeth E. New insights into iron regulation and erythropoiesis. Curr Opin Hematol. 2015;22(3):199-205.

haematologica | 2018; 103(1)


ARTICLE

Myeloproliferative Disorders

Treatment of essential thrombocythemia in Europe: a prospective long-term observational study of 3649 high-risk patients in the Evaluation of Anagrelide Efficacy and Long-term Safety study Gunnar Birgegård,1 Carlos Besses,2 Martin Griesshammer,3 Luigi Gugliotta,4 Claire N. Harrison,5 Mohamed Hamdani,6 Jingyang Wu,6 Heinrich Achenbach7 and Jean-Jacques Kiladjian8

Department of Haematology, Institute for Medical Sciences, Uppsala University, Sweden; Department of Hematology, Hospital del Mar-IMIM, Barcelona, Spain; 3Hematology and Oncology, Johannes Wesling Medical Center, Minden, Germany; 4Department of Haematology, ‘L e A Seragnoli’, Sant'Orsola-Malpighi Hospital, Bologna, Italy; 5Department of Haematology, Guy’s and St Thomas’ NHS Foundation Trust, London, UK; 6Global Biometrics, Shire Pharmaceuticals, Lexington, MA, USA; 7Research & Development, Shire GmbH, Zug, Switzerland and 8APHP, Hopital Saint-Louis, Paris, France 1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(1):51-60

2

ABSTRACT

E

valuation of Anagrelide (Xagrid®) Efficacy and Long-term Safety, a phase IV, prospective, non-interventional study performed in 13 European countries enrolled high-risk essential thrombocythemia patients treated with cytoreductive therapy. The primary objectives were safety and pregnancy outcomes. Of 3721 registered patients, 3649 received cytoreductive therapy. At registration, 3611 were receiving: anagrelide (Xagrid®) (n=804), other cytoreductive therapy (n=2666), or anagrelide + other cytoreductive therapy (n=141). The median age was 56 vs. 70 years for anagrelide vs. other cytoreductive therapy. Event rates (patients with events/100 patient-years) were 1.62 vs. 2.06 for total thrombosis and 0.15 vs. 0.53 for venous thrombosis. Anagrelide was more commonly associated with hemorrhage (0.89 vs. 0.43), especially with anti-aggregatory therapy (1.35 vs. 0.33) and myelofibrosis (1.04 vs. 0.30). Other cytoreductive therapies were more associated with acute leukemia (0.28 vs. 0.07) and other malignancies (1.29 vs. 0.44). Post hoc multivariate analyses identified increased risk for thrombosis with prior thrombohemorrhagic events, age ≥65, cardiovascular risk factors, or hypertension. Risk factors for transformation were prior thrombohemorrhagic events, age ≥65, time since diagnosis, and platelet count increase. Safety analysis reflected published data, and no new safety concerns for anagrelide were found. Live births occurred in 41/54 pregnancies (76%). clinicaltrials.gov Identifier: 00567502. Introduction Essential thrombocythemia (ET) is associated with an increased risk of thrombohemorrhagic complications and transformation to myelofibrosis (MF) or acute leukemia (AL).1 Cytoreductive therapy (CRT) is used to reduce thrombosis and hemorrhage in high-risk ET.2 European LeukemiaNet recommends hydroxycarbamide (HC) as first-line therapy, but advises caution in patients <40 years.2 Anagrelide (Xagrid®) is licensed in Europe for patients with ET intolerant/refractory to HC,2,3 and in some countries (i.e., USA, Japan) it is authorized as first-line therapy. CRT decreases thrombosis upon the reduction and control of platelet counts.4 HC carries a potential leukemogenic risk; however, this remains a matter of debate.5-7 Few prospective studies have been conducted to compare the effect of different CRT therapies on complications in ET8,9 or to report on the long-term safety outcomes. Herein, we report final data from the Evaluation of Anagrelide Efficacy and Long-term Safety (EXELS) study, the largest prospective study in highhaematologica | 2018; 103(1)

Correspondence: gunnar.birgegard@medsci.uu.se

Received: June 20, 2017. Accepted: October 25, 2017. Pre-published: October 27, 2017. doi:10.3324/haematol.2017.174672 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/51 ©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.

51


G. Birgegård et al.

risk ET patients treated with CRT, including post hoc multivariate analyses identifying risk factors for thrombohemorrhagic events, and transformation to MF and AL.

Methods Trial design EXELS is a phase IV, prospective, non-interventional, postauthorization, multicenter cohort study which was conducted in 125 centers in 13 European countries from May 2005 to April 2014. The results represent the final data from the 5-year observation period. High-risk ET patients (age >60 years, prior thrombosis/hemorrhage, or platelets >1000x109/L) were eligible if receiving/scheduled to receive CRT. The Polycythemia Vera Study Group (PVSG)10 or the World Health Organization (WHO) 2001/200811 diagnostic criteria were used. As an observational study, the protocol could not mandate mutation analysis; however, investigators were encouraged to report JAK2 status. Written informed consent was obtained from all participants in accordance with the Declaration of Helsinki.

Treatment CRT was determined prior to registration. Patients could be newly diagnosed or continuing prior CRT, including anagrelide, HC, busulfan, interferon-α (IFN), pegylated interferon, pipobroman and sodium phosphate (P32). Patients could receive concomitant anti-aggregatory (A-A) therapy and have CRT changed at any time at the investigator’s discretion.

Objectives The primary objective was safety and pregnancy outcomes with anagrelide vs. other CRTs in high-risk ET patients in routine clinical practice. The secondary objective was to evaluate efficacy, as measured by the incidence of thrombotic/hemorrhagic events, and platelet counts.

Data capture Data were collected at registration and every 6 months for 5 years. Suspected serious adverse reactions (SSARs) and predefined events (PDEs) of specific interest in this study were recorded. An independent event validation panel blinded to CRT validated PDEs for consistent/correct PDE group allocation. Study conduct/monitoring was overseen by a steering committee and an independent data and safety monitoring board.

Statistical methods

Patients receiving a treatment for ≥1 day were allocated to treatment groups. The safety population included patients receiving ≥1 dose of CRT, analyzed as: 1) first-treatment analysis; PDEs or SSARs that occurred before a patient switched CRT, and 2) overall-treatment analysis; PDEs or SSARs allocated to the treatment at time of event. Patients could be included in more than one treatment group; as such the number of events is higher in the overall - vs. first-treatment group. Data were analyzed according to ‘anagrelide only’, ‘other CRT’, and ‘anagrelide + other CRT’. Patients with multiple events of the same PDE/SSAR category were counted once for each treatment received. Selected PDEs were combined into the following five endpoints: 1) all major thrombotic events (arterial plus venous), 2) arterial thrombotic events, 3) venous thrombotic events, 4) major hemorrhagic events, and 5) thrombohemorrhagic events (all major thrombotic/hemorrhagic events). Event rates were calculated as the number of affected patients per 100 patient-years exposure. 52

Multivariate analysis Two post hoc multivariate analyses were performed, first for the overall group in order to evaluate potential risk factors (except treatment) for thrombotic/hemorrhagic events, and death, and second for the first-treatment group to investigate risk factors and treatment effect for thrombotic/hemorrhagic events and MF transformation. Regression analysis followed the strategy for model selection.12 The final model was fit by using a significance level of 0.05. To facilitate interpretation, “significant” is used for “variable” or “variable treatment” interaction term significant at P≤0.001 level as no multiplicity adjustment was made.

Results Patients The EXELS study included 3721 patients from 13 European countries. At time of registration, 110 patients were not receiving CRT and were excluded from the first-treatment safety population (n=3611). Of these, 38 subsequently received CRT and were included in the overall-treatment safety population (n=3649; Online Supplementary Figure S1). 66% of the patients were diagnosed according to WHO diagnostic criteria, 29% were diagnosed by PVSG diagnostic criteria, and 5% were diagnosed by unknown criteria. Baseline characteristics in the first-treatment safety population are shown in Table 1. Anagrelide was more frequently used in younger patients (Figure 1). Median age was therefore lower in the ‘anagrelide’ (55.5 years, range: 18–89) and ‘anagrelide + other CRT’ (59.0 years, range: 22–88) vs. the ‘other CRT’ groups (70.0 years, range: 17–95). At baseline, the proportion of patients with prior thrombohemorrhagic events were similar at 25–30% across treatment groups (Table 1). More than 80% of patients received either HC (n=2341) or anagrelide (n=804) therapy. In the safety population, median duration of exposure was 1717.0 days (range: 1–2573) and 1481.0 days (range: 1–2677) in the ‘HC’ and ‘anagrelide’ groups, respectively. Exposure to HC and anagrelide was 10377 and 4320 patient-years, respectively. IFN was used by 136 patients, and monotherapy with busulfan, pipobroman or P32 by <5% together; therefore, outcome analyses were not performed for these subgroups. A-A was used by 58.1% in the ‘anagrelide’ group and 72.8% in the ‘other CRT’ group. JAK2V617F mutation status was reported in too few patients to allow for statistical analysis.

Complications as predefined events The recorded PDEs included complications of the disease, adverse effects of treatment and non-related adverse events (Table 2). In the first treatment analysis population the most common PDEs (apart from thrombosis and hemorrhage) in the ‘anagrelide’ group were cardiovascular symptoms (i.e., palpitations and tachycardia), and in the ‘other CRT’ group non-hematological malignancies and non-PDE death. The same pattern was seen in the overall-treatment population (Online Supplementary Table S1). The most common cardiovascular events, tachycardia (2.0% vs. 0.1%) and palpitations (1.7% vs. 0.2%), were reported with more frequency in the ‘anagrelide’ vs. ‘other CRT’ group (Online Supplementary Table S2). Arrhythmias consisted almost exclusively of atrial fibrillahaematologica | 2018; 103(1)


EXELS: long-term study in high-risk ET patients

tion, which showed similar event rates (0.30 vs. 0.33) in both groups (Online Supplementary Table S1). Ventricular tachycardia and ventricular fibrillation was recorded in one patient each in the ‘other CRT’ group. Cardiac failure was reported in a very low proportion of patients in the ‘anagrelide’ and ‘other CRT’ groups (both <0.1%, event rates: 0.37 vs. 0.32; Table 2).

Suspected serious adverse reactions The SSAR event rate was low across all treatments, albeit slightly higher in the ‘anagrelide’ vs. ‘other CRT’ group, at 0.86 vs. 0.60, respectively (Online Supplementary Table S3). This difference was predominantly caused by cardiac events, and to a lesser extent by gastrointestinal disorders. No unexpected side effects were noted for anagrelide.

Death In the overall safety population, 439 patients (12.0%) died. ET-related deaths included transformation (70; 1.9%), myocardial infarction (33; 0.9%), major hemorrhagic event (21; 0.6 %) and stroke (19; 0.5%). The causes of non-ET-related death included non-hematological malignancy (57; 1.6%) and cardiac failure (20; 0.5%). In 188 patients (5.2%), death was not assigned a PDE category and comprised of: an unknown reason (108), sepsis/infection (24), respiratory/pulmonary diseases (22), natural death/deterioration of health status (20), cardiovascular disease (11), and cancer (3).

Pregnancy Fifty-four pregnancies (40 patients) occurred. Nine patients had no therapy, 24 had IFN (one patient had

Table 1. Patient demographic and baseline characteristics: first treatment analysis population.

Treatment at registration

Anagrelide N=804

Age, years Median 55.5 Range 18–89 Age categories (years), n (%) <65 years 575 (71.5) 65 – <75 years 133 (16.5) ≥75 years 96 (11.9) Gender, n (%) Male 303 (37.7) Female 501 (62.3) Clinically significant vascular risk factor present, n (%) Yes 441 (54.9) Hypertension 268 (33.3) Diabetes 41 (5.1) Hypercholesterolemia 114 (14.2) Smoking 122 (15.2) Other 96 (11.9) Overall normal cardiac function, n (%) Yes 587 (73.0) Currently taking aspirin, n (%) Yes 467 (58.1) Any significant hemorrhagic/thrombotic events prior to registration, n (%) Yes 202 (25.1) Baseline platelet count n 703 Median (109//L) 453 Range (109//L) 36–2226 Baseline white blood cells n 230 Median (109/L) 8.7 Range (109//L) 3.22–31.30 Baseline hemoglobin n 281 Median (g/dL) 12.8 Range (g/dL) 5.5–17.3

Other CRT N=2666

Anagrelide + other CRT N=141

70 17–95

59 22–88

913 (34.3) 897 (33.7) 854 (32.1)

89 (63.1) 30 (21.3) 22 (15.6)

1032 (38.7) 1632 (61.3)

61 (43.3) 80 (56.7)

1814 (68.0) 1260 (47.3) 211 (7.9) 500 (18.8) 346 (13.0) 351 (13.2)

79 (56.0) 48 (34.0) 8 (5.7) 16 (11.3) 34 (24.1) 16 (11.3)

1649 (61.9)

104 (73.8)

1942 (72.8)

91 (64.5)

768 (28.8)

43 (30.5)

2344 432 114–2020

125 494 220–1816

616 6.1 2.0–34.19

43 7.6 2.9–43.50

743 13.0 8.2–18.6

52 12.3 8.0–16.0

CRT: cytoreductive therapy.

haematologica | 2018; 103(1)

53


G. Birgegård et al.

HC+IFN), 10 had anagrelide, eight had A-A, and three were receiving anticoagulant without CRT at the time that pregnancy was reported. Most patients on IFN or no therapy continued without change during pregnancy. There were five terminations in five patients; three patients had six miscarriages (all but one during the first trimester) and the outcome was missing for two pregnancies. There were 41 live births reported (75.9% of the pregnancies), all with normal birth weight (range: 2.4–4.7 kg).

Thrombotic and hemorrhagic events In the first-treatment analysis group, the event rate for major thrombosis was lower in the ‘anagrelide’ vs. the ‘other CRT’ group (1.62 vs. 2.06; Table 3). Arterial thrombotic event rates were similar between the ‘anagrelide’ and ‘other CRT’ group (1.47 vs. 1.55), whereas the venous thrombotic rate was lower in the ‘anagrelide’ compared with the ‘other CRT’ group (0.15 vs. 0.53). The major hemorrhagic event rate was higher in the ‘anagrelide’ group (0.89 vs. 0.43), especially in patients treated with A-A (1.35 vs. 0.33, respectively). The composite thrombohemorrhagic event rate was slightly higher in the ‘anagrelide’ (2.47) compared to ‘other CRT’ (2.41) group, largely due to hemorrhagic events. Similar results for thrombosis and hemorrhage event rates were observed in the overall treatment analysis group (Online Supplementary Table S4).

Transformation and malignancies In the first-treatment analysis, 64 patients transformed to MF, 31 to AL, and 12 to myelodysplastic syndrome (MDS; Table 4). The rate of transformation to MF was higher in the ‘anagrelide’ compared with the ‘other CRT’ group (1.04 vs. 0.30). Transformation to AL was higher in the ‘other CRT’ compared with the ‘anagrelide’ group (0.28 vs. 0.07), and transformation to MDS only occurred in the ‘other CRT’ group (event rate: 0.12). Similar proportions were seen in the overall treatment analysis (Table 4). For patients who had only ever received one type of CRT, the rate of transformation to MF was still higher in the ‘anagrelide’ compared with the ‘HC’ group (0.61 vs. 0.14); the median time to transformation was similar, around 7 years (range: 0.95–20.62). The rate of transformation to

AL in the ‘anagrelide’ group was 0 vs. 0.22 in patients treated with HC alone, while the median time to transformation was 6.48 years from diagnosis (range: 1.21–22.09) for the ‘HC’ group. High platelet levels at baseline were identified in the multivariate analysis as a risk factor for MF transformation (hazard ratio [HR] 1.18, P=0.0004); the risk of MF increased for each increased platelet count of 100x109/L. The rate of non-hematological malignancies was lower in the ‘anagrelide’ vs. ‘other CRT’ group, both in the first-treatment (0.44 vs. 1.29; Table 2) and overall-treatment analysis groups (0.49 vs. 1.35; Online Supplementary Table S1).

Blood counts Platelet levels were well controlled throughout the study across treatment groups (Figure 2). The median counts for ‘anagrelide’ and ‘other CRT’ were 431x109/L and 413x109/L at 6 months, respectively, and 390x109/L and 404x109/L at 5 years, respectively. In patients who experienced a major thrombotic event, median platelet counts ranged from 402 to 430x109/L at the time of event, and the percentages of patients with a platelet count ≤450x109/L or >600x109/L were similar in patients with and without thrombosis (data not shown). As expected, median white blood cell (WBC) counts were lower at registration in the ‘other CRT’ group (6.1x109/L) compared with the ‘anagrelide’ (8.7x109/L) and ‘anagrelide + other CRT’ (7.6x109/L) groups (Table 1) throughout the study, and remained relatively stable over time (Figure 2). The WBC count >15x109/L at any time prior to thrombosis was seen in 4.3% (n=14/327) compared with 4.9% (n=162/3322) in patients without thrombosis. The amount of reported WBC data was too low to allow for a Cox regression analysis of any correlation between WBC counts and thrombosis. Notably, 11.4% (n=50/439) of patients who died during the study had a WBC count >15x109/L at any time prior to death vs. 4.2% (n=135/3210) of patients who remained alive. Median hemoglobin levels remained stable throughout the study across all treatment groups (Figure 2), with ranges of 12.3– 13.0 g/dL at registration and 12.0–13.0 g/dL at 5 years.

Table 2. Cumulative event rates of other predefined events for the first-treatment analysis population.

Treatment at registration Predefined event

Anagrelide N=804 Patients Event rate (events) n

Other CRT N=2666 Patients Event rate (events) n

Anagrelide + other CRT N=141 Patients Event rate (events) n

Congestive heart failure Cardiomyopathy Other cardiovascular symptoms Severe mucocutaneous disorders Pulmonary hypertension Pulmonary fibrosis/interstitial pneumonia Pancreatitis Rhabdomyolysis/myalgia Non-hematological malignancy Non-PDE death*

10 (10) 4 (4) 47 (61) 4 (4) 4 (4) 1 (1) 0 2 (2) 12 (13) 18 (18)

31 (37) 9 (9) 86 (117) 63 (68) 5 (5) 9 (10) 3 (3) 3 (4) 123 (140) 105 (105)

2 (2) 0 12 (12) 2 (2) 0 0 0 0 2 (2) 2 (2)

0.37 0.15 1.79 0.15 0.15 0.04 0 0.07 0.44 0.66

0.32 0.09 0.90 0.65 0.05 0.09 0.03 0.03 1.29 1.08

0.53 0 3.30 0.54 0 0 0 0 0.54 0.53

*Deaths not recorded as an outcome of a predefined event. CRT: cytoreductive therapy; PDE: predefined event.

54

haematologica | 2018; 103(1)


EXELS: long-term study in high-risk ET patients

Multivariate analysis, risk factors for thrombohemorrhagic and transformation events

‘Prior thrombohemorrhagic events’ and ‘age ≥65 years’ at baseline were identified as risk factors for major thrombotic, arterial thrombotic, venous thrombotic and total thrombohemorrhagic events (Online Supplementary Table S5). The presence of baseline cardiovascular risk factors was associated with a higher risk of arterial thrombotic events. A higher platelet count at baseline correlated with an increased risk for transformation to MF (HR 1.18, confidence interval [CI] 1.08 to 1.30, P=0.0004) for each platelet count increase of 100x109/L at baseline. A lower risk for MF transformation was indicated for patients diagnosed with ET by WHO criteria than those diagnosed by PVSG criteria (HR 0.55, P=0.03; Table 5). As expected, a time from diagnosis >10 years indicated a higher risk for MF transformation (HR 4.38, CI 1.49 to 12.88 compared with <1 year, P=0.0073; Table 6). In a separate multivariate analysis conducted in the first treatment analysis group (‘anagrelide’ vs. ‘other CRT’), the influence of treatment on risk for thrombosis and hemorrhage, as well as transformation to MF, was analyzed (Table 5). Anagrelide was associated with a higher risk for major thrombosis (HR 1.68, CI 1.09 to 2.60, P=0.02) and arterial thrombosis (HR 1.91, CI 1.20 to 3.04, P=0.0067). There was no difference between WHO-defined and PVSG-defined ET in this respect (HR 1.49 vs. 1.45). There was also no difference in risk for arterial thrombosis in general between WHO-defined and PVSG-defined ET. In a Cox regression analysis with arterial events as the outcome and ET diagnosis as the single explanatory variable, HR was 1.06 (CI 0.77 to 1.46, P=0.73). The risk for MF transformation was greater in the ‘anagrelide’ vs. ‘other CRT’ group (HR 3.33, CI 1.94 to 5.73, P<0.0001), but this difference was smaller in WHOdefined than in PVSG-defined ET (HR 2.79, CI 1.24 to 6.27, P=0.0132 vs. HR 4.67, CI 2.10 to 10.37, P=0.0002). The risk for MF transformation was also generally lower in WHO-defined than in PVSG-defined ET (Table 5). The HR for venous thrombosis, on the other hand, was lower (HR 0.43, CI 0.15 to 1.21, P=0.11), although it did not reach statistical significance. In anagrelide patients, the concomitant use of A-A therapy showed an increased risk

of bleeding (HR 3.55, CI 1.96 to 6.44, P<0.0001), which increased the total thrombohemorrhagic risk (HR 2.46, CI 1.65 to 3.66, P<0.0001). An increased risk for total thrombohemorrhagic events was also seen for anagrelide patients who were smokers (HR 2.34, CI 1.21 to 4.54, P=0.0118). A similar effect was seen for arterial thrombotic events, where the risk was also greater for patients receiving anagrelide who were smokers (HR 3.18, CI 1.41 to 7.16, P=0.005). Furthermore, in patients with no prior thrombohemorrhagic events, the risk for thrombohemorrhagic events was higher with anagrelide treatment than with other CRT (HR 2.20, CI 1.38 to 3.53, P=0.001). It was confirmed that age >65 and previous thrombosis increased the risk of thrombohemorrhagic events. An initial platelet count of >1000x109/L and hypertension were associated with a risk for hemorrhage.

Discussion EXELS represents the largest prospective cohort of patients with ET reported to date, and as such it provides several important real-world insights into the therapeutic management of this disease. Patients receiving anagrelide were younger than those receiving another CRT, likely due to investigators choosing to use anagrelide as first-line therapy in younger patients in order to avoid a transformation related to the potential leukemogenic risk associated with long-term HC treatment. Differences in age may confound analysis of the event rates in the various treatment groups.

Thrombohemorrhagic events Importantly, overall thrombohemorrhagic event rates were low with a rate of <2.5 independent of therapy, in line with other large studies, and fairly comparable between the ‘anagrelide’ and ‘other CRT’ group, consistent with the Anagrelide vs. Hydroxyurea - Efficacy and Tolerability Study in Patients With Essential Thrombocythaemia (ANAHYDRET) study8 and the Primary Thrombocythaemia 1 (PT-1) trial.9 The total thrombotic event rate was lower in the ‘anagrelide’ group, which may reflect the age difference between the groups.

Figure 1. Treatment at registration vs. age. Patients in various age groups were treated at registration with anagrelide, hydroxycarbamide, interferon, combination therapy or other monotherapy. *Includes busulfan, interferon, pipobroman, P32, thromboreductin (anagrelide).

haematologica | 2018; 103(1)

55


G. Birgegård et al.

The venous thrombotic event rate was lower in the ‘anagrelide’ group, which supports findings from both the PT1 trial and the ANAHYDRET study.8,9 Currently there is no validated explanation for such disparity, but the confirmation in two large studies suggests that the difference warrants further investigation. The hemorrhagic event rate was higher in the ‘anagrelide’ vs. ‘other CRT’ group, especially when anagrelide was combined with A-A therapy, which also corresponds with findings from the PT-1 trial9 and underlines the importance of cautious use of this combination, especially in patients with previous hemorrhages.3 A protective effect of A-A therapy for thrombosis was

seen in the ‘other CRT’ and ‘anagrelide + other’ CRT groups, but not in the ‘anagrelide’ group. A possible explanation for this is that the ‘other CRT’ patients had more benefit from A-A therapy due to their older age. These results are consistent with current guidelines recommending A-A therapy in high-risk ET patients. A recent systematic review of A-A therapy in ET13 reported significant uncertainty regarding evidence for a protective effect against thrombosis. The study herein gives some evidence for such an effect, at least in HC-treated patients. The multivariate analysis did not detect any difference in the efficacy of anagrelide between WHO- vs. PVSG-diagnosed ET.

Table 3. Cumulative event rates of thrombohemorrhagic events by first-treatment analysis population.

Treatment at registration Predefined event

Anagrelide N=804 Patients Event rate (events) n

Other CRT N=2666 Patients Event rate (events) n

Major thrombotic events* With A-A Without A-A Arterial thrombotic events With A-A Without A-A Venous thrombotic events With A-A Without A-A Major hemorrhagic events With A-A Without A-A Total thrombohemorrhagic events With A-A Without A-A

43 (52) 26 (32) 12 (13) 39 (48) 25 (31) 10 (11) 4 (4) 1 (1) 2 (2) 24 (29) 19 (22) 3 (3) 65 (81) 43 (54) 15 (16)

194 (231) 129 (151) 47(55) 147 (175) 102 (122) 33 (36) 51 (56) 28 (29) 15 (19) 42 (47) 23 (24) 15 (17) 226 (278) 146 (175) 61 (72)

1.62 1.88 1.11 1.47 1.81 0.92 0.15 0.07 0.18 0.89 1.35 0.27 2.47 3.13 1.38

2.06 1.88 2.36 1.55 1.48 1.65 0.53 0.40 0.74 0.43 0.33 0.74 2.41 2.13 3.09

Anagrelide + other CRT N=141 Patients Event rate (events) n 4 (5) 0 4 (4) 4 (5) 0 4 (4) 0 0 0 1 (1) 0 1 (1) 5 (6) 0 5 (5)

1.09 0 3.34 1.09 0 3.34 0 0 0 0.27 0 0.84 1.37 0 4.32

*Includes both arterial and venous events. A-A: anti-aggregatory; CRT: cytoreductive therapy.

Table 4. Cumulative transformation event rates by first- and overall-treatment analysis populations.

Treatment at registration (first-treatment analysis)

Anagrelide N=804 Patients Event rate (events) n

Other CRT N=2666 Patients Event rate (events) n

Myelofibrosis Myelodysplasia Acute leukemia Other leukemia*

28 (28) 0 2 (2) 5 (5)

29 (29) 12 (12) 27 (27) 13 (13)

Treatment at time of event (overall-treatment analysis) Myelofibrosis Myelodysplasia Acute leukemia Other leukemia*

1.04 0 0.07 0.18

Anagrelide N=1127 45 (45) 1 (1) 6 (6) 5 (5)

1.31 0.03 0.17 0.14

0.30 0.12 0.28 0.13

Other CRT N=2909 35 (35) 14 (14) 36 (36) 13 (13)

Anagrelide + other CRT N=141 Patients Event rate (events) n 7 (7) 0 2 (2) 0

1.91 0 0.53 0

Anagrelide + other CRT N=451 0.32 0.13 0.33 0.12

11 (11) 2 (2) 4 (4) 1 (1)

1.27 0.23 0.46 0.11

*“Other leukemia” includes chronic myelogenous leukemia and unclassified leukemia. CRT: cytoreductive therapy.

56

haematologica | 2018; 103(1)


EXELS: long-term study in high-risk ET patients

Transformation Transformation to AL was more frequent in the ‘other CRT’ vs. the ‘anagrelide’ group. All six patients in the ‘anagrelide’ group who developed AL had previously been treated with HC. In patients who had only ever received anagrelide, there were no cases of AL. Non-hematological malignancies were also less frequent in the ‘anagrelide’ than in either the ‘other CRT’ or the ‘HC’ groups. This warrants further analysis and will be addressed in a separate publication. The multivariate analysis identified both a history of thrombohemorrhagic events and an age of ≥65 years at baseline as risk factors for predicting transformation to AL/MDS events. Transformation to MF was higher in the ‘anagrelide’ vs. ‘HC’ group, in line with data from the PT-1 trial.14 In both

A

studies, patients included those with both ‘true ET’ according to the WHO classification, as well as ET diagnosed by PVSG criteria, some of whom may have had early MF according to the new WHO criteria11 with significant bone marrow fibrosis at entry. This does not explain the difference in the MF transformation event rate, but illustrates that anagrelide does not seem to hinder fibrosis development. Considering the much lower rate of MF development in correctly diagnosed ET,2,15 also supported by our results, the optimal patient group for anagrelide treatment could be those with true ET rather than early MF. This is further supported by our finding that the risk for transformation to MF in anagrelide-treated patients was more pronounced in PVSG-defined than in WHOdefined ET. Regular surveillance for features of MF should be considered with anagrelide, especially in patients with primary MF 0–1. A higher platelet count at baseline was identified as a risk factor for MF transformation in the multivariate analysis, a new and interesting finding that requires further confirmation.

Platelet control, white blood cell count and thrombosis

B

Generally, platelet counts were well controlled throughout the study. No obvious advantage for protection against thrombosis by reducing platelet counts to values below 450x109/L was seen. Of note, in EXELS, platelet levels were already well controlled at registration. These results suggest that platelet counts do not provide a prognostic value for thrombosis in patients with reasonable platelet control. In contrast, other studies indicate an effect on thrombosis rate by platelet control,16-18 and two recent studies show a higher platelet count at time of event for patients with thrombosis.19,20 Taken together, it remains uncertain whether reducing platelets to normal levels gives better protection against thrombosis compared with a more tolerant treatment goal. This does not alter the treatment goal in general, but is of importance in situations with moderate treatment efficacy or side effects. Notably, WBC >15x109/L did not correlate with thrombotic events, and anagrelide was not associated with a time-dependent hemoglobin-lowering effect.

Safety

C

Figure 2. Median blood cell counts over time by first-treatment analysis population. An ad hoc analysis was performed to exclude extreme laboratory data attributed to data entry errors based on the following thresholds: platelet counts of <10x109/L or >10,000x109/L, white blood cell counts of <0.5x109/L or >150x109/L, hemoglobin of <5 g/dL or >22 g/dL, and hematocrit at <10% or >70%. Removal of outliers resulted in similar median results. ANA: anagrelide.

haematologica | 2018; 103(1)

The most common cardiovascular adverse events associated with anagrelide in our study were palpitation and tachycardia, in line with previous reports9,17,21 and caused by the PDEIII inhibition and positive chronotropic effect of anagrelide. However, the rate of atrial fibrillation was similar for both groups (0.30 vs. 0.33), and other, more severe arrhythmias were rare in both groups. Likewise, heart failure had similar event rates in both groups, which contrast with previous reports that anagrelide may worsen cardiac failure due to the positive chronotropic effect.22 The difference in median age between the treatment groups may have contributed to this. Overall, the SSAR event rate was low, with a slightly higher rate in the ‘anagrelide’ compared with the ‘other CRT’ group, predominantly caused by the difference in cardiac event rates for events like tachycardia and palpitations. Cardiac symptoms reported in this study are consistent with the known safety profile for anagrelide.3

Pregnancy Most subjects became pregnant while receiving IFN or no CRT. Of the 54 pregnancies, 75.9% resulted in live births, a 57


G. Birgegård et al.

better result than has been seen in many other reports, which may reflect a development in patient management during pregnancy and the selection of patients who chose to become pregnant. Almost all miscarriages occurred during the first trimester, which supports the review of Valera

et al.23 There were too few miscarriages to make any comparison between treatment groups. The high percentage of successful pregnancies in this large study may support clinicians’ decisions to give their patients positive information with regard to the chance for planned childbirth.

Table 5. Results of multivariate regression analysis, first-treatment analysis group, of baseline risk factors for thrombohemorrhagic events and myelofibrosis transformation. Factors with relevant HR difference from 1.

Outcome event

Factor

Major thrombotic events

Treatment (anagrelide vs. other) Smoking (yes vs. no) Prior hemorrhagic or vascular events (yes vs. no) Age group (≥65 vs. <65 years) Vascular risk factors (yes vs. no) Smoking=yes, anagrelide vs. other Smoking=no, anagrelide vs. other Treatment (anagrelide vs. other) Smoking (yes vs. no) Prior hemorrhagic or vascular events (yes vs. no) Age group (≥65 vs. <65 years) Vascular risk factors (yes vs. no) Smoking=yes, anagrelide vs. other Smoking=no, anagrelide vs. other Treatment (anagrelide vs. other) Prior hemorrhagic or vascular events (yes vs. no) Age group (≥65 vs. <65 years) Treatment (anagrelide vs. other) Receiving A-A at start of study (yes vs. no) No A-A at start of study, anagrelide vs. other A-A at start of study, anagrelide vs. other Initial platelet count >1000x109/L (yes vs. no) Prior hemorrhagic or vascular event (yes vs. no) Hypertension (yes vs. no) Treatment (anagrelide vs. other) Smoking (yes vs. no) Receiving A-A at start of study (yes vs. no) Prior hemorrhagic or vascular event (yes vs. no) Age group (≥65 vs. <65 years) Vascular risk factors (yes vs. no) Smoking=yes, anagrelide vs. other Smoking=no, anagrelide vs. other No A-A at start of study, anagrelide vs. other A-A at start of study, anagrelide vs. other Prior hemorrhagic or vascular events = yes, anagrelide vs. other Prior hemorrhagic or vascular events = no, anagrelide vs. other Treatment (anagrelide vs. other) Sex (female vs. male) Sex=female, anagrelide vs. other Sex=male, anagrelide vs. other ET diagnosis (WHO criteria vs. PVSG criteria) ET diagnosis (Other vs. PVSG criteria)

Arterial thrombotic events

Venous thrombotic events

Major hemorrhagic events

Thrombohemorrhagic events

Myelofibrosis transformation

Hazard ratio (95% CI)

P

1.68 (1.09, 2.60) 1.18 (0.76, 1.83) 2.14 (1.65, 2.78) 2.19 (1.59, 3.01) 1.40 (1.02, 1.93) 3.01 (1.41, 6.44) 0.94 (0.63, 1.39) 1.91 (1.20, 3.04) 1.18 (0.74, 1.88) 1.83 (1.37, 2.45) 2.04 (1.43, 2.91) 1.67 (1.15, 2.42) 3.18 (1.41, 7.16) 1.14 (0.75, 1.74) 0.43 (0.15, 1.21) 3.29 (1.88, 5.75) 3.04 (1.50, 6.16) 1.29 (0.69, 2.43) 1.37 (0.73, 2.56) 0.47 (0.16, 1.42) 3.55 (1.96, 6.44) 2.36 (1.42, 3.93) 1.64 (1.00, 2.69) 1.69 (1.02, 2.79) 1.51 (1.01, 2.26) 1.22 (0.85, 1.77) 1.28 (0.93, 1.75) 1.52 (1.14, 2.02) 1.82 (1.38, 2.39) 1.38 (1.03, 1.84) 2.34 (1.21, 4.54) 0.98 (0.68, 1.41) 0.93 (0.51, 1.70) 2.46 (1.65, 3.66) 1.04 (0.62, 1.73) 2.20 (1.38, 3.53) 3.33 (1.94, 5.73) 0.74 (0.43, 1.27) 6.04 (2.93, 12.45) 1.84 (0.82, 4.12) 0.55 (0.33, 0.94) 0.23 (0.03, 1.68)

0.0194 0.4516 <0.0001 <0.0001 0.0381 0.0045 0.7508 0.0067 0.4935 <0.0001 <0.0001 0.0074 0.0053 0.5342 0.1099 <0.0001 0.002 0.4271 0.3221 0.1806 <0.0001 0.001 0.0498 0.04 0.0435 0.2859 0.124 0.0041 <0.0001 0.0305 0.0118 0.9034 0.818 <0.0001 0.8845 0.001 <0.0001 0.2699 <0.0001 0.14 0.0285 0.1464

A-A: anti-aggregatory; CI: confidence interval; ET: essential thrombocythemia; PVSG: Polycythemia Vera Study Group; WHO: World Health Organization.

58

haematologica | 2018; 103(1)


EXELS: long-term study in high-risk ET patients

Table 6. Multivariate analysis, overall-treatment population, and significant baseline risk factors for predicting a thrombohemorrhagic or transformation event. Factors with relevant HR difference from 1.

Outcome event

Factor

Hazard ratio (95% CI)

P

2.05 (1.58, 2.66) 2.14 (1.60, 2.86) 1.38 (1.01, 1.87) 1.81 (1.36, 2.41) 1.90 (1.38, 2.62) 1.65 (1.16, 2.35) 3.70 (1.86, 7.35) 3.27 (1.87, 5.72) 1.91 (1.52, 2.41) 1.67 (1.29, 2.17) 1.33 (1.04, 1.69) 4.38 (1.49, 12.88) 3.38 (1.16, 9.81) 1.54 (0.52, 4.59) 1.18 (1.08, 1.30) 2.17 (1.18, 4.01) 3.36 (1.56, 7.26)

<0.0001 <0.0001 0.0404 <0.0001 <0.0001 0.0055 0.0002 <0.0001 <0.0001 <0.0001 0.0207 0.0073 0.0254 0.4333 0.0004 0.0127 0.002

Major thrombotic events

Prior hemorrhagic or vascular events (yes vs. no) Age group: ≥65 years vs. <65 years Vascular risk factors: yes vs. no Arterial thrombotic events Prior hemorrhagic or vascular events: yes vs. no Age group: ≥65 years vs. <65 years Vascular risk factors: yes vs. no Venous thrombotic events Age group: ≥65 years vs. <65 years Prior hemorrhagic or vascular events : yes vs. no Thrombohemorrhagic events Prior hemorrhagic or vascular events : yes vs. no Age group: ≥65 years vs. <65 years Hypertension : yes vs. no Myelofibrosis transformation Time since diagnosis : ≥10 years vs. 0–<1 years Time since diagnosis : 5-<10 years vs. 0–<1 years Time since diagnosis : 1-<5 years vs. 0–<1 years Platelets at baseline (100 units increase) AL/MDS transformation Prior hemorrhagic or vascular events : yes vs. no Age group: ≥65 years vs. <65 years AL: acute leukemia; CI: confidence interval; MDS: myelodysplastic syndrome.

Multivariate risk factor analysis Multivariate analyses were performed both in the overall- and first-treatment populations. These were post hoc analyses and results must therefore be interpreted with caution; however, given the age imbalance between the patient populations, the data are important to assess alongside the event rates. Risk factors identified for thrombohemorrhagic events and for major thrombosis included age, history of thrombosis or hemorrhage, and cardiovascular risk factors, previously reported in retrospective studies.24-28 In line with the International Prognostic Score of thrombosis in World Health Organization-essential thrombocythemia (IPSET-thrombosis) risk score model,24 the WBC count was not indicative of higher thrombosis risk. Platelets >1000x109/L indicated a higher risk for MF transformation, consistent with both previous retrospective studies and the role of clonal megakaryocytes in the development of fibrosis. WBC counts >15x109/L during the study were more common in patients who died. With regard to thrombosis, bleeding and MF transformation, some treatment-related findings arose from the multivariate analysis. Both major thrombosis and arterial thrombosis were higher in the ‘anagrelide’ group (HR 1.68, CI 1.09 to 2.60, P=0.02 and HR 1.91, CI 1.20 to 3.04, P=0.0067, respectively), in line with results of the PT-1 trial, but not the ANAHYDRET study.8,9 The main statistical analysis of event rates performed in the overall group showed a lower event rate for major thrombosis and similar event rates for arterial thrombosis in the ‘anagrelide’ vs. the ‘other CRT’ groups. However, one has to take into consideration that on the one hand the age difference between the groups could influence the event rate analysis, and on the other hand the uncertainty of a post hoc multivariate analysis. The HR for venous thrombosis was lower with anagrelide treatment (HR 0.43, CI 0.15 to 1.21, haematologica | 2018; 103(1)

P=0.12, not significant), consistent with the PT-1 trial and the ANAHYDRET study. A new finding was that the combination of anagrelide and smoking may increase the risk for arterial thrombosis vs. smoking plus other CRT (HR 3.18, CI 1.41 to 7.16, P=0.005, not significant). Another unexpected finding was the increased combined thrombohemorrhagic risk in anagrelide-treated patients (vs. other CRT) with no prior vascular event (HR 2.20, CI 1.38 to 3.53, P=0.001). The increased risk of hemorrhage with concomitant use of anagrelide and A-A therapy has been previously shown.9 The higher risk for MF in the anagrelide group, especially in women, is consistent with the event rate analysis and with the PT-1 trial.9 The higher risk for MF transformation in patients diagnosed according to the PSVG criteria rather than the WHO criteria is in line with previous reports.

Limitations EXELS is not a randomized study, thus treatment arms were not balanced for risk factors or baseline characteristics. The age difference between treatment groups was due to clinicians’ treatment preferences. Assessments were only conducted as part of local routine clinical practice, therefore data were missing for variables such as laboratory values and mutation status. Events are few in ET, and even though EXELS is the largest prospective ET study performed, caution must be advocated in the interpretation of statistics, especially for PDEs with low numbers.

Conclusions The EXELS study provides important observations that may influence clinical practice for patients with high-risk ET. It is the largest ever prospective real-world study in 59


G. Birgegård et al.

this condition. Overall, total thrombohemorrhagic event rates were low (<2.5/100 patient-years), but a lower rate of venous thrombotic and a higher rate of hemorrhagic events were observed in the ‘anagrelide’ group. A post hoc multivariate analysis confirmed these differences, but also suggested a higher arterial thrombosis risk in anagrelidetreated patients, especially smokers. Transformation to MF was more frequent in PVSG-defined than in WHOdefined ET and was more frequent in the ‘anagrelide’ than in the ‘other CRT’ group. Transformation to AL and other malignancies was more common with other CRTs. A history of thrombohemorrhagic events, age ≥65 years, cardiovascular risk factors, or hypertension were identified as baseline risk factors for thrombohemorrhagic events. These factors, time since diagnosis and an increase in platelet counts above normal were identified as risk factors for transformation to MF. Acknowledgments The authors would like to thank the contribution of all investigators who participated in this study (Appendix I). Under the direction of the authors, Donna Tillotson, PhD, employee of iMed Comms, an Ashfield Company, provided writing assistance for this publication. Editorial assistance in formatting, proofreading, copy

References 1. Swerdlow SH, Campo E, Harris NL, et al. WHO classification of tumours of haemopoietic and lymphoid tissues, 2nd edition. 4th ed. Lyon: IARC Press; 2008. 2. Barbui T, Barosi G, Birgegard G, et al. Philadelphia-negative classical myeloproliferative neoplasms: critical concepts and management recommendations from European LeukemiaNet. J Clin Oncol. 2011; 29(6):761-770. 3. Xagrid Summary of Product Characteristics, Shire Pharmaceuticals Ltd. 2014. (Available from; http://www. ema. europa. eu/ docs /en_GB/ document_library/EPAR__Product_Information/human/000480/WC5 00056557.pdf. Last accessed 9 February 2017) 4. Cortelazzo S, Finazzi G, Ruggeri M, et al. Hydroxyurea for patients with essential thrombocythemia and a high risk of thrombosis. N Engl J Med. 1995;332(17):11321136. 5. Spivak JL, Hasselbalch H. Hydroxycarbamide: a user's guide for chronic myeloproliferative disorders. Expert Rev Anticancer Ther. 2011; 11(3):403-414. 6. Cerquozzi S, Tefferi A. Blast transformation and fibrotic progression in polycythemia vera and essential thrombocythemia: a literature review of incidence and risk factors. Blood Cancer J. 2015; 5:e366. 7. Kiladjian JJ, Chevret S, Dosquet C, Chomienne C, Rain JD. Treatment of polycythemia vera with hydroxyurea and pipobroman: final results of a randomized trial initiated in 1980. J Clin Oncol. 2011; 29(29):3907-3913. 8. Gisslinger H, Gotic M, Holowiecki J, et al. Anagrelide compared with hydroxyurea in WHO-classified essential thrombocythemia: the ANAHYDRET Study, a randomized controlled trial. Blood. 2013; 121(10):17201728. 9. Harrison CN, Campbell PJ, Buck G, et al. Hydroxyurea compared with anagrelide in high-risk essential thrombocythemia. N Engl J Med. 2005;353(1):33-45.

60

editing, and fact checking was also provided by iMed Comms. Shire Pharmaceutical Development Ltd provided funding to iMed Comms for support in writing and editing this manuscript. Funding The study was funded by the Sponsor, Shire Pharmaceutical Development Ltd. The study was designed by the international EXELS steering committee (GB, J-JK, CB, MG, LG, CH), chaired by GB, who is also the lead and corresponding author, and developed the final draft of the manuscript. CB, MG, LG, CH, J-JK all enrolled patients to the study, analyzed the data and critically reviewed the manuscript. HA contributed to the study management, interpretation of the data and critically reviewed the manuscript. MH and JW undertook the statistical and multivariate analysis and critically reviewed the manuscript for data accuracy. All authors provided final approval of the manuscript, and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Although the Sponsor was involved in the design, collection, analysis, interpretation, and fact checking of information, the content of this manuscript, the ultimate interpretation, and the decision to submit it for publication in Haematologica was made by the authors independently.

10. Michiels JJ. Diagnostic criteria of the myeloproliferative disorders (MPD): essential thrombocythaemia, polycythaemia vera and chronic megakaryocytic granulocytic metaplasia. Neth J Med. 1997; 51(2):57-64. 11. Tefferi A, Thiele J, Orazi A, et al. Proposals and rationale for revision of the World Health Organization diagnostic criteria for polycythemia vera, essential thrombocythemia, and primary myelofibrosis: recommendations from an ad hoc international expert panel. Blood. 2007;110(4):1092-1097. 12. Collett D. Modeling survival data in medical research, 2nd edition. Boca Raton, FL: Chapman & Hall/CRC; 2003. 13. Chu DK, Hillis CM, Leong DP, Anand SS, Siegal DM. Benefits and risks of antithrombotic therapy in essential thrombocythemia: a systematic review. Ann Intern Med. 2017;167(3):170-180. 14. Harrison C. Rethinking disease definitions and therapeutic strategies in essential thrombocythemia and polycythemia vera. Hematology Am Soc Hematol Educ Program. 2010;2010:129-134. 15. Ejerblad E, Kvasnicka HM, Thiele J, et al. Diagnosis according to World Health Organization determines the long-term prognosis in patients with myeloproliferative neoplasms treated with anagrelide: Results of a prospective long-term followup. Hematology. 2013;18(1):13-18. 16. Laguna MS, Kornblihtt LI, Marta RF, Michiels JJ, Molinas FC. Effectiveness of anagrelide in the treatment of symptomatic patients with essential thrombocythemia. Clin Appl Thromb Hemost. 2000;6(3):157161. 17. Storen EC, Tefferi A. Long-term use of anagrelide in young patients with essential thrombocythemia. Blood. 2001;97(4):863866. 18. Lahuerta-Palacios JJ, Bornstein R, FernandezDebora FJ, et al. Controlled and uncontrolled thrombocytosis. Its clinical role in essential thrombocythemia. Cancer. 1988;61(6):12071212. 19. Schmitz S, Stauch M, Schlag R. Anagrelide for the treatment of thrombocythaemia in

20.

21.

22.

23.

24.

25.

26.

27.

28.

daily clinical practice: a post-marketing observational survey on efficacy and safety performed in Germany. Onkologie. 2010; 33(1-2):39-44. Buxhofer-Ausch V, Steurer M, Sormann S, et al. Influence of platelet and white blood cell counts on major thrombosis - analysis from a patient registry in essential thrombocythemia. Eur J Haematol. 2016; 97(6):511516. Birgegard G, Bjorkholm M, Kutti J, et al. Adverse effects and benefits of two years of anagrelide treatment for thrombocythemia in chronic myeloproliferative disorders. Haematologica. 2004;89(5):520-527. Engel PJ, Johnson H, Baughman RP, Richards AI. High-output heart failure associated with anagrelide therapy for essential thrombocytosis. Ann Intern Med. 2005; 143(4):311-313. Valera MC, Parant O, Vayssiere C, Arnal JF, Payrastre B. Essential thrombocythemia and pregnancy. Eur J Obstet Gynecol Reprod Biol. 2011;158(2):141-147. Barbui T, Finazzi G, Carobbio A, et al. Development and validation of an International Prognostic Score of thrombosis in World Health Organization-essential thrombocythemia (IPSET-thrombosis). Blood. 2012;120(26):5128-5133. Passamonti F, Rumi E, Arcaini L, et al. Prognostic factors for thrombosis, myelofibrosis, and leukemia in essential thrombocythemia: a study of 605 patients. Haematologica. 2008;93(11):1645-1651. Campbell PJ, Maclean C, Beer PA, et al. Correlation of blood counts with vascular complications in essential thrombocythemia: analysis of the prospective PT1 cohort. Blood. 2012;120(7):1409-1411. Lekovic D, Gotic M, Sefer D, Mitrovic-Ajtic O, Cokic V, Milic N. Predictors of survival and cause of death in patients with essential thrombocythemia. Eur J Haematol 2015;95(5):461-466. Tefferi A, Gangat N, Wolanskyj A. The interaction between leukocytosis and other risk factors for thrombosis in essential thrombocythemia. Blood. 2007; 109(9):4105.

haematologica | 2018; 103(1)


ARTICLE

Myelodysplastic Syndromes

Erythropoietin inhibits osteoblast function in myelodysplastic syndromes via the canonical Wnt pathway

Ferrata Storti Foundation

Ekaterina Balaian,1* Manja Wobus,1* Heike Weidner,1 Ulrike Baschant,2,3 Maik Stiehler,4 Gerhard Ehninger,1 Martin Bornhäuser,1,5 Lorenz C Hofbauer,2,3,5 Martina Rauner2,3** and Uwe Platzbecker1,3,5**

1 Medical Clinic I, University Hospital Carl Gustav Carus Dresden; 2Medical Clinic III, University Hospital Carl Gustav Carus Dresden; 3Center for Healthy Aging, University Hospital Carl Gustav Carus Dresden; 4University Centre for Orthopaedics & Trauma Surgery and Centre for Translational Bone, Joint and Soft Tissue Research, University Hospital Carl Gustav Carus Dresden and 5German Cancer Consortium (DKTK), partner site Dresden and German Cancer Research Center (DKFZ), Heidelberg, Germany

Haematologica 2018 Volume 103(1):61-68

*EB and MW contributed equally to this work.

**MR and UP contributed equally to this work.

ABSTRACT

T

he effects of erythropoietin on osteoblasts and bone formation are controversial. Since patients with myelodysplastic syndromes often display excessively high erythropoietin levels, we aimed to analyze the effect of erythropoietin on osteoblast function in myelodysplastic syndromes and define the role of Wnt signaling in this process. Expression of osteoblast-specific genes and subsequent osteoblast mineralization was increased in mesenchymal stromal cells from healthy young donors by in vitro erythropoietin treatment. However, erythropoietin failed to increase osteoblast mineralization in old healthy donors and in patients with myelodysplasia, whereas the basal differentiation potential of the latter was already significantly reduced compared to that of age-matched controls (P<0.01). This was accompanied by a significantly reduced expression of genes of the canonical Wnt pathway. Treatment of these cells with erythropoietin further inhibited the canonical Wnt pathway. Exposure of murine cells (C2C12) to erythropoietin also produced a dose-dependent inhibition of TCF/LEF promoter activity (maximum at 500 IU/mL, -2.8-fold; P<0.01). The decreased differentiation capacity of erythropoietin-pretreated mesenchymal stromal cells from patients with myelodysplasia could be restored by activating the Wnt pathway using lithium chloride or parathyroid hormone. Its hematopoiesis-supporting capacity was reduced, while reactivation of the canonical Wnt pathway in mesenchymal stromal cells could reverse this effect. Thus, these data demonstrate that erythropoietin modulates components of the osteo-hematopoietic niche in a context-dependent manner being anabolic in young, but catabolic in mature bone cells. Targeting the Wnt pathway in patients with myelodysplastic syndromes may be an appealing strategy to promote the functional capacity of the osteo-hematopoietic niche.

Correspondence: uwe.platzbecker@uniklinikum-dresden.de

Received: May 24, 2017. Accepted: October 18, 2017. Pre-published: October 27, 2017 doi:10.3324/haematol.2017.172726 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/61 Š2018 Ferrata Storti Foundation

Introduction Erythropoietin is a glycoprotein mostly known for its function as a hematopoietic hormone, which is secreted from the adult kidneys in response to hypoxia. Erythropoietin receptor has been detected not only in hematopoietic cells, but also in multiple non-hematopoietic tissues, such as endothelial, skeletal muscle, and neuronal compartments, heart, kidney, pancreas and uterus, suggesting cytoprotective effects on non-erythroid cells.1 In contrast to low concentrations that are sufficient for proper erythropoiesis, other tissues appear to require relatively high concentrations of erythropoietin that are not normally reached in the circuhaematologica | 2018; 103(1)

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

61


E. Balaian et al.

lation.2 The structure of the erythropoietin receptor in some non-hematopoietic cells is distinct from that of the receptor responsible for erythropoiesis, being a heterodimer consisting of the beta common receptor subunit (CD131) in combination with the erythropoietin receptor subunit.2,3. The skeletal system also appears to be influenced by erythropoietin; however, the action of this glycoprotein on bone is still under debate. In most studies, erythropoietin has been shown to stimulate mesenchymal stromal cell (MSC) differentiation towards osteoblasts in vitro4-7 as well as to increase bone formation and the number of osteoblasts in vivo,8-10 especially if ephrinB2/EphB4 signaling is concomitantly activated.6 Moreover, certain observations in patients with fractures suggested accelerated fracture healing after erythropoietin treatment,11 and erythropoietin treatment of hemodialysis patients also improved the formation of bone matrix.12 Typically, the anabolic action of erythropoietin in animal models has been explained through indirect mechanisms, such as an increase in vascularization and tissue oxygenation due to an increase in hemoglobin level. Another possible indirect mechanism represents the concept of coupling hematopoiesis with bone formation through the stimulation of BMP2 secretion by hematopoietic stem and progenitor cells (HSPC).5 The impact of erythropoietin on the communication of osteoblasts and HSPC-derived osteoclasts through the ephrinB2/EphB4 signaling pathway may also explain the stimulation of the osteoblastic phenotype.6 Contradictory observations have been made by other groups,13,14 who have demonstrated bone loss after erythropoietin treatment. In particular, erythropoietin has been shown to directly stimulate osteoclastogenesis both in vitro and in vivo.15 In addition, OPG expression is increased after erythropoietin treatment, suggesting a compensatory mechanism aimed at attenuating the bone resorption.16 It is important to notice that different erythropoietin dosages can lead to opposite effects, as supraphysiological erythropoietin concentrations induced mineralization, whereas moderate concentrations suppressed bone formation and inhibited osteoblast differentiation in mice.17 It is very likely that erythropoietin may have both anabolic and catabolic actions in bone depending on experimental conditions. Moreover, it has been suggested that response to erythropoietin is more robust in younger animals than in older animals.18 Myelodysplastic syndromes (MDS) represent clonal disorders, mainly of the elderly, characterized by ineffective hematopoiesis and an increased risk of transformation into acute myeloid leukemia. The diverse interactions within the osteo-hematopoietic niche in MDS and the potential contribution of the niche to the pathogenesis of MDS have only recently been appreciated.19 Several studies have reported on the dysregulation of the Wnt pathway in MSC from MDS patients, with activation of non-canonical and suppression of canonical Wnt pathway.20,21 Moreover, the methylation status of Wnt antagonist genes has been shown to correlate with a poor prognosis in MDS.22 However, the potential connection of increased erythropoietin levels due to ineffective erythropoiesis or erythropoietin supplementation in MDS patients and a deregulated Wnt pathway has not been yet evaluated. Keeping in mind the contradictory data about the action of erythropoietin on bone in grown vertebrates, and based on the observation that MDS patients, who 62

are mostly elderly and often display excessive erythropoietin levels, as well as osteoporosis,23 we aimed to evaluate the influence of erythropoietin on osteoblasts derived from patients with MDS to clarify the potential association between erythropoietin levels and their effects on bone formation.

Methods Patients MSC were collected from young (22-49 years, both genders) and old (55-89 years, both genders) healthy donors and MDS patients (51-90 years, both genders) following Institutional Review Board approval and having obtained written informed consent. The MDS patients’ characteristics are presented in Online Supplementary Table S1). Heparinized bone marrow samples were obtained at diagnostic bone marrow puncture or during total hip arthroplasty operations.

Culture of human mesenchymal stromal cells The culture conditions of MSC are described in detail in the Online Supplementary Materials and Methods. Cells were treated with erythropoietin alfa (10-100 IU/mL, EPREX, Janssen-Cilag). In most experiments, differentiation medium was applied for 10 days. The dosage of erythropoietin was chosen according to previous publications,3 as well as personal experience showing that 50 IU/mL erythropoietin are required to induce anabolic effects in MSC. In certain experiments, cells were treated with parathyroid hormone intermittently for 8 h three times a week (PTH, 100 ng/mL, Preotact, Nycomed) and/or lithium chloride continuously (25 mM).

Alizarin red staining Osteoblast mineralization was assessed using Alizarin red staining. The details of the method are described in the Online Supplementary Materials and Methods.

RNA isolation, reverse transcription, and real-time polymerase chain reaction Total RNA was isolated with the High Pure RNA Isolation kit (Roche, Mannheim, Germany) according to the manufacturer`s protocol. Five-hundred nanograms of RNA were reverse transcribed using Superscript II (Invitrogen, Darmstadt, Germany) und subsequently used for SYBR green-based real-time polymerase chain reactions (PCR) (Applied Biosystems, Carlsbad, CA, USA). The primer sequences and PCR conditions are listed in the Online Supplementary Materials and Methods.

Wnt profiler polymerase chain reaction array Cells were differentiated with osteogenic medium for 7 days and treated with 50 IU/mL erythropoietin for 24 h. Afterwards, RNA was isolated as described above, reverse transcribed using the RT2 First Strand Kit (SABioscience) and 500 ng cDNA were subjected to the Wnt profiler PCR array (PAHS-043Z) containing 84 Wnt-related genes according to the manufacturer’s protocol (SABioscience). Genes were normalized to the mean of five house-keeping genes (β-actin, GAPDH, B2M, HPRT1, RPLP0).

TCF/LEF-reporter assay A TCF/LEF-reporter assay (Qiagen) was done using the murine myoblast C2C12 cell line, which is commonly used to study BMP and Wnt signaling. These cells were treated with erythropoietin (10-500 IU/mL) for 24 h. Details of this method can be found in the Online Supplementary Material. haematologica | 2018; 103(1)


Erythropoietin inhibits osteoblasts in MDS

Co-culture of CD34+ cells with primary human mesenchymal stromal cells Primary MSC were plated at a density of 1–2 x 104/cm2 in DMEM with 10% fetal calf serum and pre-treated with erythropoietin 50 IU/mL and/or lithium chloride 25 mM for 7 days. Afterwards, CD34+ cells were co-cultured with MSC for 7 days and counted using a hemocytometer. Moreover, a 4-week cobblestone area forming-cell assay was performed with or without pretreatment of the MSC layer. A detailed description of the culture conditions is given in the Online Supplementary Material.

Statistical analysis Statistical analysis was performed using GraphPad Prism software version 5.01 (GraphPad Software, La Jolla, CA, USA). Data are presented as mean ± standard deviation (SD). Statistical evaluations of two group comparisons were performed using a twosided Student t-test or two-way ANOVA test. A P-value of less than 0.05 was regarded as statistically significant.

Results Distinct effects of erythropoietin on differentiation of osteoblasts from young and old healthy donors and patients with myelodysplastic syndromes In order to determine the influence of erythropoietin on the differentiation potential of MSC towards osteoblasts, we treated human MSC from young and old healthy donors, as well as from patients with MDS, with erythropoietin. Using MSC from young donors, we could confirm

A

previously published data3-7,24 and showed an increased mineralization upon erythropoietin treatment, with the strongest induction being seen at concentrations above 50 IU/mL (Figure 1A). Expression of ALP and OPG was also significantly increased up to 1.5-fold after treatment with 50 IU/mL erythropoietin (Figure 1B), whereas the expression of RUNX2 and OSTERIX did not change (data not shown). In contrast, when we evaluated the effect of the same concentration of erythropoietin on MSC from old healthy donors and MDS patients, we did not observe an induction in matrix mineralization (Figure 1C). None of the erythropoietin concentrations ranging from 10 IU/mL to 100 IU/mL was able to increase the mineralization. Furthermore, the lack of effect was also independent of the MDS subtype or risk group (data not shown). In order to exclude the possibility that the lack of effect was due to the absence of erythropoietin receptor in MSC from elderly patients, we determined the level of EPOR expression and found no significant difference between the three groups (Online Supplementary Figure S1). The basal differentiation potential of MDS-MSC was significantly reduced compared to that of MSC from agematched healthy donors (Figure 1C), which is in line with previous reports.20

Erythropoietin inhibits canonical Wnt signaling which is already suppressed in mesenchymal stromal cells from patients with myelodysplastic syndromes Based on previous reports showing deregulated Wnt signaling in MDS patients and the prominent role of this

B

C

Figure 1. Different effects of erythropoietin on differentiation of osteoblasts from young and old healthy donors and patients with myelodysplastic syndromes. Human mesenchymal stromal cells from young healthy donors (A-B), MDS patients and age-matched healthy donors (C) were differentiated towards osteoblasts in the presence of various concentrations of erythropoietin (Epo). The mineralization was visualized with Alizarin red S staining and quantified after elution with cetylpyridinium chloride (A, C). Gene expression analysis of alkaline phosphatase (ALP) and osteoprotegerin (OPG) using real-time polymerase chain reaction in MSC from young healthy donors after treatment with 50 IU/mL Epo for 10 days (B). N=3-5. *P<0.05, **P<0.01, ns – not significant vs. control (CO).

haematologica | 2018; 103(1)

63


E. Balaian et al.

pathway in osteoblastic differentiation, we compared the expression of Wnt-related genes in differentiated osteoblasts from MDS patients and age-matched healthy donors using a PCR-based Wnt array. A change of at least 2-fold in gene expression between patients and the control group was considered significant. Transcripts of 18 genes out of 84 Wnt-related genes were differentially expressed between patients and controls. Four genes were validated using quantitative PCR (Figure 2A; Online Supplementary Table S2). To validate the data obtained from the PCR array, we evaluated the mRNA expression of differentially regulated genes, namely, target genes of the canonical Wnt pathway JUN, FOSL1 and CCND1 and a receptor of the canonical Wnt pathway FZD4 (Online Supplementary Figure S2). In accordance with the PCR array data, the mean relative mRNA expression of genes related to the canonical Wnt pathway was significantly downregulated in MDS patients compared to controls. Additionally, there was a tendency to an increased concentration of the Wnt inhibitor Dkk1 in the MSC culture supernatants from the patients (data not shown). We then treated MDS-MSC with erythropoietin to evaluate whether it could influence the canonical Wnt pathway, even in the absence of its action on osteoblastic differentiation. Interestingly, we observed a strong upregulation of GSK3B, which is a known inhibitor of the canonical Wnt pathway, whereas MMP7 and CCND1 â&#x20AC;&#x201C; target genes of the canonical Wnt pathway â&#x20AC;&#x201C; were downregulated (Figure 2A; Online Supplementary Table S3). In addition to gene expression analyses, we also examined the effect of erythropoietin on Wnt promoter activity using a TCF/LEF promoter assay. Wnt promoter activity was induced in C2C12 cells using the supernatants from Wnt3a-overexpressing L-cells. After treatment with increasing concentrations of erythropoietin, the TCF/LEF promoter activity was significantly reduced at high concentrations (100 IU/mL) (Figure 2B). Hence, erythropoi-

A

etin targets several components of the canonical Wnt pathway, which most likely contributes to its inhibition in MDS-MSC.

Activators of the canonical Wnt pathway can restore the attenuated osteoblast differentiation of erythropoietin-treated mesenchymal stromal cells from patients with myelodysplastic syndromes To further explore the role of a deregulated Wnt pathway in MDS-MSC after erythropoietin treatment, we evaluated osteoblast differentiation following reactivation of canonical Wnt signaling. We, therefore, used lithium chloride due to its known ability to inhibit GSK3B,25 which was significantly upregulated in erythropoietintreated osteoblasts. In addition, intermittent parathyroid hormone treatment was used, as this hormone is applied therapeutically as a bone-forming agent in severely osteoporotic patients and exerts its effects, among other ways, through stimulating the canonical Wnt pathway.26 Parathyroid hormone further increased matrix mineralization in osteoblasts from young healthy donors treated with erythropoietin (Figure 3A). In erythropoietin-treated MDS-MSC, parathyroid hormone also improved the mineralization capacity (Figure 3B). A similar stimulatory effect was observed after concomitant treatment of erythropoietin with lithium chloride, which indicates that reactivation of the canonical Wnt pathway improves the differentiation capacity of osteoblasts from MDS patients.

Hematopoietic support by erythropoietin-pretreated mesenchymal stromal cells One of the main functions of MSC in the bone marrow is the support of hematopoiesis. Disturbances in its function as a consequence of excessive erythropoietin levels can also affect the support of HSPC. We, therefore, performed co-culture experiments with HSPC and MSC from healthy individuals or MDS patients pretreated with

B

Figure 2. Erythropoietin inhibits canonical Wnt signaling, which is already suppressed in mesenchymal stromal cells from patients with myelodysplastic syndromes. (A) Human mesenchymal stromal cells from old healthy donors and MDS patients were compared using Wnt profiler polymerase chain reaction (white bars). Afterwards, osteoblasts from MDS patients were treated with 50 IU/mL erythropoietin (Epo) for 24 h and were compared to untreated cells (black bars). All genes presented in the bar graph were significantly regulated (P<0.05). N=3. (B) Murine myoblast C2C12 cell line was transfected with the Signal TCF/LEF Reporter and treated with increasing Epo concentrations for 24 h. Luciferase activity was assayed 24 h after treatment. N=4. *P<0.05, **P<0.01 vs. control.

64

haematologica | 2018; 103(1)


Erythropoietin inhibits osteoblasts in MDS

Discussion

erythropoietin with or without lithium chloride, as a reactivator of the deranged canonical Wnt pathway. After 7 days of co-culture with erythropoietin-pretreated MSC, we observed a tendency for a reduction of CD34+ HSPC (Figure 4A,B), whereas CD38 expression was slightly increased or not affected in most cases suggesting a differentiation-supportive effect of erythropoietin. When the canonical Wnt pathway was reactivated in MSC using lithium chloride the numbers of CD34+ and CD38+ cells were mostly expanded and even surpassed the initial level without MSC pretreatment in some cases (Figure 4A,B). CD90 (Thy-1), which marks more immature HSPC, was not affected by MSC pre-treatment (data not shown). Next, we investigated the colony potential of pre-treated MSC in long-term co-cultures with CD34+ HSPC. The number of cobblestone area forming-cells as readout for active stromal support was significantly reduced in erythropoietin-pretreated MSC from young and old healthy donors (8.25±1.89 versus 3.5±1.9 and 8.0±1.4 versus 3.5±0.7) whereas the already diminished potential in MDS MSC was only slightly decreased (4.2±0.83 versus 2.2±0.84). Interestingly, lithium chloride completely restored the hematopoietic support capacity, which even rose above the basal level (12.0±1.4/13.0±1.5/7.0±2.4) (Figure 4C). To study the clonogenic potential of HSPC cultured on pretreated MSC layers, a colony-forming unit assay was performed. Compared to controls, the total number of colonies was significantly reduced in co-cultures with erythropoietin-pretreated MSC from all three groups. Again, lithium chloride abrogated this effect and the number of colonies was comparable to those of untreated controls (Figure 4D-F). Moreover, significant differences could be detected for clonogenic progenitors of granulocytes and macrophages of MSC from old healthy donors and MDS patients, respectively (Figure 4E,F).

Erythropoietin and erythropoiesis-stimulating agents are among the main treatment options for anemia in patients with low-risk MDS.27 However, the erythroid response is observed in a relatively low number of patients and often not sustained,28 which is in part due to endogenously elevated erythropoietin levels because of ineffective erythropoiesis in many patients.29 We speculated that a sustained endogenous elevation of erythropoietin levels or iatrogenic erythropoietin therapy in MDS patients might contribute to the dysregulation of the osteo-hematopoietic niche, at least partly due to its effects on MSC and osteoblasts. In fact, we could confirm previous publications, which postulated the stimulation of osteoblastic differentiation of MSC when exposed to erythropoietin.4-7 However, we could not observe the same effect in MSC from older healthy donors or patients with MDS, although the level of EPOR expression did not differ. Of note, effective erythropoietin concentrations in our experiments were higher than previously published, which can be explained by a different source of cells, as we used primary donor material cultured in our laboratory, whereas in other papers rodent cells5-7,24 or commercially available MSC3,4 were used. Interestingly, effects were only seen when we used the α-form of erythropoietin, whereas we could not observe the biological activity in MSC using a mixture of different erythropoietin isoforms (data not shown). This raises the question whether the effect of endogenous erythropoietin on bone metabolism in patients may differ from that of exogenous erythropoietin treatment in MDS and other disorders, which requires further studies. Further, the basal rate of osteoblast differentiation was significantly lower in MDS patients, which was associated with the downregulation of the canonical Wnt pathway.20 The genes coding for the Wnt receptor FZD4 as well as for

A

B

Figure 3. Activators of the canonical Wnt pathway can restore the attenuated osteoblastic differentiation of erythropoietin-treated mesenchymal stromal cells from patients with myelodysplastic syndromes. Human mesenchymal stromal cells from young healthy donors (A) and MDS patients (B) were differentiated towards osteoblasts in the presence of 50 IU/ml erythropoietin (Epo) with intermittent 100 ng/mL parathyroid hormne (PTH) or 25 mM lithium chloride (LiCl). The mineralization was visualized with Alizarin red S staining and quantified after elution with cetylpyridinium chloride. N=3-5. *P<0.05, **P<0.01, ***P<0.001 vs. control (CO).

haematologica | 2018; 103(1)

65


E. Balaian et al.

target genes of the canonical Wnt pathway (JUN,30 FOSL1,31 CCND1) were suppressed in the MDS group. Interestingly, another study found an upregulation of JUN and CCND1 in MDS-MSC compared to the control group.21 However, in that study, patients with lymphoproliferative disorders without bone marrow involvement were selected as a control group, which may influence the results due to the known role of the Wnt pathway in lymphocyte development.32 We considered patients without hematological disorders who underwent total hip arthroplasty as a control age-matched group, which allowed us to exclude the bias due to hematological diseases. Whereas the signaling pathways activated upon erythropoietin treatment in HSPC are relatively well known, signaling cascades in non-hematopoietic cells including

osteoblasts are still a matter of debate. Activation of mTOR, AKT,4 Jak2 and PI3K3 has been shown to be involved in the anabolic action of erythropoietin in human MSC. Activation of the canonical Wnt pathway was also detected upon erythropoietin treatment in MSC during neuronal differentiation33 and kidney epithelial cells.34 Despite the inability of erythropoietin to promote the differentiation of MDS-MSC, we observed a molecular derangement in the canonical Wnt pathway following erythropoietin treatment. As such, GSK3B was strongly upregulated, whereas target genes of the canonical Wnt pathway (MMP7, CCND1) were downregulated. Supporting this observation, we demonstrated the inhibition of the canonical Wnt pathway in the C2C12 myoblast cell line upon erythropoietin treatment.

A

B

C

D

E

F

Figure 4. The hematopoietic support by erythropoietin-treated mesenchymal stromal cells is inhibited but can be restored by Wnt pathway activation. Human mesenchymal stromal cells (MSC) were pretreated with erythropoietin (Epo) 50 IU/mL and/or lithium chloride (LiCl) 25 mM for 7 days and co-cultured with freshly isolated CD34+ HSPC. After 7 days, flow cytometric analysis was performed for the non-adherent cells. (A) Number of CD34+ cells and (B) number of CD38+ cells. Each dot represents the percentage of positive-stained cells; each graph represents one donor/patient. (C) After 4 weeks of co-culture, the number of cobblestone areaforming cells (CAF-C) was determined in each well. A colony-forming unit (CFU) assay was performed in methylcellulose medium for 2 additional weeks and the colonies were classified under a microscope for HSPC co-cultured with (D) young healthy MSC, (E) old healthy MSC and (F) MDS MSC. N=3-5. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001

66

haematologica | 2018; 103(1)


Erythropoietin inhibits osteoblasts in MDS

However, this effect was detected only at high erythropoietin concentrations (starting from 100 IU/mL) reflecting different cellular sensitivities to erythropoietin. It is already known, that pharmacological activation of canonical Wnt signalling in MDS-MSC with a GSK3B inhibitor increases their proliferative potential and upregulates the expression of early osteoblastic genes.20 We evaluated if the same effect could be observed when reactivation of the canonical Wnt pathway is reached by compounds used in clinical practice, such as lithium salts used in bipolar disorders and, especially, parathyroid hormone, approved for the treatment of osteoporosis. Both medications in standard in vitro concentrations showed their effectiveness in restoration of osteoblastic differentiation of erythropoietin-treated MDS-MSC. Importantly, it is already known that parathyroid hormone does not further increase erythropoietin levels in vivo,5 which is relevant, based on the knowledge that excessive erythropoietin concentrations negatively influence the osteo-hematopoietic microenvironment. Interestingly, we could recently show an increased risk of osteoporosis in patients with MDS.23 Pretreatment of MSC with erythropoietin reduced the number of early CD34+ progenitors. Therefore, because MDS-MSC already have a reduced capacity to support myeloid and erythroid colony forming potential,20,35 excessive erythropoietin levels could further deteriorate hematopoiesis. In turn, reactivation of the canonical Wnt pathway in MSC can restitute or even surpass the initial number of CD34+ cells and colony potential in the co-cul-

References 1. Ogunshola OO, Bogdanova AY. Epo and non-hematopoietic cells: what do we know? Methods Mol Biol. 2013;982:13-41. 2. Brines M, Cerami A. The receptor that tames the innate immune response. Mol Med. 2012;18:486-496. 3. Rolfing JH, Baatrup A, Stiehler M, Jensen J, Lysdahl H, Bunger C. The osteogenic effect of erythropoietin on human mesenchymal stromal cells is dose-dependent and involves non-hematopoietic receptors and multiple intracellular signaling pathways. Stem Cell Rev. 2014;10(1):69-78. 4. Kim J, Jung Y, Sun H, et al. Erythropoietin mediated bone formation is regulated by mTOR signaling. J Cell Biochem. 2012;113(1):220-228. 5. Shiozawa Y, Jung Y, Ziegler AM, et al. Erythropoietin couples hematopoiesis with bone formation. PloS One. 2010;5(5): e10853. 6. Li C, Shi C, Kim J, et al. Erythropoietin promotes bone formation through EphrinB2/EphB4 signaling. J Dent Res. 2015;94(3):455-463. 7. Nair AM, Tsai YT, Shah KM, et al. The effect of erythropoietin on autologous stem cell-mediated bone regeneration. Biomaterials. 2013;34(30):7364-7371. 8. Rolfing JH, Bendtsen M, Jensen J, et al. Erythropoietin augments bone formation in a rabbit posterolateral spinal fusion model. J Orthop Res. 2012;30(7):1083-1088. 9. Mihmanli A, Dolanmaz D, Avunduk MC,

haematologica | 2018; 103(1)

10.

11.

12.

13.

14.

15.

16.

ture. Thus, in addition to its positive influence on bone metabolism and known effectiveness in osteoporosis, activators of the canonical Wnt pathway might also improve hematopoiesis in MDS patients, potentially in those with endogenously elevated erythropoietin levels. In conclusion, erythropoietin failed to stimulate MSC from patients with MDS to differentiate towards osteoblasts, at least in part due to the inhibition of the canonical Wnt pathway, which is intrinsically less activated in MDS-MSC compared to age-matched healthy controls, suggesting age-dependent erythropoietin resistance. Reactivation of canonical Wnt signaling does not only restore osteoblastic differentiation capacity, but also promotes hematopoiesis, which provides a potential rationale for the therapeutic use of Wnt-activators in MDS patients. Acknowledgment The authors would like to thank Eva Schubert, Patrick Böhme, Nicole Pacyna, Marie-Christin Mehnert, Anja Liebkopf, and Ivonne Habermann for technical assistance. Funding UP and LCH are supported by a grant within the German Consortium of translational cancer research (DKTK), GE/UP/MB/LCH are supported by the Sonderforschungsbereich (SFB) 655 from the Deutsche Forschungsgemeinschaft (DFG), UP/MR are supported by a grant from the Deutsche José Carreras Leukämie-Stiftung and LCH/MR are supported by a seed grant from the CRTD. The work of EB has been supported by a Gerok rotation position within the SFB 655.

Erdemli E. Effects of recombinant human erythropoietin on mandibular distraction osteogenesis. J Oral Maxillofac Surg. 2009;67(11):2337-2343. Sun H, Jung Y, Shiozawa Y, Taichman RS, Krebsbach PH. Erythropoietin modulates the structure of bone morphogenetic protein 2-engineered cranial bone. Tissue Eng. 2012;18(19-20):2095-2105. Bakhshi H, Kazemian G, Emami M, Nemati A, Karimi Yarandi H, Safdari F. Local erythropoietin injection in tibiofibular fracture healing. Trauma Mon. 2013;17(4):386-388. Takenaka T, Itaya Y, Ishikawa I, Kobayashi K, Tsuchiya Y. Skeletal effects of erythropoietin in hemodialysis patients. Int Urol Nephrol. 2003;35(3):407-413. Singbrant S, Russell MR, Jovic T, et al. Erythropoietin couples erythropoiesis, Blymphopoiesis, and bone homeostasis within the bone marrow microenvironment. Blood. 2011;117(21):5631-5642. Dewamitta SR, Russell MR, Nandurkar H, Walkley CR. Darbepoietin-alfa has comparable erythropoietic stimulatory effects to recombinant erythropoietin whilst preserving the bone marrow microenvironment. Haematologica. 2013;98(5):686-690. Hiram-Bab S, Liron T, Deshet-Unger N, et al. Erythropoietin directly stimulates osteoclast precursors and induces bone loss. FASEB J. 2015;29(5):1890-1900. Deshet-Unger N, Hiram-Bab S, HaimOhana Y, Mittelman M, Gabet Y, Neumann D. Erythropoietin treatment in murine multiple myeloma: immune gain and bone loss. Sci Rep. 2016;6:30998.

17. Rauner M, Franke K, Murray M, et al. Increased EPO levels are associated with bone loss in mice lacking PHD2 in EPOproducing cells. J Bone Miner Res. 2016;31(10):1877-1887. 18. McGee SJ, Havens AM, Shiozawa Y, Jung Y, Taichman RS. Effects of erythropoietin on the bone microenvironment. Growth factors. 2012;30(1):22-28. 19. Bulycheva E, Rauner M, Medyouf H, et al. Myelodysplasia is in the niche: novel concepts and emerging therapies. Leukemia. 2015;29(2):259-268. 20. Pavlaki K, Pontikoglou CG, Demetriadou A, et al. Impaired proliferative potential of bone marrow mesenchymal stromal cells in patients with myelodysplastic syndromes is associated with abnormal WNT signaling pathway. Stem Cells Dev. 2014;23(14): 1568-1581. 21. Falconi G, Fabiani E, Fianchi L, et al. Impairment of PI3K/AKT and WNT/betacatenin pathways in bone marrow mesenchymal stem cells isolated from patients with myelodysplastic syndromes. Exp Hematol. 2016;44(1):75-83. 22. Wang H, Fan R, Wang XQ, et al. Methylation of Wnt antagonist genes: a useful prognostic marker for myelodysplastic syndrome. Ann Hematol. 2013;92(2): 199-209. 23. Weidner H, Rauner M, Trautmann F, et al. Myelodysplastic syndromes and bone loss in mice and men. Leukemia. 2017;31(4): 1003-1007. 24. Guo L, Luo T, Fang Y, et al. Effects of erythropoietin on osteoblast proliferation and

67


E. Balaian et al. function. Clin Exp Med. 2014;14(1):69-76. 25. Galli C, Piemontese M, Lumetti S, Manfredi E, Macaluso GM, Passeri G. GSK3b-inhibitor lithium chloride enhances activation of Wnt canonical signaling and osteoblast differentiation on hydrophilic titanium surfaces. Clin Oral Implants Res. 2013;24(8):921-927. 26. Chandra A, Lin T, Zhu J, et al. PTH1-34 blocks radiation-induced osteoblast apoptosis by enhancing DNA repair through canonical Wnt pathway. J Biol Chem. 2015;290(1):157-167. 27. Hellstrom-Lindberg E, van de Loosdrecht A. Erythropoiesis stimulating agents and other growth factors in low-risk MDS. Best Pract Res Clin Haematol. 2013;26(4):401-410. 28. Park S, Grabar S, Kelaidi C, et al. Predictive factors of response and survival in

68

29.

30.

31.

32.

myelodysplastic syndrome treated with erythropoietin and G-CSF: the GFM experience. Blood. 2008;111(2):574-582. Suzuki T, Oh I, Ohmine K, et al. Distribution of serum erythropoietin levels in Japanese patients with myelodysplastic syndromes. Int J Hematol. 2015;101(1):32-36. Sassi N, Laadhar L, Allouche M, et al. The roles of canonical and non-canonical Wnt signaling in human de-differentiated articular chondrocytes. Biotech Histochem. 2014;89(1):53-65. Heo JS, Lee SY, Lee JC. Wnt/beta-catenin signaling enhances osteoblastogenic differentiation from human periodontal ligament fibroblasts. Mol Cells. 2010;30(5):449-454. Lento W, Congdon K, Voermans C, Kritzik M, Reya T. Wnt signaling in normal and malignant hematopoiesis. Cold Spring

Harb Perspect Biol. 2013;5(2). 33. Danielyan L, Schafer R, Schulz A, et al. Survival, neuron-like differentiation and functionality of mesenchymal stem cells in neurotoxic environment: the critical role of erythropoietin. Cell Death Differ. 2009;16(12):1599-1614. 34. Chen X, Wang CC, Song SM, et al. The administration of erythropoietin attenuates kidney injury induced by ischemia/reperfusion with increased activation of Wnt/betacatenin signaling. J Formos Med Assoc. 2015;114(5):430-437. 35. Ferrer RA, Wobus M, List C, et al. Mesenchymal stromal cells from patients with myelodyplastic syndrome display distinct functional alterations that are modulated by lenalidomide. Haematologica. 2013;98(11):1677-1685.

haematologica | 2018; 103(1)


ARTICLE

Myelodysplastic Syndromes

Labile plasma iron levels predict survival in patients with lower-risk myelodysplastic syndromes

Louise de Swart,1 Chloé Reiniers,2 Timothy Bagguley,3 Corine van Marrewijk,1 David Bowen,4 Eva Hellström-Lindberg,5 Aurelia Tatic,6 Argiris Symeonidis,7 Gerwin Huls,2 Jaroslav Cermak,8 Arjan A. van de Loosdrecht,9 Hege Garelius,10 Dominic Culligan,11 Mac Macheta,12 Michail Spanoudakis,13 Panagiotis Panagiotidis,14 Marta Krejci,15 Nicole Blijlevens,1 Saskia Langemeijer,1 Jackie Droste,1 Dorine W. Swinkels,16 Alex Smith2 and Theo de Witte17 on behalf of the EUMDS Steering Committee

Ferrata Storti Foundation

Haematologica 2018 Volume 103(1):69-79

Department of Hematology, Radboud university medical center, Nijmegen, the Netherlands; 2Department of Hematology, University Medical Centre, Groningen, the Netherlands; 3Epidemiology and Cancer Statistics Group, University of York, UK; 4St. James's Institute of Oncology, Leeds Teaching Hospitals, UK; 5Department of Medicine, Division of Hematology, Karolinska Institutet, Stockholm, Sweden; 6Center of Hematology and Bone Marrow Transplantation, Fundeni Clinical Institute, Bucharest, Romania; 7Department of Medicine, Division of Hematology, University of Patras Medical School, Greece; 8Department of Clinical Hematology, Institute of Hematology & Blood Transfusion, Prague, Czech Republic; 9Department of Hematology – Cancer Center Amsterdam VU University Medical Center, The Netherlands; 10Department of Medicine, Section of Hematology and Coagulation, Sahlgrenska University Hospital, Göteborg, Sweden; 11Department of Haematology, Aberdeen Royal Infirmary, UK; 12 Department of Haematology, Blackpool Victoria Hospital, Lancashire, UK; 13 Department of Haematology, Airedale NHS trust, UK; 14Department of Hematology, Laikon General Hospital, National and Kapodistrian University of Athens, Greece; 15 Department of Internal Medicine, Hematology and Oncology, University Hospital Brno and Masaryk University, Czech Republic; 16Department of Laboratory Medicine, Hepcidinanalysis.com, and Radboudumc expertise center for iron disorders, Radboud university medical center, Nijmegen, the Netherlands and 17Nijmegen Center for Molecular Life Sciences, Department of Tumor Immunology, Radboud university medical center, the Netherlands 1

ABSTRACT

R

ed blood cell transfusions remain one of the cornerstones in supportive care of lower-risk patients with myelodysplastic syndromes. We hypothesized that patients develop oxidant-mediated tissue injury through the formation of toxic iron species, caused either by red blood cell transfusions or by ineffective erythropoiesis. We analyzed serum samples from 100 lower-risk patients with myelodysplastic syndromes at six-month intervals for transferrin saturation, hepcidin-25, growth differentiation factor 15, soluble transferrin receptor, non-transferrin bound iron and labile plasma iron in order to evaluate temporal changes in iron metabolism and the presence of potentially toxic iron species and their impact on survival. Hepcidin levels were low in 34 patients with ringed sideroblasts compared to 66 patients without. Increases of hepcidin and non-transferrin bound iron levels were visible early in follow-up of all transfusion-dependent patient groups. Hepcidin levels significantly decreased over time in transfusion-independent patients with ringed sideroblasts. Increased soluble transferrin receptor levels in transfusion-independent patients with ringed sideroblasts confirmed the presence of ineffective erythropoiesis and suppression of hepcidin production in these patients. Detectable labile plasma iron levels in combination with high transferrin saturation levels occurred almost exclusively in patients with ringed sideroblasts and all transfusiondependent patient groups. Detectable labile plasma iron levels in transfusion-dependent patients without ringed sideroblasts were associated with decreased survival. In conclusion, toxic iron species occurred in all transfusion-dependent patients and in transfusion-independent patients with ringed sideroblasts. Labile plasma iron appeared to be a clinically relevant measure for potential iron toxicity and a prognostic factor for survival in transfusion-dependent patients. clinicaltrials.gov Identifier: 00600860. haematologica | 2018; 103(1)

Correspondence: theo.dewitte@radboudumc.nl

Received: May 26, 2017. Accepted: October 27, 2017. Pre-published: November 9, 2017. doi:10.3324/haematol.2017.171884 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/69 ©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.

69


L. de Swart et al.

Introduction Myelodysplastic syndromes (MDS) are a heterogeneous group of acquired clonal hematopoietic stem cell disorders that are characterized by abnormal differentiation and maturation of hematopoietic cells, bone marrow failure and genetic instability with an enhanced risk of progression to acute myeloid leukemia.1 The European MDS (EUMDS) registry is a prospective, observational registry which was established in 2007 in order to collect data on low and intermediate-1-risk MDS patients, who represent the lower-risk MDS population, comprising approximately seventy percent of the overall MDS population.2,3 The majority of lower-risk MDS patients (51% in the EUMDS Registry)3 become transfusion-dependent, usually early after diagnosis. With an expected median survival of 2.4 to 11.8 years, these patients are prone to long-term accumulation of iron due to red blood cell (RBC) transfusions.4-8 Iron overload may also occur in MDS patients who do not receive RBC transfusions, due to the stimulation of intestinal iron absorption, mediated through the suppression of hepcidin production in patients with ineffective erythropoiesis.9 Patients with ringed sideroblasts (MDS-RS) are of special interest in this context, considering their pronounced ineffective erythropoiesis.6,7,10,11 The toxic effects of iron overload in other iron loading diseases, such as hereditary hemochromatosis11 and the thalassemia syndromes12 are well known, but the consequences in MDS remain to be elucidated. MDS patients are generally older than patients with other iron loading disorders.13 Their exposure may not be long enough to develop classical tissue damage due to iron overload, but they may suffer from oxidative stress caused by toxic iron molecules. Moreover, iron toxicity might be restricted to specific subgroups of MDS patients; those receiving RBC transfusions and a subgroup of patients with MDS-RS and increased ineffective erythropoiesis.5,13

A greater insight into the pathophysiology of iron metabolism in MDS might be obtained through an optimized diagnostic work-up and monitoring by specific iron metabolism markers, including hepcidin, growth differentiation factor 15 (GDF15), soluble transferrin receptor (sTFR), and the recently introduced serum toxic iron species, namely non-transferrin bound iron (NTBI) and labile plasma iron (LPI).14-18 The most important regulator of systemic iron metabolism is hepcidin, a 25-aminoacid peptide hormone, produced predominantly by the hepatocytes. Hepcidin triggers internalization and lysosomal degradation of ferroportin, a membrane bound cellular iron exporter present on macrophages and the basolateral site of enterocytes that releases iron into the circulation.19,20 Hepcidin is suppressed in hypoxia and with increased erythropoietic iron demand and is upregulated in case of inflammation and increased circulating iron levels and elevated body iron stores.5,20,21 GDF15 is a protein produced by erythroid precursors and has been reported to be involved in the communication between bone marrow and liver in case of an increased erythroid demand, functioning as a suppressor of hepcidin synthesis, as shown for β-thalassemia.5,9,22 However, its role in MDS is still a matter of debate due to conflicting results.5,11,22-25 Twisted gastrulation factor 1 (TWSG1) and erythroferrone (ERFE) are also reported to have a suppressive function in hepatic hepcidin production, however, validated human assays are not available.9 Of additional interest in iron homeostasis is sTFR. The serum concentration of sTFR is proportional to the quantity of the transferrin receptors 1 (TfR1) on cellular membranes, especially on erythroid precursors, and is a valuable parameter of erythroid mass and iron supplies.26,27 Among others, sTFR levels are elevated in case of high erythroid proliferation rates, especially in combination with adequate iron supply,27 as in diseases characterized by ineffective erythropoiesis, such as β-thalassemia syn-

Table 1. Frequency, median and quartiles of iron substudy parameters overall, by transfusion status and MDS subtype at first sample.

N Hemoglobin (g/dl) White blood cells (109/L) Platelets (109//L) Serum Iron (µmol/L) Ferritin (μg/L) Transferrin saturation (%) Hepcidin (nmol/L) Soluble transferrin receptor (mg/L) C-reactive protein (mg/L) Non transferrin bound iron (µmol/L) Labile plasma iron (μmol/L) Growth differentiation factor 15 (ng/L)

70

100 100 99 100 100 100 99 100 100 100 100 100

Total Median (p10-p90) 10.2 (8.3 - 12.4) 4.8 (2.4 - 8.7) 212 (94 - 475) 20 (12 - 38) 287 (48 - 982) 36 (19 - 87) 4.5 (1.1 - 21.7) 1.3 (0.7 - 2.8) 5.0 (4.0 - 11.5) 0.7 (0.1 - 3.0) 0.1 (0.0 - 0.2) 2193 (952 - 5663)

N

Transfusion Independent Dependent Median N Median (p10-p90) (p10-p90)

85 10.3 (8.6 - 12.6) 85 5.1 (2.5 - 8.7) 84 218 (97 - 475) 85 19 (12 - 34) 85 264 (49 - 692) 85 35 (19 - 81) 84 4.2 (1.2 - 13.8) 85 1.3 (0.8 - 2.8) 85 5.0 (4.0 - 11.0) 85 0.6 (0.1 - 2.9) 85 0.1 (0.0 - 0.2) 85 2140 (921 - 6084)

15 9.3 (6.4 - 10.9) 15 3.8 (2.3 - 10.7) 15 158 (87 - 463) 15 26.0 (4.0 - 47.0) 15 634 (20 - 1897) 15 52 (13 - 93) 15 6.8 (0.5 - 53.7) 15 0.9 (0.6 - 3.0) 15 5.0 (4.0 - 139.0) 15 1.0 (0.1 - 3.4) 15 0.1 (0.0 - 0.3) 15 2823 (1232 - 5026)

N

Ring Sideroblasts No Yes Median N Median (p10-p90) (p10-p90)

66 10.4 (8.5 - 12.5) 66 3.9 (2.3 - 7.4) 66 168 (89 - 341) 66 17 (10 - 26) 66 246 (36 - 665) 66 31 (17 - 61) 66 4.7 (1.1 - 24.2) 66 1.2 (0.7 - 2.7) 66 5.0 (4.0 - 13.0) 66 0.5 (0.1 - 1.8) 66 0.1 (0.0 - 0.2) 66 1844 (921 - 4828)

34 34 33 34 34 34 33 34 34 34 34 34

9.9 (7.3 - 12.1) 6.0 (3.9 - 11.4) 316 (169 - 501) 30 (16 - 45) 376(127 -1242) 59 (25 - 93) 4.2 (1.2 - 10.3) 1.5 (0.8 - 3.1) 5.0 (4.0 - 10.0) 1.2 (0.3 - 3.8) 0.1 (0.0 - 0.3) 2888 (1026 -10361)

haematologica | 2018; 103(1)


Toxic iron species in lower-risk MDS

dromes, and levels are suppressed in case of decreased erythropoietic activity, as in anemia of chronic disease, and diseases with erythroid hypoplasia.20,25,28 Earlier studies showed that sTFR levels are increased in MDS-RS,5 including SF3B1-positive MDS patients.11 NTBI concentrations are only sporadically present with transferrin saturations (TSAT) <70% and increase sharply when the saturation of transferrin with iron exceeds 70%.29 Chemically, NTBI consists of iron that is rather loosely bound to albumin or low molecular weight metal complexing groups.30,31 The NTBI complexes may be taken up by specific NTBI transporters in the liver, pancreas, and heart and contribute to oxidant-mediated cellular injury in these tissues.17,32 LPI is thought to be the NTBI fraction that is mostly responsible for tissue injury, since it is readily available to participate in redox cycling, causing oxidative damage to cellular membranes, proteins and DNA.15,33 It has been proposed that plasma NTBI is an important early indicator of extra-hepatic iron toxicity in β-thalassemia major.34,35 Improved insights in the levels and roles of key players of iron metabolism during treatment with transfusions in the various MDS subtypes may provide leads for novel diagnostic and iron reducing treatment strategies. The prospective study of the EUMDS registry was initiated in order to provide a better understanding of the pathophysiology and prognostic value of iron overload and ironmediated oxidative stress as well as possibly important

markers in iron homeostasis over time in MDS. To this end, we evaluated serum ferritin, iron, transferrin saturation, hepcidin-25, GDF15, sTFR, NTBI and LPI levels over time in lower-risk MDS patients and their relation with regard to the World Health Organization (WHO) 2001 subtype and transfusion history. We identified detectable LPI levels as a new important prognostic factor for survival in patients with MDS-RS or lower-risk MDS patients treated with regular RBC transfusions.

Methods Study design and participants Patients were eligible to be included in the EUMDS registry if they were newly diagnosed with MDS according to the WHO 2001 classification and a low or intermediate-1 score according to the International Prognostic Scoring System (IPSS). Two patients with IPSS intermediate-2 or high-risk patients with secondary or therapyrelated MDS were excluded from this registry. The ethics committees of all participating countries and centers approved the protocol. Patients were required to provide written informed consent. Serum samples were collected prospectively, at registration and at 6-month intervals, from 109 patients from six countries who participated in this study from April 2008 to December 2010. Samples from nine patients had to be excluded due to technical reasons, see Online Supplementary Information for details. The total number of analyzed serum samples was 454.

A

C

B

D

Figure 1. LPI and NTBI correlated to TSAT and ferritin in different patient groups. (A) Relation between LPI and TSAT. (B) Relation between NTBI and TSAT. (C) Relation between LPI and ferritin. (D) Relation between NTBI and ferritin. Each dot represents one sample (median: 5 samples/patient). RS: ring sideroblastic; TI: transfusion-independent; TD: transfusion-dependent.

haematologica | 2018; 103(1)

71


L. de Swart et al.

Biochemical assays The iron parameters in this substudy were analyzed centrally at the department of Laboratory Medicine of the Radboudumc, Nijmegen, The Netherlands. Detailed information regarding these iron parameters is described in the Online Supplementary Information. Measurement of serum NTBI consisted of the chelation-ultrafiltration-detection approach based on the prior mobilization of serum NTBI by weak iron-mobilizing chelators, such as nitrilotriacetate (NTA), at 80 mM. The chelated NTBI was separated from transferrin-bound iron by ultrafiltration and detected by colorimetry.36 The lower limit of detection (LLOD) of the NTBI assay was 0.47 μmol/L. The LPI measurement was based on the measurement of the redox-active and readily chelatable fraction of NTBI. This assay measures iron-catalyzed radical generation in the presence of a low ascorbate concentration. Radical generation was measured with the fluorogenic redox sensitive probe dihydrorhodamine (DHR) 123, and iron-catalyzed radical generation was calculated by subtracting the radical generation in the presence of 50 μmol/L of the bidendate iron chelator deferiprone (DFO, the LPI DHR oxidation that is NOT iron dependent).37 The LLOD of the LPI assay was 0.24 μmol/L.

Statistical analysis Standard descriptive techniques were used to assess the association between the iron parameters including Spearman's rank correlation coefficients. Where NTBI or LPI was below LLOD, values were randomly drawn from a univariate distribution in the range from zero to the LLOD. Overall survival (OS) was defined as the time from date of diagnosis to death, or for subjects still alive and censored, to the date of the last visit when a sample was available. Cox proportional hazards regression models and Kaplan–Meier survival curves with time-dependent covariates38 were used in time-to-event analyses to assess the impact of LPI levels, NTBI and TSAT by transfusion status on survival. All variables were treated as time-varying covariates in the model by assessing the levels of the parameters (LPI, NTBI: <LLOD vs. elevated, TSAT <80% vs. ≥80%) and transfusion status (transfused vs. not transfused) at each visit. LPI and NTBI levels >LLOD were considered abnormal. Once a subject had received a transfusion, they were classified as transfused for the remaining time. Hazard ratios (HR) and 95% confidence intervals (95% CI) are reported for both univariate and multivariate models. In the case of multivariate analyses, the additional covariates included were age at diagnosis, IPPS-revised (IPSS-R) category and usage of erythroid stimulating agents (ESA). All analyses were undertaken in Stata 14 (StataCorp, College Station, TX, USA).

received iron chelation therapy at the time of registration. Six patients received iron chelation therapy during this observation period (Online Supplementary Table S1). The median number of samples available per patient was 5 (range: 1-7), and the median follow-up period was 5.8 years. OS and progression-free survival (PFS) in our study population were 4.8 and 4.6 years, respectively. Nineteen patients died, including 5 patients after progression and 9 patients from causes possibly related to MDS (hemorrhage 2, infection 5, and cardiovascular 2) (Online Supplementary Table S2).

Iron parameters

Median ferritin levels were elevated (>250 μg/l) at registration in all patient groups, but the highest median levels were observed in the transfusion-dependent (TD) groups (Table 1). Median serum iron levels were within reference range (12-30 μmol/L) in all patient groups at registration. Overall, median TSAT was within reference range (<45%) at registration, with the exception of TD MDS-RS patients (Table 2). Median hepcidin levels were within reference range in all patient groups at registration, but TD patients had significantly higher hepcidin levels compared to transfusion-independent (TI) patients (P<0.001). Ferritin levels correlated significantly with hepcidin levels (r=0.55,

A

B

Results Patient characteristics The median age of all patients at registration was 73 years (range: 43-95 years). The majority of the patients were male; 64% (n=64). The IPSS risk groups of the 100 patients in the study were: low 47%, intermediate-1 41%, and unknown 12%. The IPSS-R risk groups were: very low 32%, low 41%, intermediate 8%, high 3%, and unknown 16%. WHO 2001 MDS-subtypes were refractory cytopenia with multilineage dysplasia (RCMD; 37%), refractory anemia with ring sideroblasts (RARS; 30%), refractory anemia (RA; 18%), refractory anemia with excess blasts (RAEB; 7%), 5q-syndrome (4%) and refractory cytopenia with multilineage dysplasia and ringed sideroblasts (RCMD-RS; 4%). Fourteen percent of the patients were transfusion-dependent at registration (n=14). No patients 72

Figure 2 Survival according to LPI (A) or NTBI (B) and transfusion status. LPI, NTBI and transfusion status were analyzed as time dependent factors, implicating that patients may switch groups over time according to the LPI/NTBI and transfusion status at each specific time point. LLOD: lower limit of detection; TI: transfusion-independent; TD: transfusion-dependent; LPI: labile plasma iron; NTBI: non-transferrin bound iron.

haematologica | 2018; 103(1)


Toxic iron species in lower-risk MDS

P<0.001). The median GDF15 levels were elevated in the RS subgroup only. NTBI levels above LLOD (>0.47 µmol/L) occurred in all patient groups at registration with the highest levels present in MDS-RS patients. sTFR levels were within the reference range (0.8-1.8 mg/L) at registration, and the highest levels were observed in TI MDS-RS patients (Table 2). The median LPI levels were below LLOD in all patient groups at registration (<0.24 mol/L), except in TD MDS-RS patients. Median C-reactive protein (CRP) levels were below the upper limit of the reference range (<10 mg/L) in all groups at all time points (Table 1) and the majority of patients with CRP levels above 50 mg/L were TD. CRP levels correlated positively with hep-

cidin levels (r = 0.30, P<0.001) and ferritin levels (r=0.22, P<0.001).

Impact of MDS subtype and transfusions on iron parameters over time The impact of transfusions and MDS subtype (RS vs. non-RS) on TSAT, hepcidin, GDF15, NTBI and LPI levels over time is shown in Table 2. Both serum ferritin and serum iron levels increased significantly (r=0.59, P<0.001 and r=0.32, P<0.001, respectively) with a cumulative number of transfused units over time in TD patients (Online Supplementary Table S3) as well as in RS patients (Online Supplementary Table S4). TSAT remained stable and within

Table 2. Frequency, median and quartiles of iron parameters by transfusion status per MDS subtype at registration, 1 year and 2 years follow-up.

Transferrin saturation (%) MDS non-RS: TI MDS non-RS: TD MDS-RS: TI MDS-RS: TD Hepcidin (nmol/L) MDS non-RS: TI MDS non-RS: TD MDS-RS: TI MDS-RS: TD Growth differentiation factor 15 (ng/L) MDS non-RS: TI MDS non-RS: TD MDS-RS: TI MDS-RS: TD Soluble transferrin receptor (mg/L) MDS non-RS: TI MDS non-RS: TD MDS-RS: TI MDS-RS: TD Non transferrin bound iron (μmol/L) MDS non-RS: TI MDS non-RS: TD MDS-RS: TI MDS-RS: TD Labile plasma iron (μmol/L) MDS non-RS: TI MDS non-RS: TD MDS-RS: TI MDS-RS: TD

N

Registration Median(p10-p90)

N

1 year follow-up Median(p10-p90)

N

2 years follow-up Median(p10-p90)

100 56 10 29 5 99 56 10 28 5

35.6 (19.0 - 87.4) 32.8 (17.1 - 55.6) 28.7 (8.5 - 77.9) 48.8 (24.6 - 92.5) 90.0 (53.1 - 120.4) 4.5 (1.1 - 21.7) 4.5 (1.7 - 22.1) 4.9 (0.5 - 75.9) 3.8 (1.0 - 8.7) 10.3 (3.8 - 15.9)

78 32 21 16 9 78 32 21 16 9

34.4 (16.4 - 92.9) 28.4 (17.4 - 59.1) 36.8 (14.0 - 89.1) 36.4 (20.8 - 86.4) 93.6 (42.1 - 110.6) 5.6 (1.2 - 19.6) 4.3 (1.5 - 11.8) 17.3 (0.5 - 29.2) 3.4 (0.5 - 5.8) 9.2 (3.8 - 14.4)

64 26 17 9 12 65 26 17 9 13

37.5 (22.2 - 94.3) 30.1 (18.8 - 54.2) 39.3 (20.4 - 97.7) 35.6 (23.9 - 92.6) 93.1 (71.7 - 97.6) 5.2 (1.0 - 19.6) 4.6 (0.9 - 13.6) 9.2 (1.3 - 28.4) 2.9 (0.8 - 12.2) 5.2 (1.0 - 14.6)

100 56 10 29 5

2193 (952 - 5663) 1777 (731 - 4658) 2306 (1218 - 4927) 2619 (996 - 11083) 2893 (2113 - 5370)

77 32 20 16 9

2479 (1016 - 7982) 1653 (615 - 5684) 2583 (1725 - 7166) 2694 (1223 - 10303) 3866 (830 - 15167)

63 26 17 8 12

2576 (1045 - 7746) 1685 (633 - 5736) 2998 (1398 - 8037) 2780 (1331 - 9554) 5361 (1053 - 8399)

100 56 10 29 5

1.3 (0.7 - 2.8) 1.2 (0.8 - 2.7) 1.0 (0.6 - 2.8) 1.6 (0.8 - 3.3) 0.9 (0.4 - 3.1)

78 32 21 16 9

1.4 (0.7 - 3.0) 1.4 (0.9 - 2.8) 1.1 (0.4 - 3.1) 2.0 (1.1 - 2.8) 1.2 (0.6 - 3.1)

62 26 16 8 12

1.3 (0.8 - 2.7) 1.2 (0.9 - 2.7) 1.2 (0.6 - 2.2) 2.2 (1.0 - 2.8) 1.4 (0.4 - 3.6)

100 56 10 29 5

0.65 (0.14 - 3.03) 0.41 (0.10 - 1.51) 0.80 (0.05 - 2.73) 0.88 (0.26 - 3.99) 3.03 (1.90 - 3.40)

77 31 21 16 9

0.59 (0.15 - 3.64) 0.42 (0.03 - 0.91) 0.69 (0.16 - 3.64) 0.70 (0.16 - 3.52) 3.60 (0.15 - 8.64)

65 26 17 9 13

0.64 (0.14 - 5.42) 0.50 (0.18 - 1.78) 1.00 (0.12 - 7.25) 0.52 (0.05 - 5.42) 2.86 (0.46 - 7.57)

100 56 10 29 5

0.09 (0.02 - 0.22) 0.10 (0.03 - 0.19) 0.06 (0.01 - 0.18) 0.10 (0.02 - 0.32) 0.08 (0.00 - 0.35)

77 31 21 16 9

0.13 (0.03 - 0.38) 0.10 (0.02 - 0.17) 0.17 (0.06 - 0.38) 0.09 (0.05 - 0.24) 0.47 (0.06 - 1.26)

65 26 17 9 13

0.13 (0.02 - 0.38) 0.11 (0.01 - 0.30) 0.14 (0.02 - 1.08) 0.10 (0.03 - 0.17) 0.19 (0.08 - 1.39)

MDS: Myelodysplastic syndromes; RS: ring sideroblastic; TI: transfusion-independent; TD: transfusion-dependent.

haematologica | 2018; 103(1)

73


L. de Swart et al.

reference range in the TI patients, with the exception of a minority of RS patients (Online Supplementary Figure S1), and increased over time in the TD patients, with up to 94.9% in patients with >10 RBC units transfused (Online Supplementary Table S3). Hepcidin levels increased with the number of units transfused; in contrast, hepcidin levels significantly decreased over time in TI MDS-RS patients (Online Supplementary Table S4). GDF15 levels were not associated with transfusion status alone, but did increase over time in TD MDS-RS patients with a median of 2893 ng/L at registration compared to 5361 ng/L at 2 years follow up. STFR levels increased significantly (P<0.001) over time in both TI and TD MDS-RS patients (P=0.01) (Table 2). STFR levels did not change over time in non-RS MDS patients. The lowest sTFR levels were observed in patients who had received more than 10 units (Online Supplementary Table S3). TD MDS-RS patients had the most elevated levels of NTBI and LPI over time (Online Supplementary Table S3 and S4).

Correlation between markers of iron overload Both elevated NTBI and LPI levels (>LLOD) showed a threshold effect with TSAT of >70% and >80%, respectively (Figure 1A,B). Detectable LPI levels occurred almost exclusively in patients with MDS-RS and/or patients who had received transfusions. NTBI and LPI levels above the LLOD were mutually positively correlated (r=0.46; P<0.001). Both NTBI and LPI showed a linear relationship (P<0.001) with ferritin, but no threshold levels could be detected (Figure 1C,D). The highest values were observed

in TD MDS patients; subgroup analyses showed mainly a positive correlation in the TD and/or RS subgroup (Figure 1C,D).

Prognostic impact of iron overload markers Time-dependent, multivariate analysis of overall survival, adjusted for age and IPSS-R risk groups revealed no significant effect on overall survival for NTBI (HR=0.56, 95%CI 0.21-1.52; P=0.26) and for TSAT (HR=0.91, 95%CI 0.29-2.86; P=0.88) (Table 3, Figure 2B and Online Supplementary Figure S1). Ten out of 19 patients who died during this study had detectable LPI. The majority (7 patients) died from progression or MDS-related causes (Online Supplementary Table S2). Kaplan-Meier curves demonstrate prognostic impact on survival of detectable LPI levels by transfusion status (Figure 2), but no significant effect in the multivariate analysis adjusted for age and IPSS-R risk (HR=2.1, 95%CI 0.7-6.2; Table 3). Once LPI was increased in both TD and TI patients, survival time decreased, with the greatest impact observed in patients who were TD and had increased LPI levels (adjusted HR=3.0, 95%CI 0.713.3). Since 41 patients were also treated with erythropoietin stimulating agents (ESA), we repeated the analyses adjusted for whether or not the patient had been treated with ESA at each visit (Figure 3). These adjustments did not significantly alter the magnitude of the risk estimates on OS (HR=3.0, 95%CI 0.7-13.5) (Table 3). Because the survival of patients with RS-MDS is usually considered better than in the non-RS MDS population, we repeated the analyses in the largest group of 66 non-RS

Table 3. Cox model of OS by labile plasma iron, non-transferrin bound iron and transferrin saturation along with transfusion status as time varying variable for all patients (n=100).

Unadjusted Hazard ratio P (95% CI)

LPI (μmol/L) <LLOD ≥LLOD LPI<LLOD, TI LPI≥LLOD, TI LPI <LLOD, TD LPI ≥LLOD, TD NTBI (μmol/L) <LLOD ≥LLOD NTBI<LLOD, TI NTBI≥LLOD, TI NTBI<LLOD, TD NTBI≥LLOD, TD TSAT <80% >80% TSAT <80%, TI TSAT≥80%, TI TSAT <80%, TD TSAT≥80%, TD

1 2.2 (0.8 – 6.2) 0.14 1 4.6 (0.5 – 42.4) 0.18 4.1 (1.2 – 13.6) 0.02 4.7 (1.1 – 19.7) 0.03 1 0.7 (0.3 – 1.7) 0.39 1 0.6 (0.1 – 3.8) 0.61 4.7 (1.1 – 19.0) 0.03 2.2 (0.5 – 8.6) 0.27 1 1.3 (0.4 – 3.6) 0.66 1 2.5 (1.0 – 6.2) 0.04 1.9 (1.2 – 3.0) 0.003 1.3 (0.9 – 2.0) 0.19

Hazard ratio (95% CI) 1 2.0 (0.7 – 6.0) 1 3.2 (0.3 – 30.2) 2.0 (0.5 – 7.1) 3.0 (0.7 – 13.3) 1 0.6 (0.2 – 1.6) 1 0.7 (0.1 – 4.0) 2.6 (0.6 – 11.6) 1.1 (0.3 – 5.0) 1 0.9 (0.3 – 2.9) 1 2.3 (0.9 – 5.7) 1.6 (1.0 – 2.5) 1.1 (0.7 – 1.7)

Adjusted1 P 0.21 0.31 0.30 0.15 0.27 0.65 0.22 0.86 0.88 0.08 0.05 0.70

Adjusted2 Hazard ratio P (95% CI) 1 2.0 (0.7 – 5.8) 1 3.3 (0.4 – 31.1) 2.2 (0.6 – 8.1) 3.0 (0.7 – 13.5) 1 0.5 (0.2 – 1.5) 1 0.7 (0.1 – 4.2) 3.1 (0.7 – 14.4) 1.2 (0.3 – 5.4) 1 0.9 (0.3 – 2.8) 1 2.5 (1.0 – 6.5) 1.7 (1.1 – 2.7) 1.1 (0.7 – 1.7)

0.23 0.30 0.24 0.14 0.24 0.67 0.14 0.80 0.85 0.05 0.03 0.67

Adjusted3 Hazard ratio P (95% CI) 1 2.0 (0.7 – 6.2) 1 3.2 (0.3 – 30.4) 2.0 (0.5 – 7.1) 3.0 (0.7 – 13.4) 1 0.6 (0.2 – 1.5) 1 0.6 (0.1 – 4.0) 2.5 (0.6 – 11.5) 1.1 (0.3 – 4.9) 1 1.0 (0.3 – 3.1) 1 2.3 (0.9 – 5.9) 1.6 (0.99 – 2.5) 1.1 (0.7 – 1.7)

0.20 0.31 0.31 0.15 0.26 0.62 0.22 0.89 0.97 0.10 0.053 0.70

1 Adjusted for age at diagnosis and IPSS-R 2Adjusted for age at diagnosis, IPSS-R and ESA treatment status at each visit.3Adjusted for age, IPSS-R and RS status. CI: confidence interval; LLOD: lowest level of detection; LPI: labile plasma iron; TI: transfusion-independent; TD: transfusion-dependent; NTBI: non-transferrin bound iron; TSAT: transferrin saturation.

74

haematologica | 2018; 103(1)


Toxic iron species in lower-risk MDS

patients (Table 4). Detectable LPI levels had a remarkable impact on survival in the whole non-RS group, but the impact was only significant in the TD subgroup (HR=17.0, 95%CI 2.0-146.6). TSAT levels had a borderline impact on survival in TI patients. Six patients received iron chelation in this study (Online Supplementary Table S1). LPI levels during treatment with deferasirox decreased below LLOD (4 patients), even in those patients with high TSAT. Only 3 patients were treated with lenalidomide. Ferritin levels and elevated CRP are time-dependent variables, which correlate closely with transfusion burden/transfusion intensity, and presumably with infections

(data not shown). Ferritin levels and elevated CRP predict survival when adjusted for age and IPSS-R group only, but the prognostic impact is less clear when transfusion intensity was added to the model (data not shown).

Discussion This study among 100 European lower-risk MDS patients showed that both RBC transfusions and the presence of RS increased the occurrence of the toxic iron species NTBI and LPI in serum. Our data on iron parameters over time suggest that body iron accumulation and

Table 4. Cox model of OS by labile plasma iron, non-transferrin bound iron and transferrin saturation along with transfusion status as time varying variable for non-RS patients only (n=66).

Unadjusted Hazard ratio (95% CI) LPI (μmol/L) <LLOD Elevated LPI<LLOD, TI LPI≥LLOD, TI LPI <LLOD, TD LPI ≥LLOD, TD NTBI (μmol/L) <LLOD Elevated NTBI<LLOD, TI NTBI≥LLOD, TI NTBI<LLOD, TD NTBI≥LLOD, TD TSAT <80 Elevated TSAT <80, TI TSAT≥80, TI TSAT <80, TD TSAT≥80, TD

1 4.9 (1.4 – 16.8) 1 10.2 (0.9 – 115.4) 4.6 (0.9 – 23.5) 11.8 (1.9 – 74.0) 1 0.6 (0.2 – 1.9) 1 0.6 (0.1 – 6.9) 5.7 (1.1 – 30.3) 2.1 (0.4 – 12.3) 1 2.1 (0.6 – 7.8) 1 3.8 (1.1 – 12.7) 1.9 (1.1 – 3.2) 1.5 (0.9 – 2.5)

P 0.01 0.06 0.07 0.008 0.37 0.70 0.04 0.39 0.28 0.03 0.02 0.12

Adjusted1 Hazard ratio (95% CI) 1 5.4 (1.5 – 19.6) 1 5.3 (0.4 – 68.9) 2.0 (0.3 – 12.0) 10.3 (1.3 – 79.5) 1 0.6 (0.2 – 2.0) 1 1.1 (0.1 – 14.4) 5.7 (0.8 – 42.2) 1.4 (0.2 – 8.2) 1 1.1 (0.2 – 5.4) 1 3.7 (0.98 – 13.8) 1.7 (0.9 – 3.2) 1.1 (0.6 – 1.9)

P 0.01 0.20 0.47 0.03 0.38 0.92 0.09 0.74 0.90 0.053 0.13 0.80

Adjusted2 Hazard ratio (95% CI) 1 9.3 (2.0 – 43.3) 1 5.9 (0.4 – 86.2) 1.4 (0.2 – 8.9) 17.0 (2.0 – 146.6) 1 0.6 (0.2 – 2.2) 1 1.1 (0.09 – 14.3) 5.4 (0.7 – 43.7) 1.4 (0.2 – 8.2) 1 1.5 (0.3 – 8.6) 1 3.7 (0.99 – 14.1) 1.6 (0.8 – 3.2) 1.1 (0.6 – 1.9)

P 0.004 0.19 0.70 0.01 0.46 0.92 0.11 0.74 0.63 0.052 0.21 0.77

Adjusted for age at diagnosis and IPSS-R. 2Adjusted for age at diagnosis, IPSS-R and ESA treatment status at each visit. LLOD: lowest level of detection; LPI: labile plasma iron; TI: transfusion-independent; TD: transfusion-dependent; TSAT: transferrin saturation; CI: confidence interval.

1

Figure 3 Flow diagram of patients treated with transfusions and erythropoietin stimulating agents (ESAs). In total, 10 patients became transfusion-independent after starting ESA treatment

haematologica | 2018; 103(1)

75


L. de Swart et al.

toxic iron species (NTBI and LPI) in RS-MDS patients occur along the axis of ineffective erythropoiesis, characterized by elevated sTFR, increased GDF15, low hepcidin, and increased circulating and parenchymal iron levels (Figure 4A). Interestingly we found detectable LPI, but not

NTBI, to be associated with a significantly decreased OS in non-sideroblastic MDS patients. Hepcidin levels were significantly elevated in all TD patient categories immediately after the initiation of transfusions, and remained elevated during transfusion

A

B

Proposed pathogenesis of iron toxicity in lower-risk MDS: the impact of ineffective erythropoiesis (A) and of transfusions (B). Ineffective erythropoiesis , especially in RS MDS, results in increased bone marrow production of GDF15 and possibly twisted gastrulation 1 and erythroferrone. These factors inhibit hepcidin production by the hepatocytes. Low hepcidin levels increase iron absorption from intestinal mucosa and increase iron release from the macrophages. Eventually, this may lead to toxic levels of NTBI and LPI, causing damage in solid organs, the immune system and the marrow. During transfusions hepcidin levels increase, despite higher GDF15 levels, leading to lower iron absorption in the gut. However, transfusions cause massive iron loading of RES-macrophages leading to elevated, circulating stored iron levels and toxic iron species - despite elevated hepcidin levels - and subsequent toxicities. Figure adapted from ML Cuijpers, et al.6 RS: ring sideroblastic; GDF15: growth differentiation factor 15; TWSG1: twisted gastrulation 1; LPI: labile plasma iron; NTBI: non-transferrin bound iron; sTFR: soluble transferrin receptor; RES: reticuloendothelial system; TSAT: transferrin saturations; EPO: erythropoietin.

76

haematologica | 2018; 103(1)


Toxic iron species in lower-risk MDS

dependency, confirming recent studies in transfused MDS patients and illustrated in Figure 4B.5,7 However, the elevated hepcidin levels showed a tendency to decrease during continued exposure to transfusions. In addition, sTFR levels decreased over time in TD patients, compatible with previously reported suppression of erythropoiesis by continued transfusions.20,25 Interestingly, GDF15 increased over time in TD MDS patients and especially in those categorized as TD RS-MDS. Increased GDF15 has previously been associated with ineffective erythropoiesis, but not with TD-mediated suppression of erythropoiesis.5 This suggests that TD-mediated suppression of ineffective erythropoiesis may be less effective during prolonged transfusions. This is supported by the gradual decline over time of the initially elevated hepcidin levels during prolonged transfusions. These data show that previous conflicting observations on the relationship of GDF15 and hepcidin can be explained by the impact of transfusions on GDF15 and hepcidin levels, especially in RS-MDS patients.5 Hepcidin levels decreased over time in TI patients of the RS subtype. An earlier study in 107 untransfused patients observed generally elevated hepcidin levels in MDS, but low hepcidin/ferritin ratios in the RS subtypes, compatible with the low hepcidin levels in the RS patients of our study.39 In addition, RS patients showed elevated sTFR levels and decreased hepcidin levels compared to TI non-RS at all time points. These observations confirm the previously reported association between sTFR and ineffective erythropoiesis, resulting in an increased uptake of dietary iron and iron release by macrophages, subsequently leading to increased circulating iron levels, elevated parenchymal iron stores and toxic iron species.7 Interestingly, recently developed hepcidin agonists prevented low hepcidin-induced toxicity, preclinically, thus demonstrating the potential of these compounds to prevent iron loading erythropoietic activity in MDS, especially in RS-MDS.25,40 Taken together, our data suggest a worsening over time of the ineffective erythropoiesis along with lower hepcidin levels in RS patients.41,42 Elevated NTBI levels could be demonstrated in our study early in the follow-up period of all patient groups. In iron loading anemias, such as thalassemia syndromes, it has been suggested that iron species, such as NTBI and LPI, serve as early indicators of iron toxicity and as measures for the effectiveness of iron chelation therapy in reducing potentially toxic iron molecules in the plasma.7,43 Excess toxic iron species catalyze the cellular generation of reactive oxygen species (ROS). Oxidative stress and high TSAT, as in combination with a subsequent decrease in cellular antioxidants, may lead to the oxidation of lipids, proteins and DNA, causing cell and tissue damage.44,45 Biomarkers of oxidative stress have been found to be increased in patients with MDS and iron overload.3,46-49 The combination of high serum ferritin levels as well as the presence of NTBI and LPI was noted to be more frequent in RS patients compared to non-RS patients in our study. Herein, it is important to realize that in general practice, including our study, serum samples are collected immediately prior to transfusions. LPI levels are usually elevated for a few days after transfusion (except when transferrin is highly saturated) in contrast to the more stable NTBI which have been reported to have a longer half-life.50,51 These free haematologica | 2018; 103(1)

iron molecules are easily translocated intracellularly and cause oxidative stress as shown in thalassemia.33 Oxidative stress may explain why elevated LPI levels are associated with an increased risk of dying prematurely; too early to die from causes related to classical iron overload in the lungs, liver and heart as observed in young thalassemia patients after long-term transfusions. Less is known about the pathophysiology and tissue toxicity of iron overload caused by ineffective erythropoiesis in MDS. We observed that high NTBI and LPI levels also occurred in RS patients not receiving transfusions, indicating that iron toxicity (oxidative stress) may also occur in this category of MDS patients (Figure 4), similar to TI β-thalassemia intermedia, ι-thalassemia (Hb-H disease), and X-linked sideroblastic anemia.52,53 Previously, we reported that detectable LPI occurred almost exclusively in samples with TSAT >80%.29 Interestingly, in the study herein, survival of patients with TSAT >80% was not different from the survival of patients with a TSAT below this level (Online Supplementary Figure S1). The lowest hepcidin levels have been observed in RS patients,5 similar to our observations. The elevation of LPI in TI patients occurred exclusively in RS patients as expected in view of the low hepcidin levels leading to increased serum iron levels, through increased intestinal iron absorption and increased iron release from macrophages. Non-RS patients with SF3B1 mutations may show a similar iron pathophysiology since they appear to have a similar outcome compared to RS-MDS patients with SF3B1 mutations.54 In addition, significant relationships were found between SF3B1 mutations and marrow erythroblasts (P=0.001) or soluble transferrin receptor factor 15 (P=0.033).11 Our data show that elevated LPI levels - in contrast to elevated NTBI levels and TSAT - associate with decreased survival. The risk of dying prematurely in patients with detectable LPI levels occurred too early in this study to explain this risk by classical iron overload due to organ toxicity (lungs, liver and heart) after long term transfusions, but this indicates a direct effect associated with elevated LPI levels. The impact of detectable LPI was only significant in the large non-RS group, but the same tendency was observed in the smaller RS subpopulation. This effect was independent of ESA treatment, indicating that the effect of LPI on outcome is not simply an effect of the interaction of LPI with ESA, as a previously described outcome modifier.55,56 The widely used parameter TSAT cannot serve as a parameter to predict survival. However, TSAT can be used as a prescreening method to identify patients who are at risk to develop detectable LPI levels and associated poor prognosis. This approach may reduce the number of LPI determinations substantially. Ferritin levels have been reported as a prognostic indicator in MDS, but ferritin as a marker of iron toxicity may be compromised by the stage of MDS, the cumulative transfusional load and its properties as an acute phase protein.57-59 Moreover, the level of ferritin does not indicate whether iron is stored in parenchymal cells or in the reticuloendothelial system (RES), of which the former is considered to be a more toxic form of iron overload. The foregoing is reflected by the weaker correlation of ferritin levels with LPI when compared with the correlation between TSAT and LPI levels. The positive correlation between CRP and hepcidin in the study here77


L. de Swart et al.

in suggests that inflammation also influences iron homeostasis in some MDS patients, as reported for patients with other inflammatory diseases.4 Similar to ferritin, CRP had a significant impact on survival, potentially reflecting the impact of infections and autoimmune diseases on survival in this patient group. Finally, we demonstrated, in the limited number of patients treated with iron chelators in the study herein, that LPI levels decreased below LLOD, even in patients with high TSAT during treatment with deferasirox. These data corroborate with the post hoc data from a large chelation study in MDS.43 In conclusion, we demonstrated a disturbed iron homeostasis both in transfusion dependent MDS patients and in the subgroup of transfusion independent RS patients. This is the first clinical study that identifies LPI as a relevant marker for the potentially toxic fraction of iron species and its impact on OS. Increased LPI levels were restricted to patients with TSAT percentages exceeding 80%. However, TSAT exceeding 80% alone was not prognostic for survival. Therefore, we propose TSAT as a screening parameter to assess risk for detectable LPI. Additional studies are warranted to show that intervention with iron chelation improves survival, comorbidities and quality of life in lower-risk MDS patients by lowering LPI levels.

References 1. Bennett JM, Catovsky D, Daniel MT, et al. Proposals for the classification of the myelodysplastic syndromes. Br J Haematol. 1982;51(2):189-199. 2. Greenberg P, Cox C, LeBeau MM, et al. International scoring system for evaluating prognosis in myelodysplastic syndromes. Blood. 1997;89(6):2079-2088. 3. de Swart L, Smith A, Johnston TW, et al. Validation of the revised international prognostic scoring system (IPSS-R) in patients with lower-risk myelodysplastic syndromes: a report from the prospective European LeukaemiaNet MDS (EUMDS) registry. Br J Haematol. 2015;170(3):372383. 4. Zipperer E, Post JG, Herkert M, et al. Serum hepcidin measured with an improved ELISA correlates with parameters of iron metabolism in patients with myelodysplastic syndrome. Ann Hematol. 2013; 92(12):1617-1623. 5. Santini V, Girelli D, Sanna A, et al. Hepcidin levels and their determinants in different types of myelodysplastic syndromes. PLoS One. 2011;6(8):e23109. 6. Cuijpers ML, Raymakers RA, Mackenzie MA, de Witte TJ, Swinkels DW. Recent advances in the understanding of iron overload in sideroblastic myelodysplastic syndrome. Br J Haematol. 2010;149(3):322333. 7. Shenoy N, Vallumsetla N, Rachmilewitz E, Verma A, Ginzburg Y. Impact of iron overload and potential benefit from iron chelation in low-risk myelodysplastic syndrome. Blood. 2014;124(6):873-881. 8. Greenberg P, Cox C, LeBeau MM, et al. International scoring system for evaluating prognosis in myelodysplastic syndromes.

78

Acknowledgments The authors would like to thank the other members of the EUMDS Steering Committee: Pierre Fenaux, France; Moshe Mittelman, Israel; Reinhard Stauder, Austria; Guillermo Sanz, Spain; Luca Malcovati, Italy; Ulrich Germing, Germany; Krzysztof Mądry, Poland; Mette Skov Holm, Denmark; Antonio Medina Almeida, Portugal; Aleksandar Savic, Republic of Serbia and Njetočka Gredelj Šimec, Croatia. The authors and members of the steering committee of the EUMDS registry would like to thank all local investigators and operational team members for their contribution. The authors wish to thank Erwin Wiegerinck of the Radboudumc expertise center for iron disorders for the measurement of LPI, NTBI and hepcidin-25, and Siem Klaver, Margot Rekers and Karin van der Linden for sample handling and Elise van Pinxten-van Orsouw and Linda van der Landen for data entry of all iron parameters. Funding The work of the EUMDS Registry for low and intermediate-1 MDS is supported by an educational grant from Novartis Pharmacy B.V. Europe. This work is part of the MDS-RIGHT activities, which has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 634789 - “Providing the right care to the right patient with MyeloDysplastic Syndrome at the right time”.

Blood. 1997;89(6):2079-2088. 9. Kautz L, Nemeth E. Molecular liaisons between erythropoiesis and iron metabolism. Blood. 2014;124(4):479-482. 10. Ramirez JM, Schaad O, Durual S, et al. Growth differentiation factor 15 production is necessary for normal erythroid differentiation and is increased in refractory anaemia with ring-sideroblasts. Br J Haematol. 2009;144(2):251-262. 11. Ambaglio I, Malcovati L, Papaemmanuil E, et al. Inappropriately low hepcidin levels in patients with myelodysplastic syndrome carrying a somatic mutation of SF3B1. Haematologica. 2013;98(3):420-423. 12. Rund D, Rachmilewitz E. Beta-thalassemia. N Eng J Med. 2005;353(11):11351146. 13. Gattermann N, Rachmilewitz EA. Iron overload in MDS-pathophysiology, diagnosis, and complications. Ann Hematol. 2011;90(1):1-10. 14. Ganz T. Systemic iron homeostasis. Physiol Rev. 2013;93(4):1721-1741. 15. Cabantchik ZI. Labile iron in cells and body fluids: physiology, pathology, and pharmacology. Front Pharmacol. 2014;5:45. 16. Breuer W, Hershko C, Cabantchik ZI. The importance of non-transferrin bound iron in disorders of iron metabolism. TransfusSci. 2000;23(3):185-192. 17. Brissot P, Ropert M, Le Lan C, Loreal O. Non-transferrin bound iron: a key role in iron overload and iron toxicity. Biochim Biophys Acta. 2012;1820(3):403-410. 18. Hershko C, Graham G, Bates GW, Rachmilewitz EA. Non-specific serum iron in thalassaemia: an abnormal serum iron fraction of potential toxicity. Br J Haematol. 1978;40(2):255-263. 19. Ganz T. Hepcidin and iron regulation, 10 years later. Blood. 2011;117(17):4425-4433. 20. Girelli D, Nemeth E, Swinkels DW. Hepcidin in the diagnosis of iron disorders.

Blood. 2016;127(23):2809-2813. 21. Fleming RE, Ponka P. Iron overload in human disease. N Eng J Med. 2012;366(4):348-359. 22. Tanno T, Bhanu NV, Oneal PA, et al. High levels of GDF15 in thalassemia suppress expression of the iron regulatory protein hepcidin. Nat Med. 2007;13(9):1096-1101. 23. Porter JB, de Witte T, Cappellini MD, Gattermann N. New insights into transfusion-related iron toxicity: Implications for the oncologist. Crit Rev Oncol Hematol. 2016;99:261-271. 24. Nemeth E. Hepcidin and beta-thalassemia major. Blood. 2013;122(1):3-4. 25. Metzgeroth G, Rosee PL, Kuhn C, et al. The soluble transferrin receptor in dysplastic erythropoiesis in myelodysplastic syndrome. Eur J Haematol. 2007;79(1):8-16. 26. Speeckaert MM, Speeckaert R, Delanghe JR. Biological and clinical aspects of soluble transferrin receptor. Crit Rev Clin Lab Sci. 2010;47(5-6):213-228. 27. Huebers HA, Beguin Y, Pootrakul P, Einspahr D, Finch CA. Intact transferrin receptors in human plasma and their relation to erythropoiesis. Blood. 1990; 75(1):102-107. 28. Khatami S, Dehnabeh SR, Mostafavi E, et al. Evaluation and comparison of soluble transferrin receptor in thalassemia carriers and iron deficient patients. Hemoglobin. 2013;37(4):387-395. 29. de Swart L, Hendriks JC, van der Vorm LN, et al. Second international round robin for the quantification of serum non-transferrin-bound iron and labile plasma iron in patients with iron-overload disorders. Haematologica. 2016;101(1):38-45. 30. Evans RW, Rafique R, Zarea A, et al. Nature of non-transferrin-bound iron: studies on iron citrate complexes and thalassemic sera. J Biol Inorg Chem. 2008;13(1):57-74. 31. Silva AM, Hider RC. Influence of non-

haematologica | 2018; 103(1)


Toxic iron species in lower-risk MDS

32.

33.

34.

35.

36.

37.

38.

39.

40.

enzymatic post-translation modifications on the ability of human serum albumin to bind iron. Implications for non-transferrinbound iron speciation. Biochim Biophys Acta. 2009;1794(10):1449-1458. Nam H, Wang CY, Zhang L, et al. ZIP14 and DMT1 in the liver, pancreas, and heart are differentially regulated by iron deficiency and overload: implications for tissue iron uptake in iron-related disorders. Haematologica. 2013;98(7):1049-1057. Esposito BP, Breuer W, Sirankapracha P, Pootrakul P, Hershko C, Cabantchik ZI. Labile plasma iron in iron overload: redox activity and susceptibility to chelation. Blood. 2003;102(7):2670-2677. Le Lan C, Loreal O, Cohen T, et al. Redox active plasma iron in C282Y/C282Y hemochromatosis. Blood. 2005; 105(11):4527-4531. Pootrakul P, Breuer W, Sametband M, Sirankapracha P, Hershko C, Cabantchik ZI. Labile plasma iron (LPI) as an indicator of chelatable plasma redox activity in ironoverloaded beta-thalassemia/HbE patients treated with an oral chelator. Blood. 2004;104(5):1504-1510. Zhang D, Okada S, Kawabata T, Yasuda T. An improved simple colorimetric method for quantitation of non-transferrin-bound iron in serum. Biochem Mol Biol Int. 1995;35(3):635-641. Esposito BP, Breuer W, Sirankapracha P, Pootrakul P, Hershko C, Cabantchik ZI. Labile plasma iron in iron overload: redox activity and susceptibility to chelation. Blood. 2003/10/1;102(7):2670-2677. Schultz LR, Peterson EL, Breslau N. Graphing survival curve estimates for timedependent covariates. Int J Methods Psychiatr Res. 2002;11(2):68-74. Cui R, Gale RP, Zhu G, et al. Serum iron metabolism and erythropoiesis in patients with myelodysplastic syndrome not receiving RBC transfusions. Leuk Res. 2014; 38(5):545-550. Bowen DT, Culligan D, Beguin Y, Kendall R, Willis N. Estimation of effective and total erythropoiesis in myelodysplasia using serum transferrin receptor and erythropoietin concentrations, with automated reticulocyte parameters. Leukemia.

haematologica | 2018; 103(1)

1994;8(1):151-155. 41. Sasu BJ, Cooke KS, Arvedson TL, et al. Antihepcidin antibody treatment modulates iron metabolism and is effective in a mouse model of inflammation-induced anemia. Blood. 2010;115(17):3616-3624. 42. Poli M, Girelli D, Campostrini N, et al. Heparin: a potent inhibitor of hepcidin expression in vitro and in vivo. Blood. 2011;117(3):997-1004. 43. Gattermann N, Finelli C, Della Porta M, et al. Hematologic responses to deferasirox therapy in transfusion-dependent patients with myelodysplastic syndromes. Haematologica. 2012;97(9):1364-1371. 44. Rachmilewitz EA, Weizer-Stern O, Adamsky K, et al. Role of iron in inducing oxidative stress in thalassemia: Can it be prevented by inhibition of absorption and by antioxidants?. Ann N Y Acad Sci. 2005;1054:118-123. 45. Hershko C, Link G, Cabantchik I. Pathophysiology of iron overload. Ann N Y Acad Sci. 1998;850:191-201. 46. Ghoti H, Amer J, Winder A, Rachmilewitz E, Fibach E. Oxidative stress in red blood cells, platelets and polymorphonuclear leukocytes from patients with myelodysplastic syndrome. Eur J Haematol. 2007;79(6):463-467. 47. De Souza GF, Ribeiro HL Jr., De Sousa JC, et al. HFE gene mutation and oxidative damage biomarkers in patients with myelodysplastic syndromes and its relation to transfusional iron overload: an observational cross-sectional study. BMJ Open. 2015;5(4):e006048. 48. Saigo K, Takenokuchi M, Hiramatsu Y, et al. Oxidative stress levels in myelodysplastic syndrome patients: their relationship to serum ferritin and haemoglobin values. J Int Med Res. 2011;39(5):1941-1945. 49. Bulycheva E, Rauner M, Medyouf H, et al. Myelodysplasia is in the niche: novel concepts and emerging therapies. Leukemia. 2015;29(2):259-268. 50. Hod EA, Brittenham GM, Billote GB, et al. Transfusion of human volunteers with older, stored red blood cells produces extravascular hemolysis and circulating non-transferrin-bound iron. Blood. 2011;118(25):6675-6682.

51. Hod EA, Zhang N, Sokol SA, et al. Transfusion of red blood cells after prolonged storage produces harmful effects that are mediated by iron and inflammation. Blood. 2010;115(21):4284-4292. 52. Gardenghi S, Marongiu MF, Ramos P, et al. Ineffective erythropoiesis in beta-thalassemia is characterized by increased iron absorption mediated by down-regulation of hepcidin and up-regulation of ferroportin. Blood. 2007;109(11):5027-5035. 53. Taher AT, Porter J, Viprakasit V, et al. Deferasirox reduces iron overload significantly in nontransfusion-dependent thalassemia: 1-year results from a prospective, randomized, double-blind, placebocontrolled study. Blood. 2012;120(5):970977. 54. Malcovati L, Karimi M, Papaemmanuil E, et al. SF3B1 mutation identifies a distinct subset of myelodysplastic syndrome with ring sideroblasts. Blood. 2015;126(2):233-241. 55. Jadersten M, Malcovati L, Dybedal I, et al. Erythropoietin and granulocyte-colony stimulating factor treatment associated with improved survival in myelodysplastic syndrome. J Clin Oncol. 2008;26(21):36073613. 56. Park S, Grabar S, Kelaidi C, et al. Predictive factors of response and survival in myelodysplastic syndrome treated with erythropoietin and G-CSF: the GFM experience. Blood. 2008;111(2):574-582. 57. Malcovati L, Porta MG, Pascutto C, et al. Prognostic factors and life expectancy in myelodysplastic syndromes classified according to WHO criteria: a basis for clinical decision making. J ClinOncol. 2005; 23(30):7594-7603. 58. Alessandrino EP, Della Porta MG, Bacigalupo A, et al. WHO classification and WPSS predict posttransplantation outcome in patients with myelodysplastic syndrome: a study from the Gruppo Italiano Trapianto di Midollo Osseo (GITMO). Blood. 2008;112(3):895-902. 59. Chee CE, Steensma DP, Wu W, Hanson CA, Tefferi A. Neither serum ferritin nor the number of red blood cell transfusions affect overall survival in refractory anemia with ringed sideroblasts. Am J Hematol. 2008;83(8):611-613.

79


ARTICLE

Acute Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2018 Volume 103(1):80-90

Depletion of SIRT6 enzymatic activity increases acute myeloid leukemia cells’ vulnerability to DNA-damaging agents

Antonia Cagnetta,1,2* Debora Soncini,1* Stefania Orecchioni,3 Giovanna Talarico,3 Paola Minetto,1 Fabio Guolo,1 Veronica Retali,1,2 Nicoletta Colombo,1 Enrico Carminati,1 Marino Clavio,1,2 Maurizio Miglino,1,2 Micaela Bergamaschi,1 Aimable Nahimana,4 Michel Duchosal,4 Katia Todoerti,5 Antonino Neri,6,7 Mario Passalacqua,8 Santina Bruzzone,8 Alessio Nencioni,2,9 Francesco Bertolini,3 Marco Gobbi,1,2 Roberto M. Lemoli1,2** and Michele Cea1,2**

Chair of Hematology, Department of Internal Medicine (DiMI), University of Genova, Italy; 2Hematology Unit, Policlinico San Martino, Genova, Italy; 3European Institute of Oncology, Milan, Italy; 4Service and Central Laboratory of Hematology, University Hospital of Lausanne, Switzerland; 5Laboratory of Pre-Clinical and Translational Research, IRCCS-CROB, Referral Cancer Center of Basilicata, Rionero in Vulture, Potenza, Italy; 6Department of Oncology and Hemato-Oncology, University of Milan, Italy; 7 Hematology Unit, Fondazione Cà Granda, Ospedale Maggiore Policlinico, Milan, Italy; 8 Department of Experimental Medicine, University of Genova, Italy and 9Department of Internal Medicine, University of Genova, Italy 1

*AC and DS are co-first authors. **MC and RML are co-senior authors

ABSTRACT

G

Correspondence: michele.cea@unige.it

Received: July 11, 2017. Accepted: October 9, 2017. Pre-published: October 12, 2017. doi:10.3324/haematol.2017.176248 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/80 ©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.

80

enomic instability plays a pathological role in various malignancies, including acute myeloid leukemia (AML), and thus represents a potential therapeutic target. Recent studies demonstrate that SIRT6, a NAD+-dependent nuclear deacetylase, functions as genome-guardian by preserving DNA integrity in different tumor cells. Here, we demonstrate that also CD34+ blasts from AML patients show ongoing DNA damage and SIRT6 overexpression. Indeed, we identified a poor-prognostic subset of patients, with widespread instability, which relies on SIRT6 to compensate for DNA-replication stress. As a result, SIRT6 depletion compromises the ability of leukemia cells to repair DNA double-strand breaks that, in turn, increases their sensitivity to daunorubicin and Ara-C, both in vitro and in vivo. In contrast, low SIRT6 levels observed in normal CD34+ hematopoietic progenitors explain their weaker sensitivity to genotoxic stress. Intriguingly, we have identified DNA-PKcs and CtIP deacetylation as crucial for SIRT6-mediated DNA repair. Together, our data suggest that inactivation of SIRT6 in leukemia cells leads to disruption of DNA-repair mechanisms, genomic instability and aggressive AML. This synthetic lethal approach, enhancing DNA damage while concomitantly blocking repair responses, provides the rationale for the clinical evaluation of SIRT6 modulators in the treatment of leukemia.

Introduction Acute myeloid leukemia (AML) is an aggressive form of cancer with an estimated incidence in Europe of 3-5 cases per 100,000 people.1,2 It is a highly heterogeneous disease, both biologically and clinically, with variable prognosis. Despite the improvement in our understanding of the biology of AML, the last 20 years have seen no improvement in treatment.3,4 Chemotherapy remains the backbone of therapy whereas stem cell transplantation is mainly offered to young patients (age <60 years).5,6 Therefore, the majority of AML patients (e.g. elderly patients), who are often unable to tolerate intensive treatments, face a particularly poor prognosis.7 Thus, there is an urgent need to overcome biological mechanisms underlying drug resistance in AML, to enhance the efficacy of existing treatments, and to facilitate the design of novel approaches. Several studies have shown that AML oncogenes, such as MLL fusions, N-RAS, and FLT3-ITD can lead to DNA damage accumulation by promoting replication and oxidative stress.8-12 In these cases, upregulation of DNA damage response (DDR) provides AML cells with a selective survival advanhaematologica | 2018; 103(1)


SIRT6 inhibition increases anti-leukemic activity of chemotherapy

tage, but also creates room for synthetic lethal interventions. Sirtuins are a family of NAD+-dependent deacetylase modifying enzymes that are up-regulated in a wide range of tumors and have a central role in integrating growth signals that regulate a number of cellular pathways including metabolism, genome stability, cell proliferation, and survival.13,14 Recently, we have demonstrated that multiple myeloma (MM) cells exhibit constitutive overexpression of SIRT6, a member of this family with a critical role for DNA damage repair, which provides implications for both tumorigenesis and treatment of this tumor.15 Here, we show that SIRT6 has biological relevance also in AML being frequently up-regulated in tumor cells compared with normal CD34+ hematopoietic progenitors. Importantly, such a feature is associated with a signature of chromosomal instability (CIN) which in turn confers poor prognosis to a subgroup of AML patients.16 Consistent with its observed role, SIRT6 loss unleashes genomic instability thus triggering hypersensitivity to clinically used DNA-damaging agents, including daunorubicin (DNR) and cytarabine (ARA-C), both in vitro and in vivo. Mechanistically, SIRT6 binds DNA damage sites, recruits and activates, by deacetylation, DNA-PKcs and CtIP promoting overall DNA repair. Taken together, our findings suggest that hematologic cancers, including AML, have constitutive ongoing DNA damage as well as a steadily activated DNA repair response. As a result, strategies aimed at shifting the balance towards high DNA damage and reduced DNA repair by SIRT6 inhibition can decrease tumor growth and may benefit patients with otherwise unfavorable outcomes.

Methods For a more detailed description of the methods used, see the Online Supplementary Appendix.

Cell lines and reagents The AML cell lines U937, MOLM-14, MV4-11, HL60, HEL, THP-1, NOMO-1, OCI-AML2, OCI-AML3 and NB4 were provided by collaborators or were purchased from ATCC or DSMZ (Braunschweig, Germany). All cell lines were cultured in RPMI1640 medium containing 10% FBS (GIBCO, Life Technologies, Carlsbad, CA, USA), 2 μM l−1 glutamine, 100 U mL−1 penicillin, and 100 μg ml−1 streptomycin (GIBCO, Life Technologies, Carlsbad, CA, USA). The 293T cell line was purchased from ATCC and cultured in DMEM containing 10% FBS (GIBCO, Life Technologies, Carlsbad, CA, USA), 2 μM l−1 glutamine, 100 U mL−1 penicillin, and 100 μg mL−1 streptomycin (GIBCO, Life Technologies, Carlsbad, CA, USA). Daunorubicin (DNR) and cytarabine (ARA-C) were purchased from Selleck Chemicals LLC (Houston, TX, USA) and Sigma-Aldrich (St. Louis, MO, USA), respectively; SIRT6 chemical inhibitor [2,4-dioxo-N-(4-(pyridin-3yloxyphenyl)-1,2,3,4-tetrahydroquinazoline-6-sulfonamide, henceforth named compound 1] was obtained from MolPort (Riga, Latvia).

Primary cell isolation from patient samples All studies involving human samples were performed after informed consent under institutional review board protocols of San Martino Hospital (Genova, Italy). De-identified samples were utilized. Patient AML cells (n=20) were obtained from bone marrow (BM) samples with a high disease load (>90% CD34+ blasts haematologica | 2018; 103(1)

in the marrow) and mononuclear cells were isolated by FicollHypaque gradient separation as described previously.17 Normal mononuclear cells (MNCs) were isolated from BM healthy donors by Ficoll-Hypaque centrifugation. In some experiments, normal peripheral blood (PB) MNCs were processed by MiniMacs highgradient magnetic separation column (Miltenyi Biotec, Bergisch Gladbach, Germany) to obtain highly purified CD34+ cells. Cells were either used immediately for viability assays or for mRNA isolation, or stored at −80°C in medium containing 50% FBS and 10% DMSO.

Statistical analyses All data are shown as means±Standard Deviation (s.d.). Student t-test was used to compare two experimental groups using GraphPad Prism software. Correlation of SIRT6 expression with disease progression and overall survival (OS) were measured using the Kaplan-Meier method, and the log rank test was used for group comparison. P<0.05 was considered statistically significant.

Results SIRT6 is consistently over-expressed in CD34+ blasts of AML patients SIRT6 is a chromatin remodeling-deacetylase involved in tumorigenesis.15,18-21 In order to explore its function in AML, we tested a panel of leukemia cell lines and patientderived tumor cells to evaluate the presence of this protein. All tested tumor cells showed higher SIRT6 staining than normal cells, regardless of their genetic landscape (Figure 1A). Notably, immunofluorescence analysis of selected AML cell lines (Figure 1B) confirmed a prominent, although not restricted, nuclear localization of this protein, as already reported in different tumors.15,22-24 Next, SIRT6 expression was further analyzed by querying publicly available data sets.25,26 A significantly higher SIRT6 mRNA level was found in tumor samples (n=300) compared with PB and BM normal hematopoietic and stem cells, including CD34+ stem/progenitor cells (HSPCs), more primitive CD34+ CD38– cells and unselected mononuclear cells (from BM or PB) (Figure 1C). Correlative analysis of SIRT6 levels with clinico-pathological features suggested significant association between SIRT6 expression and French-American-British (FAB) classification (Online Supplementary Figure S1A). Indeed, among AML groups, SIRT6 was higher in FAB M0 and M5 whilst the FAB M6 subgroup was enriched in patients with SIRT6 low levels. The increased SIRT6 expression in tumors versus normal controls was further verified by performing a similar analysis on primary CD34+ blast cells obtained from AML patients (n=200) collected at our Hematology Unit, compared with BM as well as peripheral blood mononuclear cells (PBMCs) from healthy donors (n=10). (Figure 1D) A subsequent investigation focusing on molecular features showed that SIRT6 high levels were significantly censured in FLT3-ITD mutant than in FLT3 wild type (P=0.034), otherwise no correlations were observed between SIRT6 expression and further abnormalities including NPM1, BAALC and WT1 (Online Supplementary Figure S1B). Among these 200 AML patients, detailed survival information was available for 100 cases. As a result, we retrospectively analyzed the prognostic significance of baseline SIRT6 expression from BM aspirate samples on OS. Results show a statistically significant inverse correlation between SIRT6 levels and 81


A. Cagnetta et al. A

B

C

D

E

Figure 1. SIRT6 is highly enriched in acute myeloid leukemia (AML) and its expression confers poor prognosis. (A) Protein lysates from a panel of AML cell lines (left), primary patient-derived AML cultures or normal peripheral blood mononuclear cells (PBMCs) (right) were analyzed for SIRT6 expression by Western blot. GAPDH was used as loading control. The quantification of SIRT6/GAPDH ratios is shown below. One experiment of two is shown. (B) Six AML cell lines were analyzed for SIRT6 (green) by immunofluorescence. 4′,6-diamidino-2-phenylindole nuclear stain is shown in blue. Original magnification x20. (C) Box plot distributions of SIRT6 gene expression levels in normal hematopoietic cells from healthy donors and leukemic CD34+ blasts from AML patients (n=300), combining data from GSE1159, GSE9476 and GSE30377 (using the probe set 219613_s_at). Normal hematopoietic samples included CD34+ selected cells [n=18: 8 from bone marrow (BM) and 10 from PB], CD34+ CD38– cells (n=10), unselected bone marrows (n=10), and unselected peripheral blood (n=6). Numbers below the graph are the number of samples per group. ns: not significant; **P<0.05; ***P<0.001 [one-way analysis of variance (ANOVA) with expression data log2 transformed]. (D) Increased SIRT6 mRNA expression is observed in AML patients (CD34+ blast cells) relative to normal controls (CD34+ cells collected from PB or BM of healthy donors; **=0.006; *P=0.02, unpaired t-test). GAPDH mRNA expression was used as an internal control. (E) Survival curves relative to SIRT6 expression in 100 individuals affected by AML diagnosed at our clinic. The patient group with higher SIRT6 expression (red line) had shorter overall survival than the patient cohort with lower SIRT6 expression (blue line) (log rank test). Median SIRT6 mRNA value was used to divide AML patients.

OS, with high SIRT6 expression associated with shorter survival rates than low expression (median survival 16 vs. 32 months; P=0.025) (Figure 1E). These results were also observed by analyzing other publicly available AML patient data sets,27 which confirmed the higher SIRT6 expression in tumors as well as its prognostic significance (Online Supplementary Figure S2A and B). Taken together, our data suggest SIRT6 plays a role in the pathogenesis of AML, and also provide a rationale for its targeting.

SIRT6 controls AML cells proliferation and makes them vulnerable to DNA-Damage Agents To further elucidate the possible oncogenic role of SIRT6 in AML, we investigated the effect of its genetic depletion by employing a lentiviral-mediated long-term gene knockdown with two shRNA constructs targeting SIRT6 (Figure 2A). We chose two AML cell lines with robust SIRT6 expression and the role of SIRT6 in cell viability and proliferation was assessed. Surprisingly, introduction of SIRT6-targeted shRNA induced a significant 82

increase in cell numbers and cell-cycle progression; these were proportional to the reduction in protein levels (Figure 2A and B); while SIRT6 overexpression did not affect cell count, due to the high SIRT6 levels at baseline (data not shown). These findings, as already observed in MM and in various solid tumors, are likely to account for the discrepancies in the tumor burden, but clearly contrast with SIRT6 overexpression in AML patients.15,28 Such paradoxical behavior prompted us to hypothesize a tumor-specific role for this NAD+-dependent histone deacetylase. As SIRT6 has been found to play a key role in mediating DNA repair mechanisms,22,29-32 we investigated whether it acts as genome-guardian also in AML blasts. SIRT6depleted cell lysates subjected to western blot analysis, showed an increased γ-H2A.X staining, suggesting that downregulation of SIRT6 expression enhances instability of AML cells (Figure 2C and Online Supplementary Figure S3) Importantly, these changes were not associated with DNA response activation, since pATM, pATR, pCHK1 and pCHK2 were almost unchanged after SIRT6 silencing. haematologica | 2018; 103(1)


SIRT6 inhibition increases anti-leukemic activity of chemotherapy

A

B

C

D

E

F

Figure 2. SIRT6 affects proliferation and vulnerability to DNA-damage agents in acute myeloid leukemia (AML) cells. (A) SIRT6 silencing in THP-1 and U937 cells using a lentiviral delivery system. (Left) Western blot analysis of pGIPz-infected cells after 48 hours (h) of selection with 1 μg mL-1 puromycin. (Center) Cell numbers evaluated by cell counting with trypan blue exclusion. (Right) AML-engineered cells were assessed for cell number and (B) cell-cycle progression. All data throughout the panel are shown as mean±Standard Deviation (s.d.) of triplicates. ns: not significant; *P<0.01; **P<0.001, Student t-test. (C) Representative western blots showing DDR pathway deregulation in THP-1 cells depleted of SIRT6 compared with control. GAPDH was used as loading control. One representative blot of two is shown. (D) OCI-AML-2 and OCI-AML-3 cells were transduced with a scrambled shRNA (CTR) or with an anti-SIRT6 shRNA (#911). Cells were used for immunoblotting detection of SIRT6 or γ-tubulin expression (top) or in viability experiments. For the latter, 2x104 cells/well were plated in 96-well plates and incubated for 48 hours (h) with or without DNR or ARA-C at the indicated concentration. Thereafter, dead cells were detected by propidium iodide staining and flow cytometry. 2x104 OCIAML2 (E) and primary AML (F) cells/well were plated in 96-well plates and incubated for 72 h with (w) / or without (w/o) DNR/ARA-C (at the indicated concentration) w / or w/o compound 1, as in Sociali et al.34 Thereafter, dead cells were detected by propidium iodide staining and flow cytometry. *0.04<P<0.01; **0.009<P<0.001; ***<0.0001.

Similarly, replicative stress markers, including RAD51, resulted unaffected by gene-knockdown in AML cells (Figure 2C and Online Supplementary Figure S3). Overall, these data indicate that SIRT6 depletion freezes DNA repair mechanisms, which in turn leads to greater damage. Lack of DNA repair efficiency sensitizes cancer cells to DNA damaging agents (DDAs).33 Based on the observation that SIRT6 affects such mechanisms in AML, we hypothesized that cells depleted of SIRT6 would be more sensitive to the genotoxic agents DNR and Ara-C. We therefore incubated SIRT6 depleted cells with clinically relevant concentrations of either agents and assessed their viability. Significantly more cytotoxicity was observed in the absence of SIRT6 compared with scramble control transfectants (Figure 2D). Consistent with these data, the SIRT6 chemical inhibitor compound 134,35 was also found to sensitize cell lines as well as primary AML cells to DDAs (Figure 2E and F). Together, these results are consistent with a leading role played by SIRT6 in regulating AML cell sensitivity to chemotherapy. haematologica | 2018; 103(1)

SIRT6 loss affects ATM/CHK2 pathway, as well as recruitment of repair factors to sites of DNA damage As SIRT6-depleted cells are more sensitive to genotoxic stress due to failure of DNA repair mechanisms, we next measured levels of proteins mediating DNA DSBs response after SIRT6 silencing. Although SIRT6 depletion did not affect the protein level of ATM, CHK2 or RPA, after DDAs treatment it markedly diminished their functional activity. Specifically, in scramble control, DDAs treatment induced RPA phosphorylation on Ser4 and Ser8, as well as increased ATM and CHK2 phosphorylation together with accumulation of lower-molecular-weight protein γH2AX. DDAs treatment did not induce the same effects (in term of phosphorylation of CHK2, RPA32, and ATM) in SIRT6-knockdown cells. Similarly, the increase in γH2AX level was more pronounced in SIRT6-depleted OCI-AML2 and OCI-AML3 cells (Figure 3A and Online Supplementary Figure S4). Overall, these observations identify a crucial role of SIRT6 in preserving genome integrity of AML cells through promotion of DNA repair mecha83


A. Cagnetta et al. A

B

C

D

Figure 3. SIRT6 affects the ATM/CHK2 pathway, as well as recruitment of repair factors to sites of DNA damage in acute myeloid leukemia (AML) cells. (A) Indicated AML cells were engineered to express an anti-SIRT6 shRNA (#911). Next, cells were incubated for 3 hours (h) with (w) or without (w/o) DNR (0.1 uM), or Ara-C (1 uM). Subsequently, total and phosphorylated ATM, Chk2, and RPA as well as γH2AX levels were detected by immunoblotting. Detection of Rad51 and γH2AX (B), pRPA (C), 53BP1 (D), and DAPI was measured by confocal microscopy in OCI-AML2 cells expressing shRNA (clone #911) targeting SIRT6 or control and cultured with or without treatment with Ara-C (1 μM) or DNR (0.1 μM) for 1 h (magnification x40). Each panel includes representative foci-containing cells graph, over three experiments. *0.04<P<0.01; **0.009<P<0.001.

nisms. Next, we asked whether SIRT6 also mediates the recruitment of DNA repair factors to damage sites, which represents an attempt to preserve genomic integrity. We employed immunofluorescence to measure ability of AML cells expressing SIRT6 shRNA to recruit repair factors, including 53BP1, Rad51, RPA and γH2AX, to the sites of DNA damage following DDAs treatment. Genotoxic stress resulted in increased γH2AX foci formation as well as impaired Rad51, pRPA and 53BP1 foci formation in SIRT6-knockdown compared with SIRT6-wt AML cells (Figure 3B-D). Therefore, the simultaneous presence of increased DNA damage and decreased DNA DSBs repair explains the observed hypersensitivity of these cells to DDAs.

SIRT6 maintains genome integrity by deacetylation of DNA-PKcs and CtIP in AML cells To gain insights into specific function of SIRT6 in the context of DNA damage to AML cells, we characterized SIRT6-interacting proteins.30,36,37 GFP-tagged SIRT6 was expressed in OCI-AML3 cells and then immunoprecipitated with anti-GFP antibody. Western blot analysis revealed that DNA-PKcs and CtIP were enriched in the GFP-SIRT6 immunoprecipitates (IPs), mainly after DDAs treatment. Importantly, SIRT6 inhibition by compound 1 heavily reduced levels of both proteins, also in the presence of genotoxic stress (Figure 4A). Other SIRTs family 84

proteins, such as SIRT1, did not associate with GFPSIRT6 under these conditions, validating the specificity of the assay. Analysis of endogenous SIRT6 IPs confirmed this association, as well as its resistance to ethidium bromide, indicating that it is not due to DNA bridging (Figure 4B and Online Supplementary Figure S5). Our data, therefore, indicate that SIRT6 interacts physically with DNAPKcs and CtIP in AML cells, and that this interaction increases rapidly upon genotoxic stress. Since SIRT6 is a histone deacetylase, we next tested whether acetylation status of interacting proteins was affected by SIRT6 depletion. Each endogenous protein was pulled down separately after treatment with DDAs in both SIRT6-wt and SIRT6-KD AML cells. Although we readily detected acetylation of DNA-PKcs as well as CtIP in SIRT6 wildtype cells, their acetylation was abrogated after DNR and Ara-C treatment. In contrast, DNA damage-induced deacetylation of these proteins was totally abolished in SIRT6-depleted cells (Figure 4C and Online Supplementary Figure S6). These data suggest that DNA-PKcs and CtIP are constitutively acetylated in AML cells, and are deacetylated by SIRT6 following genotoxic stimuli, thereby promoting DNA damage repair. This observation was further confirmed by treating AML cells over-expressing human SIRT6(H133Y) catalytic mutant with increased doses of DDAs. DDAs treatment resulted in a more pronounced anti-tumor effect in AML cells over-expressing haematologica | 2018; 103(1)


SIRT6 inhibition increases anti-leukemic activity of chemotherapy

A

B

C

D

Figure 4. SIRT6 depletion/inhibition sensitizes acute myeloid leukemia (AML) cells to genotoxic agents by disrupting DNA repair machinery. (A) OCI-AML2 cells were engineered to express a GFP-tagged SIRT6. Cells were stimulated with compound 1, DNR (0.1 uM), or Ara-C (1 uM) for three hours (h). Thereafter, cells were used for protein lysate generation. SIRT6 in the different samples was co-immunoprecipitated using an anti-GFP antibody. Finally, GFP, CtIP, DNA-PKcs and SIRT1 levels were detected by immunoblotting. (B) OCI-AML2 cells were stimulated with compound 1, with (w) / or without (w/o) DNR (0.1 uM) for 3 h. Thereafter, cells were used for protein lysate generation. Endogenous SIRT6 in the different samples was immunoprecipitated using an anti-SIRT6 antibody and CtIP, DNA-PKcs, SIRT6, and SIRT1 levels were detected by immunoblotting. (C) OCI-AML2 engineered to express an shRNAs targeting SIRT6 (#911) were stimulated w / or w/o DNR (0.1 uM), or Ara-C (1 uM) for 3 h. CtIP (top) and DNA-PKcs (bottom) were immunoprecipitated and CtIP, DNA-PKcs, acetylated proteins (pan-acetyl-antibody) and Îł-tubulin were detected by immunoblotting. (D) Viability assays after Ara-C (left) or DNR (right) treatment of OCI-AML2 non-transfected cells, as well as in OCI-AML2 cells overexpressing SIRT6 wild-type or mutant (H133Y). *P=0.02; **P<0.001.

the catalytically inactive mutant than the wild-type form of SIRT6 (Figure 4D), indicating that enzymatic activity is required for SIRT6 to maintain genomic stability of AML cells.

Ongoing DNA damage is associated with intense replicative stress and high SIRT6 expression in AML cells Several studies have recently demonstrated a pervasive dysregulation of genomic stability in several cancers, including AML.8,38 To explore whether observed high SIRT6 expression was related to the constitutive DNA damage and intense replicative stress observed in AML cells, we used a chromosomal instability signature (CIN)16 to categorize AML cell lines included in a published dataset (GSE59808). A subset of approximately 40% AML cell lines demonstrated overexpression of probe sets belonging to CIN-signature (Figure 5A). To confirm this finding, we next explored a panel of AML cell lines together with primary tumor cells. Six of 9 AML cell lines, as well as primary cells derived from 10 AML patients, showed high Îł-H2A.X staining (Figure 5B and C) as well as activated DDR (Figure 5D). Remarkably, this pattern was absent in normal PBMCs derived from healthy individuals (Figure 5C), as already reported.39 Thus, such ongoing DNA damage observed in tumor cells did not induce an extensive cell death under basal condihaematologica | 2018; 103(1)

tions, suggesting existence of alternative mechanisms to escape apoptotic cell death triggered in normal cells. We had previously reported that SIRT6 preserves DNA integrity in MM cells.15 To investigate whether such deacetylase affects instability also in AML cells, we categorized leukemia cell lines included in GSE59808 according to their SIRT6 expression levels. AML cell lines with high CIN-signature exhibited greater SIRT6 mRNA levels (P=0.01) (Figure 5E). As a measure of specificity of this effect, we assessed gene expression profiles of AML cells based on their SIRT6 levels using Gene Set Enrichment Analysis.40 Remarkably, the gene expression profile defined by Carter et al.16 significantly correlates with SIRT6 expression in AML cells (P=0.02) (Figure 5F). In parallel, analysis of the entire set of transcription target gene signatures available from the Molecular Signatures Database (MSigDB) showed gene sets included in DNA replication and the cell-cycle regulatory gene pathway as also being significantly deregulated in these cells (data not shown), suggesting that SIRT6 drives DNA damage and activation of DNA damage response also in AML cells.

AML patients with SIRT6 overexpression show features of genomic instability and poor prognosis We next examined whether the broad DNA damage observed in AML patient-derived cells is also associated with SIRT6 mRNA levels. To this end, we probed sam85


A. Cagnetta et al.

A

B

D

C

E

F

Figure 5. Ongoing DNA damage and high CIN signature are associated with intense replicative stress and SIRT6 overexpression in acute myeloid leukemia (AML) cells. (A) Expression levels in a panel of 32 human AML cell lines for the probe sets corresponding to the chromosomal instability signature described by Carter et al.16 using GSE59808. Red: gene expression over the median; blue: expression under the media. (B) Immunofluorescence staining of γ-H2A.X in AML cell lines and primary tumor cells; magnification x40. (C) Western blot analysis (1 representative blot of 3) of γ-H2A.X in AML cell lines (top), AML patients’ cells, and peripheral blood mononuclear cells (PBMCs) from healthy donors (bottom). GAPDH, glyceraldehyde 3-phosphate dehydrogenase. (D) 53BP1 and RAD51 number of foci in a panel of AML cells presenting with high (red bracket) and low (blue bracket) DNA damage. (E) SIRT6 expression was compared to CIN signature among AML cell lines in the GSE59808 data set; *P=0.04. (F) GSEA enrichment profile for AML cell lines (included in GSE59808) divided in high and low SIRT6 expression groups of chromosomal instability signature, as reported by Carter et al.16 The analysis pointed to an association between high SIRT6 levels and CIN in AML cells.

A

B

C

D

E

Figure 6. High CIN gene expression signature confers poor prognosis in acute myeloid leukemia (AML) and correlates with SIRT6 expression. (A) Heat map showing CIN signature in 1157 AML patients compared to CD34+ cells derived from healthy individuals (GSE1159, GSE7186, GSE425, GSE12417 and GSE37642). Red: gene expression over the median; blue: gene expression under the media. (B) Expression levels for the probe sets corresponding to the chromosomal instability signature using GEP data of 524 AML patients (GSE14468). (C) Kaplan-Meier survival curves of AML patients showed in (B) based on their CIN gene expression signature. (D) GSEA enrichment profiles for AML patients included in GSE14468, divided into high and low SIRT6 expression groups of CIN signature as described by Carter et al.16 The analysis pointed to an association between high SIRT6 levels and CIN in AML cells. (E) SIRT6 expression was compared to CIN signature in AML patients described in the GSE14468 data set. **P=0.001; ****P=0.0001.

86

haematologica | 2018; 103(1)


SIRT6 inhibition increases anti-leukemic activity of chemotherapy

ples from 5 data sets, including tumor and CD34+ cells from healthy donors (GSE1159, GSE7186, GSE425, GSE12417 and GSE37642), for CIN gene expression signature. This analysis sharply divided samples into two groups, with AML patients over-expressing probe sets belonging to CIN signature compared with cells derived from healthy individuals (Figure 6A). To further characterize these data, we next investigated a publically available data set of 524 cases of de novo AML,41 observing that tumor samples can be split up into three groups according to the expression of genes included in CIN signature: low, intermediate and high (Figure 6B). Importantly, this arrangement did not overlap with other features, including cytogenetic abnormalities and FLT3 mutations (data not shown). Next, we analyzed the prognostic significance of these findings, observing that patients displaying higher CIN signature demonstrated poor prognosis compared with remaining patients (P<0.001) (Figure 6C). Finally, we analyzed these AML patients using GSEA. As observed in AML cell lines, this analysis revealed that CIN signature was the most significantly altered pathway measured in patients classified on the basis of their SIRT6 expression level [P=0.03, false discovery rate (FDR)=0.04] (Figure 6D). The DNA repair pathway and the BRCAness signature42 also differed in these patient subgroups (Online Supplementary Figure S7A and B). In line with these data,

higher SIRT6 levels were observed in patients with high CIN signature than those with intermediate or low SIRT6 expression level (Figure 6E). Taken together our results suggest a link between SIRT6 and genomic instability also in AML patient-derived samples, justifying the highest SIRT6 levels observed in more aggressive disease subtypes.

SIRT6 inhibition makes AML blasts more sensitive to DNR treatment in NSG mice To assess whether the biological results observed in vitro also occur in vivo, we used two different xenotransplant mouse models of AML. First, U937 scramble or SIRT6-KD stably transduced cells were injected subcutaneously into NSG mice (n=20). After tumor engraftment, mice (n=5) of each group were randomly assigned to receive either 3 mg/kg of DNR administered intraperitoneally (at day 1 and 5) or vehicle control.15 As in the in vitro setting, SIRT6 depletion made AML cells more sensitive to genotoxic agents, with a significant reduction of tumor growth in mice bearing these cells compared with tumors induced by AML cells carrying normal SIRT6 levels. Indeed, at day 30 after tumor injection, mean tumor volume was 60 versus 40 mm2, respectively (P=0.03) (Figure 7A). In a second in vivo model, we intravenously injected

A

B

C

D

Figure 7. SIRT6 inhibition makes acute myeloid leukemia (AML) blasts more sensitive to DNR treatment in NSG mice. (A) Growth of U937 control and SIRT6-depleted xenografts in mice treated with vehicle or DNR (3 mg/kg i.p. day 1 and day +4) at day 20 after tumor engraftment. *P=0.036. Data are mean tumor volumeÂąStandard Deviation (s.d.). (B) 1x106 of scramble or shSIRT6-expressing HL-60 cells were engrafted into NSG mice (n=20) by tail-vein injection. Once a systemic xenograft was confirmed, mice were randomized to receive DNR (1.5 mg/kg for 3 days) (treated group) or vehicle (control group). Histogram represents percentage of human CD45+ cells in mice, at day 31 post engraftment. Data are represented as meanÂąStandard Error of Mean (SEM); **P=0.006. (C) Representative flow cytometric dot plots representing tumor engraftment evaluated at day 40 post injection. (D) Kaplan-Meier survival plot showing median survival of mice injected with tumors with (w) / without (wo) SIRT6 before and after treatment with vehicle or DNR.

haematologica | 2018; 103(1)

87


A. Cagnetta et al.

human HL-60 cells, scramble or SIRT6 shRNA-transduced, into NSG mice (n =20; 5 mice per condition). Once a systemic xenograft was confirmed (>0.1% in peripheral blood of mice) the treatment regimen was initiated (1.5 mg/kg of DNR administered intraperitoneally, for 3 days, or vehicle control). At day 31 after cell transfer, flow cytometry evaluation of the circulating human CD45+ cells in the murine PB was performed to assess AML engraftment. This analysis revealed a significantly lower leukemia burden after DNR-treatment than vehicle (Figure 7B), with SIRT6 depletion making these cells more sensitive to chemotherapy (% of human engraftment: 0.9Âą0.1% and 0.16Âą0.01%, respectively; P=0.006), as observed in vitro. Tumor cell engraftment was measured also at day 40 and results showed that SIRT6-depleted treated mice had significantly fewer tumor cells compared with relative control (Figure 7C). Furthermore, Kaplan-Meier analyses indicated that DNR-treated mice injected with SIRT6 survived significantly longer than those bearing tumors with normal SIRT6 levels (56 vs. 39 days; P=0.004) (Figure 7D). Overall these data show that AML blasts depleted of SIRT6 are more sensitive to DDAs agents also in an in vivo environment, suggesting, therefore, evaluation of SIRT6 inhibition to be a novel strategy to enhance DDAs sensitivity in AML patients.

Discussion The efficiency of DNA-repair and DNA damageresponse pathways, affects both cancer susceptibility and responses to genotoxic agent-based therapies.33 As a result, synthetic lethal approaches to specifically kill cancer cells, that are dependent on compensatory DNA repair pathways, are emerging as a vulnerability that can be therapeutically targeted.39,43-45 In this context, we have recently shown that the chromatin-bound factor, SIRT6, safeguards the genome of MM cells.15 Here, we further extend these observations to AML cells and demonstrate that SIRT6 controls leukemogenesis and tumor growth by struggling with their instability. Indeed, we show that defects in SIRT6 expression or activity sensitize AML cells to genotoxic agents, leading to a significant reduction in blast-cell count, and to prolonged survival in AML mice models. Co-IP experiments have also demonstrated that SIRT6 deacetylates DNA-PKcs and CtIP, resulting in efficient DNA repair mechanisms and integrity of AML cells. In contrast, loss of SIRT6 enzymatic activity enhances instability, which in turn sensitizes leukemia cells to DDAs. Overall, our data suggest an innovative strategy to enhance efficacy of chemotherapy, which still remain the backbone for treatment, in AML. Additionally, based on low SIRT6 levels detected in normal CD34+ hematopoietic progenitors, a favorable therapeutic index of such an approach is also warranted. Genomic instability is one of the distinctive markers of tumor cells providing them with additional capabilities crucial for tumorigenesis.46-50 In hematologic cancers, the relevance of such features, and the mechanisms underlying instability are largely unknown.15,30,51-57 Based on our data, we assume that pervasive DNA damage observed in AML cells is reliant on genes such as SIRT6 that, when disrupted, lead to further instability.58,59 The prominent role exerted by SIRT6 on leukemogenesis is reinforced by its prognostic relevance, as observed in primary AML samples. Indeed, 88

SIRT6 overexpression is associated with greater instability and a worse prognosis. As a result, genetic inactivation of this chromatin remodeler triggers growth advantage and DNA repair weakening that in turn cause greater DDAs sensitivity. A comprehensive genomic analysis revealed that AML patients harbor several genetic alterations, including FLT3-ITD which primes leukemic cells to become genotoxic stress-induced.12 Here we observed higher SIRT6 mRNA expression levels in AML patients carrying FLT3ITD mutant, providing further evidence for a direct link between SIRT6 and genomic instability in AML. Nevertheless, these effects were not related to other specific genetic makeup, suggesting that SIRT6 acts on the genomic stability of AML regardless of its specific genomic landscape. As the cancer genome is itself reflective of phenotypic properties, specific gene signatures have been used to predict clinical outcome and identify prognostically relevant features in different cancer subtypes.60,61 Similarly, measurement of the degree of genomic instability, by leveraging specific gene signature, provides a valuable tool for prognostic stratification.62 Based on our data, here we asked whether consequences of aberrant DNA repair are reflected in genomic features, and how these events are associated with SIRT6 expression levels in AML cells. Therefore, we analyzed published databases for abnormal expression of genes belonging to chromosomal instability signature,14 recently identified as instability biomarker.43 The CIN-based stratification highly correlated with SIRT6 mRNA levels (Figure 6); greater instability was observed in patients harboring the highest SIRT6 levels which results in poor prognosis. Thus, our data identify SIRT6 as a valuable feature to segregate AML patients into distinct molecular and biological classes. Besides SIRT6, also SIRT1 promotes genomic integrity of tumor cells, proposing an over-lapping function.52,63 In such a scenario, a broad gene expression analysis of SIRTs family members revealed SIRT6 and SIRT1 to be at the top of the list, thus supporting the crucial role of these two proteins for AML cells (Online Supplementary Figure S8A and B). SIRT6 is a chromatin-bound deacetylase that participates in DNA double-strand break repair by affecting activity of several proteins, including CtIP, PARP1, DNA-PK complex and SNF2H at DNA damage sites.30,32,36,56 Here we show that, after genotoxic stress, AML cells rapidly recruit SIRT6 to DNA damaged sites where it deacetylates and promotes activity of DNA-PKcs and CtIP. In contrast, compromising SIRT6 activity decreases repair mechanisms, resulting in greater DDAs cytotoxicity both in vitro and in murine xenograft models of human AML. In summary, among the potential mechanisms that could cause instability, the disruption of the DNA repair complex is an intriguing avenue of research that should be pursued in anti-cancer therapies to increase the activity of currently used therapeutics. While an analysis of larger cohorts of patients may yet identify additional data on the specific impact of SIRT6 on genomic instability, here we identify such deacetylase action as a vulnerability to be exploited in developing future intervention strategies, and speculate as to its role as a surrogate genetic marker for instability in AML patients. Overall, our study provides proof-of-concept that depletion of SIRT6 represents a novel strategy to selectively target AML cells haematologica | 2018; 103(1)


SIRT6 inhibition increases anti-leukemic activity of chemotherapy

in order to enhance their sensitivity to currently used chemotherapies. Acknowledgments The authors thank Dr. Barbara Parodi and Dr. Paola Visconti (bio banking core facility at IRCCS AOU San Martino Hospital, Genova) for providing AML cell lines. In addition we thank all clinicians for their helpful suggestions.

References 1. Sant M, Allemani C, Tereanu C, et al. Incidence of hematologic malignancies in Europe by morphologic subtype: results of the HAEMACARE project. Blood. 2010; 116(19):3724-3734. 2. Ferrara F, Schiffer CA. Acute myeloid leukaemia in adults. Lancet. 2013;381(9865):484-495. 3. Dombret H, Gardin C. An update of current treatments for adult acute myeloid leukemia. Blood. 2016;127(1):53-61. 4. De Kouchkovsky I, Abdul-Hay M. Acute myeloid leukemia: a comprehensive review and 2016 update. Blood Cancer J. 2016;6(7):e441. 5. Guolo F, Minetto P, Clavio M, et al. High feasibility and antileukemic efficacy of fludarabine, cytarabine, and idarubicin (FLAI) induction followed by risk-oriented consolidation: A critical review of a 10-year, single-center experience in younger, non M3 AML patients. Am J Hematol. 2016;91(8):755-762. 6. Burnett AK, Hills RK, Milligan DW, et al. Attempts to optimize induction and consolidation treatment in acute myeloid leukemia: results of the MRC AML12 trial. J Clin Oncol. 2010;28(4):586-595. 7. Lowenberg B, Rowe JM. Introduction to the review series on advances in acute myeloid leukemia (AML). Blood. 2016; 127(1):1. 8. Esposito MT, So CW. DNA damage accumulation and repair defects in acute myeloid leukemia: implications for pathogenesis, disease progression, and chemotherapy resistance. Chromosoma. 2014;123(6):545-561. 9. Liu H, Takeda S, Kumar R, et al. Phosphorylation of MLL by ATR is required for execution of mammalian Sphase checkpoint. Nature. 2010; 467(7313):343-346. 10. Morgado-Palacin I, Day A, Murga M, et al. Targeting the kinase activities of ATR and ATM exhibits antitumoral activity in mouse models of MLL-rearranged AML. Sci Signal. 2016;9(445):ra91. 11. Fan J, Li L, Small D, Rassool F. Cells expressing FLT3/ITD mutations exhibit elevated repair errors generated through alternative NHEJ pathways: implications for genomic instability and therapy. Blood. 2010;116(24):5298-5305. 12. Sallmyr A, Fan J, Datta K, et al. Internal tandem duplication of FLT3 (FLT3/ITD) induces increased ROS production, DNA damage, and misrepair: implications for poor prognosis in AML. Blood. 2008;111(6):3173-3182. 13. Kugel S, Mostoslavsky R. Chromatin and beyond: the multitasking roles for SIRT6.

haematologica | 2018; 103(1)

Funding This work was supported in part by the Associazione Italiana per la Ricerca sul Cancro (AIRC, My First Grant #18491, to MC), Italian Ministry of Health (5 x 1000 Funds of IRCCS San Martino-IST 2014, to MC), Associazione Italiana Leucemie & SocietĂ Italiana di Ematologia Sperimentale fellowship (AIL-SIES, to DS) and University of Genova, Italy.

Trends Biochem Sci. 2014;39(2):72-81. 14. Finkel T, Deng CX, Mostoslavsky R. Recent progress in the biology and physiology of sirtuins. Nature. 2009;460(7255):587-591. 15. Cea M, Cagnetta A, Adamia S, et al. Evidence for a role of the histone deacetylase SIRT6 in DNA damage response of multiple myeloma cells. Blood. 2016;127(9):1138-1150. 16. Carter SL, Eklund AC, Kohane IS, Harris LN, Szallasi Z. A signature of chromosomal instability inferred from gene expression profiles predicts clinical outcome in multiple human cancers. Nat Genet. 2006;38(9):1043-1048. 17. Salvestrini V, Orecchioni S, Talarico G, et al. Extracellular ATP induces apoptosis through P2X7R activation in acute myeloid leukemia cells but not in normal hematopoietic stem cells. Oncotarget. 2017;8(4):5895-5908. 18. Bae JS, Park SH, Jamiyandorj U, et al. CK2alpha/CSNK2A1 Phosphorylates SIRT6 and Is Involved in the Progression of Breast Carcinoma and Predicts Shorter Survival of Diagnosed Patients. Am J Pathol. 2016;186(12):3297-3315. 19. Ran LK, Chen Y, Zhang ZZ, et al. SIRT6 Overexpression Potentiates Apoptosis Evasion in Hepatocellular Carcinoma via BCL2-Associated X Protein-Dependent Apoptotic Pathway. Clin Cancer Res. 2016;22(13):3372-3382. 20. Azuma Y, Yokobori T, Mogi A, et al. SIRT6 expression is associated with poor prognosis and chemosensitivity in patients with non-small cell lung cancer. J Surg Oncol. 2015;112(2):231-237. 21. Ming M, Han W, Zhao B, et al. SIRT6 promotes COX-2 expression and acts as an oncogene in skin cancer. Cancer Res. 2014;74(20):5925-5933. 22. Mostoslavsky R, Chua KF, Lombard DB, et al. Genomic instability and aging-like phenotype in the absence of mammalian SIRT6. Cell. 2006;124(2):315-329. 23. Michishita E, Park JY, Burneskis JM, Barrett JC, Horikawa I. Evolutionarily conserved and nonconserved cellular localizations and functions of human SIRT proteins. Mol Biol Cell. 2005;16(10):4623-4635. 24. Liszt G, Ford E, Kurtev M, Guarente L. Mouse Sir2 homolog SIRT6 is a nuclear ADP-ribosyltransferase. J Biol Chem. 2005;280(22):21313-21320. 25. Stirewalt DL, Meshinchi S, Kopecky KJ, et al. Identification of genes with abnormal expression changes in acute myeloid leukemia. Genes Chromosomes Cancer. 2008;47(1):8-20. 26. Eppert K, Takenaka K, Lechman ER, et al. Stem cell gene expression programs influence clinical outcome in human leukemia. Nat Med. 2011;17(9):1086-1093. 27. Bullinger L, Dohner K, Bair E, et al. Use of

28.

29.

30.

31.

32.

33. 34.

35.

36.

37.

38.

39.

40.

41.

gene-expression profiling to identify prognostic subclasses in adult acute myeloid leukemia. N Engl J Med. 2004; 350(16):1605-1616. Liu Y, Xie QR, Wang B, et al. Inhibition of SIRT6 in prostate cancer reduces cell viability and increases sensitivity to chemotherapeutics. Protein Cell. 2013;4(9):702-710. Huen MS, Chen J. The DNA damage response pathways: at the crossroad of protein modifications. Cell Res. 2008;18 (1):8-16. Kaidi A, Weinert BT, Choudhary C, Jackson SP. Human SIRT6 promotes DNA end resection through CtIP deacetylation. Science. 2010;329(5997):1348-1353. Van Meter M, Mao Z, Gorbunova V, Seluanov A. Repairing split ends: SIRT6, mono-ADP ribosylation and DNA repair. Aging. 2011;3(9):829-835. Toiber D, Erdel F, Bouazoune K, et al. SIRT6 recruits SNF2H to DNA break sites, preventing genomic instability through chromatin remodeling. Mol Cell. 2013;51(4):454-468. Curtin NJ. DNA repair dysregulation from cancer driver to therapeutic target. Nat Rev Cancer. 2012;12(12):801-817. Sociali G, Galeno L, Parenti MD, et al. Quinazolinedione SIRT6 inhibitors sensitize cancer cells to chemotherapeutics. Eur J Med Chem. 2015;102:530-539. Sociali G, Magnone M, Ravera S, et al. Pharmacological Sirt6 inhibition improves glucose tolerance in a type 2 diabetes mouse model. FASEB J. 2017;31(7):31383149. McCord RA, Michishita E, Hong T, et al. SIRT6 stabilizes DNA-dependent protein kinase at chromatin for DNA double-strand break repair. Aging. 2009;1(1):109-121. Simeoni F, Tasselli L, Tanaka S, et al. Proteomic analysis of the SIRT6 interactome: novel links to genome maintenance and cellular stress signaling. Sci Rep. 2013;3:3085. Rebechi MT, Pratz KW. Genomic instability is a principle pathologic feature of FLT3 ITD kinase activity in acute myeloid leukemia leading to clonal evolution and disease progression. Leuk Lymphoma. 2017:1-11. Cottini F, Hideshima T, Xu C, et al. Rescue of Hippo coactivator YAP1 triggers DNA damage-induced apoptosis in hematological cancers. Nat Med. 2014;20(6):599-606. 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. Taskesen E, Bullinger L, Corbacioglu A, et al. Prognostic impact, concurrent genetic mutations, and gene expression features of AML with CEBPA mutations in a cohort of

89


A. Cagnetta et al.

42.

43.

44.

45.

46. 47. 48.

90

1182 cytogenetically normal AML patients: further evidence for CEBPA double mutant AML as a distinctive disease entity. Blood. 2011;117(8):2469-2475. Konstantinopoulos PA, Spentzos D, Karlan BY, et al. Gene expression profile of BRCAness that correlates with responsiveness to chemotherapy and with outcome in patients with epithelial ovarian cancer. J Clin Oncol. 2010;28(22):3555-3561. Cottini F, Hideshima T, Suzuki R, et al. Synthetic Lethal Approaches Exploiting DNA Damage in Aggressive Myeloma. Cancer Discov. 2015;5(9):972-987. Ka NL, Na TY, Na H, et al. NR1D1 recruitment to sites of DNA damage inhibits repair and is associated with chemosensitivity of breast cancer. Cancer Res. 2017; 77(9):2453-2463. Brown JS, O'Carrigan B, Jackson SP, Yap TA. Targeting DNA Repair in Cancer: Beyond PARP Inhibitors. Cancer Disc. 2017;7(1):20-37. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011; 144(5):646-674. Friedberg EC. DNA damage and repair. Nature. 2003;421(6921):436-440. Abraham RT. Checkpoint signaling: epigenetic events sound the DNA strand-breaks alarm to the ATM protein kinase. BioEssays. 2003;25(7):627-630.

49. Jackson SP, Bartek J. The DNA-damage response in human biology and disease. Nature. 2009;461(7267):1071-1078. 50. Stratton MR, Campbell PJ, Futreal PA. The cancer genome. Nature. 2009; 458(7239):719-724. 51. Kim HS, Vassilopoulos A, Wang RH, et al. SIRT2 maintains genome integrity and suppresses tumorigenesis through regulating APC/C activity. Cancer Cell. 2011; 20(4):487-499. 52. Wang RH, Sengupta K, Li C, et al. Impaired DNA damage response, genome instability, and tumorigenesis in SIRT1 mutant mice. Cancer Cell. 2008;14(4):312-323. 53. Jeong SM, Xiao C, Finley LW, et al. SIRT4 has tumor-suppressive activity and regulates the cellular metabolic response to DNA damage by inhibiting mitochondrial glutamine metabolism. Cancer Cell. 2013; 23(4):450-463. 54. Marquardt JU, Fischer K, Baus K, et al. Sirtuin-6-dependent genetic and epigenetic alterations are associated with poor clinical outcome in hepatocellular carcinoma patients. Hepatology. 2013;58(3):10541064. 55. Michishita E, McCord RA, Berber E, et al. SIRT6 is a histone H3 lysine 9 deacetylase that modulates telomeric chromatin. Nature. 2008;452(7186):492-496. 56. Mao Z, Hine C, Tian X, et al. SIRT6 pro-

57.

58.

59.

60. 61.

62. 63.

motes DNA repair under stress by activating PARP1. Science. 2011;332(6036):14431446. Mao Z, Tian X, Van Meter M, Ke Z, Gorbunova V, Seluanov A. Sirtuin 6 (SIRT6) rescues the decline of homologous recombination repair during replicative senescence. Proc Natl Acad Sci USA. 2012; 109(29):11800-11805. Sperlazza J, Rahmani M, Beckta J, et al. Depletion of the chromatin remodeler CHD4 sensitizes AML blasts to genotoxic agents and reduces tumor formation. Blood. 2015;126(12):1462-1472. Cea M, Soncini D, Fruscione F, et al. Synergistic interactions between HDAC and sirtuin inhibitors in human leukemia cells. PloS One. 2011;6(7):e22739. Stratton MR. Exploring the genomes of cancer cells: progress and promise. Science. 2011;331(6024):1553-1558. Negrini S, Gorgoulis VG, Halazonetis TD. Genomic instability--an evolving hallmark of cancer. Nature reviews Mol Cell Biol. 2010;11(3):220-228. Chin L, Gray JW. Translating insights from the cancer genome into clinical practice. Nature. 2008;452(7187):553-563. Sasca D, Hahnel PS, Szybinski J, et al. SIRT1 prevents genotoxic stress-induced p53 activation in acute myeloid leukemia. Blood. 2014;124(1):121-133.

haematologica | 2018; 103(1)


ARTICLE

Acute Myeloid Leukemia

Variable outcome and methylation status according to CEBPA mutant type in double-mutated acute myeloid leukemia patients and the possible implications for treatment

Ferrata Storti Foundation

Dima El-Sharkawi,1 Duncan Sproul,2 Christopher G. Allen,1 Andrew Feber,3 Melissa Wright,4 Robert K. Hills,4 David C. Linch1 and Rosemary E. Gale1

Department of Haematology, UCL Cancer Institute, London; 2MRC Human Genetics Unit and Edinburgh Cancer Research Centre, MRC Institute of Genetics and Molecular Medicine, University of Edinburgh, Edinburgh; 3Medical Genomics, UCL Cancer Institute, London and 4Centre for Trials Research, Cardiff University, UK

1

Haematologica 2018 Volume 103(1):91-100

ABSTRACT

A

lthough CEBPA double-mutated (CEBPADM) acute myeloid leukemia is considered to be a favorable-risk disease, relapse remains a major cause of treatment failure. Most CEBPADM patients have a classic biallelic mutant combination with an N-terminal mutation leading to production of p30 protein plus a C-terminal loss-offunction in-frame indel mutation (CEBPAClassic-DM), but approximately onethird of cases have one or more non-classic mutations, with diverse combinations reported, and there is little information on the consequences of such mutants. We evaluated outcome in a cohort of 104 CEBPADM patients, 79 CEBPAClassic-DM and 25 with non-classic mutants, and found that the latter may have poorer survival (5-year overall survival 64% vs. 46%; P=0.05), particularly post relapse (41% vs. 0%; P=0.02). However, for this analysis, all non-classic cases were grouped together, irrespective of mutant combination. As CEBPADM cases have been reported to be hypermethylated, we used methylation profiling to assess whether this could segregate the different mutants. We developed a CEBPAClassic-DM methylation signature from a preliminary cohort of 10 CEBPADM (including 8 CEBPAClassic-DM) and 30 CEBPA wild-type (CEBPAWT) samples, and independently validated the signature in 17 CEBPAClassic-DM cases. Assessment of the signature in 16 CEBPADM cases with different non-classic mutant combinations showed that only 31% had a methylation profile equivalent to CEBPAClassic-DM whereas for 69% the profile was either intermediate between CEBPAClassic-DM and CEBPAWT or equivalent to CEBPAWT. These results suggest that CEBPADM cases with non-classic mutants may be functionally different from those with CEBPAClassic-DM mutants, and should not automatically be included in the same prognostic group. (AML12 is registered under ISRCTN17833622 and AML15 under ISRCTN17161961).

Introduction The CEBPA gene encodes CCAAT/enhancer binding protein-α (C/EBPα), a basic leucine zipper (bZIP) transcription factor that is essential for hematopoietic stem cell regulation and myeloid development.1,2 The gene is mutated in approximately 8% of acute myeloid leukemia (AML) patients with intermediate-risk cytogenetics, and presence of biallelic double mutations (CEBPADM) in the absence of an FLT3 internal tandem duplication (FLT3ITD) is associated with a favorable prognosis.3-9 In the current risk-adapted therapy strategies for AML,10-12 these patients are classified as good-risk and therefore not usually considered for consolidation of first remission by allogeneic transplantation.11,13 Mutations occur throughout the single exon gene but predominate at the N and C termini.14 N-terminal mutations are nearly always frameshift or nonsense mutahaematologica | 2018; 103(1)

Correspondence: rosemary.gale@ucl.ac.uk

Received: May 19, 2017. Accepted: October 10, 2017. Pre-published: October 12, 2017. doi:10.3324/haematol.2017.173096 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/91 ©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.

91


D. El-Sharkawi et al.

tions causing increased translation from an internal ATG start site and production of a truncated p30 protein that retains the same reading-frame as the full-length p42 protein but lacks the first transactivation domain (TAD1). The most common C-terminal mutations are in-frame indels in the bZIP DNA binding domain (DBD) or leucine zipper domain (LZD) that lead to loss of the ability to bind to DNA or dimerize, classified here as C-terminal loss-offunction (C-LOF). However, many other mutations have also been reported, including missense mutations in the DBD or LZD, missense mutations and in-frame indels in the mid-region, and frameshift or nonsense mutations in the mid-region or C-terminus. Some also present as homozygous alterations due to chromosome 19 uniparental disomy.15,16 The most common combination of mutations is an N-terminal frameshift on one allele plus a C-terminal in-frame indel on the other allele that together are predicted to lead to complete loss of normal p42 C/EBPα activity,14 hereafter called the classic CEBPADM combination (CEBPAClassic-DM). This combination was identified in 204 of 305 CEBPADM cases (67%) with defined mutants reported in six studies containing 20 or more CEBPADM cases.3,6,8,17-19 The remaining 101 cases had multiple different mutant combinations with diverse consequences: 54 (18% of total CEBPADM) would be predicted to produce just p30 due to the presence of a mid-region or Cterminal truncation, 12 (4%) would only produce a classic C-LOF protein, 19 (6%) p30 plus a C-terminal missense mutant, 5 (2%) just a C-missense mutant, and 11 (4%) other mutant combinations. Understanding the consequence of the different types of CEBPADM mutants is needed as this may impact on clinical outcome, but there is limited information available on the specific mutations. CEBPADM cases have gene expression profiles that are significantly different from single-mutated (CEBPASM) and wild-type (CEBPAWT) cases.3,7 However, recent data suggest that CEBPADM cases with non-classic combinations may not always cluster with other CEBPADM cases,7,20 with 3 of 7 such cases classified as negative for the CEBPADM expression profile in one report.20 Furthermore, genotype stratification according to expression profiling may be confounded by CEBPAWT cases with completely silenced CEBPA expression (CEBPASIL) due to methylation of the CEBPA promoter as these cases cluster together with CEBPADM cases.21 CEBPADM cases also form distinct epigenetic clusters with a markedly hypermethylated profile.22,23 But although CEBPASIL cases also have a hypermethylated profile, this segregates from the CEBPADM cluster and, interestingly, they appear to be associated with a biologically distinct subtype of AML with a poor prognosis.21,22,24 Potential methylation differences according to the underlying combination of CEBPA mutants, however, have not been reported. In order to investigate potential differences between CEBPADM mutant combinations, we evaluated clinical outcome in 104 CEBPADM cases (79 CEBPAClassic-DM and 25 with non-classic mutants) and observed that the non-classic cases may have a lower overall survival (OS). However, the number of cases was relatively small and all cases with a non-classic mutant were included in this group, irrespective of the mutant type. We therefore investigated whether methylation profiling of samples from doublemutated patients could assist in segregating the different mutant combinations. 92

Methods Patient cohorts The patients investigated were younger adults entered into the UK MRC AML10, AML12 and AML15 trials. Informed patient consent was obtained in accordance with the Declaration of Helsinki, and ethical approval for tissue use from the Wales Research Ethics Committee 3. Clinical outcome was evaluated in 104 CEBPADM patients, all under 60 years of age, and methylation profiling was performed on 135 patients: 132 (98%) of them under 60 years of age (Online Supplementary Figure S1). Selected tests were performed on a further 82 samples with specific cytogenetic and molecular abnormalities.

Therapy, clinical end points and statistical methods Details of clinical protocols, end points and statistical methods are defined in the Online Supplementary Appendix.

Methylation arrays and data processing DNA was bisulfite-converted using the EZ DNA MethylationGold Kit (Zymo Research, California, USA) and random samples checked by methylation-specific PCR to ensure efficient conversion (see the Online Supplementary Appendix). Methylation profiling was performed using the Illumina Infinium HumanMethylation27 (n=40, cohort 1) and HumanMethylation450 (n=95, cohort 2) BeadChip arrays (Illumina Inc., California, USA). Details of data processing are given in the Online Supplementary Appendix. Derived β values were expressed as the percentage methylation at a given CpG probe. Selected CpG sites were further analyzed using pyrosequencing assays (see the Online Supplementary Appendix).

Unsupervised analysis of patient methylation profiles

In a given sample, probes were defined as methylated if the β value was more than 0.3, unmethylated if 0.3 or under.25 Samples were clustered based on their β values at probes displaying significant variation (variable probes), as previously defined.25,26 They were defined as variable if methylated in 1 sample or more plus unmethylated in 1 sample or more. CpG islands (CGIs) were located as previously defined;27 their methylation levels were derived by calculating the mean β value of probes at these locations. Hierarchical clustering was performed using the Euclidian distance of β values and Ward algorithm in R.

Derivation and analysis of CEBPA signature A methylation signature of CEBPA genotype was derived from variably methylated CpGs in cohort 1. Signature CpGs were selected as the top 25 ranked probes based on the mean rank of P-values of Wilcoxon tests and the absolute median difference in β value between CEBPAClassic-DM and CEBPAWT samples. Probes were included in the analysis if more than 90% of samples and 2 or more CEBPAClassic-DM and 2 or more CEBPAWT samples had observable data. Methylation signatures of CEBPAClassic-DM and CEBPAWT samples were then defined as the median β values observed at these 25 probes. Samples were scored relative to these signatures by calculating the Euclidian distance between the signatures and their profiles at signature probes.

Mutant CEBPA level and confirmation of biallelic status Mutant CEBPA level was quantified as previously described6 or approximated from the sequence chromatogram (average height of ≥5 peaks). Monoallelic/biallelic status was investigated by sequencing clones derived from full-length CEBPA amplicons as previously described.6 haematologica | 2018; 103(1)


Variable outcome and methylation in CEBPADM AML

Table 1. CEBPA genotype of investigated cohorts.

CEBPA genotype* Clinical cohort (n=104)

Methylation Cohort 1 (n=40)

Methylation Cohort 2 (n=95)

CEBPAClassic-DM Non-classic CEBPADM Classic N + C-missense Classic N + mid-frameshift Classic N + C-frameshift Homozygous classic N Homozygous classic C Homozygous C-missense Classic C + C-frameshift Classic C + C-missense Mid-frameshift + C-missense CEBPAClassic-DM Non-classic CEBPADM Classic C + C-frameshift Homozygous C-missense CEBPAWT CEBPAClassic-DM Non-classic CEBPADM Classic N + C-missense Classic N + mid-frameshift Classic N + C-frameshift Homozygous classic C Homozygous C-missense Classic C + C-missense CEBPASM Classic N Classic C Mid-indel Mid-frameshift Mid-missense C-frameshift C-missense CEBPAWT

Number

Predicted functional consequence

79

p30 + C-LOF

4 5 2 1 4 3 4 1 1 8

p30 + C-LOF p30 p30 p30 C-LOF C-LOF C-LOF C-LOF C-LOF p30 + C-LOF

1 1 30 17

C-LOF C-LOF WT p30 + C-LOF

3 5 1 2 2 1

p30 + C-LOF p30 p30 C-LOF C-LOF C-LOF

9 5 3 9 5 2 5 26

p30 + WT C-LOF + WT UNK + WT Null**+ WT UNK + WT Null** + WT C-LOF + WT WT

*Details of the specific mutations are given in Online Supplementary Table S1.**Mid-region or C-terminal mutants with a truncating frameshift or nonsense mutation. C: C-terminal mutation; C-LOF: C-terminal loss-of-function; DM: double mutant; indel: in-frame insertion and/or deletion; N: N-terminal mutation; SM: single mutant; UNK: unknown; WT: wild-type.

Results Clinical outcome in CEBPADM cases according to mutant type Of the 104 CEBPADM cases evaluated, 79 had classic and 25 had non-classic mutants (Table 1 and Online Supplementary Table S1). The latter included 4 patients predicted to produce p30 plus a C-terminal missense mutant, 8 with combinations predicted to produce just p30, and 13 with different combinations predicted to produce just a CLOF protein. There was no difference in the baseline characteristics between the classic and non-classic cases, including white cell count, sex, World Health Organization (WHO) performance status, and AML type (de novo or secondary), although the non-classic cases were haematologica | 2018; 103(1)

older (median 35 vs. 47 years; P=0.001) (Online Supplementary Table S2). All classic cases had intermediaterisk cytogenetics compared to 89% of the non-classic cases (P=0.05). There was no significant difference in the incidence of FLT3ITD, NPM1, IDH1, WT1, TET2 and GATA2 mutations between the groups, although the non-classic cases had more IDH2 mutations (1% vs. 20%, P=0.003; all IDH2R140Q) and DNMT3A mutations (3% vs. 16%, P=0.03). Median follow up was 9.5 years (range: 0.1-22 years). Neither the proportion of transplanted patients, the type of transplant, nor the stage of transplantation differed between the groups (Online Supplementary Table S2). CEBPAClassic-DM cases had a slightly higher but statistically non-significantly different complete remission (CR) rate to non-classic cases (95% vs. 88%, P=0.2) (Table 2). In uni93


D. El-Sharkawi et al.

A

B

P=0.02

P=0.05

C

D

P=0.004 P=0.08

Figure 1.Clinical outcome in CEBPADM patients according to mutant combination. (A) Overall survival in the total cohort of 104 CEBPADM patients. (B) Survival post relapse in the 39 patients who relapsed. (C) Survival from second remission in the 23 relapsed patients who achieved a second remission. (D) Overall survival in the total cohort excluding patients with FLT3ITD and DNMT3A mutations.

variate analysis, there was no significant difference in relapse-free survival or relapse rate (RR). However, there was a trend towards a lower OS in the non-classic cases (64% vs. 46% at 5 years, P=0.05) (Figure 1A), which was largely due to the worse outcome post-relapse in the nonclassic cases. The proportion of relapsing patients who achieved a second remission did not differ between the groups (61% vs. 55%), but 5-year survival post relapse was 41% versus 0% (P=0.02) (Figure 1B), and 59% versus 0% from second remission (P=0.004) (Figure 1C). The survival differences were not, however, statistically different in multivariate analysis (Table 2), but this is not surprising given the small group sizes. Although the significantly higher proportion of DNMT3A-mutated cases in the nonclassic group could have contributed to their worse outcome, as both these mutations and FLT3ITD adversely impact on the favorable outcome associated with CEBPADM AML,6,28 the trend towards a lower OS in the non-classic cases was still present when patients with these mutations were excluded (70% vs. 53%; Hazard Ratio 1.95, 95% confidence intervals 0.95-4.01; P=0.08) (Figure 1D).

Development of a CEBPAClassic-DM methylation signature The clinical evaluation grouped all non-classic cases together in order to obtain an adequate number of patients for analysis, but this cohort, therefore, included patients with many mutant combinations predicted to have differing functional consequences. In order to 94

explore potential methods for discriminating between these combinations, we investigated the impact of mutant type on methylation profiles. A preliminary cohort of samples from 40 normal karyotype (NK) FLT3WTNPM1WT patients were investigated using the Illumina Infinium HumanMethylation27 array: 10 were CEBPADM and 30 were CEBPAWT (Methylation cohort 1) (Table 1), approximating the relative proportions of such cases in NK FLT3WTNPM1WT patients in our earlier study.6 The array data were validated by pyrosequencing assays at four differentially methylated CpG sites (Online Supplementary Figure S2). Most CpG sites analyzed showed little variation in methylation levels across the whole cohort, but unsupervised cluster analysis according to levels of the most variably methylated probes revealed two main clusters. All 10 CEBPADM samples, including two non-classic cases, fell in the cluster of 16 samples with significantly higher levels of mean CGI methylation (Figure 2A), and the mean level of CGI methylation was significantly different between CEBPAClassic-DM and CEBPAWT samples (Figure 2B). A supervised approach was then used to create CEBPAClassic-DM and CEBPAWT methylation signatures based on the 25 most differentially methylated sites between the CEBPAClassic-DM and CEBPAWT samples (Figure 2C and Online Supplementary Table S3). Two distance scores were calculated for each sample based on the Euclidian distance between their methylation levels at these signature probes and the median profile of the CEBPAClassic-DM and CEBPAWT samples. This confirmed that, when assessed according to haematologica | 2018; 103(1)


Variable outcome and methylation in CEBPADM AML

Table 2. Outcome according to CEBPADM mutant combination.

Outcome

Classic CEBPADM (n=79)

Non-classic CEBPADM (n=25)

CR 5-year OS 5-year RFS 5-year RR

95% 64% 49% 41%

88%** 46% 45% 47%

CEBPAClassic-DM vs. non-classic CEBPADM OR or HR (95% CI), P Univariate Multivariate* 3.20 (0.53-19.47), P=0.2 1.84 (1.01-3.37), P=0.05 1.15 (0.61-2.16), P=0.7 1.32 (0.66-2.66), P=0.4

Not evaluable*** 1.44 (0.74-2.79), P=0.3 1.03 (0.52-2.06), P=0.9 1.09 (0.47-2.55), P=0.8

*Adjusted for age, white blood cell count, World Health Organization performance status, type of leukemia, sex, FLT3 and DNMT3A genotype. **Remission status was missing for one patient. ***Insufficient events for analysis. P: P-value; n: number; CI: confidence intervals; CR: complete remission; DM: double mutant; HR: hazard ratio; OR: odds ratio; OS: overall survival; RFS: relapse-free survival; RR: relapse rate.

their distance scores, the CEBPAClassic-DM samples formed a distinct group that were clearly separate from the CEBPAWT samples (Figure 3A). The CEBPAClassic-DM methylation signature was validated with samples from a further 17 CEBPAClassic-DM and 26 CEBPAWT cases (Methylation cohort 2) (Table 1) using the HumanMethylation450 array. Sixteen of the 17 CEBPAClassic-DM cases (94%) fell in the same cluster in unsupervised analysis, with a relatively more hypermethylated profile, and all CEBPAWT cases fell in the hypomethylated cluster (Online Supplementary Figure S3). Using supervised analysis according to the derived CEBPAClassic-DM and CEBPAWT signatures, the same 16 CEBPADM cases had a methylation profile that was closest to the CEBPAClassic-DM signature with a wide difference between the two signatures (Figures 3B and 4), indicating that their profile was equivalent to the CEBPAClassic-DM cases in cohort 1. Further analysis of the one CEBPAClassic-DM case that fell closer to the CEBPAWT group indicated that the mutations were biallelic but only approximately half of the cells in the sample carried mutations; mean mutant level was 28% for the pair, which was the lowest mean level of all 25 CEBPAClassic-DM cases (median 44%; range: 28-50%). The methylation profile of this case could, therefore, have been affected by the presence of a significant proportion of non-leukemic cells and it was excluded from further analyses. Using the distance scores (mean±2SD) for CEBPAClassic-DM cases in cohort 1, tests showed that these scores classified CEBPAClassic-DM and CEBPAWT genotypes in the second cohort with 95% accuracy, 88% sensitivity, and 100% specificity. The distance scores of the 24 CEBPAClassic-DM cases from the combined cohorts were then used to define a CEBPAClassic-DM quadrant that segregated all CEBPAClassic-DM cases from CEBPAWT cases (Figure 3C).

Investigation of the CEBPAClassic-DM methylation signature in other good-risk patients In order to examine whether the CEBPAClassic-DM methylation signature could simply reflect ‘good-risk’ disease or be due to a lack of C/EBPα expression, methylation levels at three differentially methylated CpG sites from the signature were quantified by pyrosequencing using samples from 21 patients with inv(16) and 19 with t(8;21), both associated with down-regulated C/EBPα expression,29-31 and 42 with NPM1MUTFLT3WT. KHNYN and VAMP5 were more hypermethylated in the CEBPAClassic-DM signature, while LY9 was more hypomethylated. Each subgroup differed significantly from the 24 CEBPAClassic-DM cases at two of the three sites (Online Supplementary Figure S4). A comhaematologica | 2018; 103(1)

posite methylation score was calculated for each sample; it was statistically significantly different from the CEBPAClassic-DM cases for all three subgroups (Figure 5), indicating that the methylation profile was a distinct feature of the mutant proteins that lead to total loss of normal C/EBPα function rather than absence of C/EBPα per se.

Investigation of non-classic CEBPADM and CEBPASM mutants Having established that CEBPAClassic-DM cases have a methylation profile that is distinct from CEBPAWT cases, profiles of 14 CEBPADM cases with a variety of different non-classic combinations were investigated using the HumanMethylation450 array (Methylation cohort 2) (Table 1). On unsupervised analysis, 9 (64%) were hypermethylated and 5 (36%) hypomethylated (Online Supplementary Figure S3). To assess their impact on the CEBPA methylation signatures, these 14 cases and the 2 non-classic cases from the initial cohort were considered according to the predicted functional consequence of the combination. Six cases were predicted to produce just p30 protein, with a classic N plus null mutant (mid-region or C-terminal frameshift/nonsense) combination. Only 1 fulfilled the CEBPAClassic-DM criteria; 4 fell outside this quadrant but were still distinct from CEBPAWT cases, and 1 grouped with the CEBPAWT cases (Figure 6A). Three cases were predicted to just give rise to classic C-LOF proteins without p30 (2 homozygous classic C, 1 compound heterozygous with a classic C plus C-frameshift combination). Two fulfilled the CEBPAClassic-DM criteria and 1 was intermediate between CEBPAClassic-DM and CEBPAWT. The 7 cases with missense mutations were also highly variable. Three had a classic N plus C-missense combination; none fulfilled the CEBPAClassic-DM criteria, 2 were intermediate between the signatures, and 1 grouped with CEBPAWT cases. Mutant levels were indicative of 80% or more mutated cells in all 3 cases; 2 were biallelic by cloning but no full-length amplicons could be obtained in 1 case. Similarly, only 1 of the 3 homozygous C-missense cases fell in the CEBPAClassic-DM quadrant, the other 2 were equivalent to CEBPAWT cases. The remaining case was compound heterozygous with a classic C and C-missense combination, and this did fulfil the CEBPAClassic-DM criteria. These results suggest that the functional consequence of double-mutated cases producing at least 1 non-classic mutant protein can be highly variable and difficult to predict. Of note, when outcome was assessed in the non-classic cases according to their methylation profile, there was 95


D. El-Sharkawi et al.

a suggestion that those that fulfilled the CEBPAClassic-DM methylation criteria were less likely to relapse, with only 1 of 4 cases (25%) relapsing compared to 7 of 10 cases (70%) that fell outside the CEBPAClassic-DM quadrant (5-year RR 25% vs. 54%, respectively). These numbers were too small for meaningful statistical analysis, and the results do not necessarily indicate a causal link between the methylation pattern and outcome. They do suggest, however,

that the methylation pattern could be a useful biomarker, and thereby act as a surrogate for response and selection of therapy. Methylation cohort 2 also included 38 CEBPASM cases with a wide range of classic and non-classic mutants (Table 1). On unsupervised analysis, 31 (82%) were in the hypomethylated cluster, with no apparent segregation between the CEBPASM and CEBPAWT cases (Online

A

B

C

Figure 2. Methylation profile of the preliminary cohort of 40 samples. (A) Unsupervised cluster analysis of the cohort. Each column represents one patient. Genotype is given in the upper panel. Samples were clustered based on their methylation levels at 7679 variable probes, and the heatmap in the middle panel shows the variable CpG probes located within CpG islands (CGIs). The latter were used to calculate the mean % CpG methylation shown in the lower panel; red and blue bars indicate a predominantly hyper- or hypo-methylated profile, respectively. (B) The mean level CGI methylation for CEBPAWT and CEBPAClassic-DM samples. CGI methylation levels were calculated from all autosomal CGI probes. ***P<0.001 (Wilcoxon test). (C) Supervised cluster analysis showing the derived CEBPAClassic-DM signature (left) and the CEBPAWT signature (right). Samples are ordered according to their level of similarity to the CEBPAClassic-DM signature using Euclidian distance. (Bottom) Distance scores indicating the distance from the CEBPAClassic-DM (green circles) and CEBPAWT (white circles) signatures; the lower the y-axis value, the more closely the sample matches that particular signature.

96


Variable outcome and methylation in CEBPADM AML

Supplementary Figure S3). On supervised analysis, all except 2 cases were equivalent to CEBPAWT cases, with no obvious grouping according to the type of mutant (Figures 4 and 6B). The remaining 2 cases fell in or close to the CEBPAClassic-DM quadrant; both had a classic C mutation. It is possible that the CEBPAWT allele had been silenced in these cases, but RNA samples were not available to check this.

Discussion Molecular genotyping is increasingly used to risk stratify patients with AML, but clinical application of this information needs to be accurate and robust for optimal patient therapy. This is particularly important for good-risk patients such as those with biallelic CEBPA mutations, where the current recommendation is not to proceed to transplantation in first remission, as for some patients this could lead to undertreatment. Identifying those who are at greater risk of relapse and poorer survival may, therefore, guide patient management. Very limited information is available on the impact of the different CEBPA mutations, with all CEBPADM cases currently being classified as goodrisk, irrespective of the underlying mutant. However, mutations identified in approximately one-third of CEBPADM patients do not conform to the classic combination of N-terminal frameshift or nonsense mutation plus C-terminal in-frame indel. As many of the other mutations are unique and of unknown functional consequence, determining their significance is challenging, particularly in view of the multiple functions attributed to C/EBPÎą.1,2 Although we had access to a database of 2162 patients with known CEBPA genotype and available clinical data from three consecutive UK MRC trials of younger adult patients with AML, 67% with intermediate-risk cytogenetics, we were still able to evaluate long-term outcome in a cohort of only 79 CEBPAClassic-DM cases and 25 CEBPADM cases with at least one non-classic mutant. The results suggested that the non-classic cases may have a poorer outcome; in particular, none of these cases survived after relapse. These results may be influenced by differences in the coincident mutations in the two groups. The majority of recurrent mutations with known prognostic significance in AML are uncommon in CEBPADM cases,32 and for many of them their impact in this subgroup is, therefore, not well defined. Both FLT3ITD and DNMT3A mutations adversely impact on the favorable outcome of CEBPADM AML.6,28 But even when patients with these mutations were excluded from the analysis, OS for the non-classic cases was still worse. The incidence of GATA2 mutations was non-significantly lower in the non-classic cases, but although one study reported that they are associated with better OS,33 most studies, including our own, observed no difference.34,35 There was a non-significantly higher incidence of TET2 mutations in the non-classic cases, but their impact is unclear, with only one study reported specifically for CEBPADM AML that showed a worse OS but not event-free survival in TET2MUT cases.33 Clearly, many more cases would need to be analyzed in order to take into account coincident mutations other than FLT3ITD and DNMT3A mutations. A wide range of non-classic mutations was observed and, for the outcome analysis, sufficient patient numbers could only be obtained by grouping all the non-classic haematologica | 2018; 103(1)

patients together, which precluded evaluation of specific mutant combinations. We, therefore, sought alternative methods of assessment and, as CEBPADM patients have a distinct hypermethylated profile,22,23 investigated whether genome-wide methylation profiling could provide information on the more broad-spectrum functional consequence of different mutants. We confirmed the relatively hypermethylated profile associated with a CEBPADM geno-

A

B

C

Figure 3. Analysis of the methylation profiles of the samples according to their distances from the derived CEBPAClassic-DM and CEBPAWT methylation signatures. Difference between the distance scores (CEBPAClassic-DM - CEBPAWT) compared with the distance from the CEBPAClassic-DM (MUT) signature of (A) the preliminary cohort and (B) the CEBPAClassic-DM and CEBPAWT cases in the follow-up cohort. (C) Criteria derived for CEBPAClassic-DM using meanÂą2Standard Deviation (SD) of the distance scores of all 24 CEBPAClassic-DM cases.

97


D. El-Sharkawi et al.

Figure 4. Analysis of the follow-up cohort of 95 samples. Supervised analysis according to the CEBPAClassic-DM and CEBPAWT methylation signatures. Patient CEBPA genotype is given above the heatmap. (Bottom) Distance scores indicating the distance from the CEBPAClassic-DM (green circles) and CEBPAWT (white circles) signatures; the lower the y-axis value, the more closely the sample matches that particular signature.

type in a preliminary cohort and then derived methylation signatures for CEBPAClassic-DM and CEBPAWT cases using the 25 most differentially methylated CpG sites. These were validated in an independent cohort of samples, with 16 of the 17 CEBPAClassic-DM cases studied clearly separating from the CEBPAWT cases. The remaining case contained a significant proportion of non-mutated cells, which provided indirect evidence that the signatures reflected CEBPA genotype. The signatures were a distinct feature of mutant C/EBPα and could not simply be attributed to a lack of C/EBPα, as they were not replicated in samples from patients with core-binding factor leukemias that are associated with downregulation of CEBPA expression. Presence of at least a minimal level of C/EBPα activity is thought to be necessary for the development of leukemia as Cebpa-/- mice with totally absent C/EBPα accumulate immature myeloid progenitors but do not develop AML,36,37 and AML patients have not been reported with mutations leading to complete absence of C/EBPα. These results, therefore, suggest that it is the functionally aberrant C/EBPα protein that underlies the hypermethylated profile detected in the CEBPADM cases. The CEBPAClassic-DM cases provided a framework for assessing the methylation profiles of mutant combinations with at least one non-classic mutant. Only 31% had a methylation profile equivalent to CEBPAClassic-DM, 25% were equivalent to CEBPAWT, and 44% were intermediate between the two. Similar heterogeneity has been reported for gene expression profiles of non-classic cases, with 3 of 7 such cases segregating from cases with classical mutants.20 This variability is not surprising considering the diversity of the mutant combinations. Although the p30 isoform is thought to play an important role in allowing commitment of the leukemic stem cell to the myeloid lineage,38 the mechanism by which it promotes AML is not 98

Figure 5. Methylation levels in other good-risk groups compared to the CEBPAClassic-DM cases. A composite methylation score was calculated by summing the difference between the % methylation for samples and the median for the 24 CEBPAClassic-DM cases (excluding the outlier) for three differentially methylated CpG sites: KHNYN, VAMP5 and LY9. Mean values±95% confidence intervals are shown. The CEBPAClassic-DM and CEBPAWT results were β values from the arrays; results for the three comparative groups were obtained by pyrosequencing. Significance refers to difference from the CEBPAClassic-DM group. ***P≤0.001.

clearly defined. It has a lower affinity for some C/EBP sites than p42 and induces multiple genes that are not affected by p42,39-41 and this may have influenced the methylation profile. Knock/in mice expressing just N-terminal mutant developed leukemia but more slowly than the N+C combination.38,42 This presumably reflects the additional influence of an aberrant C-terminally mutated protein that might not bind to DNA but can still bind to other C/EBP interacting proteins, such as PU.1 and the SWI/SNF comhaematologica | 2018; 103(1)


Variable outcome and methylation in CEBPADM AML

plex. Classic C-terminal mutants are associated with hyperproliferation due to loss of cell-cycle regulation and a block in myeloid differentiation.43 Although knock/in mice with classic C mutant alone do develop leukemia, it is with slower latency than the N+C and mutant N mice.42 Since these mutants may still bind and potentially sequester other interacting factors, it has been suggested that this could limit the ability of other C/EBPs to rescue the effect of the aberrant C/EBPα,2 as shown for C/EBPβ in the C/EBPα-deficient situation.44,45 This more global cellular impact of the C-terminal mutants may have a greater consequence for signaling events downstream of C/EBPα, which may, therefore, be reflected in the methylation profile. The methylation profiles did not group according to the predicted functional consequence of the mutant, whether N- or C-terminally mutated. For example, considerable variability was observed for the cases with a C-missense mutation. Of the three classic N plus C-missense combinations assayed, 1 case had a methylation profile equivalent to CEBPAWT and 2 cases had intermediate profiles. Two homozygous C-missense cases grouped with CEBPAWT cases, whereas another homozygous C-missense case grouped with the CEBPAClassic-DM cases. In the 51 cases documented with C-missense mutations in the COSMIC database (http://cancer.sanger.ac.uk/cosmic), most mutations are unique, reported in 1 (n=27, 53%) or 2 (n=9, 18%) patients, and only two residues (R297 and R300) are recorded as being variably mutated in 5 patients. Critical amino acids at the bZIP/DNA interface have been identified from the C/EBPα crystal structure, but many additional hydrogen bonds and van der Waals contacts are implicated in the stabilization of these interactions.46 Thus, predicting the functional consequence of these mutants is difficult. From a clinical perspective, risk management requires evaluation of all the available information, and the data presented here suggest that CEBPADM patients with a nonclassic mutation should not automatically be included in the same favorable-risk prognostic group as CEBPAClassic-DM cases; it might be appropriate to consider them for allogeneic transplantation in first remission. Ultimately, this can only be proven by analysis of clinical trial outcomes and, with the increasing availability of large data sets using targeted next-generation sequencing panels, such analysis may be feasible in the future. This will also promote a better understanding of the mutational background of classic and non-classic CEBPADM cases and whether there are differences that impact on their prognosis. Our studies also raise the possibility that methylation profiling may identify those non-classic cases that behave in a similar manner to classic mutants; although we can-

References 1. Friedman AD. C/EBPalpha in normal and malignant myelopoiesis. Int J Hematol. 2015;101(4):330-341. 2. Ohlsson E, Schuster MB, Hasemann M, Porse BT. The multifaceted functions of C/EBPalpha in normal and malignant haematopoiesis. Leukemia. 2016;30(4):767775.

haematologica | 2018; 103(1)

A

B

Figure 6. Supervised analysis of CEBPADM cases with non-classic mutations and CEBPASM cases. Difference between the distance scores (CEBPAClassic-DM - CEBPAWT) compared with the distance from the CEBPAClassic-DM (MUT) signature for (A) CEBPADM cases with a non-classic mutant. (B) CEBPASM cases.

not directly attribute a causal link between the methylation pattern and chemosensitivity, and further studies are required before this is introduced into clinical practice. Acknowledgments The authors would like to thank Kerra Pearce at UCL Genomics for technical assistance. Funding This work was supported by Leukaemia and Lymphoma Research, now called Bloodwise, the UK Medical Research Council and Cancer Research UK, and was undertaken at UCL, which receives a proportion of funding from the Department of Health’s NIHR Biomedical Research Centre’s funding scheme.

3. Wouters BJ, Lowenberg B, ErpelinckVerschueren CA, van Putten WL, Valk PJ, Delwel R. Double CEBPA mutations, but not single CEBPA mutations, define a subgroup of acute myeloid leukemia with a distinctive gene expression profile that is uniquely associated with a favorable outcome. Blood. 2009;113(13):3088-3091. 4. Dufour A, Schneider F, Metzeler KH, et al. Acute myeloid leukemia with biallelic CEBPA gene mutations and normal kary-

otype represents a distinct genetic entity associated with a favorable clinical outcome. J Clin Oncol. 2010;28(4):570-577. 5. Pabst T, Eyholzer M, Fos J, Mueller BU. Heterogeneity within AML with CEBPA mutations; only CEBPA double mutations, but not single CEBPA mutations are associated with favourable prognosis. Br J Cancer. 2009;100(8):1343-1346. 6. Green CL, Koo KK, Hills RK, Burnett AK, Linch DC, Gale RE. Prognostic significance

99


D. El-Sharkawi et al.

7.

8.

9.

10.

11. 12.

13.

14. 15.

16.

17.

18.

19.

100

of CEBPA mutations in a large cohort of younger adult patients with acute myeloid leukemia: impact of double CEBPA mutations and the interaction with FLT3 and NPM1 mutations. J Clin Oncol. 2010; 28(16):2739-2747. Taskesen E, Bullinger L, Corbacioglu A, et al. Prognostic impact, concurrent genetic mutations, and gene expression features of AML with CEBPA mutations in a cohort of 1182 cytogenetically normal AML patients: further evidence for CEBPA double mutant AML as a distinctive disease entity. Blood. 2011;117(8):2469-2475. Fasan A, Haferlach C, Alpermann T, et al. The role of different genetic subtypes of CEBPA mutated AML. Leukemia. 2014; 28(4):794-803. Li HY, Deng DH, Huang Y, et al. Favorable prognosis of biallelic CEBPA gene mutations in acute myeloid leukemia patients: a meta-analysis. Eur J Haematol. 2015; 94(5):439-448. Dohner H, Estey EH, Amadori S, et al. Diagnosis and management of acute myeloid leukemia in adults: recommendations from an international expert panel, on behalf of the European LeukemiaNet. Blood. 2010;115(3):453-474. O'Donnell MR, Abboud CN, Altman J, et al. Acute myeloid leukemia. J Natl Comp Cancer Network. 2012;10(8):984-1021. Ofran Y, Rowe JM. Genetic profiling in acute myeloid leukaemia--where are we and what is its role in patient management. Br J Haematol. 2013;160(3):303-320. Cornelissen JJ, Gratwohl A, Schlenk RF, et al. The European LeukemiaNet AML Working Party consensus statement on allogeneic HSCT for patients with AML in remission: an integrated-risk adapted approach. Nat Rev Clin Oncol. 2012; 9(10):579-590. Nerlov C. C/EBPalpha mutations in acute myeloid leukaemias. Nat Rev Cancer. 2004;4(5):394-400. Fitzgibbon J, Smith LL, Raghavan M, et al. Association between acquired uniparental disomy and homozygous gene mutation in acute myeloid leukemias. Cancer Res. 2005;65(20):9152-9154. Wouters BJ, Sanders MA, Lugthart S, et al. Segmental uniparental disomy as a recurrent mechanism for homozygous CEBPA mutations in acute myeloid leukemia. Leukemia. 2007;21(11):2382-2384. Shih LY, Liang DC, Huang CF, et al. AML patients with CEBPalpha mutations mostly retain identical mutant patterns but frequently change in allelic distribution at relapse: a comparative analysis on paired diagnosis and relapse samples. Leukemia. 2006;20(4):604-609. Benthaus T, Schneider F, Mellert G, et al. Rapid and sensitive screening for CEBPA mutations in acute myeloid leukaemia. Br J Haematol. 2008;143(2):230-239. Behdad A, Weigelin HC, Elenitoba-Johnson KS, Betz BL. A clinical grade sequencingbased assay for CEBPA mutation testing: report of a large series of myeloid neo-

plasms. J Mol Diag. 2015;17(1):76-84. 20. Lavallee VP, Krosl J, Lemieux S, et al. Chemo-genomic interrogation of CEBPA mutated AML reveals recurrent CSF3R mutations and subgroup sensitivity to JAK inhibitors. Blood. 2016;127(24):3054-3061. 21. Wouters BJ, Jorda MA, Keeshan K, et al. Distinct gene expression profiles of acute myeloid/T-lymphoid leukemia with silenced CEBPA and mutations in NOTCH1. Blood. 2007;110(10):3706-3714. 22. Figueroa ME, Lugthart S, Li Y, et al. DNA methylation signatures identify biologically distinct subtypes in acute myeloid leukemia. Cancer Cell. 2010;17(1):13-27. 23. Cancer Genome Atlas Research Network, Ley TJ, Miller C, et al. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. New Engl J Med. 2013;368(22):2059-2074. 24. Figueroa ME, Wouters BJ, Skrabanek L, et al. Genome-wide epigenetic analysis delineates a biologically distinct immature acute leukemia with myeloid/T-lymphoid features. Blood. 2009;113(12):2795-2804. 25. Sproul D, Kitchen RR, Nestor CE, et al. Tissue of origin determines cancer-associated CpG island promoter hypermethylation patterns. Genome Biol. 2012;13(10):R84. 26. Sproul D, Nestor C, Culley J, et al. Transcriptionally repressed genes become aberrantly methylated and distinguish tumors of different lineages in breast cancer. Proc Natl Acad Sci USA. 2011; 108(11):4364-4369. 27. Illingworth RS, Gruenewald-Schneider U, Webb S, et al. Orphan CpG islands identify numerous conserved promoters in the mammalian genome. PLoS Gen. 2010; 6(9):e1001134. 28. Gale RE, Lamb K, Allen C, El-Sharkawi D, Stowe C, Jenkinson S, et al. Simpson's Paradox and the Impact of Different DNMT3A Mutations on Outcome in Younger Adults With Acute Myeloid Leukemia. J Clin Oncol. 2015;33(18):20722083. 29. Pabst T, Mueller BU, Harakawa N, et al. AML1-ETO downregulates the granulocytic differentiation factor C/EBPalpha in t(8;21) myeloid leukemia. Nat Med. 2001; 7(4):444-451. 30. Ptasinska A, Assi SA, Martinez-Soria N, et al. Identification of a dynamic core transcriptional network in t(8;21) AML that regulates differentiation block and selfrenewal. Cell Rep. 2014;8(6):1974-1988. 31. Helbling D, Mueller BU, Timchenko NA, et al. CBFB-SMMHC is correlated with increased calreticulin expression and suppresses the granulocytic differentiation factor CEBPA in AML with inv(16). Blood. 2005;106(4):1369-1375. 32. Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic Classification and Prognosis in Acute Myeloid Leukemia. New Engl J Med. 2016;374(23):2209-2221. 33. Grossmann V, Haferlach C, Nadarajah N, et al. CEBPA double-mutated acute myeloid leukaemia harbours concomitant molecular mutations in 76.8% of cases with TET2

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

and GATA2 alterations impacting prognosis. Br J Haematol. 2013;161(5):649-658. Green CL, Tawana K, Hills RK, et al. GATA2 mutations in sporadic and familial acute myeloid leukaemia patients with CEBPA mutations. Br J Haematol. 2013;161(5):701-705. Theis F, Corbacioglu A, Gaidzik VI, et al. Clinical impact of GATA2 mutations in acute myeloid leukemia patients harboring CEBPA mutations: a study of the AML study group. Leukemia. 2016;30(11):2248-2250. Heath V, Suh HC, Holman M, et al. C/EBPalpha deficiency results in hyperproliferation of hematopoietic progenitor cells and disrupts macrophage development in vitro and in vivo. Blood. 2004;104(6):16391647. Zhang P, Iwasaki-Arai J, Iwasaki H, et al. Enhancement of hematopoietic stem cell repopulating capacity and self-renewal in the absence of the transcription factor C/EBP alpha. Immunity. 2004;21(6):853863. Kirstetter P, Schuster MB, Bereshchenko O, et al. Modeling of C/EBPalpha mutant acute myeloid leukemia reveals a common expression signature of committed myeloid leukemia-initiating cells. Cancer Cell. 2008;13(4):299-310. Cleaves R, Wang QF, Friedman AD. C/EBPalphap30, a myeloid leukemia oncoprotein, limits G-CSF receptor expression but not terminal granulopoiesis via siteselective inhibition of C/EBP DNA binding. Oncogene. 2004;23(3):716-725. Geletu M, Balkhi MY, Peer Zada AA, et al. Target proteins of C/EBPalphap30 in AML: C/EBPalphap30 enhances sumoylation of C/EBPalphap42 via up-regulation of Ubc9. Blood. 2007;110(9):3301-3309. Wang C, Chen X, Wang Y, Gong J, Hu G. C/EBPalphap30 plays transcriptional regulatory roles distinct from C/EBPalphap42. Cell Res. 2007;17(4):374-383. Bereshchenko O, Mancini E, Moore S, et al. Hematopoietic stem cell expansion precedes the generation of committed myeloid leukemia-initiating cells in C/EBPalpha mutant AML. Cancer Cell. 2009;16(5):390400. Kato N, Kitaura J, Doki N, et al. Two types of C/EBPalpha mutations play distinct but collaborative roles in leukemogenesis: lessons from clinical data and BMT models. Blood. 2011;117(1):221-233. Jones LC, Lin ML, Chen SS, et al. Expression of C/EBPbeta from the C/ebpalpha gene locus is sufficient for normal hematopoiesis in vivo. Blood. 2002; 99(6):2032-2036. Hirai H, Zhang P, Dayaram T, et al. C/EBPbeta is required for 'emergency' granulopoiesis. Nat Immunol. 2006; 7(7):732-739. Miller M, Shuman JD, Sebastian T, Dauter Z, Johnson PF. Structural basis for DNA recognition by the basic region leucine zipper transcription factor CCAAT/enhancerbinding protein alpha. J Biol Chem. 2003; 278(17):15178-15184.

haematologica | 2018; 103(1)


ARTICLE

Acute Myeloid Leukemia

Randomized study of continuous high-dose lenalidomide, sequential azacitidine and lenalidomide, or azacitidine in persons 65 years and over with newly-diagnosed acute myeloid leukemia

Bruno C. Medeiros,1 Kelly McCaul,2 Suman Kambhampati,3,4 Daniel A. Pollyea,5 Rajat Kumar,6 Lewis R. Silverman,7 Andrea Kew,8 Lalit Saini,9 CL Beach,10 Ravi Vij,11 Xiwei Wang,10 Jim Zhong,10 and Robert Peter Gale10,12

Stanford University School of Medicine, CA, USA; 2Avera Cancer Institute, Sioux Falls, SD, USA; 3Sarah Cannon Cancer Institute at Research Medical Center, Kansas City, MO, USA; 4Kansas University Medical Center, Kansas City, KS, USA; 5University of Colorado School of Medicine Division of Hematology, Aurora, CO, USA; 6CancerCare Manitoba, Winnipeg, MB, Canada; 7Mount Sinai Hospital, New York, NY, USA; 8Queen Elizabeth II Health Sciences Center, Halifax, NS, Canada; 9University of Alberta Hospital (Adult Hematology Research), Edmonton, AB, Canada; 10Celgene Corporation, Summit, NJ, USA; 11Section of Bone Marrow Transplant and Leukemia, Washington University School of Medicine, St Louis, MO, USA and 12Imperial College London, UK 1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(1):101-106

ABSTRACT

T

herapy of acute myeloid leukemia in older persons is associated with poor outcomes because of intolerance to intensive therapy, resistant disease and co-morbidities. This multi-center, randomized, open-label, phase II trial compared safety and efficacy of three therapeutic strategies in patients 65 years or over with newly-diagnosed acute myeloid leukemia: 1) continuous high-dose lenalidomide (n=15); 2) sequential azacitidine and lenalidomide (n=39); and 3) azacitidine only (n=34). The efficacy end point was 1-year survival. Median age was 76 years (range 66-87 years). Thirteen subjects (15%) had prior myelodysplastic syndrome and 41 (47%) had adverse cytogenetics. One-year survival was 21% [95% confidence interval (CI): 0, 43%] with high-dose lenalidomide, 44% (95%CI: 28, 60%) with sequential azacitidine and lenalidomide, and 52% (95%CI: 35, 70%) with azacitidine only. Lenalidomide at a continuous high-dose schedule was poorly-tolerated resulting in a high rate of early therapy discontinuations. Hazard of death in the first four months was greatest in subjects receiving continuous high-dose lenalidomide; hazards of death thereafter were similar. These data do not favor use of continuous high-dose lenalidomide or sequential azacitidine and lenalidomide over the conventional dose and schedule of azacitidine only in patients aged 65 years or over with newly-diagnosed acute myeloid leukemia. (clinicaltrials.gov identifier: 01358734). Introduction The incidence of acute myeloid leukemia (AML) increases dramatically with age. Survival of older people with AML is poor;1-3 in the US, fewer than 50% of subjects with AML over 65 years of age receive therapy within three months of diagnosis despite considerable data indicating therapy improves their survival.4-6 Reasons for not treating older patients with AML include poor performance score, co-morbidities, frailty, an antecedent hematologic disorder (AHD), presumed therapy-related leukemia, and/or adverse risk biological features.4,7 One study of more than 3000 subjects reported a significant adverse impact of increasing age on outcomes even after adjusting for these variables, suggesting age is an independent prognostic variable associated with poor outcomes.8 Therapy recommendations for older subjects with AML vary with no standardof-care.7 Commonly-used treatments include intensive therapy (typically cytarabine and an anthracycline) and less intensive approaches such as low-dose cytarabine or DNA hypo-methylating drugs including azacitidine and decitabine.9 haematologica | 2018; 103(1)

Correspondence: brunom@stanford.edu

Received: May 14, 2017. Accepted: October 26, 2017. Pre-published: November 2, 2017. doi:10.3324/haematol.2017.172353 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/101 Š2018 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.

101


B.C. Medeiros et al.

Azacitidine was recently approved in Europe in subjects 65 years and over with AML and more than 30% bone marrow blasts judged unfit for a hematopoietic cell transplant.8 Approval was based on a 4-month improvement in median survival compared with conventional therapy or supportive care [10.4 vs. 6.5 months; Hazard Ratio (HR) 0.85, 95% confidence interval (95%CI): 0.69, 1.03]. Considerable data indicate high-dose lenalidomide, an immune-modulatory drug (IMiD®), may be effective in patients with myelodysplastic syndrome (MDS) with excess blasts and patients with AML judged unfit for intensive therapy.10,11 Phase I and II studies report concomitant or sequential azacitidine and lenalidomide is effective in MDS, AML and in chronic myelomonocytic leukemia (CMML).12-15 Several studies of lenalidomide used 50 mg/day continuously without dose modification. Based on these data, we designed a randomized phase II study comparing safety and efficacy of continuous high-dose lenalidomide, sequential azacitidine and lenalidomide or azacitidine only in persons 65 years or over with newly-diagnosed AML.

Methods We conducted a phase II randomized, open-label, parallel-group study at 25 sites in North America (clinicaltrials.gov identifier: 01358734). The study was approved by the relevant institutional review boards or an independent ethics committee, and followed guidelines set out in the Declaration of Helsinki. Subjects gave written informed consent. Statistical analyses were carried out by Celgene Corporation, Summit, NJ, USA, and were reviewed by the authors with access to all study data.

Subjects Eligible subjects were 65 years and over with newly-diagnosed AML including those with de novo AML, AML with prior MDS or with presumed therapy-related AML. Subjects could not have previously received lenalidomide, azacitidine, decitabine or cytarabine and were judged ineligible to receive a transplant at study entry. Subjects had to have an Eastern Co-operative Oncology Group (ECOG) performance score ≤2, white blood cell (WBC) count ≤10x109/L (hydroxyurea use was permitted to lower the WBC count), adequate liver and kidney function, and no uncontrolled infection or other cancer within two years of study entry. Subjects who had received prior therapy for AML other than hydroxyurea were not eligible for the study.

Study design Screening procedures were completed no more than 28 days before randomization. Confirmation of AML for study entry was based on local pathology review. A separate central review of all bone marrow aspirates, bone marrow biopsies and blood smears was conducted by a blinded pathologist (Dr. Daniel Arber, Stanford University). Cytogenetic testing on diagnostic samples was carried out locally and reviewed centrally (Prof. Athena Cherry, Stanford University). Subjects were randomized equally to the therapy arms by central interactive voice response system (IVRS). Randomization was stratified for performance score (0-1 vs. 2) and levels of blood myeloblasts (<1 vs. ≥1x109/L).

Therapy Therapy regimens are shown in Figure 1. Therapy in any arm could be interrupted or delayed because of hematologic or nonhematologic therapy-related adverse events (AEs). Lenalidomide dose-reductions were not allowed in cycles 1-4 in the continuous high-dose lenalidomide cohort or in courses 1 and 2 in the azacitidine and lenalidomide cohort. Dose reductions of lenalidomide

Figure 1. Study design and therapy regimens. Stratification factors were Eastern Co-operative Oncology Group (ECOG) performance score (0-1 vs. 2) and blood blast level (<1 vs. ≥1x109/L). Randomization in the lenalidomide regime was suspended 11th September 2013 and permanently closed on 15th April 2014 as per Amendment 2 because of the high rate of discontinuation of the study treatment in subjects receiving high-dose continuous lenalidomide. AML: acute myeloid leukemia; PO: per oral administration; SC: subcutaneous injection; SPM: secondary primary malignancies.

102

haematologica | 2018; 103(1)


High-dose LEN and AZA in newly-diagnosed AML

could be made thereafter in both cohorts. Dose reductions of azacitidine were permitted as described. Dose reductions of lenalidomide or azacitidine were continued in subsequent cycles unless additional adverse events occurred; in such cases, additional dose reductions were permitted. Subjects were seen weekly for the first 2 courses, every two weeks for the next 2 courses, and on the first day of subsequent courses. Microscope slides of blood and bone marrow samples and cytogenetic testing were obtained pre-randomization, on day 15 of course 1, within seven days of beginning courses 2 and 3, and before starting every third subsequent course.

Safety and efficacy end points Safety assessments included vital signs, physical exam, weight, ECOG performance score, complete blood count (CBC) and differential, blood chemistries, urinalyses, coagulation, thyroid function tests and electrocardiograms. AEs were recorded using Common Terminology Criteria for Adverse Events (CTCAE) v.4.0. Study subjects were kept under surveillance for the development of new cancers. Events of interest (EOIs) were recorded and reported as serious AEs (SAEs) and considered medically-important even if not meeting SAE criteria. The primary end point was 1-year survival. Secondary end points included rates of complete remission (CR), complete remission with incomplete recovery of neutrophils or platelets (CRi), and remission duration. Subjects discontinuing the study could receive subsequent therapy during the follow-up interval at their physicianâ&#x20AC;&#x2122;s discretion and with their consent.

Statistical analysis The study was not designed or powered for a formal statistical comparison of the therapy cohorts and a sample size calculation was not made. Target enrollment was 40 subjects per cohort.

Table 1. Baseline variables for the intent-to-treat populations.

Lenalidomide Azacitidine+ Azacitidine (n=15) lenalidomide (n=34) (n=39) Age (years) Median (range) >75 years (n) Male (n) ECOG performance score (n) 0-1 2 Prior MDS (n) Adverse cytogenetics (n) Blood blasts >1x10E9/L (n) Bone marrow blasts (%) Median (range) Interval from diagnosis to treatment (days) Median (range) Platelets (x109/L) Median (range) Neutrophils (x109/L) Median (range)

80 (68-85) 10 12

76 (66-87) 21 22

75 (66-85) 14 19

13 2 3 4 11

32 7 5 20 31

27 6 5 17 27

56 (22-95)

37 (12-84)

34 (14-70)

15 (7-42)

18 (4-41)

16 (6-50)

54 (8-162)

72 (13-204)

56 (8-211)

0.2 (0-1.6)

0.4 (0-3)

0.5 (0-3)

n: number; ECOG: Eastern Co-operative Oncology Group; MDS: myelodysplastic syndromes; d: days.

haematologica | 2018; 103(1)

Preliminary analyses indicated a high rate of discontinuations in the continuous high-dose lenalidomide cohort with 10 of the first 13 subjects receiving less than 56 days of therapy. Based on these data, the Data Safety Monitoring Committee (DSMC) recommended closing randomization into this cohort and the planned study sample size was reduced to 95 subjects. Time-to-death from any cause was defined as the interval from randomization to death. Living subjects were censored at date of last contact, withdrawal of consent, loss to follow up, or study completion. There was no pre-specified statistical plan to compare survival between the cohorts. However, an exploratory analysis was made comparing 1-year survival Kaplan-Meier estimates between the cohorts using the log rank test and which indicated non-proportional hazards of death. Consequently, we estimated HR and 95% confidence intervals for the intervals 0-4 months and >4-12 months using a Cox proportional hazards model. P-values were derived from the Cox model.

Results Subjects Enrollment began on 27th April 2012 with database lock on 1st May 2015. Because of poor accrual, only 88 subjects were randomized, all of whom constitute the intent-totreat (ITT) population including 15 subjects randomized to receive high-dose continuous lenalidomide, 39 randomized to receive sequential azacitidine and lenalidomide, and 34 randomized to receive azacitidine only. Subjects in the high-dose lenalidomide cohort were more likely to be over 75 years of age and to have had prior MDS, had a higher percent of bone marrow blasts, and lower levels of blood neutrophils. This cohort also had fewer subjects with adverse cytogenetics than the other cohorts (Table 1). There were 6 subjects with del(5/5q), including one each in the high-dose lenalidomide and in the azacitidine only cohorts, and 4 in the sequential azacitidine and lenalidomide cohort.

Efficacy Median therapy duration was six weeks (range 1-48 weeks) for the continuous high-dose lenalidomide cohort, eight weeks (0-78 weeks) for the sequential azacitidine and lenalidomide cohort, and 29 weeks (1-143 weeks) for

Table 2. Hazard Ratios for death. 0-4 months Lenalidomide vs. azacitidine Azacitidine + lenalidomide vs. azacitidine Lenalidomide vs. azacitidine + lenalidomide >4-12 months Lenalidomide vs. azacitidine Azacitidine + lenalidomide vs. azacitidine Lenalidomide vs. azacitidine + lenalidomide

Hazard Ratio (95% CI)

P

5.73 (1.90, 17.20) 2.51 (0.89, 7.05)

0.002 0.081

2.19 (0.94, 5.13)

0.071

0.853 (0.187, 3.90) 0.790 (0.312, 2.00)

0.838 0.620

1.02 (0.22, 4.80)

0.982

CI: Confidence Interval; P: P-value; vs.: versus.

103


B.C. Medeiros et al.

Figure 2. CONSORT study-flow diagram. *Four randomized subjects were not treated because of withdrawal of consent, adverse events, or other reasons, e.g. hospitalization. Percents are based on the intention-to-treat population.

the azacitidine only cohort. The most common reason for discontinuation was leukemia progression in 5, 11, and 13 subjects (Figure 2). One-year survivals were 21% (95%CI: 0, 43%), 44% (95%CI: 28, 60%) and 52% (95%CI: 35, 70%) (Figure 3) in the lenalidomide, azacitidine and lenalidomide and azacitidine only cohorts. Subjects in the high-dose continuous lenalidomide cohort had a higher hazard of death than subjects in the azacitidine cohort [HR 5.73 (1.91, 17.20); P=0.002] or subjects in the sequential azacitidine and lenalidomide cohort [HR 2.19 (95%CI: 0.94, 5.13); P=0.071] (Table 2) in the first four months post randomization. Hazard of death was also higher in the sequential azacitidine and lenalidomide cohort compared with the azacitidine cohort [HR 2.51 (0.89, 7.05); P=0.081]. Hazards of death >4-12 months were similar in the 3 cohorts. There were 7 deaths in the continuous high-dose lenalidomide cohort in the first four months. Median duration of therapy of these subjects was 36 days (range 7-84 days). Two subjects discontinued therapy because of an adverse event and 5 because of AML progression. Causes of death were AML and/or its complications such as infection. One subject had a pulmonary infarct. Rates of CR and CRi for the cohorts are shown in Table 3. 104

Table 3. Proportion of subjects with complete remission with (CR) and without (CRi) complete hematologic recovery. Lenalidomide (n=15) Azacitidine + lenalidomide (n=39) Azacitidine (n=34)

CR

CRi

CR/CRi

2 11 6

3 4 8

5 15 14

n: number.

Interestingly, 2 subjects receiving high-dose lenalidomide only, a non-cytotoxic drug, achieved a complete remission and 3, a complete remission with incomplete recovery of blood levels of neutrophils and/or platelets. There were too few subjects with del(5/5q) in each cohort to compare outcomes of this subset.

Adverse events Eighty-four subjects received 1 dose or more of study drug and constitute the safety population including 14 subjects receiving high-dose continuous lenalidomide, 38 receiving sequential azacitidine and lenalidomide, and 32 receiving azacitidine only. Treatment was discontinued within the first 2 cycles in 10 subjects in the high-dose continuous lenalidomide, 19 in the sequential azacitidine and lenalidomide cohort, and 8 in the azacitidine only haematologica | 2018; 103(1)


High-dose LEN and AZA in newly-diagnosed AML

cohort. The most common reasons for discontinuation of study therapy within 2 courses across the 3 cohorts were adverse events (3, 6 and 2, respectively), death (2, 6 and 1, respectively) and disease progression (3, 2 and 2, respectively). During the study the proportion of treatmentemergent adverse events (TEAEs) resulting in therapy discontinuation was highest in the high-dose continuous lenalidomide cohort (n=4) followed by the azacitidine and lenalidomide cohort (n=7), and lowest in the azacitidine only cohort (n=3). Infections were the most frequent cause of permanent therapy discontinuation and were most frequent in the high-dose continuous lenalidomide cohort (n=4 compared with 3 subjects in the azacitidine and lenalidomide cohort and 1 subject in the azacitidine only cohort). There were 3 new cancers (vulvar cancer, stage 0; lung adenocarcinoma; and central nervous system cancer), all in the azacitidine only cohort. These cancers were diagnosed at approximately, 6, 9 and less than 1 month post randomization. The most frequent TEAEs (any grade) considered drugrelated were hematologic and gastrointestinal. The most frequent TEAE grade 3 or over was febrile neutropenia (n=6, 17 and 9, respectively) (Table 4). Frequencies of treatment-emergent serious AEs (TE-SAEs) was higher in the high-dose continuous lenalidomide cohort (n=10) than in the other cohorts (n=16 and 7, respectively). Pneumonia was the most frequent infectious TE-SAE reported in 1, 2 and 0 subjects, respectively. The most common causes of death on therapy were AML progression and infections. Deaths from AML progression were reported in 8, 12 and 11 subjects and deaths from infections were reported in 2, 5 and 2 subjects, respectively. Median hospital days for the 3 cohorts were ten days (range 2-54 days), 11 days (range 2-43 days), and seven days (range 2-24 days).

Discussion Our randomized phase II study in patients over 65 years with newly-diagnosed AML was designed to numerically compare 1-year survivals between the cohorts. It was not

designed nor powered to make statistical comparisons in outcomes. The sequential azacitidine and lenalidomide and azacitidine only regimens were reasonably well-tolerated. Proportions of subjects alive at one year in these cohorts were 52% (35, 70%) and 44% (28, 60%), similar to azacitidine only in a large phase III trial in similar patients.16 However, the HR for death of 2.51 (0.89, 7.05) in the lenalidomide plus azacitidine cohort in months 0-4 suggests azacitidine only may be the best regimen. Lenalidomide at the dose and schedule used in the highdose continuous lenalidomide cohort was less well-tolerated, perhaps because no dose reduction was permitted, resulting in a high proportion of early treatment discontinuation and in subsequent leukemia progression. Subjects Table 4. Most common (≥3 subjects) grade ≥3 treatment-emergent adverse events.

Preferred term ≥1 CTCAE ≥grade-3 TEAE Febrile neutropenia Thrombocytopenia Anemia Neutropenia Pneumonia Cellulitis Neutrophil count decreased WBC decreased Platelet count decreased Dyspnea Fatigue Syncope Atrial fibrillation

Lenalidomide Azacitidine + Azacitidine (n=14) lenalidomide (n=32) (n=38) 14 6 4 4 3 3 0 0 0 0 2 4 3 0

34 17 12 10 5 2 5 5 4 6 4 6 1 4

29 9 11 8 9 8 4 4 5 2 1 5 3 0

CTCAE: Common Terminology Criteria for Adverse Events; TEAE: treatment-emergent adverse event; WBC: white blood cell count. Subjects with multiple occurrences of the same TEAE were counted only once and assigned the highest grade reported. Data cut-off date: 1st May 2015. Preferred terms are according to MedDRA dictionary, v. 17.0 and listed in descending order of overall frequency.

Figure 3. Kaplan-Meier estimates of 1year survival. Subjects in the high-dose continuous lenalidomide cohort had a higher hazard of death in 0-4 months than subjects in the azacitidine alone cohort (P=0.002) and subjects in the sequential azacitidine and lenalidomide cohort (P=0.071).

haematologica | 2018; 103(1)

105


B.C. Medeiros et al.

enrolled in this cohort had an increased hazard of death in the first four months of therapy compared with the other cohorts. Hazards of death thereafter were similar but favor the azacitidine cohort. This early difference in hazard of death should be viewed cautiously as these analyses were not specified in the pre-study statistical plan and sample sizes are small. Proportions of subjects achieving a CR/CRi were similar with wide 95% confidence intervals. It is also important to emphasize that the small sample sizes may have resulted in an imbalance in baseline variables. For example, subjects in the high-dose continuous lenalidomide cohort were older and more likely to have prior MDS and severe bone marrow failure but less frequent adverse cytogenetics. This cohort also had a greater proportion of subjects with cardio-vascular disease risk factors (data not shown). None of these differences were tested for statistical significance according to the prestudy statistical plan. Finally, discontinuation rates were generally similar to those in other clinical studies of continuous high-dose lenalidomide in older persons with AML.14 Randomization with or without stratification does not guarantee cohorts have similar distributions of known and latent prognostic variables, especially when there are few subjects in each cohort. Typically, potential imbalances are accounted for by calculating P-values and indicating confidence intervals. Because we did not perform these calculations for outcomes other than for 0-4 and >412-month survival intervals between the cohorts, differences in outcomes should not be assumed to result from therapy assignment. Sequential azacitidine and lenalidomide did not increase the frequency of >grade 3 or over TEAEs compared with continuous high-dose lenalidomide. Subjects in the continuous high-dose lenalidomide cohort were scheduled to receive 50 mg/day for 56 days whereas those in the sequential azacitidine and lenalidomide cohort were scheduled to receive the same dose of lenalidomide for 21 days followed by 14 days off therapy. Consequently, there

References 1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin. 2013;63(1):11-30. 2. Appelbaum FR, Gundacker H, Head DR, et al. Age and acute myeloid leukemia. Blood. 2006;107(9):3481-3485. 3. Buchner T, Krug UO, Gale RP, et al. Age, not therapy intensity, determines outcomes of adults with acute myeloid leukemia. Leukemia. 2016;30(8):1781-1784. 4. Medeiros BC, Satram-Hoang S, Hurst D, Hoang KQ, Momin F, Reyes C. Big data analysis of treatment patterns and outcomes among elderly acute myeloid leukemia patients in the United States. Ann Hematol. 2015;94(7):1127-1138. 5. Juliusson G. Older patients with acute myeloid leukemia benefit from intensive chemotherapy: an update from the Swedish Acute Leukemia Registry. Clin Lymphoma Myeloma Leuk. 2011;11 (Suppl 1):S54-S59. 6. Juliusson G. Most 70- to 79-year-old patients with acute myeloid leukemia do benefit from intensive treatment. Blood. 2011;117(12):3473-3474.

106

are significant differences in schedule and total dose of lenalidomide between the cohorts. Neither regimen with high-dose lenalidomide, continuous or sequential, was as well-tolerated as azacitidine only. In conclusion, data from unplanned survival analysis indicate high-dose continuous lenalidomide given without dose reduction resulted in a high rate of early discontinuations and an early but not late increased hazard of death compared with sequential azacitidine and lenalidomide and with azacitidine only in subjects 65 years and over with newly-diagnosed AML. Our data also suggest therapy with azacitidine only may be better than sequential lenalidomide and azacitidine based on the 95% confidence interval of the HR for death. Although continuous high-dose lenalidomide is active in AML and sometimes produces prolonged complete remissions,10 we cannot recommend the dose and schedule we tested for subjects 65 years and over with newly-diagnosed AML. Adding lenalidomide to azacitidine did not improve 1-year survival and may have reduced it in the first four months of therapy. However, our study was neither designed nor powered to test this question. Whether a different dose, schedule and/or dose-modification of lenalidomide alone or with azacitidine might be better tolerated and/or more effective than the dose and schedule we studied is unknown. However, the achievement of complete remission with or without complete recovery of blood levels of neutrophils and/or platelets by a non-cytotoxic drug is noteworthy. Acknowledgments Medical Communication Company (Wynnewood, PA, USA) provided editorial support in the form of arranging references for an advanced draft which was subsequently approved by all authors. Funding The study was funded by Celgene Corporation.

7. National Comprehensive Cancer Network. Clinical Practice Guidelines in Oncology. Acute Myeloid Leukemia, version 1.2017, Available at: http://www.nccn.org/professionals/physician_gls/f_guidelines.asp. 8. Krug U, Berdel WE, Gale RP, et al. Increasing intensity of therapies assigned at diagnosis does not improve survival of adults with acute myeloid leukemia. Leukemia. 2016;30(6):1230-1236. 9. National Comprehensive Cancer Network (NCCN) Clinical Practice Guidelines in Oncology. Acute Myeloid Leukemia. Version 2.2014, Available at: http://www.nccn.org/professionals/physician_gls/f_guidelines.asp; 2014 [Last accessed 5th August 2014] 10. Fehniger TA, Uy GL, Trinkaus K, et al. A phase 2 study of high-dose lenalidomide as initial therapy for older patients with acute myeloid leukemia. Blood. 2011; 117(6):1828-1833. 11. Pollyea DA, Zehnder J, Coutre S, et al. Sequential azacitidine plus lenalidomide combination for elderly patients with untreated acute myeloid leukemia. Haematologica. 2013;98(4):591-596. 12. Ramsingh G, Westervelt P, Cashen AF, et al.

A phase 1 study of concomitant high-dose lenalidomide and 5-azacitidine induction in the treatment of AML. Leukemia. 2013;27(3):725-728. 13. Wei A, Tan P, Perruzza S, et al. Maintenance lenalidomide in combination with 5-azacitidine as post-remission therapy for acute myeloid leukaemia. Br J Haematol. 2015; 169(2):199-210. 14. Sekeres MA, Gundacker H, Lancet J, et al. A phase 2 study of lenalidomide monotherapy in patients with deletion 5q acute myeloid leukemia: Southwest Oncology Group Study S0605. Blood. 2011; 118(3):523-528. 15. Sekeres M, Othus M, List AF, et al. Randomized phase II study of azacitidine alone or in combination with lenalidomide or with vorinostat in higher-risk myelodysplastic syndromes and chronic myelomonocytic leukemia: North American Intergroup Study SWOG S1117. J Clin Oncol. 2017;35(24):2745-2753. 16. 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.

haematologica | 2018; 103(1)


ARTICLE

Acute Lymphoblastic Leukemia

Predictive value of minimal residual disease in Philadelphia-chromosome-positive acute lymphoblastic leukemia treated with imatinib in the European intergroup study of post-induction treatment of Philadelphia-chromosome-positive acute lymphoblastic leukemia, based on immunoglobulin/T-cell receptor and BCR/ABL1 methodologies Giovanni Cazzaniga,1 Paola De Lorenzo,1,2 Julia Alten,3 Silja Röttgers,3 Jeremy Hancock,4 Vaskar Saha,4 Anders Castor,5 Hans O. Madsen,5 Virginie Gandemer,6 Hélène Cavé,6 Veronica Leoni,7 Rolf Köhler,3 Giulia M.Ferrari,7 Kirsten Bleckmann,3 Rob Pieters,8 Vincent van der Velden,8 Jan Stary,9 Jan Zuna,9 Gabriele Escherich,10 Udo zur Stadt,10 Maurizio Aricò,11 Valentino Conter,7 Martin Schrappe,3 Maria Grazia Valsecchi*2 and Andrea Biondi*1,7,

Ferrata Storti Foundation

Haematologica 2018 Volume 103(1):107-115

Associazione Italiana Ematologia Oncologia Pediatrica (AIEOP), Centro Ricerca Tettamanti, Pediatric Department, University of Milano-Bicocca, Monza, Italy; 2European intergroup study of post-induction treatment of Philadelphia-chromosome-positive acute lymphoblastic leukemia (EsPhALL) Trial Data Center, School of Medicine and Surgery, University of Milano-Bicocca, Monza, Italy; 3Berlin-Frankfurt-Münster Group Germany (BFMG), Germany and Switzerland; 4Children's Cancer and Leukaemia Group (CCLG), UK; 5 Nordic Society of Paediatric Haematology and Oncology (NOPHO), Sweden, Denmark, Norway, Finland and Iceland; 6French Acute Lymphoblastic Leukemia Study Groups (French Acute Lymphoblastic Leukemia Study Group, FRALLE and European Organisation for Research and Treatment of Cancer, EORTC); 7Associazione Italiana Ematologia Oncologia Pediatrica (AIEOP), Pediatric Department, University of Milano-Bicocca, Fondazione MBBM, Monza, Italy; 8Dutch Childhood Oncology Group (DCOG), the Netherlands; -9Czech Pediatric Hematology Working Group (CPH), Czech Republic; 10Cooperative study group for treatment of ALL (COALL), Germany and 11Associazione Italiana Ematologia Oncologia Pediatrica (AIEOP), Azienda Sanitaria Provinciale, Ragusa, Italy 1

*

MGV and AB contributed equally to this work

Correspondence: ABSTRACT

T

he prognostic value of minimal residual disease (MRD) in Philadelphia-chromosome-positive (Ph+) childhood acute lymphoblastic leukemia (ALL) treated with tyrosine kinase inhibitors is not fully established. We detected MRD by real-time quantitative polymerase chain reaction (RQ-PCR) of rearranged immunoglobulin/T-cell receptor genes (IG/TR) and/or BCR/ABL1 fusion transcript to investigate its predictive value in patients receiving Berlin-Frankfurt-Münster (BFM) high-risk (HR) therapy and post-induction intermittent imatinib (the European intergroup study of post-induction treatment of Philadelphiachromosome-positive acute lymphoblastic leukemia (EsPhALL) study). MRD was monitored after induction (time point (TP)1), consolidation Phase IB (TP2), HR Blocks, reinductions, and at the end of therapy. MRD negativity progressively increased over time, both by IG/TR and BCR/ABL1. Of 90 patients with IG/TR MRD at TP1, nine were negative and none relapsed, while 11 with MRD<5x10-4 and 70 with MRD≥5x104 had a comparable 5-year cumulative incidence of relapse of 36.4 (15.4) and 35.2 (5.9), respectively. Patients who achieved MRD negativity at TP2 had a low relapse risk (5-yr cumulative incidence of relapse (CIR)=14.3[9.8]), whereas those who attained MRD negativity at a later date showed higher CIR, comparable to patients with positive MRD at any level. BCR/ABL1 MRD negative patients at TP1 had a relapse risk similar to those who were IG/TR MRD negative (1/8 relapses). The overall concordance between the two methods is 69%, with significantly higher positivity by BCR/ABL1. In conclusion, MRD monitoring by both methods may be functional not only for measuring response but also for guiding biological studies aimed at investigating causes for discrepancies, although from our data IG/TR MRD monitoring appears to be more reliable. Early MRD negativity is highly predictive of favorable outcome. The earlier MRD negativity is achieved, the better the prognosis. haematologica | 2018; 103(1)

gianni.cazzaniga@hsgerardo.org

Received: July 19, 2017. Accepted: October 25, 2017. Pre-published: October 27, 2017. doi:10.3324/haematol.2017.176917 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/107 ©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.

107


G. Cazzaniga et al.

Introduction The t(9;22)(q34;q11) translocation resulting in the Philadelphia chromosome (Ph) occurs in about 3% of children with ALL.1,2 In the past, this translocation was consistently associated with poor outcome, with a 5-year event-free survival (EFS) of 40%, despite intensive chemotherapy regimens and allogeneic hematopoietic stem cells transplantation (HSCT).3,4 The introduction of tyrosine kinase inhibitors (TKI) has markedly improved outcome, but relapse remains the main cause of treatment failure.5-8 Several studies have shown that detection of MRD by IG/TR somatic rearrangements is a strong and independent prognostic factor in all subgroups of childhood ALL, including Ph+ ALL treated with conventional chemotherapy.9-11 In this context, whether BCR/ABL1 could be a more appropriate MRD marker for pediatric Ph+ ALL is still a matter of debate. Moreover, data on the predictive value of early MRD response in Ph+ ALL treated with TKIs is limited or inconclusive.5-8 Therefore, it remains relevant to compare MRD based on a clonospecific marker versus the oncogenic marker (BCR-ABL1) in patients treated with TKIs. In the intergroup EsPhALL study, imatinib was started after the first induction phase, which lasted from five to seven weeks, depending on national frontline protocols, and administered intermittently thereafter until the beginning of the maintenance phase. Most patients, however, underwent HSCT before reinduction therapy.8 Herein, we report the results of molecular MRD monitoring based on IG/TR and/or BCR/ABL1 transcript as PCR markers and their predictive value in patients treated with imatinib in the EsPhALL study.

Methods Study population Between January 2004 and December 2009, 160 Ph+ ALL patients were enrolled into the EsPhALL study (EudraCT 2004001647-30 and clinicaltrials.gov Identifier: 00287105) in the centers of ten national study groups, which obtained approval from their Institutional Review Boards. Written informed consent to participate in the study was provided for all patients by parents or legal guardians. Treatment details and outcome have already been published.7 As described, patients defined as good-risk (GR; good response to prednisone at day eight, or ≤25% bone marrow (BM) blast cells at day 15, or ≤5% BM blast cells at day 21, and in complete remission at the end of induction) were randomized to receive imatinib or not, whereas those defined as poor-risk (PR) were given imatinib intermittently during intensive therapeutic phases, for a total of 18 weeks for patients treated with chemotherapy only. The duration of treatment with imatinib was shorter (usually ten weeks) in the 80% of patients who underwent HSCT, but some patients received imatinib post-HSCT as per institutional policies. Overall, 128 patients received imatinib. An ancillary study was conducted to assess the predictive value of MRD detection by RQ-PCR analysis of both IG/TR genes and/or by BCR/ABL1 transcript at each BM aspiration, i.e., at the beginning of each treatment phase and before HSCT, as shown in Figure 1A. No clinical decision was based on MRD. Here follows our report on the MRD results in patients who received imatinib. MRD at TP1 (day 33 of treatment) was available for 90 and 61 patients for IG/TR and BCR/ABL1, respectively. Among the 90 patients with IG/TR MRD data at TP1, 51 (56.7%) of them were 108

PR, representing 72.9% of PR patients treated with imatinib (51/70). No difference in outcome by MRD availability at TP1 was observed.7 In the report herein, subsequent time points, namely TP2 (day 78 of treatment), TP3 (day 120) and TP4 (day 141) were included only if patients had MRD evaluations at all previous time points (Figure 1B,C).

MRD methodology In each national reference laboratory, diagnostic DNA samples obtained from BM mononuclear cells at diagnosis were screened by PCR amplification for somatic gene rearrangements as per laboratory specific strategy, within the frame of the Euro-MRD. Briefly, after identifying patient-specific IG/TR junctional region sequences, complementary allele-specific oligonucleotide primers were designed, and PCR-MRD targets were tested for specificity and sensitivity. Quantitative PCR analysis was performed and interpreted according to the guidelines developed within the EuroMRD network.11,12 In contrast to the corresponding serial dilution standard curve, a quantitative range (QR) for each marker was defined as the lowest dilution which gave a specific and reproducible amplification (DeltaCT of all replicates less than or equal to 1.5; all cycle threshold (CT) values at least 1.0 Ct less than the lowest CT value of the amplification observed in normal mononuclear cell (MNC) DNA). Moreover, the mean CT value had to be within a range defined by the mean CT value of the previous dilution point: 2.6–4.0 CT between 10-fold dilutions (e.g., 10-3 vs. 10-4), and 0.5–1.5 CT between 2-fold dilutions (e.g,. 10-3 vs. 5 times 10-4). The majority of national referral laboratories for BCR/ABL1 monitoring followed the protocol recommended by the Europe Against Cancer (EAC) consortium.13 All laboratories participated in the development of guidelines for the interpretation of BCR/ABL1 RQ-PCR data, and participated in annual quality control rounds in the frame of EuroMRD (Pfeifer H. et al., unpublished observations). Briefly, a standard curve was produced by plasmid dilutions in order to define a QR, similarly to that for IG/TR (see above). A sample was termed ‘positive’ if at least one replicate was positive, and it was quantified if the PCR amplification was within the QR; any MRD-positive sample outside the quantitative range was scored as ‘positive not quantifiable’ (i.e., below the value of the QR). A sample was defined as ‘negative’ when all replicates were negative with at least 10000 ABL1 copies detected. The ratio between BCR/ABL1 to 10000 ABL1 copies was calculated at diagnosis and at each follow-up time point. The MRD value of each follow-up sample was calculated as the logarithmic reduction with respect to the diagnostic value. The concordance between IG/TR and BCR/ABL1 results were assessed within the cohort of patients in whom MRD was measured with both methodologies. Results were considered discordant in the case of at least one log difference if both transpired to be positive; in the case of a negative result by only one of the two methodologies, results were considered discordant if the difference between the positive result and the sensitivity (QR) of the negative one was more than one log (although this could overestimate concordance).

Statistical analysis EFS was calculated as the time from start of EsPhALL treatment (i.e., Phase IB) to first failure, defined as resistance, relapse, death from any cause, or second malignant neoplasm (SMN). Observation periods were censored at date of last contact when no event was observed. The final follow-up was on June 30, 2014 and the median follow-up time was 6.4 years (range: 0.2 – 10.4). EFS curves were estimated according to Kaplan-Meier (with Greenwood standard error) and compared with the log-rank test. haematologica | 2018; 103(1)


Predictive value of MRD in Ph+ ALL children treated with imatinib

A

B

D

C

E

The Cox model, including MRD at TP2 (positive vs. negative) and EsPhALL risk stratification (GR vs. PR) was used for multivariate analysis. CIR was estimated adjusting for competing risks of other events and compared with the Gray test.14 The two methodologies used for MRD measurement were compared using the BlandAltman approach for analyses of agreement between two different assays.15 The differences between the two log-transformed measures on each subject were plotted against their average value. After excluding any dependence, the 95% range for the difference, calculated from twice the standard deviation and the hypothesis of zero mean difference (bias), was examined with a paired t-test. All tests were two-sided. All analyses were performed with SAS software (version 9.2).

Results Overall, the 5-year EFS (standard error [SE]) of 128 EsPhALL patients who received imatinib was 62.0 (4.3). Of note, all patients eventually achieved first complete remission (CR1) by the end of HR Block3. The outcome of 108 (84%) transplanted and 20 (16%) non-transplanted patients is described in Online Supplementary Table S1. The size of each cohort of patients with available MRD data by methodology and BM time point is shown in Figure 1. In a cascade order, patients who had TP1, the first two time points, the first three time points, or all four time points (TP1, TP2, TP3 and TP4) decreased progressively from 90 to 44 for IG/TR MRD (Figure 1B) and from 61 to 23 for BCR/ABL1 MRD (Figure 1C). The distribution of haematologica | 2018; 103(1)

Figure 1. EsPhALL treatment schema and MRD results. EsPhALL treatment schema with minimal residual disease (MRD) sampling time points (time point, panel A). Sample size at different follow-up time points, by IG/TR (panel B) and BCR/ABL1 (panel C). MRD load at different follow-up time points, by IG/TR (panel D) and BCR/ABL1 (panel E): MRD negative (white), low positive (<5x10-4, gray) and highly positive (â&#x2030;Ľ5x10-4, black). BM: bone marrow; HSCT: hematopoietic stem cells transplantation; IG/TR: immunoglobulin/Tcell receptor; G-CSF: granulocyte-colony stimulating factor.

patients as MRD negative (with QR of at least 1x10-4), MRD low positive (<5x10-4) and MRD highly positive (i.e., â&#x2030;Ľ5x10-4) is shown in Figure 1D,E. As expected, MRD negativity increased progressively, from 10% at TP1 to 57% at TP4 by IG/TR (Figure 1D) and from 13% at TP1 to 30% at TP4, by BCR/ABL1 (Figure 1E).The proportion of patients with highly positive MRD decreased from 78% to 14% and from 80% to 57% at TP1 to TP4 with IG/TR and BCR/ABL1 techniques, respectively. The probability of IG/TR negativity at any time point was higher in patients classified as GR vs. those at PR, 18% vs. 4%, respectively, at TP1 (before any exposure to imatinib, P=0.0368), 39% vs. 16% at TP2 (after exposure to imatinib during Protocol IB, P=0.0533), 60% vs. 36% (P=0.1429) at TP3, and 70% vs. 46% at TP4 (P=0.1350; Online Supplementary Figure S1A,C). Corresponding figures for BCR/ABL1 were 25% vs. 3%, 37% vs. 4%, 60% vs. 10% and 50% vs. 20%, respectively (Online Supplementary Figure S1B,D). Among the 90 patients with TP1 available for IG/TR, 51 carried the p190 BCR/ABL1 variant, 6 carried that of p210, while for the remaining 33 patients the variant was unknown; concerning the 61 patients with BCR/ABL1 monitored at TP1, the same distribution was 41, 7, and 13, respectively. Figure 2 shows the impact of MRD levels measured by IG/TR PCR on EFS and CIR at TP1, TP2, TP3, and TP4, respectively. Ninety patients were tested at TP1 (Figure 2A,B). The minority of patients who were MRD negative (N=9, 10%) had a favorable outcome with no relapses, 109


G. Cazzaniga et al.

whereas patients with high or low MRD positivity had similar 5-year CIR (SE) of 35.2 (5.9) and 36.4 (15.4), respectively. The difference in CIR between MRD negative and positive patients was significant (0.0 vs. 35.3 (5.4), P=0.0368). With respect to TP2, TP3 and TP4, the outcome of patients who achieved negativity at the respective time point (new negative) was analyzed separately from the outcome of those who achieved negativity earlier on. As shown, MRD negativity at TP2 (N=14, Figure

A

B

C

D

E

F

G

110

H

2C,D), in both GR and PR patients, is associated with low risk of relapse (two relapses, 5-year CIR (SE) 14.3 [9.8]), while the four patients who had achieved negativity at TP1 did not present relapses. Achieving negativity at TP3 or TP4 is associated with a CIR of 36.4 (15.5) and 42.9 (21.6), respectively, comparable to that of patients with low or high positivity at any time point. In a multivariate analysis adjusting for risk group stratification, MRD negativity within TP2 was associated with a non-statistically

Figure 2. Event-free survival and cumulative incidence of relapse according to IG/TR MRD levels. Event-free survival (EFS; panels A, C, E and G) and cumulative incidence of relapse (CIR; panels B, D, F and H) according to IG/TR MRD levels at time points 1 through 4 (TP1-TP4). At TP1, there were seven good-risk and two poor-risk (panel B) negative patients with no relapses. At TP2, among the 14 new negative patients, nine were good-risk (1 relapse) and five were poor-risk (1 relapse). CUM: cumulative; EsPhALL: European intergroup study of post-induction treatment of Philadelphia-chromosome-positive acute lymphoblastic leukemia; SE: standard error.

haematologica | 2018; 103(1)


Predictive value of MRD in Ph+ ALL children treated with imatinib

significant reduced risk of failure as compared to positivity at any level (hazard ratio (HR)=0.50, 95%CI (0.19 - 1.38), P=0.1811). Table 1 shows the kinetics of MRD levels from TP1 to TP2 in 68 patients with IG/TR evaluation at both time points. Out of ten cases with low MRD at TP1, 7 (70%) became negative at TP2, while in patients with high MRD, only 7 out of 54 (13%) became negative. Figure 3 shows that BCR/ABL1 MRD negativity at TP1 or TP2, although based on small numbers, is reached less frequently when compared with IG/TR MRD, but is associated with a similar very low risk of relapse (only one relapse in eight MRD patients negative at TP1). Patients with low or high positivity have, however, a high risk of relapse, again similar to that observed for positive IG/TR MRD. At subsequent time points, the number of patients for which BCR/ABL1 MRD data are available is too small to allow us to draw any conclusions. The comparison between MRD sample values obtained by IG/TR vs. BCR/ABL1 is shown in Table 2 and Figure 4. The overall level of concordance is 69% (Table 2 and Figure 4A). The concordance rate between sample values at each time point (see Figure 4B,C for TP1 and TP2, respectively) were similar, ranging from 65% at TP3 to 71% at TP2 (Table 2). For patients with positive

MRD by both techniques, where an adequate number of samples were available, we closely examined the differences between the two methods using the Bland-Altman algorithm for TP1 and TP2 (Figure 4D,E). Following logarithmic transformation, the estimated mean differences of BCR/ABL1 versus IG/TR results were significantly greater than zero at both time points (P=0.03 and P=0.001 at TP1 and TP2, respectively), signifying that the BCR/ABL1 value tended to be higher than the IG/TR (the estimated average difference was 0.25 (SD 0.66) at TP1, and 0.43 (SD 0.73) at TP2). There were also four patients who were negative by IG/TR but positive at low or high Table 1. Distribution of patients according to IG/TR MRD levels at TP1 and TP2.

IG/TR MRD TP1 Neg Low MRD High MRD Total

Neg

Low MRD

IG/TR MRD TP2 High MRD

Total

4 (0) 7 (1) 7 (1) 18 (2)

0 3 (2) 14 (4) 17 (6)

0 0 33 (14) 33 (14)

4 (0) 10 (3) 54 (19) 68 (22)

Patients by IG/TR MRD levels at TP1 and TP2 (number of relapses in parenthesis). Neg = MRD negative, low MRD = MRD <5x10-4 and high MRD = MRD≥5x10-4.TP1: time point 1; TP2: time point 2; IG/TR: immunoglobulin/T-cell receptor; MRD: minimal residual disease.

A

B

C

D

Figure 3. Event-free survival and cumulative incidence of relapse according to BCR/ABL1 MRD levels. Event-free survival (EFS; panels A, C) and cumulative incidence of relapse (CIR; panels B, D) according to BCR/ABL1 MRD levels at time points 1 (TP1) and 2 (TP2). The patient who was negative at TP1 and relapsed is not represented at TP2 because their MRD at this time point was not available. At TP2 only three patients were ‘new negative’. CUM: cumulative; EsPhALL: European intergroup study of post-induction treatment of Philadelphia-chromosome-positive acute lymphoblastic leukemia; SE: standard error.

haematologica | 2018; 103(1)

111


G. Cazzaniga et al.

level by BCR/ABL1, as shown by the individual patterns in Figure 5. Two of these patients were negative by IG/TR at TP1, TP2, TP3, and TP4 (Figure 5A,B), and the remaining two became negative either at TP2 (Figure 5C)

or TP3 (Figure 5D), and remained negative thereafter. All received HSCT in CR1, and only the patient in panel C relapsed. No cases with persistently negative BCR/ABL1 and positive IG/TR were observed.

Table 2. MRD detection by methodology at each time point.

BCR/ABL1 MRD

≥5x10 <5x10-4 Neg concordance rate -4

TP1 N=50 ≥5x10-4 <5x10-4 Neg 30/35 0/1 0/1

0/4 0/3 3/3 0/0 1/2 1/1 35/50=70%

IG/TR MRD TP2 TP3 N=56 N=40 ≥5x10-4 <5x10-4 Neg ≥5x10-4 <5x10-4 Neg 24/28 0/2 0/0

4/8 0/6 2/2 3/3 0/0 7/7 40/56=71%

4/8 0/1 0/0

3/5 0/6 4/4 3/4 5/5 7/7 26/40=65%

TP4 N=37 ≥5x10-4 <5x10-4 4/5 0/1 0/0

Overall N=183 Neg≥5x10-4<5x10-4

3/7 2/8 1/1 5/5 0/0 10/10 25/37=68%

Neg

62/76 10/24 2/23 0/5 10/10 11/12 0/1 6/7 25/25 126/183=69%

In each cell, the number of concordant over total number of samples analyzed are reported. Discordance was based on the following criteria: if both MRD results were positive, the difference between MRD results was at least 1 logarithm, and if MRD was negative by one methodology and positive by the other, the difference between the positive result and the sensitivity (quantitative range) of the negative one was more than 1 logarithm. Neg = MRD negative. TP1: time point 1; TP2: time point 2; TP3: time point 3; TP4: time point 4. IG/TR: immunoglobulin/T-cell receptor; MRD: minimal residual disease.

A

B

C

D

E

N=37, P=0.03

112

N=34, P=0.001

Figure 4. Overall concordance of IG/TR vs. BCR/ABL1 MRD. Scatterplot of IG/TR- and BCR/ABL1-based minimal residual disease (MRD) sample values on the logarithmic scale. Positive samples with MRD below the quantitative range ('positive, not quantifiable' – POS, NQ) were conventionally represented with 10-6; MRD negative samples are labeled with NEG. The black diagonal line represents the exact agreement; the area within the green lines includes concordant samples with acceptable agreement, defined as less than 1 log difference between the two measurements. Red boxes include discrepant cases, while green boxes include cases with either ‘POS,NQ’ or ‘NEG’ results via at least one of the two methods, for which assessment of concordance was based on sensitivity. n=number of samples; numbers of patients are indicated in parenthesis. Panel A includes all samples for all patients, while panels B-D and C-E show only samples at TP1 and TP2, respectively. Panel D and E show the comparison of MRD measurements by IG/TR and BCR/ABL1 according to Bland-Altman, with the continuous line representing zero difference, and the dashed line representing the estimated mean difference ±2 SD. Among ten IG/TR negative and BCR/ABL1 positive patients (who were concordant according to the definition above), seven carried the p190 fusion protein, one carried the p210 fusion protein and two did not have this information available. The fusion protein detected in nine IG/TR POS<QR and BCR/ABL1 positive patients (who were concordant according to the definition above) was p190 in eight patients, and not known in the remaining patient. IG/TR: immunoglobulin/T-cell receptor.

haematologica | 2018; 103(1)


Predictive value of MRD in Ph+ ALL children treated with imatinib

Discussion Ph+ ALL in pediatric age is very rare, occurring in only 3% of cases, thus intergroup studies are needed to continue investigations into this ALL subtype.1 The predictive value of MRD response in children with Ph+ ALL treated with TKIs is still unknown. A recent study reporting on solely nine children with Ph+ ALL showed that TKIs added to chemotherapy allowed for a marked increase in the rate of patients with negative MRD and improved outcome as compared to patients treated with chemotherapy alone.16 In adult patients treated with TKIs it has been reported that MRD negativity before HSCT in CR1, and its persistence following HSCT, are associated with a significantly better outcome.17-22 EsPhALL was one of the first studies to introduce imatinib in childhood Ph+ ALL treated with a BFM-type therapy,7 and the results reported herein on MRD monitoring provide novel information on the relevance and prognostic impact of MRD response. An earlier study conducted by the Children's Oncology Group (COG) analyzed the prognostic impact of MRD;8 their results suggested that a better MRD response was associated with a more favorable outcome, but the difference was not statistically significant, probably due to the small number of patients investigated. In our study, data are available on a relatively large number of patients after induction (TP1) and consolidation (TP2) phases, as well as after the two blocks of intensification therapy (TP3 and TP4).

The probability of achieving MRD negativity at early time points was predominantly lower than in other populations of BCP-ALL, in keeping with previous reports.9 This probability was higher in EsPhALL patients who qualified as GR vs. PR: for instance, at TP2 after exposure to imatinib in consolidation phase IB, MRD negativity was 39% vs. 16% when measured by IG/TR (P=0.0533) and of 37% vs. 4% by BCR/ABL1 (P=0.0047). Of note, the definition of MRD negativity was herewith based on strict criteria, requiring a QR of at least 1x10-4. In a multivariate analysis, MRD negativity at TP2 was associated with a reduced risk of failure, yet was not considered statistically significant. Definitive conclusions on the role of MRD as an independent prognostic factor in this setting are difficult due to the limited sample size. These data suggest that traditional unfavorable prognostic criteria remain associated with less favorable MRD response and, as reported, with poorer final outcome.7 Without accounting for the kinetics of MRD levels across subsequent time points, MRD negativity at an early time point is associated with a low number of relapses, both in GR and PR patients (Online Supplementary Figure S1). In our experience, the 5-year CIR was very low for patients achieving MRD negativity subsequent to either induction or consolidation phases. When MRD negativity was obtained later (TP3 and TP4), CIR was instead quite high, and not different from that observed for positive MRD, suggesting that the earlier negative MRD is achieved, the lower the risk of relapse. Taken together, our data suggest that BFM-type back-

A

B

C

D

Figure 5. MRD in discordant cases. MRD pattern from TP1 through TP4 (or TP5 if available) of four patients with persistently discordant results by IG/TR (dashed lines) and BCR/ABL1 (continuous lines). Patients identified with ID 184, ID 111 and ID 44 carried the p190 variant protein; for patient ID 25, this information was not available.

haematologica | 2018; 103(1)

113


G. Cazzaniga et al.

bone chemotherapy defines two main subsets of children with Ph+ ALL, i.e., those who are capable of achieving molecular remission within TP2, and those who will achieve this result only at a later stage, or never, which will make little to no difference in terms of finding a possible cure and thus eradicating the disease. MRD results based on BCR/ABL1 confirm that an early achievement of MRD negativity, i.e., at TP2, has a similar predictive value of MRD detected by IG/TR. However, no conclusions can be drawn on subsequent time points due to the small number of patients analyzed by BCR/ABL1. In this study, the IG/TR and BCR/ABL1 RT-PCR methodologies for MRD detection were compared for the first time in a relatively large and multicentric setting. We showed that BCR/ABL1 MRD levels tend to be higher than those based on IG/TR at any time point; as a consequence, the proportion of patients achieving MRD negativity by BCR/ABL1 is lower when compared to IG/TR (i.e., 30% vs. 57% at TP4). Moreover, there are several discrepant samples (31%) in the majority of cases which had a higher MRD level by BCR/ABL1 than that by IG/TR. This is in keeping with a previous single-center report on 17 childhood ALL patients,23 and a more recent report from the same group,24 wherein 20% and 23% of samples were BCR/ABL1-positive and IG/TR-negative, respectively. Interestingly, the testing of cell-sorted hematopoietic subpopulations at diagnosis of patients with discrepant MRD results revealed BCR-ABL1-positivity in non–ALL B lymphocytes, T cells, and/or myeloid cells. This indicates the multilineage involvement of the BCR-ABL1-positive clone in some patients diagnosed with BCR-ABL1-positive ALL, in which a multipotent hematopoietic progenitor might have been the target of the BCR-ABL1 translocation. These patients have a BCR-ABL1-positive clonal hematopoiesis resembling a chronic myeloid leukemia (CML) disease, although other features (i.e., lower white blood count (WBC) level at diagnosis and a lower proportion of non-B cells positive for BCR/ABL1 than in CML blast crisis) are different from typical CML, thus suggesting that those Ph+ ALL with discordant IG/TR vs. BCR/ABL1 MRD might represent a rare subgroup with a ‘CML-like’ feature.24 Whether this biological heterogeneity may have an impact on patient outcomes and optimal treatment (early stem cell transplantation SCT vs. continuous TKI therapy) as well as on MRD testing needs further prospective investigations in larger series. Of note, all of our four discordant patients with high BCR/ABL1 and negative IG/TR

References 1. Pui CH, Evans WE. Treatment of acute lymphoblastic leukemia. N Engl J Med. 2006;354(2):166-178. 2. Schrappe M, Camitta B, Pui CH, et al. Longterm results of large prospective trials in childhood acute lymphoblastic leukemia. Leukemia. 2000;14(12):2193-2320. 3. Aricò M, Schrappe M, Hunger SP, et al. Clinical outcome of children with newly diagnosed Philadelphia chromosome-positive acute lymphoblastic leukemia treated between 1995 and 2005. J Clin Oncol. 2010;

114

MRD levels (two of them with two IG/TR markers) at several consecutive time points received HSCT. In addition, the evidence that t(9;22) is the primary, but not sufficient leukemogenic event, underlines the risk of monitoring pre-leukemic cells by BCR/ABL1 only.25 Thus, MRD monitoring by both methods may be functional for both measuring response and guiding biological studies aimed at investigating the causes of discrepancies; from our data IG/TR MRD monitoring appears to be more reliable. In conclusion, the present study in a large multicenter series of children with Ph+ ALL indicates, for the first time, that MRD is also highly predictive of prognosis in the context of treatment with imatinib and HSCT, and in particular, that early negativity is strongly associated with good prognosis. On the contrary, clearance of MRD after TP2 did not show a prognostic impact on CIR, which is a useful indication for the design of future trials as it suggests that reliable clinical decisions, particularly regarding indications of individual treatment intensification, may be made in a timely and safe fashion based on the TP2 results of the patients. It would be challenging to investigate whether these findings will be confirmed in the current protocols, where a lower proportion of patients is undergoing HSCT and imatinib is administered earlier and continuously. If so, MRD monitoring could be used to refine clinical decisions, including how to optimize the use of TKIs and whether to perform HSCT. Acknowledgments We thank Jacques Van Dongen as a Chair of the EuroMRD network and Heike Pfeifer as a leader of the Ph+ALL branch of EuroMRD, Jocelyne Vivent for data management of French patients, and all technicians from the referral laboratories, doctors from all Centers and all patients Funding This project was partially funded by the following grants: Italian Association for Cancer Research (AIRC, to GC and AB), AIRC 2013-14634 (to MGV), grants from the Ministry of Health of the Czech Republic #16-30186A (Czech Health Research Council) and #00064203 (University Hospital Motol) (to JZ and JS), the "Programme Hospitalier de Recherche Clinique" (PHRC) and the association "Enfants et Santé" (to HC and VG). MRD was supported by Bloodwise and trial management by Cancer Research UK; VS is a Margdarshi Fellow of the Wellcome-DBT India Alliance.

28(31):4755-4761. 4. Aricò M, Valsecchi MG, Camitta B, et al. Outcome of treatment in children with Philadelphia chromosome-positive acute lymphoblastic leukemia. N Engl J Med. 2000;342 (14):998-1006. 5. Schultz KR, Bowman WP, Aledo A, et al. Improved early event-free survival with imatinib in Philadelphia chromosome-positive acute lymphoblastic leukemia: a children's oncology group study. J Clin Oncol. 2009;27(31):5175-5181. 6. Slayton WB, Schultz KR, Jones T, et al. Continuous dose dasatinib is safe and fea-

sible in combination with intensive chemotherapy in pediatric Philadelphia chromosome positive acute lymphoblastic leukemia (Ph+ ALL): Children's Oncology Group (COG) trial AALL0622. Blood. 2012; 120(21):137. 7. Biondi A, Schrappe M, De Lorenzo P, et al. Imatinib after induction for treatment of children and adolescents with Philadelphiachromosome-positive acute lymphoblastic leukaemia (EsPhALL): a randomised, openlabel, intergroup study. Lancet Oncol. 2012;13(9):936-945. 8. Schultz KR, Carroll A, Heerema NA, et al.

haematologica | 2018; 103(1)


Predictive value of MRD in Ph+ ALL children treated with imatinib

9.

10.

11.

12.

13.

Long-term follow-up of imatinib in pediatric Philadelphia chromosome-positive acute lymphoblastic leukemia: Children’s Oncology Group Study AALL0031. Leukemia. 2014;28(7):1467–1471. Conter V, Bartram CR, Valsecchi MG, et al. Molecular response to treatment redefines all prognostic factors in children and adolescents with B-cell precursor acute lymphoblastic leukemia: results in 3184 patients of the AIEOP-BFM ALL 2000 study. Blood. 2010;115(16):3206–3214. Schrappe M, Valsecchi MG, Bartram CR, et al. Late MRD response determines relapse risk overall and in subsets of childhood Tcell ALL: results of the AIEOP-BFM-ALL 2000 study. Blood. 2011;118(8):2077–2084. Flohr T, Schrauder A, Cazzaniga G, et al. Minimal residual disease-directed risk stratification using real-time quantitative PCR analysis of immunoglobulin and T-cell receptor gene rearrangements in the international multicenter trial AIEOP-BFM ALL 2000 for childhood acute lymphoblastic leukemia. Leukemia. 2008;22(4):771–782. Van der Velden VH, Cazzaniga G, Schrauder A, et al. Analysis of minimal residual disease by Ig/ TCR gene rearrangements: guidelines for interpretation of realtime quantitative PCR data. Leukemia. 2007;21(4):604-611. 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

haematologica | 2018; 103(1)

14.

15.

16.

17.

18.

19.

Cancer program. Leukemia. 2003;17(12): 2318–2357. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(446):496509. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476):307-310. Jeha S, Coustan-Smith E, Pei D, et al. Impact of tyrosine kinase inhibitors on minimal residual disease and outcome in childhood Philadelphia chromosome-positive acute lymphoblastic leukemia. Cancer. 2014;120(10):1514-1519. Nishiwaki S, Imai K, Mizuta S, et al. Impact of MRD and TKI on allogeneic hematopoietic cell transplantation for Ph+ALL: a study from the adult ALL WG of the JSHCT. Bone Marrow Transplant. 2016; 51(1):43-50. Lee S, Kim DW, Cho B, et al. Risk factors for adults with Philadelphia-chromosomepositive acute lymphoblastic leukaemia in remission treated with allogeneic bone marrow transplantation: the potential of real-time quantitative reverse-transcription polymerase chain reaction. Br J Haematol. 2003;120(1):145-153. Yanada M, Sugiura I, Takeuchi J, et al. Prospective monitoring of BCR-ABL1 transcript levels in patients with Philadelphia chromosome-positive acute lymphoblastic leukaemia undergoing imatinib-combined chemotherapy. Br J Haematol. 2008; 143(4):503-510.

20. Ottmann OG, Wassmann B, Pfeifer H, et al. GMALL Study Group. Imatinib compared with chemotherapy as front-line treatment of elderly patients with Philadelphia chromosome-positive acute lymphoblastic leukemia (Ph+ALL). Cancer. 2007; 109(10):2068-2076. 21. Radich JP, Gehly G, Gooley T, et al. Polymerase chain reaction detection of the BCR-ABL fusion transcript after allogeneic marrow transplantation for chronic myeloid leukemia: results and implications in 346 patients. Blood. 1995; 85(9):2632– 2638. 22. Stirewalt DL, Guthrie KA, Beppu L, et al. Predictors of relapse and overall survival in Philadelphia chromosome-positive acute lymphoblastic leukemia after transplantation. Biol Blood Marrow Transplant. 2003;9 (3):206–212. 23. Zaliova M, Fronkova E, Krejcikova K, et al. Quantification of fusion transcript reveals a subgroup with distinct biological properties and predicts relapse in BCR/ABL-positive ALL: implications for residual disease monitoring. Leukemia. 2009;23(5):944–951. 24. Hovorkova L, Zaliova M, Venn NC, et al. Monitoring of childhood ALL using BCRABL1 genomic breakpoints identifies a subgroup with CML-like biology. Blood. 2017;129(20):2771-2781. 25. Cazzaniga G, van Delft FW, Lo Nigro L, et al. Developmental origins and impact of BCR-ABL1 fusion and IKZF1 deletions in monozygotic twins with Ph+ acute lymphoblastic leukemia. Blood. 2011;118(20): 5559-5564.

115


ARTICLE

Non-Hodgkin Lymphoma

Ferrata Storti Foundation

Inhibition of focal adhesion kinase overcomes resistance of mantle cell lymphoma to ibrutinib in the bone marrow microenvironment Martina Rudelius,1,2 Mathias Tillmann Rosenfeldt,1 Ellen Leich,1 Hilka Rauert-Wunderlich,1 Antonio Giovanni Solimando,3 Andreas Beilhack,3 German Ott4 and Andreas Rosenwald1

Institute of Pathology, University of Würzburg and CCC-Mainfranken, Würzburg; Institute of Pathology, University of Duesseldorf; 3Medizinische Klinik II, University Hospital of Würzburg; 4Department of Clinical Pathology, Robert-Bosch-Krankenhaus, and Dr. Margarete Fischer-Bosch Institute of Clinical Pharmacology, Stuttgart, Germany 1

Haematologica 2018 Volume 103(1):116-125

2

ABSTRACT

M

Correspondence: rosenwald@mail.uni-wuerzburg.de

Received: July 29, 2017. Accepted: October 16, 2017. Pre-published: October 27, 2017.

antle cell lymphoma and other lymphoma subtypes often spread to the bone marrow, and stromal interactions mediated by focal adhesion kinase frequently enhance survival and drug resistance of the lymphoma cells. To study the role of focal adhesion kinase in mantle cell lymphoma, immunohistochemistry of primary cases and functional analysis of mantle cell lymphoma cell lines and primary mantle cell lymphoma cells co-cultured with bone marrow stromal cells (BMSC) using small molecule inhibitors and RNAi-based focal adhesion kinase silencing was performed. We showed that focal adhesion kinase is highly expressed in bone marrow infiltrates of mantle cell lymphoma and in mantle cell lymphoma cell lines. Stroma-mediated activation of focal adhesion kinase led to activation of multiple kinases (AKT, p42/44 and NF-κB), that are important for prosurvival and proliferation signaling. Interestingly, RNAi-based focal adhesion kinase silencing or inhibition with small molecule inhibitors (FAKi) resulted in blockage of targeted cell invasion and induced apoptosis by inactivation of multiple signaling cascades, including the classic and alternative NF-κB pathway. In addition, the combined treatment of ibrutinib and FAKi was highly synergistic, and ibrutinib resistance of mantle cell lymphoma could be overcome. These data demonstrate that focal adhesion kinase is important for stroma-mediated survival and drug resistance in mantle cell lymphoma, providing indications for a targeted therapeutic strategy.

Introduction doi:10.3324/haematol.2017.177162 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/116 ©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.

116

Mantle cell lymphoma (MCL) is an aggressive B-cell lymphoma with a poor prognosis, and a significant number of patients relapse after treatment.1 Promising results can be achieved in relapsed or refractory MCL with ibrutinib, a small molecule inhibitor of Bruton tyrosine kinase (BTK), with a significant improvement in progression-free survival. However, despite this, primary resistance to ibrutinib occurs in one-third of all patients. Acquired secondary resistance has also been described.2-4 Although some mechanisms of resistance, such as activation of the alternative NF-αB signaling pathway,5 mutations in the BTK binding site and others6 have been identified, most mechanisms of ibrutinib resistance remain unclear, and multiple mechanisms are likely to be involved. In several B-cell malignancies, stromal interactions support cell survival, and it has been shown that in MCLs bone marrow (BM) stromal interaction can increase drug resistance.7 Over 90% of MCL patients have extranodal manifestations, and especially the aggressive blastoid variant of MCL is characterized by bone marrow involvement. Homing to the BM requires the expression of adhesion molecules on the lymphoma cells and intact intracellular signaling, with the classic and alternative NF-κB signaling pathway being some of the major components.7 Recently, focal adhesion kinase (FAK), a major signaling molecule that functions downstream of integrins and that translates signals from the extracellular matrix,8,9 has gained attention as a drug target in the treatment of solid tumors. Several studhaematologica | 2018; 103(1)


Focal adhesion kinase in MCL

ies have demonstrated that FAK can enhance cell proliferation, survival and migration in response to stromal interaction.10,11 Therefore, we chose to study the role of FAK in BM stroma-mediated enhancement of MCL proliferation and survival. We identified FAK inhibition as a possible mechanism of restoring the ibrutinib response, which makes it an attractive target for combination treatment, especially in patients who present with BM involvement.

genome, was cloned into lentiviral Tet-regulated expression vector (pRRLT3GmiR-E, a kind gift from J. Zuber13 to MTR).

Molecular targets for therapy test and combination treatment The molecular targets for therapy (MTT) test was performed as previously described.13 Synergism was assessed by the ChouTalalay method,14 calculating combination index (CI) values using CalcuSyn software (Biosoft, Cambridge, UK).

Invasion assay Methods Primary cases and cell lines Thirty primary MCL cases [10 typical MCLs, 10 MCLs of the blastoid variant, and 10 paired typical MCL samples of BM infiltrates and extramedullary infiltrates (lymph node or gastro-intestinal tract)] were selected from the files of the Institute of Pathology, University of Wuerzburg, Germany. The cases were classified according to the World Health Organization (WHO) classification as typical MCL or as blastoid variant. All human specimens were processed after informed consent in compliance with the institutional review board of the Faculty of Medicine of the University of Wuerzburg, Germany, and conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report. Nine well-characterized and widely used MCL cell lines were used in this study: Granta 519, Z138C, HBL-2, REC-1, JEKO, MINO, MAVER, JVM-2 and UPN-1. BM stromal cells (BMSC) were isolated from BM samples from patients as previously described.12 For co-culture experiments, BMSC were plated overnight, and after confirming the confluence of the stroma layer, medium was replaced by 5x105 MCL cells in RPMI-1640. Drugs were added after 4 hours (h) of incubation and ibrutinib was preincubated for 30 minutes (min) before addition of VS-6063.

Immunoreagents and inhibitors The following antibodies were used for immunoblotting and immunohistochemistry: FAK, pFAK (Tyr397), pPaxillin (Tyr118), pAKT (Ser473), actin, p-p42/44 (Tyr202/204), pGSK3β (Ser9), pIκBα (Ser32/36), IKKα, pIKKα/β (Ser176/180), p52, cleaved caspase-3, anti-mouse and anti-rabbit IgG horseradish peroxidase (HRP)-linked from Cell Signaling (Beverly, MA, USA). Cyclin D1 was obtained from Thermo Scientific (Waltham, MA, USA); cMyc was from Abcam (Cambridge, UK). Immunodetection was performed with the DAKO REAL detection kit (DAKO GmbH, Hamburg, Germany). The following inhibitors and immunoreagents were used: VS6063 (Selleckchem, Muenchen, Germany), ibrutinib (Selleckchem, Muenchen, Germany), and rhCXCL-12 (R&D Systems, Wiesbaden, Germany).

Western blot analysis, immunoprecipitation and immunohistochemistry Western blot analysis, immunoprecipitation and immunohistochemistry were performed as previously described.13

microRNA sequences, plasmid constructs and lentiviral transduction A short hairpin oligonucleotide directed against FAK was designed. The sense strand of the FAK-microRNA sequence was TGCTGTTGACAGTGAGCGAAGCGATTATGTTAGAGATAT AGTGAAGCCACAAGATGTATATCTAACATATAATCGCTCT GCCTACTGCCTCGGA. FAK oligonucleotide and a control microRNA, which lacks complementary sequences in the human haematologica | 2018; 103(1)

Matrigel-coated nucleopore filter inserts in a 24-well transwell chamber (Corning Biocoat, New York, USA) were used for the invasion assays. Cells (treated or untreated with VS-6063) were seeded at a density of 40,000 cells/well into the upper part of the Matrigel-coated filter, and RPMI-1640 supplemented with 10% heat-inactivated fetal bovine serum and rhCXCL-12 (R&D Systems, Wiesbaden, Germany) (1 ng/mL) was added to the lower part. After 24 h, the cells that had migrated through the Matrigel and the 8-μm pore-size membrane were fixed, stained, and counted under a light microscope.

Statistical analysis Continuous variables and categorical variables were compared by t-test or Fisher exact test. All reported P-values were two-sided; P<0.05 was considered statistically significant.

Results FAK is highly expressed in bone marrow infiltrates of MCL and in MCL cell lines We first examined the expression of FAK in primary MCL samples by immunohistochemistry. Cases selected were lymph node infiltrates of MCL that were classified according to the WHO as 10 typical MCLs and 10 MCLs of the blastoid variant. They showed only mild (n=14) or no (n=5) expression of FAK, with only one typical case displaying high FAK expression. FAK expression did not correlate with the WHO classification (typical or blastoid) (Figure 1A). As FAK can be induced and activated by integrin and extracellular matrix signaling,8,15 we next compared FAK expression in BM infiltrates versus an extramedullary (lymph node or gastro-intestinal tract) manifestation of MCL. We performed immunohistochemistry of 10 paired samples (medullary and extramedullary infiltrates) of typical MCLs. All BM infiltrates were characterized by high FAK expression, whereas only one infiltrate in the colon showed high expression of FAK. All other extramedullary infiltrates displayed no or only weak staining of FAK (Figure 1C). Thus, high FAK expression correlated with infiltration of the BM (P=0.0001). We then evaluated FAK expression in MCL cell lines. We performed western blot analysis of 9 well-characterized MCL cell lines (Granta 519, Z138C, REC-1, HBL-2, JEKO, MAVER, MINO, JVM-2, UPN-1). All cell lines showed clear FAK expression with the highest levels of FAK expression in MINO and HBL-2. They also showed high expression of pFAK (Tyr397), which is the autophosphorylation and binding site for Src family kinase members (PI3K, PLCα) (Figure 1D).

FAK expression and activation in MCL can be induced by CXCL12 Lymphoma migration and homing requires the co-oper117


M. Rudelius et al.

ation between chemokines, adhesion molecules, ligands and their receptors expressed by stromal and lymphoma cells. MCLs express G-protein-coupled chemokine receptors such as CXCR4 and CXCR5 that bind CXCL12.16 Hence, we determined whether FAK expression and activation could be induced by CXCL12. We cultured two different MCL cell lines, JEKO and Z138C, in the presence of 1 ng/mL rhCXCL12. Western blot analysis was performed after 15, 30 and 60 min. Expression of total FAK and pFAK (Tyr397) increased after incubation with rhCXCL12. In addition, rhCXCL12 induced increased phosphorylation of the direct downstream target of FAK, Paxillin, confirming FAK activation (Figure 2A). Focal adhesion kinase can mediate cell proliferation and

A

survival in many types of solid and non-solid tumors17 and, therefore, we analyzed downstream targets affected by FAK signaling in MCL. Indeed, FAK expression and activation led to phosphorylation and activation of several downstream targets. An increase in the phosphorylation of p42/44 and AKT, as well as a constant upregulation of c-MYC and Cyclin D1, could be observed (Figure 2B).

Co-culture with BMSC activates FAK signaling in MCL cell lines and primary MCL-cells We could observe a high FAK expression in BM infiltrates, and incubation of MCL cell lines with rhCXCL12 led to FAK upregulation and activation. As CXCL12 is highly expressed by BMSCs,18,19 we determined whether

B

C

Figure 1. Focal adhesion kinase (FAK) is highly expressed in bone marrow (BM) infiltrates of mantle cell lymphoma (MCL) with only weak to moderate expression in extramedullary manifestation sites, and FAK and pFAK are highly expressed in MCL cell lines. Immunohistochemistry for FAK was performed in typical (MCL1) and blastoid (MCL2) MCL cases (A), and in patients presenting with BM infiltrates and extramedullary manifestations (C); control BM without infiltrate is shown in (B). Displayed are H&E, Cyclin D1 immunohistochemistry, and FAK immunohistochemistry. (D) Western blot analysis with FAK and phospho-FAK (Tyr397) are shown for Granta 519 (1), JVM-2 (2), REC-1 (3), HBL-2 (4), JEKO (5), MAVER (6), MINO (7), Z138C (8), UPN-1 (9). Actin is shown as loading control.

118

haematologica | 2018; 103(1)


Focal adhesion kinase in MCL

co-culture of MCL with BMSCs could activate FAK signaling. Co-culture of the MCL cell lines Z138C and JEKO with BMSC resulted in FAK upregulation. Western blot analysis revealed a steady increase of FAK and phosphoPaxillin after 8, 24 and 48 h. FAK activation led to the phosphorylation and activation of multiple downstream targets, such as p42/44 and AKT, and to the upregulation of Cyclin D1 (Figure 3A). Interestingly, we could observe interactions of FAK with the NF-κB signaling pathway. FAK activation resulted in an increase in the phosphorylation of IKKα as shown by the immunoprecipitation of IKKα and detection with pIKKα/β. (To our knowledge, there is no phospho-specific IKKα antibody commercially available). IKKα is an important component of the canonical and non-canonical NF-κB pathway.20,21 Indeed, phosphorylation of IkBα led to the phosphorylation of IkBα and to an increase in p52 in MCL cell lines (Figure 3C). To rule out cell culture artefacts due to the use of MCL cell lines, we repeated the experiment with primary MCL cells. Western blot analysis showed similar results with an upregulation and activation of FAK after 72 h accompanied by an increase in p52 and increase in the phosphorylation of IkBα and AKT (Figure 3B).

FAK silencing suppresses multiple signaling pathways and FAK is essential for CXCL12 induced activation of several downstream targets Our results indicate that incubation with rhCXCL12 or co-culture with BMSCs can activate FAK and several downstream targets in MCLs. Although the downstream targets studied are well-established FAK targets, it is possible that these proteins could have been activated through alternative kinases. Therefore, we studied the effect of FAK silencing on the activation of downstream targets. We used a Tet-inducible microRNA system in the two MCL cell lines HBL-2 and MINO with high constitutive FAK expression and activation. Western blot analysis showed that, after 72 and 96 h of incubation with doxicyclin in both cell lines, FAK expression was completely abolished, whereas expression of the kinase PYK2 with high homology to FAK did not change. As expected, FAK downregulation dramatically decreased or resulted in a complete loss of activation or phosphorylation of the downstream targets Paxillin, p42/44, IkBα, p52, AKT and GSK3β, as well as in downregulation of Cyclin D1 (Figure 4A).

A

B

Figure 2. Focal adhesion kinase (FAK) expression and activation can be induced by CXCL12. Western blot analysis of JEKO and Z138C was performed 15 minutes (min), 30 min and 60 min after rhCXCL12 treatment and compared to basal levels (co) without rhCXCL12 treatment. Increasing levels of phosphorylation of FAK, Paxicillin, p42/44 and AKT were detected (A and B), accompanied by an increase in FAK, c-Myc and Cyclin D1 expression (A and B). Actin is shown as loading control.

haematologica | 2018; 103(1)

119


M. Rudelius et al.

Interestingly, incubation with doxycyclin and silencing of FAK also reduced the upregulation of phosphorylation of AKT, IkBα and p42/44 and upregulation of p52 expression after rhCXCL12 treatment, as shown by western blot analysis for HBL-2 (Figure 4B). This result clearly demonstrates that FAK participates in CXCL12-mediated activation of the AKT, p42/44 and NF-κB signaling pathways.

FAK inhibition leads to inhibition of the alternative and classical NF-κB signaling pathway Silencing of FAK resulted in the inhibition of multiple downstream targets. There are several FAK-inhibitors available that are currently being evaluated for use in clinical trials for the treatment of solid tumors.22,23 Therefore, we determined whether we could achieve a blockage of the downstream targets of FAK in MCL with VS-6063. VS6063 is a 2nd-generation inhibitor that is in clinical trials for solid tumors and that targets the FAK/PYK2 kinase domain in an (ATP)-competitive, reversible way.24 We were especially interested in the regulation of the NF-κB signaling pathway, as this pathway was highly activated by co-culture with BMSCs. We performed experiments with different MCL cell lines with activation of the classic or alternative NF-κB signaling pathway. We chose the MINO and HBL-2 cell lines, which are characterized by the activation of the classic NF-κB signaling pathway, and which showed high expression levels of FAK in our western blot analysis. In addition, we chose the cell line Z138C, with a TRAF2 mutation. The TRAF2 mutation leads to an activation of the alternative NF-κB pathway, so

Z138C is characterized by the activation of the classic and alternative NF-κB signaling pathway. The cell lines were co-cultured with BMSCs and treated with VS-6063 (100 nM). After 15 min, 8 h, and 24 h, the cells were harvested and western blots were performed. VS-6063 treatment inhibited the phosphorylation of Paxillin, p42/44, AKT and GSK3β. We also observed a downregulation of the total protein levels of Cyclin D1 and c-MYC after 15 min and an increase in cleaved caspase-3 after 8 h (Figure 5A). In each cell line, even in Z138C (carrier of a TRAF2 mutation), an inhibition of the activation of the classic and alternative NF-κB signaling pathway could be observed (Figure 5B and C). There was a clear inhibition of the phosphorylation of IKKα (shown by immunoprecipitation) and IkBα as components of the classic NF-κB signaling pathway. In addition, downregulation of p52 indicated the inhibition of the alternative NF-κB signaling pathway. We confirmed the western blot results with immunofluorescence microscopy. HBL-2 and MINO were treated with VS-6063 (100 nM) for 8 h, the cells were fixed and immunofluorescence staining for p65 was performed. After VS-6063 treatment, the p65 subunit accumulated in the cytoplasma and decreased in the cell nucleus in both cell lines (Figure 5D).

FAK inhibition results in suppression of targeted cell invasion One major escape mechanism of lymphoma cells treated with chemotherapy is homing to the BM niche. For this reason, we analyzed whether treatment with VS-6063

B A

C Figure 3. Co-culture with bone marrow stromal cells (BMSC) activates FAK signaling in mantle cell lymphoma (MCL) cell lines and primary MCL cells. Western blot analysis was performed 8 hours (h), 24 h, 48 h and 72 h after co-culture with BMSCs. An increase in phosphorylation of FAK, AKT, p42/44 and Paxcillin, as well as an increase in expression level of FAK and Cyclin D1 could be observed in Z138C and JEKO (A) and primary MCL cells (P1 and P2) (B). In addition, activation of the classic and alternative NF-κB signaling pathway (phospho-IκBα, phospho-IκKα and p52) could be detected (B and C), shown by western blot analysis and immunoprecipitation (IP).

120

haematologica | 2018; 103(1)


Focal adhesion kinase in MCL

could inhibit the targeted migration of MCL cells. VS-6063 treatment (100 nM) inhibited the invasion of the MCL cell line MINO, HBL-2, JEKO and Z138C towards medium supplemented with rhCXCL12. After 24 h almost no invasion could be observed in the VS-6063-treated group, whereas untreated cells showed a high invasion towards rhCXCL12 (P=0.0001) (Figure 6).

FAK inhibition acts highly synergistically with ibrutinib treatment In the treatment of refractory MCL, promising results can be achieved with ibrutinib. However, activation of the alternative NF-κB signaling pathway can lead to primary or secondary resistance to ibrutinib. As FAK inhibition resulted in the suppression of the classic and alternative NF-κB signaling pathway, we questioned whether resistance to ibrutinib could be overcome with VS-6063 treatment, and whether ibrutinib and VS-6063 treatment are synergistic. We chose four MCL cell lines whose resistance to ibrutinib is known: JEKO and MINO (which are responsive to ibrutinib), Z138C (with a TRAF2 mutation), and HBL-2 (which are resistant to ibrutinib).5 Cells co-cultured with BMSCs were treated with escalating doses of ibrutinib and VS-6063 (0-1 µM) for 48 h and MTT-assays were performed. We observed a highly synergistic effect in all four MCL cell lines with CI values less than 0.6 (Figure 7A) (ED50, ED75, ED90) and resistance to ibrutinib could be overcome by combination treatment.

A

Combination of VS-6063 and ibrutinib leads to complete abrogation of the NF-κB signaling pathway We observed a high synergistic effect by combination treatment (ibrutinib and VS-6063) of different MCL cell lines. As ibrutinib and VS-6063 can both inhibit the classic and alternative NF-κB signaling pathway, we performed western blot analysis to study the effect on this pathway by the various treatments (VS-6063 or ibrutinib alone (100 nM) or in combination (10 nM). After 48 h of single or combination treatment, western blot assays were performed (Figure 7B). As expected, single treatment with ibrutinib or VS-6063 resulted in the downregulation of the phosphorylation of IkBα, AKT and p42/44, whereas combination treatment resulted in complete abrogation of the phosphorylation. In addition, treatment with VS-6063 alone or in combination with ibrutinib led to the downregulation or complete loss of p52 expression. Therefore, combination treatment resulted in the complete inhibition of the classic and alternative NF-κB signaling pathways.

Discussion Mantle cell lymphoma and other lymphoma subtypes frequently spread to the BM, and stromal interactions often lead to enhanced survival and drug resistance.16,25-27 Therefore, targeting deregulated kinases activated by the microenvironment has emerged as a promising strategy

B

Figure 4. Focal adhesion kinase (FAK) silencing suppresses multiple signaling pathways and FAK is essential for CXCL12 induced activation of several downstream targets. A tet-inducible miroRNA system was used to functionally knock-down FAK by treatment with doxycyclin (Dox). Western blot analysis of HBL-2 and MINO was performed 72 or 96 hours after addition of doxycyclin and compared to basal levels without doxycyclin (co) and microRNA with a scramble sequence (scr). A clear downregulation of FAK and multiple downstream targets (phospho-Paxicillin, phospho-p42/44, phospho-IκBα, p52, phospho-AKT, phospho-GSK3β and Cyclin D1) could be observed, whereas PYK2 levels remained stable (A). In addition, pre-incubation of HBL-2 with doxycyclin resulted in blockage of the CXCL12-induced-upregulation of phospho-p42/44, phospho-AKT, phospho-IkBα and p52, as shown by western blot analysis (B).

haematologica | 2018; 103(1)

121


M. Rudelius et al. A

B

C

Figure 5. Focal adhesion kinase (FAK) inhibition results in inhibition of multiple downstream targets including the classic and alternative NF-κB pathway. Western blot analysis and immunoprecipitation (IP) (A-C) was performed 15 minutes, 8 hours (h) and 24 h after treatment with VS6063 of HBL-2, MINO or Z138C co-cultured with bone marrow stromal cells. In all cell lines, an abrogation or downregulation of the phosphorylation of IκBα and IKKα accompanied by a decrease in p52 could be observed. Actin is shown as loading control. In addition, immunofluorescence microscopy (1000x magnification) with p65 (red fluorescence, Cy3) revealed a cytoplasmic retention of p65 after VS-6063 treatment in HBl-2 and MINO (D). For fluorescence microscopy a ZEISS Apotome microscope, ZEISS Z1 camera and ZEISS acquisition software was used.

D

A

B

Figure 6. Focal adhesion kinase (FAK) inhibition suppresses targeted cell invasion. Invasion assays were performed with rhCXCL12 directed invasion. Invasion assays without (co) or with treatment with VS-6063 (VS6063), for MINO, HBL-2, JEKO and Z138C are shown (A). After 24 hours, directed invasion could be observed in the untreated controls (blue bars) whereas treatment with VS-6063 (green bars) significantly inhibited invasion with P-values of P<0.0001, as calculated by Student t-test (B).

P<0.0001*** P<0.0001***

122

P<0.0001***

P<0.0001***

haematologica | 2018; 103(1)


Focal adhesion kinase in MCL

for the treatment of lymphomas. It is well known that integrin-mediated signaling cascades play important roles in cell adhesion and interaction with the microenvironment.28-30 FAK kinase is activated in response to chemokines, cytokines or ligand-binding, and plays a key role in integrin signaling.8,9,11,31 Here, we show that FAK was highly expressed in BM infiltrates of MCLs with only weak to moderate expression in lymph node infiltrates and moderate to high expression in MCL cell lines, consistent with previous studies demonstrating weak or moderate expression of FAK in the majority (63%) of lymph node infiltrates of MCLs.32 Earlier studies of MCLs showed that FAK could be activated by BCR or Hedgehog-signaling.33,34 A recently published paper35 demonstrated that FAK expression in MCL is regulated by SOX11, and FAK expression leads to invasion of MCL and homing to the BM. As we observed high FAK expression, especially in BM infiltrates, we investigated whether FAK could also be activated by stromal interaction and integrin signaling in MCL. Indeed, FAK activation and expression could be induced in MCL by incubation with rhCXCL12 or co-culture with BMSC. Incubation with rhCXCL12 or co-culture with BMSCs resulted in a significant increase in FAK, and in an increase in the phosphorylation of the autophosphorylation site of FAK and an increase in the phosphorylation of Paxillin, a direct downstream target of FAK. In addition to cell adhesion and cytoskeletal reorganization, FAK mediates proliferation and survival signals.8,9,11,36,37 Here, we show, that FAK activation by rhCXCL12 or co-culture with BMSCs leads to the activation of AKT and p42/44, two kinases that are important for cell proliferation and survival. In addition, the classic and alternative NF-κB pathway was activated, as shown by the phosphorylation of pIkBα, IKKα and by an increase in p52 expression. Previous studies in MCL, CLL and solid tumors emphasize the importance of the NF-κB signaling pathway in stromal-lymphoma interactions, supporting long-term expansion and drug resistance.7,38-40 In our experiments, FAK was essential for the activation of multiple downstream targets, including the classic and alternative

A

NF-κB signaling cascade, as FAK silencing resulted in a nearly complete abrogation of the activation of AKT, p44/42, pIkBα, and IKKα or in downregulation of p52. This is in line with Balsas et al.35 who were also able to demonstrate a blockage of the PI3K/AKT pathway after FAK inhibition. These data are clinically highly relevant, given that there are several FAK inhibitors22,24 currently being tested in patients. We chose the 2nd-generation inhibitor VS-6063 to perform the inhibition experiments, as this targets the FAK kinase domain with a better pharmacodynamics profile than first-line inhibitors. In MCL, enhanced activation of several downstream targets (AKT, c-Myc and p42/44) by co-culture with BMSCs could be blocked by the inhibition of FAK with VS-6063. In addition, the downregulation of Cyclin D1, which is up-regulated in MCLs by the t(11;14)(q13;q32) translocation could be achieved after 824 h. This effect could be due to reactivation of GSK3β kinase, which targets Cyclin D1 for degradation41-43 After 24 h, we observed an increase in cleaved caspase-3 as an indicator that cells underwent apoptosis, which makes FAK an attractive target in MCLs. As previously shown in combination with sorafenib treatment,33 inhibition of FAK also resulted in a dramatic decrease in targeted cell invasion, underlining its key role in tissue microenvironmental regulation of lymphoma dissemination to the BM. Mesenchymal stromal cells protect MCL from apoptosis through the activation of the classic and alternative NF-κB signaling pathways.7 These pathways have gained great attention as promising new therapeutic targets in lymphoma with activated BCR/NF-κB pathway,44-47 and with available BTK-inhibitors such as ibrutinib, a significant efficacy could be achieved in the treatment of refractory or relapsed MCL.2,3,48 However, somatic mutations in NF-κB regulatory genes can confer resistance to ibrutinib treatment in MCLs.5,49 Interestingly, FAK inhibition resulted in the inhibition of the classic and alternative NF-κB signaling pathways not only in the MCL cell lines MINO and HBL2, but also in Z138C, which has a known TRAF2 mutation and is characterized by the upregulation of p52.5

B

Figure 7. Focal adhesion kinase (FAK) inhibition acts highly synergistic with ibrutinib treatment. JEKO, HBL-2, MINO and Z138C co-cultured with bone marrow stromal cells (BMSCs) were treated with escalating doses of ibrutinib and VS-6063 alone or in combination, and MTT-assays were performed after 48 hours. Combination index (CI) values calculated were less than 0.6 indicating a high synergistic effect for all cell lines. In addition, western blot analysis after 48 hours of single treatment with ibrutinib or VS-6063 demonstrates decreased phosphorylation of downstream targets (AKT, p42/44, IkKα and IKKα) and complete inhibition by combination treatment.

haematologica | 2018; 103(1)

123


M. Rudelius et al.

This result could be achieved by abrogation of the phosphorylation of IKKα, which has been previously described as a substrate of FAK.50 IKKα is a component of both the classic and the alternative NF-κB pathway by functioning within an IkB kinase complex to directly phosphorylate the negative regulator IkB and by facilitating the cleavage of p100 to p52/Rel B.20,21 Combined treatment with ibrutinib and FAK inhibition turned out to be highly synergistic, and ibrutinib resistance in HBL2 and Z138C could be overcome by complete inhibition of the alternative and classic NF-κB signaling pathway. This supports previously published data35 demonstrating that FAK confers cell-adhesion-mediated drug resistance and contributes to a more aggressive phenotype.

References 1. Dreyling M, Ferrero S. The role of targeted treatment in mantle cell lymphoma: is transplant dead or alive? Haematologica. 2016;101(2):104-114. 2. Dreyling M, Jurczak W, Jerkeman M, 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):770-778. 3. Wang ML, Rule S, Martin P, Goy A, et al. Targeting BTK with Ibrutinib in relapsed or refractory mantle-cell lymphoma. N Engl J Med. 2013;369 (6):507-516. 4. Martin P, Maddocks K, Leonard JP, et al. Postibrutinib outcomes in patients with mantle cell lymphoma. Blood. 2016; 127(12):1559-1563. 5. Rahal R, Frick M, Romero R, et al. Pharmacological and genomic profiling identifies NF-κB-trageted treatment strategies for mantle cell lymphoma. Nat Med. 2014;20(1):87-92. 6. Chiron D, Di Liberto M, Martin P, et al. Cellcycle reprogramming for PI3K inhibition overrides a relapse-specific C481S BTK mutation revealed by longitudinal functional genomics in mantle cell lymphoma. Cancer Discov. 2014;4(9):1022-1035. 7. Medina DJ, Goodell L, Glod J, et al. Mesenchymal stromal cells protect mantle cell lymphoma cells from spontaneous and drug-induced apoptosis through secretion of B-cell activating factor and activation of the canonical and non-canonical nulear factor κB pathways. Hematologica. 2012; 97(8):1255-1263. 8. Desgrosellier JS, Cheresh DA. Integrins in cancer: biological implications and therapeutic opportunities. Nat Rev Cancer. 2010;10(1):9-22. 9. Guan JL. Integrin signaling through FAK in the regulation of mammary stem cells and breast cancer. IUBMB Life. 2010;62(4):268276. 10. Mitra SK, Schlaepfer DD. Integrin-regulated FAK-Src signaling in normal and cancer cells. Curr Opin Cell Biol. 2006;18(5):516-523. 11. Schaller MD. Cellular functions of FAK kinases: insight into molecular mechanisms and novel functions. J Cell Sci. 2010; 123(7):1007-1013. 12. Stühmer T, Zöllinger A, Siegmund D, et al. Signaling profile and antitumour activity of

124

13.

14.

15.

16.

17.

18.

19.

20.

21.

22. 23.

24.

In conclusion, our results provide evidence that FAK modulates the migratory and prosurvival signals mediated by the microenvironment in MCL. Furthermore, FAK inhibition, especially in combination with ibrutinib, may represent a promising approach to treat patients with advanced MCL presenting with BM involvement. Acknowledgments The excellent technical assistance of Sabine Roth and Barbara Leibbrandt is gratefully acknowledged. Funding This work was supported in part through a grant to M.T.R. (DFG-FOR2314) from the German Research Foundation (DFG).

the novel Hsp90 inhibitor NVP-AUY922 in multiple myeloma. Leukemia. 2008; 22(8):1604-1612. Fellmann C, Hoffmann T, Sridhar V, et al. An optimized microRNA backbone for effective single-copy RNAi. Cell Rep. 2013;5(6):1704-1713. Chou TC, Talalay P. Quantitative analysis of dose-effect relationships: The combined effects of multiple drugs or enzyme inhibitors. Adv Enzyme Regul. 1984;22:2755. Hanks SK, Calalb M, Harper MC, et al. Focal adhesion protein-tyrosine phosphorylated in response to cell attachment to fibronectin. Proc Natl Acad Sci USA. 1992; 89(18):8487-8491. Kurtova AV, Tamayo A, Ford RJ, et al. Mantle cell lymphoma cells express high levels of CXCR4, CXCR5, and VLA-4 (CD49d): importance for interactions with the stromal microenvironment and specific targeting. Blood. 2009;113(19):4604-4613. Gabarra-Niecko V, Schaller MD, Dunty JM. FAK regulates biological processes important for the pathogenesis of cancer. Cancer Metastasis Rev. 2003;22(4):359-374. Alsayed Y, Ngo H, Rummels J, et al. Mechanisms of regulation of CXCR4/SDF1 (CXCL12)-dependent migration and homing in multiple myeloma. Blood. 2007; 109(7):2708-2717. Jakubikova J, Cholujova D, Hideshima T, et al. A novel 3D mesenchymal stem cell model of the multiple myeloma bone marrow niche: biologic and clinical applications. Oncotarget. 2016;7(47):77326-77341. Karin M, Yamamoto Y, Wang QM. The IKK NF-kappa B system: a treasure trove for drug development. Nat Rev Drug Discov. 2004;3(1):17-21. Kendellen MF, Bradford J, Lawrence CL, et al. Canonical and non-canonical NFkappaB signaling promotes breast cancer tumor-initiating cells. Oncogene. 2014; 33(10):1297-1305. Golubovskaya VM. Targeting FAK in human cancer: from finding to first clinical trials. Front Biosci. 2014;1(19):687-706. Jones SF, Siu LL, Bendell JC, et al. A phase I study of VS-6063, a second-generation fokal adhesion kinase inhibitor, in patients with advanced solid tumors. Invest New Drugs. 2015;33(5):1100-1107. Infante JR, Camidge D, Mileshkin LR, et al. Safety, pharmacokinetic, and pharmacody-

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

namic phase I dose-escalation trial of PF00562271, an inhibitor of focal adhesion kinase, in advanced solid tumors. J Clin Oncol. 2012;30(13):1527-1533. Burger JA, Ford R. The microenvironment in mantle cell lymphoma: cellular and molecular pathways and emering targeted therapies. Semin Canc Biol. 2011;21(5):308312. Mraz M, Zent CS, Church AK, et al. Bone marrow stromal cells protect lymphoma Bcells from Rituximab-induced apoptosis and trageting integrin alfa-4-beta-1 (VLA-4) with Natalizumab can overcome its resistance. Br J Haematol. 2011;155(1):53-64. Burger JA, Ghia P, Rosenwald A, et al. The microenvironment in mature B-cell malignancies: a target for new treatment strategies. Blood. 2009;114(16):3367-3375. Faraldo MM, Deugnier M, Lukashev M, et al. Pertubation of beta 1-integrin function alters the development of murine mammary gland. EMBO J. 1998;17(8):2139-2147. Yamada KM, Aota S, Akiyama SK, et al. Mechanisms of fibronectin and integrin functions during cell adhesion and migration. Cold Spring Harb Symp Quant Biol. 1992;57:203-212. Jones JI, Doerr M, Clemmons DR. Cell migration: interactions among integrins, IGFs and IGFBPs. Prog Growth Factor Res. 1995;6(2-4):319-327. Shen Y, Schaller MD. Focal adhesion targeting: the critical determinat of FAK regulation and substrate phosphorylation. Mol Biol Cell. 1999;10(8):2507-2518. Ozkal S, Paterson J, Tedoldi S, et al. Focal adhesion kinase (FAK) expression in normal and neoplastic lymphoid tissues. Pathol Res Pract. 2009;205(11):781-788. Xargay-Torrent S, Lopez-Guerra M, Montraveta A., et al. Sorafenib inhibits cell migration and stroma-mediated bortezomib resistance by interfering B-cell receptor signaling and protein translation in mantle cell lymphoma. Clin Cancer Res. 2013;19(3):586-597. Zhang H, Chen Z, Neelapu SS, et al. Hedgehog inhibitors selectively target cell migration and adhesion of mantle cell lymphoma in bone marrow microenvironment. Oncotarget. 2016;7(12):14350-14365. Balsas P, Palomero J, Eguileor A, et al. SOX11 promotes tumor protective microenvironment interactions through CXCR4 and FAK regulation in mantle cell

haematologica | 2018; 103(1)


Focal adhesion kinase in MCL

lymphoma. Blood. 2017;130(4):501-513. 36. You D, Xin J, Volk A, Wei W, et al. FAK mediates a compensatory survival signal parallel to PI3K-AKT in PTEN-null T-ALL cells. Cell Rep. 2015;10(12):2055-2068. 37. Itoh S, Maeda M, Shimada S, et al. Role of expression of focal adhesion kinase in progression of hepatocellular carcinoma. Cancer Res. 2004;10(8):2812-2807. 38. Lutzny G, Kocher T, Schmidt-Supprian M, et al. Protein kinase c-ß-dependent activation of NF-kB in stromal cells is indispensable for the survival of chronic lymphocytic leukemia B cells in vivo. Cancer Cell. 2013; 23(1):77-92. 39. Wu XB, Liu Y, Wang GH, et al. Mesenchymal stem cells promote colorectal cancer progression through AMPK/mTOR- mediated NF-kB activation. Sci Rep. 2016;19(6):21420. 40. Koliaraki V, Pasparakis M, Kollias G. IKKß in intestinal mesenchymal cells promotes initiation of colitis-associated cancer. J Exp

haematologica | 2018; 103(1)

Med. 2015;212(13):2235-2251. 41. Rudelius M, Pittaluga S, Nishizuka S, et al. Constitutive activation of Akt contributes to the pathogenesis and survival of mantle cell lymphoma. Blood. 2006;108(5):16681676. 42. Yu XJ, Han QB, Wen ZS, et al. Gambogenic acid induces G1 arrest via GSK3ß-dependent cyclin D1 degradation and triggers autophagy in lung cancer cells. Cancer Lett. 2012;322(2):185-194. 43. Diehl JA, Cheng M, Roussel MF, et al. Glycogen synthase kinase-3beta regulates cyclin D1 proteolysis and subcellular localization. Genes Dev. 1998;12(22):3499-3511. 44. Yang Y, Shaffer AL, Emre NC, et al. Exploiting synthetic lethality for the therapy of ABC diffuse large B cell lymhoma. Cancer Cell. 2012;21(6):723-737. 45. Jares P, Colomer D, Campo E. Molecular pathogenesis of mantle cell lymphoma. J Clin Invest. 2012;122(10):3416-3423. 46. Pham LV, Tamayo AT, Yoshimura LC, et al.

47.

48.

49.

50.

Inhibition of constitutive NF-kappa B activation in mantle cell lymphoma B cells leads to induction of cell cycle arrest and apoptosis. J Immunol. 2003;171(1):88-95. Rossi D, Deaglio S, Dominguez-Sola D, et al. Alteration of BIRC3 and multiple other NF-kB pathway genes in splenic marginal zone lymphoma. Blood. 2011;118(18): 4930-4934. Wang ML, Blum KA, Martin P, et al. Longterm follow-up of MCL patients treated with single-agent ibrutinib: updated safety and efficacy results. Blood. 2015; 126(6):739-745. Bea S, Valdes-Mas R, Navarro A, et al. Landscape of somatic mutations and clonal evolution in mantle cell lymphoma. Proc Natl Acad Sci USA. 2013;110(45):1825018255. Dwyer SF, Gao L, Gelman IH. Identification of novel focal adhesion kinase substrates: role of FAK in NFkB signaling. Int J Biol Sci. 2015;11(4):404-410.

125


ARTICLE

Non-Hodgkin Lymphoma

Ferrata Storti Foundation

Histone deacetylase inhibitors downregulate CCR4 expression and decrease mogamulizumab efficacy in CCR4-positive mature T-cell lymphomas

Akihiro Kitadate,1,3 Sho Ikeda,1 Fumito Abe,1 Naoto Takahashi,1 Norio Shimizu,2 Kosei Matsue3 and Hiroyuki Tagawa1

Haematologica 2018 Volume 103(1):126-135

Department of Hematology, Nephrology, and Rheumatology, Akita University Graduate School of Medicine; 2Division of Virology and Immunology, Medical Research Institute, Tokyo Medical and Dental University, Tokyo and 3Division of Hematology/Oncology, Department of Medicine, Kameda Medical Center, Kamogawa, Japan

1

ABSTRACT

H

Correspondence: htagawa0279jp@me.com

Received: July 30, 2017. Accepted: October 10, 2017. Pre-published: October 12, 2017.

doi:10.3324/haematol.2017.177279 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/126 Š2018 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.

126

istone deacetylase inhibitors are promising agents for various Tcell lymphomas, including cutaneous T-cell lymphoma, peripheral T-cell lymphoma, and adult T-cell lymphoma/leukemia. CCR4 is an important therapeutic target molecule because mogamulizumab, an anti-CCR4 antibody, has shown promising efficacy against various T-cell lymphomas. In this study, we examined the in vitro synergistic effects of mogamulizumab and histone deacetylase inhibitors against various T-cell lymphomas. First, we examined the expression of CCR4 mRNA and surface CCR4 in various T-cell lymphoma cell lines and found that it was downregulated upon treatment with vorinostat, a pan-histone deacetylase inhibitor. Next, we used isoform-specific histone deacetylase inhibitors and short-interfering RNA to determine the histone deacetylase isoform involved in the regulation of CCR4, and demonstrated that romidepsin, a class I selective histone deacetylase inhibitor, reduced CCR4 most efficiently. Moreover, among class I histone deacetylases, histone deacetylase 2 knockdown led to a reduction of CCR4 in lymphoma cells, suggesting that CCR4 expression is mainly regulated by histone deacetylase 2. When we examined the CCR4 expression in skin samples from primary cutaneous T-cell lymphoma, obtained from the same patients before and after vorinostat treatment, we found that CCR4 expression was greatly reduced after treatment. Finally, when we conducted an antibody-dependent cellmediated cytotoxicity assay with mogamulizumab by using various lymphoma cells, we found that the efficacy of mogamulizumab was significantly reduced by pretreatment with vorinostat. Altogether, our results suggest that the primary use of histone deacetylase inhibitors before treatment with mogamulizumab might not be suitable to obtain synergistic effects. Moreover, these results have potential implications for optimal therapeutic sequences in various CCR4-positive T-cell lymphomas.

Introduction Mature T-cell neoplasms comprise approximately 20 sub-classified categories of non-Hodgkin lymphomas and are broadly divided into cutaneous T-cell lymphomas (CTCL) and peripheral T-cell lymphomas (PTCL).1-3 For instance, according to the World Health Organization (WHO) classification, PTCL includes peripheral T-cell lymphoma not otherwise specified (PTCL-NOS), angioimmunoblastic T-cell lymphoma, anaplastic large cell lymphoma (ALCL), adult T-cell leukemia/lym-

haematologica | 2018; 103(1)


HDAC inhibition decreases mogamulizumab efficacy in ML

phoma (ATLL), and others. CTCL mainly consist of mycosis fungoides and Sézary syndrome.1-3 In addition, the main mature natural killer (NK)-cell neoplasms include extranodal NK/T-cell lymphoma, nasal type and NK-cell leukemia.1-3 Combination chemotherapy, including cyclophosphamide, hydroxydoxorubicin, vincristine, and prednisone (CHOP) as well as CHOP-like regimens, has usually been the standard first-line treatment for patients with PTCL and advanced CTCL.4 Except for anaplastic lymphoma kinase-positive ALCL, however, the efficacy of these combination therapies is unsatisfactory, and most patients have a poor prognosis.5 Improvement of the survival of patients with malignant lymphoma has recently been expected on the basis of the appearance of various molecular targeted therapeutic drugs. Novel molecular targeted therapies have also been developed against T-cell and NK-cell neoplasms. Two particularly noteworthy therapies are mogamulizumab, an anti-CCR4 antibody, and histone deacetylase (HDAC) inhibitors, including vorinostat and romidepsin. These two promising therapies are currently being applied separately for the treatment of T-cell and NK-cell lymphomas. Mogamulizumab is a humanized anti-CCR4 antibody developed against ATLL that highly expresses CCR4, a chemokine receptor. Mogamulizumab prompts potent antibody-dependent cellular cytotoxicity (ADCC) activity against malignant cells.6-8 CCR4 is expressed in ATLL and in approximately 38% of PTCL.9 In addition, expression of CCR4 is promoted in CTCL with the progression of the disease.10 In recent years, mogamulizumab has been shown to be clinically effective against CCR4-positive CTCL and PTCL.11 Moreover, mogamulizumab has been shown to be effective against T-cell and NK-cell lymphomas in preclinical studies.12 Mogamulizumab is, therefore, a promising agent for the treatment of CCR4-positive T-cell and NK-cell lymphomas. Eighteen isoforms of HDAC are known.13,14 In particular, class I HDAC (HDAC1, HDAC2, and HDAC3) are considered to inhibit the transcription of tumor-suppressor genes and additional related genes (e.g., p21, miR-16).14-17 The inhibition of class I HDAC could, therefore, restore the expression of tumor suppressor genes and exert an antitumor effect.17,18 HDAC inhibitors can be classified into two types, namely pan-HDAC inhibitors and isoformspecific HDAC inhibitors. While pan-HDAC inhibitors broadly inhibit multiple HDAC, isoform-specific HDAC inhibitors target specific HDAC. The pan–HDAC inhibitor, vorinostat/suberoylanilide hydroxamic acid (SAHA), is a first-line therapy against advanced CTCL19 and HBI-8000, a new pan-HDAC inhibitor, has also been suggested from preclinical studies to be active against ATLL.20 The class I-specific HDAC inhibitor, romidepsin, has shown promising efficacy against PTCL.21 In addition, a novel pan-HDAC inhibitor, belinostat was recently approved for use in relapsed or refractory PTCL in the USA.22 As described above, we can expect the clinical application of HDAC inhibitors in various T-cell lymphomas. The synergistic effects of molecular targeted drugs could be studied for future therapeutic strategies. The efficacy of HDAC inhibition in combination with anti-PD-1 antibodies,23 bortezomib,24 DNA methyltransferase inhibitors,25 a Bruton tyrosine kinase inhibitor,26 and other treatments has been suggested. These are expected to have a synerhaematologica | 2018; 103(1)

gistic effect through the combined use of molecular targeted drugs. However, there is a risk that molecular targeted drugs with different modes of action may adversely affect each other. Among these combinations, in this study, we clarify the effect of the combined use of mogamulizumab and HDAC inhibitors on various T-cell and NK-cell lymphomas. Based on our findings, we discuss what benefits or adverse effects might be assumed for patients if these molecular targeting agents are used in clinical practice.

Methods Primary lymphoma and control samples We collected samples from patients with ATLL (n=1), PTCLNOS (n=2), and advanced CTCL (n=6). Six samples were obtained before and after treatment of SAHA/vorinostat. For quantitative reverse transcriptase polymerase chain reaction (RT-PCR) analysis of normal control cells, CD4 positive cells were collected from healthy donors using a magnetic cell sorting system (Miltenyi Biotec, Bergisch Gladbach, Germany) or cell sorter (Dako Cytomation MoFlo, Tokyo, Japan). Samples were collected under a protocol approved by the Institutional Review Boards of Akita University.

Cell lines We used the following 15 cell lines for this study: My-La, HH, MJ, and HUT78 (CTCL cell lines), MT-1, MT-2, MT-4, and TL-Su (ATLL cell lines), SR786 and Karpass299 (K299) (ALCL cell lines) and Kai3, SNK6, HANK1, SNK10 and KHYG1 (NK-cell lymphoma/leukemia cell lines). HH, HUT78, and MJ cell lines were purchased from the American Type Tissue Collection. My-La was from the European Collection of Cell Cultures. K299 and SR786 were purchased from the Deutsche Sammlung von Mikroorganismen und Zellkulturen. KHYG1, MT-1, MT-2, MT-4, TL-Su, and Kai3 were bought from the Japanese Collection of Research Bioresources Cell Bank. HANK1 was a gift from Dr Yoshitoyo Kagami (Toyota Kosei Hospital, Japan). Cells were cultured in Arteimis-1 medium (with or without 2% inactivated human serum), which is a chemically defined, serum-free medium purchased from NihonTechno Service Co. Ltd. (Ibaraki, Japan). It contains recombinant human insulin (5.0 μg/L), recombinant human interleukin-2 (250 IU/mL), and human serum albumin (2 g/L). It does not contain any other cytokines or growth factors.

Real-time quantitative reverse transcriptase polymerase chain reaction analysis Real-time quantitative RT-PCR was performed with the Taqman method (Life Technology) using the Light Cycler 480 probe master (Roche Diagnostics, Basel, Switzerland). The Taqman probes for CCR4 (Hs00747615_s1) and GAPDH (Hs02758991_g1) were purchased from Applied Biosystems (Foster City, USA). mRNA levels were normalized to those of GAPDH and the relative level of expression of specific mRNA was presented by 2–ΔCt or 2–ΔΔCt. Quantitative stem-loop reverse transcription was then performed using a First-Strand cDNA Synthesis Kit (GE Healthcare, Buckinghamshire, UK).

Flow cytometric analyses For flow cytometric analysis, cells were stained at 4°C with fluorescein isothiocyanate-conjugated anti-human CCR4 (R&D Systems, Minneapolis, USA). The appropriate isotype controls used were unlabeled mouse IgG1 antibody (R&D Systems). After washing, cells were analyzed using a FACS Canto flow cytometer (BD Biosciences, San Jose, CA, USA). 127


A. Kitadate et al.

Immunohistochemical analysis Immunostaining for CD4 and CCR4 in paraffin-embedded blocks of primary T-cell lymphoma samples was conducted according to the protocols of the kits’ manufacturers. CONFIRM anti-CD4 (SP35) was purchased from Roche Diagnostics. The CCR4 staining kit was bought from Kyowa Medex (Tokyo, Japan).

Chemicals Vorinostat was purchased from R&D Systems. Romidepsin, CI994, RGFP966, and ricolinostat were bought from Cosmo Bio Co., Ltd. (Tokyo, Japan). PCI-34051 was purchased from Santa Cruz Biotechnology (Santa Cruz, USA).

Transient short interfering RNA transfection For transient knockdown of HDAC1 (product n. L-003493-00), HDAC2 (product n. L-003495-02), and HDAC3 (product n. L003496-00), we used Dharmacon ON-TARGET plus SMART pool short interfering (si) RNA. For a non-targeting control, we used Dharmacon ON-TARGET plus Non-targeting pool (product n. D001810-10). These were purchased from GE Health Care Japan (Tokyo, Japan). siRNA transfection was performed using the Amaxa cell optimization kit V (Amaxa, Koeln, Germany) according to Amaxa guidelines.

Antibody-dependent cell-mediated cytotoxicity assay Cell lines were used as target cells. Human peripheral blood mononuclear cells from healthy donors were used as effector cells. Target cells (2.5x103) and effector cells (1.25x105) were co-cultured in 96-well plates with mogamulizumab or solvent alone (control) for 4 h in Artemis-1 medium. After incubation, the supernatant of each well was obtained, and percentage cell death was calculated by measuring the lactate dehydrogenase concentration in the supernatant using the CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega, Madison, USA).

Cell migration assay In vitro cell migration was assayed using a CytoSelect 96-Well Cell Migration Assay kit (5 μm, Fluorometric Format) (Cell Biolabs, Inc. San Diego, CA, USA) according to the manufacturer's protocol. Migration was stimulated by CCL22 (500 ng/mL) in the lower chamber; no serum was added to the upper chamber. The incubation period was 16 h and the cell lysis buffer was transferred to a 96-well plate. Relative fluorescence units were measured by a plate reader at 480 nm/520 nm. The migration index was calculated as the number of cells migrating toward the concentration gradient of chemokines divided by the number of cells migrating toward medium only.

Western blot analysis HDAC1 (#5356), HDAC2 (#5113) and HDAC3 (#3949) were purchased from Cell Signaling (Danvers, USA). Tubulin (MS-581P0) was acquired from NeoMarkers (Fremont, CA, USA).

Statistical analysis A Student t-test was used to examine the statistical significance of the findings.

Results Vorinostat downregulates the expression of the CCR4 chemokine receptor in various T-cell lymphoma cell lines We first examined the expression of CCR4 in 15 T-cell and NK-cell lymphoma cell lines and a sample of periph128

eral blood mononuclear cells to investigate the effect of vorinostat, a pan-HDAC inhibitor, on CCR4 expression. The expression of CCR4 was analyzed by quantitative RT-PCR using the Taqman method. The chemokine receptor was expressed in most (11 out of 15) cell lines, including ATLL (MT-1, MT-2, MT-4, and TL-SU), CTCL (My-La, HH, and MJ), and NK/T-cell lymphoma cell lines (Kai3, SNK6, HANK1, and SNK10) (1.0- to 7.6-fold change). However, as previously reported, anaplastic lymphoma kinase-positive ALCL cell lines (SR786 and K299) had lower expression compared with CD4-positive T cells.27 Next, we investigated the effect of vorinostat on CCR4 expression in T-cell and NK-cell lymphoma cell lines. Since the IC50 (24 h) of vorinostat in T-cell lymphomas is 5 μM, as previously reported,16 an exposure experiment at this concentration was conducted. When quantitative RT-PCR was performed with the vorinostattreated cells, the mRNA levels of CCR4 were markedly reduced, except for those in the cell lines HUT78, KHYG1, SR786, and K299, which did not express CCR4 (Figure 1A). This result was independent of the presence or absence of a human T-cell leukemia virus type-1 infection. Next, we examined the changes in the expression of CCR4 protein using flow cytometry. Similar to the change in CCR4 mRNA, the CCR4 mean fluorescence intensity decreased markedly in the 11 CCR4-expressing cell lines after vorinostat treatment (Figure 1B,C). To investigate whether vorinostat affects the function of CCR4, we carried out in vitro CCL22-dependent chemotaxis assays. In this experiment, cell lines were plated onto the upper chamber of transwell plates, and the cell migration capability from the upper chamber to the lower chamber, where CCL22 was bound, was examined (Figure 1D). There was no fetal calf serum in either chamber. The migration was quantified on the basis of relative fluorescence units, as previously described.28 Control cells migrated from the upper to the lower chamber, but the migration was significantly suppressed by vorinostat (Figure 1D). From the foregoing outcomes, we concluded that vorinostat decreases mRNA expression and surface expression of CCR4, and as a result, suppresses the function and migration of T-cell lymphoma cells.

CCR4 expression in cutaneous T-cell lymphoma and adult T-cell leukemia/lymphoma cells is regulated by HDAC2 Given the results shown in Figure 1, we speculated that the expression of CCR4 is regulated by HDAC. Among the known 18 HDAC, class I HDAC (HDAC1, HDAC2, and HDAC3) are involved in gene transcription within the cell nucleus.29 In order to narrow down which HDAC could control CCR4 expression, we used various isoformor class-selective HDAC inhibitors. In detail, we used the following class-specific HDAC inhibitors: romidepsin as a class I-selective HDAC inhibitor,30 CI-994 as an HDAC1/HDAC2-selective inhibitor,31 and RGFP966 as an HDAC3-selective inhibitor.32 We also studied HDAC6 and HDAC8 inhibitors, which are expected to have therapeutic effects on hematopoietic tumors:33,34 ricolinostat as an HDAC6-specific inhibitor and PCI-34051 as an HDAC8specific inhibitor.35,36 When these CTCL cells were exposed to these drugs at the IC50, romidepsin and CI-994 strongly suppressed CCR4 expression (Figure 2A). These results suggest that class I HDAC might control CCR4 expression. haematologica | 2018; 103(1)


HDAC inhibition decreases mogamulizumab efficacy in ML

and elicits powerful ADCC against tumor cells.37 We previously reported that pan-HDAC inhibitors (vorinostat and panobinostat) decreased the expression of chemokine receptor CCR6/CCR6 in CTCL.17 Together with these reports, we hypothesized that HDAC inhibitors may decrease the expression of CCR4, leading to the negative effect of mogamulizumab on ADCC activity. We therefore examined whether the combination of HDAC inhibitor and mogamulizumab might affect ADCC activity against T-cell lymphomas. We conducted a mogamulizumab-induced ADCC assay against cells pretreated with vorinostat (for 24 h) (Figure 3A). Firstly, viability with mogamulizumab was confirmed by an MTT assay and there was no change in viability of cell lines (data not shown). Compared to control cells treated with dimethylsulfoxide, we found that cytotoxicity induced by mogamulizumab was significantly reduced in vorinostat-treated cells. In particular, in the cell lines in which expression of CCR4 was weakly positive (MJ, TL-Su, and SNK10), the ADCC activity of mogamulizumab almost disappeared (Figure 3B). These results indicate that when an HDAC inhibitor and mogamulizumab are used in combi-

We further performed knockdown experiments using siRNA against HDAC1, HDAC2, and HDAC3. We confirmed the suppression of expression of these proteins by western blot analysis (Figure 2B). When we compared the change in expression of CCR4 mRNA in HDAC-knockdown cells, HDAC2 knockdown cells showed the most significantly decreased expression of CCR4 mRNA (Figure 2C). Moreover, when we examined the surface expression of CCR4 in HDAC-knockdown cells by flow cytometry, HDAC2-knockdown cells showed the greatest decrease in CCR4 mean fluorescence intensity (Figure 2D). This knockdown experiment was also performed on ATLL cell lines (MT-1 and MT-4), and confirmed a significant decrease in CCR4 mean fluorescence intensity (Figure 2D). These results suggest that class I HDAC, especially HDAC2, might be deeply involved in the regulation of CCR4 expression.

Pretreatment with histone deacetylase inhibitors significantly decreased mogamulizumab-induced antibody-dependent cell-mediated cytotoxicity Mogamulizumab binds strongly to CCR4-positive cells

A

C

B

D

Figure 1. The pan-histone deacetylase inhibitor, vorinostat reduces CCR4 expression in various T-cell and natural killer- cell lymphoma cell lines. Bars indicate the mean ± standard error of the mean (SEM) of three independent experiments. Asterisks (*) indicate statistical significance: *0.01 ≤ P < 0.05, **0.001 ≤ P < 0.01, ***P < 0.001, n.s: not siginificant. (A) Quantitative RT-PCR analysis of CCR4 in vorinostat-treated T- and NK-cell lymphoma cell lines (5 μM for 24 h). The Student ttest was used to examine statistical significance. x-axis: cell lines. y-axis: 2-ΔCt values for microRNA expression. (B) Flow cytometry analysis of CCR4 in indicated T- and NK-cell lymphoma cells treated with vorinostat. Cells were stained with CCR4-FITC 24 h after vorinostat treatment (5 μM). Representative flow cytometry histograms are shown. ΔMFI (mean fluorescence intensity) values were obtained after subtraction of the isotype control, MFI from CCR4 MFI. SAHA: suberoylanilide hydroxamic acid. (C) ΔMFI of CCR4 in vorinostat-treated T- and NK-cell lymphoma cell lines (5 μM for 24 h). x-axis: cell lines. y-axis: ΔMFI of CCR4 in dimethylsulfoxide (DMSO) or vorinostat-treated T- and NK-cell lymphoma cell lines. (D) Migration assay of lymphoma cells treated with 5 μM vorinostat for 24 h. A schematic illustration of the migration assay is also shown. RFU: relative fluorescence units.

haematologica | 2018; 103(1)

129


A. Kitadate et al.

nation, the HDAC inhibitor downregulates CCR4 expression of lymphoma cells, resulting in a decrease of ADCC induced by mogamulizumab.

CCR4 expression was greatly reduced after vorinostat treatment in primary cutaneous T-cell lymphoma To examine the effect of CCR4 downregulation by HDAC inhibitors in clinical cases, we examined the CCR4 expression of CTCL skin samples, which were obtained from the same patients before and after vorinostat treatment. In patient #1, we confirmed strong concordance between CCR4 positivity and CD4-positive T cells before vorinostat treatment. However, we found that CCR4 expression was greatly reduced in the specimen at the time of relapse after 2 years of vorinostat treatment (Figure 4A). In Figure 4B, we show the clinical course of patient #1. This case showed weak positivity for CCR4 at relapse and when mogamulizumab was administered, but the response was not effective and the patient had progressive disease. Similarly, CCR4 expression was evaluated by immunohistochemical staining for six cases for which samples before and after vorinostat treatment were avail-

able (Table 1). As a result, we found that CCR4 expression decreased significantly in the specimens after vorinostat treatment (Figure 4C). Furthermore, when we examined CCR4 mRNA changes in the primary samples (patient #7: ATLL, patient #8: CTCL, and patient #9: PTCL-NOS) treated with vorinostat, we also confirmed a marked downregulation of CCR4 (Figure 4D). Significant downregulation of CCR4 was confirmed when the surface expression was analyzed in an ATLL primary sample by flow cytometry (Figure 4E). Moreover, when we conducted an ADCC assay of mogamulizumab using this primary ATLL sample, we found that the efficacy of mogamulizumab was significantly reduced by vorinostat pretreatment (Figure 4F). These results strongly suggest that ADCC of lymphoma cells could not be expected from pre-treatment with HDAC inhibitors, even in the primary sample.

Discussion HDAC inhibitors interfere with histone tail modifications, thus altering chromatin structure and epigenetically

A

B

D

130

C

Figure 2. Effect of isoform-specific histone deacetylase inhibitors and histone deacetylase knockdown on CCR4 expression. Bars indicate the mean ± standard error of the mean (SEM) of three independent experiments. Asterisks (*) indicate statistical significance: *0.01 ≤ P < 0.05, ***P < 0.001. (A) Flow cytometry analysis of CCR4 in CTCL cells (My-La) treated with various HDAC inhibitors. Cells were stained with CCR4-FITC 24 h after treatment (5 μM vorinostat, 10 nM romidepsin, 50 μM CI994, 20 μM RGFP966, 10 μM ricolinostat, and 75 μM PCI-34051). Left panel: representative flow cytometry histograms are shown. Right panel: ΔMFI of CCR4 in My-La cells treated with HDAC inhibitors. (B) Transient knockdown of HDAC (HDAC1, HDAC2, and HDAC3) in CTCL (My-La and HH) cells. Western blot analysis of HDAC1, HDAC2, and HDAC3 in cells after transient transduction of siHDAC or a scrambled control (scr). Tubulin: protein positive control. (C) Quantitative RT-PCR analysis of CCR4 after transient transfection of siHDAC and a scr in CTCL cells. x-axis: cell lines; yaxis: expression relative to control cells that were assigned a value of 1.0. (D) ΔMFI of CCR4 after transient transfection of siHDAC and scr in CTCL (My-La and HH) and ATLL (MT-1 and MT4) cells. x-axis: cell lines; y-axis: ΔMFI of CCR4. MFI: mean fluorescence intensity.

haematologica | 2018; 103(1)


HDAC inhibition decreases mogamulizumab efficacy in ML

controlled pathways. Pan-HDAC inhibitors, such as vorinostat, can restore the expression of its target molecules. HDAC inhibition directly suppresses the transcription of genes regulated by HDAC, while there are genes whose expression increases with HDAC inhibition. Mediator molecules, including various microRNA which are a direct target of HDAC inhibitors, may cause enhanced gene transcription by HDAC inhibitors. The main role of HDAC inhibition in cancer treatment is the

restoration of tumor suppressor gene expression, which is suppressed by HDAC. HDAC inhibitors can restore coding and non-coding genes, such as CDKN1A/p21 and miR-16.13,16 p21 is a transcriptional target molecule of tumor suppressor protein p53. miR-16 is a well-known tumor-suppressive microRNA.16 Restoration of these molecules induces cell growth arrest, cellular senescence, and apoptosis in lymphoma cells.13,16 In addition, we expect that the restoration of tumor suppressive genes downreg-

A

B

Figure 3. Pretreatment of vorinostat significantly decreases mogamulizumab-induced antibody-dependent cell-mediated cytotoxicity. Bars indicate the mean ± standard error of the mean (SEM) of three independent experiments. Asterisks (*) indicate statistical significance: **0.001 ≤ P < 0.01, ***P < 0.001, n.s: not siginificant. (A) A schematic illustration of the ADCC assay is shown. Cells were treated with 5 μM vorinostat for 24 h or dimethylsulfoxide (DMSO) as a control before mogamulizumab treatment. Cytotoxicity was measured using the lactate dehydrogenase assay in the presence of effector cells (peripheral blood mononuclear cells. PBMC) obtained from healthy volunteers and mogamulizumab (10 mg/mL) or the same volume of solvent (control). The ratio of target:effector cells was fixed at 1:50. (B) The ADCC assay against CTCL (upper panel: MyLa, HH, MJ, and HUT78), ATLL (middle panel: MT-1, MT-2, MT-4, and TLSu), and NK/T-cell lymphoma (lower panel: SNK6, Kai3, HANK1, and SNK10) cell lines. x-axis: cell lines; y-axis: percent cell lysis.

haematologica | 2018; 103(1)

131


A. Kitadate et al.

ulates oncogenes and the translation of proteins. For example, BMI1/Bmi-1, a proto-oncogene, is suppressed by the restoration of miR-16.16,38 Thus, the role of HDAC inhibitors in eradicating cancer cells involves restoring tumor suppressive genes by downregulation of oncogenes. However, considering the effects of HDAC inhibition on chemokines, chemokine receptors, and cell surface antigens, its respective tumor-suppressive and tumor-promoting aspects must be analyzed. When considering the anti-tumor effects, enhancement of the expression of chemokine receptors or chemokines may be desired in some cases, while their suppression may be desirable in other cases. As an example of desirable enhanced expression, Zheng et al. recently showed that HDAC inhibition can enhance T-cell infiltration and T-cell dependent tumor

regression by increasing the expression of CCL5, CXCL9, and CXCL10, thereby augmenting the immune effect of anti-PD-1.23 As an example of desirable reduced expression by HDAC inhibition, we recently reported that HDAC inhibition decreased CCR6 expression via upregulation of miR-150 and consequently inhibited multiple metastases of CTCL cells.17 However, as we have shown in this study, HDAC inhibitors also decrease the expression of CCR4. As a result, mogamulizumab-inducible ADCC was remarkably decreased, and a synergistic effect between mogamulizumab and HDAC inhibition could not be confirmed. A similar synergistic effect was expected from combining HDAC inhibitors with antibodies for cell surface antigens, but is not advisable. For example, Hasanali et al. previously showed that vorinostat suppresses CD30 expression and attenuates the efficacy of

A

B

D

132

C

E

F

Figure 4. CCR4 expression of clinical lymphoma samples before and after vorinostat treatment. Bars indicate the mean ± standard error of the mean (SEM) of three independent experiments. Asterisks (*) indicate statistical significance: *0.01 ≤ P < 0.05, **0.001 ≤ P < 0.01, ***P < 0.001, n.s: not siginificant. (A) CCR4 expression determined by immunohistochemical staining before and after vorinostat treatment. Representative results from samples (skin and lymph nodes) from patient with mycosis fungoides (MF) are shown. Specimens were stained with hematoxylin and eosin (HE), CD4, and CCR4 for a pathological diagnosis and confirmation: 200x magnification. LN: lymph node. (B) Clinical time course in patient (Pt) #1 with MF before and after vorinostat treatment. (C) CCR4-positive lymphocytes (% of total infiltrating lymphocytes) in specimens from six patients (Pt #1 - 6) before and after vorinostat treatment. The Student t-test was used to examine statistical significance. (D) Quantitative RT-PCR analysis of CCR4 in samples of primary T-cell lymphomas (ATLL, CTCL, and PTCL-NOS) (Pt #7 9) exposed to 5 μM vorinostat for 24 h. (E) ΔMFI of CCR4 in a sample of primary ATLL (Pt #7) treated with 5 μM vorinostat for 24 h. (F) ADCC assay against a sample of primary ATLL. Cells were treated with 5 μM vorinostat for 24 h or control cells treated with dimethylsulfoxide (DMSO) before mogamulizumab treatment. Cytotoxicity was measured using the lactate dehydrogenase assay in peripheral blood mononuclear cells obtained from healthy volunteers and mogamulizumab (10 mg/mL) or the same volume of solvent (control). The ratio of target: effector cells was fixed at 1:50.

haematologica | 2018; 103(1)


HDAC inhibition decreases mogamulizumab efficacy in ML

the anti-CD30, brentuximab vedotin, in ALCL.39 Thus, when expecting a synergistic effect from HDAC inhibition and other molecular targeted drugs, the influence of the target molecule of HDAC inhibitors must be examined sufficiently. Accumulated knowledge of the targets of HDAC inhibition could lead us to establish effective strategies for the administration order of molecular targeting agents. For example, CCR4 expression of a vorinostat-treated clinical sample from patient #1 recovered slightly 2 weeks after completion of treatment. This might be because of the recovery of epigenetically repressed CCR4 expression. Although the diminished expression of CCR4 by vorinostat was epigenetic, it appeared to be reversible.40 To test this notion, we monitored CCR4 expression of CTCL cell lines at specified time points (24, 48, and 72 h) after vorinostat treatment (5 μM, 24 h). Following the removal

of vorinostat, CCR4 expression returned to pretreatment levels after 72 h (data not shown). However, it is unknown how long CCR4 expression would take to recover after long-term vorinostat exposure in vivo. The restoration of CCR4 expression to its original level after vorinostat treatment completion may require some time. For patient #1, we could not exclude the possibility that the ADCC of effector cells was attenuated by five cycles of CHOP therapy. In addition, mogamulizumab has been shown to decrease CCR4-positive regulatory T cells and increase CD8-positive T cells and NK-cell numbers.41 The effect on such tumor microenvironments is also an important factor and the positivity of CCR4 alone may not, therefore, determine the response to mogamulizumab. Nevertheless, at least, it may be beneficial for patients to use mogamulizumab followed by vorinostat as a treatment strategy for CCR4-positive T-cell lymphomas.

Table 1. Information about the patients with T-cell lymphoma.

Case

Age

Sex Primary diagnosis

Pt #1

58

M

Pt #2

62

M

Biopsy

Stage CCR4+ Treatment at biopsy T cells and therapy in lesions (%)

Mycosis fungoides Skin IV Skin (after vorinostat) IV LN (after vorinostat) IV

95 5 15

nbUVB→ vorinostat→ CHOP→ mogamulizumab

24

Under treatment

-

2 weeks

Mycosis fungoides

60

nbUVB→ vorinostat→ CHOP

-

-

Skin

III A

Skin III A (after vorinostat) Pt #3

Pt #4

Pt #5

41

68

82

M

F

F

Duration Post-vorinostat Clinical Immunophenotype of vorinostat treatment response to (months) interval at mogamulizumab re-biopsy

Mycosis fungoides

Skin

IV

20 5

Skin (after vorinostat)

IV

5

LN

IV

55

PTCL-NOS

LN (after vorinostat)

IV

10

Skin

IV

30

Mycosis fungoides

nbUVB→ vorinostat→ CHOP gemcitabine CHOP→ GDP→vorinostat gemcitabine mogamulizumab

12

1 week

-

-

20

2 weeks

-

-

IV

5

IV

45

CD3(+)CD4(+)CD5(+) CD7(-)CD8(-)CD30(-) ALK(-)PD1(-)EBER(-)

-

CD3(+)CD4(+)CD5(-) CD7(-)CD8(-)CD30(-) ALK(-)PD1(-)EBER(-)

-

CD3(+)CD4(+)CD5(-) CD7(-)CD8(-)CD30(-) ALK(-)PD1(-)EBER(-)

PD 3

1 week

-

-

Vorinostat Skin

PD

6

Under treatment

-

-

5

Under treatment

-

-

-

PR

Other

CD3(+)CD4(+)CD8(-) CD20(+)CD30(-)CD79a(-)

CD3(+)CD4(+)CD5(-) CD7(-)CD8(-)CD30(-) ALK(-)PD1(-)EBER(-)

(after vorinostat) Pt #6

67

M

Mycosis fungoides

Skin

Interferon-γ, PUVA→radiation, nbUVB→vorinostat

Skin

IV

10

Pt #7

65

F

Adult T-cell lymphoma

Pleural effusion

IV

95

Pt #8

78

M

Mycosis fungoides

LN

IV

60 nbUVB→PUVA→CHOP

-

-

-

CD3(+)CD4(+)CD5(+) CD8(-)CD30(-)ALK(-)

Pt #9

67

M

PTCL-NOS

LN (at relapse)

IV

40

-

-

-

CD3(+)CD4(+)CD5(+) CD8(-)CD30(+)EBER(-)

Dose-intensified chemotherapy→ mogamulizumab

CHOP→GDP→ camptothecin 11→ brentuximab vedotin

-

CD3(+)CD4(+)CD5(+) CD8(-)CD30(-)ALK(-) CD3(+)CD4(+)CD5(+) Monoclonal CD8(-)CD25(+)CD30(-) integration ALK(-)Foxp3(+) of HTLV-1 proviral DNA was confirmed

LN: lymph node; nbUVB: narrow-band ultraviolet beta wave therapy; CHOP: cyclophosphamide, doxorubicin, vincristine, and prednisone therapy; PD: progressive disease; GDP: gemcitabine, dexamethasone, and cisplatin therapy; PR: partial response. M: male; F: female; CCR4-positive cell numbers as a percentage of total infiltrating lymphocytes.

haematologica | 2018; 103(1)

133


A. Kitadate et al.

In this study, we showed that HDAC2 mainly regulates CCR4 expression. Avoiding unnecessary inhibition of target genes by HDAC inhibitors may be important for the reduction of side effects. The development and clinical application of isoform-specific HDAC inhibitors are progressing.42 Romidepsin, a class I-specific HDAC inhibitor has been shown to be clinically effective against PTCL.21 Currently, however, HDAC inhibitors that can be used against malignant lymphomas, including vorinostat, romdepsin, and belinostat, all suppress HDAC2. Consequently, the suppression of CCR4 by class-specific or pan-HDAC inhibitors is currently inevitable. In contrast, the HDAC6-selective inhibitor, ricolinostat, has potential for therapeutic application against multiple myeloma.43 It has also been shown that PCI-34051, an HDAC8-specific inhibitor, induces apoptosis specifically in T-cell lymphomas.36 Because these HDAC inhibitors did not suppress CCR4 expression in our study, their clinical efficacy in lymphomas should be examined for future clinical trials. In addition, because HDAC inhibitors directly restore target gene expression, they might decrease CCR4 expression by some undetermined mediating molecules

References 1. WHO Classification of Tumours of Haematopoietic and Lymphoid tissues 2008. 4th ed. Lyon, France. International Agency for Research on Cancer (IARC). 2. Swerdlow SH, Campo E, Pileri SA, et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390. 3. O'Leary HM, Savage KJ. Update on the World Health Organization classification of peripheral T-cell lymphomas. Curr Hematol Malig Rep. 2009;4(4): 227-235. 4. Savage KJ. Therapies for peripheral T-cell lymphomas. Hematology Am Soc Hematol Educ Program. 2011;2011:515-524. 5. Vose J, Armitage J, Weisenburger D. International peripheral T-cell and natural killer/T-cell lymphoma study: pathology findings and clinical outcomes. J Clin Oncol. 2008;26(25):4124-4130. 6. Ishii T, Ishida T, Utsunomiya A, et al. Defucosylated humanized anti-CCR4 monoclonal antibody KW-0761 as a novel immunotherapeutic agent for adult T-cell leukemia/lymphoma. Clin Cancer Res. 2010;16(5):1520-1531. 7. Yamamoto K, Utsunomiya A, Tobinai K, et al. Phase I study of KW-0761, a defucosylated humanized anti-CCR4 antibody, in relapsed patients with adult T-cell leukemialymphoma and peripheral T-cell lymphoma. J Clin Oncol. 2010;28(9):1591-1598. 8. Ishida T, Joh T, Uike N, et al. Defucosylated anti-CCR4 monoclonal antibody (KW-0761) for relapsed adult T-cell leukemia-lymphoma: a multicenter phase II study. J Clin Oncol. 2012;30(8):837-842. 9. Ishida T, Inagaki H, Utsunomiya A, et al. CXC chemokine receptor 3 and CC chemokine receptor 4 expression in T-cell and NK-cell lymphomas with special reference to clinicopathological significance for peripheral T-cell lymphoma, unspecified. Clin Cancer Res. 2004;10(16):5494-5500.

134

such as microRNA. Alternatively, the decrease may be mediated by undetermined transcription factor(s) that regulate CCR4 expression, although we did not determine these factors. Further research is required to elucidate the specific mediator. In summary, our results suggest that the use of HDAC inhibitors before mogamulizumab against CTCL and PTCL might reduce the benefits of mogamulizumab in patients. Conversely, mogamulizumab followed by an HDAC inhibitor may be effective. In developing a therapeutic strategy, the effect of HDAC inhibitors on combined molecular targeted drugs directly affects the benefits that patients gain. Acknowledgments The authors would like to thank Ms. Hiromi Kataho, Yuko Chiba, and Yukiko Abe for their outstanding technical assistance. We thank Drs K. Teshima, H. Oyagi, M. Kume (Hiraka General Hospital), Y. Abe & K. Narita (Kameda Medical Center) for collecting primary samples. This work was supported by JSPS KAKENHI [Grant-in-Aid for Scientific Research (with funds to AK & HT)].

10. Yagi H, Seo N, Ohshima A, et al. Chemokine receptor expression in cutaneous T cell and NK/T-cell lymphomas: immunohistochemical staining and in vitro chemotactic assay. Am J Surg Pathol. 2006;30(9):1111-1119. 11. Ogura M, Ishida T, Hatake K, et al. Multicenter phase II study of mogamulizumab (KW-0761), a defucosylated anti-cc chemokine receptor 4 antibody, in patients with relapsed peripheral T-cell lymphoma and cutaneous T-cell lymphoma. J Clin Oncol. 2014;32(11):1157-1163. 12. Kanazawa T, Hiramatsu Y, Iwata S, et al. Anti-CCR4 monoclonal antibody mogamulizumab for the treatment of EBV-associated T- and NK-cell lymphoproliferative diseases. Clin Cancer Res. 2014;20(19):50755084. 13. Richon VM, Sandhoff TW, Rifkind RA, Marks PA. Histone deacetylase inhibitor selectively induces p21WAF1 expression and gene-associated histone acetylation. Proc Natl Acad Sci USA. 2000;97(18):1001410019. 14. Lane AA, Chabner BA. Histone deacetylase inhibitors in cancer therapy. J Clin Oncol. 2009; 27(32): 5459-5468. 15. Matthews GM, Mehdipour P, Cluse LA, et al. Functional-genetic dissection of HDAC dependencies in mouse lymphoid and myeloid malignancies. Blood. 2015;126(21): 2392-2403. 16. Kitadate A, Ikeda S, Teshima K, et al. MicroRNA-16 mediates the regulation of a senescence-apoptosis switch in cutaneous T-cell and other non-Hodgkin lymphomas. Oncogene. 2016;35(28):3692-3704. 17. Abe F, Kitadate A, Ikeda S, et al. Histone deacetylase inhibitors inhibit metastasis by restoring a tumor suppressive microRNA150 in advanced cutaneous T-cell lymphoma. Oncotarget. 2017;8(5):7572-7585. 18. Ju R, Muller MT. Histone deacetylase inhibitors activate p21(WAF1) expression via ATM. Cancer Res. 2003;63(11):28912897.

19. Duvic M, Talpur R, Ni X, et al. Phase 2 trial of oral vorinostat (suberoylanilide hydroxamic acid, SAHA) for refractory cutaneous T-cell lymphoma (CTCL). Blood. 2007;109 (1):31-39. 20. Hasegawa H, Bissonnette RP, Gillings M, et al. Induction of apoptosis by HBI-8000 in adult T-cell leukemia/lymphoma is associated with activation of Bim and NLRP3. Cancer Sci. 2016;107(8):1124-1133. 21. Coiffier B, Pro B, Prince HM, et al. Results from a pivotal, open-label, phase II study of romidepsin in relapsed or refractory peripheral T-cell lymphoma after prior systemic therapy. J Clin Oncol. 2012;30(6):631-636. 22. O'Connor OA, Horwitz S, Masszi T et al. Belinostat in patients with relapsed or refractory peripheral T-cell lymphoma: results of the pivotal phase II BELIEF (CLN19) Study. J Clin Oncol. 2015;33(23):24922499. 23. Zheng H, Zhao W, Yan C, et al. HDAC Inhibitors enhance T-cell chemokine expression and augment response to PD-1 immunotherapy in lung adenocarcinoma. Clin Cancer Res. 2016;22(16):4119-4132. 24. Heider U, Rademacher J, Lamottke B, et al. Synergistic interaction of the histone deacetylase inhibitor SAHA with the proteasome inhibitor bortezomib in cutaneous T cell lymphoma. Eur J Haematol. 2009;82(6): 440-449. 25. Rozati S, Cheng PF, Widmer DS, Fujii K, Levesque MP, Dummer R. Romidepsin and Azacitidine synergize in their epigenetic modulatory effects to induce apoptosis in CTCL. Clin Cancer Res. 2016;22(8):20202031. 26. Mondello P, Brea EJ, De Stanchina E, et al. Panobinostat acts synergistically with ibrutinib in diffuse large B cell lymphoma cells with MyD88 L265 mutations. JCI Insight. 2017;2(6):e90196. 27. Vermeer MH, Dukers DF, ten Berge RL, et al. Differential expression of thymus and activation regulated chemokine and its receptor CCR4 in nodal and cutaneous anaplastic

haematologica | 2018; 103(1)


HDAC inhibition decreases mogamulizumab efficacy in ML

28.

29.

30.

31.

32.

33.

large-cell lymphomas and Hodgkin's disease. Mod Pathol. 2002;15(8):838-844. Ito M, Teshima K, Ikeda S, et al. MicroRNA150 inhibits tumor invasion and metastasis by targeting the chemokine receptor CCR6, in advanced cutaneous T-cell lymphoma. Blood. 2014;123(10):1499-1511. Haberland M, Montgomery RL, Olson EN. The many roles of histone deacetylases in development and physiology: implications for disease and therapy. Nat Rev Genet. 2009;10(1):32-42. Furumai R, Matsuyama A, Kobashi N, et al. FK228 (depsipeptide) as a natural prodrug that inhibits class I histone deacetylases. Cancer Res. 2002;62(17):4916-4921. Moradei OM, Mallais TC, Frechette S, et al. Novel aminophenyl benzamide-type histone deacetylase inhibitors with enhanced potency and selectivity. J Med Chem. 2007;50(23):5543-5546. Malvaez M, McQuown SC, Rogge GA, et al. HDAC3-selective inhibitor enhances extinction of cocaine-seeking behavior in a persistent manner. Proc Natl Acad Sci USA. 2013;110(7):2647-2652. Hideshima T, Qi J, Paranal RM, et al. Discovery of selective small-molecule

haematologica | 2018; 103(1)

34.

35.

36.

37. 38.

HDAC6 inhibitor for overcoming proteasome inhibitor resistance in multiple myeloma. Proc Natl Acad Sci USA. 2016;113(46): 13162-13167. Higuchi T, Nakayama T, Arao T, Nishio K, Yoshie O. SOX4 is a direct target gene of FRA-2 and induces expression of HDAC8 in adult T-cell leukemia/lymphoma. Blood. 2013;121(18):3640-3649. Santo L, Hideshima T, Kung AL, et al. Preclinical activity, pharmacodynamic, and pharmacokinetic properties of a selective HDAC6 inhibitor, ACY-1215, in combination with bortezomib in multiple myeloma. Blood. 2012;119(11):2579-2589. Balasubramanian S, Ramos J, Luo W, Sirisawad M, Verner E, Buggy JJ. A novel histone deacetylase 8 (HDAC8)-specific inhibitor PCI-34051 induces apoptosis in Tcell lymphomas. Leukemia. 2008;22(5): 1026-1034. Ishida T, Ueda R. CCR4 as a novel molecular target for immunotherapy of cancer. Cancer Sci. 2006;97(11):1139-1146. Teshima K, Nara M, Watanabe A, et al. Dysregulation of BMI1 and microRNA-16 collaborate to enhance an anti-apoptotic potential in the side population of refractory

39.

40.

41.

42.

43.

mantle cell lymphoma. Oncogene. 2014;33(17):2191-2203. Hasanali ZS, Epner EM, Feith DJ, Loughran TP Jr, Sample CE. Vorinostat downregulates CD30 and decreases brentuximab vedotin efficacy in human lymphocytes. Mol Cancer Ther. 2014;13(12):2784-2792. Selcuklu SD, Spillane C. Translational epigenetics: clinical approaches to epigenome therapeutics for cancer. Epigenetics. 2008;3(2):107-112. Ni X, Jorgensen JL, Goswami M, et al. Reduction of regulatory T cells by mogamulizumab, a defucosylated anti-CC chemokine receptor 4 antibody, in patients with aggressive/refractory mycosis fungoides and SĂŠzary syndrome. Clin Cancer Res. 2015;21(2):274-285. Falkenberg KJ, Johnstone RW. Histone deacetylases and their inhibitors in cancer, neurological diseases and immune disorders. Nat Rev Drug Discov. 2014;13(9):673691. Yee AJ, Bensinger WI, Supko JG, et al. Ricolinostat plus lenalidomide, and dexamethasone in relapsed or refractory multiple myeloma: a multicentre phase 1b trial. Lancet Oncol. 2016;17(11):1569-1578.

135


ARTICLE

Chronic Lymphocytic Leukemia

Ferrata Storti Foundation

Microenvironmental stromal cells abrogate NF-κB inhibitor-induced apoptosis in chronic lymphocytic leukemia Carl Philipp Simon-Gabriel,1* Katharina Foerster,1* Shifa Saleem,1 Dorothee Bleckmann,1 Marco Benkisser-Petersen,1 Nicolas Thornton,1 Kazuo Umezawa,2 Sarah Decker,1 Meike Burger,3 Hendrik Veelken,4 Rainer Claus,1 Christine Dierks,1 Justus Duyster1 and Katja Zirlik1,5

Haematologica 2018 Volume 103(1):136-147

Department of Hematology, Oncology and Stem Cell Transplantation, University Medical Center, Faculty of Medicine, University of Freiburg, Germany; 2Department of Molecular Target Medicine, Aichi Medical University School of Medicine, Nagakute, Japan; 3Furtwangen University, Faculty of Medical and Life Sciences, Schwenningen Campus, Villingen-Schwenningen, Germany; 4Department of Hematology, Leiden University Medical Centre, the Netherlands and 5Tumor and Breast Center ZeTuP, St. Gallen, Switzerland 1

*CS-G and KF contributed equally to this work

ABSTRACT

uclear factor κ-light-chain-enhancer of activated B cells (NF-κB) is known to play an important role in the pathogenesis of chronic lymphocytic leukemia (CLL). Several NF-κB inhibitors were shown to successfully induce apoptosis of CLL cells in vitro. Since the microenvironment is known to be crucial for the survival of CLL cells, herein, we tested whether NF-κB inhibition may still induce apoptosis in these leukemic cells in the presence of protective stromal interaction. We used the specific NF-κB inhibitor dehydroxymethylepoxyquinomicin (DHMEQ). Microenvironmental support was mimicked by co-culturing CLL cells with bone marrow-derived stromal cell lines (HS-5 and M210B4). NF-κB inhibition by DHMEQ in CLL cells could be confirmed in both the monoculture and co-culture setting. In line with previous reports, NF-κB inhibition induced apoptosis in the monoculture setting by activating the intrinsic apoptotic pathway resulting in poly (ADP-ribose) polymerase (PARP)-cleavage; however, it was unable to induce apoptosis in leukemic cells co-cultured with stromal cells. Similarly, small interfering ribonucleic acid (siRNA)-mediated RELA downregulation induced apoptosis of CLL cells cultured alone, but not in the presence of supportive stromal cells. B-cell activating factor (BAFF) was identified as a microenvironmental messenger potentially protecting the leukemic cells from NF-κB inhibition-induced apoptosis. Finally, we show improved sensitivity of stroma-supported CLL cells to NF-κB inhibition when combining the NF-κB inhibitor with the SYK inhibitor R406 or the Bruton's tyrosine kinase (BTK) inhibitor ibrutinib, agents known to inhibit the stroma-leukemia crosstalk. We conclude that NFκB inhibitors are not promising as monotherapies in CLL, but may represent attractive therapeutic partners for ibrutinib and R406.

N Correspondence: katja.zirlik@zetaup.ch/ katja.zirlik@uniklinik-freiburg.de Received: January 26, 2017. Accepted: October 26, 2017. Pre-published: November 9, 2017. doi:10.3324/haematol.2017.165381 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/136 ©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.

136

Introduction Although progress has been made with the introduction of new therapeutic agents in the treatment of CLL, the disease remains mostly incurable, highlighting the need for new therapeutic targets and substances. NF-κB is a key factor contributing to CLL pathology and has thus been suggested as a treatment target.1-4 The five subunits of NF-κB (RELA, RELB, NFκB1, NFκB2 and c-REL) reside in the cytoplasm. Once activated, they translocate into the nucleus and bind to promotor regions on the DNA, modifying gene expression.5,6 Indeed, NF-κB is constitutively activated in CLL cells and up-regulates antiapoptotic genes (e.g., TRAF1, BCL2, BCL2L1, IAPs), increasing apoptosis resistance, a major feature of CLL cells.7,8 NF-κB signaling is also involved in increased resistance to anti-cancer therapies9 and mediates signals from crucial pathways in haematologica | 2018; 103(1)


NF-κB inhibition in CLL

the crosstalk between CLL cells and their protective microenvironment, such as the B-cell receptor (BCR) pathway, the BAFF/BAFF-receptor axis, or the CD40L/CD40 axis.7,10-12 Different mouse models also highlight the major role of NF-κB in CLL biology.1 Finally, high RELA DNA-binding is associated with advanced Binet stage, shorter time to first or subsequent treatment, and overall survival.13 Several inhibitors of NF-κB have been successfully tested on CLL cells in vitro. Amongst others (e.g., parthenolide, curcumin, IMD-0354, BAY11-7082),12,14-18 the specific NF-κB inhibitor DHMEQ efficiently induced apoptosis in CLL cells in vitro.19 However, the efficacy of NF-κB inhibitors has never been tested on CLL cells cultured in the presence of their protective microenvironment, crucial for CLL cell survival. This prompted us to assess the ability of NF-κB inhibition to induce apoptosis in CLL cells co-cultured with supportive bone marrow stromal cells (BMSCs) using DHMEQ. The latter was chosen for its unique mode of action: it covalently binds to cysteine residues of NF-κB subunits, thereby inhibiting the interaction of NF-κB with its DNA-binding site.20 In addition, DHMEQ has also been shown to inhibit NF-κB translocation into the nucleus, making it a highly potent NF-κB inhibitor.21 Herein, we provide evidence that although NF-κB inhibition is highly effective at inducing apoptosis of monocultured CLL cells, NF-κB inhibition alone is not sufficient to induce apoptosis of CLL cells cultured with supportive BMSCs. Furthermore, our results suggest that in CLL treatment, NF-κB inhibitors should not be used as single agents, but rather in combination with substances that disrupt the crosstalk between CLL cells and their microenvironment.

Quantification of viable and apoptotic cells Viability was measured by flow cytometry (CyAn ADP, Beckmann) after staining CLL cells with fluorescein isothiocyanate (FITC) Annexin V Apoptosis Detection kit I (BD Biosciences). Annexin V (ANX5)/propidium iodide (PI) doublenegative cells were regarded as live cells, ANX5 positive/PI negative cells as early apoptotic cells, and ANX5/PI double-positive cells as late apoptotic/necrotic cells. Results were analyzed with FlowJo software (FlowJo, LLC).

NF-κB DNA-binding activity

NF-κB DNA-binding activity was measured from whole cell lysates using the TransAM® NF-κB Family Kit (Active Motif), according to the manufacturer’s instructions.

Immunoblotting Total cell protein was extracted from CLL cells and subjected to western blotting as described previously.22 Subcellular fractionation to obtain cytosolic and nuclear protein fractions for western blotting is described in the Online Supplementary Materials. Blots were incubated with primary and secondary antibodies overnight at 4°C or for 1h at room temperature. Antibodies used are listed in the Online Supplementary Materials. Densitometric analysis was performed using LabImage 1D (Kapelan Bio-Imaging). Whole cell protein expression was normalized to β-Actin and nuclear protein expression to Histone 3.

Transfection Freshly isolated CLL cells were transfected with 2μM RELA or non-targeting siRNA (GE Healthcare Dharmacon) using an Amaxa Human B cell Nucleofector Kit and nucleofector program U013 (Nucleofector 2b Device) following the manufacturer’s instructions (Amaxa).

Statistical Analysis Methods Patients This study was approved by the Institutional Review Board of the University Medical Center Freiburg. Blood samples were obtained from untreated CLL patients (off therapy for at least 6 months), following written informed consent.

Cell preparation and culture Peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation (Ficoll-Histopaque, PAA Laboratories GmbH). In PBMC samples containing fewer than 90% CD5/CD19 positive cells as measured by flow cytometry (CD5-APC, BD Pharmingen; CD19-PE-Texas Red, Southern Biotech), purification by negative selection (B cell isolation kit II, Milteny Biotec) was performed. Freshly isolated and thawed CLL cells were used explaining slight variability in the viability of untreated controls between experiments. CLL cells were treated and cultured alone or with BMSCs. BMSCs M2-10B4 (mouse) and HS-5 (human) were used, both obtained from the American Type Culture Collection (ATCC). Culture medium and treatment were replenished every other day. Reagents are listed in the Online Supplementary Materials. BMSCs were plated (2,5 x 105 cells/mL) and CLL cells (5 x 106 cells/mL) were added the following day on confluent BMSC layers. After treatment, CLL cells were harvested by careful pipetting to rinse floating CLL cells off adherent BMSCs, taking care not to damage the BMSC layer. To distinguish between the effects of direct BMSC-CLL contact versus soluble factors-induced effects, Corning® HTS Transwell® plates were used. haematologica | 2018; 103(1)

Analysis of synergistic drug effects was performed with CompuSyn (ComboSyn, Inc.). GraphPad Prism software was used for statistical analysis (version 6.0, GraphPad Software, Inc.). Data are represented as mean ± standard error of the mean (SEM). For comparisons between two parameters, a 2-tailed, paired Student's t-test was applied; for more than two parameters, one-way ANOVA with correction for multiple comparisons with Turkey’s test was used. P<0.05 was considered statistically significant.

Results DHMEQ induces apoptosis of primary CLL cells in monoculture but not in co-culture with bone marrow stromal cells CLL cells cultured alone or in co-culture with protective BMSCs (HS-5 and M2-10B4) were treated with 2 or 5μg/ml of the NF-κB inhibitor DHMEQ in vitro for up to 144h (Figure 1 and Figure 2). DHMEQ induced apoptosis of monocultured CLL cells in a dose- and time-dependent manner (Figure 1A-C). Also, in monocultured DHMEQ-treated CLL cells, the major cell population evolved from being ANX5 singlepositive after 24h, to being ANX5/PI double-positive after 48h, suggesting apoptosis induction rather than direct toxicity (Figure 1D). Surprisingly, the viability of DHMEQ-treated CLL cells co-cultured with BMSCs remained unchanged, irrespective of DHMEQ dosage (2 or 5μg/ml for 48h) or treatment 137


C.P. Simon-Gabriel et al.

time. DHMEQ-induced apoptosis was suppressed by both BMSCs tested (HS-5, M2-10B4) (Figure 1A-C). Of note, CLL cell viability of untreated controls after 4 and 6 days was higher than on day 2 (Figure 1A). We reasoned that the process of thawing led to apoptosis and the death of some CLL cells, which is supported by western blot analysis of PARP/cleaved PARP expression on day 0 (data not shown), leading to a comparably low viability on day 2. After cell components were processed and degraded, the percentage of live cells increased on day 4 in vehicle-treated cohorts (monoculture and co-culture).

Reduced viability in monocultured CLL cells is accompanied by activation of the intrinsic apoptotic pathway We next analyzed whether the reduction of CLL cell viability relies on activation of the intrinsic apoptotic pathway by downregulation of the NF-κB target genes TRAF1, BCL2L1 (also known as BCL-xL), and MCL1, and increased PARP cleavage. TRAF1 is a recognized NF-κB target gene,23,24 BCL2L1 and MCL1 represent two antiapoptotic BCL2 family members known to be regulated by NF-κB. PARP cleavage is frequently used as a surrogate marker for caspase-3 activation. Monocultured CLL cells treated with DHMEQ (2μg/ml) for 48h showed a significant downregulation of TRAF1 expression (P=0.0194) accompanied by a notable increase of PARP cleavage (P=0.067) compared to untreated controls, while BCL2L1 and MCL1 remained unaffected

(Figure 2A). Interestingly, TRAF1 downregulation occurred before the increase in PARP cleavage (Figure 2B). In contrast, in CLL cells co-cultured with M2-10B4 cells, DHMEQ treatment for 2 or 6 days did not induce changes in PARP cleavage. In fact, PARP cleavage was almost undetectable in CLL cells co-cultured with BMSCs (Figure 2A,C). Under co-culture conditions, no significant downregulation of BCL2L1 and MCL1 was detected upon treatment. Only TRAF1 expression tended to decrease (Figure 2A). Notably, BCL2L1 expression was increased in co-cultured CLL cells. Similar results were observed after 6 days of treatment with 2 μg/ml of DHMEQ. Although significant TRAF1 downregulation was seen in both monocultured and co-cultured CLL cells, PARP cleavage was only induced in monocultured cells. Additional analysis of BAX, a proapoptotic protein, showed increased expression in monocultured CLL cells after DHMEQ treatment, but no change in the co-culture setting (Figure 2C).

DNA-binding activity of all five NF-κB subunits is strongly suppressed by DHMEQ treatment in monocultured CLL cells and also in those cells co-cultured with supportive stromal cells We next tested whether DHMEQ inhibited NF-κB activity in the various conditions by using the TransAM® NFκB Family Kit (Active Motif), a DNA-binding enzyme-linked immunosorbent assay (ELISA) which enabled us to test the DNA-binding activity of each NF-κB subunit. Additionally, western blot analyses of the expression of

A

B

C

D

Figure 1. DHMEQ reduces viability of CLL cells in monoculture but not in co-culture with stromal cells. Cell viability as measured by flow cytometry with ANX5/PI staining of CLL cells cultured in vitro (A) alone or in co-culture with M2-10B4 cells after 2, 4, and 6 days of treatment with 2μg/ml of DHMEQ, (B) alone or in coculture with HS-5 cells after 2 days of treatment with 2μg/ml or (C) 5μg/ml of DHMEQ. (D) Cell populations as measured by flow cytometry after ANX5/PI staining of in vitro monocultured CLL cells with or without 5μg/ml of DHMEQ. Fold changes of CLL cell viability are indicated above the signs for significance. ****P<0.0001, ***P<0.001, **P<0.01 and *P<0.05. CLL: chronic lymphocytic leukemia; DHMEQ: dehydroxymethylepoxyquinomicin; ns: not significant.

138

haematologica | 2018; 103(1)


NF-κB inhibition in CLL

the different NF-κB subunits in nuclear and cytosolic extracts were performed (Figure 3). DNA-binding activity of all five NF-κB subunits was significantly suppressed by DHMEQ treatment in monoculture in addition to CLL cells co-cultured with M2-10B4 cells (Figure 3A). DHMEQ treatment reduced nuclear translocation of all NF-κB subunits in both the monoculture and the co-culture setting (Figure 3B). Furthermore, western blot analyses of the expression of the different NF-κB subunits in whole cell lysates showed significant reduction after 6 days of treatment with 2μg/ml DHMEQ in co-culture with stromal cells (Figure 3C). Taken together, these results validate the efficiency of NF-κB inhibition by DHMEQ in both monoculture and co-culture.

siRNA-mediated knockdown of the NF-κB subunit RELA does not induce apoptosis of CLL cells in co-culture with stromal cells but appears to do so in monocultured CLL cells To confirm the finding that NF-κB inhibition is not sufficient to induce apoptosis of CLL cells co-cultured with BMSCs, siRNA-mediated knockdowns of the NF-κB subunit RELA (also known as p65) were performed. From the different NF-κB subunits, RELA was chosen as the knockdown target as it has been shown that the high binding activity of RELA to its DNA-binding site is predictive of a short time to first treatment, time to subsequent treatment, and overall survival from the date of diagnosis.13 The knock-

A

down efficiency in the analyzed samples was assessed by western blot, and the effect of RELA knockdown on viability of CLL cells in monoculture and co-culture with M210B4 cells was assessed by flow cytometric apoptosis measurements (Figure 4). In the western blot analysis, the average expression of RELA in RELA-transfected CLL cells in the monoculture setting was 79.2% as compared to 90.9% in the non-targeting transfected controls, and as such did not reach significance. In contrast, in the co-culture setting RELA expression in CLL cells was reduced to 38% in RELA-transfected cells compared to 73.2% in cells transfected with non-targeting siRNA (Figure 4A,B). However, when analyzing the apoptosis measurements, the relative reduction of RELA by about 50% had no visible effect on the viability of CLL cells co-cultured with M2-10B4 cells, whereas the less efficient knockdown in monocultured CLL cells had a marked tendency to induce apoptosis of these cells as compared to respective controls. This is in accordance with our previous data suggesting that NF-κB inhibition is not sufficient to induce apoptosis in CLL cells in the presence of supportive BMSCs.

Soluble factors and cell bound factors from the BMSCs are likely to protect CLL cells from apoptosis induction by NF-κB inhibition with DHMEQ This discovery led us to question which factors from the BMSCs were responsible for CLL cell protection from apoptosis induction by NF-κB inhibition.

C

B

Figure 2. DHMEQ treatment activates the intrinsic apoptotic pathway. Densitometrically analyzed protein expression normalized to β-Actin of CLL cells cultured in vitro (A) alone or in co-culture with M2-10B4 cells after 2 days of treatment with 2µg/ml of DHMEQ, (B) alone after 0.5, 1, 2, 4, 8 and 24 hours of treatment with 2 μg/ml of DHMEQ and (C) alone or in co-culture with M2-10B4 cells after 6 days of treatment with 2μg/ml of DHMEQ shown with exemplary western blot. ****P<0.0001, ***P<0.001, **P<0.01 and *P<0.05. D: DHMEQ (dehydroxymethylepoxyquinomicin); D2: day 2; D6: day 6; SD: stroma + DHMEQ; SV: stroma + vehicle control; v: vehicle control; CLL: chronic lymphocytic leukemia; ns: not significant; ACTB: β-Actin.

haematologica | 2018; 103(1)

139


C.P. Simon-Gabriel et al.

First, we used Corning® Transwell® inserts offering minuscule pores impermeable for cells, but permeable for culture medium and soluble factors for co-culture experiments. Hence, half of the CLL cells were placed in the lower compartment, in direct contact with the M2-10B4 cells, while the other half was placed in the inserts so that they could only be influenced by the BSMCs via soluble factors (Figure 5A). The viability of CLL cells from the lower compartments and from the inserts was measured separately via flow cytometry after ANX5/PI staining (Figure 5B). As before, the monocultured CLL cells were very sensi-

A

tive to DHMEQ treatment and exhibited a strong decrease in viability, while the co-cultured CLL cells in the lower compartment with direct cell-cell contact showed virtually no apoptosis after DHMEQ treatment. In contrast, the co-cultured CLL cells in the inserts (with contact between stromal and CLL cells through soluble ligands only) showed an intermediary sensitivity to DHMEQ treatment, as their viability was significantly decreased by DHMEQ treatment while staying significantly better than the viability of DHMEQ-treated CLL cells in monoculture (Figure 5B). This indicates that the protection afforded by BMSCs from NF-κB-inhibition-induced apoptosis is partly

B

C

Figure 3. NF-κB DNA-binding activity, nuclear and cytosolic expression are significantly reduced by DHMEQ treatment in both monocultured and co-cultured CLL cells. (A) DNA-binding activity of the different NF-κB subunits in monocultured or M2-10B4 co-cultured CLL cells treated with 5μg/ml of DHMEQ or vehicle control for 8h. (B) Western blot analysis of cytosolic and nuclear NF-κB expression in monocultured or M2-10B4 co-cultured CLL cells after 2 days of treatment with 2μg/ml of DHMEQ. (C) Densitometrically analyzed NF-κB expression normalized to β-Actin of CLL cells cultured in co-culture with M2-10B4 cells after 6 days of treatment with 2μg/ml of DHMEQ shown with exemplary western blot. Since the viability of the DHMEQ-treated CLL cells under monoculture conditions was very poor, densitometric analysis was not possible in this set of experiments. ****P<0.0001, ***P<0.001, **P<0.01 and *P<0.05. ACTB: β-Actin; C: cytosolic; D6: day 6; N: nuclear; SD: stroma + DHMEQ; SV: stroma + vehicle control; P1: patient 1; P2: patient 2; CLL: chronic lymphocytic leukemia; ns: not significant.

140

haematologica | 2018; 103(1)


NF-κB inhibition in CLL

mediated via soluble factors. However, direct contact or closer proximity of CLL cells appears to increase the protective effect of the stromal cells.

BAFF but not a proliferation-inducing ligand (APRIL), CD40L, or C-X-C motif chemokine 12 (CXCL12) can protect CLL cells from DHMEQ-induced apoptosis To identify factors which might contribute to CLL cell protection from NF-κB-inhibition-induced apoptosis, we

assessed the protective effect of four acknowledged factors involved in the crosstalk between the microenvironment and CLL cells. CLL cells were cultured alone for 144h with or without DHMEQ (2 μg/ml) and with or without one of the following: CD40L (1 μg/ml), CXCL12 (100ng/ml), BAFF (50ng/ml), and APRIL (500 μg/ml). Cell viability was measured by flow cytometry of ANX5/PIstained CLL cells (Figure 5C,D, Online Supplementary Figure S1). While APRIL did not protect CLL cells from DHMEQ-

A

B

C

Figure 4. siRNA-mediated RELA downregulation seems to induce apoptosis in monocultured but not in co-cultured CLL cells. (A) Exemplary western blots of RELA in monocultured and M2-10B4 co-cultured CLL cells after transfection with non-targeting siRNA or RELA siRNA. (B) Densitometric western blot analysis illustrating the knockdown efficiency obtained in monocultured and co-cultured CLL cells 2 days after transfection. (C) CLL cell viability 3 days after transfection with non-targeting siRNA or RELA siRNA. ***P<0.001, **P<0.01 and *P<0.05. ACTB: β-Actin; D2: day 2; D3: day 3; n-t-siRNA: non-targeting small interfering ribonucleic acid; untr: untreated; CLL: chronic lymphocytic leukemia; ns: not significant.

haematologica | 2018; 103(1)

141


C.P. Simon-Gabriel et al.

induced apoptosis, CD40L, CXCL12 and BAFF slightly reduced DHMEQ’s proapoptotic effect. A combination of these four ligands did not result in better protection than with BAFF alone (Figure 5C, Online Supplementary Figure S1). Indeed, the viability of BAFF- and DHMEQ-treated cells was 16% as compared to 1% for solely DHMEQtreated cells and 48% for M2-10B4 co-cultured CLL cells (Figure 5D). The decreased effectivity of DHMEQ treatment when CLL cells were stimulated with BAFF was significant when using a t-test (Figure 5E). These results suggest that, amongst other factors, BAFF could be involved in protecting CLL cells from NF-κB-inhibition-induced apoptosis.

The combination of DHMEQ with fludarabine, idelalisib, R406, or ibrutinib partly restores the DHMEQ-sensitivity of stroma-supported CLL cells As mentioned before, NF-κB inhibition has been proposed as a promising strategy in CLL treatment. However,

A

our results indicate that NF-κB inhibitors alone might not be as effective as previously assumed in the treatment of CLL patients. Therefore, we investigated whether DHMEQ could be effectively combined with established chemotherapeutic agents (fludarabine) or substances which have been shown to disrupt the crosstalk between the microenvironment and CLL cells (idelalisib, ibrutinib, R406). Indeed, apart from inhibiting the BCR pathway, ibrutinib, idelalisib, and R406 have been shown to disrupt at least a part of the interactions between CLL cells and their microenvironments.10,11,22 Therefore, cells in monoculture or co-culture with M2-10B4 cells were treated with or without DHMEQ and with or without ibrutinib, idelalisib, R406 or fludarabine. In the monoculture setting, CLL cell viability was significantly decreased by every substance used alone, including DHMEQ. In addition, combining DHMEQ with ibrutinib, R406, idelalisib, and fludarabine, respectively, led to significantly reduced CLL cell viability. In the co-culture set-

B

C

D

E

Figure 5. Protection from DHMEQ-induced apoptosis involves direct cell-cell interaction as well as soluble factors such as BAFF. (A) Illustration of the transwell experiment. (B) Cell viability after DHMEQ treatment of CLL cells in monoculture and co-culture, depending on their localization in the transwell experiment. (C) CLL cell viability with or without DHMEQ in monoculture or in the presence of APRIL, BAFF, CXCL12, CD40L, all ligands, or M2-10B4 cells. (D) Viability of CLL cells cultured alone, with BAFF, or with M2-10B4 cells, and treated with or without DHMEQ. (E) Additional t-test analysis of monocultured CLL cells versus BAFF-supported CLL cells, both treated with DHMEQ. ****P<0.0001, ***P<0.001 and *P<0.05. CLL: chronic lymphocytic leukemia; DHMEQ: dehydroxymethylepoxyquinomicin; BAFF: B-cell activating factor; APRIL: a proliferation-inducing ligand; CXCL12: C-X-C motif chemokine 12.

142

haematologica | 2018; 103(1)


NF-ÎşB inhibition in CLL

Figure 6. Combining DHMEQ with stroma crosstalk inhibiting substances partly restores DHMEQ sensitivity in stroma-supported CLL cells. Viability of CLL cells in monoculture or co-culture with M2-10B4 cells treated for 2 days with or without DHMEQ, with or without fludarabine, R406, idelalisib, or ibrutinib, or a combination of DHMEQ and one of these substances.****P<0.0001, ***P<0.001, **P<0.01 and *P<0.05. D2: day 2; Ibru: ibrutinib; Idel: idelalisib; Flud: fludarabine; CLL: chronic lymphocytic leukemia; DHMEQ: dehyd r o x y m e t h y l e poxyquinomicin; ns: not significant.

haematologica | 2018; 103(1)

143


C.P. Simon-Gabriel et al.

ting, as expected, DHMEQ monotherapy did not induce a significant reduction in CLL cell viability, while fludarabine, ibrutinib, R406, and idelalisib did. However, combining DHMEQ with ibrutinib or R406, but not with fludarabine, led to a significant reduction in the viability of CLL cells cultured with stromal cells, compared to CLL cells treated with the respective single agents (Figure 6). There was no significant difference in the obtained effect when drugs were added concomitantly (DHMEQ + ibrutinib or DHMEQ + R406 added at the same time) versus sequentially (DHMEQ added 4 hours before ibrutinib or R406 and vice versa, Online Supplementary Figure S2). To determine whether the combined effects of DHMEQ + ibrutinib and of DHMEQ + R406 on CLL cells cultured with stromal cells are synergistic, additive, or antagonistic, combination indices were calculated (see Online Supplementary Methods). A combination index (CI) <1, =1, and >1 indicates synergism, additive effect and antagonism, respectively. The combination of DHMEQ and ibrutinib showed moderate

synergism (0.79) at the median effective dose (ED50), and frank synergism at higher effective doses (CI< 0.7 at ED75, 90 and 95, respectively) (Figure 7A,C). The combination of DHMEQ and R406, diversely, showed nearly additive effects at ED50, moderate synergism at ED75 and ED90, and frank synergism at ED95 (Figure 7A,C). Additionally, dose-reduction indices (DRIs) were calculated to investigate whether adding DHMEQ to R406 or ibrutinib treatment could lower the doses of R406 or ibrutinib needed to obtain a given effect, thus potentially decreasing side-effects. A DRI>1 indicates favorable dosereduction. Figure 7B shows that at fractions affected (Fa) > 0.1, a favorable dose-reduction of ibrutinib or R406 can be obtained by combining these drugs with DHMEQ.

Discussion Although previous studies using NF-κB inhibitors indicated that NF-κB could be a promising therapeutic target

A

B

C

Figure 7. Combination Index studies reveal synergistic effect for the combination of DHMEQ and ibrutinib or R406 respectively on M2-10B4 co-cultured CLL cells. Combination index (CI) studies were performed by treating CLL cells co-cultured with M2-10B4 cells with DHMEQ (titrated from 8μg/ml), ibrutinib (titrated from 15μM) or R406 (titrated from 10μM) and combinations of R406+DHMEQ (at a ratio of 1.25:1) or ibrutinib + DHMEQ (at a ratio of 1.875:1) for 48h. Drug-induced cytotoxicities were calculated using the CLL cell viabilities to generate CIs. (A) CI plots for DHMEQ+ ibrutinib (D&I) and DHMEQ+R406 (D&R). CI-values <1 indicate synergism, CI=1 indicates additive effects and CI-values >1 indicate antagonism. (B) Dose-reduction indices (DRIs) were calculated along with the CI. A DRI>1 indicates a favorable dose reduction, while a DRI<1 indicates an unfavorable dose reduction. Above a fraction affected of 0.1, all DRIs are >1, indicating that the doses of R406 or ibrutinib needed for a given effect level in single drug treatment can be reduced by adding DHMEQ. (C) The obtained CI values for the combination D&I and D&R at different effective doses (ED50, ED75, ED90, ED95) are given. Fa: fraction affected; DHMEQ: dehydroxymethylepoxyquinomicin.

144

haematologica | 2018; 103(1)


NF-κB inhibition in CLL

in CLL, our data suggest that the role of the protective CLL microenvironment might have been underestimated in these reports. In this study we used the specific NF-κB inhibitor DHMEQ, which has been shown to induce apoptosis in CLL cells but not in normal B-cells,19 and has been used in vivo in mice with promising therapeutic results and no serious side effects for the treatment of rheumatoid arthritis,25 systemic lupus erythematosus,26 and most interestingly in human T-cell leukemia27,28 and multiple myeloma.29 In our study, we confirmed that DHMEQ is a potent NF-κB inhibitor blocking both nuclear translocation and DNA-binding of the NF-κB subunits, regardless of the presence of a protective microenvironment imitated by BMSCs. While NF-κB inhibition using DHMEQ led to a strong induction of the intrinsic apoptotic pathway by BAX upregulation, increased PARP cleavage, and a downregulation of TRAF1 in CLL cells cultured without BMSCs, expression of the NF-κB target genes BCL2L1 and MCL1 remained unaffected. Albeit others have observed the downregulation of BCL2L1 and/or MCL1 after NF-κB inhibition,17,19 our results are in line with those of Pickering et al. who reported that NF-κB inhibition did not alter BCL2L1, BCL2 and MCL1 expression in CLL cells in vitro.16 In the co-culture setting, DHMEQ efficiently inhibited NF-κB activity and caused downregulation of the NF-κB subunits themselves, either by DHMEQ directly,30 or by breaking the self-renewing cycle of NF-κB expression.31–36 Surprisingly, neither CLL cell survival nor the expression of BAX or cleaved PARP was influenced by DHMEQ in the co-culture setting. In the NF-κB gene-silencing performed to support the findings from pharmacological NFκB inhibition, a loss of importance of the RELA subunit was specifically seen for the survival of CLL cells co-cultured with BMSCs. It seems that in the presence of BMSCs, NF-κB is not essential for CLL cell survival while other pathways become more relevant, thus questioning the efficacy of NF-κB inhibitors as monotherapeutic agents in CLL. As we observed an upregulation of the proapoptotic BCL2 family member BAX upon DHMEQ treatment in the monoculture setting only, and the anti-apoptotic protein BCL2L1 was upregulated in the CLL-BMSC co-culture setting, it is possible that BCL2 family members could be key players in the NF-κB-independent resistance to apoptosis seen in CLL cells co-cultured with BSMCs. Adding BCL2 inhibitors to DHMEQ treatment might diminish the microenvironment’s protective effect and represent an attractive combination of targeted therapeutics. Fittingly, López-Guerra et al. reported strong synergism between BMS-345541 (a selective IKK inhibitor) and the pan-BCL2 inhibitor obatoclax.17 Combinations with the BCL2 inhibitors venetoclax, which received breakthrough therapy designation,37,38 and navitoclax39 could also offer good prospects of success. Furthermore, the study herein shows that both direct cell-cell interactions between CLL cells and BMSCs as well as soluble ligands secreted by the BSMCs are involved in mediating the protective effect against NF-κB inhibition. In particular, soluble BAFF, secreted in vivo by nurse-like cells (NLC)12 and BMSCs40 was identified. CLL cells themselves have been shown to express and secrete BAFF, implying an autocrine loop.41 BAFF binds to three receptors: BCMA (B-cell maturation antigen), TACI (transmembrane activator of the calcium modulator and cyclophylin haematologica | 2018; 103(1)

ligand interactor), and BAFF-Receptor (BR3).12,42 APRIL is known to bind to BCMA and TACI, but not to BR3.12,43,44 Since APRIL did not protect CLL cells from DHMEQinduced apoptosis to the same extent as BAFF, a central role for the receptor BR3 can be hypothesized. A possible explanatory pathway is a recently described influence of gene expression through Histone H3 phosphorylation: the binding of BAFF to BR3 induces receptor translocation to the nucleus, where the receptor can bind IKKβ. The kinase then phosphorylates Histone H3, leading to the activation of a variety of transcription factors and cofactors.45 Since DHMEQ binds to the NF-κB subunits but not to the IKKs, a regulation of anti-apoptotic genes through BAFF may exist despite DHMEQ treatment.21 Interestingly, the BAFF inhibitor belimumab, which is currently only approved for the treatment of systemic lupus erythematosus, has recently been tested on CLL cells;46 Wild et al. report the neutralization of BAFF’s CLL-protecting effect by belimumab, sensitizing CLL cells to lysis. Combining DHMEQ with belimumab could therefore be a promising approach. Moreover, by disrupting the interactions between the microenvironment and CLL cells through the addition of ibrutinib (a BTK inhibitor) or R406 (a SYK inhibitor), DHMEQ’s proapoptotic effect was partially restored, while the proapoptotic effects of the combined treatments were also significantly higher than those of the single drug treatments with ibrutinib or R406. Indeed, CIs for both the combination of DHMEQ + ibrutinib as well as DHMEQ + R406 indicate synergism. Dose-reduction indices for R406 and ibrutinib further imply that adding DHMEQ could lower the doses of R406 and ibrutinib needed for a given effect level, thus potentially reducing side effects. It is known that ibrutinib is able to overcome pro-survival signals derived from BCR stimulation, NLCcontact, and BAFF, among others.47,48 Ibrutinib also releases tumor cells from the tissue compartment into the peripheral blood, a drug effect that can be seen only in in vivo experiments.49 Since ibrutinib inhibits BCR signaling, potentially bypassing the NF-κB pathway in the presence of the microenvironment, and further disrupts molecular crosstalk between CLL cells and their microenvironment by redistributing CLL cells into the peripheral blood, a combination of DHMEQ and ibrutinib could possibly show an even better synergistic effect in vivo. Furthermore, BTK influences CLL cell survival by activating the NF-κB pathway. Hence, inhibiting those pathways could not only restore DHMEQ’s proapoptotic effect by disrupting interaction between CLL cells and their microenvironment, but also increase NF-κB inhibition. We and others previously identified SYK as a candidate for targeted therapy in CLL due to its enhanced expression and activity and the apoptotic effects of pharmacological SYK inhibition.22,50 Entospletinib, a selective SYK inhibitor, demonstrated promising clinical activity in patients with relapsed or refractory CLL.51 Our results suggest that combining DHMEQ and entospletinib might also be a promising therapeutic strategy. Our study has some limitations. First, our results are based solely on experiments using the NF-κB inhibitor DHMEQ. This is because DHMEQ is quite unique, as, unlike the plethora of other NF-κB inhibitors, its exact and selective mechanism of action is well documented,20 and it disrupts the final step of the NF-κB signaling pathway, ensuring that no cross regulation via other signaling 145


C.P. Simon-Gabriel et al.

pathways downstream of the targeted signaling step is possible. To our knowledge, no other NF-κB inhibitor meets these criteria. Second, the ibrutinib concentration used in the combination experiment was relatively high (10 μM), a concentration, however, which is not uncommon for in vitro experiments.47 In vivo, the maximum concentration which is obtained with a dose of 840mg ibrutinib daily does not exceed circa 0.48 μM (210ng/ml).52 However, it has been shown that, in vivo, the primary effect obtained with the aforementioned ibrutinib con-

References 1. Pekarsky Y, Zanesi N, Aqeilan RI, Croce CM. Animal models for chronic lymphocytic leukemia. J Cell Biochem. 2007; 100(5):1109–1118. 2. Pekarsky Y, Zanesi N, Croce CM. Molecular basis of CLL. Semin Cancer Biol. 2010;20(6):370–376. 3. Zanesi N, Balatti V, Bottoni A, M. Croce C, Pekarsky Y. Novel insights in molecular mechanisms of CLL. Curr Pharm Des. 2012;18(23):3363–3372. 4. Gaidano G, Foà R, Dalla-Favera R. Molecular pathogenesis of chronic lymphocytic leukemia. J Clin Invest. 2012; 122(10):3432–3438. 5. Karin M, Cao Y, Greten FR, Li Z-W. NF-κB in cancer: from innocent bystander to major culprit. Nat Rev Cancer. 2002;2(4):301–310. 6. Aggarwal BB. Nuclear factor-κB: the enemy within. Cancer Cell. 2004;6(3):203–208. 7. Furman RR, Asgary Z, Mascarenhas JO, Liou H-C, Schattner EJ. Modulation of NFκB activity and apoptosis in chronic lymphocytic leukemia B cells. J Immunol. 2000; 164(4):2200–2206. 8. Romano MF, Lamberti A, Tassone P, et al. Triggering of CD40 antigen inhibits fludarabine-induced apoptosis in B chronic lymphocytic leukemia cells. Blood. 1998; 92(3):990–995. 9. Hertlein E, Byrd JC. Signalling to drug resistance in CLL. Best Pract Res Clin Haematol. 2010;23(1):121–131. 10. Burger JA, Gribben JG. The microenvironment in chronic lymphocytic leukemia (CLL) and other B cell malignancies: Insight into disease biology and new targeted therapies. Semin Cancer Biol. 2014;24:71–81. 11. ten Hacken E, Burger JA. Microenvironment interactions and B-cell receptor signaling in chronic lymphocytic leukemia: implications for disease pathogenesis and treatment. Biochim Biophys Acta BBA - Mol Cell Res. 2016;1863(3): 401–413. 12. Endo T, Nishio M, Enzler T, et al. BAFF and APRIL support chronic lymphocytic leukemia B-cell survival through activation of the canonical NF-κB pathway. Blood. 2007;109(2):703–710. 13. Hewamana S, Lin TT, Rowntree C, et al. Rel A is an independent biomarker of clinical outcome in chronic lymphocytic leukemia. J Clin Oncol. 2009; 27(5):763– 769. 14. Steele AJ, Jones DT, Ganeshaguru K, et al. The sesquiterpene lactone parthenolide induces selective apoptosis of B-chronic lymphocytic leukemia cells in vitro.

146

centration is the blockage of cell proliferation, while apoptosis only occurs at higher doses.53 As our primary endpoint was CLL cell apoptosis, we deliberately chose this relatively high dose. To conclude, considering that our results strongly indicate that NF-κB inhibition alone is not able to induce apoptosis in CLL with a supportive microenvironment, we suggest that DHMEQ should be tested in combination with R406, ibrutinib, the BAFF inhibitor belimumab, potentially, or BCL2 inhibitors in subsequent studies.

Leukemia. 2006;20(6):1073–1079. 15. Kanduri M, Tobin G, Åleskog A, Nilsson K, Rosenquist R. The novel NF-κB inhibitor IMD-0354 induces apoptosis in chronic lymphocytic leukemia. Blood Cancer J. 2011;1(3):e12. 16. Pickering BM, de Mel S, Lee M, et al. Pharmacological inhibitors of NF-κB accelerate apoptosis in chronic lymphocytic leukaemia cells. Oncogene. 2006; 26(8):1166–1177. 17. López-Guerra M, Roué G, Pérez-Galán P, et al. p65 activity and ZAP-70 status predict the sensitivity of chronic lymphocytic leukemia cells to the selective IκB kinase inhibitor BMS-345541. Clin Cancer Res. 2009;15(8):2767–2776. 18. Hewamana S, Alghazal S, Lin TT, et al. The NF-κB subunit Rel A is associated with in vitro survival and clinical disease progression in chronic lymphocytic leukemia and represents a promising therapeutic target. Blood. 2008;111(9):4681–4689. 19. Horie R, Watanabe M, Okamura T, et al. DHMEQ, a new NF-κB inhibitor, induces apoptosis and enhances fludarabine effects on chronic lymphocytic leukemia cells. Leukemia. 2006;20(5):800–806. 20. Yamamoto M, Horie R, Takeiri M, Kozawa I, Umezawa K. Inactivation of NF-κB components by covalent binding of (−)-dehydroxymethylepoxyquinomicin to specific cysteine residues. J Med Chem. 2008; 51(18):5780–5788. 21. Ariga A, Namekawa J, Matsumoto N, Inoue J, Umezawa K. Inhibition of tumor necrosis factor-α-induced nuclear translocation and activation of NF-κB by dehydroxymethylepoxyquinomicin. J Biol Chem. 2002;277(27):24625–24630. 22. Buchner M, Baer C, Prinz G, et al. Spleen tyrosine kinase inhibition prevents chemokine- and integrin-mediated stromal protective effects in chronic lymphocytic leukemia. Blood. 2010;115(22):4497–4506. 23. Dunn IF, Sannikova TY, Geha RS, Tsitsikov EN. Identification and characterization of two CD40-inducible enhancers in the mouse TRAF1 gene locus. Mol Immunol. 2000;37(16):961–973. 24. Carpentier I, Beyaert R. TRAF1 is a TNF inducible regulator of NF-κB activation. FEBS Lett. 1999;460(2):246–250. 25. Wakamatsu K, Nanki T, Miyasaka N, Umezawa K, Kubota T. Effect of a small molecule inhibitor of nuclear factor-κB nuclear translocation in a murine model of arthritis and cultured human synovial cells. Arthritis Res Ther. 2005;7(6):R1348-1359. 26. Qu H, Bian W, Xu Y. A novel NF-κB inhibitor, DHMEQ, ameliorates pristaneinduced lupus in mice. Exp Ther Med. 2014;8(1):100–104.

27. Ohsugi T, Horie R, Kumasaka T, et al. In vivo antitumor activity of the NF-κB inhibitor dehydroxymethylepoxyquinomicin in a mouse model of adult T-cell leukemia. Carcinogenesis. 2005; 26(8):1382–1388. 28. Watanabe M, Ohsugi T, Shoda M, et al. Dual targeting of transformed and untransformed HTLV-1-infected T cells by DHMEQ, a potent and selective inhibitor of NF-κB, as a strategy for chemoprevention and therapy of adult T-cell leukemia. Blood. 2005;106(7):2462–2471. 29. Watanabe M, Dewan MZ, Okamura T, et al. A novel NF-κB inhibitor DHMEQ selectively targets constitutive NF-κB activity and induces apoptosis of multiple myeloma cells in vitro and in vivo. Int J Cancer.;114(1):32–38. 30. Dabaghmanesh N, Matsubara A, Miyake A, et al. Transient inhibition of NF-κB by DHMEQ induces cell death of primary effusion lymphoma without HHV-8 reactivation. Cancer Sci. 2009;100(4):737–746. 31. Oeckinghaus A, Ghosh S. The NF-κB family of transcription factors and its regulation. Cold Spring Harb Perspect Biol. 2009;1(4): a000034. 32. Perkins ND. Integrating cell-signalling pathways with NF-κB and IKK function. Nat Rev Mol Cell Biol. 2007;8(1):49–62. 33. Bren GD, Solan NJ, Miyoshi H, Pennington KN, Pobst LJ, Paya CV. Transcription of the RelB gene is regulated by NF-κB. Oncogene. 2001;20(53):7722–7733. 34. Neumann M, Wohlleben G, Chuvpilo S, et al. CD40, but not lipopolysaccharide and anti-IgM stimulation of primary B lymphocytes, leads to a persistent nuclear accumulation of RelB. J Immunol Baltim Md 1950.;157(11):4862–4869. 35. Ten RM, Paya CV, Israël N, et al. The characterization of the promoter of the gene encoding the p50 subunit of NF-κB indicates that it participates in its own regulation. EMBO J. 1992;11(1):195–203. 36. Lombardi L, Ciana P, Cappellini C, et al. Structural and functional characterization of the promoter regions of the NFKB2 gene. Nucleic Acids Res. 1995;23(12):2328–2336. 37. Ruefli-Brasse A, Reed JC. Therapeutics targeting Bcl-2 in hematological malignancies. Biochem J. 2017; 474(21):3643-3657. 38. Gentile M, Petrungaro A, Uccello G, et al. Venetoclax for the treatment of chronic lymphocytic leukemia. Expert Opin Investig Drugs. 2017;26(11):1307-1316. 39. Kipps TJ, Eradat H, Grosicki S, et al. A phase 2 study of the BH3 mimetic BCL2 inhibitor navitoclax (ABT-263) with or without rituximab, in previously untreated B-cell chronic lymphocytic leukemia. Leuk Lymphoma. 2015;56(10):2826–2833.

haematologica | 2018; 103(1)


NF-κB inhibition in CLL

40. Lwin T, Crespo LA, Wu A, et al. Lymphoma cell adhesion-induced expression of B cell-activating factor of the TNF family in bone marrow stromal cells protects non-Hodgkin’s B lymphoma cells from apoptosis. Leukemia. 2008; 23(1):170–177. 41. Kern C, Cornuel J-F, Billard C, et al. Involvement of BAFF and APRIL in the resistance to apoptosis of B-CLL through an autocrine pathway. Blood. 2004; 103(2):679–688. 42. Morrison MD, Reiley W, Zhang M, Sun SC. An atypical tumor necrosis factor (TNF) receptor-associated factor-binding motif of B cell-activating factor belonging to the TNF family (BAFF) receptor mediates induction of the noncanonical NF-κB signaling pathway. J Biol Chem. 2005; 280(11):10018–10024. 43. Day ES, Cachero TG, Qian F, et al. Selectivity of BAFF/BLyS and APRIL for binding to the TNF family receptors BAFFR/BR3 and BCMA. Biochemistry (Mosc). 2005;44(6):1919–1931. 44. Haiat S, Billard C, Quiney C, Ajchenbaum-

haematologica | 2018; 103(1)

45.

46.

47.

48.

Cymbalista F, Kolb J-P. Role of BAFF and APRIL in human B-cell chronic lymphocytic leukaemia. Immunology. 2006; 118(3):281–292. Fu L, Lin-Lee Y-C, Pham LV, Tamayo AT, Yoshimura LC, Ford RJ. BAFF-R promotes cell proliferation and survival through interaction with IKKβ and NF-κB/c-Rel in the nucleus of normal and neoplastic B-lymphoid cells. Blood. 2009; 113(19):4627–4636. Wild J, Schmiedel BJ, Maurer A, et al. Neutralization of (Nκ-cell-derived) B-cell activating factor by Belimumab restores sensitivity of chronic lymphoid leukemia cells to direct and rituximab-induced NK lysis. Leukemia. 2015;29(8):1676–1683. Herman SEM, Gordon AL, Hertlein E, et al. Bruton tyrosine kinase represents a promising therapeutic target for treatment of chronic lymphocytic leukemia and is effectively targeted by PCI-32765. Blood. 2011; 117(23):6287–6296. Ponader S, Chen S-S, Buggy JJ, et al. The Bruton tyrosine kinase inhibitor PCI-32765 thwarts chronic lymphocytic leukemia cell survival and tissue homing in vitro and in

vivo. Blood. 2012;119(5):1182–1189. 49. Herman SEM, 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 Off J Am Assoc Cancer Res. 2015;21(20): 4642–4651. 50. Buchner M, Fuchs S, Prinz G, et al. Spleen tyrosine kinase is overexpressed and represents a potential therapeutic target in chronic lymphocytic leukemia. Cancer Res. 2009;69(13):5424–5432. 51. Sharman J, Hawkins M, Kolibaba K, et al. An open-label phase 2 trial of entospletinib (GS-9973), a selective spleen tyrosine kinase inhibitor, in chronic lymphocytic leukemia. Blood. 2015;125(15):2336–2343. 52. 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. 53. Cheng S, Ma J, Guo A, et al. BTK inhibition targets in vivo CLL proliferation through its effects on B-cell receptor signaling activity. Leukemia. 2014;28(3):649–657.

147


ARTICLE

Platelet Biology & Its Disorders

Ferrata Storti Foundation

Haematologica 2018 Volume 103(1):148-162

Introducing high-throughput sequencing into mainstream genetic diagnosis practice in inherited platelet disorders

José M. Bastida,1,3 María L. Lozano,2,3 Rocío Benito,4 Kamila Janusz,4 Verónica Palma-Barqueros,2 Mónica Del Rey,4 Jesús M. Hernández-Sánchez,4 Susana Riesco,5 Nuria Bermejo,6 Hermenegildo González-García,7 Agustín Rodriguez-Alén,8 Carlos Aguilar,9 Teresa Sevivas,10 María F. López-Fernández,11 Anna E. Marneth,12 Bert A. van der Reijden,12 Neil V. Morgan,13 Steve P. Watson,13 Vicente Vicente,3 Jesús M. Hernández-Rivas,1,4 José Rivera*2,3 and José R. González-Porras*1

Servicio de Hematología, Hospital Universitario de Salamanca-IBSAL-USAL, Spain; Servicio de Hematología y Oncología Médica, Hospital Universitario Morales Meseguer, Centro Regional de Hemodonación, Universidad de Murcia, IMIB-Arrixaca, CB15/00055-CIBERER, Spain; 3On behalf of the Project “Functional and Molecular Characterization of Patients with Inherited Platelet Disorders” of the Hemorrhagic Diathesis Working Group of the Spanish Society of Thrombosis and Haemostasis; 4 IBSAL, IBMCC, CIC, Universidad de Salamanca-CSIC, Spain; 5Servicio de Pediatría, Hospital Universitario de Salamanca-IBSAL, Spain; 6Servicio de Hematología, Complejo Hospitalario San Pedro Alcántara, Cáceres, Spain; 7Servicio de Pediatría, Hospital Clínico Universitario de Valladolid, Spain; 8Servicio de Hematología y Hemoterapia, Hospital Virgen de la Salud, Complejo Hospitalario de Toledo, Spain; 9Servicio de Hematología, Complejo Asistencial de Soria, Spain; 10Serviço de Imunohemoterapia, Sangue e Medicina Transfusional do Centro Hospitalar e Universitário de Coimbra, EPE, Portugal; 11Servicio Hematología y Hemoterapia, Complejo Hospitalario Universitario A Coruña, Spain; 12Department of Laboratory Medicine, Laboratory of Hematology, Radboud Institute for Molecular Life Sciences, Radboud University Medical Centre, Nijmegen, the Netherlands and 13Birmingham Platelet Group, Institute of Cardiovascular Sciences, College of Medical and Dental Sciences, University of Birmingham, UK 1 2

*

JR and JRG-P contributed equally to this work

Correspondence: jmbastida@saludcastillayleon.es/ jose.rivera@carm.es Received: April 20, 2017. Accepted: September 29, 2017. Pre-published: October 5, 2017. doi:10.3324/haematol.2017.171132 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/148 ©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.

148

ABSTRACT

I

nherited platelet disorders are a heterogeneous group of rare diseases, caused by inherited defects in platelet production and/or function. Their genetic diagnosis would benefit clinical care, prognosis and preventative treatments. Until recently, this diagnosis has usually been performed via Sanger sequencing of a limited number of candidate genes. High-throughput sequencing is revolutionizing the genetic diagnosis of diseases, including bleeding disorders. We have designed a novel highthroughput sequencing platform to investigate the unknown molecular pathology in a cohort of 82 patients with inherited platelet disorders. Thirty-four (41.5%) patients presented with a phenotype strongly indicative of a particular type of platelet disorder. The other patients had clinical bleeding indicative of platelet dysfunction, but with no identifiable features. The high-throughput sequencing test enabled a molecular diagnosis in 70% of these patients. This sensitivity increased to 90% among patients suspected of having a defined platelet disorder. We found 57 different candidate variants in 28 genes, of which 70% had not previously been described. Following consensus guidelines, we qualified 68.4% and 26.3% of the candidate variants as being pathogenic and likely pathogenic, respectively. In addition to establishing definitive diagnoses of well-known inherited platelet disorders, high-throughput sequencing also identified rarer disorders such as sitosterolemia, filamin and actinin deficiencies, and G protein-coupled receptor defects. This included disease-causing variants in DIAPH1 (n=2) and RASGRP2 (n=3). Our study reinforces the feasibility of introducing high-throughput sequencing technology into the mainstream laboratory for the genetic diagnostic practice in inherited platelet disorders. haematologica | 2018; 103(1)


Genetic diagnosis of IPDs by HTS

Introduction Inherited platelet disorders (IPDs) are a heterogeneous group of rare diseases of variable clinical severity, usually characterized by mucocutaneous bleeding and excessive blood loss after trauma or surgery. Some IPDs are due to defects in genes that encode proteins that play critical roles in megakaryopoiesis and proplatelet production, leading to an inherited thrombocytopenia (IT). Alternatively, the molecular pathology may affect the development or maintenance of the platelet ultrastructure, the formation and cargo of granules, or platelet responses to agonists, which are known collectively as inherited platelet function disorders (IPFDs). Not infrequently, IPDs combine thrombocytopenia and impaired platelet function.1,2 Although phenotype-guided diagnosis is straightforward in some IPDs, including Bernard Soulier syndrome (BSS), Glanzmann Thrombasthenia (GT) and HermanskyPudlak syndrome (HPS), due to the severity of the bleeding and a readily available flow cytometry test, or to the

syndromic nature of the disorder, most IPDs lack distinctive clinical and laboratory features. The diagnosis of this group remains a challenge even under expert analysis. Consequently, many IPDs remain underdiagnosed, despite the many diagnostic algorithms that have been proposed.3 Moreover, until recently, IPD diagnosis at the genetic level has been attained in fewer than half of patients, with the greatest success being seen in specialized centers.4,5 Genetic diagnosis for IPD families would facilitate better clinical care, prognosis and preventive treatments, which are especially important for clinically severe IPDs, or for those platelet syndromes that associate with an increasing risk of malignancy, and which require genetic counseling.6-8 While genotyping based on Sanger sequencing of candidate genes has been used successfully since the early 1990s for some IPDs,4,5 the approach is, at present, costly, time-consuming and not applicable to disorders whose phenotype-based diagnosis is not straightforward and for which there is no obvious candidate gene.6-8 The recent advent of high-throughput sequencing (HTS),

Table 1. Genes included in the HTS platform for molecular screening of IPDs.

Target protein

Transcription Factors

Agonist platelet receptors

Platelet granules

Cytoskeletal assembly and structural proteins

Signal transduction

haematologica | 2018; 103(1)

Genes

Chromosome location

Genes

Chromosome location

RBM8A USF1 MPL HOXA11 CYCS GP9 P2RY12 P2RY1 GP5 ITGA2 CD36 ADRA2A A2M LYST MLPH NBEAL2 HPS3 AP3B1 DTNBP1 STX11 PRF1 PLAU HPS1 HPS6 ABCG5 ABCG8 DIAPH1 PRKACG ABCA1 ANKRD26 MASTL GNAI3 PLA2G4 RGS2 DHCR24 TBXAS1 GNAQ

1q21.1 1q23.3 1p34.2 7p15.2 7p15.3 3q21.3 3q25.1 3q25.2 3q29 5q11.2 7q21.11 10q25.2 12p13.31 1q42.3 2q37.3 3p21.31 3q24 5q14.1 6p22.3 6q24.2 10q22.1 10q22.2 10q24.2 10q24.32 2p21 2p21 5q31.3 9q21.11 9q31.1 10p12.1 10p12.1 1p13.3 1q31.1 1q31.2 1p32.3 7q34 9q21.2

GFI1B STIM1 FLI1 RUNX1 GATA1 P2RX1 GP1BA ITGA2B ITGB3 TBXA2R GP6 GP1BB

9q34.13 11p15.4 11q24.3 21q22.12 Xp11.23 17p13.2 17p13.2 17q21.31 17q21.32 19p13.3 19q13.42 22q11.21

HPS5 ANO6 VIPAS39 BLOC1S6 MYO5A RAB27A VPS33B UNC13D STXBP2 BLOC1S3 HPS4 FERMT3 ACTN1 TUBB1 MYH9 WAS FLNA

11p15.1 12q12 14q24.3 15q21.1 15q21.2 15q21.3 15q26.1 17q25.1 19p13.2 19q13.32 22q12.1 11q13.1 14q24.1 20q13.32 22q12.3 Xp11.23 Xq28

PTGS1 RASGRP2 PTS DPAGT1 PLCB2 GNAS

9q33.2 11q13.1 11q23.1 11q23.3 15q15.1 20q13.32 149


J.M. Bastida et al. Table 2. General characteristics of 34 patients with a clinical and biological phenotype suggesting a particular type of IPD.

Case

Sex

Age ISTH-BAT Consanguinity# Platelets (x109/L)

11.

F

49

4

No

72

12.

F

1

1

No

20

13.

M

11

1

No

90

14.

M

37

3

No

50

15.

F

27

3

No

30

16.

F

26

2

No

87

17.

F

48

3

No

30

18.

F

11

0

No

88

19.

M

39

4

No

116

20.

F

62

9

No

25

21.

F

4

2

No

89

22.

F

13

10

No

3 50

23.

M

62

12

No

4

24.

F

15

8

No

A 27

25.

M

32

5

No

45

26.

F

21

1

No

36

27.

M

28

6

No

74

28.

F

40

3

Yes

88

Clinical symptoms, familial background, platelet phenotype

Suspected IPD

Lifelong macrothrombocytopenia with dominant inheritance Deafness Lifelong mild macrothrombocytopenia Neutrophil inclusion bodies in blood smear Lifelong macrothrombocytopenia with dominant inheritance Neutrophil inclusion bodies in blood smear Lifelong macrothrombocytopenia Neutrophil inclusion bodies in blood smear Lifelong macrothrombocytopenia with dominant inheritance Neutrophil inclusion bodies in blood smear Mild thrombocytopenia detected at pre-surgical screening Deafness, mild renal disease Lifelong macrothrombocytopenia in several family members Sensorineural hearing loss in several family members Neutrophil inclusion bodies in blood smear Mild macrothrombocytopenia and neutropenia in several family members Sensorineural hearing loss in several family members No apparent neutrophil inclusion bodies in blood smear Normal expression of major GPs. Normal platelet aggregation and granule secretion to moderate doses of several agonists Mild macrothrombocytopenia and neutropenia in several family members Sensorineural hearing loss in several family members No apparent neutrophil inclusion bodies in blood smear Normal expression of major GPs Lifelong macrothrombocytopenia Severely impaired platelet aggregation with ristocetin Undetectable expression of GPIb/IX Lifelong macrothrombocytopenia Severely impaired platelet aggregation with ristocetin Normal platelet aggregation with other agonist 75% reduction in GPIb/IX platelet expression 2-fold increased expression of GPIIIa and GPIa 30% ristocetin-induced VWF binding Lifelong macrothrombocytopenia <10% GPIbα expression <10% ristocetin-induced VWF binding Lifelong severe macrothrombocytopenia Absent GPIb/IX platelet expression Mild increased expression of GPIIa, GPIa and GPVI Absent ristocetin-induced VWF binding Lifelong macrothrombocytopenia Reduced agonist-induced P-selectin exposure Decreased α-granules under electron microscopy Lifelong macrothrombocytopenia Reduced number of α-granules under electron microscopy Lifelong macrothrombocytopenia Normal or mild reduced platelet aggregation with several agonists Marked impairment of agonist-induced P-selectin secretion Lifelong macrothrombocytopenia Mildly reduced platelet aggregation with several agonists Impaired agonist-induced P-selectin secretion Lifelong macrothrombocytopenia Presence of stomatocytes in blood smear Xanthelasma

MYH9-RD

MYH9-RD MYH9-RD

MYH9-RD MYH9-RD MYH9-RD MYH9-RD

DIAPH1-RD

DIAPH1-RD

BSS

BSS

BSS

BSS

GPS

GPS GPS

GPS

STSL

continued on the next page

150

haematologica | 2018; 103(1)


Genetic diagnosis of IPDs by HTS continued from the previous page

Age ISTH-BAT Consanguinity# Platelets (x109/L)

Case

Sex

Clinical symptoms, familial background, platelet phenotype

Suspected IPD

29.

M

1

2

No

35

Syndromic thrombocytopenia Mental retardation, corpus callosum hypoplasia, facial and cardiac abnormalities

IT DiGeorge syndrome

30. 31.

F F

46 13

10 2

No No

84 40

TAR FLNA-RD

5

No

59

4

4

No

87

M

45

5

Yes

160

35.

F

27

10

Yes

175

36.

M

7

6

Yes

220

37.

F

8

5

Unknown

238

Lifelong macrothrombocytopenia and absence of radii Lifelong thrombocytopenia with mildly increased platelet size Facial abnormalities and corpus callosum hypoplasia Lifelong thrombocytopenia with normal platelet size Family history of acute myeloid leukemia Mildly reduced platelet aggregation with several agonists Slightly prolonged PFA-100 closure times Lifelong microthrombocytopenia Recurrent infections Decreased WASP expression by flow cytometry Severely impaired platelet aggregation with TRAP, ADP, epinephrine, arachidonic acid, and collagen. 50% reduced ristocetin-induced platelet aggregation PFA-100 >300s; abnormal clot retraction Undetectable GPIIb/IIIa platelet expression. Normal expression of GPIb/IX, GPIa and GPVI Absence of agonist-induced PAC-1 binding Severely reduced platelet aggregation with several agonists except ristocetin. Marked defect in GPIIb/IIIa expression. Normal expression of GPIb/IX, GPIa and GPVI Severely reduced platelet aggregation with several agonists except ristocetin. Marked defect in GPIIb/IIIa expression. Normal expression of GPIb/IX, GPIa and GPVI Severely impaired platelet aggregation with TRAP, ADP, epinephrine, arachidonic acid and collagen. 50% reduced ristocetin-induced platelet aggregation PFA-100 >300s; Abnormal clot retraction Undetectable GPIIb/IIIa platelet expression Normal expression of GPIb/IX, GPIa and GPVI <20% agonist-induced fibrinogen binding

32.

F

42

33.

M

34.

38.

F

50

22

No

158

39.

M

55

23

No

40.

F

21

5

No

41.

F

18

12

No

RUNX1-RD

WAS GT

GT

GT

GT

Severely impaired platelet aggregation with TRAP, ADP, epinephrine, arachidonic acid, and collagen. Impaired ristocetin-induced platelet aggregation PFA-100 >300s; Abnormal clot retraction Undetectable GPIIb/IIIa platelet expression Normal expression of GPIb/IX, GPIa and GPVI Markedly impaired agonist-induced fibrinogen Mildly impaired agonist-induced P-selectin and CD63 secretion GT 313 Severely impaired platelet aggregation with TRAP, ADP, epinephrine, arachidonic acid, and collagen. Normal ristocetin-induced GT platelet aggregation PFA-100 >300s; Abnormal clot retraction 85% reduction in GPIIb/IIIa platelet expression. Normal expression of GP1b/IX, GPIa and GPVI Markedly impaired agonist-induced fibrinogen binding 230 Reduced platelet aggregation with low dose collagen and CRP GPVI defect Normal platelet aggregation with other agonists Normal expression of GP IIb/IIIa, GPIb/IX and GPIa 50% expression of GPVI . Severely reduced CRP-induced fibrinogen binding and P-selectin secretion, but normal with other agonists Similar platelet phenotype in mother 201 Albinism and granulomatous colitis Mildly impaired platelet aggregation with low dose of ADP, HPS epinephrine and ristocetin. Impaired serotonin uptake Reduced number of dense granules under electron microscopy continued on the next page

haematologica | 2018; 103(1)

151


J.M. Bastida et al. continued from the previous page

Case

Sex

Age ISTH-BAT Consanguinity# Platelets (x109/L)

42.

M

40

10

Yes

200

43.

M

27

4

Unknown

312

44.

M

1

1

Unknown

180

Clinical symptoms, familial background, platelet phenotype

Suspected IPD

Prolonged PFA-100 closure times Albinism and pulmonary fibrosis Albinism and strabismus Reduced serotonin uptake Impaired agonist-induced CD63 secretion Reduced dense granules by electron microscopy Albinism Neutrophil inclusion bodies in blood smear Impaired agonist-induced CD63 secretion. Normal agonist-induced P-selectin release

HPS

HPS

CHS

# No information was provided regarding familiar consanguinity and no further investigations were carried out due to ethical reasons. MYH9-RD: MYH9 related disease; IT: inherited thrombocytopenia; RUNX1-RD: RUNX1-related disease; GPS: Gray platelet syndrome; HPS: Hermansky-Pudlak syndrome; GT: Glanzmann thrombasthenia; BSS: Bernard Soulier syndrome; FLNA-RD: filaminopathy or filamin-related disease; CHS: Chediak-Higashi syndrome; WAS: Wiskott-Aldrich syndrome; GPVI: glycoprotein VI; STSL: sitosterolemia; TAR: thrombocytopenia with absent radius syndrome; WASP: Wiskott-Aldrich syndrome protein; TRAP: thrombin receptor activating peptide; ADP: adenosine diphosphate; CRP: collagen-related peptide; PFA: platelet function analyser;VWF: von Willebrand factor; GP: glycoprotein; IPD: inherited platelet disorder; ISTH-BAT: International Society on Thrombosis and Haemostasis Bleeding Assessment Tool.

either whole-genome sequencing (WGS), whole-exome sequencing (WES), and targeted sequencing of pre-specified genes, has begun to revolutionize the field of genetic diagnosis and is rapidly becoming established in clinical practice.9,10 With respect to bleeding disorders, HTS is emerging as a valuable tool for the molecular diagnosis of hemophilia and other rare coagulation disorders, and also for genotyping in IPDs.9,11 Thus, HTS is increasingly important as the first-line of the diagnostic investigation of these diseases.3,6,8,12,13 Herein, we report the design and implementation of a multigenic HTS platform for genetic analysis of IPD patients. The application of this novel approach has greatly aided our diagnostic process, resulting in a conclusive molecular diagnosis in the largest series of IPD patients investigated in the Iberian Peninsula.

Methods Patient enrollment, blood sampling, platelet phenotyping and DNA isolation We studied 92 unrelated patients with suspected IPD, enrolled prospectively within the aims of the multicenter “Functional and Molecular Characterization of Patients with Inherited Platelet Disorders” project, under the scientific sponsorship of the Spanish Society of Thrombosis and Haemostasis. Patients with known acquired disorders or clotting factor defects were excluded. Investigations abided by the Declaration of Helsinki and were approved by the Local Ethical Committees of the Instituto de Investigación Biomédica (IBSAL, Salamanca, Spain) and Hospital Universitario Reina Sofía (Murcia, Spain). All patients gave their written informed consent. Clinical data of all patients were reviewed by the investigating team and their bleeding symptoms were reassessed and scored using the International Society on Thrombosis and Haemostasis Bleeding Assessment Tool (ISTH-BAT) questionnaire. Venous blood samples were obtained for platelet phenotyping4 and molecular studies (Online Supplementary Methods). 152

Custom HTS panel design, sequencing and data analysis Using Design Studio (Illumina, San Diego, CA, USA), we designed a HTS platform with 1399 probes targeting 1106 regions, including exons, untranslated regions (UTRs) and flanking regions, of 72 genes that are associated with IPDs and/or are significant in platelet physiology (Table 1). Sequence data generated by HTS was mapped to the Reference Human Genome (hg19) with MiSeq integrated computer software (MiSeq Reporter, Illumina), which uses a Burrows-Wheeler-Aligner (BWA) and Genome Analysis Tool Kit (GATK) for variant-calling of single nucleotide variants (SNVs) and short insertions/deletions (InDels).14,15 Secondary data analysis, sequence alignment and variant detection was performed with Variant Studio Data Analysis and Integrative Genomics Viewer (IGV) (Broad Institute, Cambridge, MA, USA) software. The coverage per base was ≥100 reads. The first step consisted of a quality filter based on a Phred score >20, Quality >20 and Read coverage >30 at each position within the reads, to indicate high sequence quality. Data was then filtered according to the severity of the consequence, in order to prioritize variants leading to an amino acid change in the protein sequence (missense, nonsense, frameshift) and those in the splice site and UTRs. Apart from exceptional cases, synonymous and intronic variants were disregarded. Minor allelic frequencies (MAFs) were consulted in the Exome Variant Server, 1000 Genomes Browser and exome aggregation consortium (ExAC) databases; variants with a MAF of <0.05 were selected for further analysis. The other variants were searched for across sources, such as the dbSNP147, the Catalog of Somatic Mutations in Cancer (COSMIC), the National Center for Biotechnology Information (NCBI) ClinVar, the HGMD professional database, PubMed, Online Mendelian Inheritance in Man (OMIM), and locus-specific mutation databases in an attempt to identify variants known to cause IPDs (Online Supplementary Figure S1).16 Several in silico tools, Polymorphism Phenotyping v2 (PolyPhen-2), Sorting Intolerant From Tolerant (SIFT), Mutation Taster, MutationAssessor and Functional Analysis Through Hidden Markov Models (fathmm), were used to predict the functional effects and pathogenicity of the novel variants. We followed the guidelines of the American College of haematologica | 2018; 103(1)


Genetic diagnosis of IPDs by HTS

Figure 1. Classification of the 92 IPD patients sequenced with a novel HTS platform. Ninety-two unrelated patients with a suspicion of IPD were enrolled in the project “Functional and Molecular Characterization of Patients with Inherited Platelet Disorders”. Patients fell into 2 main groups: on the left, a validation group comprising 10 IPD patients harboring known pathogenic variants identified by Sanger sequencing (Online Supplementary Table S1), and on the right, a study group of 82 IPD patients with unknown molecular pathology. DNA from all patients was sequenced with an HTS platform targeting 72 genes (Table 1), as described in the Methods. The identified genetic variants were prioritized and assessed for pathogenicity, as stated in the Methods. IPD: inherited platelet disorder; HTS: high-throughput sequencing.

Medical Genetics and Genomics and Association for Molecular Pathology,17 to qualify each identified variant as a “pathogenic variant” (PV), “likely pathogenic variant” (LPV) or “variant of uncertain significance”(VUS; Online Supplementary Figure S2). For more methodological details, see the Online Supplementary Methods.

Results Patient characteristics Patients were divided into 2 groups (Figure 1): i) a small validation group comprising 10 patients previously characterized at the functional and molecular level,4,18 harboring known pathogenic variants identified by Sanger sequencing (Online Supplementary Table S1), and ii) a larger study group of 82 patients who were enrolled in the project because of variable bleeding diathesis and abnormalities in the number or function of platelets, or both, but with unknown molecular pathology. Only 3 (3.7%) of these patients had undergone Sanger sequencing of candidate genes, and this had failed to identify candidate disease-causing variants. The major clinical and biological characteristics of the 82 patients in the study group are summarized in Table 2 and Table 3. The majority (62.2%) were female; median age was 29 (1-82) years; median bleeding score was 5 (023); median platelet count was 96 (4-617) x109/L. Fiftythree cases (64.6%) presented with lifelong thrombocytopenia as the main inclusion criterion and the others had haematologica | 2018; 103(1)

laboratory abnormalities consistent with an IPFD. Forty patients (50%) had a family history of bleeding, thrombocytopenia and/or hematological malignancies. In 34 cases (41.5%), clinical background and centralized laboratory assessment supported the diagnosis of a particular type of IT or IPFD: MYH9 related disease (MYH9-RD), n=7; DIAPH1 related disease (DIAPH1-RD), n=2; BSS, n=4; Gray platelet syndrome (GPS), n=4; sitosterolemia (STSL), n=1; DiGeorge syndrome, n=1; thrombocytopenia with absent radius syndrome (TAR), n=1; filaminopathy or filamin-related disease (FLNA-RD), n=1; RUNX1-related disease (RUNX1-RD), n=1; Wiskott-Aldrich syndrome (WAS), n=1; GT, n=6; glycoprotein VI (GPVI) signalling defect, n=1; HPS, n=3; Chediak-Higashi syndrome (CHS=1), n=1. The remaining 48 patients (58.5%) either had low platelet counts (n=30) and/or platelet function abnormalities (n=18) of uncertain etiology (Table 2 and Table 3).

General performance of the HTS assay and validation of the HTS multigenic platform The 72 genes included in the panel (Table 1) were analyzed for all patients. We successfully sequenced 95.6% of the 1106 target regions at a minimum coverage depth of 100 for each nucleotide base-pair position of interest. The remaining 4.4% of regions could not be sequenced with adequate coverage using the NGS platform (Online Supplementary Table S2). In addition, the mean fraction of exonic bases covered at 20X and 50X was 0.991 and 0.987, respectively. 153


J.M. Bastida et al.

Table 3. Clinical and biological characteristics of 48 patients with IPD of uncertain etiology.

Case Sex Age ISTH-BAT Consanguinity# Platelets (x109/L) 45.

F

53

2

No

138

46. 47.

M M

80 30

1 6

No No

98 28

48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58.

F F M F F F F M F F F

22 35 39 31 14 21 22 74 16 4 15

5 4 0 6 2 2 5 2 2 4 3

No No No No No No No No No No No

100 101 87 67 85 70 123 69 80 115 94

59. 60. 61.

F M F

53 37 43

13 0 11

No No No

139 35 135

62.

F

54

13

No

140

63.

F

40

7

No

99

64.

F

43

1

No

60

65.

M

45

9

No

25

66.

M

5

6

No

9

67.

M

56

4

No

50

68.

M

6

4

No

86

69.

F

41

7

No

72

70.

M

40

10

Unknown

65

Clinical symptoms, familial background, platelet phenotype

Suspected IPD

Lifelong macrothrombocytopenia in several family members Normal platelet GPs expression Normal platelet aggregation Lifelong macrothrombocytopenia in several family members Lifelong macrothrombocytopenia in several family members Severely reduced platelet aggregation with ristocetin. Normal platelet aggregation with other agonists 50% reduction in GPIb/IX expression. Increased expression of GPIIa and GPIa 50% reduced ristocetin-induced VWF binding Lifelong macrothrombocytopenia in several family members Lifelong macrothrombocytopenia in several family members Lifelong macrothrombocytopenia in several family members Lifelong macrothrombocytopenia in severalfamily members Lifelong macrothrombocytopenia in several family members Lifelong macrothrombocytopenia in several family members Lifelong macrothrombocytopenia in several family members Lifelong macrothrombocytopenia in several family members Lifelong macrothrombocytopenia in several family members Lifelong thrombocytopenia in several family members Lifelong thrombocytopenia in several family members Normal platelet size Lifelong thrombocytopenia in several family members Lifelong thrombocytopenia with dominant inheritance Lifelong mild thrombocytopenia in several family members Mildly reduced platelet aggregation with low dose ADP and epinephrine Lifelong macrothrombocytopenia Mildly reduced platelet aggregation with most agonists Lifelong thrombocytopenia with slightly increased platelet size Mildly reduced platelet aggregation with low dose ADP and epinephrine Lifelong macrothrombocytopenia in the patient and her mother Mildly reduced GPIb/IX expression Lifelong thrombocytopenia in several family members Reduced number of Îą-granules by electron microscopy Lifelong macrothrombocytopenia Lack of response to previous ITP treatments Lifelong thrombocytopenia in several family members Renal disease in the patient and his father Lifelong thrombocytopenia Mildly reduced platelet aggregation with ADP and TRAP Reduced TRAP-induced P-selectin and CD63 expression Lifelong thrombocytopenia with slightly increased platelet size Significantly reduced platelet aggregation with epinephrine and mildly decreased with other agonist Impaired PAR1-induced granule secretion Lifelong macrothrombocytopenia Mitral regurgitation, Pericentral retinitis pigmentosa Mental retardation Impaired agonist-induced fibrinogen binding and P-selectin release One sister also with thrombocytopenia and retinitis pigmentosa

IT

IT IT

IT IT IT IT IT IT IT IT IT IT IT IT IT IT IT IT IT IT IT IT IT

IT

IT

continued on the next page

154

haematologica | 2018; 103(1)


Genetic diagnosis of IPDs by HTS continued from the previous page

Case Sex Age ISTH-BAT Consanguinity# Platelets (x109/L) 71.

F

72

22

Unknown

72.

M

45

12

No

73.

M

26

2

No

74.

M

16

5

Unknown

75.

F

82

9

No

76.

F

17

3

Yes

77.

F

21

6

No

78.

F

54

7

No

79.

F

55

10

No

80.

F

66

9

No

81.

M

36

5

No

82.

F

53

10

No

83.

F

9

3

No

84.

F

36

8

No

85.

M

1

4

No

86.

F

13

3

No

87.

F

6

3

No

88.

F

4

2

Unknown

89.

F

39

14

No

haematologica | 2018; 103(1)

Clinical symptoms, familial background, platelet phenotype

Suspected IPD

36

Lifelong thrombocytopenia IT Relatives affected by hemophilia Normal GPs expression Slightly reduced platelet aggregation with several agonists 12 Lifelong thrombocytopenia in several family members IT Normal platelet size, myelodysplasia 108 Lifelong thrombocytopenia, normal platelet volume IT Reduced platelet aggregation with low dose of ADP, epinephrine, and collagen Normal platelet GPs expression except GPIa (30%) Normal agonist-induced platelet a P-selectin and CD63 secretion 302 Mildly reduced platelet aggregation with ADP Platelet signaling/ and epinephrine secretion defect 12 Lifelong macrothrombocytopenia IT & Normal expression of major GPs Platelet signaling/ Defective agonist-induced P-selectin and CD63 secretion secretion defect 280 Reduced platelet aggregation with, ADP, epinephrine and low dose collagen Normal platelet aggregation with other agonists PFA-100 >300s; Normal clot retraction Normal expression of GPIIb/IIIa, GPIb/IX, and GPIa Mildly reduced agonist-induced P-selectin, CD63 and serotonin release Impaired PAC-1 binding with ADP but normal with PMA GPIIb/IIIa activation defect 215 Slightly reduced platelet aggregation with ADP and epinephrine Platelet signaling/ secretion defect 195 Slightly reduced platelet aggregation with ADP, Platelet signaling/ epinephrine and arachidonic acid secretion defect 315 Slightly reduced platelet aggregation with collagen, Platelet signaling/ ADP, and epinephrine secretion defect 220 Mildly reduced platelet aggregation with most agonists Platelet signaling/ secretion defect 198 Mildly reduced platelet aggregation with epinephrine and collagen Platelet signaling/ secretion defect 190 Mildly reduced platelet aggregation with ADP, and ristocetin Platelet signaling/ secretion defect 185 Mildly reduced platelet aggregation with ADP and ristocetin Platelet signaling/ secretion defect 226 Mildly reduced platelet aggregation with ADP and epinephrine Platelet signaling/ secretion defect 617 Mildly reduced platelet aggregation with ADP and epinephrine Platelet signaling/ secretion defect 182 Mildly reduced platelet aggregation with ADP and epinephrine Platelet signaling/ secretion defect 294 Slightly reduced platelet aggregation with ADP and epinephrine Platelet signaling/ secretion defect 223 Mildly reduced platelet aggregation with most agonists Platelet signaling/ Normal expression of major platelet GPs secretion defect Defective agonist-induced P-selectin, CD63 and serotonin release 267 Markedly reduced platelet aggregation with low dose ADP, Platelet signaling collagen and CRP. Normal platelet aggregation with other agonists defect Normal expression of GP IIb/IIIa, GPIb/IX, GPIa, and GPVI Reduced agonist-induced fibrinogen binding with low ADP and CRP Normal agonist induced P-selectin and serotonin secretion continued on the next page 155


J.M. Bastida et al. continued from the previous page

Case Sex Age ISTH-BAT Consanguinity# Platelets (x109/L) 90.

M

9

7

Yes

315

91.

M

8

7

Unknown

193

92.

F

4

9

NO

309

Clinical symptoms, familial background, platelet phenotype

Suspected IPD

Markedly reduced platelet aggregation with ADP, epinephrine and low-dose collagen Mildly reduced aggregation with other agonists PFA-100 >300s; normal clot retraction Normal expression of GPIIb/IIIa, GPIb/IX, GPIa and GPVI Severely reduced fibrinogen binding with all agonists but PMA Markedly reduced platelet aggregation with ADP, epinephrine, low-dose collagen and CRP Mildly reduced aggregation with other agonists PFA-100 >300s; normal clot retraction Normal expression of GPIIb/IIIa, GPIb/IX, GPIa and GPVI Reduced fibrinogen binding with all agonists but PMA Mildly reduced platelet secretion Markedly reduced platelet aggregation with all agonists PFA-100 >300s; normal clot retraction Normal expression of GPIIb/IIIa, GPIb/IX, GPIa and GPVI Reduced fibrinogen binding with all agonists but PMA Mildly reduced platelet secretion

GPIIb/IIIa activation defect43

GPIIb/IIIa activation defect

GPIIb/IIIa activation defect

# No information was provided regarding familiar consanguinity and no further investigations were carried out due to ethical reasons. IT: inherited thrombocytopenia; WAS: Wiskott-Aldrich syndrome; GPVI: glycoprotein VI; TRAP: thrombin receptor activating peptide; ADP: adenosine diphosphate; CRP: collagen-related peptide; PFA: platelet function analyser; VWF: von Willebrand factor; GP: glycoprotein; IPD: inherited platelet disorder; ISTH-BAT: International Society on Thrombosis and Haemostasis Bleeding Assessment Tool; ITP: Idiopathic Thrombocytopenic Purpura; PMA: phorbol 12-myristate 13-acetate.

To validate the accuracy of our HTS platform for detecting causative variants within these genes, we assayed, in a blind manner, DNA from the 10 patients with previously ascertained pathogenic variants by Sanger sequencing. These included 12 SNV (9 missense variants and 3 nonsense changes) and 1 deletion within 8 genes (Online Supplementary Table S1).4,18 In each case the HTS test, and accompanying data analysis, identified the previously known genetic variant, thus demonstrating the high sensitivity of the platform. Reproducibility studies were also performed, in which 4 DNA samples from the validation groups (Cases 1, 4, 8 and 10) were assayed in triplicate in 3 separate runs. After applying the prioritization protocol, we found 100% concordance between runs in detecting the correspondent variants present in each DNA, demonstrating high intra-batch and inter-batch reproducibility of the platform.

General performance of the HTS multigenic platform in the IPD study group The high sensitivity and reproducibility displayed by our HTS platform in the validation group prompted us to use it as the first genotyping method in the 82 patients of the study group. As stated above, this included 34 (41.5%) cases with a strong suspicion of a defined IPD, and 48 (58.5%) patients with phenotypes not suggestive of a particular type of IPD (Figure 1). Overall, by applying the bioinformatic tools described in the Methods, there was a median of 169 (range: 128230) sequence variations across the 72 genes in the 82 patients. We prioritized 62 of these candidate variants in 56 (68.3%) index cases (Table 4 and Table 5). This includ156

ed 40 missense variants, 8 nonsense variants, 11 frameshift variants (deletion=11, Insertion-Deletion=1), 2 variants in the UTR region and 1 splice site variant within 29 genes. All these genetic changes were confirmed by Sanger sequencing and segregated into available family members. Variants were inherited in a homozygous/hemizygous manner in 14 patients, 4 cases were compound heterozygous defects and the other 38 were heterozygous. Forty-one (72%) variants had not been previously reported in public databases and are therefore novel variants. Assessment of variant pathogenicity, following consensus guidelines,17 led to 43 (69%) variants being classified as PV, 16 (26%) as LPV and 3 as VUS (5%) (Table 4 and Table 5). As expected, the HTS platform was highly sensitive in detecting causal variants among patients with a strong suspicion of defined IPDs. In 30 out of 34 (88.2%) of these patients we identified 34 different candidate variants in 17 genes, which were identified in most cases as PVs (n=28), but in 4 cases as LPVs and in 2 as VUS (Table 4). These genetic findings gave rise to confirmation of diagnosis of BSS (n=4), MYH9-RD (n=7), TAR (n=1), GPS (n=1), FLNA-RD (n=1), RUNX1-RD, (n=1), GT (n=5), HPS (n=3), CHS (n=1), GPVI defect (n=1), and 2 rare cases of WAS and STSL that we have recently reported in detail (Online Supplementary Table S3). In 2 patients (Cases 18 and 19) with macrothrombocytopenia, mild neutropenia and familiar sensorineural hearing loss, the HTS identified the p.Arg1213* variant in DIAPH1. In these families, the p.Arg1213* variant displayed full penetrance with deafness, minor impact on platelet functional status, and a moderate effect on the platelet levels of DIAPH1 and haematologica | 2018; 103(1)


Genetic diagnosis of IPDs by HTS Tubulin β1 (Online Supplementary Figure S3, S4, S5 and S6). Moreover, HTS failed to identify PV and LPV variants in 1 patient with what was previously thought to be a straightforward diagnosis of GT (Case 34) and in 3 cases with a suspicion of GPS (Cases 24-27; Table 2). Furthermore, in Case 27, who had suspected GPS, we found 2 novel missense variants in STIM1 and RUNX1 (Table 4). These variants, similarly to that of a variant in GNAS in Case 86 (Table 5), were classified as VUS on the basis of their identification only in the index case, since no relatives were

available for screening, there are no previous reports in other patients, and no specific studies demonstrating a deleterious functional effect.17 The overall sensitivity of the HTS platform was significantly lower (Z-score=3.0599; P=0.00222) among the subgroup of IPD of uncertain etiology. Herein, we identified 26 different variants located in 16 genes in 26 cases (54.2%). Most of them were classified as PV or LPVs (12 and 13, respectively) and one was identified as VUS. Accordingly, these patients were assigned a diagnosis of

Table 4. Genetic variants identified with the HTS test in patients with suspicion of particular IPDs according to their clinical and biological phenotype.

Case

Gene

Status

cDNA mutation

Protein change

Reference

MAF ExAC

MAF 1000 Genome

Variant assessment*

11 12 13 14 15 16 17 18 19 20 21 22 23

MYH9 MYH9 MYH9 MYH9 MYH9 MYH9 MYH9 DIAPH1 DIAPH1 GP1BA GP1BA GP1BB GP1BB

Het Het Het Het Het Het Het Het Het Hom Hom Hom C. Het

Pathogenic Pathogenic Pathogenic Pathogenic Pathogenic Pathogenic Pathogenic Pathogenic Pathogenic Pathogenic Pathogenic Pathogenic Both pathogenic

28

Hom Het Het C. Het

Novel Novel Novel 33 33 33 Novel 34 34 4,26 Novel Novel Novel Novel 48 Novel Novel 36

No data No data No data No data No data No data No data No data No data No data No data No data No data

NBEAL2 STIM1 RUNX1 ABCG5

30

RBM8A

Hom

FLNA RUNX1 WAS ITGA2B ITGA2B ITGA2B

Het Het Hem Hom Hom C. Het

No data No data No data No data 0.000015 0,034 No data No data No data No data 0 No data

No data No data No data No data 0,00004143 0,056 No data No data No data No data 0 No data

Pathogenic Both uncertain significance Pathogenic Likely pathogenic Pathogenic#

31 32 33 35 36 37

Likely pathogenic Likely pathogenic Pathogenic Pathogenic Pathogenic Both Pathogenic

38 39 40 41 42 43 44

ITGA2B ITGB3 GP6 HPS4 HPS4 HPS3 LYST

Hom Hom Het Hom Hom Het C. Het

p.Arg903Gly p.Lys74Glu p.Arg1417Gln p.Asp1925Thr fs*23 p.Arg1933* p.Ser96Leu p.Arg1162Ser p.Arg1213* p.Arg1213* p.Cys225Ser p.Leu155Val p.Pro27Leu p.Trp148* p.Leu167Pro p.Arg2071Pro p.Gln188Arg p.Gly199Ala p.Phe630Leu fs*8 p.Thr305Arg p.Thr1232Ile p.Asn159Ser --p.Gln778Pro p.Ala688Val p.Ala989Pro fs*142 p.Val982Met p.Ala989Pro fs*142 p.Met144Arg p.Asn236Lys fs*105 p.Pro685Leu fs*17 p.Leu91Pro p.Arg822* p.Lys3367Arg fs*34 p.Lys3365Asn p.Leu2379Pro

No data No data No data No data No data No data No data No data No data No data No data No data No data

24 27

c.2707C>G c.220A>G c.4250G>A c.5773delG c.5797C>T c.287C>T c.3486G>T c.3637C>T c.3637C>T c.673T>A c.463C>G c.80C>T c.443G>A c.500T>C c.6212G>C c.563A>G c.596G>C c.1890delT c.914C>G c.-21G>A (+deletion 1q21.1) c.3695C>T c.476A>G c.449+5G>A c.2333A>C c.2063C>T c.2965delG c.2944G>A c.2965delG c.431T>G c.708_711delCGAA c.2054delC c.272T>C c.2464C>T c.10100delA c.10095G>C c.7136T>C

No data No data 0.000083 No data No data No data No data

No data No data No data No data No data No data No data

Pathogenic Pathogenic Pathogenic Pathogenic Pathogenic Likely pathogenic All Pathogenic

22 Novel Novel 49 50 Novel Novel 51 Novel 52 Novel 53 Novel Novel All Novel

*Pathogenicity assessment was assessed following the guidelines of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.17 Pathogenic in combination with deletion of 1q21.1.22 MAF: minor allele frequency; cDNA: complementary DNA; ExAC: The Exome Aggregation Consortium.

#

haematologica | 2018; 103(1)

157


J.M. Bastida et al. monoallelic BSS (n=5); tubulin β1-related thrombocytopenia (TUBB1-RT), n=5; RUNX1-RD, n=2; actinin-1 related thrombocytopenia (ACTN1-RT), n=1; type-2 thrombocytopenia (ANKRD26-RT), n=1; WAS, n=1. The other cases presumably had congenital defects in the gene encoding for ADP receptor P2Y12 (P2RY12)(n=1); thromboxane A2 receptor gene (TBXA2R) (n=1); Ca2+ DAG-regulated guanine nucleotide exchange factor I or CalDAG-GEFI (RASGRP2) (n=3) (Online Supplementary Table S3); G-protein α subunit or Gs-α (GNAS) (n=1); phospholipase β2 (PLCB2) (n=1); prostaglandin-endoperoxide synthase 1 (PTGS1) (n=2); thromboxane A synthase 1 (TBXAS1) (n=1); and integrin α2 (ITGA2)(n=1); (Table 5).

Discussion In recent years, the identification of the underlying molecular pathology has become the cornerstone for establishing a conclusive diagnosis of IPDs, leading to better clinical care and follow-up of these patients.6,7 Until recently, Sanger sequencing of candidate genes and linkage studies have been the main tools for IPD genetic diagnosis, providing outstanding but limited results.4,5 In 2010, HTS emerged

as the herald of the revolution in genetic diagnosis of human diseases, including IPDs.6-8,13 The study herein demonstrates the feasibility of an inhouse designed HTS platform for rapid genetic characterization of patients with IPDs in a clinical setting. Indeed, 60% of patients in our study had no diagnostic features allowing for a straightforward selection of candidate genes and the use of Sanger sequencing. Further, for those patients who presented with a phenotype indicative of a particular type of disorder, most were associated with diseases that can be due to defects in several genes and/or in large genes, thus hampering Sanger sequencing (Table 4). We have already used a detailed phenotyping and Sanger sequencing approach for this group of IPD patients,4 and it compares negatively with our current HTS test both in terms of time and cost, although we did not undertake a full cost-benefit analysis. The current version of this platform allows for multiplex analysis of coding and selected non-coding regions of 72 genes including those previously associated with IPDs (Table 1). However, it can be easily modified to include additional genes such as those recently identified: SLFN14, ETV6, and SCR.2,6,13 None of the patients in this IPD study had phenotypes consistent with these genes, but they are

Table 5. Genetic variants identified with the HTS test in the cohort of patients with IPDs of uncertain etiology on the basis of clinical and biological phenotype.

Case 45 46 47 48 49 51 52 53 54 55 64 66 72 73 74 77 78 79 82 84 86 87 89 90 91 92

Gene

Status

cDNA mutation

Protein change

Reference

MAF ExAC

MAF 1000 Genome

Variant assessment*

GP1BA GP1BA GP1BA GP1BB GP1BB TUBB1 TUBB1 TUBB1 TUBB1 TUBB1 ACTN1 WAS ANKRD26 RUNX1 RUNX1 P2RY12 TBXA2R ITGA2 TBXAS1 PTGS1 PTGS1 GNAS PLCB2 RASGRP2 RASGRP2 RASGRP2

Het Het Het Het Het Het Het Het Het Het Het Hem Het Het Het Het Het Het Het Het Het Het Het Hom Hom Hom

c.463C>G c.673T>A c.1474delA c.119G>A c.1A>T c.1075C>T c.319A>C c.319A>C c.1267C>T c.35delG c.137G>A c.802delC c.-118C>T c.802C>T c.1205C>T c.835G>A c.620C>T c.3472T>C c.1181C>T c.35_40delTCCTGC c.428A>G c.1276G>C c.1303G>A c.1142C>T c.706C>T c.887G>A

p.Leu155Val p.Cys225Ser p.Thr494Pro fs*59 p.Gly40Glu p.Met1Leu p.Arg359Trp p.Thr107Pro p.Thr107Pro p.Gln423* p.Cys12Leu fs*12 p.Arg46Gln p.Arg268Gly fs*40 p.Gln268* p.Ser402Phe p.Val279Met p.Ser207Leu p.Phe1158Leu p.Thr394Ile p.Leu13_Leu14del p.Asn143Ser p.Ala462Pro p.Phe435Lys p.Ser381Phe p.Gln236* p.Cys296Tyr

Novel 4,26 Novel Novel 31 Novel Novel Novel Novel Novel 54 32 38 Novel Novel Novel Novel Novel Novel Novel Novel Novel Novel 43 Novel Novel

No data No data No data 0,0044 No data No data No data No data No data No data No data No data No data No data 0 No data No data No data No data No data No data No data 0 No data No data No data

No data No data No data 0,0047 No data No data No data No data No data No data No data No data No data No data 0 No data No data No data No data No data No data No data No data No data No data No data

Pathogenic Pathogenic Likely pathogenic Likely pathogenic Likely pathogenic Pathogenic Likely Pathogenic Likely Pathogenic Pathogenic Pathogenic Pathogenic Pathogenic Pathogenic Likely pathogenic Likely pathogenic Likely pathogenic Likely pathogenic Likely pathogenic Likely pathogenic Likely pathogenic Likely pathogenic Uncertain significance Likely pathogenic Pathogenic Pathogenic Pathogenic

*Pathogenicity assessment was performed following the guidelines of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.17 MAF: minor allele frequency; cDNA: complementary DNA; ExAC: The Exome Aggregation Consortium.

158

haematologica | 2018; 103(1)


Genetic diagnosis of IPDs by HTS

of interest for novel cases referred to our project. The ThromboGenomics platform does not currently include these genes, but patients with gain-of-function variants in SCR have been reported.19 We first performed a validation study of previously ascertained genetic variants in blinded samples, which confirmed the strong analytical sensitivity and reproducibility of the HTS platform, and the appropriateness of our variant-filtering strategy. We detected the 13 known variants in eight genes, including missense and nonsense variants and small deletions, from 10 patients (Online Supplementary Table S1). Subsequently, we used the HTS multigenic platform for the first genetic analysis of a cohort of 82 patients prospectively enrolled from different hospitals in order to realize the aim of the collaborative project “Functional and Molecular Characterization of Patients with Inherited Platelet Disorders”. The clinical and platelet phenotypic presentation was highly variable among these patients, which is consistent with the widely recognized heterogeneity of IPDs. Most patients (65%) presented with thrombocytopenia as their main hematological feature. Many of them also displayed a variable degree of platelet dysfunction consistent with recent findings in other IT series.20 Not surprisingly, the majority of patients were women (62.2%), as more frequent bleeding complications in females facilitate their clinical identification. Moreover, most individuals (67%) were adults (aged >18 years) at the time of our centralized phenotypic evaluation and selection for molecular analysis. We have previously shown that, at least in our clinical setting, there is a significant delay between the time patients are suspected to have an IPD and the time confirmatory phenotypic diagnosis and molecular characterization are performed.4 The significance of this should not be underestimated, as it implies that many patients remain without a conclusive diagnosis for years, and thus are at risk of receiving inappropriate treatments. Multicentre collaboration, as supported by our project, and availability of HTS are expected to change this. Overall, our HTS approach enabled a molecular diagnosis in 68.3% of the patients (Figure 1), which is a much higher success rate than we achieved using only candidate gene sequencing.4 Remarkably, this sensitivity increased to nearly 90% for patients presenting with a well-defined clinical and laboratory phenotype indicative of a particular type of IPD. These results resemble those recently reported for the ThromboGenomics HTS platform, which covers the molecular screening of 63 genes.11 However, among patients presenting with an unclear phenotype other than bleeding or low platelet count, i.e., IPDs of uncertain etiology, our HTS platform was only able to identify candidate variants in about 50% of patients. This value is higher than that reported by the ThromboGenomics consortium in the same category of patients,11 but similar to the sensitivity recently achieved by WES in a limited cohort of IT patients with unknown etiology.20 Several factors could potentially contribute to this difference. Our study recruited only patients with established IPDs, even though the etiology of about half of them could not be inferred from clinical and laboratory data. In contrast, the ThromboGenomics study enrolled, among “cases with a highly uncertain etiology”, patients with bleeding problems but normal platelet function tests, and a few patients who had experienced thrombotic events. In addition, the gene content in both platforms is different, with only 33 genes from our platform haematologica | 2018; 103(1)

(Table 1) being present in the ThromboGenomics platform, which also included genes involved in coagulation disorders.11 The failure to identify candidate genetic defects in about 30% of cases may be due to intrinsic limitations of our HTS platform. First, the causative gene may not be present in the panel. Second, HTS methods can miss large deletions or duplications (>200bp), copy number variants involving >1000bp, or big structural chromosomal variants, translocations and aneuploidy, unless they have been specifically designed for such a purpose.21 Thus, for successful molecular diagnosis of certain cases, the HTS test must be combined with other molecular approaches such as comparative genomic hybridization (CGH) array, quantitative polymerase chain reaction (q-PCR), or multiplex ligation-dependent probe amplification (MLPA). For instance, in Case 30, who was diagnosed with TAR, the HTS test detected the uncommon rs139428292 single nucleotide polymorphism (SNP), inherited from the father (Online Supplementary Figure S7), but was insensitive to the pathogenic microdeletion in 1q21.1 which is associated with the disorder,22 and which was later detected by CGH-array analysis (Online Supplementary Figure S7). Standard analysis of HTS results also failed to identify candidate variants in Case 29, who had a clinical suspicion of DiGeorge syndrome (Table 2). However, massive parallel sequencing of CNVs analysis by HTS identified a RUNX1 deletion (Online Supplementary Figure S8), and hybridization in situ analysis revealed a 21q22 microdeletion (data not shown), resulting in RUNX1 haploinsufficiency.23 Moreover, our HTS platform also left a few target regions with insufficient coverage (<20x). This affected up to 21 genes, although none of them appeared to be related to the phenotypes of the corresponding patients (Online Supplementary Table S2). In Case 34, who had an unambiguous phenotype of GT but no HTS findings, no β3 messenger ribonucleic acid (mRNA) was detected, but Sanger sequencing of the candidate ITGB3 in the patient DNA also yielded negative results. It has been suggested that the few GT patients in whom no ITGA2B and ITGB3 variants are detected might benefit from whole genome analysis. This may unravel defects in regulatory elements and deep intronic regions that adversely affect the transcription or post-translational modifications and trafficking of αIIbβ3 integrin.24 In addition, for most IPDs of uncertain etiology and in a few suspected ones, such as our Cases 2527 who had a suspicion of GPS, the lack of genetic variants in our study highlights that many genes which cause IPDs or pathogenic variants affecting noncoding regulatory regions of the genome remain unidentified. Large-scale HTS projects, such as the 100,000 Genomes project in the UK and other novel approaches to gene discovery6,7 will help to overcome this limitation. In this series of patients, we found 57 different candidate variants in 28 genes, 70% of which were absent from the main reference databases, thus emphasizing the great heterogeneity of the molecular pathology underlying IPDs. Appropriate interpretation of the pathogenicity of candidate genetic variants found by HTS in IPDs remains a major challenge, especially for novel variants, even if present in well-established IPD genes. To prevent misinterpretation, the use of consensus guidelines is highly recommended,13,17 although there remains significant discordance between laboratories.25 Herein, following established guidelines,17 we classified 68.4% and 26.3% of the identified candidate variants as PV and LPV, respectively. In 2 159


J.M. Bastida et al.

cases (27 and 87) we found 3 (5.2%) novel variants affecting STIM1, RUNX1 and GNAS which qualified as VUS. Lack of segregation of novel candidate variants in the pedigrees is critical to prevent over-interpretation of pathogenicity. As an example, we disregarded novel variants in GP1BA (Case 45, c.1022C>G, p.Ser341Cys), FLNA (Case 63, c.5933 C>T, p.Thr1978Met), MASTL (Case 58, (c.836C>G; p.Pro279Arg) and NBEAL2 (Case 70, c.3424G>T, p.Gly1142Cys), since they were present in family members exhibiting no platelet defects. The genetic findings of this study are of clinical and scientific relevance. We established a conclusive diagnosis of autosomal recessive severe IPDs or X-linked disorders in about 25% of the patients, which would have informed decisions regarding their clinical care. These included diagnosis of BSS (n=4), GT (n=6), HPS (n=3), CHS (n=1), GPS (n=1), TAR (n=1) and WAS (n=2). One BSS patient carried a missense p.Cys225Ser variant in GP1BA, which has been previously identified in other patients from the Iberian peninsula, thereby suggesting a common ancestry.4,26 Another patient carried the novel change p.Leu155Val, also in GP1BA, which resembles the previously reported Bolzano variant p.Ala156Val, as it associates with severe biallelic BSS and nearly asymptomatic monoallelic BSS.26 It is worth mentioning that heterozygous variants in the GP1BA and GP1BB genes were a common cause of dominant IT in our series of patients, lending further weight to the idea that this condition might be more common than previously recognized.26,27 Six GT patients were diagnosed in this study. HTS revealed no pathogenic variants in ITGA2B and ITGB3 in 1 of these (Case 34), although the patient presented with an obvious type I GT platelet phenotype and no detectable levels of ITGB3 mRNA by quantitative real-time (qRT)-PCR. This case, along with those few GT patients previously reported to have no detectable mutations in ITGA2B and ITGB3,24,28 might benefit from whole-genome analysis with the aim of identifying defects in regulatory elements and deep intronic regions that adversely affect the transcription or post-translational modifications and trafficking of αIIbβ3. HTS cannot fully replace clinical evaluation and platelet phenotyping as it may result in misdiagnosis, but molecular characterization should be a component of an integral protocol for diagnosis. A similar argument may be valid for the 3 patients presenting with a phenotype suggestive of GPS (Cases 25-27), in whom we found no candidate variants in NBEAL or GFI1B. This is in accordance with multiple unknown genotypic alterations that may underlie the GPS phenotype. The clinical value of patient care and an accessible HTS test is also well exemplified in the case of severe multi-system IPDs such as HPS (Cases 41-43), CHS (Case 44) and WAS (Cases 33 and 66), in which the gene affected, as in the cases of HPS, or the type of mutation, regarding CHS and WAS, is likely to predict phenotype.29-31 Early identification of patients with these potentially life-threatening IPDs, depending on the genotype, is critical to the successful application of hematopoietic stem cell transplantation, and possibly gene therapy in some cases. Remarkably, in this project 3 young children had their molecular diagnosis confirmed as having either CHS or WAS defects, and 1 of the WAS patients (Case 66) atypically presented with macrothrombocytopenia of uncertain phenotype.32 The importance of genotype in predicting a patient´s clinical phenotype is also apparent in MYH9-RD.33 Herein, our HTS platform identified 7 pathogenic variants in 160

MYH9 in 7 unrelated cases. Three were unreported variants affecting the head domain of the protein, the SH3/MDi region, and another novel variant affected the tail domain. Notably, 5 of these patients were young individuals (<30 years old) who presented with no extra-hematological manifestations, and who could therefore benefit from close follow-up. Two notable pedigrees, Cases 18 and 19, were originally referred to us with a clinical suspicion MYH9-RD on the basis of mild macrothrombocytopenia and autosomal dominant sensorineural hearing loss. Subsequent analysis of the clinical and biological records in all relatives affected by deafness showed variable thrombocytopenia, absence of inclusion bodies in neutrophils and mild neutropenia. These data did not support the suspicion of MYH9-RD, but prompted us to suspect an underlying molecular pathology in DIAPH1. In OMIM this gene is usually linked to autosomal dominant deafness with/without thrombocytopenia, and thus is considered to be a phenocopy of MYH9. Remarkably, our HTS test revealed, in both pedigrees, the variant p.Arg1213* in DIAPH1, with full penetrance and with deafness presenting as the main clinical abnormality. This is a gain-of-function variant affecting autoregulation of DIAPH1 activity and proplatelet formation, recently identified by the ThromboGenomics consortium in 2 unrelated pedigrees with a similar phenotype to that in our cases.34 Functional studies of our patients revealed a minor effect of the p.Arg1213* variant in platelet aggregation and secretion, while it associates with a mildly reduced platelet expression of the DIAPH1 protein and a slightly higher level of tubulin β1 (Online Supplementary Material). Our current data generally concord with the previously reported phenotype in p.Arg1213* carriers,34 and gives further support to the idea that DIAPH1-RD is a novel type of IPD. Few other rare cases were identified with a novel molecular pathology affecting genes encoding cytoskeletal proteins involved in proplatelet formation, such as TUBB1 (Cases 51-55), ACTN1 (Case 64) and FLNA (Case 31).2 The later patient, a 13-year-old girl who displayed moderate to severe thrombocytopenia, had required sporadic blood transfusions and had been investigated for suspicion of WAS and CAMT, despite some clinical signs suggesting a filaminopathy (Table 2). Our identification of the novel missense mutation (c.3695C>T; p.Thr1232Ile) in exon 22 of FLNA supported the diagnosis of filaminopathy and warrants a more specific clinical investigation. This novel FLNA variant appears to be a de novo variant in the patient (Online Supplementary Figure S9), a phenomenon that occurs in about 20-30% of cases with sporadic bilateral periventricular nodular heterotopias.35 Additionally, 1 patient (Case 28) was characterized with STSL, a rare inherited sterol storage disorder.36 Functional in vitro studies are underway to explore the potential deleterious effect of these variants in platelet formation. Three novel variants in RUNX1 were found in Cases 32, 73 and 74, who all presented with mild platelet dysfunction, whilst 2 of them also had thrombocytopenia. Only members of 1 of these families had a history of hematological malignancy. These variants are expected to affect the function of RUNX1, as 2 (p.Gln268* and p.Asn159Ser) lie within the RUNT homology domain which mediates DNA binding and heterodimerization with CBFβ and the remaining (p.Ser402Phe) in the C-terminal inhibitory domain of RUNX1.37 Comprehensive interpretation of variants in this transcription factor, like those in ANKRD26 haematologica | 2018; 103(1)


Genetic diagnosis of IPDs by HTS

(Case 72 in our series), is relevant as such variants may increase the risk of developing myeloid malignancies.37,38 In Case 40, the selective platelet dysfunction at the GPVI level, which likely had autosomal dominant inheritance, matched well with the finding of a novel in-frame deletion in GP6 detected by HTS. Among the subgroup of patients presenting with mild platelet dysfunction of uncertain etiology, 8 (44%) pedigrees exhibited likely pathogenic heterozygous variants in genes encoding other platelet receptors (ITGA2, TBXA2R, P2YR12) or enzymes involved in second messenger release and platelet signal transduction (GNAS, PLCB2, PTGS1, TBXAS1; Table 5). Further studies are required to determine the contribution, if any, of these genetic defects to the platelet dysfunction and bleeding tendency of these patients. Interestingly, other patients previously reported to have inherited defects in these platelet proteins were also heterozygous.39,40 Bleeding diathesis in these patients may be facilitated by co-inheritance with other genetic disorders of hemostasis, such as type 1 von Willebrand disease.41 A key protein for integrin signaling in platelets and neutrophils is the guanine nucleotide exchange factor CalDAG-GEFI. Recently a variant in RASGRP2, the gene encoding CalDAG-GEFI, was identified in 3 siblings with impaired platelet function and bleeding diathesis.42 Herein, our HTS test identified 3 novel variants in RASGRP2 (p.Ser381Phe, p. Gln236* and p. Cys296Tyr) from 3 unrelated children with lifelong severe bleeding complications and reduced platelet aggregation with most agonists. Further functional studies demonstrated that these novel variants severely affect CalDAG-GEFI expression and activity, leading to defective agonist-induced integrin activation in platelets and neutrophils.43,44 Of late other patients harboring pathogenic variants in RASGRP2 have been identified,45-47 indicating that this type of IPFD might occur more frequently than previously thought.

Conclusions This study demonstrates that our HTS platform is an accurate, reproducible and reliable tool for the genetic characterization of IPDs. Using this approach, we can achieve a molecular diagnosis in most patients with a suspected etiology, and in about half the cases presenting with a disease of highly uncertain biological cause. Our findings reinforce the feasibility of introducing this technology into mainstream genetic testing for diagnosing IPDs. Patients with an IPD in which the HTS platform fails to identify the underlying molecular pathology are candidates for examination using less restrictive molecular approaches, such as WES or WGS. The use of human phenotype ontology codification, consensus guidelines for interpreting genetic variants, and in silico bioinformatics analysis tools to facilitate the identification of candidate causative variants will be important in aiding this process. However, definitive pathogenicity

References 1. Nurden AT, Nurden P. Congenital platelet disorders and understanding of platelet function. Br J Haematol. 2014;165(2):165-178. 2. Savoia A. Molecular basis of inherited

haematologica | 2018; 103(1)

assignment of novel rare variants must be established on the basis of their identification in unrelated pedigrees with similar phenotype and/or demonstrative functional studies. Funding This study was supported by research grants from the Gerencia Regional de Salud (GRS 1370/A/16), ISCIII & Feder (PI14/01956), CIBERER CB15/00055, Fundación Séneca (19873/GERM/15) and Sociedad Española de Trombosis y Hemostasia (SETH). SPW holds a British Heart Foundation chair. Acknowledgments We acknowledge all the patients and their families for providing samples. We thank Dr Phil Mason for his help with technical aspects. We are grateful to the following clinicians: Members of the Castilla y León Society of Thrombosis and Haemostasis Group: RM Fisac (Hospital General, Segovia), MP Martínez-Badas (Complejo Asistencial de Ávila), LJ García-Frade and E Fontecha (Hospital Universitario Río Hortega, Valladolid), JM MartínAntorán (Complejo Asistencial de Palencia), C Aguilera (Hospital de El Bierzo, Ponferrada), B Pérez (Complejo Asistencial de León) MJ Cebeira (Hospital Clínico de Valladolid), TJ González-López (Complejo Asistencial de Burgos), RM Henar-Cantalejo (Hospital General de Aranda de Duero), R Campos (Hospital de Jerez), E Pardal (Hospital Virgen del Puerto, Plasencia), R Ramos (Hospital Infanta Cristina, Badajoz), R Vidal and MP Llamas (Fundación Jiménez Díaz, Madrid), M Salces (Hospital Universitario La Paz, Madrid), P Olivera (Hospital Vall d´Hebron, Barcelona), A Repáraz (Unidad de Citogenética y Genética Molecular, Hospital Álvaro Cunqueiro, Vigo), G Iruin (Hospital de Cruces, Bilbao), AR Cid (Hospital Universitario La Fe, Valencia), E Bardón (Hospital Universitario de Torrejón, Madrid), A Galera (Hospital Universitario Virgen de la Arrixaca, Murcia), JL Fuster and ME LLinares (Hospital Universitario Virgen de la Arrixaca, Murcia), S Riesco, MC Mendoza, A Benito and A Hortal (Hospital Universitario de Salamanca), MT Alonso (Hospital Universitario de Valladolid), J Huertas (Hospital Gregorio Marañón, Madrid), I Astigarraga (Hospital de Cruces, Bilbao), D Jaimes (Hospital de Donostia), H Cano (Hospital Los Arcos, Murcia), J Mateo (Hospital San Pablo, Barcelona), T Iturbe (Hospital Santa Lucia, Cartagena), R Berrueco (Hospital Sant Joan de Déu, Barcelona), M Lozano (Hospital Clinic, Barcelona), N Fernandez Mosteririn (Hospital Miguel Servet, Zaragoza), C Muñoz (Hospital Virgen de la Macarena, Sevilla), I Ancin (H. Cruces, Bilbao), T Jover (Hospital Universitario Virgen de la Arrixaca, Murcia), E Roselló (Hospital de Universitario de Bellvitge, Barcelona), EM Mingot, Hospital Universitario Carlos Haya de Málaga), RM Campos (Hospital de Jérez), JM Guinea (Hospital de Araba), M Trapero (Clínica Puerta de Hierro, Madrid), N Rollón (Hospital Virgen de la Salud, Toledo M) and Karkucak (Dpt. Medical Genetics, Sakarya University Training and Research Hospital, Turkey). We are also grateful to Irene Rodríguez, Sara González, Sandra Santos, Sandra Pujante, José Padilla, Ana Isabel Antón, Isabel Sánchez-Guiu, Eva Caparrós, Nerea Mota and Constantino Martínez for their help in isolating and processing DNA and for carrying out some of the platelet assays.

thrombocytopenias: an update. Curr Opin Hematol. 2016;23(5):486-492. 3. Gresele P. Diagnosis of inherited platelet function disorders: guidance from the SSC of the ISTH. J Thromb Haemost. 2015; 13(2):314-322.

4. Sanchez-Guiu I, Anton AI, Padilla J, et al. Functional and molecular characterization of inherited platelet disorders in the Iberian Peninsula: results from a collaborative study. Orphanet J Rare Dis. 2014;9:213. 5. Watson SP, Lowe GC, Lordkipanidze M,

161


J.M. Bastida et al.

6.

7. 8.

9.

10.

11.

12.

13.

14.

15. 16.

17.

18.

19.

20.

21.

162

Morgan NV. Genotyping and phenotyping of platelet function disorders. J Thromb Haemost. 2013;11 Suppl 1:351-363. Lentaigne C, Freson K, Laffan MA, Turro E, Ouwehand WH. Inherited platelet disorders: toward DNA-based diagnosis. Blood. 2016;127(23):2814-2823. Westbury SK, Mumford AD. Genomics of platelet disorders. Haemophilia. 2016;22 Suppl 5:20-24. Sivapalaratnam S, Collins J, Gomez K. Diagnosis of inherited bleeding disorders in the genomic era. Br J Haematol. 2017; 179(3):363-376. Bastida JM, Del Rey M, Lozano ML, et al. Design and application of a 23-gene panel by next-generation sequencing for inherited coagulation bleeding disorders. Haemophilia. 2016;22(4):590-597. de Koning TJ, Jongbloed JD, SikkemaRaddatz B, Sinke RJ. Targeted next-generation sequencing panels for monogenetic disorders in clinical diagnostics: the opportunities and challenges. Expert Rev Mol Diagn. 2015;15(1):61-70. Simeoni I, Stephens JC, Hu F, et al. A highthroughput sequencing test for diagnosing inherited bleeding, thrombotic, and platelet disorders. Blood. 2016;127(23):2791-2803. Westbury SK, Turro E, Greene D, et al. Human phenotype ontology annotation and cluster analysis to unravel genetic defects in 707 cases with unexplained bleeding and platelet disorders. Genome Med. 2015;7(1):36. Freson K, Turro E. High-throughput sequencing approaches for diagnosing hereditary bleeding and platelet disorders. J Thromb Haemost. 2017;15(7):1262-1272. DePristo MA, Banks E, Poplin R, et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet. 2011;43(5):491-498. Li H, Durbin R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics. 2010;26(5):589-595. Stenson PD, Mort M, Ball EV, et al. The Human Gene Mutation Database: towards a comprehensive repository of inherited mutation data for medical research, genetic diagnosis and next-generation sequencing studies. Hum Genet. 2017;136(6):665-677. 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. Monteferrario D, Bolar NA, Marneth AE, et al. A dominant-negative GFI1B mutation in the gray platelet syndrome. N Engl J Med. 2014;370(3):245-253. Turro E, Greene D, Wijgaerts A, et al. A dominant gain-of-function mutation in universal tyrosine kinase SRC causes thrombocytopenia, myelofibrosis, bleeding, and bone pathologies. Sci Transl Med. 2016;8(328):328ra330. Johnson B, Lowe GC, Futterer J, et al. Whole exome sequencing identifies genetic variants in inherited thrombocytopenia with secondary qualitative function defects. Haematologica. 2016;101(10):1170-1179. Daber R, Sukhadia S, Morrissette JJ. Understanding the limitations of next generation sequencing informatics, an approach to clinical pipeline validation using artificial data sets. Cancer Genet. 2013;206(12):441448.

22. Albers CA, Paul DS, Schulze H, et al. Compound inheritance of a low-frequency regulatory SNP and a rare null mutation in exon-junction complex subunit RBM8A causes TAR syndrome. Nat Genet. 2012; 44(4):435-439, S431-432. 23. Christensen RD, Wiedmeier SE, Yaish HM. A neonate with congenital amegakaryocytic thrombocytopenia associated with a chromosomal microdeletion at 21q22.11 including the gene RUNX1. J Perinatol. 2013;33(3):242-244. 24. Kannan M, Saxena R. No genetic abnormalities identified in alpha2IIb and beta3: phenotype overcomes genotype in Glanzmann thrombasthenia. Int J Lab Hematol. 2017; 39(2):e41-e44. 25. Amendola LM, Jarvik GP, Leo MC, et al. Performance of ACMG-AMP variant-interpretation guidelines among nine laboratories in the Clinical Sequencing Exploratory Research Consortium. Am J Hum Genet. 2016;98(6):1067-1076. 26. Savoia A, Kunishima S, De Rocco D, et al. Spectrum of the mutations in BernardSoulier syndrome. Hum Mutat. 2014; 35(9):1033-1045. 27. Sivapalaratnam S, Westbury SK, Stephens JC, et al. Rare variants in GP1BB are responsible for autosomal dominant macrothrombocytopenia. Blood. 2017;129(4):520-524. 28. Nurden AT, Pillois X, Fiore M, et al. Expanding the mutation spectrum affecting alphaIIbbeta3 integrin in glanzmann thrombasthenia: screening of the ITGA2B and ITGB3 genes in a large international cohort. Hum Mutat. 2015;36(5):548-561. 29. Buchbinder D, Nugent DJ, Fillipovich AH. Wiskott-Aldrich syndrome: diagnosis, current management, and emerging treatments. Appl Clin Genet. 2014;7:55-66. 30. Lozano ML, Rivera J, Sanchez-Guiu I, Vicente V. Towards the targeted management of Chediak-Higashi syndrome. Orphanet J Rare Dis. 2014;9:132. 31. Sanchez-Guiu I, Torregrosa JM, Velasco F, et al. Hermansky-Pudlak syndrome. Overview of clinical and molecular features and case report of a new HPS-1 variant. Hamostaseologie. 2014;34(4):301-309. 32. Bastida JM, Del Rey M, Revilla N, et al. Wiskott-Aldrich syndrome in a child presenting with macrothrombocytopenia. Platelets. 2017;28(4):417-420. 33. Pecci A, Klersy C, Gresele P, et al. MYH9related disease: a novel prognostic model to predict the clinical evolution of the disease based on genotype-phenotype correlations. Hum Mutat. 2014;35(2):236-247. 34. Stritt S, Nurden P, Turro E, et al. A gain-offunction variant in DIAPH1 causes dominant macrothrombocytopenia and hearing loss. Blood. 2016;127(23):2903-2914. 35. Parrini E, Ramazzotti A, Dobyns WB, et al. Periventricular heterotopia: phenotypic heterogeneity and correlation with Filamin A mutations. Brain. 2006;129(Pt 7):1892-1906. 36. Bastida JM, Benito R, Janusz K, et al. Two novel variants of the ABCG5 gene cause xanthelasmas and macrothrombocytopenia: a brief review of hematologic abnormalities of sitosterolemia. J Thromb Haemost. 2017;15(9):1859-1866. 37. Daly ME. Transcription factor defects causing platelet disorders. Blood Rev. 2017;31(1):110. 38. Noris P, Favier R, Alessi MC, et al. ANKRD26related thrombocytopenia and myeloid malignancies. Blood. 2013;122(11): 19871989.

39. Jones ML, Norman JE, Morgan NV, et al. Diversity and impact of rare variants in genes encoding the platelet G protein-coupled receptors. Thromb Haemost. 2015;113(4): 826-837. 40. Lecchi A, Razzari C, Paoletta S, et al. Identification of a new dysfunctional platelet P2Y12 receptor variant associated with bleeding diathesis. Blood. 2015;125(6):10061013. 41. Stockley J, Nisar SP, Leo VC, et al. Identification and characterization of novel variations in platelet G-protein coupled receptor (GPCR) genes in patients historically diagnosed with Type 1 von Willebrand Disease. PLoS One. 2015; 10(12):e0143913. 42. Canault M, Ghalloussi D, Grosdidier C, et al. Human CalDAG-GEFI gene (RASGRP2) mutation affects platelet function and causes severe bleeding. J Exp Med. 2014;211(7): 1349-1362. 43. Lozano ML, Cook A, Bastida JM, et al. Novel mutations in RASGRP2, which encodes CalDAG-GEFI, abrogate Rap1 activation, causing platelet dysfunction. Blood. 2016;128(9):1282-1289. 44. Sevivas T, Bastida JM, Paul DS, et al. Identification of two novel mutations in RASGRP2 affecting platelet CalDAG-GEFI expression and function in patients with bleeding diathesis. Platelets. 2017:1-4. 45. Bermejo E, Alberto MF, Paul DS, et al. Marked bleeding diathesis in patients with platelet dysfunction due to a novel mutation in RASGRP2, encoding CalDAG-GEFI (p.Gly305Asp). Platelets. 2017:1-3. 46. Kato H, Nakazawa Y, Kurokawa Y, et al. Human CalDAG-GEFI deficiency increases bleeding and delays alphaIIbbeta3 activation. Blood. 2016;128(23):2729-2733. 47. Westbury SK, Canault M, Greene D, et al. Expanded repertoire of RASGRP2 variants responsible for platelet dysfunction and severe bleeding. Blood. 2017;130(8):10261030. 48. Bottega R, Nicchia E, Alfano C, et al. Gray platelet syndrome: Novel mutations of the NBEAL2 gene. Am J Hematol. 2017;92 (2):E20-E22. 49. Yoon SH, Cho T, Kim HJ, et al. IVS6+5G>A found in Wiskott-Aldrich syndrome and Xlinked thrombocytopenia in a Korean family. Pediatr Blood Cancer. 2012;58(2):297-299. 50. Ambo H, Kamata T, Handa M, et al. Novel point mutations in the alphaIIb subunit (Phe289-->Ser, Glu324-->Lys and Gln747->Pro) causing thrombasthenic phenotypes in four Japanese patients. Br J Haematol. 1998;102(3):829-840. 51. Nurden AT, Breillat C, Jacquelin B, et al. Triple heterozygosity in the integrin alphaIIb subunit in a patient with Glanzmann's thrombasthenia. J Thromb Haemost. 2004;2(5):813-819. 52. Jallu V, Dusseaux M, Panzer S, et al. AlphaIIbbeta3 integrin: new allelic variants in Glanzmann thrombasthenia, effects on ITGA2B and ITGB3 mRNA splicing, expression, and structure-function. Hum Mutat. 2010;31(3):237-246. 53. Bachli EB, Brack T, Eppler E, et al. Hermansky-Pudlak syndrome type 4 in a patient from Sri Lanka with pulmonary fibrosis. Am J Med Genet A. 2004; 127A(2):201-207. 54. Gueguen P, Rouault K, Chen JM, et al. A missense mutation in the alpha-actinin 1 gene (ACTN1) is the cause of autosomal dominant macrothrombocytopenia in a large French family. PLoS One. 2013; 8(9):e74728.

haematologica | 2018; 103(1)


ARTICLE

Platelet Biology & Its Disorders

Comparison of up-front treatments for newly diagnosed immune thrombocytopenia a systematic review and network meta-analysis

Ferrata Storti Foundation

Yasuyuki Arai,1,2 Tomoyasu Jo,1 Hiroyuki Matsui,1 Tadakazu Kondo1 and Akifumi Takaori-Kondo1

Department of Hematology and Oncology, Graduate School of Medicine, Kyoto University, Japan; 2Laboratory of Clinical Immunology and Microbiology, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD, USA 1

Haematologica 2018 Volume 103(1):163-171

ABSTRACT

C

orticosteroids such as prednisolone and dexamethasone have been established as up-front therapy for the treatment of newly diagnosed immune thrombocytopenia. Recent studies have indicated that other treatments such as rituximab or thrombopoietin receptor agonist can also be effective choices. We performed a systematic review and network meta-analysis to establish a clinically meaningful hierarchy of efficacy and safety of treatments for newly diagnosed primary immune thrombocytopenia in adults. Randomized controlled trials evaluating medical treatments for newly diagnosed immune thrombocytopenia were included. Reviewers independently extracted data and assessed the risk of bias. The main outcome was the sustained response (platelet count >30Ă&#x2014;109/L for 3-6 months after completion of treatments), while overall response (platelet count >30Ă&#x2014;109/L for 2-4 weeks after initiation of the up-front treatment) and therapy-related adverse events were the secondary endpoints. A total of 21 randomized controlled trials (1898 patients) were included in this study. Our main findings were a significantly better sustained response in the recombinant human thrombopoietin+dexamethasone and rituximab+dexamethasone arms compared to those of conventional therapies (prednisolone and dexamethasone monotherapy). Moreover, recombinant human thrombopoietin+dexamethasone and +prednisolone improved early overall response compared to prednisolone, dexamethasone, and rituximab-containing regimens. Therapy-related adverse events showed similar profiles and were tolerable in all treatment arms. Regimens containing recombinant human thrombopoietin agonist may be beneficial up-front therapies in addition to the conventional corticosteroid monotherapies. Future head-to-head trials including these regimens and rituximab-containing treatments are necessary in order to overcome the limitations of the small number in our study and determine the most suitable initial therapies for newly diagnosed immune thrombocytopenia.

Introduction Newly diagnosed primary immune thrombocytopenia (ITP) is characterized by platelet destruction due to acquired autoantibodies against the platelets and relatively impaired platelet production in the bone marrow,1,2 leading to an increased risk of bleeding.3 The conventional treatments for symptomatic newly diagnosed ITP include corticosteroids; previous guidelines recommended either prednisolone/prednisone (PSL) or high-dose dexamethasone.4,5 In addition, intravenous immune globulins, eradication of Helicobacter pylori, and methylPSL have long been used solely or in haematologica | 2018; 103(1)

Correspondence: tadakazu@kuhp.kyoto-u.ac.jp

Received: June 15, 2017. Accepted: September 27, 2017. Pre-published: September 29, 2017

doi:10.3324/haematol.2017.174615 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/163 Š2018 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.

163


Y. Arai et al.

combination with other corticosteroids.5 Recently, rituximab in addition to dexamethasone has been shown to be more effective than dexamethasone monotherapy.6 Moreover, thrombopoietin receptor agonists (TPO-RA), including eltrombopag, romiplostim, and recombinant human thrombopoietin (rhTPO), may be reasonable choices given their ability to enhance platelet production in the bone marrow.7 Thus, while there are multiple therapeutic options for up-front treatment of newly diagnosed ITP, there is little evidence to identify the best option for early and sustained recovery of platelet counts without severe adverse events. Several randomized controlled trials (RCT) have compared two specific treatment regimens (e.g., dexamethasone versus PSL and rituximab versus placebo); related systematic reviews and meta-analyses have summarized these results.6,8 However, as there are so many therapeutic options, it is nearly impossible to cover all the combinations in direct head-to-head RCT in order to determine the optimal choice. Recent developments in the application of network meta-analysis may possibly be helpful to overcome this limitation.9 In contrast to conventional pairwise metaanalyses, this method enables indirect comparisons between treatments used in different trials.10 We, therefore, performed a systematic review and network metaanalysis in order to establish a clinically meaningful hierarchy for the efficacy and safety of newly diagnosed ITP treatments in adults through the integration and synthesis of all available evidence.

Methods Search strategies and study selection The search strategies are outlined in Online Supplementary Tables S1 and S2. The patients were limited to adults (16 years or older) with newly diagnosed primary ITP; RCT using other definitions of ITP (such as “acute ITP”) were included only if their definitions matched the “newly diagnosed ITP” defined as ITP of less than 3 months’ duration11 without preceding treatments. Those with secondary thrombocytopenia or those who had had previous therapeutic interventions for ITP were excluded. Two review authors scanned the titles and abstracts of the studies identified by the electronic search strategies in order to assess their eligibility. The two authors then independently evaluated the full-text versions of each potentially relevant study for inclusion in the meta-analysis. Disagreements between authors were resolved by discussion. If necessary, arbitration was provided by the senior authors. When missing information inhibited the evaluation of a study, further information was sought from the original authors or other possible sources. The study selection process is reported in a PRISMA flow diagram (Figure 1).

ments were discussed with the senior authors until a consensus was obtained.

Data synthesis and analysis We performed a random effects network meta-analysis using STATA13 software (StataCorp, College Station, TX, USA). Evidence from both direct (head-to-head trials) and indirect (using common comparators without actual head-to-head trials) comparisons was combined in the analysis. The primary outcome was the incidence of long-term sustained response (SR; platelet counts >30×109/L for 3 - 6 months after the completion of treatments). The secondary endpoints included the incidence of (i) early overall response (OR; platelet count >30×109/L in 2 – 4 weeks after the initiation of the up-front therapy) and (ii) severe adverse events [grade 3 or more according to the Common Terminology Criteria for Adverse Events (CTCAE) version 4.0]. All of our treatment effects were measured as dichotomous data, and were presented as the summaries of the risk ratios (RR) with 95% confidence intervals (CI). A surface under the cumulative ranking curve (SUCRA) was also used to provide a hierarchy of the efficacy and the risk of adverse events of the treatments,12 where SUCRA values of 100% indicated the most effective treatment or the treatment with the least risk of adverse events, and values of 0% indicated the least effective and the highest risk treatment.

Results Identification of studies The study identification and selection process is illustrated in Figure 1. From the primary search, 11 studies were excluded because of rare and uncommon therapeutic interventions; these studies dealt with gamma globulins in specific solvent,13,14 a proton pump inhibitor (monotherapy to eradicate Helicobacter pylori),15 vinblastine,16 deflazacort,17 and various Chinese herbal medicines.18-21 Two RCT deal-

Data extraction and quality assessment Data from the included trials were independently extracted by two review authors using a structured, pilot-tested, data extraction form (Online Supplementary Table S3). Differences in data extraction were resolved either by discussion or by consultation with the senior authors. These review authors also independently assessed the eligible studies for bias using the tool described in the Cochrane Handbook for Systematic Reviews of Interventions. We evaluated the risk of bias as low, high, or unclear using an assessment form designed for the topic of this review (Online Supplementary Table S4). Any disagree164

Figure 1. Study schema. Flow diagram showing the process of identifying and selecting relevant studies.

haematologica | 2018; 103(1)


Network meta-analysis for newly diagnosed ITP

ing with intravenous anti-D globulin were excluded from data integration, because one (comparing anti-D versus routine care) did not report either early OR or long-term SR,22 and the other compared low- and high-dose anti-D without reference to other treatment arms,23 resulting in a node unconnected to the other treatment network. A search of conference proceedings revealed one additional relevant study (abstract only). Thus, in total, we included 21 RCT that involved 1898 adult patients with newly diagnosed primary ITP. Individual patients’ data were not available in any study.

Study characteristics The studies were conducted worldwide including North

America, Europe, Africa, and Asia, and were published between 1985 and 2016 (Table 1).24-44 The median age of the participants ranged from 25 to 54 years. Only patients with primary newly diagnosed ITP were included, and RCT dealing with secondary ITP were excluded. Inclusion criteria in each RCT for the initial platelet count or actual platelet count were reported in 19 RCT, with most having a platelet count <30×109/L as an inclusion criterion (Table 1). The interventions were initiated rapidly after the diagnosis (generally within 2 weeks), and they consisted mainly of PSL, dexamethasone, rituximab, and their combinations; while intravenous immunoglobulin, eradication of Helicobacter pylori (using amoxicillin, clarithromycin, and rabeprazole), methylprednisolone, and rhTPO were also

Table 1. Summary of included studies.

Study ID

Country

Mazzucconi 198524 Bellucci 198825 Jacobs 199426

Number of participants (I/C)

Sex Age (y), Initial Plt (M/F) median (range) criteria/ median (×109/L)

Italy France South Africa

37/32 111/112 26/17

17/52 66/157 9/34

27(13-65) 41.5 33(16-66)

<60/ND <10/ND ND/ND

Godeau 200227

France

56/60

39/77

38(24-59)

<20/7

Kong 200828

China

65/35

43

ND/ND

Thailand Italy

18/18 49/52

8/28 41/60

42 48

<20/9 <20/ND

Bae 201030 Cui 201132 Li 201133

Korea China China

76/75 30/29 31/31

45/105 22/37 25/37

44 31(16-65) 25(18-59)

<30/16 <30/11 ND/6

Arnold 201234 Mashhadi 201235

Canada Iran

33/27 30/30

25/35 13/47

40(30-59) 26(18-46)

<30/15 <20/12

Gu 201336

China

31/31

24/38

50(21-84)

ND/7

62/71

63/70

54(33-70)

<25/13

Praituan 200929 Zaja 201031

Gudbrandsdottir 201337 Denmark Li 201338

China

45/44/49

55/63

36(18-70)

<30/11

Xing 201339

China

38/36

35/39

34

<30/ND

Din 201540 Li 201641

China China

61/29 23/25

40/50 26/22

30(16-64) 45(20-69)

<20/11 ND/4

Germany

13/9

13/9

44(22-77)

ND/3

Sun 201643

China

30/29

29/30

30

<20/ND

Wei 201644

China

95/97

56/136

43(18-75)

<30/7

Matschke 201642

Intervention

Regimen (I)

Comparison

Regimen (C)

PSL(LD) PSL(LD) IVIG±PSL

PSL 0.5 mg/kg×1m PSL PSL 1.5 mg/kg×1m PSL 0.25 mg/kg×3w PSL PSL 1 mg/kg×3w IVIG 400 mg/kg×5 PSL PSL 1 mg/kg (→ PSL 1 mg/kg) IVIG±PSL IVIG 0.7 mg/kg×3 mPSL±PSL mPSL 15 mg/kg×3 (→ PSL 1 mg/kg×18d) (→ PSL 1 mg/kg×18d) HP±PSL AMPC/CAM/RAB PSL PSL 1 mg/kg (→ PSL 1 mg/kg) Dex Dex 40 mg×4d PSL PSL 60 mg×14d RTX+Dex RTX 375 mg/m2×4w Dex Dex 40 mg×4d → Dex 40 mg×4d Dex Dex 40 mg×4d×1-2 PSL PSL 1 mg/kg×28d Dex Dex 40 mg×4d×2 PSL PSL 1-1.5 mg/kg×4w RTX+Dex RTX 100 mg×4w Dex Dex 40 mg×4d → Dex 40 mg×4d RTX+PSL RTX 375mg/m2×4w+PSL PSL PSL Dex Dex 40 mg×4d PSL PSL1 mg/kg×28d → PSL 1 mg/kg×7d rhTPO+PSL rhTPO 15000U×5-7d PSL PSL 60 mg/d → PSL 60 mg/d RTX+Dex RTX 375 mg/m2×4w Dex Dex 40 mg×4d → Dex 40 mg×4d Dex or Dex 40 mg×4d or PSL PSL 1.5 mg/kg×2-4w RTX+Dex Dex 40 mg×4d → RTX 100 mg×4w RTX+Dex+PSL Dex 40 mg×4d → RTX+Dex Dex 40 mg×4d RTX 100 mg×4w → RTX 100mg×4w → PSL 60 mg Dex Dex 40 mg×4d×3 PSL PSL 1 mg/kg×4w rhTPO+Dex rhTPO 15000U×14d Dex Dex 40 mg×4d → Dex 40 mg×4d Dex PSL 1 mg/kg×1w PSL PSL 1 mg/kg×2w → Dex 0.6 mg/kg×4d×6 rhTPO+Dex rhTPO 300 U/kg×14d Dex Dex 40 mg×4d → Dex 40 mg×4d Dex Dex 40 mg×4d×1-2 PSL PSL1 mg/kg×28d

I/C: intervention/comparison; M/F: male/female; y: year; ND: not determined; PSL: prednisolone or prednisone; LD: low-dose; m: month; w: week; IVIG: intravenous gamma globulin; d: day; mPSL: methylprednisolone; HP: eradicating agent for Helicobacter pylori; AMPC: amoxicillin; CAM: clarithromycin; RAB: rabeprazole; Dex: dexamethasone; RTX: rituximab; rhTPO: human recombinant thrombopoietin. Dex: dexamethasone.

haematologica | 2018; 103(1)

165


Y. Arai et al.

used. Rituximab was administered weekly at the dose of either 375 mg/m2 2,37 or 100 mg/body33,38,39 as combination therapy with dexamethasone. Low-dose (100 mg/body) rituximab was selected mainly because of concerns regarding infectious risk33. All three RCT dealing with rhTPO used TPIAO® manufactured by 3SBio Inc. in China, and rhTPO was administered for 1 – 2 weeks. As for the PSL arm, four studies29,30,35,38 used prednisolone, while prednisone was used in nine studies;24-27,32,34,40,42,44 these data were combined in our meta-analysis. PSL was categorized as low dose if the daily dosage was less than 1.0 mg/kg. Among these studies, 16 RCT (1583 patients) covering nine types of interventions reported results on the primary endpoint (long-term SR; Figure 2A).25-27,30-35,37-

of the studies lacked sufficient blinding of participants and personnel, and assessors (classified as “high risk”). These data indicate that some RCT, especially those performed before 2010’s, were performed under poorly designed protocols. Some of the articles were written in Chinese28,32,36,38,39,41,43 and English version were not available; one of the authors (YA) was able to understand the Chinese language, and we also asked the translation service to confirm the detailed meanings. The translated versions of these papers were analyzed by multiple authors just as we did for the other papers written in English. The risk analyses and outcome measure assessment in these Chinese papers were, therefore, properly performed just as for other studies.

40,42-44

Consistency between direct and indirect evidence Risk of bias in the included studies The risk of bias is graphically summarized in Online Supplementary Figure S1. Information on random sequence generation and allocation concealment was not sufficiently described in many studies (classified as “unclear”). Most

Loop-specific tests revealed no significant inconsistency in one available loop (formed by PSL, dexamethasone, and rituximab + dexamethasone arms) within the data network for long-term SR. Based on a design-by-treatment interaction model,45 no significant inconsistency between

A

B

C

Figure 2. Results of the network of long-term sustained response comparison. (A) The network of comparisons included in the network meta-analysis for long-term sustained response (SR; platelet counts > 30×109/L at 3 - 6 months). The circle size is proportional to the total number of patients in the treatment group. The line width is proportional to the number of trials comparing the treatment groups. (B) The summary effect estimate (risk ratio of SR) for each combination of treatments. Risk rations are indicated by dots, and 95% confidence intervals by bars. (C) The surface under the cumulative ranking curve (SUCRA) is shown for each treatment.

166

haematologica | 2018; 103(1)


Network meta-analysis for newly diagnosed ITP

direct and indirect evidence was identified within the evidence network as a whole (P=0.18). These data support the consistency model in the following analyses.

Outcomes Long-term sustained response We analyzed the SR (platelet count >30×109/L11) at 3 - 6 months after the completion of therapies as a dichotomous outcome. The total numbers of patients and the numbers of those who obtained a SR along with its definition in each RCT are displayed in Online Supplementary Table S5. The pooled results demonstrated that (i) rhTPO + dexamethasone and rituximab + dexamethasone produced significantly better responses than dexamethasone or PSL, and (ii) rhTPO + dexamethasone was significantly superior to rituximab + PSL (RR, 5.22; 95% CI: 1.61 – 16.9; P< 0.01) and the RR was also higher when compared to that of rituximab + dexamethasone, though without significance (RR, 2.01; 95% CI: 0.75 – 5.40; P=0.16) (Figure 2B). Here, in the rhTPO + dexamethasone arm, the rhTPO was given for 14 days followed by 4 days of dexamethasone;43 even with such a short therapeutic period, the SR at 3 months after the completion of this up-front treatment marked the fairly good outcome (76.7%) (Online Supplementary Table S5). rhTPO + PSL was not included in this network, and rituximab + PSL did not show any significant difference in comparison with PSL and dexamethasone. The dosage of rituximab (100 mg/body versus 375 mg/m2) was not related to a significant change in longterm SR (RR, 1.10; 95% CI: 0.51 – 2.34; P=0.81). SUCRA rankings for long-term SR were also evaluated (Figure 2C and Table 2); rhTPO + dexamethasone (97.1%), rituximab + dexamethasone (81.3%), and rituximab + dexamethasone + PSL (81.1%) showed the highest values, followed by dexamethasone (56.2%), PSL (41.5%), and rituximab + PSL (34.6%), while low-dose PSL (27.7%), methylprednisolone (18.5%), and intravenous immune globulins (11.9%) had the lowest values in this ranking. These data indicate that the highest probability of achieving a long-term SR is when ITP is treated with regimens including rhTPO and rituximab in combination with dexamethasone. However, attention should be paid to the small size of the rhTPO + dexamethasone arm (1 study, n=30) when drawing conclusions from these findings (Online Supplementary Table S5).

for dexamethasone, and RR, 1.17; 95% CI 0.94 – 1.46; P=0.16 for PSL). The dosage of rituximab (100 mg/body versus 375 mg/m2) was not related to a significant change in early OR (RR, 1.13; 95% CI: 0.76 – 1..68; P=0.55). According to the SUCRA values of early OR (Figure 3C and Table 2), rhTPO + PSL (98.8%) had the highest value followed by rhTPO + dexamethasone (82.4%). Rituximab + dexamethasone + PSL (74.7%), H. pylori eradication (only in H. pylori-positive patients; 59.3%), rituximab + dexamethasone (54.2%), dexamethasone (49.3%), and PSL (31.1%) had moderate values, while intravenous immune globulins (24.2%), methylprednisolone (15.1%), and low-dose PSL (10.8%) showed the lowest SUCRA values. These data indicate that regimens combining rhTPO and corticosteroids (rhTPO + dexamethasone or rhTPO + PSL) may be the optimal choice for obtaining an early OR in newly diagnosed ITP. The small size of the rhTPO + PSL arm (1 study, n=31 patients) and the rhTPO + dexamethasone arm (2 studies, n=53 patients) should be considered as a limitation (Online Supplementary Table S6). This tendency was also confirmed even when OR was determined as a platelet count >50×109/L instead of a platelet count >30×109/L, although the number of included studies was small (Online Supplementary Table S6) and the results were not robust enough (data not shown).

Incidence of adverse events A total of 16 studies including 1325 patients provided data regarding acute- or chronic-phase adverse events related to each intervention.27,29-37,39-44 Severe adverse events (CTCAE grade 3 or more) were divided into non-hemorrhagic and hemorrhagic events, and the numbers of patients who experienced them are shown in detail and compared among seven interventions (Online Supplementary Table S7). As for non-hemorrhagic adverse events, the pooled data showed no significant differences in incidence (Online Supplementary Figure S2A,B). The SUCRA rankings for adverse events revealed the best score for dexamethasone (69.3%; the lowest risk of adverse events), while rituximab + PSL had the lowest score (19.3%; the highest risk) and other interventions were associated with a modest risk (Online Supplementary Figure S2C). In the dexamethasone arm (12 RCT in total), the incidence of adverse events was 4.1% (22/531), com-

Early overall response Next, we compared the early OR (platelet count >30×109/L within 2 - 4 weeks after the initial therapies11). Data regarding the incidence of early OR were extracted from 20 studies with 1838 patients;24-33,35-44 the network maps are shown in Figure 3A. The total numbers of patients and the numbers of those who obtained an OR along with its definition in each RCT are presented in Online Supplementary Table S6. The pooled results indicate that the rhTPO arm (both rhTPO + dexamethasone and rhTPO + PSL) produced significantly superior responses compared to the PSL or dexamethasone monotherapy arms (Figure 3B). These two rhTPO regimens also offered better responses than the rituximab + dexamethasone regimen (RR, 1.39; 95% CI: 0.97 – 2.01; P=0.07 for rhTPO + dexamethasone, and RR, 2.56; 95% CI: 1.31 – 5.01; P<0.01 for rhTPO + PSL). The efficacy of rituximab + dexamethasone was almost the same as those of dexamethasone and PSL monotherapies (RR, 1.04: 95% CI: 0.86 – 1.25; P=0.69 haematologica | 2018; 103(1)

Table 2. Ranking of each arm according to the SUCRA values of sustained response and overall response.

SR Ranking Treatment 1 2 3 4 5 6 7 8 9 10

rhTPO+Dex RTX+Dex RTX+Dex+PSL Dex PSL RTX+PSL PSL(LD) mPSL±PSL IVIG±PSL PSL(LD)

SUCRA

Ranking

OR Treatment

SUCRA

97.1 81.3 81.1 56.2 41.5 34.6 27.7 18.5 11.9 10.8

1 2 3 4 5 6 7 8 9

rhTPO+PSL rhTPO+Dex RTX+Dex+PSL HP±PSL RTX+Dex Dex PSL IVIG±PSL mPSL±PSL

98.8 82.4 74.7 59.3 54.3 49.3 31.1 24.2 15.1

SUCRA: Surface under the cumulative ranking curve. Others are shown in Table 1.

167


Y. Arai et al.

pared to 6.0% (2/33) in the rituximab + PSL arm (including only 1 RCT); almost all of the adverse events were manageable. In studies using rhTPO, emergence of anti-thrombopoietin antibodies was not reported. As for hemorrhagic events, a total of nine events was observed in the PSL (n=4), rituximab + dexamethasone (n=3), and dexamethasone (n=2) arms ranging from petechiae to intracranial hemorrhage (Online Supplementary Table S7). No clear relationship with each arm and the incidence was detected.

Discussion

Publication bias was assessed using a funnel plot for the network of long-term SR and early OR (Online Supplementary Figure S3). All the included studies were symmetrically distributed around the vertical line, indicating that there was no significant publication bias in this network analysis.

This systematic review and network meta-analysis of the efficacy of up-front treatments for newly diagnosed ITP included 21 trials with 1898 randomly assigned participants. Our main findings showed significantly superior SR for the rhTPO + dexamethasone and rituximab + dexamethasone arms compared to those for the conventional therapies (PSL and dexamethasone monotherapy). Moreover, both rhTPO + dexamethasone and rhTPO + PSL also improved early OR compared to PSL, dexamethasone, or rituximab-containing regimens. Therapy-related adverse events showed similar profiles, and were tolerable in all treatment arms. The superior efficacy of rhTPO + dexamethasone compared to PSL or dexamethasone monotherapy as an initial treatment for newly diagnosed ITP with regards to longterm SR has been firstly shown in this meta-analysis. This

A

B

Publication bias

C

Figure 3. Results of the network of long-term sustained response comparison. (A) The network of comparisons included in the network meta-analysis for long-term sustained response (SR; platelet counts > 30Ă&#x2014;109/L at 3 - 6 months). The circle size is proportional to the total number of patients in the treatment group. The line width is proportional to the number of trials comparing the treatment groups. (B) The summary effect estimate (risk ratio of SR) for each combination of treatments. Risk ratios are indicated by dots, and 95% confidence intervals by bars. (C) The surface under the cumulative ranking curve (SUCRA) is shown for each treatment.

168

haematologica | 2018; 103(1)


Network meta-analysis for newly diagnosed ITP

methodology demonstrates more formally what some clinicians have suspected, and the result suggests that TPORA may possibly be a first choice for the treatment of newly diagnosed ITP in addition to therapy-resistant or chronic ITP, in which the additive effects of TPO-RA have previously been shown in various RCT and systematic reviews.7,46 Our results are partially supported by previous studies that analyzed blood thrombopoietin concentrations; serum thrombopoietin levels in patients with newly diagnosed ITP are within the normal range or only minimally elevated compared to those of healthy people,11,47 although data on serum thrombopoietin levels were not available in any RCT included in our study. Attention should be paid to the fact that in all the studies rhTPO was combined with PSL or dexamethasone; relatively shortterm administration of rhTPO alone (2 weeks) may not have resulted in such good responses. Moreover, it should be noted that rhTPO may be better than even rituximabcontaining regimens, which have been widely used recently for the treatment of newly diagnosed ITP after several RCT and systematic reviews reported that these latter are associated with higher efficiency and lower relapse risk than corticosteroid monotherapy.6 The results of this network meta-analysis, which allowed comparisons to be made even without head-to-head RCT (rhTPO versus rituximab), support the necessity for RCT directly comparing rhTPO and rituximab in order to validate our analysis and to determine the first-choice regimen that offers the best results in terms of long-term SR. However, it should be noted that rhTPO, but not romiplostim or eltrombopag, was used as the TPO-RA in this study. In addition to superior SR, the higher incidence of early OR with rhTPO regimens (compared to PSL, dexamethasone, and rituximab) in this study is also a novel finding. Previous studies have shown that rituximab regimens are not expected to accelerate platelet recovery or to improve early OR compared to corticosteroid monotherapy.31,37 However, rhTPO shortens the platelet recovery period in each RCT;36,41,43 as a result, the data synthesized in this network meta-analysis clearly demonstrated the improvement of early OR. This prominent effect of rhTPO + dexamethasone or + PSL can be obtained because each drug has a different mechanism of action; the former enhances platelet production in the bone marrow,48 while the latter increases the rate of apoptosis of autoantibody-producing lymphocytes and down-regulates macrophage activity responsible for platelet phagocytosis.49 This synergistic effect cannot be obtained from any combinations of various immunosuppressing agents (such as rituximab + dexamethasone). The serious adverse events were manageable in every treatment arm, even those with rhTPO or rituximab. This may be because (i) rituximab and rhTPO are drugs that generally have low incidences of severe adverse effects, and (ii) these additional treatments promote rapid and sustained platelet recovery, leading to shorter treatment periods and smaller corticosteroid dosages; as a result, steroidderived adverse events, such as hypertension, glucose intolerance, and infection, may be decreased in these treatment arms. Other treatment strategies, such as methylprednisolone and intravenous immune globulins, did not show any significant additive effects compared to conventional PSL or dexamethasone monotherapies. Among them, intravenous immune globulins induced platelet recovery withhaematologica | 2018; 103(1)

in days,26,27 indicating its therapeutic advantages for patients requiring rapid platelet recovery due to extremely severe thrombocytopenia and/or high risk of critical hemorrhage. However, this rapid recovery seems to be temporal and was not related to superior OR and SR. So far, we have described a methodology that reveals new insights into the efficacy and safely of treatments for newly diagnosed ITP, but this study has some limitations. First, only a few RCT were included for some treatment arms; for example, only one RCT was included for the H. pylori eradication,28 rhTPO + PSL,36 and rituximab + dexamethasone followed by consolidation PSL39 treatment arms, although the number of participants in each RCT was relatively large. This limitation generally results in a higher β error (lower power to detect differences) and a statistically unstable model. Sensitivity analysis excluding these three RCT (H. pylori eradication, rhTPO + PSL, and rituximab + dexamethasone + PSL) resulted in the same conclusions as for SR and OR results (data not shown), supporting the reliability of our results. Future RCT, however, are warranted to overcome this limitation. Including patients after splenectomy may be useful for a more comprehensive analysis. Second, meta-analyses of the outcomes were performed only for SR at 3 - 6 months and OR at 2 – 4 weeks. Other outcomes, such as SR at later time-points (1 – 3 years) or normalization of platelet counts after therapy (>100×109/L instead of 30×109/L), should be analyzed. Unfortunately, the numbers of RCT reporting these outcomes were very limited, so we could not carry out meta-analysis. Nevertheless, our outcomes, such as SR at 3- 6 months or OR at 2 – 4 weeks, are reasonable enough to evaluate the potency of these regimens because both of these outcomes represent sustained and early therapeutic effects in the management of newly diagnosed ITP.31,44 Third, regarding TPO-RA, only studies dealing with rhTPO were included in our analysis, because we were unable to identify RCT comparing other TPO-RA such as romiplostim and eltrombopag. rhTPO is now commercially available only in China, which will limit the geographical generalization of the findings of this study. Sensitivity analysis excluding these studies using rhTPO36,41,43 resulted in the same conclusions regarding the superiority of rituximab combined with corticosteroid compared to the monotherapy (Online Supplementary Table S8). The other type of rhTPO (pegylated, recombinant humsn megakaryocyte growth and development factor; PEG-rHuMGDF) was developed and tested in the 1990s, but the development of this type ended soon because of the emergence of autoantibodies to PEG-rHuMGDF,50 and no RCT involving this drug were found. Moreover, the efficacy of TPO-RA may differ according to the specific agent used.7 Hence, further RCT are necessary to confirming the benefits of TPO-RA as an initial treatment for newly diagnosed ITP. Lastly, insufficiency in adverse event reporting is another limitation. Detailed information on adverse events was obtained only from 16 RCT (76.2% of the total). The protocol to follow up toxicities may be different in each country, especially in the older studies. Judging from the data obtained, adverse events in the clinical course of ITP treatment are generally manageable, and treatment-related deaths were rarely observed in any of the RCT included in this study. These data suggest that any regimens analyzed in this study are feasible as up-front treatments for newly diagnosed ITP. In summary, this systematic review and network meta169


Y. Arai et al.

analysis demonstrated that this approach is reasonable for the analysis of ITP, and it indicated the efficacy of rhTPO in newly diagnosed ITP; the results suggest that rhTPO + dexamethasone or + PSL regimens may be up-front therapeutic options along with conventional corticosteroid monotherapy or rituximab. No significant differences in long-term SR or early OR were detected between patients given the two dosages of rituximab (100 mg/body versus 375 mg/m2), which may have some impact on the current

References 14. 1. Psaila B, Bussel JB. Immune thrombocytopenic purpura. Hematol Oncol Clin North Am. 2007;21(4):743-759. 2. Stasi R, Evangelista ML, Stipa E, Buccisano F, Venditti A, Amadori S. Idiopathic thrombocytopenic purpura: current concepts in pathophysiology and management. Thromb Haemost. 2008;99(1):4-13. 3. Cohen YC, Djulbegovic B, Shamai-Lubovitz O, Mozes B. The bleeding risk and natural history of idiopathic thrombocytopenic purpura in patients with persistent low platelet counts. Arch Intern Med. 2000;160(11): 1630-1638. 4. Provan D, Stasi R, Newland AC, et al. International consensus report on the investigation and management of primary immune thrombocytopenia. Blood. 2010;115(2):168-186. 5. Neunert C, Lim W, Crowther M, et al. The American Society of Hematology 2011 evidence-based practice guideline for immune thrombocytopenia. Blood. 2011;117(16): 4190-4207. 6. Chugh S, Darvish-Kazem S, Lim W, et al. Rituximab plus standard of care for treatment of primary immune thrombocytopenia: a systematic review and meta-analysis. Lancet Haematol. 2015;2(2):e75-81. 7. Wang L, Gao Z, Chen XP, et al. Efficacy and safety of thrombopoietin receptor agonists in patients with primary immune thrombocytopenia: a systematic review and metaanalysis. Sci Rep. 2016;6:39003. 8. Mithoowani S, Gregory-Miller K, Goy J, et al. High-dose dexamethasone compared with prednisone for previously untreated primary immune thrombocytopenia: a systematic review and meta-analysis. Lancet Haematol. 2016;3(10):e489-e496. 9. Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of multiple treatments: combining direct and indirect evidence. BMJ. 2005;331(7521):897-900. 10. Bucher HC, Guyatt GH, Griffith LE, Walter SD. The results of direct and indirect treatment comparisons in meta-analysis of randomized controlled trials. J Clin Epidemiol. 1997;50(6):683-691. 11. Lambert MP, Gernsheimer TB. Clinical updates in adult immune thrombocytopenia. Blood. 2017;129(21):2829-2835. 12. Chaimani A, Higgins JP, Mavridis D, Spyridonos P, Salanti G. Graphical tools for network meta-analysis in STATA. PLoS One. 2013;8(10):e76654. 13. Bussel JB, Eldor A, Kelton JG, et al. IGIV-C, a novel intravenous immunoglobulin: evaluation of safety, efficacy, mechanisms of

170

15.

16.

17.

18.

19.

20.

21.

22.

23.

usage of rituximab. However, the lower number of included RCT and patients and higher risk of bias in some RCT could be a limitation in our analysis. The high risk of bias for blinding can work in favor of new interventions (rituximab and rhTPO). Thus, future worldwide head-to-head RCT including these regimens (including romiplostim and eltrombopag) and rituximab-containing treatments are essential in order to determine the most suitable initial therapeutic strategies for newly diagnosed ITP in adults.

action, and impact on quality of life. Thromb Haemost. 2004;91(4):771-778. Wolf HH, Davies SV, Borte M, et al. Efficacy, tolerability, safety and pharmacokinetics of a nanofiltered intravenous immunoglobulin: studies in patients with immune thrombocytopenic purpura and primary immunodeficiencies. Vox Sang. 2003;84(1):45-53. Tsutsumi Y, Kanamori H, Yamato H, et al. Randomized study of Helicobacter pylori eradication therapy and proton pump inhibitor monotherapy for idiopathic thrombocytopenic purpura. Ann Hematol. 2005;84(12):807-811. Facon T, Caulier MT, Wattel E, Jouet JP, Bauters F, Fenaux P. A randomized trial comparing vinblastine in slow infusion and by bolus i.v. injection in idiopathic thrombocytopenic purpura: a report on 42 patients. Br J Haematol. 1994;86(3):678-680. Ferrari A, Pasqualetti D, Del Bianco P, Gandolfo GM, Chistolini A, Mazzucconi MG. Prednisone versus deflazacort in the treatment of autoimmune thrombocytopenic purpura: evaluation of clinical response and immunological modifications. Haematologica. 1991;76(4):342-345. Luo YG, Liu YQ, Hu J. [Clinical study on effect of recombinant roasted licorice decoction combined with low-dose glucocorticoids in treating idiopathic thrombocytopenic purpura]. Zhongguo Zhong Xi Yi Jie He Za Zhi. 2001;21(7):501-503. Yu XQ, Chen HM, Sun JH, Luo M, Lu YL. [Therapeutic efficacy of multigly-cosidorum tripterygium combined with rhIL-11 for immune thrombocytopenia]. Zhongguo Shi Yan Xue Ye Xue Za Zhi. 2015;23(5):14001403. Zeng FC, Hong F, Yang YQ. [Clinical study of zhinu-I,-II in treating 61 patients with idiopathic thrombocytopenic purpura]. Zhongguo Zhong Xi Yi Jie He Za Zhi. 1996;16(4):207-209. Zhou YM, Hu MH, Yang JM, et al. [Clinical study on idiopathic thrombocytopenic purpura treated with Shengxueling Granule]. Zhong Xi Yi Jie He Xue Bao. 2004;2(6):421425. George JN, Raskob GE, Vesely SK, et al. Initial management of immune thrombocytopenic purpura in adults: a randomized controlled trial comparing intermittent antiD with routine care. Am J Hematol. 2003;74(3):161-169. Newman GC, Novoa MV, Fodero EM, Lesser ML, Woloski BM, Bussel JB. A dose of 75 microg/kg/d of i.v. anti-D increases the platelet count more rapidly and for a longer period of time than 50 microg/kg/d in adults with immune thrombocytopenic purpura. Br J Haematol. 2001;112(4):1076-1078.

24. Mazzucconi MG, Francesconi M, Fidani P, et al. Treatment of idiopathic thrombocytopenic purpura (ITP): results of a multicentric protocol. Haematologica. 1985;70(4): 329-336. 25. Bellucci S, Charpak Y, Chastang C, Tobelem G. Low doses v conventional doses of corticoids in immune thrombocytopenic purpura (ITP): results of a randomized clinical trial in 160 children, 223 adults. Blood. 1988;71(4):1165-1169. 26. Jacobs P, Wood L, Novitzky N. Intravenous gammaglobulin has no advantages over oral corticosteroids as primary therapy for adults with immune thrombocytopenia: a prospective randomized clinical trial. Am J Med. 1994;97(1):55-59. 27. Godeau B, Chevret S, Varet B, et al. Intravenous immunoglobulin or high-dose methylprednisolone, with or without oral prednisone, for adults with untreated severe autoimmune thrombocytopenic purpura: a randomised, multicentre trial. Lancet. 2002;359(9300):23-29. 28. Kong R, Qiu HC, Wu PF, Niu XH, Shen WX, Wang Y. [Clinical significance of Helicobacter pylori in pathogenesis of idiopathic thrombocytopenic purpura]. Zhongguo Shi Yan Xue Ye Xue Za Zhi. 2008;16(5):1222-1226. 29. Praituan W, Rojnuckarin P. Faster platelet recovery by high-dose dexamethasone compared with standard-dose prednisolone in adult immune thrombocytopenia: a prospective randomized trial. J Thromb Haemost. 2009;7(6):1036-1038. 30. Bae SH, Ryoo H-M, Lee WS, et al. High dose dexamethasone vs. conventional dose prednisolone for adults with immune thrombocytopenia: a prospective multicenter phase III Trial. Blood. 2010;116:3687. 31. Zaja F, Baccarani M, Mazza P, et al. Dexamethasone plus rituximab yields higher sustained response rates than dexamethasone monotherapy in adults with primary immune thrombocytopenia. Blood. 2010; 115(14):2755-2762. 32. Cui ZG, Wei Y, Hou M, Zhao HG, Wang HY. [The efficacy and safety of 2 cycles' high-dose dexamethasone treatment adult primary immune thrombocytopenia]. Zhonghua Nei Ke Za Zhi. 2011;50(5):401403. 33. Li Z, Mou W, Lu G, et al. Low-dose rituximab combined with short-term glucocorticoids up-regulates Treg cell levels in patients with immune thrombocytopenia. Int J Hematol. 2011;93(1):91-98. 34. Arnold DM, Heddle NM, Carruthers J, et al. A pilot randomized trial of adjuvant rituximab or placebo for nonsplenectomized patients with immune thrombocytopenia.

haematologica | 2018; 103(1)


Network meta-analysis for newly diagnosed ITP

Blood. 2012;119(6):1356-1362. 35. Mashhadi MA, Kaykhaei MA, Sepehri Z, Miri-Moghaddam E. Single course of high dose dexamethasone is more effective than conventional prednisolone therapy in the treatment of primary newly diagnosed immune thrombocytopenia. Daru. 2012; 20(1):7. 36. Gu SY, Zhuang JL, Zou SH, et al. [A clinical comparative study on treatment of severe newly diagnosed immune thrombocytopenia by recombinant human thrombopoietin combined with glucocorticoid]. Zhonghua Xue Ye Xue Za Zhi. 2013;34 (10):883-886. 37. Gudbrandsdottir S, Birgens HS, Frederiksen H, et al. Rituximab and dexamethasone vs dexamethasone monotherapy in newly diagnosed patients with primary immune thrombocytopenia. Blood. 2013;121(11): 1976-1981. 38. Li ZY, Li DP, Yan ZL, et al. [Effect of different therapeutic regimens on regulatory T cells in patients of primary immune thrombocytopenia]. Zhonghua Xue Ye Xue Za Zhi. 2013;34(6):478-481. 39. Xing WW, Li ZY, Yan ZL, et al. [Efficacy and safety of low-dose rituximab combined with different dosage of glucocorticoids for immune thrombocytopenia]. Zhonghua

haematologica | 2018; 103(1)

Xue Ye Xue Za Zhi. 2013;34(5):409-412. 40. Din B, Wang X, Shi Y, Li Y. Long-term effect of high-dose dexamethasone with or without low-dose dexamethasone maintenance in untreated immune thrombocytopenia. Acta Haematol. 2015;133(1):124-128. 41. Li Y, Huang Q, Wang C, Muhebaier, An L, Wang X. [Efficacy and safety of high-dose dexamethasone combined with rhTPO for newly diagnosed adults with severe immune thrombocytopenia]. Zhonghua Xue Ye Xue Za Zhi. 2016;37(2):134-137. 42. Matschke J, Muller-Beissenhirtz H, Novotny J, et al. A Randomized trial of daily prednisone versus pulsed dexamethasone in treatment-naive adult patients with immune thrombocytopenia: EIS 2002 Study. Acta Haematol. 2016;136 (2):101-107. 43. Sun M, Wang X, Jiang M, et al. [A clinical analysis of treatment with recombinant human thrombopoietin combined with large doses of dexamethasone in primary immune thrombocytopenia]. Zhonghua Nei Ke Za Zhi. 2016;55(3):202-205. 44. Wei Y, Ji XB, Wang YW, et al. High-dose dexamethasone vs prednisone for treatment of adult immune thrombocytopenia: a prospective multicenter randomized trial. Blood. 2016;127(3):296-302.

45. Higgins JP, Jackson D, Barrett JK, Lu G, Ades AE, White IR. Consistency and inconsistency in network meta-analysis: concepts and models for multi-arm studies. Res Synth Methods. 2012;3(2):98-110. 46. Cooper KL, Fitzgerald P, Dillingham K, Helme K, Akehurst R. Romiplostim and eltrombopag for immune thrombocytopenia: methods for indirect comparison. Int J Technol Assess Health Care. 2012;28(3):249258. 47. Makar RS, Zhukov OS, Sahud MA, Kuter DJ. Thrombopoietin levels in patients with disorders of platelet production: diagnostic potential and utility in predicting response to TPO receptor agonists. Am J Hematol. 2013;88(12):1041-1044. 48. Provan D, Newland AC. Current management of primary immune thrombocytopenia. Adv Ther. 2015;32(10): 875-887. 49. Mizutani H, Furubayashi T, Imai Y, et al. Mechanisms of corticosteroid action in immune thrombocytopenic purpura (ITP): experimental studies using ITP-prone mice, (NZW x BXSB) F1. Blood. 1992;79(4):942947. 50. Kuter DJ. Milestones in understanding platelet production: a historical overview. Br J Haematol. 2014;165(2):248-258.

171


ARTICLE

Coagulation & Its Disorders

Ferrata Storti Foundation

Comparative profiling of HLA-DR and HLA-DQ associated factor VIII peptides presented by monocyte-derived dendritic cells

Ivan Peyron,1* Robin B. Hartholt,1* Laura Pedró-Cos,1 Floris van Alphen,2 Anja ten Brinke,3 Neubury Lardy,4 Alexander B. Meijer1,2,5 and Jan Voorberg1,6

Department of Plasma Proteins, Sanquin Research and Landsteiner Laboratory, Academic Medical Center, University of Amsterdam; 2Department of Research Facilities, Sanquin Research Amsterdam; 3Department of Immunopathology, Sanquin Research and Landsteiner Laboratory, Academic Medical Center, University of Amsterdam; 4 Department of Immunogenetics, Sanquin Research and Landsteiner Laboratory, Academic Medical Center, Amsterdam; 5Pharmaceutics, Utrecht Institute for Pharmaceutical Sciences, Utrecht University and 6Department of Vascular Medicine, Academic Medical Center, University of Amsterdam, the Netherlands 1

Haematologica 2018 Volume 103(1):172-178

*These authors contributed equally to this work.

ABSTRACT

T

Correspondence: j.voorberg@sanquin.nl

Received: June 22, 2017. Accepted: October 4, 2017. Pre-published: October 27, 2017.

doi:10.3324/haematol.2017.175083 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/172

he development of anti-factor VIII antibodies is a major complication of the treatment of patients with hemophilia A. Generation of high affinity anti-factor VIII antibodies is dependent on help provided by CD4+ T cells that recognize factor VIII-derived peptides presented on class II major histocompatibility complex on the surface of antigen-presenting cells. In order to identify the immune-dominant epitopes that can be presented to CD4+ T cells, we previously developed a mass spectrometry-based method to identify factor VIII-derived peptides that are presented on human leukocyte antigen (HLA)-DR. In the present work, we compared the repertoire of FVIII-derived peptide presented on HLA-DR and HLA-DQ. Monocyte-derived dendritic cells from nine HLA-typed healthy donors were pulsed with recombinant factor VIII. HLA-DR and HLA-DQ molecules were purified using monoclonal antibodies. Our data show that HLA-DQ and HLA-DR present a similar repertoire of factor VIII-derived peptides. However, the number of peptides associated with HLA-DQ was lower than that with HLADR. We also identified a peptide, within the acidic a3 domains of factor VIII, which is presented with higher frequency on HLA-DQ. Interestingly, this peptide was found to have a higher predicted affinity for HLA-DQ than for HLA-DR. Taken together, our data suggest that HLA-DQ participates in the presentation of factor VIII peptides, thereby contributing to the development of inhibitory antibodies in a proportion of patients with severe hemophilia A.

©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.

172

Introduction Hemophilia A is an X-linked coagulation disorder characterized by decreased levels of functional factor VIII (FVIII) in circulation. In the most severe form of hemophilia A (FVIII residual activity <1 IU/dL), the absence of functional endogenous FVIII leads to spontaneous bleeding episodes and life-threatening hemorrhages.1 To compensate for the lack of endogenous FVIII, therapeutic FVIII is administered intravenously to the patients either on-demand or under prophylaxis. While the current treatment protocols are successful in most patients and lead to a symptomfree, near-normal life expectancy2, about 30% of patients develop an immune response to the administered FVIII.3–6 The humoral response against therapeutic FVIII results in the generation of anti-FVIII antibodies that inhibit the pro-coagulant activity of FVIII (FVIII inhibitors). This is a serious complication of the treatment of haematologica | 2018; 103(1)


Presentation of FVIII on HLA-DR and HLA-DQ

patients with hemophilia A.7 FVIII inhibitors are predominantly of the IgG1 and IgG4 isotypes,8 suggesting that the anti-FVIII immune response is dependent on help provided by CD4+ T cells. Activation of FVIII-specific CD4+ T cells requires the internalization of FVIII by professional antigen-presenting cells, such as dendritic cells, macrophages or B cells. After intracellular processing, FVIII-derived peptides are presented at the cell surface in association with major histocompatibility class II (MHCII) molecules. The first signal leading to the activation of CD4+ T cells is provided by the interaction of the T-cell receptor with peptide-bound MHCII on the surface of antigen-presenting cells. Together with the expression of co-stimulatory molecules, the presentation of FVIII peptides by antigen-presenting cells primes and activates FVIII-specific CD4+ T cells. Subsequently, the FVIII-specific CD4+ T cells recognize peptide/MHCII complexes on the B-cell surface resulting in the activation of FVIII-specific B cells that differentiate into anti-FVIII IgG producing plasma cells or FVIII-specific memory B cells.9 Several genetic and non-genetic risk factors have been associated with the incidence of inhibitor development.10–17 Among them, the HLA haplotype of patients has been linked to the presence of FVIII inhibitors.13–16 Located on the short arm of chromosome 6, the class II HLA gene complex contains three loci, DP, DQ and DR. Each of these loci encodes at least one alpha chain (DPA, DQA and DRA, respectively) and a variable number of beta chain polypeptides (DPB, DRB and DQB, respectively). As of December 2016, 4,230 HLA class II alleles had been assigned, half of which are attributed to variations in DRB.18 Less than 10% of these alleles are commonly identified in unrelated individuals as described in the Common and Well Documented (CWD) catalogue assembled by the American Society for Histocompatibility and Immunogenetics (ASHI).19 A larger allele variation can be identified in Europe.20 Hence, the 2017 European Federation for Immunogenetics (EFI) CWD HLA catalogue reported a total of 1,048 CWD alleles. Sanchez-Mazas and co-workers identified 130 DRB1, 20 DQA1 and 86 DQB1 alleles.20 Since HLA class II molecules arise from the noncovalent association of non-identical alpha and beta chains, up to 130 different HLA-DR and 1720 HLA-DQ haplotypes can be found in the general European population.20 As mentioned earlier, various studies have investigated the association of HLA haplotypes with the presence of inhibitors. In several studies the haplotypes HLADRB1*15 and DQB1*0602 were significantly associated with inhibitor development.13,16 In general it appeared that HLA profile was not a major determinant for inhibitor development.13 This may potentially be due to the wide repertoire of FVIII peptides that can be presented on HLA class II. Alternatively, promiscuous binding of immunodominant peptides to multiple HLA molecules may explain the absence of a strong link between HLA profile and inhibitor development. Identification of the immunodominant T-cell epitopes of FVIII which are presented on HLA molecules is still a major challenge.21 Using a mass spectrometry-based approach, our group previously identified the immunodominant peptides derived from FVIII presented on HLA-DR.22,23 The aim of the present work was to address the repertoire of FVIII-derived peptides presented on HLA-DQ and to compare it to that presented on HLA-DR. Using monoclonal antibodies specific for HLA-DR or HLA-DQ, we characterized for the first time haematologica | 2018; 103(1)

the repertoire of FVIII-derived peptides presented on HLA-DQ.

Methods Purification of peptide/MHC complexes Blood was obtained after approval from the Sanquin Ethical Advisory Board, in accordance with the Declaration of Helsinki. FVIII-loaded mature monocyte-derived dendritic cells were harvested and resuspended in 500 μL of lysis buffer (10 mM Tris-HCl pH 8.0, 0.25% octyl-β-D-glucopyranoside, 1% sodium deoxycholate and complete protease inhibitor) for 30 min at 4°C. Lysates were then centrifuged at 20,000xg for 15 min at 4°C and incubated with 300 μL of Sepharose beads coupled with either anti-HLA-DR (L243) or anti HLA-DQ (SPV-L3) monoclonal antibodies. Following overnight incubation at 4°C, Sepharose beads were washed twice with lysis buffer, and five times with 10 mM Tris-HCl pH 8.0. Bound MHCII molecules were eluted using 10% acetic acid for 10 min at room temperature. Eluates were collected and heated for 15 min at 70°C to dissociate the peptide/MHCII complexes.

Mass spectrometry Samples were desalted using C18 STAGE tips prepared inhouse. STAGE tips were equilibrated with 100% acetonitrile and subsequently washed with 1% formic acid. Samples were loaded on STAGE tips and washed once with 1% formic acid and once with 1% formic acid supplemented with 5% acetonitrile. Peptides were eluted from STAGE tips with 60 μL 1% formic acid supplemented with 30% acetonitrile and concentrated to a final volume of 5 μL using vacuum centrifugation. Eluted peptides were separated using columns filled with 1.9 μm C18 particles (New Objective type FS360-75-8-N-5-C20, Inc., Woburn, MA, USA) at a flow rate of 300 nL/min, with a step-wise gradient from 0 to 72 % (v/v) acetonitrile in 0.1 M acetic acid. Column eluate was sprayed directly into the Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher Scientific Inc., Bremen, Germany) using a nanoelectrospray source with a spray voltage of 2.15 kV. Higher-energy collisional dissociation was performed in top-speed mode with 3 second cycles (400-1500 m/z, resolving power 120,000). The mass spectrometer was calibrated on a regular basis as recommended by the manufacturer in order to ensure a high mass accuracy.

Data analysis Peptides were identified using Proteome Discoverer 1.4 (Thermo Scientific, Bremen, Germany). Raw Xcalibur data files were screened against the UniprotKB non-redundant protein 25.H_sapiens.fase database with a mass deviation of 20 ppm, a fragment mass tolerance of 0.8 Da and a false positive discovery rate of 95%. All identified FVIII-derived peptides with high and medium confidence were grouped and aligned for each donor. The NetMHCIIpan 3.1 server24 was used to determine the binding core sequence of overlapping FVIII peptides identified by mass spectrometry as outlined in the Online Supplementary Information. The binding core sequence with highest affinity was used as the representative peptide for each group of identified peptides irrespectively of the associated HLA allele. Differences in peptide numbers were statistically assessed with a two-sided non-parametric Mann-Whitney U-test using Graphpad Prism 7.0. The heat map was generated with Perseus 1.5.6.0.25 The absolute numbers of uniquely identified peptides for each protein were clustered with the following settings: distance - Euclidean, linkage - average, number of clusters - 300, maximal number of iterations - 10, number of restarts - 1. 173


I. Peyron et al.

Results Comparative analysis of peptide presentation by HLA-DR and HLA-DQ In order to explore the repertoire of FVIII-derived peptides presented on HLA-DR and HLA-DQ, we generated monocyte-derived dendritic cells from nine healthy donors. The dendritic cells were pulsed with 100 nM of recombinant full-length FVIII and maturation was induced using lipopolysaccharide to stabilize the expression of HLA-peptide complexes at the cell surface. Subsequently, cells were lysed and HLA-DR or HLA-DQ molecules were purified using Sepharose beads conjugated with the monoclonal antibodies L243 (HLA-DR) or SPV-L3 (HLA-DQ). Samples were then analyzed using mass spectrometry. As depicted in Figure 1A, the total number of peptides found in the case of HLA-DQ was consistently lower than that in the case of HLA-DR (733.7 ± 216.3 versus 1724 ± 491.4, mean ± SD). Similarly, the number of FVIII-derived peptides eluted from HLA-DQ was lower than that eluted from HLA-DR (7.8 ± 7.6 versus 36.7 ± 17.2, mean ± SD, Figure 1B). We then investigated the relative expression of HLA-DR and HLADQ. As shown in Figure 1C, we found a 4-fold lower expression of HLA-DQ compared to HLA-DR on monocyte-derived dendritic cells, suggesting that the lower abundance of the peptides eluted from HLA-DQ compared to HLA-DR is due to the lower expression of HLA-DQ. In order to determine the selectivity of peptide presentation by HLA-DR and HLA-DQ, the proteins identified by their unique peptides were clustered based on their absolute peptide count for all donors. The top 40 clustered proteins identified in the case of HLA-DR and HLADQ were visualized in a heat map (Figure 2). Consistent with our previous observation, the overall number of peptides found was lower in the case of HLA-DQ compared to HLA-DR. Peptides derived from positively charged histones were preferentially associated with HLA-DQ. In contrast, HLA-derived peptides were primarily found to be presented on HLA-DR. Based on these findings, HLA-DR and HLA-DQ appear to present distinct but overlapping peptide repertoires. In addition, FVIII was the sixth hit in the case of HLA-DR, but was found only at position 35 in the heat map generated for HLA-DQ, suggesting that FVIII-derived peptides preferentially associate with HLA-DR.

that presented on HLA-DR. Of note, with the exception of one donor, no B domain-derived peptides were found to be presented on HLA-DQ.

A peptide derived from the acidic a3 domains is efficiently presented on HLA-DQ To investigate the reasons underlying the reduced number of FVIII peptides found to be presented on HLA-DQ, we compared the predicted affinities of the FVIII-derived peptides identified in both the HLA-DR and HLA-DQ

A

B

C

The repertoire of factor VIII-derived peptides presented on HLA-DQ overlaps with that on HLA-DR In order to compare the profiles of FVIII-derived peptides presented on HLA-DR and HLA-DQ, the binding cores of the unique FVIII-derived peptides were determined using NetMHCIIpan 3.1.24 For the nine donors, core peptides were predicted with respect to their HLA haplotypes. FVIII-derived peptides sharing the same predicted core were grouped. An overview of the predicted cores is shown in Figure 3 for the nine donors tested (A to I) for HLA-DR and HLA-DQ. The complete set of FVIII-derived peptides is provided in Online Supplementary Figure S1. To compare the repertoire of FVIII-derived peptides presented on HLA-DR and HLA-DQ, the predicted cores of FVIIIderived peptides identified in the nine donors were grouped. As shown in Figure 4A, most of the peptides identified to be presented on HLA-DQ were also identified on HLA-DR, suggesting that the repertoire of FVIIIderived peptides presented on HLA-DQ overlaps with 174

Figure 1. Comparative profiling of peptide presentation by HLA-DR and HLA-DQ. Five million monocyte-derived dendritic cells were pulsed with 100 nM FVIII and matured for 24 h. Peptides associated with HLA-DR and HLA-DQ were identified using mass spectrometry. (A) Panel A shows the total number of peptides identified after elution from either HLA-DR or HLA-DQ. (B) Panel B depicts the FVIIIspecific peptides found on either HLA-DR or HLA-DQ. Statistical differences were determined using the non-parametric Mann-Whitney U-test. (C) Two hundred thousand immature monocyte-derived dendritic cells were incubated with 5 μg/mL of monoclonal antibody L243 (HLA-DR) or SPV-L3 (HLA-DQ). The relative binding of L243 and SPV-L3 was determined by flow cytometry using Alexa Fluor 488-conjugated anti-mouse IgG2a secondary antibody. Inset: representative histogram of the flow cytometry analysis of HLA-DR and HLA-DQ expression.

haematologica | 2018; 103(1)


Presentation of FVIII on HLA-DR and HLA-DQ

pull-downs with respect to the HLA type of the corresponding donors. As depicted in Figure 4B, the majority of the FVIII-derived peptides identified have a higher predicted affinity for HLA-DR than for HLA-DQ. Together with the lower expression of HLA-DQ, the relatively low affinity of these FVIII-derived peptides for HLA-DQ could explain their under-representation in the HLA-DQ pulldowns. Interestingly, a peptide, derived from the acidic a3 domain displayed a higher affinity for HLA-DQ than for HLA-DR (Figure 4B). This peptide was identified with a similar frequency in HLA-DQ and HLA-DR eluates (Figure 4A). These observations suggest that this a3-derived peptide is preferentially presented on HLA-DQ.

Discussion In the present work we describe for the first time the repertoire of FVIII-derived peptides presented on HLADQ by monocyte-derived dendritic cells. In parallel, we also determined the repertoire of FVIII-derived peptides presented on HLA-DR. Overall, fewer peptides were presented on HLA-DQ than on HLA-DR. This observation was correlated with a 4-fold lower surface expression of HLA-DQ compared to HLA-DR. Our results are consistent with previous publications reporting lower RNA levels of the DQA1 chain compared to DRA126 and a lower surface expression of HLA-DQ compared to HLA-DR.27 In addition, while the dendritic cells used for HLA-DR and HLA-DQ immunoprecipitation were incubated with equal amounts of FVIII, comparative analysis of the total

A

FVIII-derived peptides presented on HLA-DR and HLADQ revealed a preferential presentation of FVIII-derived peptides on HLA-DR. This suggests that following FVIII processing, HLA-DR and HLA-DQ molecules within the endolysosomal compartment compete for the binding of FVIII-derived peptides. Consistent with previous work describing the repertoire of naturally presented peptides by mature monocyte-derived dendritic cells, we identified a large collection of peptides derived from proteins from intracellular compartments which were presented on HLA-DR and HLA-DQ.23,28–30 Peptides derived from endogenous proteins are expected to compete with FVIIIderived peptides for binding to MHCII. Indeed, our analysis of the peptides presented on HLA-DQ revealed the presentation of peptides derived from several other endogenous proteins prior to FVIII, further limiting the presentation of FVIII-derived peptides on HLA-DQ. While the overall presentation of FVIII-derived peptides seemed to be more limited on HLA-DQ compared to HLA-DR, we identified a set of peptides within the acidic a3 domain which was equally presented on both HLA-DR and HLADQ. Interestingly, the predicted affinities of this peptide identified an MHCII binding core with higher affinities for HLA-DQ than for HLA-DR alleles. This suggests that the lower expression of HLA-DQ can be compensated by a higher affinity of these peptides for HLA-DQ. The predicted affinity of the FVIII peptides identified in this study for HLA-DR and HLA-DQ varies considerably (see Figure 4B). A subset of HLA-DR-presented peptides is predicted to bind with high affinity to HLA-DR. In contrast, the predicted affinity for FVIII peptides presented on HLA-

B

Figure 2. Cluster analysis of proteins identified for HLA-DR and HLA-DQ. The absolute counts of unique peptide from all donors were grouped using hierarchical clustering for HLA-DR (left panel) and HLADQ (right panel). The top 40 proteins are displayed (black ≥ 80 peptides, blue = 40 peptides, red ≤ 10 peptides). Gene names (referred to as UniProt) are used to represent the proteins.

haematologica | 2018; 103(1)

175


I. Peyron et al.

Figure 3. Overview of factor VIII peptides presented by HLA-DR and HLA-DQ. FVIII-derived peptides presented on HLA-DR and HLA-DQ were identified using mass spectrometry. Core-peptide sequences and affinities were determined using NetMHCIIpan 3.1 for the different HLA-DR and HLA-DQ combinations. The core peptide with highest affinity was selected as the representative sequence for each group of overlapping peptides. In the case of overlapping discrepancies in the core-peptide sequence between HLA-DR and HLA-DQ, the core peptide determined for HLA-DR was used for the graphic representation. Core peptides were sorted based on their sequence localization. Each column represents the results for an individual donor. Green: peptide identified in the HLA-DR condition. Yellow: peptide identified in the HLA-DR and HLA-DQ condition. Red: peptide identified only in the HLA-DQ condition. Numbers within boxes indicate the total number of peptides represented by the specific core peptide. Core sequences indicated in bold correspond to previously documented T-cell epitopes: 1Steinitz et al., Blood 2012; 2Hu et al., JTH, 2004; 3 Reding et al., JTH, 2004; 4Jones et al., JTH, 2005; 5Reding et al., JTH, 2003.

DQ appears to be quite low with an overall predicted affinity of >100 nM (Figure 4B). The significance of these observations is currently unknown. The strength of association between the T-cell receptor and the peptide-MHC complex was repeatedly found to dictate the activation and polarization of T cells.31,32 Very few reports have addressed the contribution of the peptide affinity for the HLA molecule to the strength of association between the T-cell receptor and peptide-MHC complex. Evidence has been obtained, in a model of influenza infection, in support of antigen signal strength during the priming of CD4+ T cells playing a crucial role for the cellsâ&#x20AC;&#x2122; effector function.33 In that study, it was argued that peptide-MHC stability has a significant impact on the functional properties of the effector CD4+ T cells.33 Similarly, the range of predicted affinities of the FVIII peptides identified in this study could guide the development of functionally diverse CD4+ T cells. The predicted lower affinity of HLA-DQ-presented peptides would need to be confirmed by in vitro peptide-bind176

ing assays. The affinity of a small number of FVIII peptides for different MHCII molecules has been experimentally determined. A C1 domain-derived peptide Ile2163Thr2180 bound with high affinity (IC50 below 20 nM) to several DRB1 molecules.34 In contrast, a much lower affinity (IC50 ranging from 0.5 to 1.0 ÎźM) was measured for binding of an A2 domain-derived peptide (Gln611Leu622) to DRB1*11.35 The range of predicted affinities for the FVIII peptides identified in this study is consistent with the experimentally determined values. It is, therefore, likely that the set of FVIII peptides identified in this study is representative of the repertoire of FVIII peptides that is presented on antigen-presenting cells of patients with hemophilia A (and most likely by default also on antigen-presenting cells of healthy individuals). Our current knowledge about the peptide repertoire recognized by FVIII-specific CD4+ T cells in hemophilia A patients with inhibitors is still limited. Nevertheless, the repertoire of experimentally determined CD4+ T-cell epitopes overhaematologica | 2018; 103(1)


Presentation of FVIII on HLA-DR and HLA-DQ

A

B

Figure 4. Comparison of the factor VIII-derived peptide repertoire presented on HLA-DR and HLA-DQ. (A) The data obtained from the nine donors were grouped based on core-peptide sequence and their frequency (out of 9). Blue and red bars represent the frequency with which the FVIII core peptide was found in the HLA-DR and HLA-DQ conditions, respectively. Core sequences indicated in bold correspond to previously documented T-cell epitopes. (B) The affinity of the peptides found in both the HLA-DR and HLA-DQ conditions was determined using NetMHCIIpan 3.1 and plotted as paired peptides. Data points corresponding to the acidic a3 peptides are depicted in red.

laps significantly with the repertoire of naturally presented FVIII peptides as identified in this and previous studies.22,23 This is exemplified by the peptides with core sequences of VITLKNMAS, ARAWPKNHT, LIIFKNQAS, FRNQASRPY and YSIRSTLRM, which were previously identified in a humanized E17 HLA-DRB1*15:01 mouse model.36 While these peptides were found associated with HLA-DRB1*15:01, we also detected these peptides on non-DRB1*15:01 HLA molecules. Conversely, several peptides identified as CD4+ T-cell epitopes by Steinitz and co-workers could not be detected by our experimental approach. This may be due to potential limitations in our current experimental protocols. In the present study, a higher-energy collision-induced dissociation strategy was employed for peptide fragmentation. Recently, Mommen et al. employed multiple peptide fragmentation technologies (electron-transfer dissociation, higher-energy collision-induced dissociation and combined electron-transfer/higher-energy collision dissociation) to successfully expand the HLA ligandome.29,37 In the present work, C18 STAGE tips were used to process HLA-DR and HLA-DQ eluates. The use of multi STAGE-Tip consisting of a hydrophobic C18 disk combined with a strong cation exchange disk could potentially increase the diversity of haematologica | 2018; 103(1)

identified peptides by allowing for binding for peptides with different chemical properties.38 While HLA haplotypes are commonly investigated as a genetic risk factor in various human diseases, very few studies have linked the expression levels of HLA gene products with the susceptibility to or outcome of autoimmune diseases. This was exemplified by Cavalli et al., who did not associate the susceptibility to auto-immune vitiligo development with a specific HLA haplotype, but with three single nucleotide polymorphisms located in a region that regulates the expression of the HLA genes.39 Several studies have investigated the relation between HLA haplotypes (especially DRB1) and the presence of inhibitor in hemophilia A patients, yielding variable results.13â&#x20AC;&#x201C;16 Whether the other three HLA-DRB encoding-genes (DRB3, DRB4 and DRB5) as well as HLA expression levels participate in the pathogenesis of inhibitor formation in patients with hemophilia A remains to be established. Acknowledgments The authors would like to thank Carmen van der Zwaan, Eduard Ebberink, Annemarie Koornneef, Arjan Hoogendijk and Maartje van den Biggelaar for their assistance in acquiring and analyzing the mass spectrometry data. 177


I. Peyron et al.

References 1. Franchini M, Mannucci PM. Past, present and future of hemophilia: a narrative review. Orphanet J Rare Dis. 2012;7(24):24. 2. Fijnvandraat K, Cnossen MH, Leebeek FWG, Peters M. Diagnosis and management of haemophilia. BMJ. 2012;344:e2707. 3. Franchini M, Tagliaferri A, Mengoli C, Cruciani M. Cumulative inhibitor incidence in previously untreated patients with severe hemophilia A treated with plasma-derived versus recombinant factor VIII concentrates: a critical systematic review. Crit Rev Oncol Hematol. 2012;81(1):82–93. 4. Gouw SC, van den Berg HM, Fischer K, et al. Intensity of factor VIII treatment and inhibitor development in children with severe hemophilia A: the RODIN study. Blood. 2013;121(20):4046–4055. 5. Hay CRM, Palmer B, Chalmers E, et al. Incidence of factor VIII inhibitors throughout life in severe hemophilia A in the United Kingdom. Blood. 2011;117(23):6367–6370. 6. Peyvandi F, Mannucci PM, Garagiola I, et al. A Randomized trial of fctor VIII and neutralizing antibodies in hemophilia A. N Engl J Med. 2016;374(21):2054–2064. 7. Kempton CL, Meeks SL. Toward optimal therapy for inhibitors in hemophilia. Blood. 2014;124(23):3365–3372. 8. Towfighi F, Gharagozlou S, Sharifian RA, et al. Comparative measurement of anti-factor VIII antibody by Bethesda assay and ELISA reveals restricted isotype profile and epitope specificity. Acta Haematol. 2005;114(2):84– 90. 9. Wroblewska A, Reipert BM, Pratt KP, Voorberg J. Dangerous liaisons: how the immune system deals with factor VIII. J Thromb Haemost. 2013;11(1):47–55. 10. Reipert BM. Risky business of inhibitors: HLA haplotypes, gene polymorphisms, and immune responses. Hematol Am Soc Hematol Educ Progr. 2014;2014(1):372–378. 11. Lövgren KM, Søndergaard H, Skov S, Wiinberg B. Non-genetic risk factors in haemophilia A inhibitor management - the danger theory and the use of animal models. Haemophilia. 2016;1–10. 12. Oldenburg J, Pavlova A. Genetic risk factors for inhibitors to factors VIII and IX. Haemophilia. 2006;12 (Suppl 6):15–22. 13. Hay CR, Ollier W, Pepper L, et al. HLA class II profile: a weak determinant of factor VIII inhibitor development in severe haemophilia A. UKHCDO Inhibitor Working Party. Thromb Haemost. 1997;77(2):234–237. 14. Ohta H, Takahashi I, Kojima T, et al. Histocompatibility antigens and alleles in Japanese haemophilia A patients with or without factor VIII antibodies. Tissue Antigens. 1999;54(1):91–97. 15. Pinto P, Parasannanavar D, Ghosh K, Shetty

178

16.

17.

18.

19.

20.

21. 22.

23.

24.

25.

26.

27.

S. The association of HLA-DRB1 and HLADQB1 alleles with the development of factor VIII inhibitors in severe haemophilia A patients in India. Tissue Antigens. 2014;84(2):235–237. Pavlova A, Delev D, Lacroix-Desmazes S, et al. Impact of polymorphisms of the major histocompatibility complex class II, interleukin-10, tumor necrosis factor-alpha and cytotoxic T-lymphocyte antigen-4 genes on inhibitor development in severe hemophilia A. J Thromb Haemost. 2009;7(12):2006– 2015. Repessé Y, Peyron I, Dimitrov JD, et al. Development of inhibitory antibodies to therapeutic factor VIII in severe hemophilia A is associated with microsatellite polymorphisms in the HMOX1 promoter. Haematologica. 2013;98(10):1650–1655. Robinson J, Halliwell JA, Hayhurst JD, Flicek P, Parham P, Marsh SGE. The IPD and IMGT/HLA database: allele variant databases. Nucleic Acids Res. 2015;43(Database issue):D423-431. Mack SJ, Cano P, Hollenbach JA, et al. Common and well-documented HLA alleles: 2012 update to the CWD catalogue. Tissue Antigens. 2013;81(4):194–203. Sanchez-Mazas A, Nunes JM, Middleton D, et al. Common and well-documented HLA alleles over all of Europe and within European sub-regions: a catalogue from the European Federation for Immunogenetics. HLA. 2017;89(2):104–113. Hartholt RB, Peyron I, Voorberg J. Hunting down factor VIII in the immunopeptidome. Cell Immunol. 2016;301:59–64. Van Haren SD, Wroblewska A, Herczenik E, et al. Limited promiscuity of HLA-DRB1 presented peptides derived of blood coagulation factor VIII. PLoS One. 2013;8(11):1–11. van Haren SD, Herczenik E, ten Brinke A, Mertens K, Voorberg J, Meijer AB. HLA-DRpresented peptide repertoires derived from human monocyte-derived dendritic cells pulsed with blood coagulation factor VIII. Mol Cell Proteomics. 2011;10(6): M110.002246. Andreatta M, Karosiene E, Rasmussen M, Stryhn A, Buus S, Nielsen M. Accurate panspecific prediction of peptide-MHC class II binding affinity with improved binding core identification. Immunogenetics. 2015;67 (11–12):641–650. Tyanova S, Temu T, Sinitcyn P, et al. The Perseus computational platform for comprehensive analysis of (prote)omics data. Nat Methods. 2016;13(9):731–740. Fernandez S, Wassmuth R, Knerr I, Frank C, Haas JP. Relative quantification of HLADRA1 and -DQA1 expression by real-time reverse transcriptase-polymerase chain reaction (RT-PCR). Eur J Immunogenet. 2003;30(2):141–148. Pickl WF, Majdic O, Kohl P, et al. Molecular

28.

29.

30. 31.

32.

33.

34.

35.

36.

37.

38.

39.

and functional characteristics of dendritic cells generated from highly purified CD14+ peripheral blood monocytes. J Immunol. 1996;157(9):3850–3859. Ciudad MT, Sorvillo N, van Alphen FP, et al. Analysis of the HLA-DR peptidome from human dendritic cells reveals high affinity repertoires and nonconventional pathways of peptide generation. J Leukoc Biol. 2017;101(1):15-27. Mommen GPM, Marino F, Meiring HD, et al. Sampling from the proteome to the human leukocyte antigen-DR (HLA-DR) ligandome proceeds via high specificity. Mol Cell Proteomics. 2016;15(4):1412–1423. Münz C. Autophagy proteins in antigen processing for presentation on MHC molecules. Immunol Rev. 2016;272(1):17–27. Gascoigne NRJ, Rybakin V, Acuto O, Brzostek J. TCR Signal strength and T cell development. Annu Rev Cell Dev Biol. 2016;32(1):327–348. Van Panhuys N. TCR signal strength alters T-DC activation and interaction times and directs the outcome of differentiation. Front Immunol. 2016;7:6. Nagaoka M, Hatta Y, Kawaoka Y, Malherbe LP. Antigen signal strength during priming determines effector CD4 T cell function and antigen sensitivity during influenza Virus Challenge. J Immunol. 2014;193(6): 2812– 2820. Jacquemin M, Vantomme V, Buhot C, et al. CD4+ T-cell clones specific for wild-type factor VIII: a molecular mechanism responsible for a higher incidence of inhibitor formation in mild/moderate hemophilia A. Blood. 2003;101(4):1351–1358. James EA, van Haren SD, Ettinger RA, et al. T-cell responses in two unrelated hemophilia A inhibitor subjects include an epitope at the factor VIII R593C missense site. J Thromb Haemost. 2011;9(4):689–699. Steinitz KN, Van Helden PM, Binder B, et al. CD4+ T-cell epitopes associated with antibody responses after intravenously and subcutaneously applied human FVIII in humanized hemophilic E17 HLA-DRB1*1501 mice. Blood. 2012;119(17):4073–4082. Mommen GPM, Frese CK, Meiring HD, et al. Expanding the detectable HLA peptide repertoire using electron-transfer/higher-energy collision dissociation (EThcD). Proc Natl Acad Sci USA. 2014;111(12):4507–4512. Rappsilber J, Mann M, Ishihama Y. Protocol for micro-purification, enrichment, pre-fractionation and storage of peptides for proteomics using StageTips. Nat Protoc. 2007;2(8):1896–1906. Cavalli G, Hayashi M, Jin Y, et al. MHC class II super-enhancer increases surface expression of HLA-DR and HLA-DQ and affects cytokine production in autoimmune vitiligo. Proc Natl Acad Sci USA. 2016;113(5):1363– 1368.

haematologica | 2018; 103(1)


ARTICLE

Coagulation & Its Disorders

Analyses of the FranceCoag cohort support differences in immunogenicity among one plasma-derived and two recombinant factor VIII brands in boys with severe hemophilia A Thierry Calvez,1 Hervé Chambost,2,3 Roseline d'Oiron,4 Vincent Dalibard,5 Virginie Demiguel,6 Alexandra Doncarli,6 Yves Gruel,7 Yoann Huguenin,8 Patrice Lutz,9 Chantal Rothschild,10 Christine Vinciguerra11 and Jenny Goudemand5,12 for FranceCoag Collaborators*

Sorbonne Universités, UPMC Paris 06, Inserm, Institut Pierre Louis d'Épidémiologie et de Santé Publique (IPLESP UMRS 1136); 2Service d'Hématologie Oncologie Pédiatrique, La Timone, APHM, Marseille; 3Aix Marseille University, INSERM, INRA, NORT, Marseille; 4 Centre Régional de Traitement de l’Hémophilie, Hôpital Bicêtre, Hôpitaux Universitaires Paris Sud, APHP, Le Kremlin Bicêtre; 5Département d'Hématologie et de Transfusion, Centre Hospitalier Universitaire de Lille, Université Lille 2, EA 2693, Faculté de Médecine; 6Santé Publique France, French National Public Health Agency, SaintMaurice; 7Centre Régional de Traitement de l’Hémophilie, Laboratoire d’Hématologie, UMR CNRS 7292, Université François Rabelais, Tours; 8Service d'Hématologie, Hôpital Pellegrin Tripode, Bordeaux; 9Unité Pédiatrique d’Hématologie Oncologie, Hôpital Hautepierre, Strasbourg; 10Centre Régional de Traitement de l’Hémophilie, Hôpital Necker, APHP, Paris; 11Service d'Hématologie Biologique, Hospices Civils de Lyon, EA 4609, Université de Lyon and 12Institut Pasteur de Lille, EGID, Inserm UMR 1011, Université Lille 2, France 1

*FranceCoag Collaborators (n=111) Adjaoud D, Aouba A, Ardillon L, Barbay V, Barro C, Bastenaire B, Bayart S, Behar C, Benz-Lemoine E, Berger C, Berny K, Bertrand MA, Beurrier P, Bianchin M, Biernat J, Biron-Andreani C, Blanc L, Borg JY, Bovet J, Briquel ME, Castet S, Coatmelec B, Codine P, Costa C, Costagliola D, De Lumley L, De Raucourt E, Demay Y, Derlon A, Desprez D, Deville A, Donadel Claeyssens S, Donadio D, Dumesnil C, Durin-Assollant A, Dutrillaux F, Falaise C, Faradji A, Ferry N, Fiks Sigaud M, Fimbel B, Fouassier M, Fressinaud E, Frotscher B, Gaillard S, Gautier P, Gay V, Gembara P, Gorde S, Grémy I, Guerois C, Guillaume Y, Guillet B, Guérin V, Harroche A, Hassoun A, Henni T, Lambert T, Laurian Y, Legrand F, Li-Thiao-Te V, Lienhart A, Macchi L, Meunier S, Micheau M, Milien V, Monlibert B, Monpoux F, Moreau P, Munzer M, Navarro R, Négrier C, Normand C, Nyombe P, Oudot C, Ounnoughene N, Pan Petesch B, Parquet A, Paugy P, Pautard B, Peynet J, Pincemaille O, Pineau-Vincent F, Polack B, Pouille Lievin O, Pouplard C, Pouzol P, Rafowicz A, Rauch A, Regina S, Ricard C, Robert V, Rospide P, Ryman A, Sainte Marie I, Sannié T, Schneider P, Schoepfer C, Schved JF, Stieltjes N, Stoven C, Tarral E, Thiercelin Legrand MF, Tintillier V, Toguyeni E, Torchet MF, Trossaërt M, Valentin JB, Vannier JP, Volot F, Wibaut B

ABSTRACT

A

round one third of boys with severe hemophilia A develop inhibitors (neutralizing antibodies) against their therapeutic factor VIII product. This adverse effect may result in more lifethreatening bleeding, disability, impaired quality of life, and costly care. We compared the incidence of inhibitors in boys treated with the three factor VIII products most used in France: one plasma-derived (Factane) and two recombinant products (Advate and Kogenate Bayer). A previously untreated cohort of patients was created in 1994 to investigate risk factors for inhibitor development. We selected boys with severe hemophilia A (factor VIII <1 IU/dL) first treated with one of the three factor VIII products studied. Details of product infusions, inhibitor assays and main fixed and time-varying inhibitor risk factors were recorded for the first 75 exposure days. Three outcomes (all inhibitors, high-titer inhibitors and subsequently treated inhibitors) were analyzed by univariate and multivariate Cox models. We studied 395 boys first treated between 2001 and 2016 (131, 137, and 127 with Factane, Advate, and Kogenate Bayer, respectively). Clinically significant inhibitors were diagnosed in 121 patients (70 high-titer). The incidence of high-titer inhibitors was significantly associated with the factor VIII product received (P=0.005): the cumulative incidence at 75 exposure days was 12.7% (95% CI: 7.7-20.6) with Factane, 20.4% (95% CI: 14.0-29.1) with haematologica | 2018; 103(1)

Ferrata Storti Foundation

Haematologica 2018 Volume 103(1):179-189

Correspondence: thierry.calvez@iplesp.upmc.fr

Received: June 16, 2017. Accepted: October 5, 2017. Pre-published: October 12, 2017.

doi:10.3324/haematol.2017.174706 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/1/179 ©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.

179


T. Calvez et al.

Advate, and 31.6% (95% CI: 23.5-41.7) with Kogenate Bayer. The low inhibitor incidence observed with Factane is concordant with recent findings from the SIPPET randomized trial. These consistent results from observational and experimental studies should lead to improved care for previously untreated patients and cost savings for healthcare systems worldwide.

Introduction Hemophilia A is a hereditary disorder caused by a quantitative and/or qualitative deficiency of a coagulation protein, factor VIII (FVIII). Prevention and treatment of bleeding rely on repeated infusions of the deficient clotting factor.1 The first FVIII products were plasma-derived. From 1993 to 2012, six recombinant products were authorized in the European Union. This group of products has become the most prescribed in Western countries.2 Since the effective prevention of transmission of human immunodeficiency virus and hepatitis C virus infections via blood products in the late 1980s, inhibitors (neutralizing antibodies against therapeutic FVIII) have been the most serious adverse effect of the treatment of hemophilia A. Around one third of boys with severe hemophilia A (FVIII <0.01 IU/dL) develop inhibitors during their first 50 exposure days (EDs). An ED is defined as a day when at least one FVIII dose is infused. In most cases, inhibitors substantially impair the outcome of hemophilia A and increase its management cost.3 The identification of modifiable inhibitor risk factors (FVIII product or other treatment modalities) is therefore a major issue. Unfortunately, to date, firm knowledge is lacking, leaving regulatory authorities, prescribers and hemophilia patients in doubt. Most knowledge comes from observational studies, of varying quality, which compare treatment modalities in previously untreated patients (PUPs). In the 2000s, the first comparative studies, albeit limited, suggested that incidence of inhibitors was higher with recombinant products than with plasmaderived ones in PUPs with severe hemophilia A4–6 (hereafter, inhibitor risk associated with FVIII products is termed "immunogenicity" regardless of the mechanisms involved). In the 2010s, two systematic reviews7,8 and a patient-level meta-analysis9 did not support a potential difference in immunogenicity between recombinant and plasma-derived products in PUPs. In 2013, the largest and most extensive international PUP cohort study found no such difference, however, only 88 PUPs were treated with 15 different plasma-derived products.10 Unexpectedly, this study showed an inhibitor incidence difference between the two most prescribed recombinant products. In 2014, this difference was confirmed by ad hoc analyses of two national PUP cohorts.11,12 However, in the absence of demonstrated pathophysiological mechanisms, these results have been hotly debated.13–17 Nevertheless, they support the concept of considering the immunogenicity of each FVIII product rather than its source (recombinant versus plasma-derived). Launched in 2010 and published in 2016, the SIPPET trial focused on immunogenicity according to product source, demonstrating a higher incidence of inhibitors in children treated with recombinant products.18,19 Until now, SIPPET remains the only randomized trial addressing product immunogenicity in children 180

with hemophilia A. Such trials are challenging as the target population is very young, and the children often require immediate treatment at diagnosis. Thus regulatory agencies and authors have recommended systematic enrollment of PUPs in standardized national or international follow-up to rapidly determine the immunogenicity of newly marketed FVIII products.20–22 However, establishing such pharmacosurveillance systems takes time and currently very few well-documented PUP cohorts are available worldwide. In 1994, a national PUP cohort dedicated to the study of genetic and non-genetic inhibitor risk factors was established in France,23 where a single plasma-derived product has been overwhelmingly used since 2001. In this context, we compared the inhibitor incidence in PUPs with severe hemophilia A treated with this plasma-derived product and those treated with two recombinant products during the same period.

Methods Study design In France, the public health authorities created a national pharmacosurveillance system in 1994 for FVIII and factor IX products administered to hemophiliacs.23 Clinicians of all hemophilia treatment centers were invited to include all hemophilia patients in an observational open cohort. In 2003, this system was renamed FranceCoag and inclusion was extended to other hereditary bleeding disorders. The high observed average prevalence of hemophilia A at birth (23.3 cases per 100 000 male live births for 1991-2008) compared with prevalences in other industrialized countries supports the exhaustiveness of this registry.11,24 Since 1994, PUPs with hemophilia (FVIII or factor IX <2 IU/dL) have been enrolled in a sub-cohort with detailed follow-up and data collection to investigate risk factors for inhibitor development and the impact of prophylaxis. FranceCoag is fully publicly funded, and governed by a steering committee representing all stakeholders (Online Supplementary Data). FranceCoag was authorized by the French data protection authority. Parents or legal guardians of PUPs were informed about its objectives and modalities and approved their child’s enrollment in accordance with the Declaration of Helsinki.

Patients From 2000 onward, all children diagnosed with hemophilia A and FVIII <2 IU/dL have had to be included in the FranceCoag PUP cohort before treatment initiation, or shortly thereafter. For this analysis, we selected boys with severe hemophilia A and welldocumented FVIII treatment with details of the first EDs (see the following paragraph). Among them, we selected boys first treated with a currently marketed product used sufficiently to enable informative comparisons.

Data collection The data collected have been described previously.11,16 Briefly, in the FranceCoag PUP cohort, main fixed inhibitor risk factors (e.g., haematologica | 2018; 103(1)


Inhibitor incidence according to 3 FVIII brands

F8 gene defect, family history of hemophilia and inhibitor, ethnic origin) were recorded at inclusion or shortly thereafter. Quarterly visits were recommended until ED-150. At each follow-up visit, hemorrhagic events, surgical procedures, treatments received and results of all inhibitor assays since the previous visit were accurately recorded. All data were centralized via a dedicated website. In parallel, details of the first 75 EDs [date, reason(s) for treatment, FVIII product, dose and body weight] were recorded on a spreadsheet for each patient from his own booklet and hospital records. Data were automatically checked for inconsistencies and closely monitored by three dedicated clinical research assistants who compared the database with the original files in the centers.

Follow-up and outcomes The cutoff date was December 6, 2016. Only the first 75 EDs were considered. If inhibitors developed during this observational period, EDs were counted until the last ED before their detection.

If the patient had not reached 75 EDs at the last clinical visit or in the event of death or a switch to another FVIII product, follow-up was censored after the last recorded ED with the initial product. Classic outcomes were considered: all clinically significant inhibitors defined as a positive result (titer >0.6 Bethesda units) found in two distinct samples and high-titer inhibitors with a peak titer of at least 5 Bethesda units at any time. As previously published,11 we also considered inhibitors subsequently treated at any time with a bypassing agent and/or immune tolerance induction, as these treatments impair the patient's quality of life and represent a significant economic burden for the community. Inhibitor assays were performed in each center's laboratory. All of these laboratories participate in the international external quality assessment program of the European Concerted Action on Thrombosis Foundation.25 All inhibitor cases were validated by an ad hoc committee using a standardized procedure (Online Supplementary Data).

Figure 1. Patient selection process. At the cutoff date (December 6, 2016), 649 previously untreated patients (PUPs) with hemophilia A (factor VIII <2 IU/dL) had been included in the dedicated cohort of FranceCoag. After the selection process, three groups of boys with severe hemophilia A (factor VIII <1 IU/dL) were formed based on the first factor VIII product received. MA: marketing authorization dates in European Union (or in France for FactaneÂŽ).

haematologica | 2018; 103(1)

181


T. Calvez et al.

Factor VIII products studied

Sensitivity analyses

We compared the three FVIII products still marketed in the European Union and sufficiently represented in our PUP cohort (Figure 1). Factane® (manufactured by LFB) is a plasma-derived FVIII product that originates from cryoprecipitate of large pools of plasma from blood donors. It is co-purified by ion exchange chromatography with von Willebrand factor (20-40 IU per 100 IU FVIII). Its virus removal/inactivation steps include solvent/detergent inactivation and nanofiltration on 35 and 15 nm filters.26 Advate® (Baxalta, acquired by Shire in 2016) is a third-generation recombinant FVIII product. The full-length FVIII protein is produced in Chinese hamster ovary cells. Neither human nor animal protein is used in the fermentation process and the lyophilized preparation is stabilized with trehalose and mannitol.27 Kogenate® Bayer (Bayer HealthCare) is a second-generation recombinant FVIII product. The full-length FVIII protein is produced in baby hamster kidney cells and the final preparation is stabilized with sucrose.28 This FVIII product is also distributed as Helixate® NexGen by CSL Behring. Hereafter, these three studied FVIII products are termed Factane, Advate, and Kogenate, respectively.

Two sensitivity analyses were performed, one by selecting patients first treated in the same period (from 2004 for Advate versus Factane comparisons, and until 2013 for Kogenate versus Factane comparisons) and the other by using real time instead of ED as the time unit in Cox models.

Statistical analysis We compared inhibitor incidence between PUPs treated with the plasma-derived product widely used in France since 2001 (Factane) and those treated with the two recombinant products most used in the same period (Advate and Kogenate). First, the cumulative incidences of inhibitor according to product received were represented by Kaplan-Meier curves for the three outcomes. This representation was also used to analyze the product interactions with: the first exposure period, and treatment intensity at first exposure. Associations between the products received and inhibitor incidence were analyzed by a Cox proportional hazards model. Time was measured in terms of EDs, as is usual for inhibitor risk analyses. An ED was defined as a day during which one or more infusions of FVIII were given. As in our previous article,11 in the case of a product switch, patients were considered as still being exposed to the first FVIII product for seven calendar days following the first infusion of the new product. This period was considered as the shortest latency between a risk factor exposure and a detectable related inhibitor occurrence. For the high-titer inhibitor analysis, the follow-up of patients who developed a low-titer inhibitor was censored at its onset. Similarly, for the analysis of treated inhibitors (with a bypassing agent and/or immune tolerance induction), the follow-up of patients who developed an untreated inhibitor was censored at its onset. Multivariate analysis included, individually and then together, four fixed cofactors (F8 gene defect, family history of hemophilia and inhibitor, ethnic origin, and age at first exposure) and five timevarying cofactors (calendar period, peak treatment episodes ≥5 and ≥10 consecutive EDs, severe bleeding episodes, and regular prophylaxis). Follow-up was censored upon any switch of product, so this factor was considered as fixed and Factane was chosen as the reference. The analyses were repeated, taking propensity scores into account by two methods: stratifying by quintiles of the propensity score and inverse probability of treatment weighting.29–31 Crude and adjusted hazard ratios with 95% confidence intervals (95% CI) were reported for Advate versus Factane and then for Kogenate versus Factane. Stata statistical software release 13.1 (College Station, TX StataCorp LP, USA) was used for all analyses and figures. For cofactor definitions, grouping, missing data procedures, and propensity score analyses, see the Online Supplementary Methods.

182

Results Selection and characteristics of the patients In all, 649 children with hemophilia A were included by 35 centers in the FranceCoag PUP cohort between 1994 and 2016. For this analysis, 120 ineligible patients were excluded (Figure 1). The 29 patients excluded due to insufficient data were first treated with nine well-balanced FVIII products. An inhibitor was diagnosed before ED 75 in five of them (first treated with five different FVIII products). Among the remaining 529 patients, 144 were first treated with plasma-derived FVIII products and 385 with recombinant products. We analyzed three product groups (Factane, Advate and Kogenate) which were sufficiently large to provide informative comparisons. Calendar period being a potential confounder, we excluded two patients first treated with Kogenate 2 years before its marketing authorization. Finally, 395 patients first treated with Factane, Advate and Kogenate (n=131, 137 and 127, respectively) between 2001 and 2016 were included in our analyses. Baseline characteristics and time-varying cofactors according to the FVIII product received are presented in Table 1. No significant association between product and cofactors was observed, except for calendar period of first exposure to FVIII (P<0.001) and F8 gene defect (P=0.009). Advate was marketed more than 3 years after Factane and Kogenate; initial treatment with Kogenate decreased during 20132016, probably due to consistent results published in 201310 and 201411,12 (Online Supplementary Table S1). Most PUPs with a not yet tested F8 gene defect (14 out of 18) were born in 2013 or after and were, therefore, treated with Factane or Advate. Without the undetermined F8 gene defect modality, no significant imbalance between product groups was observed (P=0.226).

Follow-up and exposure to factor VIII Overall 18,244 EDs were recorded during 559.5 personyears (Online Supplementary Table S2). Among the 274 patients without inhibitors, 194 (70.8%) were followed up until ED 75. Among the 80 patients with a censored follow-up before ED 75, two died, 24 switched to another FVIII product (see details in Online Supplementary Table S3) and 54 had not reached ED 75 at the last clinical visit. The contributions in EDs according to the FVIII product received and the studied time-varying risk factors are shown in Online Supplementary Table S4.

Inhibitor assay frequency and inhibitor incidence Altogether, 2,002 inhibitor assays were documented for the three groups of patients during the observation period. On average, these assays were performed every 6.0 EDs during the first 25 EDs and every 9.1 EDs during the overall follow-up period. The assay frequency was similar across the FVIII product groups (Online Supplementary Table S5). A clinically significant inhibitor was diagnosed in 121 patients after a median of 14 EDs (interquartile range, 8-20 EDs) and haematologica | 2018; 103(1)


Inhibitor incidence according to 3 FVIII brands

at a median age of 16.0 months (interquartile range, 12.024.0 months) (Table 2). Among them, 70 (57.9%) had hightiter inhibitors and 104 (86.0%) were subsequently treated with a bypassing agent and/or immune tolerance induction at some time during the entire FranceCoag follow-up. The global cumulative incidence at 75 EDs was 35.0% (95% CI: 30.2%-40.3%) for all inhibitors, 21.3% (95% CI: 17.2%26.2%) for high-titer inhibitors and 30.5% (95% CI: 25.9%35.7%) for subsequently treated inhibitors.

Inhibitor incidence according to factor VIII product Inhibitor incidence was significantly associated with the product received for all inhibitors (P<0.001), high-titer inhibitors (P=0.005) and treated inhibitors (P<0.001) (Figure 2A). For high-titer inhibitors, the cumulative incidence at 75 EDs was 12.7% (95% CI: 7.7-20.6) with Factane, 20.4% (95% CI: 14.0-29.1) with Advate, and 31.6% (95% CI: 23.541.7) with Kogenate (Table 3). Results were similar in weighted analyses using propensity scores (Figure 2B).

Table 1. Patients’ characteristics according to the factor VIII product received.

Fixed risk factors F8 gene defect — no. (%) Low risk High risk* Undetermined (e.g., not yet tested, unidentified) High-risk F8 gene defect known at first FVIII infusion Family history — n. (%) Hemophilia without inhibitor Hemophilia with inhibitor No family history of hemophilia Family history of hemophilia and inhibitor known at first FVIII infusion Ethnic origin — n. (%)† White only (both parents) Other, not African or Afro-American African or Afro-American (at least one grandparent) Calendar period of first exposure to FVIII — n. (%) 2001-2003 2004-2006 2007-2009 2010-2012 2013-2016 Age at first exposure to FVIII — n. (%) Less than 6 months 6-11 months At least 12 months

Factane (N = 131)

Advate (N = 137)

Kogenate (N = 127)

37 83 11 31

(28.2) (63.4) (8.4) (23.7)

33 97 7 40

(24.1) (70.8) (5.1) (29.2)

27 (21.3) 100 (78.7) 0 (0.0) 35 (27.6)

47 7 77 3

(35.9) (5.3) (58.8) (2.3)

57 13 67 9

(41.6) (9.5) (48.9) (6.6)

38 15 74 6

P 0.009

(29.9) (11.8) (58.3) (4.7)

91 (69.5) 27 (20.6) 13 (9.9)

100 (73.0) 25 (18.2) 12 (8.8)

93 (73.2) 29 (22.8) 5 (3.9)

21 16 37 18 39

0 35 36 31 35

34 22 35 31 5

0.579 0.128

0.243 0.372

<0.001 (16.0) (12.2) (28.2) (13.7) (29.8)

(0.0) (25.5) (26.3) (22.6) (25.5)

(26.8) (17.3) (27.6) (24.4) (3.9) 0.063

44 (33.6) 41 (31.3) 46 (35.1)

39 (28.5) 54 (39.4) 44 (32.1)

26 (20.5) 43 (33.9) 58 (45.7)

38 26 14 7 11

(29.0) (19.8) (10.7) (5.3) (8.4)

40 20 10 6 14

(29.2) (14.6) (7.3) (4.4) (10.2)

37 22 11 2 15

(29.1) (17.3) (8.7) (1.6) (11.8)

0.999 0.523 0.618 0.259 0.661

86 56 21 21 14

(65.6) (42.7) (16.0) (16.0) (10.7)

90 50 18 19 17

(65.7) (36.5) (13.1) (13.9) (12.4)

83 46 18 12 18

(65.4) (36.2) (14.2) (9.4) (14.2)

0.998 0.470 0.793 0.282 0.697 0.153§

Risk factors at first exposure (fixed risk factors) Peak treatment episode at first exposure — n. (%) At least 3 consecutive EDs At least 5 consecutive EDs At least 10 consecutive EDs First exposure linked to surgical procedure (with at least 3 EDs) — n. (%) First exposure linked to severe bleeding episode — n. (%)

Risk factors at any time (time-varying risk factors) History of peak treatment episodes (≥1 during follow-up) — n. (%) ≥3 consecutive EDs ≥5 consecutive EDs ≥10 consecutive EDs History of surgical procedures (with ≥3 EDs) during follow-up — n. (%) History of severe bleeding episodes during follow-up — n. (%) Initiation of regular prophylaxis‡ N. (% according to Kaplan-Meier estimator) at 25 EDs N. (% according to Kaplan-Meier estimator) at 50 EDs

51 (65.4) 73 (93.6)

63 (74.1) 80 (94.1)

40 (60.6) 61 (92.4)

*High-risk F8 gene defects include large deletions (at least 1 exon), intron 1 and 22 inversions, small deletions/insertions with stop codon (out of A-run) and nonsense mutations. †Up to four ethnic origins per patient could be recorded (one for each grandparent). ‡The initiation of regular prophylaxis was defined as the moment at which at least three consecutive prophylactic infusions of FVIII were given within a period of at least 15 days (RODIN definition).10 §P for global test using Cox proportional hazards model with exposure day as the observational time unit. FVIII: factor VIII; ED: exposure day.

haematologica | 2018; 103(1)

183


T. Calvez et al.

Interaction analyses Patients were split into two roughly equal groups according to calendar period of first exposure to FVIII. The immunogenicity differences across FVIII products were similar for both periods (Figure 3A) and the interactions between period and FVIII product were not statistically

significant (P≥0.444). Similarly, no significant interactions between peak treatment episode at first exposure and FVIII product were observed (Figure 3B).

Advate versus Factane The crude hazard ratio of Advate versus Factane was

Table 2. Characteristics of clinically significant inhibitors.

All inhibitors (N = 121) Median number of EDs at inhibitor detection (IQR) Median age at inhibitor detection (IQR) — months Median duration between ED 1 and inhibitor detection (IQR) — months Median of maximal inhibitor titer (IQR) — Bethesda units

14 16.0 5.3 8.0

Treatments received at any time during the FranceCoag follow-up Treatment with bypassing agents — n. (%) 88 Treatment with immune tolerance induction (ITI) — n. (%) 89 Treatment with bypassing agents and/or ITI — n. (%) 104

(8-20) (12.0-24.0) (1.9-10.1) (2.5-72.0)

(72.7) (73.6) (86.0)

High-titer inhibitors* (N = 70)

Low-titer inhibitors (N = 51)

14 (8-18) 13.4 (8.6-20.2) 3.0 (1.2-7.4) 46.0 (15.0-256.0)

17 (10-24) 19.3 (15.1-28.5) 7.3 (2.8-14.8) 2.0 (1.1-3.3)

64 62 70

24 27 34

(91.4) (88.6) (100.0)

(47.1) (52.9) (66.7)

* High-titer inhibitor defined as peak titer equal to or greater than 5 Bethesda units at any time during the FranceCoag follow-up. ED: exposure day; IQR: interquartile range

A

B

Figure 2. Kaplan-Meier representation of the cumulative incidence of inhibitors, with exposure day as the observational time unit, according to the factor VIII product received. Three outcomes are shown: all inhibitors, high-titer inhibitors and inhibitors subsequently treated with a bypassing agent and/or immune tolerance induction. (A) Kaplan-Meier estimates are shown for all patients. Tests used Cox proportional hazards model. (B) Weighted Kaplan-Meier estimates are shown for patients first treated between 2004 and 2012. This selection was made to avoid having patients with an extremely low probability of having received one of the two counterfactual treatments (see Online Supplementary Methods). Tests used weighted Cox proportional hazards model.

184

haematologica | 2018; 103(1)


Inhibitor incidence according to 3 FVIII brands

1.63 (95% CI: 0.84-3.17) for high-titer inhibitors (Figure 4A and Online Supplementary Table S7). Repeated analyses were adjusted first for each cofactor, then for all fixed and then for all time-varying cofactors, and showed comparable results. Finally, adjusted hazard ratios of Advate versus Factane were 1.64 (95% CI: 0.823.25) in the complete model (including all fixed and time-varying cofactors) and 1.54 (95% CI: 0.73-3.24) in the propensity score analysis using inverse probability of treatment weighting and adjustment for time-varying risk factors. Results were similar for the two sensitivity analyses and the two other outcomes: all inhibitors and treated inhibitors (Figure 4A, Online Supplementary Tables S6 and S8).

Kogenate versus Factane The crude hazard ratio for Kogenate versus Factane was 2.68 (95% CI: 1.43-5.00) for high-titer inhibitors (Figure 4B, Online Supplementary Table S7). Repeated analyses were adjusted first for each cofactor, then for all fixed and for all time-varying cofactors, with comparable results. The adjusted hazard ratios for Kogenate versus

Factane were 2.81 (95% CI: 1.44-5.49) in the complete model and 2.13 (95% CI: 1.02-4.46) in the propensity score analysis using inverse probability of treatment weighting and adjustment for time-varying risk factors. Results were also similar for the two sensitivity analyses and for the two other outcomes (Figure 4B, Online Supplementary Tables S6 and S8).

Discussion Since 1994, PUPs with severe hemophilia A have been prospectively followed up in France, notably to evaluate the immunogenicity of FVIII products. Whatever the outcome considered, adjusted results showed a highly significant difference (Pâ&#x2030;¤0.005) in the incidence of inhibitors among the groups receiving the three most used products between 2001 and 2016. To our knowledge, this study is the first to compare inhibitor incidence among large groups receiving single FVIII products, including a plasmaderived FVIII product. Firstly, we found a higher risk of inhibitor development, approximately 50%, in PUPs treat-

A

B

Figure 3. Kaplan-Meier representation of the cumulative incidence of inhibitors, with exposure day as the observational time unit, according to the factor VIII product received. Three outcomes are shown: all inhibitors, high-titer inhibitors and inhibitors subsequently treated with a bypassing agent and/or immune tolerance induction. Tests used Cox proportional hazards model. (A) Kaplan-Meier estimates according to calendar period of first exposure to factor VIII. (B) Kaplan-Meier estimates according to treatment intensity at first exposure (peak treatment episode â&#x2030;Ľ 3 consecutive exposure days).

haematologica | 2018; 103(1)

185


T. Calvez et al.

ed with Advate than in those treated with Factane. This result was stable across different models and sensitivity analyses, but not statistically significant for the three studied outcomes. However, the adjusted hazard ratio of

Advate versus Factane for high-titer inhibitors (1.64; 95% CI: 0.82-3.25) was similar to that in the SIPPET trial in which the adjusted hazard ratio for recombinant FVIII versus plasma-derived FVIII was 1.69 (95% CI: 0.96-2.98).32

A

B

Figure 4. Hazard ratios and 95% confidence intervals for (A) Advate versus Factane and (B) Kogenate versus Factane according to several models, two propensity score methods and two sensitivity analyses. Three outcomes are shown: all inhibitors, high-titer inhibitors, and inhibitors subsequently treated with a bypassing agent and/or immune tolerance induction. Except in PS analyses and in the first sensitivity analysis, 131, 137, and 127 PUPs first treated with Factane, Advate and Kogenate, respectively, were considered. In Panel A, 110 PUPs first treated from 2004 with Factane and 137 PUPs first treated with Advate were considered in PS analyses and in the first sensitivity analysis. In Panel B, 92 PUPs first treated until 2012 with Factane and 122 PUPs first treated until 2012 with Kogenate were considered in PS analyses; 99 PUPs first treated until 2013 with Factane and 127 PUPs first treated with Kogenate were considered in the first sensitivity analysis. EDs: Exposure days; HR: hazard ratio; PS: propensity score; PUP: previously untreated patient.

186

haematologica | 2018; 103(1)


Inhibitor incidence according to 3 FVIII brands

Hazard ratios were similar with propensity score analyses. Secondly, inhibitor incidence was at least twice as high in PUPs treated with Kogenate than in those treated with Factane. This result was stable across different analyses and highly significant for the three outcomes. Hazard ratios were slightly lower but still significant with propensity score analyses. This result was predictable considering the higher inhibitor incidence with Kogenate previously observed in our cohort,11 and in two other studies.10,12 We did not observe interactions between FVIII products and calendar periods or intensity of the initial FVIII treatment, unlike a recent study.9 However, our series was too small to study finer classifications. Interactions between other cofactors and FVIII products should be explored. The higher inhibitor incidence in PUPs treated with Kogenate than in those treated with Advate reported in three independent observational studies sparked a lively debate.13,14 The main alleged bias has been confounding by indication.15 Clinicians might have preferentially prescribed Kogenate to patients most at risk of inhibitors after two publications highlighting its low immunogenicity.33,34 We showed that the absence of unidirectional imbalance in known risk factors at first FVIII infusion between product groups does not support this hypothesis.16 Some interviewed French clinicians acknowledged their willingness to use various brands of FVIII products in their center. Consequently, at least two of the studied FVIII products were used in most centers (Online Supplementary Table S9). However, the determinants in choosing a FVIII product for a given patient could be diverse, involving the views of the physician and/or family and also depending on environmental conditions. As data on such determinants were not collected, they cannot be precisely understood. Thus, residual confounding related to unknown or unregistered risk factors remains possible. Since the 2000s, several studies have reported that plasma-derived products are less immunogenic in PUPs, especially those products with a high concentration of von Willebrand factor.4–6 A confounding by indication mechanism could therefore have led some French clinicians to treat most at-risk patients with Factane, inducing a higher incidence of inhibitors in this group. However, our results showed the opposite. Thus, confounding by indication related to known risk factors or subtle unrecorded patients’ characteristics cannot explain the observed lower inhibitor incidence with Factane compared with Advate and Kogenate. Moreover, our results changed only slightly after integrating propensity scores to counter possible confounding by indication. This was consistent with the lack of systematic imbalance in risk factors between the groups receiving the different products. We consider a center-related bias is more conceivable. The relative proportions of PUPs treated with each FVIII product differed among centers. If centers have their own effect on inhibitor incidence, regardless of FVIII product and other considered cofactors, residual confounding would be possible. We integrated the size of the centers in our multivariate analyses without observing substantial variations in the results. As finally selected PUPs were distributed in 32 hemophilia treatment centers (1 to 53 PUPs per center), we could not integrate centers individually in our multivariate analyses. However, this bias is unlikely to explain the concordant results observed in the three observational studies conducted in independent areas.10–12 Moreover, it could not have affected the SIPPET trial.18,19 haematologica | 2018; 103(1)

Table 3. Number of inhibitors and cumulative incidence at 75 exposure days according to the factor VIII product received.

Outcome

N.

Cumulative incidence at 75 exposure days % (95% CI)

All inhibitors Factane Advate Kogenate

25 36 60

22.5 31.6 50.1

(15.8-31.5) (23.9-41.1) (41.6-59.4)

High-titer inhibitors Factane Advate Kogenate

14 23 33

12.7 20.4 31.6

(7.7-20.6) (14.0-29.1) (23.5-41.7)

P*

<0.001

0.005

Inhibitors treated with bypassing agents and/or immune tolerance induction <0.001 Factane 19 17.4 (11.5-26.0) Advate 32 28.0 (20.7-37.2) Kogenate 53 45.5 (36.9-55.1) *P for the global test using Cox proportional hazards model with exposure day as the observational time unit.

Observational studies cannot prove a difference in immunogenicity between FVIII products. However, consistent results from several independent and rigorously conducted observational studies10,12 support the immunogenicity difference between Advate and Kogenate. Concerning plasma-derived products, the results of the randomized SIPPET trial and our study are also consistent, although SIPPET compared products according to their source whereas we compared three manufactured FVIII products (received by 47% of the SIPPET patients). Moreover, although SIPPET was mainly performed in developing countries (76% of patients were enrolled in India, Egypt, or Iran),32 while our study reflects real life in a Western country, both conjointly support the reality of a difference in immunogenicity between plasma-derived and recombinant products. Assuming an immunogenicity difference between two FVIII products as similar as Advate and Kogenate, an immunogenicity difference between all products is highly likely, particularly between products as different as Factane and recombinant products. Numerous genetic and non-genetic factors have been shown to be associated with the inhibitor incidence in PUPs.35,36 Depending on the combination of these factors, inhibitor probability may vary widely (10 to 90%).37,38 Given recent studies,10–12,18 the assumption that the nature of the product received affects the inhibitor incidence is increasingly credible. All FVIII products could potentially have their own immunogenicity levels, but extraordinary background noise generated by other cofactors could have prevented their identification until recently. The rarity of hemophilia A and the inadequacy of institutional support for comprehensive data collection in many countries are also responsible for this non-recognition. No convincing pathophysiological hypothesis has yet attempted to explain a possible higher incidence of inhibitors with Kogenate. Conversely, many recently reviewed hypotheses exist to explain a possible low inhibitor incidence with plasma-derived products.39,40 However, identifying predominant explanatory mechanisms is crucial to prove the 187


T. Calvez et al.

existence of this difference and to sustain the development of less immunogenic recombinant products. After the unexpected observation of an immunogenicity difference between Advate and Kogenate in PUPs, some authors stressed that the primacy of randomized trials and the 5% significance level (commonly required in drug efficacy assessment) were not relevant for considering drug adverse effects, a fortiori when alternative therapies exist.14 Thus, this result led to revised therapeutic recommendations.41 Oddly, although SIPPET has been the only randomized trial addressing an immunogenicity difference among FVIII products, yielding significant results, many experts advocated against broad changes in clinical practice.42–44 SIPPET and our results cannot be applied to patients with moderate/mild hemophilia A or severe hemophilia with over 50/100 EDs, when the inhibitor incidence is considerably lower (about 3 per 1000 person-years).45 However, when initiating replacement treatment in PUPs with severe hemophilia A, prescribers cannot ignore the risk/benefit of the different products, including imperfect knowledge of their

References 1. Peyvandi F, Garagiola I, Young G. The past and future of haemophilia: diagnosis, treatments, and its complications. Lancet. 2016;388(10040):187–197. 2. World Federation of Hemophilia. Report on the Annual Global Survey 2014 [Internet]. 2015 [cited Mar 17, 2016];Available from: http://www.wfh.org/en/data-collection 3. Di Minno MND, Di Minno G, Di Capua M, Cerbone AM, Coppola A. Cost of care of haemophilia with inhibitors. Haemophilia. 2010;16(1):e190-201. 4. Escuriola-Ettingshausen C, Kreuz W. Recombinant vs. plasma-derived products, especially those with intact VWF, regarding inhibitor development. Haemophilia. 2006;12(Suppl 6):102–106. 5. Goudemand J, Rothschild C, Demiguel V, et al. Influence of the type of factor VIII concentrate on the incidence of factor VIII inhibitors in previously untreated patients with severe hemophilia A. Blood. 2006;107(1):46–51. 6. Chalmers EA, Brown SA, Keeling D, et al. Early factor VIII exposure and subsequent inhibitor development in children with severe haemophilia A. Haemophilia. 2007;13(2):149–155. 7. Iorio A, Halimeh S, Holzhauer S, et al. Rate of inhibitor development in previously untreated hemophilia A patients treated with plasma-derived or recombinant factor VIII concentrates: a systematic review. J Thromb Haemost. 2010;8(6):1256–1265. 8. Franchini M, Tagliaferri A, Mengoli C, Cruciani M. Cumulative inhibitor incidence in previously untreated patients with severe hemophilia A treated with plasma-derived versus recombinant factor VIII concentrates: a critical systematic review. Crit Rev Oncol Hematol. 2012;81(1):82–93. 9. Marcucci M, Mancuso ME, Santagostino E, et al. Type and intensity of FVIII exposure on inhibitor development in PUPs with haemophilia A. A patient-level meta-analysis. Thromb Haemost. 2015;113(5):958–967.

188

immunogenicity. Moreover, at a time when several new recombinant FVIII products, including extended half-life ones, are entering the market with very little knowledge of their specific immunogenicity, it is crucial that all stakeholders organize accurate follow-up of all treated hemophilia A patients, and particularly PUPs with severe hemophilia A. Acknowledgments The authors thank Dominique Costagliola for critical review of the manuscript and helpful discussions. We also thank all the contributors from the FranceCoag Network, patients and their parents or legal guardians for their participation. TC thanks Jennifer Markovic for English editing assistance. The French hemophilia surveillance system ("Suivi thérapeutique National des Hémophiles" since 1994 and "FranceCoag" from 2003 onward) is fully supported by the public health authorities. Additional data collection for the first 75 exposure days was partially supported by the French National Clinical Research Program (2009) and by Assistance Publique Hôpitaux de Marseille.

10. Gouw SC, van der Bom JG, Ljung R, et al. Factor VIII products and inhibitor development in severe hemophilia A. N Engl J Med. 2013;368(3):231–239. 11. Calvez T, Chambost H, ClaeyssensDonadel S, et al. Recombinant factor VIII products and inhibitor development in previously untreated boys with severe hemophilia A. Blood. 2014;124(23):3398–3408. 12. Collins PW, Palmer BP, Chalmers EA, et al. Factor VIII brand and the incidence of factor VIII inhibitors in previously untreated UK children with severe hemophilia A, 20002011. Blood. 2014;124(23):3389–3397. 13. Kessler CM, Iorio A. The Rodin (Research Of Determinants of INhibitor Development among PUPs with haemophilia) study: the clinical conundrum from the perspective of haemophilia treaters. Haemophilia. 2013;19(3):351–354. 14. van der Bom JG, Gouw SC, Rosendaal FR. Second-generation recombinant factor VIII and inhibitor risk: interpretation of RODIN study findings and implications for patients with haemophilia A. Haemophilia. 2014;20(2):e171-174. 15. Berntorp E, Iorio A. Reflections on the FranceCoag report on inhibitory antibodies to factor VIII in patients with severe hemophilia A. Blood. 2015;125(24):3816–3817. 16. Calvez T, Chambost H, Lutz P, Rothschild C, Goudemand J. Response: confounding by indication is unlikely to explain the higher inhibitor incidence in boys treated with a recombinant FVIII product. Blood. 2015;125(24):3817–3819. 17. Mathew P, Dinter H, Church N, Humphries TJ, Kulkarni R. Inhibitors in haemophilia A: a perspective on clotting factor products as a potential contributing factor. Haemophilia. 2016;22(3):334–341. 18. Peyvandi F, Mannucci PM, Garagiola I, et al. A randomized trial of factor VIII and neutralizing antibodies in hemophilia A. N Engl J Med. 2016;374(21):2054–2064. 19. Peyvandi F, Mannucci PM, Palla R, Rosendaal FR. SIPPET: methodology, analysis and generalizability. Haemophilia. 2017;23(3):353-361.

20. European Medicines Agency. Consensus points - Workshop on Haemophilia Registries [Internet]. 2015 [cited Mar 29, 2016];Available from: http://www.ema. europa.eu/docs/en_GB/document_library/ Other/2015/08/WC500191378.pdf. 21. van den Berg HM, Peyvandi F. Assessment of clotting factor concentrates – pivotal studies and long-term requirements. Semin Thromb Hemost. 2015;41(8):855–859. 22. Carcao M, Re W, Ewenstein B. The role of previously untreated patient studies in understanding the development of FVIII inhibitors. Haemophilia. 2016;22(1):22–31. 23. Calvez T, Biou M, Costagliola D, et al. The French haemophilia cohort: rationale and organization of a long-term national pharmacosurveillance system. Haemophilia. 2001;7(1):82–88. 24. Stonebraker JS, Bolton-Maggs PHB, Soucie JM, Walker I, Brooker M. A study of variations in the reported haemophilia A prevalence around the world. Haemophilia. 2010;16(1):20–32. 25. Meijer P, Verbruggen B. The between-laboratory variation of factor VIII inhibitor testing: the experience of the external quality assessment program of the ECAT foundation. Semin Thromb Hemost. 2009;35(8): 786–793. 26. Chtourou S, Porte P, Nogré M, et al. A solvent/detergent-treated and 15-nm filtered factor VIII: a new safety standard for plasma-derived coagulation factor concentrates. Vox Sang. 2007;92(4):327–337. 27. Parti R, Schoppmann A, Lee H, Yang L. Stability of lyophilized and reconstituted plasma/albumin-free recombinant human factor VIII (ADVATE rAHF-PFM). Haemophilia. 2005;11(5):492–496. 28. Jiang R, Monroe T, McRogers R, Larson PJ. Manufacturing challenges in the commercial production of recombinant coagulation factor VIII. Haemophilia. 2002;8 (Suppl 2):1–5. 29. Austin PC. The relative ability of different propensity score methods to balance measured covariates between treated and untreated subjects in observational studies.

haematologica | 2018; 103(1)


Inhibitor incidence according to 3 FVIII brands

Med Decis Making. 2009;29(6):661–677. 30. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399–424. 31. Ali MS, Groenwold RHH, Belitser SV, et al. Reporting of covariate selection and balance assessment in propensity score analysis is suboptimal: a systematic review. J Clin Epidemiol. 2015;68(2):112–121. 32. Peyvandi F, Mannucci PM, Garagiola I, et al. Source of factor VIII replacement (plasmatic or recombinant) and incidence of inhibitory alloantibodies in previously untreated patients with severe hemophilia A: the multicenter randomized Sippet study [Internet]. 57th Annual Meeting & Exposition of the American Society of Hematology - Orlando. 2015 [cited Mar 16, 2016];Available from: https://ash.confex.com/ash/2015/webprogram/Paper82866.html 33. Giangrande PLF, KOGENATE Bayer Study Group. Safety and efficacy of KOGENATE Bayer in previously untreated patients (PUPs) and minimally treated patients (MTPs). Haemophilia. 2002;8(Suppl 2):19–22. 34. Kreuz W, Gill JC, Rothschild C, et al. Fulllength sucrose-formulated recombinant factor VIII for treatment of previously untreated or minimally treated young children with severe haemophilia A: results of an international clinical investigation. Thromb Haemost. 2005;93(3):457–467.

haematologica | 2018; 103(1)

35. Astermark J, Altisent C, Batorova A, et al. Non-genetic risk factors and the development of inhibitors in haemophilia: a comprehensive review and consensus report. Haemophilia. 2010;16(5):747–766. 36. Astermark J, Donfield SM, Gomperts ED, et al. The polygenic nature of inhibitors in hemophilia A: results from the Hemophilia Inhibitor Genetics Study (HIGS) combined cohort. Blood. 2013;121(8):1446–1454. 37. ter Avest PC, Fischer K, Mancuso ME, et al. Risk stratification for inhibitor development at first treatment for severe hemophilia A: a tool for clinical practice. J Thromb Haemost. 2008;6(12):2048–2054. 38. Hashemi SM, Fischer K, Moons KGM, van den Berg HM. Improved prediction of inhibitor development in previously untreated patients with severe haemophilia A. Haemophilia. 2015;21(2):227–233. 39. Oldenburg J, Lacroix-Desmazes S, Lillicrap D. Alloantibodies to therapeutic factor VIII in hemophilia A: the role of von Willebrand factor in regulating factor VIII immunogenicity. Haematologica. 2015;100(2):149–156. 40. Lai J, Hough C, Tarrant J, Lillicrap D. Biological considerations of plasma-derived and recombinant factor VIII immunogenicity. Blood. 2017;129(24):3147-3154. 41. World Federation of Hemophilia. Third publication suggests recombinant FVIII product associated with higher risk of inhibitor development in newly diagnosed, previous-

42.

43. 44.

45.

ly untreated patients with severe hemophilia A [Internet]. 2015 [cited Oct 11, 2016]; Available from: http://www.wfh.org/en/ our-work/treatment-safety/inhibitors-pupsupdate-nov2014 Medical and Scientific Advisory Council (MASAC). Recommendation #243 on SIPPET (Survey of Inhibitors in Plasma-ProductExposed Toddlers): results and recommendations for treatment products for previously untreated patients with hemophilia A [Internet]. [cited Dec 16, 2016]; Available from: https://www.hemophilia.org/ Researchers-Healthcare-Providers/Medicaland-Scientific-Advisory-CouncilMASAC/MASAC-Recommendations/ MASAC-Recommendation-On-SIPPETSurvey-of-Inhibitors-in-Plasma-ProductExposed-Toddlers. van den Berg HM, Pipe S, Ljung R. Plasma products do not solve the inhibitor problem. Haemophilia. 2017;23(3):346-347. Iorio A. Research and policy implications of a recently published controlled study in previously untreated haemophilia patients at high risk of inhibitor development. Haemophilia. 2017;23(3):350-352. Xi M, Makris M, Marcucci M, Santagostino E, Mannucci PM, Iorio A. Inhibitor development in previously treated hemophilia A patients: a systematic review, meta-analysis, and meta-regression. J Thromb Haemost. 2013;11(9):1655–1662.

189


Haematologica, Volume 103, issue 1  
Haematologica, Volume 103, issue 1