Haematologica, Volume 101, issue 9 by Haematologica

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

Editor-in-Chief Jan Cools (Leuven)

Deputy Editor Luca Malcovati (Pavia)

Managing Director Antonio Majocchi (Pavia)

Associate Editors Hélène Cavé (Paris), Ross Levine (New York), Claire Harrison (London), Pavan Reddy (Ann Arbor), Andreas Rosenwald (Wuerzburg), Juerg Schwaller (Basel), Monika Engelhardt (Freiburg), Wyndham Wilson (Bethesda), Paul Kyrle (Vienna), Paolo Ghia (Milan), 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)

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 (Rotterdam); 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 2016 are as following: Print edition

Institutional Euro 500

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: Tipografia PI-ME, via Vigentina 136, Pavia, Italy. Printed in August 2016.

haematologica calendar of events

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

EHA Scientific Conference on Bleeding Disorders Scientific Program Committee: C Balduini (Chair), A Falanga (Chair), F Rodeghiero, I Pabinger, M Makris September 14-17, 2016 Barcelona, Spain

2 International Conference on New Concepts in B-Cell Malignancies European School of Haematology (ESH) Chairs: M Hallek, L Staudt, S Stilgenbauer, A ThomasTikhonenko September 9-11, 2016 Estoril, Portugal nd

EBMT International Transplant Course European Group for Blood and Marrow Transplantation (EBMT) Chairs: M Mohty, J Kuball, R Duarte September 9-11, 2016 Barcelona, Spain

12th Edition of the Educational Course of the EBMT Lymphoma Working Party on Treatment of Malignant Lymphoma: State-ofthe Art and the Role of Stem Cell Transplantation European Group for Blood and Marrow Transplantation (EBMT) Chairs: S Montoto, P Dreger, A Sureda, E Vandenberghe September 21-23, 2016 Dublin, Ireland

Fifth International Symposium on Secondary Leukemia and Leukemogenesis Società Italiana di Ematologia (SIE) Chairs: F Lo Coco, L Pagano, MT Voso September 22-24, 2016 Rome, Italy

3rd International Conference on Multiple Myeloma European School of Haematology (ESH) Chairs: S Lonial, M Mohty, A Palumbo October 7-9, 2016 Milan, Italy

XIV Congresso Nazionale SIES - Societa' Italiana Di Ematologia Sperimentale Societa' Italiana Di Ematologia Sperimentale (SIES) Chair: M Massaia October 19-21, 2016 Rimini, Italy

Highlights of Past EHA - HOPE Dubai 2016 Chairs: R Foà, M Qari November 24-26, 2016 Dubai, UAE

EHA Scientific Meeting on Anemia Diagnosis and Treatment in the Omics Era Chair: A Iolascon February 2-4, 2017 Barcelona, Spain

EHA Hematology Tutorial on Lymphoid malignancies , Multiple myeloma and Bone Marrow Failure February 23-24, 2017 Colombo, Sri Lanka

EHA Hematology Tutorial on Lymphoid Malignancies March 17-18, 2017 Warsaw, Poland

EHA Scientific Meeting on Advances in Biology and Treatment of B Cell Malignancies with a Focus on Rare Lymphoma Subtypes Chairs: MJ Kersten and M Dreyling March 10-12, 2017 Barcelona, Spain

EHA Scientific Meeting on Aging and Hematology Chair: D Bron May 4-6, 2017 Location: TBC

22nd Congress of the European Hematology Association European Hematology Association June 22 - 25, 2017 Madrid, Spain

EHA Scientific Meeting on Challenges in the Diagnosis and Management of Myeloproliferative Neoplasms Chairs: JJ Kiladjian and C Harrison October 12-14, 2017 Location: TBC

EHA Scientific Meeting on Shaping the Future of Mesenchymal Stromal Cells Therapy Chair: W Fibbe November 23-25, 2017 Location: TBC

10th Hodgkin Symposium University hospital of Cologne Chairs: A Engert, B von Treskow, B Böll October 22-25, 2016 Cologne, Germany

2nd MEGMA Conference on Thalassaemia and Other Haemoglobinopathies Thalassaemia International Federation (TIF) Chairs: A Taher, J Porter, A Piga, A Beshlawy November 11-12, 2016 Amman, Jordan

Calendar of Events updated on August 1, 2016

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

Table of Contents Volume 101, Issue 9: September 2016 Cover Figure The link between mutations and targeted drugs - image accompanying the review article by Rosenquist et al. page 1002 (Image created by www.somersault1824.com)

Editorials 997

Circulating cell-free DNA in hematological malignancies Lieselot Buedts and Peter Vandenberghe

999

Hematopoietic stem cells meet induced pluripotent stem cells technology Maria Teresa Esposito

Review Articles 1002

Clinical impact of recurrently mutated genes on lymphoma diagnostics: state-of-the-art and beyond Richard Rosenquist, et al.

1010

The relevance of PTEN-AKT in relation to NOTCH1-directed treatment strategies in T-cell acute lymphoblastic leukemia Rui D. Mendes, et al.

Articles Red Cell Biology & Its Disorders

1018

Severe Ankyrin-R deficiency results in impaired surface retention and lysosomal degradation of RhAG in human erythroblasts Timothy J. Satchwell, et al.

1028

Altered heme-mediated modulation of dendritic cell function in sickle cell alloimmunization Emmanuelle Godefroy, et al.

Platelet Biology & Its Disorders

1039

Immune thrombocytopenia in adults: a prospective cohort study of clinical features and predictors of outcome Lamiae Grimaldi-Bensouda, et al.

Coagulation& Its Disorders

1046

Joint effects of cancer and variants in the factor 5 gene on the risk of venous thromboembolism Olga V. Gran, et al.

Myeloproliferative Disorders

1054

A novel role for nuclear factor-erythroid 2 in erythroid maturation by modulation of mitochondrial autophagy Monika Gothwal, et al.

Haematologica 2016; vol. 101 no. 9 - September 2016 http://www.haematologica.org/

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation 1065

Safety and efficacy of ruxolitinib in an open-label, multicenter, single-arm phase 3b expanded-access study in patients with myelofibrosis: a snapshot of 1144 patients in the JUMP trial Haifa Kathrin Al-Ali, et al.

Acute Myeloid Leukemia

1074

Clinical characteristics and prognosis of acute myeloid leukemia associated with DNA-methylation regulatory gene mutations Takeshi Ryotokuji, et al.

Acute Lymphoblastic Leukemia

1082

Characterization of leukemias with ETV6-ABL1 fusion Marketa Zaliova, et al.

Hodgkin Lymphoma

1094

Detection and prognostic value of recurrent exportin 1 mutations in tumor and cell-free circulating DNA of patients with classical Hodgkin lymphoma Vincent Camus, et al.

Plasma Cell Disorders

1102

Immunoparesis status in immunoglobulin light chain amyloidosis at diagnosis affects response and survival by regimen type Eli Muchtar, et al.

1110

Geriatric assessment in multiple myeloma patients: validation of the International Myeloma Working Group (IMWG) score and comparison with other commoncomorbidity scores Monika Engelhardt, et al.

Stem Cell Transplantation

1120

Letters to the Editor Letters are available online only at www.haematologica.org/content/101/9.toc

e356

Efficacy estimation of erythropoiesis-stimulating agents using erythropoietin-deficient anemic mice Norio Suzuki, et al. http://www.haematologica.org/content/101/9/e356

e361

Immunoglobulin heavy and light chain gene features are correlated with primary cold agglutinin disease onset and activity Agnieszka MaĹ&#x201A;ecka, et al. http://www.haematologica.org/content/101/9/e361

e365

Microparticle phenotypes are associated with driver mutations and distinct thrombotic risks in essential thrombocythemia AgnĂ¨s Charpentier, et al. http://www.haematologica.org/content/101/9/e365

e369

ATM mutations in major stereotyped subsets of chronic lymphocytic leukemia: enrichment in subset #2 is associated with markedly short telomeres Veronika Navrkalova, et al. http://www.haematologica.org/content/101/9/e369

Haematologica 2016; vol. 101 no. 9 - September 2016 http://www.haematologica.org/

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

e374

Chronic lymphocytic leukemia development is accelerated in mice with deficiency of the pro-apoptotic regulator NOXA Erik Slinger, et al. http://www.haematologica.org/content/101/9/e374

e378

Complementarity determining region-independent recognition of a superantigen by B-cell antigen receptors of mantle cell lymphoma Michael Fichtner, et al. http://www.haematologica.org/content/101/9/e378

e382

Significant functional difference between TNFAIP3 truncation and missense mutants Leire Escudero-Ibarz, et al. http://www.haematologica.org/content/101/9/e382

e385

Prolonged disease-free survival in elderly relapsed diffuse large B-cell lymphoma patients treated with lenalidomide plus rituximab Pier Luigi Zinzani, et al. http://www.haematologica.org/content/101/9/e385

e387

Whole exome sequencing reveals activating JAK1 and STAT3 mutations in breast implant-associated anaplastic large cell lymphoma anaplastic large cell lymphoma Piers Blombery, et al. http://www.haematologica.org/content/101/9/e387

Comments e391

Aplastic anemia: the correct nomenclature matters Branko Cuglievan, et al. http://www.haematologica.org/content/101/9/e391

Haematologica 2016; vol. 101 no. 9 - September 2016 http://www.haematologica.org/

EDITORIALS Circulating cell-free DNA in hematological malignancies Lieselot Buedts1 and Peter Vandenberghe2 1

Center for Human Genetics; and 2Department of Hematology, University Hospitals KU Leuven, Belgium

E-mail: peter.vandenberghe@uzleuven.be doi:10.3324/haematol.2015.131128

n 1948, Mandel and Métais detected nucleic acids in peripheral blood plasma, thereby invalidating the previous dogma claiming that DNA is always cell-bound.1 However, long before the double helix structure of DNA was discovered, their observation fell into oblivion for many decades. In the nineties, the interest in circulating cell-free DNA (ccfDNA) resurged, as ccfDNA was shown to contain DNA derived from the fetus, or more precisely the placenta, in pregnant women, and from cancer cells in cancer patients.2,3 The majority of ccfDNA is thought to originate from apoptosis, as evidenced by the predominant fragment length of (multiple integers of) 160 to 180 bp, typical of DNA released from apoptotic cells. However, the presence of larger fragments, up to 21 kb, suggests that ccfDNA can also be derived from necrotic cells.4 Moreover, many studies attribute a notable fraction of ccfDNA to an active secretion mechanism, releasing newly synthesized nucleic acids associated with lipid-protein complexes.5 Since these processes occur ubiquitously in the body, the blood plasma of healthy individuals by default contains ccfDNA at steady-state levels and primarily derived from hematopoietic cells.6 In cancer patients, ccfDNA also contains, besides ccfDNA derived from nonmalignant cells, a fraction of circulating tumor DNA (ctDNA) as tumor cells also shed DNA into the circulation. This creates an opportunity for “remote sampling” of the tumoral DNA, eloquently epitomized as the “liquid biopsy”. As such, analysis of ccfDNA may potentially provide insights into somatic genomic mutations of the patient’s tumor with potential diagnostic, prognostic and therapeutic value. In addition, the relative amount of ctDNA can be informative on the tumoral burden and its evolution. Generally, ctDNA constitutes only a minor fraction (less than 1%) of the total ccfDNA, although the actual fractions can range from 0.01% to more than 90% depending on the tumor size, location and vascularity, and the disease extent and stage.7 Given the high amount of background ccfDNA and the low amount of input material, molecular methods with a low detection threshold are required for the detection of rare tumor variants. For metastatic tumors such methods have been validated.8,9 In the context of early-stage malignancy or minimal residual disease, approaches with lower detection thresholds are required, which are capable of identifying mutations in limited amounts of ctDNA.8,10,11 For solid tumors, several studies have validated that somatic copy number variations and somatic mutations detected in ctDNA match with those present in the primary tumor. For instance, KRAS mutations have been detected in the ccfDNA of colorectal cancer patients with a concordance of 96% with the mutation status in tumor tissue samples,12 and HER2 amplifications have been identified in the ccfDNA of 64% of cases with HER2-amplified breast cancers.13 In this way, ccfDNA can facilitate diagnostic and prognostic cancer assessment by noninvasively retrieving the tumor’s somatic mutations. Furthermore, while conventional biopsies comhaematologica | 2016; 101(9)

prise only one lesion or part of one lesion, ctDNA DNA is derived from the entire primary lesion as well as from metastatic lesions, as they all shed ctDNA. Thus, the ctDNA is in principle representative of the whole tumor burden within the patient, and less subject to sampling bias. This is of pivotal importance for solid tumors as molecular tumor heterogeneity is known to exist between the primary and metastatic sites as well as within tumoral lesions.4 A second potential application is the follow-up of disease under therapy: somatic mutations and structural chromosomal alterations first identified in primary tumors, and confirmed in ctDNA, can subsequently be targeted in ctDNA in a quantitative matter, to estimate the evolution of the tumor burden under therapy. In this context, ccfDNA sampling offers a cheaper and less interventional approach in comparison with PET/CT scans; this may enable frequent sampling so that patients can be monitored more closely or dynamically.14 Another promising application is monitoring the emergence of resistance mutations. Diaz et al. spotted resistance KRAS mutations in the ctDNA of colorectal cancer patients 5 months before radiological evidence for tumor progression.15 Finally, the analysis of ccfDNA in the context of noninvasive prenatal testing has led to a low rate of incidental diagnoses of cancer.16,17 As such, the potential clinical applications of ccfDNA in a context of malignant diseases may, in principle, encompass almost the entire course of cancer from early detection to diagnosis, therapeutic monitoring and monitoring of relapse. Hematological malignancies generally are “liquid” malignancies, originating in the bone marrow or in the lymphoid system, and circulating in the peripheral blood, in leukemic numbers or in lower numbers detectable only by flow cytometry or molecular techniques. As such it is not surprising that hematologists have, by intuition, taken the straightforward route of examining cellular fractions from blood or bone marrow for cytological, immunophenotypic or genetic studies at diagnosis or during follow-up.4 Nevertheless, the fact that BCL-2/IgH PCR status at the end of induction treatment is not predictive for progression-free survival in relapsed/resistant follicular lymphoma clearly indicates that sampling the cellular fraction of blood has limitations.18 Indeed, more recent findings indicate that ccfDNA may hold several advantages for hematological disorders that were unanticipated until recently. In a technical prospective proof-of-principle study, we recently described that low-pass sequencing of ccfDNA from 10 consecutive patients with newly diagnosed Hodgkin lymphoma (HL) revealed copy number changes present in Hodgkin/Reed-Sternberg (HRS) cells in the majority of cases, including early as well as late stage Hodgkin lymphoma. The initial case of the above series of ten came to our attention as an incidental finding revealed by noninvasive prenatal testing. Among the initial 4000 noninvasive prenatal tests (NIPT), we identified three cases with genomic profile aberrations corresponding to tumor-related copy number variations. Further clinical evaluation revealed a maternal ovarian carci997

Editorials

noma, Hodgkin lymphoma and follicular lymphoma.16 Given that the finding of Hodgkin lymphoma was especially surprising, we initiated a pilot study in which nine additional patients with nodular sclerosis Hodgkin lymphoma at diagnosis and without concomitant malignancies, were recruited.19 Hodgkin/Reed-Sternberg cells, the malignant hallmark cells in classical Hodgkin lymphoma are embedded in a dense cellular infiltrate of B and T cells, macrophages, neutrophils and other cell types, and generally represent only 0.1-2% of the tumor mass. The low abundance of HRS cells in Hodgkin lymphoma tissue represents the major technical obstacle for the exploration of their genome. This led researchers to study HRS-derived cell lines, or HRS cells purified by laser microdissection20 or by flowsorting21 from primary lymph node tissues. Increased concentrations of ccfDNA had been observed in classical Hodgkin lymphoma patients before, although it was not formally proven that this ccfDNA contained DNA from HRS cells, along with ccfDNA from non-malignant cells.22 Our findings therefore revealed a potential novel gateway for the exploration of the genetic landscape of Hodgkin lymphoma on a wider scale than was previously possible. Exploitation of this gateway on a wider scale in the context of clinical studies may facilitate the recognition and identification of genetic subgroups of HL, and hence the development of personalized therapies. Analysis of ccfDNA may also hold promise for other hematological malignancies where the low malignant cell fraction in disease tissues may challenge their genetic characterization, e.g. in T-cell/histiocyte-rich B-cell lymphoma, or multiple myeloma, or in any situation where primary lesions are difficult to approach for technical reasons. In our study, copy number aberrations were detected in 9 out of 10 cases, two of whom had stage IVB and seven of whom had stage IIA disease. Interestingly, the aberrations were most pronounced in the two cases with extensive disease. Moreover, it was found that the genomic representation profiles normalized between day 15 and 28 of the first chemotherapy course. Both observations strongly suggest that the fraction of ctDNA in circulating cell-free DNA is related to the malignant disease burden in HL, and this is now the subject of our further investigations. In this respect, the half-life of ctDNA is known to be short, up to less than an hour:4 this may turn out to be a very attractive feature, as it might allow for the dynamic monitoring of tumor burden in time spans of hours to days instead of weeks to months. This early window by far antedates the decline of protein biomarkers, such as paraproteins or other tumor markers as well as the tumor response as evaluated by conventional and even metabolic imaging. Though the exact clearance mechanism of ccfDNA remains poorly understood, plasma nucleases are probably involved in the degradation of ccfDNA, complemented by lymphatic clearance and transrenal excretion.23 That ctDNA analysis holds great promise for the followup of the malignant burden in non-Hodgkin lymphoma (NHL), was excellently demonstrated by a retrospective analysis by Roschewski and colleagues of ctDNA in diffuse large B-cell lymphoma (DLBCL), one of the most common types of non-Hodgkin lymphoma. Because of the considerable risk (40%) of relapse and treatment failure in the first five years after diagnosis, patients are usu998

ally followed up to 5 years after first-line treatment using PET/CT scans.24 Indeed, while detecting recurrence at the subclinical stage could be beneficial as tumor burden is still likely to be low, the potential benefits of PET monitoring are counterbalanced by additional health risks for patients due to ionizing radiation exposure, increased medical costs and false positive results.14 Roschewski and colleagues hypothesized that massive parallel sequencing of VDJ immunoglobulin gene sequences in ccfDNA could serve as a more convenient and more sensitive alternative for monitoring the tumor burden. Mature B cells have uniquely rearranged immunoglobulin gene loci. Roschewski et al. identified the tumor-specific VDJ rearrangement, the ‘clonotype’, in pretreatment lymph node biopsies or cffDNA in serum samples, and measured the tumor clonotype in cffDNA during and up to 5 years after first-line treatment to quantify the fraction of ctDNA. Detectable interim circulating tumor DNA had a positive predictive value of 62·5% (95% CI 40·6–81·2) and a negative predictive value of 79·8% (69·6–87·8), respectively. Surveillance monitoring of circulating tumor DNA was carried out in 107 patients who achieved complete remission. This analysis showed that the hazard ratio for clinical disease progression was 228 (95% CI 51–1022) for patients who developed detectable circulating tumor DNA during surveillance compared with patients with undetectable circulating tumor DNA. Surveillance circulating tumor DNA had a positive predictive value of 88·2% (95% CI 63·6–98·5) and a negative predictive value of 97·8% (92·2–99·7), respectively, and identified the risk of recurrence at a median of 3·5 months (range 0–200) before evidence of clinical disease. The authors concluded that interim ctDNA is a promising biomarker for identifying patients at high risk of treatment failure. In addition, surveillance of ctDNA in remission identified patients at risk of recurrence before clinical evidence of disease relapse.25 This relatively new area of research still faces important technical challenges. The standardization of ccfDNA analysis is required, as is the elucidation of the unknowns, such as inter-individual and diurnal variation, baseline levels, and the effect of inflammatory responses or physical trauma on ccfDNA levels.4 It is equally unclear to what extent physical barriers, such as the blood-brain barrier or the blood-testis barrier, impair the release of ctDNA from tumors in these sanctuary sites.9 Moreover, the low fraction of ctDNA in early stage disease and the poor extraction efficiency require a considerable sequencing sensitivity for oncogenic mutations to be reliably distinguished from sequencing errors.10 For targeted detection of known mutations, several powerful methods are at hand based on digital PCR, assessing each ccfDNA molecule individually in emulsion droplets or microfluidic devices to reduce the background noise of normal ccfDNA.26 Finally, for unbiased mutation discovery, the error rate of the current NGS platforms, of around 0.1%, is problematic.27 Alternatively, several methods have been developed to improve template preparation or data interpretation, either by using barcodes allowing PCR- and sequencing-induced mutations to be filtered out after sequencing (TAm-Seq,8 SAFESeqS,10 Duplex Sequencing11), or by enriching tumor-specific mutations prior to library preparation (COLDhaematologica | 2016; 101(9)

Editorials

PCR,28 Nuclease-Assisted Mutation Enrichment29). The last decades have witnessed significant progress in elucidating the origin, characteristics and potential applications for the analysis of ccfDNA. Given its ability to assess dynamic and comprehensive tumor processes, ccfDNA, as a source for ctDNA, holds great promise for the implementation of personalized and dynamic cancer therapies. In addition, the low detection limit and ease of sampling may critically improve post-treatment monitoring compared to surveillance imaging. Resolving the remaining technical and practical challenges will allow ccfDNA to prove its clinical utility and revise cancer patient care.

References 1. Mandel P, Metais P. Les acides nuclĂŠiques du plasma sanguin chez l'homme. C R Seances Soc Biol Fil. 1948;142(3-4):241-243. 2. Vasioukhin V, Anker P, Maurice P, Lyautey J, Lederrey C, Stroun M. Point mutations of the N-ras gene in the blood plasma DNA of patients with myelodysplastic syndrome or acute myelogenous leukaemia. Br J Haematolo. 1994;86(4):774-779. 3. Lo D, Corbetta N, Chamberlain P, et al. Presence of fetal DNA in maternal plasma and serum. Lancet. 1997;350(9076):485-487. 4. Schwarzenbach H, Hoon DSB, Pantel K. Cell-free nucleic acids as biomarkers in cancer patients. Nat Rev Cancer. 2011;11(6):426-437. 5. Bronkhorst AJ, Wentzel JF, Aucamp J, Van Dyk E, Du Plessis L, Pretorius PJ. Characterization of the cell-free DNA released by cultured cancer cells. Biochim Biophys Acta. 2016;1863(1):157-165. 6. Lui Y, Chik K, Chiu R, Ho C, Lam C, Lo Y. Predominant hematopoietic origin of cell-free DNA in plasma and serum after sex-mismatched bone marrow transplantation. Clin Chem. 2002;48(3):421-427. 7. Diehl F, Schmidt K, Choti MA, et al. Circulating mutant DNA to assess tumor dynamics. Nat Med. 2008;14(9):985-990. 8. Forshew T, Murtaza, M. Noninvasive identification and monitoring of cancer mutations by targeted deep sequencing of plasma DNA. Sci Transl Med. 2012;4(136):136ra68. 9. Bettegowda C, Sausen M, Leary R, Kinde I, Wang Y, Agrawal N. Detection of circulating tumor DNA in early-and-late stage human malignancies. Sci Transl Med. 2014;6(224):224ra24. 10. Kinde I, Wu J, Papadopoulos N, Kinzler K, Vogelstein B. Detection and quantification of rare mtations with massively parallel sequencing. Proc Natl Acad Sci U S A. 2011;108(23):9530-9535. 11. Kennedy SR, Schmitt MW, Fo EJ, et al. Detecting ultralow-frequency mutations by Duplex Sequencing. Nat Protoc. 2014;9(11):2586-2606. 12. Thierry AR, Mouliere F, El Messaoudi S, et al. Clinical validation of the detection of KRAS and BRAF mutations from circulating tumor DNA. Nat Med. 2014;20(4):430-435. 13. Gevensleben H, Garcia-Murillas I, Graeser MK, et al. Noninvasice detection of HER2 amplification with plasma DNA digital PCR. Clin Cancer Res. 2013;19(12):3276-3284.

14. Abel GA. Does surveillance imaging after treatment for diffuse large Bcell lymphoma really work. J Clin Oncol. 2015;33(13):1427-1429. 15. Diaz L, Williams R, Wu J, et al. The molecular evolution of acquired resistance to targeted EGFR blockade in colorectal cancers. Nature. 2012;486(7404):537-540. 16. Amant F, Verheecke M, Wlodarska I, et al. Presymptomatic identification of cancers in pregnant women during noninvasive prenatal testing. JAMA Oncol. 2015;1(6):814-819. 17. Bianchi DW, Chudova D, Sehnert AJ, et al. Noninvasive prenatal testing and incidental detection of occult maternal malignancies. J Am Med Association. 2015;314(2):162-169. 18. Van Oers M, TĂśnissen E, Van Glabbeke M, et al. BCL-2/IgH polymerase chain reaction status at the end of induction treatment is not predictive for progression-free survival in relapsed/resistant folicular lymphoma: results of a prospective randomized EORTC 20981 phase III intergroup study. J Clin Oncol. 2010;28(13):2246 - 2252. 19. Vandenberghe P, Wlodarska I, Tousseyn T, et al. Non-invasive detection of genomic imbalances in Hodgkin/Reed-Sternberg cells in early and advanced stage Hodgkin's lymphoma by sequencing of circulating cell-free DNA: a technicalproof-of-principle study. Lancet Haematol. 2015;2(2):e55-65. 20. Steidl C, Diepstra A, Lee T, Chan F. Gene expression profiling of microdissected Hodgkin Reed-Sternberg cells correlates with treatment outcome in classical Hodgkin lymphoma. Blood. 2012;120(17):35303540. 21. Reichel J, Chadburn A. Flow-sorting and exome sequencing reveals the oncogenome of primary Hodgkin and Reed Sternberg cells. Blood. 2014;125(7):1061-1072. 22. Hohaus S, Giachela M, Massini G, et al. Cell-free circulating DNA in Hodgkin's and non-Hodgin's lymphomas. Ann Oncol. 2009;20(8):1408-1413. 23. Yu S, Lee S, Jiang P, et al. High-resolution profiling of fetal DNA clearance from maternal plasma by massive parallel sequencing. Clin Chem. 2013;59(8):1228-1237. 24. Liedtke M, Hamlin P, Moskowitz C, Zelenetz A. Surveillance imaging during remission identifies a group of patients with more favorable aggresive NHL at time of relapse: a retrospective analysis of a uniformly-treated patient population. Ann Oncol. 2006;17(6):909-913. 25. Roschewski M, Dunleavy K, Pittaluga S, Moorhead M, Pepin F. Circulating tumour DNA and CT monitoring in patients with untreated diffuse large B-cell lymphoma: a correlative biomarker study. Lancet Oncol. 2015;16(5):541-549. 26. Li M, Diehl F, Dressman D, Vogelstein B, Kinzler K. BEAMing up for detection and quantification of rare sequence variants. Nat Methods. 2006;3(2):95-97. 27. Glenn T. Field guide to next-generation DNA sequencers. Mol Ecol Resour. 2011;11(5):759-769. 28. Li J, Wang L, Mamon H, Kulke MH, Berbeco R, Makrogiorgos M. Replacing PCR with COLD-PCR enriches variant DNA sequences and redefines the sensitivity of genetic testing. Nat Med. 2008;14(5):579584. 29. Makrigiorgos M. Novel methods for enrichment of mutations and differentially methylated sequences from liquid biopsy genomes. Molecular Diagnostics Europe, Circulating Cell-Free DNA. Lisbon, 2016.

Hematopoietic stem cells meet induced pluripotent stem cells technology Maria Teresa Esposito* School of Health, Sport and Bioscience, University of East London, Water Lane London E15 4LZ, UK. E-mail: m.esposito@uel.ac.uk doi:10.3324/haematol.2016.150755

ematopoietic stem cells (HSCs), the source of the entire blood cells repertoire, represent the first stem cells identified in an adult tissue, the bone marrow, and the first stem cells used as a therapy in humans, through bone marrow transplantation.1 Although this therapeutic approach is well established and used to treat a variety of hematological conditions, including leukemia, several aspects limit its application. A lack of matched donors poses a serious barrier, especially for haematologica | 2016; 101(9)

patients from ethnic minority backgrounds. Finding the perfect match between donor and recipient, and obtaining a large number of HSCs, represent two of the unmet major clinical challenges. Development in a parallel area of stem cells research, induced pluripotent stem cells (iPS) technology, might provide new avenues to circumvent the limitations posed by the scarce number of HSCs available for transplantation. A decade ago the team led by Nobel prize winner Shinya 999

Editorials

Figure 1. Schematic representation of Red Blood Cells (RBCs) generation from induced pluripotent stem cells (iPS). Fibroblasts are isolated from the skin of the patient and reprogrammed in vitro into iPS. iPS can be differentiated in vitro in RBCs and be used for blood transfusion. Patient-specific iPS can be powerful tools for disease modelling or drug discovery processes.

Yamanaka authored a breakthrough study describing the generation of pluripotent stem cells from adult cells.2,3 Since then, many scientists have tried to develop hematopoietic stem cells and blood cells from adult cells via iPS. Szabo et al. showed that expression of the reprogramming gene and transcription factor OCT4 could induce the generation of HSCs from adult fibroblasts.4 With an analogous approach, Pereira et al. reprogrammed mouse fibroblasts into HSCs by introducing a set of transcription factors, including Gata2, Gfi 1b, Fos and Etv6.5 One of the major obstacles to the reprogramming is the epigenetic code that sets a strong barrier between the differentiation potential of an adult cell and a stem cell. Based on this consideration, Doulatov et al. decided to attempt the reprogramming of human committed myeloid cells into HSCs, identifying 13 essential transcription factors.6 Although these studies showed that the reprogramming of a variety of cells into HSCs is achievable, none of them were able to achieve long-term peripheral blood reconstitution in vivo and to establish the full repertoire of HSCs through serial transplantations. These are hallmarks of the ability of the cells to differentiate and self-renew and represent the gold standards for stem cells. In a more recent study, Riddell and colleagues reprogrammed differentiated somatic blood cells, pro-B cells, by using a cocktail of transcription factors, namely Hlf, Lmo2, Pbx1, Prdm5, Runx1t1, and Zfp37.7 Of course the cocktail used by Riddell and colleagues includes potent leukemic oncogenes that limit the immediate clinical use of HSCs generated by this application. Non-integrating methods and controllable expression systems of the transcription factors are currently under investigation for the clinical translation of these products. 1000

The first clinical trial with iPS derived cells started over two years ago in Japan, where a patient affected by macular degeneration of the retina was injected with iPS derived cells.8 Although the treatment proved safe and effective, last September the trial was halted because of mutations found in the cells prepared for a second patient, and due to new laws regulating regenerative medicine products in Japan.9 However the scientific community strongly believes that the iPS technology holds better promise than embryonic stem cells (ESC) for clinical translation. Currently there is no open source of human ESC (hESC) in the UK10 and hESC products are not patentable in Europe;11,12 moreover, hESC research has been restricted due to ethical consideration regarding the use of human embryos.13 Furthermore, no good manufacturing practices (GMPs) grade hESC lines of O Rh negative blood type are available, a critical aspect in order to develop widely transfusable red blood cells. Novosang is a project born from a consortium between British academic groups and blood services planning to launch the first human clinical trial using red blood cells produced by iPS in 2017.14 Stem cell research is a very hot topic that has led to hope and hype, with exaggerated predictions and claims regarding a potential clinical translation of these products. Considering the steep progress made in iPS technology in less than ten years we can optimistically expect progress for clinically proofed induced hematopoietic stem cells (iHSCs). Moreover, iHSCs hold the promise of becoming potent tools for modelling diseases in vitro and for drug discovery (Figure 1). These are two particularly important aspects in the hematology field, where the lack of appropriate humanized models and ethical concerns over the use of xenotransplantation and in vivo toxicity studies limit both the understanding of the biology of the disease and haematologica | 2016; 101(9)

Editorials

the screening of new drugs. The marriage between the HSC and iPS technology therefore represents a very promising and interesting area of stem cell research.

References 1. Till JE, Mc CE. A direct measurement of the radiation sensitivity of normal mouse bone marrow cells. Radiat Res. 1961;14:213-222. 2. Takahashi K, Yamanaka S. Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell. 2006;126(4):663-676. 3. Takahashi K, Tanabe K, Ohnuki M, et al. Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell. 2007;131(5):861-872. 4. Szabo E, Rampalli S, Risueno RM, et al. Direct conversion of human fibroblasts to multilineage blood progenitors. Nature. 2010;468 (7323):521-526. 5. Pereira CF, Chang B, Qiu J, et al. Induction of a hemogenic program in mouse fibroblasts. Cell Stem Cell. 2013;13(2):205-218.

haematologica | 2016; 101(9)

6. Doulatov S, Vo LT, Chou SS, et al. Induction of multipotential hematopoietic progenitors from human pluripotent stem cells via respecification of lineage-restricted precursors. Cell Stem Cell. 2013;13(4):459-470. 7. 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):549-564. 8. Reardon S, Cyranoski D. Japan stem-cell trial stirs envy. Nature. 2014;513(7518):287-288. 9. Garber K. RIKEN suspends first clinical trial involving induced pluripotent stem cells. Nat Biotechnol. 2015;33(9):890-891. 10. Courtney A, de Sousa P, George C, Laurie G, Tait J. Balancing open source stem cell science with commercialization. Nat Biotechnol. 2011;29(2):115-116. 11. Moran N. European court bans embryonic stem cell patents. Nat Biotechnol. 2011;29(12):1057-1059. 12. Moran N. Brustle patent holds up in Germany. Nat Biotechnol. 2013;31(2):94. 13. Hyun I. Policy: Regulate embryos made for research. Nature. 2014;509(7498):27-28. 14. Mittra J, Tait J, Mastroeni M, Turner ML, Mountford JC, Bruce K. Identifying viable regulatory and innovation pathways for regenerative medicine: a case study of cultured red blood cells. N Biotechnol. 2015;32(1):180-190.

1001

REVIEW ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2016 Volume 101(9):1002-1009

Clinical impact of recurrently mutated genes on lymphoma diagnostics: state-of-the-art and beyond

Richard Rosenquist,1 Andreas Rosenwald,2 Ming-Qing Du,3 Gianluca Gaidano,4 Patricia Groenen,5 Andrew Wotherspoon,6 Paolo Ghia,7 Philippe Gaulard,8 Elias Campo,9 Kostas Stamatopoulos,10 on behalf of the European Research Initiative on CLL (ERIC) and the European Association for Haematopathology (EAHP)

Department of Immunology, Genetics and Pathology, Science for Life Laboratory, Uppsala University, Sweden; 2Institute of Pathology, University of Würzburg, Germany and Comprehensive Cancer Center Mainfranken (CCC MF), Germany; 3Division of Molecular Histopathology, Department of Pathology, University of Cambridge, UK; 4 Division of Haematology, Department of Translational Medicine, University of Eastern Piedmont, Novara, Italy; 5Department of Pathology, Radboud University Medical Center, Nijmegen, The Netherlands; 6Department of Histopathology, Royal Marsden Hopsital, Fulham Road, London, UK; 7Division of Experimental Oncology and Department of OncoHematology, Università Vita-Salute San Raffaele and IRCCS Instituto Scientifico San Raffaele, Milan, Italy; 8Department of Pathology, AP-HP, Groupe hospitalier Henri Mondor-Albert Chenevier, Créteil INSERM U955, Université Paris-Est, Créteil, France; 9 Hemathopatology Section, Department of Pathology, Hospital Clinic and Institut d'Investigacions Biomediques August Pi i Sunyer (IDIBAPS), University of Barcelona, Spain; and 10Institute of Applied Biosciences, CERTH, Thessaloniki, Greece. 1

ABSTRACT

Correspondence: richard.rosenquist@igp.uu.se

Received: February 24, 2016 Accepted: May 3, 2016. Pre-published: no prepublication doi:10.3324/haematol.2015.134510

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1002

©2016 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights reserved to the Ferrata Storti Foundation. Copies of articles are allowed for personal or internal use. Permission in writing from the publisher is required for any other use.

1002

imilar to the inherent clinical heterogeneity of most, if not all, lymphoma entities, the genetic landscape of these tumors is markedly complex in the majority of cases, with a rapidly growing list of recurrently mutated genes discovered in recent years by next-generation sequencing technology. Whilst a few genes have been implied to have diagnostic, prognostic and even predictive impact, most gene mutations still require rigorous validation in larger, preferably prospective patient series, to scrutinize their potential role in lymphoma diagnostics and patient management. In selected entities, a predominantly mutated gene is identified in almost all cases (e.g. Waldenström’s macroglobulinemia/ lymphoplasmacytic lymphoma and hairy-cell leukemia), while for the vast majority of lymphomas a quite diverse mutation pattern is observed, with a limited number of frequently mutated genes followed by a seemingly endless tail of genes with mutations at a low frequency. Herein, the European Expert Group on NGS-based Diagnostics in Lymphomas (EGNL) summarizes the current status of this ever-evolving field, and, based on the present evidence level, segregates mutations into the following categories: i) immediate impact on treatment decisions, ii) diagnostic impact, iii) prognostic impact, iv) potential clinical impact in the near future, or v) should only be considered for research purposes. In the coming years, coordinated efforts aiming to apply targeted next-generation sequencing in large patient series will be needed in order to elucidate if a particular gene mutation will have an immediate impact on the lymphoma classification, and ultimately aid clinical decision making.

Introduction Most, if not all, lymphoma subtypes display considerable heterogeneity in their clinical and pathological characteristics, intricately linked to the underlying biological heterogeneity.1 Indeed, the genomic landscape of lymphomas is remarkably diverse, although an increasing number of ‘shared’ genetic lesions have recently emerged, affecting similar mechanisms and processes that in certain instances are haematologica | 2016; 101(9)

NGS-based diagnostics in lymphomas

important for ‘generic’ cell homeostasis (e.g. DNA repair) while in others they are ‘lymphocyte-specific’ (e.g. antigen receptor signaling) (Figure 1). Thanks to next-generation sequencing (NGS), it has become possible to appreciate the panorama of recurrently affected genes that contribute to disease pathogenesis and/or evolution, at least in major lymphoma subtypes. Mounting evidence suggests that certain gene mutations have diagnostic, prognostic and/or predictive impact. However, for most mutations, functional in vitro validation and confirmation in larger patient series are warranted in order to fully elucidate their role in the pathobiology of a particular lymphoma as well as their relevance for routine diagnostics. In few circumstances, a single recurrent mutation is identified in almost all cases of a given lymphoma and predominates by far in the genomic landscape of that particular tumor, e.g. the MYD88L265P mutation in Waldenström’s Macroglobulinemia (WM)/lymphoplasmacytic lymphoma (LPL)2 and the BRAFV600E mutation in hairy-cell leukemia (HCL).3 However, for the great majority of lymphomas, including chronic lymphocytic leukemia (CLL),4 diffuse large B-cell lymphoma (DLBCL),5 follicular lymphoma (FL),6 mantle cell lymphoma (MCL),7 Burkitt lymphoma (BL),8 splenic marginal zone lymphoma (SMZL)9 and most peripheral T-cell lymphoma (PTCL) subtypes,10-13 NGS studies have revealed a quite diverse and complex mutation pattern, with a limited number of frequently mutated genes accompanied by a long tail of genes with low-frequency mutations. In addition, while some genes are biased to certain lymphoma entities, e.g. SF3B1 mutations

in CLL,14,15 KLF2 mutations in SMZL,16-18 ID3 and TCF3 mutations in BL,19 STAT3 mutations in large granular lymphocyte (LGL) leukemia,11 and RHOA mutations in angioimmunoblastic T-cell lymphoma (AITL),20,21 other genes found recurrently mutated, such as genes involved in DNA repair, epigenetic modification and regulation of transcription (Figure 1), can be detected in multiple subtypes (Table 1) and even in other cancer types. Considering the genetic heterogeneity of lymphomas, also highlighted by diverse results reported in hitherto published NGS studies, it will require large-scale initiatives encompassing thousands of patients to clarify if a specific gene mutation has an impact on the current lymphoma classification/diagnostics and aids clinical decision-making, including therapy selection and response prediction. In an effort to summarize the current status of this everevolving field and provide guidelines regarding the clinical relevance of recent genetic findings, we have established the European Expert Group on NGS-based Diagnostics in Lymphomas (EGNL), with expertise in hematopathology, molecular pathology, clinical genetics, hematology, and oncology. The EGNL Group is supported by the European Research Initiative on CLL (ERIC; Scientific Working Group within the European Hematology Association, EHA) and the European Association for Haematopathology (EAHP). As our first task, we took advantage of the published literature on lymphomas to identify recurrently mutated genes with a potential clinical relevance that were reported in at least two independent studies; other genomic aberrations, e.g. translocations and copy number aberra-

Figure 1. Schematic overview of recurrently affected pathways (A) and processes (B) in B-cell lymphomas. Based on references listed in Table 1.

haematologica | 2016; 101(9)

1003

R. Rosenquist et al.

tions, may also be clinically relevant, though they are not discussed in this review. As the next step, we grouped the identified genes into the following categories based on: i) immediate impact on treatment decisions, ii) diagnostic impact, ii) prognostic impact, iv) potential clinical impact in the near future, or v) interest for research purposes only. In the end, only few genes were deemed to have direct therapeutic or diagnostic implications, while a sizeable proportion of genes were judged as prognostic and/or with a potential role for patient management within the next few years (Table 2). Though this latter category is evidently difficult to define, we decided to include genes for which recent evidence strongly suggests a diagnostic and/or predictive role, sometimes limited to retrospective studies, or for which targeted therapy is under development. In the following sections, we outline our arguments for including a particular gene in one of these categories; highlighted genes in each category are summarized in Table 2.

I. Genes with immediate impact on treatment decisions Currently, there are few genetic lesions with documented impact on therapy selection and patient management in patients with lymphoma. This category is best exempli-

fied by TP53 aberrations in CLL.22 Such aberrations are due to (i) deletions of chromosome 17p (covering the TP53 gene) seen in 5-10% of patients at diagnosis, which are often associated with TP53 mutations on the remaining allele; and, (ii) in a small fraction of CLL patients (3-6%), mutations within the TP53 gene only.23-25 Both types of aberrations (i.e. 17p-deletions and TP53 mutations) are equally adverse in CLL, portending for refractoriness to standard chemoimmunotherapy and poor overall survival.26-29 Notably, such patients experience major clinical benefit by the recently approved novel agents targeting B-cell receptor signaling, namely, the Bruton tyrosine kinase inhibitor ibrutinib and the phosphatidylinositol 3kinase delta inhibitor idelalisib that are approved for the treatment of patients carrying either of these lesions;30 however, these patients still constitute a high-risk group with an increased risk of disease recurrence with time. On these grounds, the assessment of TP53 status is essential for clinical decision-making. Hence, sequencing of the TP53 gene is now recommended for all CLL patients, in addition to FISH analysis, before the start of any line of therapy (except in the palliative situation).31,32 Indeed, testing for TP53 aberrations should be considered companion diagnostics for signaling inhibitor treatment in

Table 1A. Mutation frequencies in different B-cell lymphoma entities.

Pathway/cellular function B-cell receptor signaling CD79A/CD79B CARD11 Toll-like receptor signaling MYD88 NF-κB signaling pathway TNFPAI3 BIRC3 TRAF3 NFKBIE Notch signaling NOTCH1 NOTCH2 Other signaling pathways BRAF CXCR4 Transcription factors ID3 TCF3 KLF2 DNA repair/genomic integrity ATM TP53 POT1 Epigenetic modifiers TET2 EZH2 IDH2 CREBBP EP300

CLLa

MCLb

BLc

FLd

ABC-DLBCLe

GCB-DLBCLe

SMZLf

HCLg

WMh

<1% 1%

4% 10%

10-20% 10%

<5% 5%

10% -

20-30%

<5%

>90%

<3% <1% <2%

5-10% 5%

10% -

20% <5%

<5% <5%

8% 5% 5% 2%

40%* -

10% <1%

10-15% 5%

6% 15-20%

3% <1%

4% <10%

<1% <1%

>90% -

25%

35-60% 10-25% -

14%

11% 5% 5%

40-50% 15-20% <3%

35% -

5% -

10-25% -

10-20% -

6% 18% <1%

<1% <1% <1% <1%

<5% -

10-20% 50% 10-15%

5-10% 20% 40% <10%

5-10% 15-20% <5%

3% <1% 6% 4%

CLL: chronic lymphocytic leukemia; MCL: mantle cell lymphoma; BL: Burkitt lymphoma; FL: follicular lymphoma; ABC-DLBCL: activated B-cell-like diffuse large B-cell lymphoma; GCB-DLBCL: germinal center B-cell-like diffuse large B-cell lymphoma; SMZL: splenic marginal zone lymphoma; HCL: hairy-cell leukemia; WM: Waldenström’s Macroglobulinemia. *Genomic deletions. a Frequencies are based on the combined cohort included in Puente et al., Nature 201586 and Landau et al., Nature 201587, though the former encompasses more general practice CLL patients and the latter more advanced, clinical trial patients. bBased on Bea et al., PNAS 20137 Meissner et al., Blood 2013101, Rahal et al., Nat Med 201491. cBased on Schmitz et al., Nat Med 201219 d Based on Morin et al., Nat Genet 201061 Okosun et al., Nat Genet 20146 Okosun et al., Nat Genet 201660. eBased on Compagno et al., Nature 2009102, Davis et al., Nature 201093 Ngo et al., Nature 20115, Pasqualucci et al., Nat Genet 2011103, de Miranda et al., Blood 2014104, Bohers et al., Genes Chrom Cancer 2014105. fBased on Rossi et al., JEM 20129 Parry et al., Clin Cancer Res 201516, Piva et al., Leukemia 201518. gBased on Tiacci et al., NEJM 201141. hBased on Treon et al., NEJM 20122 Poulain et al., Clin Cancer Res 2015106.

1004

haematologica | 2016; 101(9)

NGS-based diagnostics in lymphomas

CLL, in order to identify patients with aberrant TP53 for whom the new targeted therapies, described above, represent the current treatment of choice. Furthermore, the fact that the frequency of TP53 mutations gradually increases as the disease becomes more aggressive and chemorefractory supports the need to repeat TP53 mutation analysis prior to any subsequent line of chemotherapy.31,32 Of note, it has also been demonstrated that TP53 microclones at diagnosis, i.e. subclones with a low-allelic burden detected only by NGS but not by Sanger sequencing, were selected by repeated rounds of chemoimmunotherapy and conferred a poor outcome similar to clonal TP53 mutations.33,34 If this finding is confirmed within the context of prospective clinical trials including signaling inhibitors, it is very likely that NGS-based protocols will become the recommended method for TP53 mutation screening in routine clinical practice. Mutations within TP53 have been described in most other lymphoid malignancies besides CLL, albeit with varying frequency. Although they have been linked to poor clinical outcome in DLBCL,35 SMZL16 and MCL,36,37 this information does not currently have any impact on treatment decisions or follow-up strategies for the individual lymphoma patient.

II. Genes of diagnostic potential Few lymphoma entities show a predominating, recurrent mutation, such as: the hotspot MYD88L265P mutation in more than 90% of WM/LPL,2,38-40 the hotspot BRAFV600E mutation in ~90% of HCL,41-43 and the STAT3 mutations detected in up to 40% of LGL leukemia.11,44 None of these mutations are pathognomonic of (i.e. exclusive to) a particular entity and can also be found in other lymphomas, though generally at lower frequencies (Table 1). For

instance, the MYD88L265P mutation is detected in a significant fraction of DLBCL of the activated B-cell-like subtype (ABC DLBCL),5 as well as in primary cutaneous,45 the central nervous system46 and testicular large B-cell lymphomas,47 but also in a minority of patients with CLL4,48-50 and SMZL.16,51 As another example, STAT3 mutations have been reported, albeit rarely, in immune, mainly hypoplastic, bone marrow failure characterizing a subset of patients with severe aplastic anemia or myelodysplastic syndrome.52 Moreover, some gene mutations are mainly found in a specific lymphoma entity at a relatively high frequency, whereas they are rare in other subtypes, e.g. ID3 and TCF3 mutations in BL,19 KLF2 in SMZL,16-18 SF3B1 mutations in CLL,14,15 RHOA mutations in AITL and other PTCL with a follicular helper T cell (TFH) phenotype,13,20,21,5359 and, very recently, the novel somatic mutations in RRAGC encoding a Rag GTPase protein (RagC) that were enriched in FL (16%) but were absent in other mature Bcell lymphomas.60 That said, a pattern has started to emerge in recent years where certain lymphoma entities ‘share’ common types of genetic events affecting selected pathways or biological mechanisms (Figure 1). For instance, germinal center derived B-cell malignancies, such as DLBCL of the germinal center B-cell-like type (GCB DLBCL) and FL, appear to have higher frequencies of aberrations in epigenetic-related genes (e.g. EZH2, CREPPB, TET2, IDH2),61,62 while other B-cell lymphomas demonstrate common mutations of members in the NOTCH, B-cell receptor (BcR), and NF-κB signaling pathways (Table 1).63,64 Similarly, different PTCL subtypes, such as mycosis fungoides and Sézary syndrome, AITL and other TFH-derived PTCL, and adult T-cell leukemia/lymphoma (ATLL) share frequent alterations affecting both epigenetic regulation and T-cell receptor

Table 1B. Mutation frequencies in different T-cell lymphoma entities.

Pathway/cellular function

AITL

Co-stimulatory/TCR signaling elements CD28 11% FYN 3% PLCG1 12% Epigenetic modifiers TET2 50-70% IDH2 20-40% DNMT3 20-30% JAK-STAT signaling pathway STAT3 STAT5B (N642H)* JAK1 JAK3 NF-κB signaling pathway PRKCB CARD11 TNFRSF1B Other genes RHOA (G17V)¤ 60-70% DDX3X CCR4 -

MF/SS

PTCL-NOS

LGL

HSTL

P-TLL

ALK-neg ALCL

NKTCL

20%

15%

----

2% 3% 36%

20-48° 10-27° -

35% <5% -

---38%# -6% 38%# -20%

25% -

35% <10% 30-40%

5-10% -

15% 6%

----

33% 24% -

7% 7%

18%° -

--20% --

<15% -

ATLL

22% -

AITL: angioimmunoblastic T-cell lymphoma; MF: mycosis fungoides, SS: Sézary syndrome; PTCL-NOS: peripheral T-cell lymphoma not otherwise specified; LGL: large granular lymphocytic leukemia; HSTL: hepatosplenic T-cell lymphoma; T-PLL: T-cell prolymphocytic leukemia; ATLL: adult T-cell leukemia/lymphoma; ALK-neg ALCL: ALK-negative anaplastic large cell lymphoma, NKTCL, extranodal NK/T-cell lymphoma, nasal-type. *STAT5B mutations were also reported in 36% of EATL type 2 (Kücück et al Nat Commun 2015107). °Mutations in TET2, RHOA and DNMT3A also reported in the subgroup of PTCL-NOS with TFH phenotype. #38% of systemic ALK-negative ALCL patients were reported to carry both JAK1 and STAT3 mutations (Crescenzo et al, Cancer Cell 201565). ¤Whereas RHOA mutations in AITL almost invariably involve the hotspot G17V, RHOA mutations in ATLL are more widely distributed (Kataoka et al, Nat Genet, 201556; Nagata et al, Blood 2016108).

haematologica | 2016; 101(9)

1005

R. Rosenquist et al.

(TCR) signaling, whereas mutations of members of the JAK-STAT signaling pathway (STAT3, STAT5B) are shared by several cytotoxic T-cell or NK-cell lymphomas, such as T-LGL, nasal NK/T-cell lymphomas, hepatosplenic T-cell lymphoma, enteropathy-associated T-cell lymphoma (type 2) and ALK-negative anaplastic large cell lymphomas11,56,65,66 (Table 1). Subgroups of patients within a lymphoma entity may also display differential mutation profiles, initially shown for the GCB vs. ABC subtypes of DLBCL (Table 1) and, more recently, also demonstrated in other lymphomas e.g. HCL, where cases expressing the IGHV4-34 gene in the clonotypic B-cell receptor lack the canonical BRAFV600E, and instead display enrichment for MAP2K1 mutations with an overall similar profile to the HCL-variant.67,68 Similarly, a subgroup of SMZLs with IGHV1-2 expressing B-cell receptor commonly harbors inactivating KLF2 mutations and 7q deletions.16,17 Potentially, this might aid future diagnostics aiming to distinguish these subtypes, however the utility and applicability of this approach have to be studied further. In summary, presently, MYD88, BRAF, ID3, TCF3, STAT3, STAT5B, RHOA, TET2, and IDH2 mutations are the only genes that can be considered as a complement to the current set-up for lymphoma diagnostics. However, the list is rapidly expanding as evidenced by the case of RRAGC mutations, for example.

III. Genes with prognostic potential Several gene mutations have been associated with clinical outcome in various lymphoma subtypes. In CLL, besides TP53 and ATM mutations, which are both known to confer poor prognosis, recent high-throughput NGS studies have revealed recurrent mutations within NOTCH1, SF3B1, and BIRC3 for example, that were reported to be associated with poor clinical outcome with higher frequencies in relapsing/treatment-refractory CLL and in Richter’s syndrome.48,69-79 More recent studies have also identified additional gene mutations that may confer a worse outcome in CLL, e.g. NKFBIE, EGR2, and RPS15, although they have been studied less.78-79 In a recent multicenter study conducted within ERIC, sequencing of TP53, NOTCH1, SF3B1, BIRC3 and MYD88 was performed in a large patient series (totaling 3,490 patients), revealing that TP53 and SF3B1 mutations, but not NOTCH1 mutations, remained as independent prognostic markers of shorter time to first treatment in multivariate analysis, even amongst patients expressing unmutated IGHV genes.50 A few published clinical trials have also pointed to a prognostic and even predictive role of SF3B1 and NOTCH1 mutations in CLL,28 where, in particular, the latter confers resistance to the anti-CD20 monoclonal antibodies rituximab29 and ofatumumab,80 however, this needs further exploration and validation. Currently, as a new ERIC project, a dedicated NGS-based gene panel (including 11 genes) has been designed with the purpose to test different targeted enrichment techniques and evaluate intercenter variability and reproducibility. Similar gene panel-based efforts have also been performed for SMZL, revealing a high frequency of TP53, KLF2, NOTCH2, TNFAIP3, and MYD88 mutations,17,18 and demonstrating NOTCH2 and TP53 mutations as independent markers of short treatment-free and overall survival, respectively.16 Nonetheless, caution is required given the retrospective nature of the published studies and the 1006

overall rarity of SMZL raising concerns about potential selection biases. MCL is characterized by a relatively high number of recurrent secondary genomic aberrations, but few of them have demonstrated additional prognostic value. From recent NGS studies, the prognostic value of NOTCH1/NOTCH2 mutations was recently highlighted7,81 in addition to TP53 defects. In DLBCL, the prognostic impact of MYC translocations critically depends on the second hit, with cases harboring TP53 mutation and MYC translocation showing the worst overall survival, followed by those cases carrying MYC and BCL2 translocation.82,83 Indeed, DLBCL with TP53 mutation and MYC translocation is a newly recognized subset of ‘double-hit’ lymphoma, accounting for one-third of MYC translocation positive DLBCL. Hence, it is pivotal to perform TP53 mutation screening, in addition to the detection of BCL2 translocation in MYC translocation positive DLBCL, in order to distinguish the double-hit DLBCL from those with an isolated MYC translocation.84 Among PTCL, TET2 mutations in AITL patients were reported to associate with a more aggressive clinical presentation and a shorter progression-free survival,53 whereas in a recent study of NK-TCL DDX3X mutations conferred a particularly poor prognosis.85 Considering the significant differences in mutation frequencies observed between different studies, best exemplified by DLBCL, large-scale efforts are now needed, in particular for rarer entities, to fully understand which genes are the most relevant and will retain independent prognostic impact in relation to other known clinical/molecular markers. This is particularly relevant in light of recent studies86-88 pointing to a prognostic relevance of particular combinations of mutations and/or an increasing

Table 2. Categorization of gene mutations based on current evidence levels.

Category

Gene mutations

1. Immediate impact on patient care 2. Diagnostic impact

TP53 mutations (exons 4-10) in CLL

3. Prognostic impact

4. Potential clinical impact in the near future

MYD88L265P mutation in WM/LPL BRAFV600E mutation in HCL KLF2 mutations in SMZL ID3 and TCF3 mutations in BL STAT3 mutations in LGLL RHOA, TET2, IDH2 and DNMT3A mutations in AITL and other TFH-derived PTCL CLL: TP53, ATM, BIRC3, NFKBIE, NOTCH1, SF3B1 MCL: TP53, NOTCH1, NOTCH2 mutations SMZL: NOTCH2, TP53 mutations DLBCL: TP53 mutation & MYC translocation NKTCL: DDX3X mutations Therapy response to BcR inhibitors: WM: MYD88, CXCR4 mutations DLBCL: CD79B mutations (responsive) CARD11, MYD88 mutations (non-responsive) Resistance to BcR inhibitors: BTKC481S, PCLG2 mutations New inhibitors under development: EZH2, SF3B1 & NOTCH1

Based on references listed in Table 1. CLL: chronic lymphocytic leukemia; WM: Waldenström’s Macroglobulinemia; LPL: lymphoplasmacytic lymphoma; HCL: hairy-cell leukemia; BL: Burkitt lymphoma; AITL: angioimmunoblastic T-cell lymphoma; MCL: mantle cell lymphoma; SMZL: splenic marginal zone lymphoma; DLBCL: diffuse large B-cell lymphoma; NKTCL: NK T-cell lymphoma. a

haematologica | 2016; 101(9)

NGS-based diagnostics in lymphomas

number of ‘driver mutations’. In addition, the time point of mutation screening may also differ depending on the disease entity and response to therapy.

IV. Genes with potential clinical impact in the near future Recent studies in different lymphoma subtypes have highlighted relevant associations between certain recurrent gene mutations and response to treatment. The most striking example is perhaps offered by MYD88 and CXCR4 mutations, which were shown to affect responses to ibrutinib in WM. Indeed, patients with MYD88L265PCXCR4WT (with WT indicating wild-type) status had 100% overall response rate (91% major response rate) as opposed to 86% (62%) and 71% (29%) for MYD88L265PCXCR4WHIM and MYD88WTCXCR4WT patients, respectively.89 Although these results were obtained from a small cohort of patients with WM, they offer a tantalizing glimpse into the future of lymphoma treatment with novel companion diagnostics tailored to novel therapeutic agents. From recent studies, we have also learnt that mutations within the BTK (C481S) and PLCG2 genes may emerge in CLL patients relapsing after and/or refractory to ibrutinib treatment, and we foresee that the assessment of these genes may soon be incorporated in the diagnostic set-up.90 Preliminary results also indicate that the divergent responses of patients with MCL or DLBCL to ibrutinib may be linked to distinct profiles of genomic aberrations, e.g. BIRC3 mutations in MCL91 and isolated MYD88 mutation in DLBCL92 as well as mutations downstream of BTK, such as activating mutations of CARD11 in DLBCL,93,94 may make these tumors resistant to BTK inhibition. Recent reports of genes linked to particular physiological processes have revealed new types of mechanisms that may be suitable for targeted therapy. For mutant BRAF, the inhibitor vemurafenib is already in clinical trials in relapsed/refractory HCL, demonstrating high activity even among heavily pre-treated patients.3,95 New types of promising inhibitors have also been developed for EZH2 (a histone methyl transferase),96 SF3B1 (a splicing factor),97 NOTCH198 and IDH2,99 and in some cases have already entered early phase clinical trials. The high frequency of TET2 and/or DNMT3A mutations in AITL and other TFHderived PTCL may also support the rationale to use demethylating agents as an alternative way to treat patients, supported by the results of a recent single report.100 More generally, targeting certain epigenetic abnormalities or alterations in pathways frequently involved in patients with PTCL (TCR, JAK-STAT, NF-κB; Table 2) represents an attractive approach, also taking into consideration the overall poor outcome of the majority of such patients with conventional chemotherapy-based approaches. In conclusion, we expect that the list of potential therapeutic targets will expand quickly in coming years, once markers have been functionally validated and new compounds have been discovered.

References 1. Swerdlow S, Campo E, Harris N, et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues: Lyon, France: IARC Press, 2008.

haematologica | 2016; 101(9)

V. Genes for research purposes only Finally, for most recurrently mutated genes identified thus far, we do not yet understand their functional role and/or their clinical association, either alone or in the presence of cooperating events. These genes might still be of interest, however, before large-scale studies are performed, it will not be possible to discern their potential contribution to disease pathobiology. Having said that, and similar to category IV above, this subgroup of genes will be a ‘moving target’ depending on the evidence level ascertained for a particular mutation in the coming years, and genes may be taken out if a certain type of mutation is judged to have minor impact for a specific lymphoma entity.

How to move forward in the diagnostic field using NGS? As mentioned, targeted NGS allows us to select and test many genes and samples simultaneously, and this approach has now been validated for CLL, SMZL and DLBCL, amongst others. In addition, targeted NGS permits a high sequence read depth, an important factor for studying minor subclones and thus clonal heterogeneity, and can also be adapted to formalin-fixed, paraffin-embedded (FFPE) tumor material, which in reality is the source of material for most diagnostic lymphoma specimens. To gain further insight into NGS-based diagnostics in lymphoma, the EGNL Group has participated in the design of a lymphoma gene panel, compatible with FFPE material, that includes 30 recurrently mutated genes. Our ambition now is to perform a multi-center validation study using a defined number of matched FFPE/fresh-frozen lymphoma specimens to set the technical requirement for NGS-based diagnostics (including metrics such as sequence coverage, specificity, sensitivity and reproducibility of the technique). Once validated, large-scale international collaborative efforts can be conducted for each lymphoma entity, in particular within ongoing or planned treatment studies, to fully understand how to adapt and implement NGS-based diagnostics in day-to-day routine diagnostics. Acknowledgments This work was supported by the Swedish Cancer Society, the Swedish Research Council, and The Lion’s Cancer Research Foundation, Uppsala, Sweden; the Special Program in Molecular Clinical Oncology, 5 x 1000 No. #9965 and 10007, AIRC, Milan, Italy, and Ricerca Finalizzata RF-2010-2318823 and RF-2011-02349712, Ministero della Salute, Rome, Italy; the Institut National du Cancer (INCa) and Association pour la Recherche contre le Cancer (ARC), France; the Spanish Ministry of Economy SAF2015-64885-R and Generalitat de Catalunya 2009SGR992, Spain; H2020 “AEGLE, An analytics framework for integrated and personalized healthcare services in Europe”, and “MEDGENET” (No.692298) by the EU. EC is an ICREA-Academia researcher of the Generalitat de Catalunya, Spain.

2. Treon SP, Xu L, Yang G, et al. MYD88 L265P somatic mutation in Waldenstrom's macroglobulinemia. N Engl J Med. 2012;367 (9):826-833. 3. Tiacci E, Park JH, De Carolis L, et al. Targeting Mutant BRAF in Relapsed or Refractory Hairy-Cell Leukemia. N Engl J Med.

2015;373(18):1733-1747. 4. Puente XS, Pinyol M, Quesada V, et al. Whole-genome sequencing identifies recurrent mutations in chronic lymphocytic leukaemia. Nature. 2011;475(7354):101-105. 5. Ngo VN, Young RM, Schmitz R, et al. Oncogenically active MYD88 mutations in

1007

R. Rosenquist et al.

10. 11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

1008

human lymphoma. Nature. 2011;470(7332): 115-119. Okosun J, Bodor C, Wang J, et al. Integrated genomic analysis identifies recurrent mutations and evolution patterns driving the initiation and progression of follicular lymphoma. Nat Genet. 2014;46(2):176-181. 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. Richter J, Schlesner M, Hoffmann S, et al. Recurrent mutation of the ID3 gene in Burkitt lymphoma identified by integrated genome, exome and transcriptome sequencing. Nat Genet. 2012;44(12):1316-1320. Rossi D, Trifonov V, Fangazio M, et al. The coding genome of splenic marginal zone lymphoma: activation of NOTCH2 and other pathways regulating marginal zone development. J Exp Med. 2012;209(9):1537-1551. Couronne L, Bastard C, Bernard OA. TET2 and DNMT3A mutations in human T-cell lymphoma. N Engl J Med. 2012;366(1):95-96. Koskela HL, Eldfors S, Ellonen P, et al. Somatic STAT3 mutations in large granular lymphocytic leukemia. N Engl J Med. 2012;366(20):1905-1913. Odejide O, Weigert O, Lane AA, et al. A targeted mutational landscape of angioimmunoblastic T-cell lymphoma. Blood. 2014;123(9):1293-1296. Sakata-Yanagimoto M, Enami T, Yoshida K, et al. Somatic RHOA mutation in angioimmunoblastic T cell lymphoma. Nat Genet. 2014;46(2):171-175. Wang L, Lawrence MS, Wan Y, et al. SF3B1 and other novel cancer genes in chronic lymphocytic leukemia. N Engl J Med. 2011;365(26):2497-2506. Quesada V, Conde L, Villamor N, et al. Exome sequencing identifies recurrent mutations of the splicing factor SF3B1 gene in chronic lymphocytic leukemia. Nat Genet. 2012;44(1):47-52. Parry M, Rose-Zerilli MJ, Ljungstrom V, et al. Genetics and Prognostication in Splenic Marginal Zone Lymphoma: Revelations from Deep Sequencing. Clin Cancer Res. 2015;21(18):4174-4183. Clipson A, Wang M, de Leval L, et al. KLF2 mutation is the most frequent somatic change in splenic marginal zone lymphoma and identifies a subset with distinct genotype. Leukemia. 2015;29(5):1177-1185. Piva R, Deaglio S, Fama R, et al. The Kruppellike factor 2 transcription factor gene is recurrently mutated in splenic marginal zone lymphoma. Leukemia. 2015;29 (2):503-507. Schmitz R, Young RM, Ceribelli M, et al. Burkitt lymphoma pathogenesis and therapeutic targets from structural and functional genomics. Nature. 2012;490(7418):116-120. Yoo HY, Sung MK, Lee SH, et al. A recurrent inactivating mutation in RHOA GTPase in angioimmunoblastic T cell lymphoma. Nat Genet. 2014;46(4):371-375. Palomero T, Couronne L, Khiabanian H, et al. Recurrent mutations in epigenetic regulators, RHOA and FYN kinase in peripheral T cell lymphomas. Nat Genet. 2014;46(2):166170. Stilgenbauer S. Prognostic markers and standard management of chronic lymphocytic leukemia. Hematology Am Soc Hematol Educ Program. 2015;2015(1):368-377. Dohner H, Stilgenbauer S, Benner A, et al. Genomic aberrations and survival in chronic lymphocytic leukemia. N Engl J Med. 2000;343(26):1910-1916.

24. Malcikova J, Smardova J, Rocnova L, et al. Monoallelic and biallelic inactivation of TP53 gene in chronic lymphocytic leukemia: selection, impact on survival, and response to DNA damage. Blood. 2009;114(26):53075314. 25. Zenz T, Eichhorst B, Busch R, et al. TP53 mutation and survival in chronic lymphocytic leukemia. J Clin Oncol. 2010;28(29): 44734479. 26. Trbusek M, Smardova J, Malcikova J, et al. Missense mutations located in structural p53 DNA-binding motifs are associated with extremely poor survival in chronic lymphocytic leukemia. J Clin Oncol. 2011;29(19):2703-2708. 27. Gonzalez D, Martinez P, Wade R, et al. Mutational status of the TP53 gene as a predictor of response and survival in patients with chronic lymphocytic leukemia: results from the LRF CLL4 trial. J Clin Oncol. 2011;29(16):2223-2229. 28. Oscier DG, Rose-Zerilli MJ, Winkelmann N, et al. The clinical significance of NOTCH1 and SF3B1 mutations in the UK LRF CLL4 trial. Blood. 2013;121(3):468-475. 29. Stilgenbauer S, Schnaiter A, Paschka P, et al. Gene mutations and treatment outcome in chronic lymphocytic leukemia: results from the CLL8 trial. Blood. 2014;123(21):32473254. 30. Stilgenbauer S, Furman RR, Zent CS. Management of chronic lymphocytic leukemia. Am Soc Clin Oncol Educ Book. 2015;164-175. 31. Hallek M, Cheson BD, Catovsky D, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood. 2008;111(12):54465456. 32. Pospisilova S, Gonzalez D, Malcikova J, et al. ERIC recommendations on TP53 mutation analysis in chronic lymphocytic leukemia. Leukemia. 2012;26(7):1458-1461. 33. Rossi D, Khiabanian H, Spina V, et al. Clinical impact of small TP53 mutated subclones in chronic lymphocytic leukemia. Blood. 2014;123(14):2139-2147. 34. Malcikova J, Stano-Kozubik K, Tichy B, et al. Detailed analysis of therapy-driven clonal evolution of TP53 mutations in chronic lymphocytic leukemia. Leukemia. 2015;29(4):877-885. 35. Xu-Monette ZY, Wu L, Visco C, et al. Mutational profile and prognostic significance of TP53 in diffuse large B-cell lymphoma patients treated with R-CHOP: report from an International DLBCL RituximabCHOP Consortium Program Study. Blood. 2012;120(19):3986-3996. 36. Nordstrom L, Sernbo S, Eden P, et al. SOX11 and TP53 add prognostic information to MIPI in a homogenously treated cohort of mantle cell lymphoma--a Nordic Lymphoma Group study. Br J Haematol. 2014;166(1):98-108. 37. Delfau-Larue MH, Klapper W, Berger F, et al. High-dose cytarabine does not overcome the adverse prognostic value of CDKN2A and TP53 deletions in mantle cell lymphoma. Blood. 2015;126(5):604-611. 38. Poulain S, Roumier C, Decambron A, et al. MYD88 L265P mutation in Waldenstrom macroglobulinemia. Blood. 2013;121(22): 4504-4511. 39. Varettoni M, Arcaini L, Zibellini S, et al. Prevalence and clinical significance of the MYD88 (L265P) somatic mutation in Waldenstrom's macroglobulinemia and relat-

40.

41. 42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

ed lymphoid neoplasms. Blood. 2013; 121(13):2522-2528. Rossi D. Role of MYD88 in lymphoplasmacytic lymphoma diagnosis and pathogenesis. Hematology Am Soc remato Educ Program. 2014;2014(1):113-118. Tiacci E, Trifonov V, Schiavoni G, et al. BRAF mutations in hairy-cell leukemia. N Engl J Med. 2011;364(24):2305-2315. Tiacci E, Schiavoni G, Forconi F, et al. Simple genetic diagnosis of hairy cell leukemia by sensitive detection of the BRAF-V600E mutation. Blood. 2012;119(1):192-195. Arcaini L, Zibellini S, Boveri E, et al. The BRAF V600E mutation in hairy cell leukemia and other mature B-cell neoplasms. Blood. 2012;119(1):188-191. Jerez A, Clemente MJ, Makishima H, et al. STAT3 mutations unify the pathogenesis of chronic lymphoproliferative disorders of NK cells and T-cell large granular lymphocyte leukemia. Blood. 2012;120(15):3048-3057. Pham-Ledard A, Cappellen D, Martinez F, Vergier B, Beylot-Barry M, Merlio JP. MYD88 somatic mutation is a genetic feature of primary cutaneous diffuse large B-cell lymphoma, leg type. J Invest Dermatol. 2012;132(8):2118-2120. Gonzalez-Aguilar A, Idbaih A, Boisselier B, et al. Recurrent mutations of MYD88 and TBL1XR1 in primary central nervous system lymphomas. Clin Cancer Res. 2012;18(19): 5203-5211. Kraan W, van Keimpema M, Horlings HM, et al. High prevalence of oncogenic MYD88 and CD79B mutations in primary testicular diffuse large B-cell lymphoma. Leukemia. 2014;28(3):719-720. Jeromin S, Weissmann S, Haferlach C, et al. SF3B1 mutations correlated to cytogenetics and mutations in NOTCH1, FBXW7, MYD88, XPO1 and TP53 in 1160 untreated CLL patients. Leukemia. 2014;28(1):108-117. Martinez-Trillos A, Pinyol M, Navarro A, et al. Mutations in TLR/MYD88 pathway identify a subset of young chronic lymphocytic leukemia patients with favorable outcome. Blood. 2014;123(24):3790-3796. Baliakas P, Hadzidimitriou A, Sutton LA, et al. Recurrent mutations refine prognosis in chronic lymphocytic leukemia. Leukemia. 2015;29(2):329-336. Martinez-Lopez A, Curiel-Olmo S, Mollejo M, et al. MYD88 (L265P) somatic mutation in marginal zone B-cell lymphoma. Am J Surg Pathol. 2015;39(5):644-651. Jerez A, Clemente MJ, Makishima H, et al. STAT3 mutations indicate the presence of subclinical T-cell clones in a subset of aplastic anemia and myelodysplastic syndrome patients. Blood. 2013;122(14):2453-2459. Lemonnier F, Couronne L, Parrens M, et al. Recurrent TET2 mutations in peripheral Tcell lymphomas correlate with TFH-like features and adverse clinical parameters. Blood. 2012;120(7):1466-1469. Cairns RA, Iqbal J, Lemonnier F, et al. IDH2 mutations are frequent in angioimmunoblastic T-cell lymphoma. Blood. 2012;119(8):1901-1903. Lee SH, Kim JS, Kim J, et al. A highly recurrent novel missense mutation in CD28 among angioimmunoblastic T-cell lymphoma patients. Haematologica. 2015;100(12):e505-507. Kataoka K, Nagata Y, Kitanaka A, et al. Integrated molecular analysis of adult T cell leukemia/lymphoma. Nature genetics. 2015;47 (11):1304-1315. Vaque JP, Gomez-Lopez G, Monsalvez V, et al. PLCG1 mutations in cutaneous T-cell lym-

haematologica | 2016; 101(9)

NGS-based diagnostics in lymphomas phomas. Blood. 2014;123(13):2034-2043. 58. Ungewickell A, Bhaduri A, Rios E, et al. Genomic analysis of mycosis fungoides and Sezary syndrome identifies recurrent alterations in TNFR2. Nat Genet. 2015;47(9): 1056-1060. 59. Wang L, Ni X, Covington KR, et al. Genomic profiling of Sezary syndrome identifies alterations of key T cell signaling and differentiation genes. Nat Genet. 2015;47(12):14261434. 60. Okosun J, Wolfson RL, Wang J, et al. Recurrent mTORC1-activating RRAGC mutations in follicular lymphoma. Nature Genet. 2016;48(2):183-188. 61. Morin RD, Johnson NA, Severson TM, et al. Somatic mutations altering EZH2 (Tyr641) in follicular and diffuse large B-cell lymphomas of germinal-center origin. Nat Genet. 2010;42(2):181-185. 62. Asmar F, Punj V, Christensen J, et al. Genomewide profiling identifies a DNA methylation signature that associates with TET2 mutations in diffuse large B-cell lymphoma. Haematologica. 2013;98(12):1912-1920. 63. Rossi D, Ciardullo C, Gaidano G. Genetic aberrations of signaling pathways in lymphomagenesis: revelations from next generation sequencing studies. Semin Cancer Biol. 2013;23(6):422-430. 64. Sutton LA, Rosenquist R. The complex interplay between cell-intrinsic and cell-extrinsic factors driving the evolution of chronic lymphocytic leukemia. Semin Cancer Biol. 2015;34(22-35. 65. Crescenzo R, Abate F, Lasorsa E, et al. Convergent mutations and kinase fusions lead to oncogenic STAT3 activation in anaplastic large cell lymphoma. Cancer Cell. 2015;27(4):516-532. 66. Nicolae A, Xi L, Pittaluga S, et al. Frequent STAT5B mutations in gammadelta hepatosplenic T-cell lymphomas. Leukemia. 2014;28(11):2244-2248. 67. Xi L, Arons E, Navarro W, et al. Both variant and IGHV4-34-expressing hairy cell leukemia lack the BRAF V600E mutation. Blood. 2012;119(14):3330-3332. 68. Waterfall JJ, Arons E, Walker RL, et al. High prevalence of MAP2K1 mutations in variant and IGHV4-34-expressing hairy-cell leukemias. Nat Genet. 2014;46(1):8-10. 69. Schnaiter A, Paschka P, Rossi M, et al. NOTCH1, SF3B1, and TP53 mutations in fludarabine-refractory CLL patients treated with alemtuzumab: results from the CLL2H trial of the GCLLSG. Blood. 2013;122(7):1266-1270. 70. Rossi D, Bruscaggin A, Spina V, et al. Mutations of the SF3B1 splicing factor in chronic lymphocytic leukemia: association with progression and fludarabine-refractoriness. Blood. 2011;118(26):6904-6908. 71. Cortese D, Sutton LA, Cahill N, et al. On the way towards a 'CLL prognostic index': focus on TP53, BIRC3, SF3B1, NOTCH1 and MYD88 in a population-based cohort. Leukemia. 2014;28(3):710-713. 72. Del Giudice I, Rossi D, Chiaretti S, et al. NOTCH1 mutations in +12 chronic lymphocytic leukemia (CLL) confer an unfavorable prognosis, induce a distinctive transcriptional profiling and refine the intermediate prognosis of +12 CLL. Haematologica. 2012;97(3): 437-441. 73. Rossi D, Fangazio M, Rasi S, et al. Disruption of BIRC3 associates with fludarabine chemorefractoriness in TP53 wild-type chronic lymphocytic leukemia. Blood. 2012;119(12):2854-2862. 74. Rossi D, Rasi S, Spina V, et al. Integrated

haematologica | 2016; 101(9)

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

91.

mutational and cytogenetic analysis identifies new prognostic subgroups in chronic lymphocytic leukemia. Blood. 2013;121(8):14031412. Weissmann S, Roller A, Jeromin S, et al. Prognostic impact and landscape of NOTCH1 mutations in chronic lymphocytic leukemia (CLL): a study on 852 patients. Leukemia. 2013;27(12):2393-2396. Villamor N, Conde L, Martinez-Trillos A, et al. NOTCH1 mutations identify a genetic subgroup of chronic lymphocytic leukemia patients with high risk of transformation and poor outcome. Leukemia. 2013;27(5):11001106. Mansouri L, Cahill N, Gunnarsson R, et al. NOTCH1 and SF3B1 mutations can be added to the hierarchical prognostic classification in chronic lymphocytic leukemia. Leukemia. 2013;27(2):512-514. Ljungstrom V, Cortese D, Young E, et al. Whole-exome sequencing in relapsing chronic lymphocytic leukemia: clinical impact of recurrent RPS15 mutations. Blood. 2015; Mansouri L, Sutton LA, Ljungstrom V, et al. Functional loss of IkappaBepsilon leads to NFkappaB deregulation in aggressive chronic lymphocytic leukemia. J Exp Med. 2015;212(6):833-843. Tausch E, BecK P, Schlenk R, et al. NOTCH1 mutation and treatment outcome in CLL patients treated with Chlorambucil (Chl) or Ofatumumab-Chl (O-Chl): Results from the Phase III Study COMPLEMENT 1 (OMB110911). Blood. 2013;122(21):527 abstr. Kridel R, Meissner B, Rogic S, et al. Whole transcriptome sequencing reveals recurrent NOTCH1 mutations in mantle cell lymphoma. Blood. 2012;119(9):1963-1971. Schiefer AI, Kornauth C, Simonitsch-Klupp I, et al. Impact of Single or Combined Genomic Alterations of TP53, MYC, and BCL2 on Survival of Patients With Diffuse Large B-Cell Lymphomas: A Retrospective Cohort Study. Medicine. 2015;94(52):e2388. Gebauer N, Bernard V, Gebauer W, Thorns C, Feller AC, Merz H. TP53 mutations are frequent events in double-hit B-cell lymphomas with MYC and BCL2 but not MYC and BCL6 translocations. Leuk Lymphoma. 2015;56(1): 179-185. Clipson A, Barrans S, Zeng N, et al. The prognosis of MYC translocation positive diffuse large B-cell lymphoma depends on the second hit. J Pathol. 2015;1(3):125–133. Jiang L, Gu ZH, Yan ZX, et al. Exome sequencing identifies somatic mutations of DDX3X in natural killer/T-cell lymphoma. Nat Genet. 2015;47(9):1061-1066. Puente XS, Bea S, Valdes-Mas R, et al. Noncoding recurrent mutations in chronic lymphocytic leukaemia. Nature. 2015;526 (7574):519-524. Landau DA, Tausch E, Taylor-Weiner AN, et al. Mutations driving CLL and their evolution in progression and relapse. Nature. 2015;526(7574):525-530. Guieze R, Robbe P, Clifford R, et al. Presence of multiple recurrent mutations confers poor trial outcome of relapsed/refractory CLL. Blood. 2015;126(18):2110-2117. Treon SP, Tripsas CK, Meid K, et al. Ibrutinib in previously treated Waldenstrom's macroglobulinemia. N Engl J Med. 2015;372(15):1430-1440. Woyach JA, Furman RR, Liu TM, et al. Resistance mechanisms for the Bruton's tyrosine kinase inhibitor ibrutinib. N Engl J Med. 2014;370(24):2286-2294. Rahal R, Frick M, Romero R, et al.

Pharmacological and genomic profiling identifies NF-kappaB-targeted treatment strategies for mantle cell lymphoma. Nature Med. 2014;20(1):87-92. 92. Wilson WH, Young RM, Schmitz R, et al. Targeting B cell receptor signaling with ibrutinib in diffuse large B cell lymphoma. Nature Med. 2015;21(8):922-926. 93. Davis RE, Ngo VN, Lenz G, et al. Chronic active B-cell-receptor signalling in diffuse large B-cell lymphoma. Nature. 2010;463(7277):88-92. 94. Young RM, Staudt LM. Targeting pathological B cell receptor signalling in lymphoid malignancies. Nat Rev Drug Discov. 2013;12(3):229-243. 95. Pettirossi V, Santi A, Imperi E, et al. BRAF inhibitors reverse the unique molecular signature and phenotype of hairy cell leukemia and exert potent antileukemic activity. Blood. 2015;125(8):1207-1216. 96. Beguelin W, Popovic R, Teater M, et al. EZH2 is required for germinal center formation and somatic EZH2 mutations promote lymphoid transformation. Cancer Cell. 2013;23(5):677692. 97. Larrayoz M, Blakemore SJ, Dobson RC, et al. The SF3B1 inhibitor spliceostatin A (SSA) elicits apoptosis in chronic lymphocytic leukaemia cells through downregulation of Mcl-1. Leukemia. 2016;30(2):351-360. 98. Lopez-Guerra M, Xargay-Torrent S, Rosich L, et al. The gamma-secretase inhibitor PF03084014 combined with fludarabine antagonizes migration, invasion and angiogenesis in NOTCH1-mutated CLL cells. Leukemia. 2015;29(1):96-106. 99. Wang F, Travins J, DeLaBarre B, et al. Targeted inhibition of mutant IDH2 in leukemia cells induces cellular differentiation. Science. 2013;340(6132):622-626. 100. Cheminant M, Bruneau J, Kosmider O, et al. Efficacy of 5-azacytidine in a TET2 mutated angioimmunoblastic T cell lymphoma. Br J Haematol. 2015;168(6):913-916. 101. Meissner B, Kridel R, Lim RS, et al. The E3 ubiquitin ligase UBR5 is recurrently mutated in mantle cell lymphoma. Blood. 2013;121 (16):3161-3164. 102. Compagno M, Lim WK, Grunn A, et al. Mutations of multiple genes cause deregulation of NF-kappaB in diffuse large B-cell lymphoma. Nature. 2009;459(7247):717-721. 103. Pasqualucci L, Trifonov V, Fabbri G, et al. Analysis of the coding genome of diffuse large B-cell lymphoma. Nat Genet. 2011; 43(9):830-837. 104. de Miranda NF, Georgiou K, Chen L, et al. Exome sequencing reveals novel mutation targets in diffuse large B-cell lymphomas derived from Chinese patients. Blood. 2014;124(16):2544-2553. 105. Bohers E, Mareschal S, Bouzelfen A, et al. Targetable activating mutations are very frequent in GCB and ABC diffuse large B-cell lymphoma. Genes Chromosomes Cancer. 2014;53(2):144-153. 106. Poulain S, Roumier C, Venet-Caillault A, et al. Genomic Landscape of CXCR4 Mutations in Waldenström Macroglobulinemia. Clin Cancer Res. 2016;22(6):1480-1488. 107. Küçük C, Jiang B, Hu X, Zhang W, et al. Activating mutations of STAT5B and STAT3 in lymphomas derived from -T or NK cells. Nat Commun. 2015;6:6025. 108. Nagata Y, Kontani K, Enami T, et al. Variegated RHOA mutations in adult T-cell leukemia/lymphoma. Blood. 2016;127(5): 596-604.

1009

REVIEW ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

The relevance of PTEN-AKT in relation to NOTCH1-directed treatment strategies in T-cell acute lymphoblastic leukemia

Rui D. Mendes,1 Kirsten Canté-Barrett,1 Rob Pieters,1,2 and Jules P.P. Meijerink1,2

Haematologica 2016 Volume 101(9):1010-1017

Department of Pediatric Oncology/Hematology, Erasmus MC Rotterdam-Sophia Children’s Hospital, Rotterdam, The Netherlands; and 2Princess Máxima Center for Pediatric Oncology, Utrecht, The Netherlands

ABSTRACT

Correspondence: j.meijerink@erasmusmc.nl

Received: March 18, 2016 Accepted: June 1, 2016 Pre-published: no prepublication doi:10.3324/haematol.2016.146381

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1110

©2016 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Copies of articles are allowed for personal or internal use. Permission in writing from the publisher is required for any other use.

1010

he tumor suppressor phosphatase and tensin homolog (PTEN) negatively regulates phosphatidylinositol 3-kinase (PI3K)-AKT signaling and is often inactivated by mutations (including deletions) in a variety of cancer types, including T-cell acute lymphoblastic leukemia. Here we review mutation-associated mechanisms that inactivate PTEN together with other molecular mechanisms that activate AKT and contribute to T-cell leukemogenesis. In addition, we discuss how Pten mutations in mouse models affect the efficacy of gamma-secretase inhibitors to block NOTCH1 signaling through activation of AKT. Based on these models and on observations in primary diagnostic samples from patients with T-cell acute lymphoblastic leukemia, we speculate that PTEN-deficient cells employ an intrinsic homeostatic mechanism in which PI3K-AKT signaling is dampened over time. As a result of this reduced PI3K-AKT signaling, the level of AKT activation may be insufficient to compensate for NOTCH1 inhibition, resulting in responsiveness to gamma-secretase inhibitors. On the other hand, de novo acquired PTEN-inactivating events in NOTCH1-dependent leukemia could result in temporary, strong activation of PI3K-AKT signaling, increased glycolysis and glutaminolysis, and consequently gamma-secretase inhibitor resistance. Due to the central role of PTEN-AKT signaling and in the resistance to NOTCH1 inhibition, AKT inhibitors may be a promising addition to current treatment protocols for T-cell acute lymphoblastic leukemia.

T-cell acute lymphoblastic leukemia T-cell acute lymphoblastic leukemia (T-ALL) is a cancer of developing T cells in the thymus. T-ALL is characterized by chromosomal rearrangements. These rearrangements can lead to the aberrant activation of oncogenic transcription factors by placing their genes under the control of promoters and/or enhancers of Tcell receptor genes, the BCL11B gene, or other genes; occasionally, these rearrangements can give rise to oncogenic fusion proteins. The activated oncogenic transcription factors include TAL1 and LMO2 (and related family members), TLX1, TLX3, NKX2-1, HOXA, and MEF2C; in addition, certain oncogenic fusion proteins can directly activate the HOXA or MEF2C genes.1,2 Oncogenic proteins facilitate the developmental arrest of pre-leukemic immature T cells. We previously proposed that these chromosomal rearrangements should be classified as type A aberrations, as they are generally considered to be the driving oncogenic event associated with unique expression profiles.2 Based upon their gene expression signatures, T-ALL can be classified into the following four major subtypes: ETP-ALL, TLX, proliferative, and TALLMO.3-5 Maturation arrest induces a pre-leukemic condition in which additional mutations can give rise to T-ALL.1,2 These secondary mutations are not necessarily clonal events and are often selected during disease progression or post-treatment haematologica | 2016; 101(9)

PTEN-AKT and NOTCH1-directed therapy

relapse.6,7 We therefore proposed that these mutations should be classified as type B aberrations.2 Type B mutations are prevalent among all T-ALL subtypes and affect a wide variety of cellular processes, including survival and proliferation, cell cycle progression, and epigenetic events. Type B mutations often affect signal transduction pathways, including the NOTCH1, IL7R-JAK-STAT, RASMEK-ERK, and PTEN-PI3K-AKT pathways. A growing body of evidence suggests that some of these signaling pathways are preferentially mutated in specific T-ALL subtypes, presumably due to the fact that developing T cells are dependent on these pathways in specific stages. For example, mutations in IL7 receptor (IL7R) and the downstream molecules JAK or RAS are prevalent among TLX and ETP-ALL patients.8-10 Although new therapeutic strategies that target oncogenic transcription factor complexes are emerging,11 several compounds that selectively inhibit altered signaling pathways are currently available. Thus, inhibiting signaling proteins such as NOTCH, IL7R, RAS and/or AKT may provide a promising new therapeutic approach for T-ALL. In this review, we describe the role of PTEN as a tumor suppressor and we discuss various PTEN-inactivating mechanisms observed in different human cancers and TALL. Besides PTEN inactivation, we describe other mechanisms that contribute to AKT activation and leukemogenesis. Finally, we discuss PTEN-AKT signaling in relation to future NOTCH1-directed therapies and provide a rationale for the use of AKT inhibitors in addition to current treatment protocols.

The PTEN tumor suppressor Mutations in the tumor suppressor gene PTEN (phosphatase and tensin homolog), which is located on chromo-

somal band 10q23, are very common in a wide range of cancers.12,13 The PTEN gene contains nine exons, and the encoded protein includes an N-terminal phosphatase domain, a central C2 lipid membrane-binding domain, and a C-terminal tail domain (Figure 1). PTEN is a phosphatase that dephosphorylates PIP3 [phosphatidylinositol (3,4,5)-triphosphate] to produce PIP2 [phosphatidylinositol (4,5)-bisphosphate], thereby opposing the function of PI3K (phosphatidylinositol 3-kinase). PI3K converts PIP2 into PIP3, which in turn activates key downstream kinases, including PDK1 and AKT (Figure 2). Thus, PTEN is an important negative regulator of PI3K-AKT signaling. Because AKT plays key roles in cellular metabolism, proliferation and survival, inactivation of PTEN by genetic aberrations drives survival and uncontrolled proliferation, ultimately leading to cancer.14 A recent study identified an alternate translation initiation site located upstream of the coding region of canonical PTEN that generates a larger form of PTEN.15 This isoform is known as PTENÎą and is described to be involved in mitochondrial energy metabolism.15

PTEN aberrations in cancer Heterozygous germline mutations in PTEN were identified initially in 60-80% of patients with a group of rare syndromes including Cowden syndrome, BannayanRiley-Ruvalcaba syndrome, and PTEN-related Proteus syndrome; these disorders are known collectively as PTEN hamartoma tumor syndrome (PHTS).16 With respect to sporadic (i.e., non-hereditary) tumors, heterozygous PTEN mutations occur in 50-80% of prostate, glioblastoma, and endometrial cancers and 30-50% of lung, colon, and breast cancers.17 Loss of both functional PTEN alleles is common among patients with prostate or breast cancer, as well as

Figure 1. Schematic representation of the human PTEN gene located on chromosome 10q23. The PTEN gene contains nine exons, and the PTEN protein contains several functional domains, including a phosphatase domain (dark gray) and a C2 lipid-binding domain (light gray). The positions of nonsense insertion and deletion mutations are indicated by closed triangles, and missense mutations are indicated by open triangles. Microdeletions and deletions in the PTEN gene are shown below the exons. The number of patients with each mutation/deletion in our cohort of T-ALL patients is indicated.21, 35

haematologica | 2016; 101(9)

1011

R.D. Mendes et al.

among those with melanoma or glioblastoma.18 The majority of these aberrations are caused by point mutations, small insertions, or deletions, all of which can occur throughout the entire PTEN gene. At the transcriptional and post-transcriptional levels, PTEN inactivation can occur via promoter methylation and through the expression of PTEN-directed microRNA.19 PTEN activity is also

regulated at the post-translational level: phosphorylation, ubiquitination, oxidation, and acetylation can regulate the phosphatase activity, subcellular localization, and degradation of PTEN.17 Defects in any of these processes may explain the absence of functional PTEN in cancer patients who apparently lack genetic aberrations in PTEN.20-22 Several ALL cases have been identified in which high lev-

Figure 2. Schematic overview of the upstream and downstream effectors of PTEN and associated molecular mechanisms that can activate AKT and lead to GSI resistance. (A) PTEN-PI3K-AKT mutations. (B) NOTCH mutations and AKT activation. (C) MYC signaling and AKT activation. (D) IL7R/IGF1R signaling and AKT activation. The molecules with activating and inactivating mutations are indicated in dark gray and light gray, respectively. Dashed lines/arrows represent processes that contribute to cellular GSI sensitivity, and that are frequently inactivated by inactivating mutations/rearrangements in PTEN or FBXW7. Solid lines/arrows represent GSI resistance mechanisms.

1012

haematologica | 2016; 101(9)

PTEN-AKT and NOTCH1-directed therapy

els of inactive PTEN are accompanied by an active PI3KAKT pathway.23,24 Although the majority of prevalent pathogenic mechanisms affect the loss of one or both PTEN alleles, subtle changes in PTEN protein levels can have a powerful effect on cancer susceptibility and/or tumor progression, as exemplified by the Pten hypomorphic mouse model.25 Thus, the level of functional PTEN affects tumor susceptibility, and PTEN function can be compromised at the DNA, mRNA, and/or protein levels.

PTEN aberrations in T-cell acute lymphoblastic leukemia PTEN deletions and mutations were initially identified in cell lines.26,27 Restoring PTEN levels in these cell lines decreased cell size and induced apoptosis by suppressing the PI3K-AKT pathway.28 Studies by others29-34 and our group21,35 revealed aberrations in the PTEN-PI3K-AKT pathway in approximately 23% of primary samples obtained from pediatric T-ALL patients. With respect to TALL subtypes, we have shown that PTEN aberrations are strongly associated with TAL- or LMO-rearranged leukemia in children21 and the same was observed in adult T-ALL cohorts.36 The vast majority of PTEN aberrations are nonsense mutations in exon 7 (which truncate the Cterminal domain) and deletions that affect nearly the entire locus (Figure 1). Although truncated PTEN proteins that lack the lipid-binding C-terminal domain retain their phosphatase activity, they are highly unstable and are degraded rapidly.37 In mice, truncated PTEN leads to decreased genomic stability and the development of multiple cancers.38 Recently, we reported that approximately 8% of T-ALL patients have a RAG-mediated microdeletion in the phosphatase domain that disrupts the reading frame (Figure 1).35 In addition, mutations have been identified in PI3K and AKT; specifically, 9% of pediatric T-ALL patients have a mutation in either the catalytic (PIK3CA) or regulatory (PIK3R1) subunit of PI3K, and 2% of patients have a mutation in AKT itself (Figure 2A).21,30 Many T-ALL patients with a heterozygous PTEN mutation also acquire a deletion21 or microdeletion35 in their remaining wild-type allele in leukemic subclones29 that may give rise to relapse. This phenomenon was demonstrated functionally by Clappier et al. who used an elegant human T-ALL xenograft transplantation model in mice and found the selection and preferential outgrowth of PTEN-inactivated leukemic cells.39 In line with this, heterozygous Pten knockout mice develop T-cell leukemia in which the remaining wild-type allele is frequently deleted.40-42 Other leukemogenic mechanisms that can inactivate PTEN at the protein level are the increased expression of casein kinase 2 (CK2) and the production of reactive oxygen species (ROS) that stabilize inactive forms of PTEN proteins and lead to impaired phosphatase activity.23,43 Collectively, these findings indicate the existence of ongoing pathogenic pressure to inactivate both PTEN alleles during disease progression, and the resulting loss of PTEN activity in turn activates the PI3K-AKT pathway.

Clinical implications We have reported that aberrations in PTEN represent a significant, independent risk factor for relapse in T-ALL patients treated using either the Dutch Childhood haematologica | 2016; 101(9)

Oncology Group or German Cooperative Study Group for Childhood ALL protocol.21,35 Similar results were reported for other cohorts of pediatric T-ALL patients treated using other protocols.32,33 In the Berlin-Frankfurt-Munster study, the presence of NOTCH1-activating mutations in addition to PTEN-inactivating mutations predicts for good outcome similar to that of patients harboring NOTCH1-activating mutations only,32 suggesting that NOTCH1-mutations can antagonize the unfavorable effect of PTEN aberrations. In a French Group for Research in Adult ALL study of T-ALL patients, those with aberrations in RAS and/or PTEN had a significantly worse outcome compared to patients without such mutations.36 This was not confirmed in the MRC UKALL2003 trial for pediatric T-ALL; RAS and/or PTEN aberrations also did not change the favorable outcome of patients with NOTCH1/FBWX7 mutations.44 Taken together, these findings suggest that PTEN aberrations may represent a general, poor prognostic factor in T-ALL.

NOTCH1 mutations lead to activation of AKT More than 65% of T-ALL patients have aberrant activation of the NOTCH1 pathway due to mutations in either the NOTCH1 gene itself or FBXW7, which encodes E3ubiquitin ligase.45,46 Thus, the NOTCH1 pathway may be an ideal target for therapeutic intervention. Furthermore, NOTCH1-directed therapies are clinically important, as they can also boost the cellular response to steroids.47,48 Gamma-secretase inhibitors (GSI), which inhibit the presenilin gamma-secretase complex, block the cleavage of NOTCH1 at its S3 site; this cleavage step is required to release the active, intracellular NOTCH1 domain (ICN1) upon ligand binding (Figure 2B). Several groups have applied GSI to cell lines derived from T-ALL patients with NOTCH1-activating mutations; although GSI treatment initially induces cell cycle arrest, the majority of cell lines adapt and ultimately stop responding to the treatment (i.e., develop GSI resistance).45,49 Nevertheless, GSI treatment effectively blocks gamma-secretase activity, resulting in reduced intracellular levels of the ICN1 domain and reduced expression of NOTCH1â&#x20AC;&#x2122;s target genes.29 GSI resistance is, therefore, caused by other mechanisms that circumvent NOTCH1 inhibition.50,51 Consistent with this notion, Palomero and co-workers found that decreased PTEN levels in cell lines are correlated with GSI resistance, and GSI-resistant lines have increased levels of activated AKT.29 Restoring the expression of functional PTEN in these GSI-resistant lines restored a GSI sensitivity response, whereas constitutively activated AKT or using shRNA to knock down PTEN expression provoked GSI resistance in a GSI-responsive line.29 This seminal study identified two important NOTCH1 downstream targets that regulate PTEN expression: HES1 and MYC. HES1 is a robust transcriptional repressor, whereas MYC is a weak transcriptional activator. Because the negative effect of HES1 prevails over the positive effect of MYC, PTEN expression is suppressed (Figure 2B).29 However, the resistance of leukemic cells to GSI resulted in disappointing results upon testing the GSI inhibitor MK-0752 in a clinical trial (DFCI-04-390).52 This trial was unsuccessful due to the compoundâ&#x20AC;&#x2122;s limited efficacy in leukemic cells and severe gastrointestinal toxicity. To overcome these issues, next-generation NOTCH1 inhibitors 1013

R.D. Mendes et al.

with reduced off-target toxicity are currently in development.53 For example, promising strategies include selectively blocking NOTCH1 using anti-NOTCH1 antibodies54,55 or chemically modified peptides that block the NOTCH transcriptional complex in the nucleus.56

PTEN is not linked a priori to resistance to gamma-secretase inhibitors in human T-cell acute lymphoblastic leukemia Despite the initial report by Palomero and co-workers,29 subsequent studies have not confirmed that loss of PTEN activity is intrinsically linked to GSI resistance.21,57,58 For example, GSI sensitivity was similar between NOTCH1driven T-cell leukemia cells obtained from wild-type mice and from PTEN knockout mice.58 However, PTEN deficiency does accelerate the disease progression of NOTCH1-driven leukemia.58 Using a different Pten knockout mouse model (Ptenflox/flox/Lck-Cre), Hagenbeek et al. found that PTEN-deficient thymocytes were just as sensitive to in vitro GSI treatment as were wild-type thymocytes.57 Moreover, several human T-ALL cell lines with mutant alleles of PTEN – different cell lines from those used by Palomero et al. – were actually sensitive to GSI.21 In diagnostic samples from patients with TALLMO, PTEN is frequently inactivated in the absence of NOTCH-activating mutations.21,32,36,59 Thus, mutations in PTEN and NOTCH1/FBXW7 are frequently independent genetic events and only co-occur in a small number of patients’ primary samples. In those patients who harbor both PTEN mutations and NOTCH1/FBXW7 mutations, the NOTCH1 mutations are usually weakly activating mutations. Because PTEN and NOTCH1 mutations are mostly independent genetic events in primary T-ALL, PTEN-deficient leukemic cells in T-ALL patients likely do not have intrinsic GSI resistance at disease presentation. Perhaps one way that PTEN-deficient T-ALL can be linked to GSI resistance is upon relapse, when NOTCH1-dependent leukemic cells may have lost PTEN activity, possibly due to clonal selection following treatment. However, there is currently no evidence to support this notion. The question remains, is it possible that immediately following PTEN loss, NOTCH1-dependent T-ALL becomes NOTCH1-independent and develops GSI-resistance? Recently, Adolfo Ferrando’s group addressed this intriguing question by generating an elegant mouse model of NOTCH1-induced T-ALL in which the Pten gene is deleted only in established tumors.60 Unlike previous Pten knockout models,57,58 deletion of Pten in this new model conferred strong resistance to dibenzazepine, a potent GSI. In this model,60 Pten loss activated expression of genes involved in cell metabolism, ribosomal RNA processing, and amino acid and nucleotide biosynthesis, genes that are normally suppressed following NOTCH1-inhibiting GSI treatment. Moreover, GSI treatment increased leukemic cells’ dependency on autophagy in order to recycle essential metabolites. Pten loss also relieved the GSIinstigated block of glycolysis and glutaminolysis, a phenotype that was copied by expressing the constitutively active myrisoylated AKT. Because both the GSI dibenzazepine and the glutaminase inhibitor BPTES [bis-2-(5phenylacetamido-1,2,4-thiadiazol-2-yl)ethyl sulfide 3] act in a synergistic fashion in inducing anti-leukemic effects, the authors proposed glutaminolysis as a major therapeu1014

tic target for treating NOTCH-activated T-ALL.60 Another unanswered question remains: why does the loss of Pten in an established NOTCH1-driven tumor cause GSI resistance,60 while NOTCH1-driven tumors that are generated in Pten knockout mice remain GSI-sensitive?57,58 The answer may lie in the ability of cells to adapt to PTEN loss by dampening PI3K-Akt signaling over time to a level that is still of advantage to leukemic cells. Unlike progressive reduction in PI3K-Akt signaling, the loss of PTEN may initially drive the rapid, high activation of Akt, resulting in cell proliferation, survival, and GSI resistance. This hypothesis predicts two consequences of GSI or other NOTCH1-inhibiting treatment. First, T-ALL patients who lack PTEN activity at disease onset may still respond to NOTCH1 inhibition. Second, NOTCH1 inhibition may trigger leukemic cells to acquire mutations such as PTEN deletions, which leads to activation of AKT and resistance to NOTCH1 inhibitors. Several key observations provide support for this hypothesis of reduced PI3K-AKT signaling over time in the absence of PTEN. For example, we found no difference in AKT phosphorylation between primary T-ALL patients with aberrant PTEN and patients without such mutations,21 indicating that patients with PTENdefective leukemia may have adapted and reduced PI3KAKT signaling. Reduced AKT activation may explain why primary PTEN-defective T-ALL cells are NOTCH1-dependent and remain GSI-sensitive.58 Although this hypothesis has not been tested formally, future NOTCH-inhibiting therapies may be more effective when combined with inhibitors of PI3K or AKT. Consistent with this, the PI3K/mTOR dual inhibitor PI-103 resulted in enhanced NOTCH-MYC activity in T-ALL cell lines.61 Additionally, T-ALL induced by retroviral insertional mutagenesis in wild-type or RasG12D-mutant mice demonstrated an initial response to PI3K-inhibitor GDC-0941 treatment.62 However, this treatment led to the survival and outgrowth of drug-resistant clones with active PI3K-AKT signaling that frequently had reduced Notch1 signaling.62 To avoid resistance when combined with NOTCH1 inhibitors, the authors propose a sequential treatment using a NOTCH1 inhibitor at diagnosis to eliminate NOTCH1 mutant clones followed by PI3K/AKT inhibitor treatment.62

Other mechanisms that can activate AKT and lead to resistance to gamma-secretase inhibitors Eighty-five percent of T-ALL patients have an activated AKT pathway accompanied by increased phosphorylation of AKT and its downstream targets GSK-3β and FOXO3a.23 Notably, this percentage is higher than the frequency of PTEN aberrations (present in 23% of patients) and, therefore, has to be explained by the activation of AKT through other mechanisms. MYC may provide an alternative mechanism to activate AKT either directly or indirectly (e.g. MYC activates the expression of mir-17-92 and mir-19, which target PTEN mRNA)63-65 (Figure 2C). Using an inducible MYC-dependent zebrafish T-ALL model, Gutierrez et al. found that established tumors regressed when MYC expression was turned off. This effect was circumvented by activating PI3K-AKT signaling,66 showing that AKT activation is an important downstream effector of MYC which may drive GSI resistance. Moreover, the MYC gene is an important haematologica | 2016; 101(9)

PTEN-AKT and NOTCH1-directed therapy

downstream target of NOTCH1, and T-ALL patients with activating mutations in NOTCH1 overexpress MYC.67,68 NOTCH binds a distal enhancer located far downstream of the MYC locus.69,70 This NOTCH-MYC enhancer region (N-Me) is duplicated in approximately 5% of T-ALL patients, acting as a “super-enhancer”.69 In another 6% of adult and childhood T-ALL patients, MYC is ectopically activated due to a MYC translocation; importantly, these patients usually do not have NOTCH1-activating mutations.71 MYC may also activate NOTCH1 via a positive feedback mechanism, as MYC suppresses the expression of miRNA-30, which targets the 3’ untranslated region of NOTCH1 (Figure 2C).72 Accordingly, treatment of T-ALL xenografted mice with the bromodomain protein inhibitor JQ1 results in decreased MYC levels and also reverses MYC-induced resistance to GSI.73,74 Furthermore, in human T-ALL cell lines, GSI-sensitive cells can be converted to being GSI-resistant by the ectopic expression of MYC.68,75 Under normal conditions, MYC is phosphorylated by the kinase GSK-3β; phosphorylated MYC is then subjected to ubiquitination by FBXW7 and proteasomemediated degradation (Figure 2C).76,77 Conversely, activated AKT can stabilize MYC protein by phosphorylating – and thereby inactivating – GSK-3β. These findings may explain the observation that MYC and PTEN are reciprocally expressed in T-ALL.78 Apart from enhancing cellular resistance to NOTCH1 inhibitors, MYC also enhances leukemia-initiating cell activity and worsens outcome in various mouse models of T-ALL. Mutant Fbxw7-R465C mice develop aggressive leukemias that acquire Notch1 mutations.79 Myc levels are stabilized in these mice, resulting in the expansion of leukemia cells that have enhanced self-renewal capacity and that express a stem cell-like expression profile.79 The Tal1/Lmo2 transgenic mouse model develops spontaneous T-cell tumors that also acquire Notch1 mutations. Because Myc is a Notch target, Notch inhibition led to reduced leukemia-initiating cell activity in these mice.80 Reducing endogenous Myc levels led to increased survival and reduced numbers of leukemia cells with leukemia-initiating cell potential in both models.79,81 Overall, these positive feedback loops between NOTCH, MYC, and AKT suggest that inhibitors of MYC or PI3K/AKT may help to prevent resistance to NOTCH1-inhibiting therapies,82 and also eliminate leukemia-initiating cell activity in T-ALL. Co-targeting the PI3K pathway and MYC remarkably enhanced the elimination of leukemia-initiating cells.83

References 1. Meijerink JP. Genetic rearrangements in relation to immunophenotype and outcome in T-cell acute lymphoblastic leukaemia. Best Pract Res Clin Haematol. 2010;23(3):307318. 2. Van Vlierberghe P, Pieters R, Beverloo HB, Meijerink JP. Molecular-genetic insights in paediatric T-cell acute lymphoblastic leukaemia. Br J Haematol. 2008;143(2):153168. 3. Homminga I, Pieters R, Langerak AW, et al. Integrated transcript and genome analyses reveal NKX2-1 and MEF2C as potential oncogenes in T cell acute lymphoblastic

haematologica | 2016; 101(9)

Another AKT activation mechanism is via the gene that encodes the IL7 receptor (IL7Ra), which also represents a direct target gene of NOTCH1.84,85 The IL7R gene is mutated in nearly 10% of T-ALL patients. These mutations cause the constitutive activation of STAT5 and AKT,8,86,87 and can provoke GSI resistance (Figure 2D). For instance, expression of the IL7Ra can overcome the effects of NOTCH1 inhibition on the cell cycle and survival, thereby contributing to resistance.84 Similar results were obtained by overexpressing IGF1R, which encodes insulin-like growth factor 1 receptor and is another NOTCH1 target (Figure 2D).88 In these cases, too, NOTCH-inhibiting therapies may be more effective when combined with AKT inhibitors. Furthermore, enhanced AKT activity may limit leukemia sensitivity to steroid treatment,89,90 one of the cornerstone drugs in the treatment of human T-ALL. AKT was shown to directly phosphorylate (S134) and inactivate the steroid receptor NR3C1.89 Combined steroid treatment with the dual PI3K-mTOR inhibitor BEZ23591 or the MK2206 AKT inhibitor89 sensitized AKT-activated leukemic cells to steroid treatment.

Conclusion As a potent tumor suppressor, PTEN is considered to be the principal negative regulator of PI3K-AKT signaling. Inactivation of PTEN indirectly activates PI3K-AKT signaling, causing the uncontrolled proliferation of thymocytes, ultimately leading to T-ALL. Regardless of PTEN, AKT can be over-activated by a variety of signaling molecules, including PI3K, AKT, MYC, IL7R and IGF1R (Figure 2). Initial activation of AKT causes resistance to NOTCH1inhibiting therapies. However, in the long-term, we suggest that AKT signaling may be dampened, thereby restoring responsiveness to NOTCH-inhibiting therapies. Overall, because AKT activation is central to a variety of leukemogenic mechanisms and crucial in the resistance to NOTCH1 inhibition, using AKT inhibitors in current treatment protocols may be a promising strategy to treat NOTCH1-mutated T-ALL. Acknowledgments RDM and KC-B were financed by the Children Cancer Free Foundation (Stichting Kinderen Kankervrij (KiKa 2008-29 and KiKa 2013-116). We thank EnglishEditingSolutions.com for editorial assistance.

leukemia. Cancer Cell. 2011;19(4):484-497. 4. Ferrando AA, Neuberg DS, Staunton J, et al. Gene expression signatures define novel oncogenic pathways in T cell acute lymphoblastic leukemia. Cancer Cell. 2002; 1(1):75-87. 5. Soulier J, Clappier E, Cayuela JM, et al. HOXA genes are included in genetic and biologic networks defining human acute Tcell leukemia (T-ALL). Blood. 2005;106 (1):274-286. 6. Notta F, Mullighan CG, Wang JC, et al. Evolution of human BCR-ABL1 lymphoblastic leukaemia-initiating cells. Nature. 2011;469(7330):362-367. 7. Anderson K, Lutz C, van Delft FW, et al. Genetic variegation of clonal architecture

10.

11.

and propagating cells in leukaemia. Nature. 2011;469(7330):356-361. Zenatti PP, Ribeiro D, Li W,et al. Oncogenic IL7R gain-of-function mutations in childhood T-cell acute lymphoblastic leukemia. Nat Genet. 2011;43(10):932-939. Zhang J, Ding L, Holmfeldt L, et al. The genetic basis of early T-cell precursor acute lymphoblastic leukaemia. Nature. 2012;481 (7380):157-163. Cante-Barrett K, Spijkers-Hagelstein JA, Buijs-Gladdines JG, et al. MEK and PI3KAKT inhibitors synergistically block activated IL7 receptor signaling in T-cell acute lymphoblastic leukemia. Leukemia. 2016 May 13 [Epub ahead of print]. Filippakopoulos P, Qi J, Picaud S, et al.

1015

R.D. Mendes et al.

12.

13.

14. 15.

16.

17. 18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

1016

Selective inhibition of BET bromodomains. Nature. 2010;468(7327):1067-1073. Li J, Yen C, Liaw D, et al. PTEN, a putative protein tyrosine phosphatase gene mutated in human brain, breast, and prostate cancer. Science. 1997;275(5308):1943-1947. Steck PA, Pershouse MA, Jasser SA, et al. Identification of a candidate tumour suppressor gene, MMAC1, at chromosome 10q23.3 that is mutated in multiple advanced cancers. Nat Genet. 1997;15(4): 356-362. Vogelstein B, Kinzler KW. Cancer genes and the pathways they control. Nat Med. 2004;10(8):789-799. Liang H, He S, Yang J, et al. PTENalpha, a PTEN isoform translated through alternative initiation, regulates mitochondrial function and energy metabolism. Cell Metab. 2014;19(5):836-848. Marsh DJ, Coulon V, Lunetta KL, et al. Mutation spectrum and genotype-phenotype analyses in Cowden disease and Bannayan-Zonana syndrome, two hamartoma syndromes with germline PTEN mutation. Hum Mol Genet. 1998;7(3):507-515. Salmena L, Carracedo A, Pandolfi PP. Tenets of PTEN tumor suppression. Cell. 2008;133(3):403-414. Song MS, Salmena L, Pandolfi PP. The functions and regulation of the PTEN tumour suppressor. Nat Rev Mol Cell Biol. 2012;13(5):283-296. Hollander MC, Blumenthal GM, Dennis PA. PTEN loss in the continuum of common cancers, rare syndromes and mouse models. Nat Rev Cancer. 2011;11(4):289-301. Leupin N, Cenni B, Novak U, et al. Disparate expression of the PTEN gene: a novel finding in B-cell chronic lymphocytic leukaemia (BCLL). Br J Haematol. 2003;121(1):97-100. Zuurbier L, Petricoin EF, Vuerhard MJ, et al. The significance of PTEN and AKT aberrations in pediatric T-cell acute lymphoblastic leukemia. Haematologica. 2012;97(9):14051413. Mutter GL, Lin MC, Fitzgerald JT, et al. Altered PTEN expression as a diagnostic marker for the earliest endometrial precancers. J Natl Cancer Inst. 2000;92(11):924-930. Silva A, Yunes JA, Cardoso BA, et al. PTEN posttranslational inactivation and hyperactivation of the PI3K/Akt pathway sustain primary T cell leukemia viability. J Clin Invest. 2008;118(11):3762-3774. Gomes AM, Soares MV, Ribeiro P, et al. Adult B-cell acute lymphoblastic leukemia cells display decreased PTEN activity and constitutive hyperactivation of PI3K/Akt pathway despite high PTEN protein levels. Haematologica. 2014;99(6):1062-1068. Alimonti A, Carracedo A, Clohessy JG, et al. Subtle variations in Pten dose determine cancer susceptibility. Nat Genet. 2010;42(5):454-458. Sakai A, Thieblemont C, Wellmann A, Jaffe ES, Raffeld M. PTEN gene alterations in lymphoid neoplasms. Blood. 1998;92(9):34103415. Shan X, Czar MJ, Bunnell SC, et al. Deficiency of PTEN in Jurkat T cells causes constitutive localization of Itk to the plasma membrane and hyperresponsiveness to CD3 stimulation. Mol Cell Biol. 2000;20(18): 6945-6957. Xu Z, Stokoe D, Kane LP, Weiss A. The inducible expression of the tumor suppressor gene PTEN promotes apoptosis and decreases cell size by inhibiting the PI3K/Akt pathway in Jurkat T cells. Cell Growth Differ. 2002;13(7):285-296. Palomero T, Sulis ML, Cortina M, et al.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

Mutational loss of PTEN induces resistance to NOTCH1 inhibition in T-cell leukemia. Nat Med. 2007;13(10):1203-1210. Gutierrez A, Sanda T, Grebliunaite R, et al. High frequency of PTEN, PI3K, and AKT abnormalities in T-cell acute lymphoblastic leukemia. Blood. 2009;114(3):647-650. Larson Gedman A, Chen Q, Kugel Desmoulin S, et al. The impact of NOTCH1, FBW7 and PTEN mutations on prognosis and downstream signaling in pediatric T-cell acute lymphoblastic leukemia: a report from the Children's Oncology Group. Leukemia. 2009;23(8):1417-1425. Bandapalli OR, Zimmermann M, Kox C, et al. NOTCH1 activation clinically antagonizes the unfavorable effect of PTEN inactivation in BFM-treated children with precursor T-cell acute lymphoblastic leukemia. Haematologica. 2013;98(6):928-936. Jotta PY, Ganazza MA, Silva A, et al. Negative prognostic impact of PTEN mutation in pediatric T-cell acute lymphoblastic leukemia. Leukemia. 2010;24(1):239-242. Remke M, Pfister S, Kox C, et al. High-resolution genomic profiling of childhood T-ALL reveals frequent copy-number alterations affecting the TGF-beta and PI3K-AKT pathways and deletions at 6q15-16.1 as a genomic marker for unfavorable early treatment response. Blood. 2009;114(5):1053-1062. Mendes RD, Sarmento LM, Cante-Barrett K, et al. PTEN microdeletions in T-cell acute lymphoblastic leukemia are caused by illegitimate RAG-mediated recombination events. Blood. 2014;124(4):567-578. Trinquand A, Tanguy-Schmidt A, Ben Abdelali R, et al. Toward a NOTCH1/FBXW7/RAS/PTEN-based oncogenetic risk classification of adult T-cell acute lymphoblastic leukemia: a Group for Research in Adult Acute Lymphoblastic Leukemia study. J Clin Oncol. 2013;31(34):4333-4342. Georgescu MM, Kirsch KH, Akagi T, Shishido T, Hanafusa H. The tumor-suppressor activity of PTEN is regulated by its carboxyl-terminal region. Proc Natl Acad Sci USA. 1999;96(18):10182-10187. Sun Z, Huang C, He J, et al. PTEN C-terminal deletion causes genomic instability and tumor development. Cell Rep. 2014;6(5):844-854. Clappier E, Gerby B, Sigaux F, et al. Clonal selection in xenografted human T cell acute lymphoblastic leukemia recapitulates gain of malignancy at relapse. J Exp Med. 2011;208(4):653-661. Di Cristofano A, Pesce B, Cordon-Cardo C, Pandolfi PP. Pten is essential for embryonic development and tumour suppression. Nat Genet. 1998;19(4):348-355. Suzuki A, de la Pompa JL, Stambolic V, et al. High cancer susceptibility and embryonic lethality associated with mutation of the PTEN tumor suppressor gene in mice. Curr Biol. 1998;8(21):1169-1178. Di Cristofano A, Kotsi P, Peng YF, CordonCardo C, Elkon KB, Pandolfi PP. Impaired Fas response and autoimmunity in Pten+/mice. Science. 1999;285(5436):2122-2125. Giambra V, Jenkins CR, Wang H, et al. NOTCH1 promotes T cell leukemia-initiating activity by RUNX-mediated regulation of PKC-theta and reactive oxygen species. Nat Med. 2012;18(11):1693-1698. Jenkinson S, Kirkwood AA, Goulden N, Vora A, Linch DC, Gale RE. Impact of PTEN abnormalities on outcome in pediatric patients with T-cell acute lymphoblastic leukemia treated on the MRC UKALL2003 trial. Leukemia. 2016;30(1):39-47.

45. Weng AP, Ferrando AA, Lee W, et al. Activating mutations of NOTCH1 in human T cell acute lymphoblastic leukemia. Science. 2004;306(5694):269-271. 46. Ferrando AA. The role of NOTCH1 signaling in T-ALL. Hematology Am Soc Hematol Educ Program. 2009:353-361. 47. Real PJ, Tosello V, Palomero T, et al. Gamma-secretase inhibitors reverse glucocorticoid resistance in T cell acute lymphoblastic leukemia. Nat Med. 2009;15(1): 50-58. 48. Samon JB, Castillo-Martin M, Hadler M, et al. Preclinical analysis of the gamma-secretase inhibitor PF-03084014 in combination with glucocorticoids in T-cell acute lymphoblastic leukemia. Mol Cancer Ther. 2012;11(7):1565-1575. 49. Lewis HD, Leveridge M, Strack PR, et al. Apoptosis in T cell acute lymphoblastic leukemia cells after cell cycle arrest induced by pharmacological inhibition of notch signaling. Chem Biol. 2007;14(2):209-219. 50. Hales EC, Taub JW, Matherly LH. New insights into Notch1 regulation of the PI3KAKT-mTOR1 signaling axis: targeted therapy of gamma-secretase inhibitor resistant Tcell acute lymphoblastic leukemia. Cell Signal. 2014;26(1):149-161. 51. Palomero T, Ferrando A. Oncogenic NOTCH1 control of MYC and PI3K: challenges and opportunities for anti-NOTCH1 therapy in T-cell acute lymphoblastic leukemias and lymphomas. Clin Cancer Res. 2008;14(17):5314-5317. 52. Deangelo DJ, Stone RM, Silverman LB, et al. A phase I clinical trial of the notch inhibitor MK-0752 in patients with T-cell acute lymphoblastic leukemia/lymphoma (T-ALL) and other leukemias. J Clin Oncol. 2006; 24(18_suppl):6585. 53. Tosello V, Ferrando AA. The NOTCH signaling pathway: role in the pathogenesis of T-cell acute lymphoblastic leukemia and implication for therapy. Ther Adv Hematol. 2013;4(3):199-210. 54. Aste-Amezaga M, Zhang N, Lineberger JE, et al. Characterization of Notch1 antibodies that inhibit signaling of both normal and mutated Notch1 receptors. PLoS One. 2010;5(2):e9094. 55. Wu Y, Cain-Hom C, Choy L, et al. Therapeutic antibody targeting of individual Notch receptors. Nature. 2010;464(7291): 1052-1057. 56. Moellering RE, Cornejo M, Davis TN, et al. Direct inhibition of the NOTCH transcription factor complex. Nature. 2009;462 (7270):182-188. 57. Hagenbeek TJ, Wu X, Choy L, et al. Murine Pten(-/-) T-ALL requires non-redundant PI3K/mTOR and DLL4/Notch1 signals for maintenance and gammac/TCR signals for thymic exit. Cancer Lett. 2014;346(2):237248. 58. Medyouf H, Gao X, Armstrong F, et al. Acute T-cell leukemias remain dependent on Notch signaling despite PTEN and INK4A/ARF loss. Blood. 2010;115(6):11751184. 59. Zuurbier L, Homminga I, Calvert V, et al. NOTCH1 and/or FBXW7 mutations predict for initial good prednisone response but not for improved outcome in pediatric T-cell acute lymphoblastic leukemia patients treated on DCOG or COALL protocols. Leukemia. 2010;24(12):2014-2022. 60. Herranz D, Ambesi-Impiombato A, Sudderth J, et al. Metabolic reprogramming induces resistance to anti-NOTCH1 therapies in T cell acute lymphoblastic leukemia. Nat Med. 2015;21(10):1182-1189.

haematologica | 2016; 101(9)

PTEN-AKT and NOTCH1-directed therapy 61. Shepherd C, Banerjee L, Cheung CW, et al. PI3K/mTOR inhibition upregulates NOTCH-MYC signalling leading to an impaired cytotoxic response. Leukemia. 2013;27(3):650-660. 62. Dail M, Wong J, Lawrence J, et al. Loss of oncogenic Notch1 with resistance to a PI3K inhibitor in T-cell leukaemia. Nature. 2014;513(7519):512-516. 63. Mu P, Han YC, Betel D, et al. Genetic dissection of the miR-17~92 cluster of microRNAs in Myc-induced B-cell lymphomas. Genes Dev. 2009;23(24):2806-2811. 64. Mavrakis KJ, Wolfe AL, Oricchio E, et al. Genome-wide RNA-mediated interference screen identifies miR-19 targets in Notchinduced T-cell acute lymphoblastic leukaemia. Nat Cell Biol. 2010;12(4):372379. 65. Olive V, Bennett MJ, Walker JC, et al. miR19 is a key oncogenic component of mir-1792. Genes Dev. 2009;23(24):2839-2849. 66. Gutierrez A, Grebliunaite R, Feng H, et al. Pten mediates Myc oncogene dependence in a conditional zebrafish model of T cell acute lymphoblastic leukemia. J Exp Med. 2011;208(8):1595-1603. 67. Palomero T, Lim WK, Odom DT, et al. NOTCH1 directly regulates c-MYC and activates a feed-forward-loop transcriptional network promoting leukemic cell growth. Proc Natl Acad Sci USA. 2006;103(48): 18261-18266. 68. Weng AP, Millholland JM, Yashiro-Ohtani Y, et al. c-Myc is an important direct target of Notch1 in T-cell acute lymphoblastic leukemia/lymphoma. Genes Dev. 2006;20(15): 2096-2109. 69. Herranz D, Ambesi-Impiombato A, Palomero T, et al. A NOTCH1-driven MYC enhancer promotes T cell development, transformation and acute lymphoblastic leukemia. Nat Med. 2014;20(10):1130-1137. 70. Yashiro-Ohtani Y, Wang H, Zang C, et al. Long-range enhancer activity determines Myc sensitivity to Notch inhibitors in T cell leukemia. Proc Natl Acad Sci USA. 2014;111(46):E4946-4953. 71. La Starza R, Borga C, Barba G, et al. Genetic

haematologica | 2016; 101(9)

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

profile of T-cell acute lymphoblastic leukemias with MYC translocations. Blood. 2014;124(24):3577-3582. Ortega M, Bhatnagar H, Lin AP, et al. A microRNA-mediated regulatory loop modulates NOTCH and MYC oncogenic signals in B- and T-cell malignancies. Leukemia. 2015;29(4):968-976. Delmore JE, Issa GC, Lemieux ME, et al. BET bromodomain inhibition as a therapeutic strategy to target c-Myc. Cell. 2011;146(6):904-917. Knoechel B, Roderick JE, Williamson KE, et al. An epigenetic mechanism of resistance to targeted therapy in T cell acute lymphoblastic leukemia. Nat Genet. 2014;46(4):364-370. Sharma VM, Calvo JA, Draheim KM, et al. Notch1 contributes to mouse T-cell leukemia by directly inducing the expression of c-myc. Mol Cell Biol. 2006;26(21):802231. Sears R, Nuckolls F, Haura E, Taya Y, Tamai K, Nevins JR. Multiple Ras-dependent phosphorylation pathways regulate Myc protein stability. Genes Dev. 2000;14(19):2501-2514. Cross DA, Alessi DR, Cohen P, Andjelkovich M, Hemmings BA. Inhibition of glycogen synthase kinase-3 by insulin mediated by protein kinase B. Nature. 1995;378(6559):785-789. Bonnet M, Loosveld M, Montpellier B, et al. Posttranscriptional deregulation of MYC via PTEN constitutes a major alternative pathway of MYC activation in T-cell acute lymphoblastic leukemia. Blood. 2011;117(24): 6650-6659. King B, Trimarchi T, Reavie L, et al. The ubiquitin ligase FBXW7 modulates leukemia-initiating cell activity by regulating MYC stability. Cell. 2013;153(7):1552-1566. Tatarek J, Cullion K, Ashworth T, Gerstein R, Aster JC, Kelliher MA. Notch1 inhibition targets the leukemia-initiating cells in a Tal1/Lmo2 mouse model of T-ALL. Blood. 2011;118(6):1579-1590. Roderick JE, Tesell J, Shultz LD, et al. c-Myc inhibition prevents leukemia initiation in mice and impairs the growth of relapsed and induction failure pediatric T-ALL cells.

Blood. 2014;123(7):1040-1050. 82. Roti G, Stegmaier K. New approaches to target T-ALL. Front Oncol. 2014;4:170. 83. Schubbert S, Cardenas A, Chen H, et al. Targeting the MYC and PI3K pathways eliminates leukemia-initiating cells in T-cell acute lymphoblastic leukemia. Cancer Res. 2014;74(23):7048-7059. 84. Gonzalez-Garcia S, Garcia-Peydro M, Martin-Gayo E, et al. CSL-MAML-dependent Notch1 signaling controls T lineagespecific IL-7R{alpha} gene expression in early human thymopoiesis and leukemia. J Exp Med. 2009;206(4):779-791. 85. Ribeiro D, Melao A, Barata JT. IL-7R-mediated signaling in T-cell acute lymphoblastic leukemia. Adv Biol Regul. 2013;53(2):211222. 86. Barata JT, Cardoso AA, Boussiotis VA. Interleukin-7 in T-cell acute lymphoblastic leukemia: an extrinsic factor supporting leukemogenesis? Leuk Lymphoma. 2005;46 (4):483-495. 87. Shochat C, Tal N, Bandapalli OR, et al. Gainof-function mutations in interleukin-7 receptor-alpha (IL7R) in childhood acute lymphoblastic leukemias. J Exp Med. 2011;208(5):901-908. 88. Medyouf H, Gusscott S, Wang H, et al. High-level IGF1R expression is required for leukemia-initiating cell activity in T-ALL and is supported by Notch signaling. J Exp Med. 2011;208(9):1809-1822. 89. Piovan E, Yu J, Tosello V, et al. Direct reversal of glucocorticoid resistance by AKT inhibition in acute lymphoblastic leukemia. Cancer Cell. 2013;24(6):766-776. 90. Blackburn JS, Liu S, Wilder JL, et al. Clonal evolution enhances leukemia-propagating cell frequency in T cell acute lymphoblastic leukemia through Akt/mTORC1 pathway activation. Cancer Cell. 2014;25(3): 366378. 91. Hall CP, Reynolds CP, Kang MH. Modulation of glucocorticoid resistance in pediatric T-cell acute lymphoblastic leukemia by increasing BIM expression with the PI3K/mTOR inhibitor BEZ235. Clin Cancer Res. 2016;22(3):621-632.

1017

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Red Cell Biology & Its Disorders

Ferrata Storti Foundation

Haematologica 2016 Volume 101(9):1018-1027

Severe Ankyrin-R deficiency results in impaired surface retention and lysosomal degradation of RhAG in human erythroblasts Timothy J. Satchwell,1,2* Amanda J. Bell,1* Bethan R. Hawley, 1,2 Stephanie Pellegrin, 1,2 Kathryn E. Mordue,1,2 Cees Th. B. M. van Deursen,3 Nicole Heitink-ter Braak,3 Gerwin Huls,4 Mathie P.G Leers,5 Eline Overwater,6 Rienk Y. J. Tamminga,7 Bert van der Zwaag,8 Elisa Fermo,9 Paola Bianchi,9 Richard van Wijk,10 and Ashley M. Toye1,2

School of Biochemistry, University of Bristol, UK; 2National Institute for Health Research Blood and Transplant Research Unit (NIHR BTRU) in Red Blood Cell Products, University of Bristol, UK; 3Department of Internal Medicine, Zuyderland Hospital, Heerlen-Sittard, The Netherlands; 4Department of Hematology, University Medical Center Groningen, The Netherlands; 5Department of Clinical Chemistry and Hematology, Atrium Medical Center Parkstad, Heerlen, The Netherlands; 6Department of Clinical Genetics, VU University Medical Center, and Department of Clinical Genetics, Academic Medical Center, Amsterdam, The Netherlands; 7Department of Pediatric Hematology, Beatrix Childrens Hospital, University Medical Center Groningen, The Netherlands; 8Department of Medical Genetics, University Medical Center Utrecht, The Netherlands; 9 Oncohematology Unit - Physiopathology of Anemias Unit, Foundation IRCCS Cà Granda, Ospedale Maggiore Policlinico, Milan, Italy; and 10Department of Clinical Chemistry and Haematology, Laboratory for Red Blood Cell Research, University Medical Center Utrecht, The Netherlands 1

*TJS and AJB contributed equally to this work.

ABSTRACT

Correspondence: ash.m.toye@bristol.ac.uk/ t.satchwell@bristol.ac.uk

Received: March 14, 2016. Accepted: May 25, 2016. Pre-published: May 31, 2016. doi:10.3324/haematol.2016.146209

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1118

1018

nkyrin-R provides a key link between band 3 and the spectrin cytoskeleton that helps to maintain the highly specialized erythrocyte biconcave shape. Ankyrin deficiency results in fragile spherocytic erythrocytes with reduced band 3 and protein 4.2 expression. We use in vitro differentiation of erythroblasts transduced with shRNAs targeting ANK1 to generate erythroblasts and reticulocytes with a novel ankyrin-R ‘near null’ human phenotype with less than 5% of normal ankyrin expression. Using this model, we demonstrate that absence of ankyrin negatively impacts the reticulocyte expression of a variety of proteins, including band 3, glycophorin A, spectrin, adducin and, more strikingly, protein 4.2, CD44, CD47 and Rh/RhAG. Loss of band 3, which fails to form tetrameric complexes in the absence of ankyrin, alongside GPA, occurs due to reduced retention within the reticulocyte membrane during erythroblast enucleation. However, loss of RhAG is temporally and mechanistically distinct, occurring predominantly as a result of instability at the plasma membrane and lysosomal degradation prior to enucleation. Loss of Rh/RhAG was identified as common to erythrocytes with naturally occurring ankyrin deficiency and demonstrated to occur prior to enucleation in cultures of erythroblasts from a hereditary spherocytosis patient with severe ankyrin deficiency but not in those exhibiting milder reductions in expression. The identification of prominently reduced surface expression of Rh/RhAG in combination with direct evaluation of ankyrin expression using flow cytometry provides an efficient and rapid approach for the categorization of hereditary spherocytosis arising from ankyrin deficiency. Introduction The ankyrins are a family of structural adaptor proteins that link membrane proteins to the cytoskeleton in a diverse range of cell types where they play a critical role in the formation and organization of physiologically important membrane proteins within specialized membrane microdomains.1 Canonical ankyrins have a modular haematologica | 2016; 101(9)

Ankyrin-R prevents RhAG degradation in erythroblasts

domain organization with an N-terminal membrane binding domain containing 24 ANK repeats (1-24), followed by a central spectrin binding domain, a death domain and a Cterminal regulatory domain.2 Erythrocyte ankyrin (or ankyrin-R), alongside protein 4.2, links the tetrameric band 3-based multiprotein complex to the underlying spectrin cytoskeleton via associations with both the D2 and D3-4 regions of the ankyrin membrane binding domain.3 Ankyrin-R can also associate with other erythrocyte membrane proteins, including CD44,4 Na+K+ATPase5 and RhAG.6 RhAG is a 50kDa glycosylated NH3 and CO2 membrane transport protein7,8 that together with the Rh proteins RhCe and RhD forms the core of the Rh subcomplex that also contains CD47, LW and GPB.9 Evidence supporting the interaction between RhAG and ankyrin includes a drop in RhAG expression in the nb/nb mouse with ankyrin deficiency, an interaction of the C-terminal tail of RhAG and the D2 subdomain of the ankyrin membrane binding domain in vitro, and the presence of the RhAG mutation D399A, shown to abrogate direct ankyrin binding in vitro in a patient with Rhnull syndrome.6 In humans, mutations in the ankyrin-R gene are responsible for 50% of hereditary spherocytosis (HS) cases,10,11 a form of hemolytic anemia affecting 1 in 2000 people of Northern European ancestry.12 Although mutations in the ANK1 gene that result in hereditary spherocytosis are relatively prevalent, such mutations are most commonly heterozygous in nature and encode an array of complex alterations that include substitutions within specific binding sites and the generation of novel spliced or truncated protein products in addition to partial protein deficiency. This heterogeneity complicates the identification of secondary protein deficiencies that result from reduced ankyrin expression in human red blood cells and the corresponding insight into the role that this protein plays in the generation of the unique erythrocyte membrane cytoskeletal architecture. Reduced expression of the ankyrin binding proteins band 3 and protein 4.2 has been noted in these patients,13-15 and in murine models of ankyrin deficient HS, secondary protein deficiencies of GPA, protein 4.2 and Rh, but not band 3 or GPC, were observed.16 However, more extensive characterization of the membrane protein composition in humans, including the effect on RhAG and other putative ankyrin associated proteins, is so far lacking. A more detailed assessment of the importance and role that ankyrin-R protein expression plays in directing the development and composition of the human erythrocyte membrane is complicated by the absence of a naturally occurring null phenotype for this protein. In this study, we employ lentiviral shRNA transduction of primary erythroid cells and demonstrate the ability to deplete ankyrin-R expression to a level considerably lower than that observed in vivo.17 These cells were used to address the influence that ankyrin deficiency exerts on the membrane composition of the erythrocyte and developing erythroblast membrane, and to gain mechanistic insight regarding the loss of dependent proteins in its absence. The observed reduction in the plasma membrane expression of the core Rh complex components Rh and RhAG was confirmed on red blood cells from patients with HS due to mutations in ANK1. Furthermore, we propose a flow cytometry based strategy for easy identification of HS erythrocytes with ankyrin deficiency that could be used to direct diagnoses of HS molecular genetic defects. haematologica | 2016; 101(9)

Methods Donor and patient blood Platelet apheresis waste blood (NHSBT, Bristol, UK) from healthy donors or peripheral blood from patients previously diagnosed with HS because of a mutation in ANK1 were obtained with written informed consent for research use in accordance with the Declaration of Helsinki and approved by Bristol Research Ethics Committee Centre reference 12/SW/0199, as previously described.18

Erythroid cell culture and lentiviral transduction Briefly, CD34+ cells were purified from peripheral blood mononuclear cells using Magnetic Activated Cell Sorting (Miltenyi Biotec) following the manufacturer’s instructions, cultured for three days in Stemspan (Stem Cell Technologies) supplemented with SCF (100 ng/mL), Epo (2 U/mL, Bristol Royal Infirmary, UK), dexamethasone (1 μM, Sigma), IGF-1 (40 ng/mL, R&D Systems), cholesterol-rich lipids (40 μg/mL, Sigma) and 1 ng/mL IL-3 (R&D Systems): Phase 1. After three days, 50 μL concentrated lentivirus was added per 1x106 cells in expansion medium - IMDMExp (Phase 2), as previously described,18 supplemented with 8 μg/mL polybrene. After 20 h, the cells were washed 3 times in PBS and plated in fresh IMDMExp. Erythroblasts were further expanded and differentiated (Phase 3), as previously described.18 Lentiviral transduction of erythroblasts was performed as previously described using pLKO.1 vector with puromycin selection. Broad Institute Repository shRNA sequences TRCN0000083720 (shRNA 1) and TRCN0000083721 (shRNA 2) targeting ANK1 were used in preliminary experiments and TRCN0000083720 used unless otherwise stated.

Flow cytometry and FACS Flow cytometry and FACS on cultured erythroid cells and erythrocytes was performed as previously described.18 For assessment of ankyrin levels by flow cytometry, donor and patient erythrocytes were fixed with 1% paraformaldehyde + 0.0075% glutaraldehyde in PBS supplemented with 1 mg/mL BSA (PBSAG), 2 mg/mL glucose for 10 min at room temperature, permeabilized with 0.05% Triton X-100 for 2 min, centrifuged for 3 min at 500 G, washed once in PBSAG, re-suspended in PBSAG containing 4% BSA, and stained sequentially with BRIC272 (anti-ankyrin) and APC-conjugated polyclonal anti-mouse IgG (Biolegend), as described.18 Pronase epitope recovery experiments were performed as previously described.19 A list of antibodies used in this study is provided in the Online Supplementary Table S1. Data were acquired on a MACSQuant flow cytometer and analyzed using FlowJo v.7.6.5 (FlowJo).

Blue Native PAGE Detergent extraction using 0.5% C12E8 in 5 mM phosphate buffer containing 2 mM PMSF and 1% protease inhibitor cocktail (Calbiochem) was performed on 2x106 T120 erythroblasts per sample for 10 min on ice. Samples were centrifuged at 150,000 G for 30 min at 4°C and protein complexes from the extract separated using the Novex Native PAGE BisTris Blue Native polyacrylamide gel electrophoresis system (Thermo Fisher Scientific). Samples were run on 4%-16% gels with final sample concentration of 0.0625% Coomassie G and light blue cathode buffer-and transferred to PVDF membrane according to the manufacturer’s instructions.

Inhibition of lysosomal degradation A combination of 10 µg/mL leupeptin, 10 µg/mL pepstatin A, 10 μg/mL E64d or DMSO alone was added to non-targeting con1019

T.J. Satchwell et al.

Figure 1. shRNA mediated knockdown of ankyrin results in a novel reticulocyte phenotype with almost complete absence of ankyrin expression accompanied by widespread secondary protein deficiencies. (A) Purified populations of ankyrin knockdown and control reticulocytes were obtained from in vitro cultures by FACS sorting of the Hoechst negative population. 5 x 105 reticulocytes were lysed, proteins separated by SDS-PAGE and immunoblotted with mouse monoclonal antibodies specific for ankyrin (BRIC274), β spectrin (BRAC65), α spectrin (BRIC172), α adducin (Santa Cruz), band 3 (BRIC170) protein 4.2 (BRIC273) or rabbit antibodies against RhAG, CD47 or GPC (in house). Data are representative of three independent experiments. (B) Table showing ankyrin knockdown reticulocyte mean surface expression of indicated proteins relative to reticulocytes derived from erythroblasts expressing a non-targeting control shRNA. Means (+ standard error) are derived from 3 independent cultures using average flow cytometry mean fluorescent intensities (MFI) of reticulocyte populations from three time points per culture.

trol or ankyrin knockdown basophilic erythroblasts (T24) for 24 h in culture, then fixed for immunofluorescence or counted and processed for SDS PAGE and immunoblotting.

Immunofluorescence Cells were fixed for 15 min in 1% PFA + 0.0075% glutaraldehyde in PBSAG, allowed to adhere to 1% PEI coated coverslips, then labeled and imaged as previously described19 using a Leica SP5 AOBS confocal laser scanning microscope with 100x oilimmersion objective (NA 1.4).

Results Ankyrin is required for normal expression of multiple erythroid membrane and cytoskeletal proteins To understand the essential role ankyrin plays in the assembly and maintenance of the human erythrocyte membrane, ankyrin-R was knocked down using shRNA lentivirally tranduced into CD34+ cells isolated from healthy donor peripheral blood and then these cells were expanded and differentiated in vitro. Ankyrin knockdown of up to 95% was achieved and maintained in reticulocytes differentiated from erythroblasts expressing two independent shRNAs (Online Supplementary Figure S1). Analysis of protein expression by immunoblotting of total cell lysates derived from equal numbers of sorted ankyrin knockdown and control reticulocytes revealed reductions in the expression of cytoskeletal proteins spectrin and adducin, the membrane proteins CD47, CD44 and band 3 and strikingly large reductions in expression of protein 4.2 and RhAG (Figure 1A). Reductions in the reticulocyte expression of band 3 and GPA, and to a greater extent CD47, Rh, RhAG and CD44, were also observed by flow cytometry (Figure 1B). 1020

Ankyrin is required for membrane protein retention during human erythroblast enucleation Mislocalization of protein during enucleation has previously been suggested to account for protein loss based on confocal immunofluorescent imaging of nb/nb mouse erythroblasts undergoing enucleation.20 By staining differentiating ankyrin deficient erythroblasts grown in our human in vitro culture system with Hoechst and using a recently published flow cytometry gating strategy,18 protein partitioning profiles were determined for localization of specific membrane proteins within the plasma membrane of the reticulocyte and pyrenocyte (the extruded condensed nucleus encased by thin layer of cytoplasm and plasma membrane) derived from non-targeting control and ankyrin knockdown erythroblasts, respectively. The bar graphs in Figure 2A demonstrate that in the ankyrin deficient cultures there is an increase in the proportion of the ankyrin-R associated complex proteins band 3, GPA, RhAG, Rh and CD47 partitioning to the plasma membrane surrounding the extruded nucleus at this stage of differentiation. Interestingly, reduced reticulocyte plasma membrane retention was also observed for the junctional complex component GPC, as well as CD99, indicating that the effects of severe ankyrin deficiency extend beyond established ankyrin associated proteins. To confirm the mis-sorting of band 3 and GPA to the pyrenocyte during enucleation, sorted populations of reticulocytes and pyrenocytes were lysed and separated by SDS-PAGE. Figure 2B confirms the increased amount of GPA and band 3 in the pyrenocytes extruded from the ankyrin deficient erythroblasts relative to those from control erythroblasts. The remaining cytoskeletal adaptor protein 4.2 (the expression of which is reduced with severe depletion of ankyrin expression) is retained within the reticulocyte haematologica | 2016; 101(9)

Ankyrin-R prevents RhAG degradation in erythroblasts

Figure 2. Mislocalization to the pyrenocyte plasma membrane during enucleation accounts for reduced reticulocyte expression of band 3/GPA but not Rh/RhAG in the absence of ankyrin. (A) Graphical representation of protein partitioning profiles between plasma membranes of reticulocytes and extruded nuclei (pyrenocytes) of indicated proteins from control and ankyrin knockdown cultures during enucleation. Partitioning data are presented as percentages derived from mean fluorescent intensities for each antibody on the pyrenocyte and reticulocyte populations divided by the total intensity for the two populations % nucleus = [MFI nucleus/(MFI nucleus + MFI reticulocyte)]x 100 and % reticulocyte = [MFI reticulocyte/(MFI reticulocyte + MFI nucleus)]x100. (B) 5x105 reticulocytes and extruded nuclei from control and ankyrin shRNA transduced cultures collected by FACS were lysed (with addition of 1 ÎźL Omnicleave for nuclei populations), separated by SDS-PAGE and immunoblotted with antibodies specific for indicated proteins. (C) Bar chart illustrating differences in gross levels of indicated proteins on pyrenocytes extruded from control or ankyrin knockdown erythroblasts. Fold change = MFI nuclei from ankyrin knockdown/MFI nuclei from control. In all cases means (+ standard error) are derived from 3 independent cultures using average mean fluorescent intensities (MFI) of populations from three time points per culture.

membrane and the membrane protein flotillin maintains the partitioning profile observed in healthy donor21 and non-targeting shRNA expressing cells. Similar experiments conducted using protein 4.2 knockdown cells did not result in reduced reticulocyte expression of Rh, RhAG or CD44 (data not shown), demonstrating that these effects are not an indirect consequence of the secondary protein 4.2 expression deficiency that occurs in response to ankyrin knockdown. Importantly, although severe ankyrin-R deficiency results in an increase in the proportion of Rh and RhAG lost with the extruded pyrenocyte during enucleation as a percentage of total levels, absolute levels of these two proteins on the pyrenocytes derived from ankyrin knockdown erythroblasts are only minimally increased relative to control (Figure 2C). This observation highlights the fact that, although impaired reticulocyte retention of RhAG during erythroblast enucleation does occur in response to severe depletion of ankyrin, the majority of RhAG protein loss due to ankyrin-R deficiency must occur prior to enucleation.

Rhesus protein complex deficiency as a result of a loss of ankyrin occurs prior to erythroblast enucleation To investigate the stage at which RhAG and other proteins are lost prior to enucleation in response to severe haematologica | 2016; 101(9)

depletion of ankyrin-R expression, erythroblasts derived from CD34+ cells transduced with non-targeting or ankyrin-R targeting shRNA were expanded and differentiated in vitro and protein expression examined at 24-h intervals by immunoblotting and flow cytometry. Representative cytospin images charting the morphological differentiation of control and ankyrin deficient erythroblasts are shown in Online Supplementary Figure S2. Figure 3A demonstrates that in the almost complete absence of ankyrin-R the cytoskeletal adaptor protein 4.2, which associates with band 3 early during erythropoiesis19 and is rapidly degraded in its absence, is strikingly reduced in its expression relative to non-targeting control vector transduced erythroblasts from the onset of terminal differentiation. Surface expression of both band 3 and GPA was maintained at normal levels until the enucleation stage (Figure 3B). In contrast, a relative reduction in the expression of the Rh complex proteins Rh, RhAG and CD47 is observed from an early stage of differentiation, which in the case of Rh/RhAG becomes more pronounced as these proteins accumulate at the plasma membrane in the control erythroblasts. The expression profile and levels of GPC in erythroblasts transduced with ankyrin targeting shRNA was similar to control, although a reduction in the reticulocytes was observed by flow cytometry. Interestingly, the 1021

T.J. Satchwell et al. A

Figure 3. Severe depletion of ankyrin impacts expression of protein 4.2, Rh and RhAG prior to enucleation. (A) 5x105 differentiating erythroblasts expressing scramble control shRNA (Control) or ankyrin shRNA (Ankyrin KD) were removed from culture at 24-h intervals throughout differentiation, lysed and proteins separated by SDS-PAGE. Lysates were immunoblotted with antibodies specific for ankyrin (BRIC274), protein 4.2 (BRIC273), band 3 (BRIC170) or rabbit antibodies against RhAG (in house) and GAPDH (Santa Cruz). (B) Surface expression of indicated proteins was monitored on differentiating erythroblasts expressing non-targeting control or ANK1 shRNA at 24-h intervals using the indicated monoclonal antibodies as detailed in the Online Supplementary Table S1. The dashed box highlights the switch from gating on erythroblasts to reticulocytes post enucleation. All data shown are representative of three independent experiments.

loss of ankyrin also caused a pronounced reduction in the steady state levels of transferrin receptor throughout terminal differentiation.

Rh/RhAG expression levels are prominently reduced in a panel of hereditary spherocytosis patients with confirmed ankyrin mutations To determine whether the secondary membrane protein deficiencies identified in our in vitro derived ankyrin near null model are also observed in naturally occurring ankyrin deficient HS patients, we obtained a blood sample from a variety of patients with confirmed or suspected ankyrin mutations/deficiency. Although mutations in the gene encoding ankyrin-R (ANK1) are known to be responsible for HS in approximately 50% of cases, direct quantification of ankyrin deficiencies in the erythrocytes of these patients is notoriously laborious.17 To facilitate the assessment of the level of ankyrin deficiency in patient erythrocytes, we developed a protocol for the detection of ankyrin by flow cytometry using the monoclonal antibody BRIC272. The epitope that BRIC272 recognizes on ankyrin-R is not known but we have shown that it does not recognize the ANK90 binding domain (data not shown). Figure 4A and Online Supplementary Figure S3 demonstrate the variability in ankyrin expression within the healthy donor population as assessed using this technique which is consistent with that previously reported by radioactive immunoassay.17 This figure also shows expression of ankyrin relative to mean healthy donor controls in a random selection of patients available to us, and provides quantitative data that defines several mild and two severe deficiencies within this cohort. Expression of major membrane proteins was assessed in these patients (Figure 4B). Note that despite the fact that ANK1 mutations (where known â&#x20AC;&#x201C; additional details on the mutations are provided in the Online Supplementary 1022

Appendix) affect different regions of the ankyrin protein, a consistent significant reduction was observed in the surface expression of Rh and RhAG, as identified in our in vitro generated ankyrin knockdowns. This Rh/RhAG alteration is consistently prominent compared to the smaller reductions in expression of other proteins that reside within the ankyrin associated multiprotein complex including band 3 and GPA. No significant difference in the degree of RhAG reduction was observed between patients with mild and severe ankyrin deficiencies. In several patients, reduced expression of the ankyrin binding protein CD44 was also observed, although in others expression of this protein was decreased to only a mild degree. In contrast, erythrocytes from HS patients with differing clinical severity resulting from defined mutations in the genes encoding band 3 and protein 4.2 display increased or unaltered CD44 expression, respectively22-25 (Online Supplementary Table S2). To determine whether loss of Rh/RhAG also occurs prior to enucleation in naturally occurring ankyrin deficient patients, peripheral blood samples were obtained from a small number of ankyrin-R deficient HS patients with a range of clinical severities. Erythroblasts were expanded and differentiated in order to produce a population containing nucleated orthochromatic erythroblasts and reticulocytes. Consistent with the ankyrin-R knockdown results, Figure 4C demonstrates a substantially reduced surface expression of Rh/RhAG in orthochromatic erythroblasts of the more severely ankyrin deficient patient (approx. 43% normal ankyrin-R) prior to enucleation that is enhanced to only a minor degree by additional loss during enucleation. Surprisingly, despite obvious reductions in the expression of RhAG in the erythrocytes provided from mildly ankyrin deficient HS patients (Figure 4B), reductions in expression of this protein were not observed in the orthochromatic erythroblasts or (and to haematologica | 2016; 101(9)

Ankyrin-R prevents RhAG degradation in erythroblasts

Figure 4. Reduced erythrocyte surface expression of Rh/RhAG is common to hereditary spherocytosis patients with ankyrin deficiency. Erythrocytes isolated from peripheral blood of healthy donors and patients diagnosed with hereditary spherocytosis arising from ankyrin mutations or deficiency were fixed and permeabilized where necessary, as detailed in Materials and Methods. Expression of (A) ankyrin or (B) indicated proteins was measured by flow cytometry and expressed relative to healthy donor control erythrocytes treated in the same way for each experiment. Data are presented as mean fluorescent intensity relative to healthy donor control for each antibody and patient. (C) Erythroblasts derived from peripheral blood samples of severe and mildly ankyrin deficient hereditary spherocytosis patients were differentiated to generate orthochromatic erythroblasts and reticulocytes. Surface expression of indicated proteins in orthochromatic erythroblast and reticulocyte populations across three consecutive days were assessed by flow cytometry and mean data presented as mean fluorescent intensity relative to healthy donor controls cultured in parallel.

only a mild degree) in cultured reticulocytes from these patients.

Severe ankyrin-R deficiency abrogates tetrameric band 3 macrocomplex formation Ankyrin can simultaneously bind to two dimers of band 3 via two distinct and co-operative sites located in the D2 and D3-4 domains forming the band 3 tetramer based core for a major membrane cytoskeletal multiprotein complex. We hypothesized that the dramatic reduction in surface expression of Rh complex components but not band 3 under conditions of severe ankyrin deficiency prior to enucleation indicated a requirement for band 3 tetramer formation for stable surface expression of these proteins that would be disrupted in the absence of ankyrin. Using Blue Native PAGE electrophoresis of C12E8 detergent extracts of healthy donor erythroblasts, we show for the first time using this approach that it is possible to distinguish two separate complexes containing band 3 (Figure 5A). Based on co-migration, the upper complex likely contains band 3, ankyrin, protein 4.2 and RhAG, and has a molecular weight consistent with the classical ankyrin associated band 3 tetrameric multiprotein complex. The second band 3, containing complex of lower molecular weight, likely represents the dimeric population of band 3. This populahaematologica | 2016; 101(9)

tion may be partially comprised of band 3 dissociated from tetrameric complexes during detergent extraction; however, immunoblotting with a GLUT1 specific antibody identified this protein to be present specifically within the region of the gel corresponding with dimeric band 3 and completely absent from that overlapping with the ankyrin associated tetrameric band 3 complex. This represents a novel demonstration that, despite its reported association with band 3,26 GLUT1 does not reside within the ankyrin associated complex of proteins in healthy donor erythroblast membranes. Figure 5B shows that detergent extracts from non-targeting control shRNA transduced erythroblasts have both the higher molecular weight (tetrameric) and smaller complexes of band 3 (two representative immunoblots shown for band 3), whereas in the severe ankyrin knockdown erythroblasts extracts only the lower molecular weight band 3 complex is detected.

RhAG loss in the absence of ankyrin occurs via lysosomal degradation Reduced expression of RhAG in the absence or severe depletion of ankyrin could be accounted for by impaired delivery or a reduced stability at the plasma membrane. To explore the molecular mechanism that underlies reduced RhAG expression in the absence of ankyrin, 1023

T.J. Satchwell et al.

pronase surface epitope recovery experiments were performed to assess the delivery efficiency of RhAG. Figure 6A shows that, following pronase treatment of control and ankyrin shRNA transduced erythroblasts, reappearance of the LA1818 epitope after a 1-h recovery period occurs to a similar degree in both control and ankyrin knockdown erythroblasts, indicating an equivalent level of delivery of RhAG to the plasma membrane in the presence and absence of ankyrin. Contrastingly, over a longer 2-h recovery time period, RhAG surface levels are reduced in the erythroblasts lacking ankyrin relative to control erythroblasts, suggesting an altered stability within the plasma membrane in the absence of ankyrin-R. In contrast, reappearance of band 3 BRIC6 epitopes following pronase treatment is increased after 1 h in ankyrin knockdown erythroblasts but not significantly different to control after 2h recovery, possibly indicating a more rapid delivery of band 3 to the plasma membrane in the absence of ankyrin mediated band 3 tetramers. Confocal imaging of immunofluorescently labeled RhAG in ankyrin knockdown erythroblasts identified weak plasma membrane labeling of RhAG together with isolated intracellular clusters of RhAG labeling that co-localize with the lysosomal protein LAMP1, indicating localization of RhAG in a lysosomal compartment, which also appears enlarged/aggregated in these cells (Figure 6B). To confirm that reduced RhAG expression in ankyrin deficient erythroblasts results from lysosomal degradation of destabilized plasma membrane RhAG, cells were treated with a combination of lysosomal hydrolase inhibitors leupeptin, pepstatin A and E64d. Figures 6C and D show that treatment of ankyrin knockdown erythroblasts with this combination of lysosomal inhibitors results in an accumulation of RhAG that can be observed by immunofluorescence and immunoblotting of total cell lysates, demonstrating that loss of ankyrin causes lysosomal degradation of RhAG due to instability at the plasma membrane. Of note, we were unable to inhibit degradation of protein 4.2 in ankyrin knockdown erythroblasts using either lysosomal (leupeptin, pepstatin A, E64d), proteosomal (MG132) or calpain (calpain inhibitor I) inhibitors (data not shown). Protein 4.2 appears to be rapidly degraded in the absence of ankyrin, but in the presence of band 3 by an as yet undefined mechanism.

Discussion Using shRNA mediated knockdown in primary erythroblasts prior to terminal erythroid differentiation we have demonstrated the ability to deplete ankyrin-R expression by up to 95% generating an effective novel near null reticulocyte phenotype. As would be predicted, the reduction of ankyrin expression predominantly impacts the surface expression of membrane proteins assigned to the band 3 tetramer based erythrocyte multiprotein complex. The greatest impact of ankyrin deficiency is not on band 3 itself, however, but on the stability of the Rhesus protein sub complex (RhAG, Rh in particular), whether mediated by shRNA or by natural ankyrin genetic mutations. Importantly, reduced surface expression of the core Rh subcomplex components Rh and RhAG precedes the loss of the more extensively studied ankyrin binding protein band 3. We also demonstrate that absence of ankyrin during erythropoiesis abrogates the ability of band 3 to form the larger of two band 3-containing macro1024

Figure 5. Ankyrin depletion disrupts formation of band 3 tetramers. C12E8 detergent extracts of healthy donor (A) or cells expressing non-targeting control or ankyrin shRNAs (B) removed from culture after 120 h of differentiation were separated by Blue Native PAGE. Protein complexes were immunoblotted using mouse monoclonal antibodies specific for ankyrin (BRIC274), protein 4.2 (BRIC273), or rabbit antibodies against band 3, RhAG or GLUT1 (in house). Two independent examples of separation of the two band 3 complex populations are shown in Figure 4B. Molecular weights are indicated in kDa.

complexes (likely the band 3 tetrameric complex) detected by Blue Native PAGE. Protein 4.2, which fails to express at any stage of erythropoiesis in the absence of band 3 (but in the presence of ankyrin),18 is severely depleted in the absence of ankyrin at a stage of erythroid differentiation at which band 3 surface expression is unaltered, highlighting a reliance of protein 4.2 on the ankyrin mediated tetrameric organization specifically of band 3 for its stable expression. Interestingly severe depletion of ankyrin also impacts upon the reticulocyte expression of α/β spectrin and α adducin, highlighting the importance of this protein for stable maintenance of the wider erythroid cytoskeletal architecture. A previous study that employed immunofluorescent staining of enucleating ankyrin deficient nb/nb murine erythroblasts attributed reduced expression of band 3, GPA and RhAG on the erythrocytes from this mouse to reduced retention of these proteins within the reticulocyte membrane during enucleation.20 The data presented here illustrate that, whilst band 3 and GPA do display increased partitioning to the pyrenocyte plasma membrane during haematologica | 2016; 101(9)

Ankyrin-R prevents RhAG degradation in erythroblasts

Figure 6. Ankyrin depletion results in lysosomal degradation of RhAG in differentiating erythroblasts. Differentiating erythroblasts (at T72) were removed from culture, washed and re-suspended in 500 μg/mL pronase for 30 min at 37°C. Cells were washed in PBS and transferred into pre-warmed differentiation media. Reappearance of LA1818 and BRIC6 epitopes was assessed at indicated times. (A) Histograms showing mean reappearance of LA1818 and BRIC6 epitopes in ankyrin knockdown erythroblasts as assessed by flow cytometry after 1 h and 2 h recovery as a % of control = (MFI ankyrin knockdown/MFI control)x 100. Error bars represent standard error of the mean derived from data from 4 independent cultures at T72. Statistical significance was assessed using Student's t-test. (B) Differentiating erythroblasts at T48 from control and ankyrin knockdown cultures were fixed, immunolabeled with antibodies specific for RhAG (LA1818) and LAMP-1 and stained with DAPI. Images were acquired using a Leica SP5 AOBS confocal laser scanning microscope using a 100x oil-immersion objective (N.A 1.4). (C) Differentiating basophilic erythroblasts from control and ankyrin knockdown cultures were treated with a combination of lysosomal inhibitors as outlined in the Methods section. After 24 h treatment, 5x105 cells were lysed, proteins separated by SDS PAGE and immunoblotted with antibodies specific for ankyrin (BRIC274), RhAG (in house) and GAPDH. Immunoblots are representative of three independent experiments. (D) Differentiating erythroblasts from control and ankyrin knockdown cultures treated with lysosomal inhibitors were fixed, immunolabeled with antibodies specific for RhAG (LA1818) and LAMP-1 and stained with DAPI. Images were acquired using a Leica SP5 AOBS confocal laser scanning microscope using a 100x oil-immersion objective (N.A 1.4).

human erythroblast enucleation, accounting for their reduced reticulocyte expression in the absence of ankyrin, the contribution of this mechanism to reduced reticulocyte RhAG expression is only minor. Intriguingly, we demonstrate that loss of the bulk of the core Rh subcomplex components, Rh and RhAG in the absence of ankyrin is temporally distinct from that of band 3/GPA, and in the case of RhAG, occurs via a distinct mechanism involving lysosomal degradation of unstable protein internalized post delivery to the plasma membrane during erythroblast differentiation. This mechanism of RhAG turnover likely also accounts for the loss of RhAG reported in Rhnull phenotype resulting from the p.(Asp399Ala) RhAG mutation within the putative ankyrin binding site,6 and is also consistent with the surprising stability of RhAG in the absence of band 3 during the early stages of terminal differentiation.18 Interestingly, ankyrin ‘near null’ reticulocytes exhibit a slightly greater reduction in the surface expression of Rh than that of RhAG. A population of RhAG is known to be capable of existing within the erythrocyte membrane independently of Rh.6 One possible explanation for this discrepancy in the levels of the Rh core components is that it is accounted for by a population of RhAG that is independent of both Rh and ankyrin, and achieves membrane stability via interaction with nonhaematologica | 2016; 101(9)

tetrameric complex band 3. Importantly, Rh/RhAG but not band 3/GPA were also found to be dramatically decreased in the pre-enucleation orthochromatic erythroblasts of a naturally occurring HS patient with a 57% deficiency in ankyrin expression, demonstrating the relevance of observations made using our ‘near null’ model to understanding secondary protein loss in ankyrin deficient HS patients. The presence of RhAG at close to wild-type levels on orthochromatic erythroblasts and reticulocytes of mildly ankyrin deficient HS patients in spite of reduced erythrocyte expression is surprising. This observation highlights the difficulty in uncovering the basis of secondary protein loss in patients with missense ANK1 mutations that may retain misfolded or truncated protein products with potentially unpredictable effects, and suggests that additional protein loss must occur during reticulocyte remodeling and membrane shedding within the circulation of these patients. The fact that mild ankyrin deficiency appears to be well tolerated (at least in the early stages of erythroid differentiation), suggests that, initially, ankyrin exists in excess of that required for band 3 tetramer and Rh subcomplex stabilization. Nevertheless, the observation that the contribution of different mechanisms of secondary protein loss (pre and post enucleation) differs according to the degree of ankyrin deficiency in 1025

T.J. Satchwell et al.

Figure 7. Summary model for impaired complex assembly and loss of proteins in the absence of ankyrin during erythropoiesis. Under normal conditions, band 3 chaperoned by GPA, associates with protein 4.2 in an early intracellular compartment. Ankyrin mediates formation of band 3 tetramers which in turn stabilize plasma membrane expression of the core Rh proteins, Rh and RhAG leading to the formation of a band 3 â&#x20AC;&#x2DC;macrocomplexâ&#x20AC;&#x2122;. In the absence of ankyrin, band 3 fails to form a tetrameric complex, protein 4.2 is destabilized and degraded, RhAG, lacking the stabilization provided by association with tetrameric band 3 and/or ankyrin is rapidly internalized post delivery to the plasma membrane and degraded via the lysosomes. Dimeric band 3 maintains its association with GPA and is expressed at normal levels within the plasma membrane for the majority of erythroid differentiation. During erythroblast enucleation, gross disruptions in cytoskeletal architecture allied to impaired band 3 connectivity to this structure in the absence of ankyrin result in increased partitioning of band 3/GPA to the pyrenocyte plasma membrane and reduced reticulocyte expression of these proteins.

erythroid cells highlights important features for consideration of the role of ankyrin in membrane protein stability and remodeling at multiple stages in the ultimate development of a mature erythrocyte. These data expand our understanding of the assembly of multiprotein complexes that occurs as part of erythroid membrane biogenesis during normal erythropoiesis, reveal the role of the cytoskeletal adaptor protein ankyrinR, and describe the mechanism by which different associated proteins are lost during erythropoiesis. We propose the model summarized in Figure 7 for the assembly of the ankyrin associated multiprotein complex in which plasma membrane expression of Rh/RhAG is reliant upon stabilization by ankyrin. This likely occurs both indirectly through the formation of band 3 tetramers (further supported by the inability of the band 3 membrane domain to rescue reticulocyte expression of RhAG in an otherwise band 3 null environment)18 and through low affinity direct association of the C-terminus of RhAG with ankyrin, as previously suggested.6 Absence or severe reduction of ankyrin-R abrogates the formation of band 3 tetramers, resulting in expression of dimeric band 3 only, which results in the loss of protein 4.2 and the rapid internalization and lysosomal degradation of RhAG in differentiating erythroblasts post initial delivery to the plasma membrane. Dimeric band 3 and associated GPA are retained within the plasma membrane throughout the initial stages of erythropoiesis, but display increased partitioning to the pyrenocyte plasma membrane in the absence of cytoskeletal association via ankyrin. Reduced reticulocyte expression of other proteins reported here without established 1026

associations with ankyrin likely results both from reduced cytoskeletal association and increased loss of plasma membrane during enucleation and subsequent remodeling brought about by gross disruption of membrane cytoskeletal integrity in the absence of ankyrin. Finally, mutations in the ANK1 gene are responsible for approximately 50% of HS cases in Europe and North America; however, current methods for identification and quantification of ankyrin deficiency are laborious and time consuming.17 In addition to the identification of consistently prominent reductions in the expression of the core Rh subcomplex proteins Rh and RhAG across a range of HS patients with ANK1 mutations, we report the application of a monoclonal anti-ankyrin antibody BRIC272 for the quantification of ankyrin expression using intracellular flow cytometry of permeabilized erythrocytes. We, therefore, propose that monitoring expression of Rh/RhAG and ankyrin levels directly by flow cytometry in patientâ&#x20AC;&#x2122;s erythrocytes provides a convenient screen for rapid preliminary identification and categorization of HS arising from mutations in ANK1. Acknowledgments The authors thank Dr. Andrew Herman and the University of Bristol Flow Cytometry Facility for assistance in performing fluorescence activated cell sorting, Dr. Paul Curnow (School of Biochemistry) for advice regarding Blue Native PAGE, Professor George Banting (School of Biochemistry) for provision of the CD99 monoclonal antibody (12E7) and Dr Rosey Mushens (Bristol Institute of Transfusion Medicine) for monoclonal antibodies. haematologica | 2016; 101(9)

Ankyrin-R prevents RhAG degradation in erythroblasts

Funding This work was supported by a Wellcome Trust project grant (094277) for TJS and a Wellcome Trust PhD studentship for AJB, KEM was supported by a BBSRC PhD DTG Case with Bristol Institute of Transfusion Sciences. This work was also funded in part by the National Institute for Health Research Blood and Transplant Research Unit (NIHR BTRU) in Red

References 1. Bennett V, Healy J. Membrane domains based on ankyrin and spectrin associated with cell-cell interactions. Cold Spring Harb Perspect Biology. 2009;1(6):a003012. 2. Bennett V, Healy J. Organizing the fluid membrane bilayer: diseases linked to spectrin and ankyrin. Trends Mol Med. 2008; 14(1):28-36. 3. Michaely P, Bennett V. The ANK repeats of erythrocyte ankyrin form two distinct but cooperative binding sites for the erythrocyte anion exchanger. J Biol Chem. 1995; 270(37):22050-22057. 4. Kalomiris EL, Bourguignon LY. Mouse T lymphoma cells contain a transmembrane glycoprotein (GP85) that binds ankyrin. J Cell Biol. 1988;106(2):319-327. 5. Nelson WJ, Veshnock PJ. Ankyrin binding to (Na+ + K+)ATPase and implications for the organization of membrane domains in polarized cells. Nature. 1987; 328(6130):533-536. 6. Nicolas V, Le Van Kim C, Gane P, et al. RhRhAG/ankyrin-R, a new interaction site between the membrane bilayer and the red cell skeleton, is impaired by Rh(null)-associated mutation. J Biol Chem. 2003; 278(28):25526-25533. 7. Endeward V, Cartron JP, Ripoche P, Gros G. RhAG protein of the Rhesus complex is a CO2 channel in the human red cell membrane. FASEB J. 2008;22(1):64-73. 8. Geyer RR, Parker MD, Toye AM, Boron WF, Musa-Aziz R. Relative CO(2)/NH(3) permeabilities of human RhAG, RhBG and RhCG. J Membr Biol. 2013;246(12):915926. 9. Cartron JP. RH blood group system and molecular basis of Rh-deficiency. Baillieres

haematologica | 2016; 101(9)

10. 11.

12. 13.

14.

15.

16.

17.

18.

Blood Cell Products at the University of Bristol in partnership with NHS Blood and Transplant (NHSBT, AMT), and the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement 602121 (CoMMiTMenT-project, RvW). The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, the Department of Health or NHSBT.

Best Pract Res Clin Haematol. 1999; 12(4):655-689. Tse WT, Lux SE. Red blood cell membrane disorders. Br J Haematol. 1999;104(1):2-13. Eber SW, Gonzalez JM, Lux ML, et al. Ankyrin-1 mutations are a major cause of dominant and recessive hereditary spherocytosis. Nature Genet. 1996;13(2):214-218. Perrotta S, Gallagher PG, Mohandas N. Hereditary spherocytosis. Lancet. 2008; 372(9647):1411-1426. Jarolim P, Rubin HL, Brabec V, Palek J. A nonsense mutation 1669Glu-->Ter within the regulatory domain of human erythroid ankyrin leads to a selective deficiency of the major ankyrin isoform (band 2.1) and a phenotype of autosomal dominant hereditary spherocytosis. J Clin Invest. 1995; 95(3):941-947. Iolascon A, Miraglia del Giudice E, Camaschella C, et al. Ankyrin deficiency in dominant hereditary spherocytosis: report of three cases. Br J Haematol. 1991; 78(4):551-554. Lanciotti M, Perutelli P, Valetto A, Di Martino D, Mori PG. Ankyrin deficiency is the most common defect in dominant and non dominant hereditary spherocytosis. Haematologica. 1997;82(4):460-462. Rank G, Sutton R, Marshall V, et al. Novel roles for erythroid Ankyrin-1 revealed through an ENU-induced null mouse mutant. Blood. 2009;113(14):3352-3362. Savvides P, Shalev O, John KM, Lux SE. Combined spectrin and ankyrin deficiency is common in autosomal dominant hereditary spherocytosis. Blood. 1993;82(10): 2953-2960. Satchwell TJ, Hawley BR, Bell AJ, Ribeiro ML, Toye AM. The cytoskeletal binding domain of band 3 is required for multipro-

19.

20.

21.

22.

23.

24.

25.

26.

tein complex formation and retention during erythropoiesis. Haematologica. 2015; 100(1):133-142. Satchwell TJ, Bell AJ, Pellegrin S, et al. Critical band 3 multiprotein complex interactions establish early during human erythropoiesis. Blood. 2011;118(1):182-191. Salomao M, Chen K, Villalobos J, Mohandas N, An X, Chasis JA. Hereditary spherocytosis and hereditary elliptocytosis: aberrant protein sorting during erythroblast enucleation. Blood. 2010;116(2):267-269. Bell AJ, Satchwell TJ, Heesom KJ, et al. Protein distribution during human erythroblast enucleation in vitro. PloS one. 2013;8(4):e60300. Bruce LJ, Beckmann R, Ribeiro ML, et al. A band 3-based macrocomplex of integral and peripheral proteins in the RBC membrane. Blood. 2003;101(10):4180-4188. Toye AM, Williamson RC, Khanfar M, et al. Band 3 Courcouronnes (Ser667Phe): a trafficking mutant differentially rescued by wild-type band 3 and glycophorin A. Blood. 2008;111(11):5380-5389. Bruce LJ, Ghosh S, King MJ, et al. Absence of CD47 in protein 4.2-deficient hereditary spherocytosis in man: an interaction between the Rh complex and the band 3 complex. Blood. 2002;100(5):1878-1885. van den Akker E, Satchwell TJ, Pellegrin S, et al. Investigating the key membrane protein changes during in vitro erythropoiesis of protein 4.2 (-) cells (mutations Chartres 1 and 2). Haematologica. 2010;95(8):12781286. Jiang W, Ding Y, Su Y, Jiang M, Hu X, Zhang Z. Interaction of glucose transporter 1 with anion exchanger 1 in vitro. Biochem Biophys Res Commun. 2006;339(4):12551261.

1027

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Red Cell Biology & Its Disorders

Ferrata Storti Foundation

Altered heme-mediated modulation of dendritic cell function in sickle cell alloimmunization

Emmanuelle Godefroy,1 Yunfeng Liu, 1 Patricia Shi,2 W. Beau Mitchell,3 Devin Cohen,4 Stella T. Chou,4 Deepa Manwani, 5 and Karina Yazdanbakhsh 1

Laboratory of Complement Biology, New York Blood Center, NY; 2Medical Services, New York Blood Center, NY; 3Laboratory of Platelet Biology, New York Blood Center, NY; 4 Children’s Hospital of Philadelphia, PA; and 5Division of Pediatric Hematology/Oncology - Children's Hospital at Montefiore, New York, NY, USA 1

Haematologica 2016 Volume 101(9):1028-1038

ABSTRACT

Correspondence: kyazdanbakhsh@nybc.org

Received: March 30, 2016. Accepted: May 19, 2016. Pre-published: May 26, 2016.

ransfusions are the main treatment for patients with sickle cell disease. However, alloimmunization remains a major life-threatening complication for these patients, but the mechanism underlying pathogenesis of alloimmunization is not known. Given the chronic hemolytic state characteristic of sickle cell disease, resulting in release of free heme and activation of inflammatory cascades, we tested the hypothesis that anti-inflammatory response to heme is compromised in alloimmunized sickle patients, increasing their risk of alloimmunization. Heme-exposed monocyte-derived dendritic cells from both non-alloimmunized sickle patients and healthy donors inhibited priming of proinflammatory CD4+ type 1 T cells, and exhibited significantly reduced levels of the maturation marker CD83. In contrast, in alloimmunized patients, heme did not reverse priming of pro-inflammatory CD4+ cells by monocyte-derived dendritic cells or their maturation. Furthermore, heme dampened NF-κB activation in non-alloimmunized, but not in alloimmunized monocyte-derived dendritic cells. Heme-mediated CD83 inhibition depended on Toll-like receptor 4 but not heme oxygenase 1. These data suggest that extracellular heme limits CD83 expression on dendritic cells in non-alloimmunized sickle patients through a Toll-like receptor 4-mediated pathway, involving NF-κB, resulting in dampening of pro-inflammatory responses, but that in alloimmunized patients this pathway is defective. This opens up the possibility of developing new therapeutic strategies to prevent sickle cell alloimmunization.

doi:10.3324/haematol.2016.147181

Introduction Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1028

1028

Sickle cell disease (SCD) results from a mutation in the β-globin gene causing hemoglobin to polymerize when deoxygenated to form rigid polymers within red blood cells (RBC), which leads to complications including chronic hemolytic anemia.1 Transfusion therapy remains an important treatment modality for patients with SCD. Despite its therapeutic benefits, 20%-60% patients with SCD develop alloantibodies with specificities against disparate antigens on transfused RBC, causing complications ranging from life-threatening hemolytic transfusion reactions, to logistical problems in finding compatible RBC for transfusion.2 The immunological basis for SCD alloimmunization remains ill-defined. Consistent with the importance of CD4+ helper T cells (TH) in driving B-cell responses, several studies have identified altered TH cell phenotypes and/or activity in alloimmunized patients with SCD.3-7 Given the ongoing hemolysis in SCD,8 we had previously investigated the effects of RBC breakdown product heme on immune responses of patients, with and without alloantibodies, undergoing chronic transfusion therapy, and found altered anti-inflammatory response to exogenous heme by monocytes from alloimmunized patients with SCD, resulting in a T-cell profile with heightened prohaematologica | 2016; 101(9)

Hemin alters dendritic cell function

inflammatory (TH1), but lower anti-inflammatory (TREG) T-cell subsets.9 These data suggested aberrant innate immune control of T-cell polarization in SCD alloimmunization, although the exact nature of the innate immune cell type or underlying molecular mechanism for these alterations remains elusive. Dendritic cells (DCs) are key antigen presenting cells in initiating/shaping T-cell immune responses.10 During an inflammatory response, they can be activated/matured by toll-like receptor (TLR) ligands. Once activated, they migrate to the lymphoid organs to activate/prime naïve T cells into effector cells.11 The DC maturation process which is key to initiate T-cell responses, involves upregulation of co-stimulation molecules, e.g. CD80, CD86, and de novo expression of CD83, as well as cytokine secretion.12 In response to a homolog of heme, TLR-matured human monocyte derived DCs (moDCs), in a non-SCD setting, were shown to display less immunogenic properties, including lower expression of DC maturation markers and proinflammatory cytokines than untreated DCs.13 Although this has not yet been tested, less immunogenic DCs are likely to dampen proinflammatory T-cell polarization profiles, thereby reducing the risk of mounting immune responses, including humoral responses. In this study, we tested the hypothesis that, in response to exogenous heme, DCs differentially shape T-cell polarization toward pro-inflammatory (TH1) phenotype in alloimmunized compared to non-alloimmunized SCD patients.

Methods Human samples All studies were approved by the Institutional Review Boards of the New York Blood Center (NYBC), the Children’s Hospital of Philadelphia, and the Montefiore Medical Center. De-identified fresh leukocyte-enriched products were obtained from NYBC’s healthy donors. For SCD patient samples, blood was obtained solely from discarded apheresis waste bags collected during erythrocytapheresis procedures from patients aged 15-34 years on chronic red cell exchange therapy (every 3-4 weeks for at least 2 years using leuko-depleted units, phenotype-matched for the C, E and K red cell antigens; see Online Supplementary Appendix).

T-cell priming and DC analysis Monocyte-derived DCs (moDCs) were prepared from peripheral blood mononuclear cells (PBMC) (see Online Supplementary Appendix). CFSE labeled purified (5x104) naïve (CD45RA+) CD4+ T cells from healthy donors were added to allogeneic moDCs (derived from SCD patients or healthy donors; 5x103). T-cell priming toward TH1, TH2 and TH17 detected after ten days was defined as the frequency of CFSElow (divided) CD4+ cells expressing IFNg, IL-4 and IL-17, respectively. Surface expression of CD80, CD83 and CD86 was determined at day 2 after maturation (or not). Intracellular cytokine expression (IL-12p40 and TNFα) was assessed after overnight stimulation in the presence of brefeldin A while IL12p70 ELISA was performed on day 2 supernatants, and NF-κB transcription factors in nuclear fractions were analyzed using TransAM technology (Active Motif), as described in the Online Supplementary Appendix.

Statistical analysis For most experiments, each patient data point is an average of at least 2-5 independent experiments using samples collected on different occasions from the same patient. Due to biological varihaematologica | 2016; 101(9)

ability between individuals, we performed a paired t-test comparing before and after heme treatment in each individual. P≤0.05 was considered statistically significant. [Data are represented as mean values±standard error of the mean (SEM) where indicated].

Results Hemin-treated moDCs down-regulate priming of IFNg-producing CD4+ T cells in non-alloimmunized, but not in alloimmunized SCD patients Because of ongoing hemolysis in SCD, we studied the effect of exogenously added hemin, as a surrogate for hemoglobin breakdown product,8,14-18 on moDCs in driving naïve T-cell differentiation, known as T-cell priming. Three different, but classical, DC maturation agents were used (LPS, LPS+ IFNg or R848) and the effects of heme exposure during DC maturation on priming of allogeneic naïve T cells from healthy donors were studied. Baseline (without hemin) frequency of primed IFNg-producing CD4+ cells (TH1) by moDCs from healthy donors and SCD patients were comparable (P>0.1). Hemin treatment (5 μM or 20 μM) resulted in significant inhibition of TH1 priming by moDCs from healthy donors as well as non-alloimmunized SCD patients irrespective of the DC maturation method used: approximately 25% less IFNg-producing cells were primed by DCs treated with 20 μM hemin as compared to untreated cells (n>7, P<0.002) (Figure 1A-D). In contrast, hemin had no effect on TH1 priming by moDCs from alloimmunized SCD patients: IFNg was not down-regulated in CD4+ T cells primed by hemin-exposed moDCs from alloimmunized patients using any of the DC maturation methods (P>0.21, n=9) (Figure 1B-D). Similarly, TNFα- (TH1) producing CD4+ T cells (Online Supplementary Figure S1A) were significantly lower when primed with heme-exposed DCs from healthy donor and non-alloimmunized SCD patients but not from the alloimmunized group. IL-4 production was not affected by hemin-treated moDCs in this system (Online Supplementary Figure S1B). Only low levels of IL-17A were detected, precluding the assessment of the effects of hemin on these cells (Online Supplementary Figure S1C). These results show that while hemin dampens the ability of moDCs to prime TH1 cells in non-alloimmunized SCD patients, it has little or no effect in alloimmunized patients.

Hemin down-regulates CD83 on moDCs derived from non-alloimmunized, but not from alloimmunized SCD patients To investigate the mechanism of altered heme-regulation of immunogenicity in alloimmunized SCD patients, we first assessed the effect of hemin on the expression of co-stimulation molecules CD80 (Online Supplementary Figure S2A) and CD86 (Online Supplementary Figure S2B) on moDCs. Both molecules were up-regulated upon maturation, but neither was affected by hemin irrespective of whether moDCs were derived from healthy donors or SCD patients, (Online Supplementary Figure S2A and B). We next tested the expression of CD83 maturation marker before and after hemin treatment (Figure 2). As expected, immature DCs expressed only very low CD83 levels (Figure 2A and B) which were up-regulated upon maturation, albeit to the same extent in both healthy donor and patient groups in the absence of heme (Figure 2A, C-E). Upon hemin treatment, CD83 expression was inhibited 1029

E. Godefroy et al.

Figure 1. Production of IFNg by CD4+ T cells primed with hemin-treated human monocyte derived dendritic cells (moDCs). CFSE-stained naïve CD4+ T cells (50x103/well) were cultured with allogeneic mature moDCs (5x103/well) pre-treated or not with hemin (0, 5, or 20 μM) in the presence of 5 UI/mL IL-2. MoDCs had been matured using 100 ng/mL lipopolysaccharide (LPS) (A and B), 100 ng/mL LPS + 500 UI/mL IFNg (C), or 5 μM R848 (D). Twelve days later, cells were stimulated with PMA/ionomycin for 5 h in the presence of brefeldin A and stained for CD4 and IFNg. (A) Representative example for non-alloimmunized patients. (B-D) The amount of cytokine-producing T cell is represented as a percentage in CFSElow/CD4+ cells. The middle data point refers to 5 μM hemin and the 3rd data point to 20 μM hemin. Two-tailed paired Student’s t-tests were used to compare the production of cytokines by T cells stimulated with hemin-treated moDCs versus untreated moDCs. *P<0.05.

on moDCs from healthy donors (P<0.02) and non-alloimmunized patients (P<0.025). In contrast, heme had no effect on CD83 expression levels on moDCs from alloimmunized SCD patients regardless of maturation method (P>0.25) (Figure 2C-E). These data indicate that downregulation of CD83 maturation marker in response to heme is defective in alloimmunized SCD patients.

IL-12 production is comparable in hemin-treated moDCs derived from alloimmunized and non-alloimmunized SCD patients We next investigated whether altered cytokine expression was involved in the differential effects of heme on moDC immunogenicity in the 2 patient groups. Because IL-12 is crucial for skewing towards TH1 IFNg-secreting cells,19 we first examined the effect of hemin on IL-12p40 production by moDCs. As expected, immature DCs did not produce IL-12p40 without or with heme treatment (Figure 3A and B). Hemin inhibited IL-12p40 production by moDCs from healthy donors matured with LPS, LPS+ IFNg and R848 (P<0.04 for 20 μM hemin, n=7) (Figure 3A and B). Similarly, IL-12p40 was down-regulated in response to hemin in moDCs not only from non-alloimmunized SCD patients, but also from alloimmunized SCD patients (Figure 3A and B). We also tested the levels of bioactive IL-12p70, which is composed of the 2 subunits IL-12p35 and IL-12-p40. Immature moDCs produced no IL-12p70. As previously reported in human settings,20,21 moDCs matured with LPS alone produced significantly less IL-12p70 relative to LPS+ 1030

IFNg and R848-matured DCs (Figure 3C; note scale difference). Overall, levels of IL-12p70 were significantly higher in matured moDCs from patients with SCD as compared to healthy donors (Figure 3C). As with IL-12p40, IL-12p70 was inhibited by hemin in moDCs from SCD patients regardless of alloimmunization status (LPS+ IFNg: non-alloimmunized, P=0.07, and alloimmunized, P=0.004; R848: non-alloimmunized, P=0.04, and alloimmunized, P=0.04) (Figure 3C). We also tested TNFα typically expressed by activated moDCs. Similar levels of TNFα were produced by moDCs in the absence or presence of heme, indicating lack of toxic effects of heme on moDCs (Online Supplementary Figure S3). Altogether, these data indicate that heme down-regulates IL-12 expression by moDCs regardless of alloimmunization status, suggesting that IL-12 inhibition is not involved in the differential heme-mediated regulation of DC immunogenicity. Furthermore, the data clearly show that moDCs from alloimmunized SCD patients are still responsive to heme at least in their ability to down-regulate IL-12 production.

HO-1, CO and biliverdin are not involved in CD83 regulation of hemin-treated moDCs HO-1 is induced by hemin/heme and is known for its immunoregulatory functions.13,22 We have previously reported that altered monocyte control of T-cell responses in alloimmunized SCD patients following exposure to exogenous heme was in part due to monocyte HO-1 levels.9 In contrast to monocytes,9 baseline HO-1 levels (in the haematologica | 2016; 101(9)

Hemin alters dendritic cell function

Figure 2. Expression of CD83 by hemin-treated human monocyte derived dendritic cells (moDCs). Immature moDCs were pre-treated or not with hemin (0, 5, or 20 μM) at day 0. After 2 h, maturation was performed or not (B) using 100 ng/mL lipopolysaccharide (LPS) (A and C), 100 ng/mL LPS + 500 UI/mL IFNg (D), or 5 μM R848 (E). At day 2, CD83 expression was assessed by immunostaining. (A) Representative example for non-alloimmunized patients. Two-tailed paired Student’s t-tests were used to compare the expression of co-stimulation markers on hemin-treated moDCs versus untreated moDCs. *P<0.05.

absence of hemin) in moDCs from alloimmunized SCD patients were not statistically different from those of nonalloimmunized and healthy donors (Figure 4A and B). Interestingly, and as already reported,22 upregulation of HO-1 in response to hemin was lower in matured moDCs than in immature moDCs (Figure 4B), and this trend was similar in both healthy donors and patients. We found higher levels of HO-1 induction following hemin treatment in immature moDCs from non-alloimmunized than from alloimmunized SCD patients (12±2fold increase vs. 6±1-fold increase; P=0.02), in LPSmatured moDCs (5±0.6-fold increase vs. 3±0.3-fold increase; P=0.02), and to some extent in LPS+ IFNgmatured moDCs (6±1-fold increase vs. 3±0-fold increase; P=0.05), but not in R848-matured moDCs (4±0.4-fold increase vs. 3±0.4-fold increase; P=0.3). To further explore the role of HO-1 induction in CD83 regulation, we first examined the direct effects of HO-1 enzymatic products. HO-1 catabolizes heme into equimolar amounts of labile Fe, biliverdin (BV) and carbon monoxide (CO). Its products, especially BV and CO, are thought to drive HO-1 immunomodulatory effects.23-27 We therefore tested whether HO-1 could down-regulate CD83 through BV or CO. MoDCs were cultured in the presence of BV or the CO-release molecule CORM-328 before being matured with TLR agonists. Whereas heminmediated downregulation of CD83 was observed in both healthy donors (Online Supplementary Figure S4A) and nonalloimmunized SCD patients (Figure 4C, left panel), HO-1 products had no effect in inhibiting CD83 expression (Figure 4C, see CORM-3 or Biliverdin lanes). As expected, CD83 was not down-regulated by hemin in alloimmunized patients (Figure 4D, right panel). Furthermore, CD83 expression was unaffected by HO-1 products in alloimmunized patients (Figure 4D, right panel). Efficacy of the hemin treatment was controlled by measuring induction of HO-1 levels (Online Supplementary Figure S4B) and CORM-3 treatment was controlled by IL-12 inhibition haematologica | 2016; 101(9)

(Online Supplementary Figure S4C).22 As a second approach to examine the potential role of HO-1 in CD83 regulation, we blocked HO-1 activity using 2 inhibitors, namely SnPPIX and ZnPPIX.29 MoDCs treated with these inhibitors were not capable of blocking CD83 downregulation in the presence of hemin either in healthy donors (Online Supplementary Figure S5A) or in non-alloimmunized patients (Figure 4D, left panel). As expected, CD83 was not down-regulated by hemin in alloimmunized patients and was unaffected by HO-1 inhibitors (Figure 4D, right panel). Efficacy of the hemin treatment was controlled by measuring HO-1 levels (Online Supplementary Figure S5B-D). Altogether, these results show that CD83 downregulation is HO-1-independent.

Hemin blocks CD83 through TLR4 triggering Heme can trigger TLR4.14,30-32 We therefore examined whether hemin down-regulates CD83 through a TLR-4mediated pathway. TLR4 also binds LPS and, as expected, TLR4 blockade using a blocking polyclonal antibody against TLR433 strongly prevented TLR4 activation by LPS in the absence of hemin, as demonstrated by inhibition of CD83 and IL-12 (Figure 5 and Online Supplementary Figure S6, respectively; compare isotype control vs. anti-TLR4). Pre-treatment of mature moDCs derived from healthy donors (Figure 5A) or non-alloimmunized SCD patients (Figure 5B) with anti-TLR4 no longer resulted in downregulation of CD83 in response to hemin, whereas the isotype control effectively inhibited CD83 expression in hemin-treated moDCs. TLR4 blockade did not further affect CD83 on the moDCs from the alloimmunized group (Figure 5C). Anti-TLR4 reversed the inhibition of IL12p40 by hemin in all groups, confirming the efficacy of TLR4 blockade (Online Supplementary Figure S6A-C). These data strongly suggest that hemin down-regulates DC maturation by signaling through TLR4. 1031

E. Godefroy et al. A

Figure 3. Production of IL-12 by hemin-treated human monocyte derived dendritic cells (moDCs). Immature moDCs were pre-treated or not with indicated doses of hemin (0, 5, or 20 μM) at day 0. When maturation was performed, 100 ng/mL lipopolysaccharide (LPS), 100ng/mL LPS + 500UI/mL IFNg, or 5 μM R848 was added 2 h later. For intracellular staining, all conditions were cultured in the presence of brefeldin A. The next day, cells were fixed and stained intracellularly for IL-12p40 (A and B). (C) For IL-12p70 detection, supernatants (without brefeldin A) were harvested two days after maturation. (A) Representative example for nonalloimmunized patients. In the case of LPS+ IFNg matured moDCs, at lower concentrations of hemin, we found a more robust downregulation of IL-12p40 in nonalloimmunized moDCs (no hemin vs. 5 μM hemin; P<0.002) as compared to alloimmunized (P>0.1), similar to our published data in LPS/ IFNg treated monocytes.9 Two-tailed paired Student’s t-tests were used to compare the expression of cytokines produced by hemin-treated moDCs versus untreated moDCs. *P<0.05.

Hemin inhibits NF-κB responses in moDCs from non-alloimmunized, but not alloimmunized SCD patients The cd83 promoter contains NF-κB binding sites and NF-κB transcription factors are involved in regulating maturation-specific CD83 expression in DCs.34 To test whether heme alters NF-κB-mediated maturation of DCs, we first examined the activation levels of NF-κB transcription factors in hemin-exposed moDCs from healthy donors. Immature moDCs were stimulated for 2 h with R848 or LPS+ IFNg in the presence or absence of hemin, and NF-κB transcription factors, namely p50, p52, p65, RelB and c-Rel, were tested in nuclear fractions (Figure 6). We found a robust and consistent downregulation of p50 activation (Figure 6A) and RelB (Figure 6D), but not of the other NF-κB subunits in non-alloimmunized hemin-treated mature moDCs. These results suggest a model whereby hemin blocks p50 and RelB activation in non-alloimmunized, but not in alloimmunized SCD patients.

Discussion In this study, we identified striking differences in innate immune response to exogenous heme in alloimmunized 1032

and non-alloimmunized SCD patients. Hemin-exposed moDCs derived from non-alloimmunized SCD patients inhibited priming of pro-inflammatory CD4+ T cells expressing IFNg (TH1) (Figure 1) and TNFα (TH1-like) (Online Supplementary Figure S1A) and exhibited a significant decrease in the levels of the maturation marker CD83 expression (Figure 2), similar to those of healthy donors. Strikingly, in alloimmunized patients, hemin did not affect DC maturation or their ability to prime TH1 cells. The underlying mechanism for CD83 regulation depended on TLR4 (Figure 7), but not on HO-1. The prototypical proinflammatory signaling pathway NF-κB transcription factor, specifically p50 and RelB subunits, were down-regulated in response to hemin in moDCs from non-alloimmunized, but not alloimmunized patients (Figure 6). These data suggest that hemin dampens DC immunogenicity through downregulation of CD83/p50/RelB and that this pathway is defective in alloimmunized SCD patients. There was no statistical difference in baseline levels of the key immune parameters tested in this study, including CD83, IL-12, HO-1, p50 and RelB levels, between alloimmunized and non-alloimmunized SCD patients, probably due to person to person variations. On the other hand, differences were observed upon treatment with exogenous heme. Specifically, DC immunogenicity was dampened in non-alloimmunized and healthy donors, but not in alloimhaematologica | 2016; 101(9)

Hemin alters dendritic cell function

munized patients with heme treatment. Furthermore, inhibition of NF-κB signaling was identified as a mechanism by which heme could suppress DC maturation and function. Out of the NF-κB transcription factors (p50, p52, p65, RelB, c-Rel), we have narrowed down two, p50 and RelB, that are strikingly differentially activated in alloimmunized and non-alloimmunized SCD patients in the presence of heme. Differences in TLR4 expression levels may be responsible for altered p50/RelB activation, but we did not directly measure TLR4 levels on DCs derived from alloimmunized or non-alloimmunized patients, partly due to the insensitivity of commercially-available anti-TLR detection antibodies. There was no difference in inflammatory cytokines, including IL-12, between the 2 patient populations in response to LPS, suggesting that functional TLR4 levels are likely to be similar in both groups of patients. Therefore, differences between both groups could be due to differential co-receptor(s) used to detect hemin or signaling molecules “located” between TLR4 and NF-κB. We believe that uncovering this pathway represents a significant and novel finding that distinguishes DC function in alloimmunized versus non-alloimmunized haematologica | 2016; 101(9)

Figure 4. Expression of HO-1 by hemin-treated human monocyte derived dendritic cells (moDCs). (A and B) Immature moDCs were pretreated or not with hemin (0, 5, or 20 μM) at day 0. When maturation was performed, 100 ng/mL lipopolysaccharide (LPS), 100 ng/mL LPS + 500 UI/mL IFNg, or 5 μM R848 was added 2 h later. After two days, HO-1 was stained intracellularly (A and B). (A) Representative examples. (C) Immature moDCs were pre-treated or not with indicated doses of hemin products or with hemin at day 0. Maturation was performed 2 h later. After two days, CD83 levels were assessed by immunostaining. (D) HO-1 inhibitors (SnPPIX and ZnPPIX) were used to treat immature moDCs. Two hours later, hemin was added at indicated doses. After two additional hours, moDCs were matured or not. CD83 was detected at day 2. Two-tailed paired Student’s t-tests were used to compare the expression of HO-1 or CD83 by hemin-treated moDCs (or treated with hemin products) versus untreated moDCs. *P<0.05 [error bars, standard error of the mean (SEM)].

patients, and should help the identification of downstream target genes encoding potential key molecules such as CD83, in alloimmunization. It may be that allosenistized patients have higher ongoing hemolysis and are, therefore, exposed to more cell-free heme which alters their DC responses. More accurate measurements of intravascular hemolysis including haptoglobin levels will be needed to investigate this possibility, although the reticulocyte counts, a surrogate marker for anemia, in our cohort of alloimmunized patients were not statistically different from the non-alloimmunized group (Table 1). It remains to be determined whether the p50/RelB inhibitory heme-effect is the cause or effect of alloimmunization in SCD. Future longitudinal studies can determine whether this pathway drives alloimmunization or not. As a working model, we hypothesize that the ongoing hemolysis in SCD can dampen DC maturation in nonalloimmunized patients, thereby lowering the immune response and subsequent risk of alloimmunization. However, when SCD DCs are defective in their response to heme/hemolysis, they are poised to trigger RBC-specific immune responses when transfusion occurs, resulting in 1033

E. Godefroy et al.

alloimmunization. Alternatively, attenuation of the NF-κB inhibitory heme-effect may be the result of alloimmunization; it may be that once patients become alloimmunized, the microenvironment changes, altering DC properties. This would be similar to the situation in other immune disorders where the microenvironment can affect immune responses and DC function.35-37 These studies also raise the provocative possibility that alloimmunized SCD patients may display higher responses, not only against RBC antigens, but also non-specifically.38 We found that

hemin down-regulates IL-12 production, the prototypical TH1-skewing cytokine, in moDCs from both alloimmunized and non-alloimmunized SCD patients (Figure 3). In LPS+IFNg-treated monocytes, we previously reported greater inhibition of IL-12p40 expression by hemin in nonalloimmunized as compared to alloimmunized patients,9 but the concentrations of hemin used in monocytes were significantly lower (<5 μM hemin) than in this study due to differences in culture conditions. It may also be that differences in molecular programing occur during differenti-

Figure 5. Hemin downregulates CD83 in a TLR4-dependent manner. Immature human monocyte derived dendritic cells (moDCs) from healthy donors (A), non-alloimmunized (B) or alloimmunized patients (C) were incubated for 1 h in the presence of either an isotype control or a blocking antibody for TLR4 (20 μg/mL). Hemin was then added for 2 h before maturation. CD83 was assessed two days later. Two-tailed paired Student’s t-tests were used to compare the expression of CD83 by hemin-treated moDCs versus untreated moDCs. *P<0.05.

Table 1. Characteristics of red blood cell non-alloimmunized versus alloimmunized patients on chronic transfusion. Age (years), median (min, max) Female Transfusion indication: stroke prevention Cumulative number of transfused RBC units, median (min, max) Iron chelation* WBC (x109/ L), median (min, max) Hgb (g/dL), median (min, max) Reticulocyte (x109/L), median (min, max) Pre-transfusion HgbS (%), median (min, max) RBC alloantibody specificities Anti-D Anti-C Anti-E Anti-e Anti-V Anti-VS Anti-Jsa Anti-Kpa Anti-Fyb Anti-S Anti-Lea Anti-COb Anti-P1 Anti-WRa RBC autoantibody Anti-HLA

Non-alloimmunized, n=6

Alloimmunized, n=12

18.4 (15.7, 21.9) 2 (33.3%) 6 (100%) 552 (160, 726) 0 (0%) 9.9 (5.6, 19.2) 9.1 (7.1, 11.4) 350 (260, 593) 32.95 (21.5, 61.7)

20.9 (16.6, 33.7) 4 (33.3%) 11 (91.7%) 629 (266, 1659) 2 (16.7%) 13.15 (6.8, 22.6) 10.65 (7.6, 12.1) 414 (174, 559) 35.2 (22.3, 54.3) 10 (83.3%) 7 (58.3%) 3 (25%) 3 (25%) 2 (11.1%) 1 (8.3%) 4 (33.3%) 1 (8.3%) 1 (8.3%) 3 (25%) 1 (8.3%) 1 (8.3%) 1 (8.3%) 1 (8.3%) 12 (100%) 11 (91.7%)

RBC: red blood cells; WBC: white blood cell count; Hgb: hemoglobin; HgbS: sickle hemoglobin;*both patients were on Exjade therapy.

1034

haematologica | 2016; 101(9)

Hemin alters dendritic cell function

ation of monocytes into moDCs. These data indicate that even though moDCs from alloimmunized SCD patients cannot down-regulate CD83 or inhibit TH1 priming following heme treatment, they are still responsive to heme, thereby arguing against an analogous state of anergy or exhaustion to that described for T cells. These data also suggest that hemin-mediated inhibition of IFNg-secreting

T-cell priming by moDCs is not solely through dampening IL-12. Instead, it is likely that the inhibition of TH1 responses is due to the downregulation of CD83. Indeed, deletion or mutation of mouse CD83 results in defective CD4+ T-cell development, characterized by a dramatic reduction in both CD4 single-positive thymocytes and peripheral CD4+ T cells,39,40 as well as inhibited DC-medi-

Figure 6. NF-κB inhibition by hemin. Immature human monocyte derived dendritic cells (moDCs) were matured with 5 μM R848 (gray) or 100 ng/mL lipopolysaccharide (LPS) + 500 UI/ml IFNg (black) for 2 h in the presence or in the absence of 20 μM hemin. Nuclei were then extracted (Active Motif) and the level of transcription factors p50 (A), p52 (B), p65 (C), RelB (D) and c-Rel (E) in the nuclear fraction of moDCs was assessed (Active Motif) in moDCs derived from healthy donors (left panels), non-alloimmunized (middle panels), or alloimmunized (right panels) patients. Results are represented as OD450 Two-tailed paired Student’s t-tests were used to compare moDCs stimulated alone versus stimulated + hemin. *P<0.05.

haematologica | 2016; 101(9)

1035

E. Godefroy et al.

Figure 7. Proposed model. Hemin triggers TLR4 on human monocyte derived dendritic cells (moDCs), which leads to inhibition of NF-κB and downstream CD83 in non-alloimmunized sickle cell disease (SCD) patients. On the other hand, this signaling pathway is defective in alloimmunized patients, thereby preventing the inhibition of NF-κB and CD83. Hemin-exposed moDCs derived from non-alloimmunized patients (CD83low) prime naïve T cell to produce lower levels of ΙFΝg than when derived from alloimmunized patients (CD83high). This pathway could, therefore, represent a mechanism by which enhanced inflammation is induced and/or maintained in alloimmunized patients.

ated T-lymphocyte activation.41 Moreover, circulating CD4+ T cells in CD83–/– mice have been shown to have an altered phenotype with lower expression of TCRβ, CD5 and CD3.39 Despite an ill-defined function for CD83, these results, together with ours, strongly suggest an instrumental role for this molecule in CD4+ T-cell function. Whether CD83 affect all CD4+T-cell subsets in a similar fashion still has to be investigated. Regarding IL-17A responses (TH17like), only low levels were observed (Online Supplementary Figure S1C), making it difficult to assess the effect of hemin on TH17 responses in our system. We found IL-4 (TH2-like) production by T cells was not affected by hemin treatment (Online Supplementary Figure S1B), suggesting that CD83 downregulation by hemin mainly affects TH1, at least in our system. Because hemin induces overexpression of HO-1, which in turn exhibits immunomodulatory function,13,22,26 we first asked whether CD83 expression was inhibited by HO-1. In one report, inhibition of HO-1 promoted DC maturation (HLA-DR and CD86) in a p38 MAPK-CREB/ATF1dependent manner.42 However, these effects were comparable to LPS-treatment and the possibility of endotoxin contamination in their HO-1 inhibited DC cultures was not ruled out. On the contrary, in our system, CD83 expression is independent of HO-1. The use of either hemin products (Figure 4C) or HO-1 inhibitors (Figure 4D) did not affect CD83 downregulation in moDCs from nonalloimmunized SCD patients (or from healthy donors), supporting the conclusion that hemin does not down-regulate CD83 through HO-1. These data do not exclude the possibility that other anti-inflammatory effects of heme are mediated through HO-1.13,22,43 For example, we found that IL-12 was inhibited by CO in DCs (Online Supplementary Figure S4C), invoking a role for the HO-1 pathway in IL-12 regulation. Since we did not observe a 1036

difference in inhibition of IL-12 at high concentrations of heme between alloimmunized and non-alloimmunized patients, the possibility remains that the HO-1 pathway is intact in these two groups. Provision of CO has been shown to down-regulate DC activation markers CD80 and CD86, as well as CD83, in healthy donors.22 We did not examine the effects of CO on CD80 and CD86, but following heme treatment, CD80 or CD86 levels were not affected, and only CD83 was inhibited (in healthy donors and non-alloimmunized patients). One can also speculate that the formation of a dimer between p50 and RelB is responsible for the specificity for CD83 expression. A key finding of our study is that heme blocks the proinflammatory signal though TLR4 signaling in DCs in healthy donors and non-alloimmunized SCD patients (Figure 7), and that this occurs through inhibition of the NF-κB activation pathway, namely p50 and RelB (Figure 6). Interestingly, most of the published literature implicates a pro-inflammatory effect of heme/TLR4 signaling.14,18,30-32 Differences in cell types at the level of co-receptors and/or signaling molecules, involving other NF-κB subunits besides p50 and RelB14,44 may explain the dichotomy as to why heme/heme analogs drive anti-inflammatory responses in DC, as we and others13 have detected, but pro-inflammatory responses in these other cell types. In in vitro studies, heme signaling induces TNFα, albeit at low levels, and only in the absence of serum.15,18,30 We have also observed hemin-induced very low levels of TNFα in monocytes, but only in the total absence of serum (data not shown). Our present study was performed in the presence of serum (5% from healthy donors), enabling us not only to detect the anti-inflammatory pathway mediated by heme, but, more importantly, the existence of a defective heme/TLR4 antiinflammatory signaling response in alloimmunized SCD patients. LPS signaling, which also occurs through TLR4 haematologica | 2016; 101(9)

Hemin alters dendritic cell function

signaling in DCs, is proinflammatory and involves upregulation of NF-κB.14,45 One may speculate that LPS and heme have different binding sites or binding affinities for TLR4, or require additional molecules that can induce a pro- versus an anti-inflammatory downstream signal in DCs. In alloimmunized patients, we found that the LPS/TLR4 proinflammatory signal is intact, since LPS (in the absence of hemin) resulted in upregulation of CD83 and IL-12 in moDCs similar to non-alloimmunized and healthy donors, but that the anti-inflammatory heme/TLR4 pathway is defective in matured (either by TLR4 or by TLR7/8 agonists) DCs (Figure 7). Deciphering the components of this defective pathway could possibly identify targets to restore the anti-inflammatory capacity of DCs when exposed to free heme, thereby potentially reducing the risk of alloimmunization. In summary, we have found that in non-alloimmunized SCD patients extracellular heme limits CD83 expression on DCs through a TLR4-mediated pathway, involving NF-κB, resulting in the dampening of pro-inflammatory responses (Figure 7). In contrast, in alloimmunized SCD patients, heme-driven TLR4-dependent pathway is defective, preventing NF-κB-mediated CD83 downregulation

References 1. Frenette PS, Atweh GF. Sickle cell disease: old discoveries, new concepts, and future promise. J Clin Invest. 2007;117(4):850-858. 2. Yazdanbakhsh K, Ware RE, Noizat-Pirenne F. Red blood cell alloimmunization in sickle cell disease: pathophysiology, risk factors, and transfusion management. Blood. 2012;120(3):528-537. 3. Bao W, Zhong H, Li X, et al. Immune regulation in chronically transfused allo-antibody responder and nonresponder patients with sickle cell disease and beta-thalassemia major. Am J Hematol. 2011; 86(12):1001-1006. 4. Godefroy E, Zhong H, Pham P, Friedman D, Yazdanbakhsh K. TIGIT-positive circulating follicular helper T cells display robust Bcell help functions: potential role in sickle cell alloimmunization. Haematologica. 2015;100(11):1415-1425. 5. Nickel RS, Horan JT, Fasano RM, et al. Immunophenotypic parameters and RBC alloimmunization in children with sickle cell disease on chronic transfusion. Am J Hematol. 2015;90(12):1135-1141. 6. Vingert B, Tamagne M, Desmarets M, et al. Partial dysfunction of Treg activation in sickle cell disease. Am J Hematol. 2014; 89(3):261-266. 7. Vingert B, Tamagne M, Habibi A, et al. Phenotypic differences of CD4(+) T cells in response to red blood cell immunization in transfused sickle cell disease patients. Eur J Immunol. 2015;45(6):1868-1879. 8. Schaer DJ, Buehler PW, Alayash AI, Belcher JD, Vercellotti GM. Hemolysis and free hemoglobin revisited: exploring hemoglobin and hemin scavengers as a novel class of therapeutic proteins. Blood. 2013; 121(8):1276-1284. 9. Zhong H, Bao W, Friedman D, Yazdanbakhsh K. Hemin controls T cell polarization in sickle cell alloimmunization.

haematologica | 2016; 101(9)

(Figure 7). Our data show that CD83 expression and T-cell priming were not inhibited by hemin in alloimmunized patients, and this could represent a mechanism by which enhanced inflammation is induced and/or maintained in this group of patients. Soluble CD83 fragments or CD83-Ig fusion protein46 to limit inflammation may be efficacious in preventing alloimmunization. Taken together, we believe that the identification of heme-driven TLR4 signaling and NF-κB inhibition as potential brakes of SCD alloimmunization may lead to the development of diagnostic strategies to predict which patients are at risk of alloimmunization, as well as of innovative treatments to prevent alloimmunization in this patient population. Acknowledgments We thank the members of the Laboratory of Complement Biology (NYBC), Drs. Lifeng Liu, and Ms. Meredith Spadaccia for technical help and discussions. Funding This work was in part supported by an NIH grant R01HL121415 (to KI), an American Heart Association grant, 14GRNT20480109 (to KI).

J Immunol. 2014;193(1):102-110. 10. Steinman RM, Hawiger D, Nussenzweig MC. Tolerogenic dendritic cells. Annu Rev Immunol. 2003;21:685-711. 11. Banchereau J, Steinman RM. Dendritic cells and the control of immunity. Nature. 1998;392(6673):245-252. 12. Hivroz C, Chemin K, Tourret M, Bohineust A. Crosstalk between T lymphocytes and dendritic cells. Crit Rev Immunol. 2012;32(2):139-155. 13. Chauveau C, Remy S, Royer PJ, et al. Heme oxygenase-1 expression inhibits dendritic cell maturation and proinflammatory function but conserves IL-10 expression. Blood. 2005;106(5):1694-1702. 14. 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. 2014;123(3):377-390. 15. Dutra FF, Alves LS, Rodrigues D, et al. Hemolysis-induced lethality involves inflammasome activation by heme. Proc Natl Acad Sci USA. 2014;111(39):E41104118. 16. Fortes GB, Alves LS, de Oliveira R, et al. Heme induces programmed necrosis on macrophages through autocrine TNF and ROS production. Blood. 2012;119(10):23682375. 17. Vercellotti GM. Look out heme, Mac is back in town. Blood. 2014;124(9):13881389. 18. Vinchi F, Costa da Silva M, Ingoglia G, et al. Hemopexin therapy reverts heme-induced pro-inflammatory phenotypic switching of macrophages in a mouse model of sickle cell disease. Blood. 2016;127(4):473-486. 19. Trinchieri G. Role of interleukin-12 in human Th1 response. Chem Immunol. 1996;63:14-29. 20. Bogunovic D, Manches O, Godefroy E, et al. TLR4 engagement during TLR3-induced proinflammatory signaling in dendritic cells promotes IL-10-mediated suppression of

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

antitumor immunity. Cancer Res. 2011; 71(16):5467-5476. Skoberne M, Somersan S, Almodovar W, et al. The apoptotic-cell receptor CR3, but not alphavbeta5, is a regulator of human dendritic-cell immunostimulatory function. Blood. 2006;108(3):947-955. Remy S, Blancou P, Tesson L, et al. Carbon monoxide inhibits TLR-induced dendritic cell immunogenicity. J Immunol. 2009; 182(4):1877-1884. Chora AA, Fontoura P, Cunha A, et al. Heme oxygenase-1 and carbon monoxide suppress autoimmune neuroinflammation. J Clin Invest. 2007;117(2):438-447. Otterbein LE, Bach FH, Alam J, et al. Carbon monoxide has anti-inflammatory effects involving the mitogen-activated protein kinase pathway. Nat Med. 2000; 6(4):422-428. Ryter SW, Choi AM. Targeting heme oxygenase-1 and carbon monoxide for therapeutic modulation of inflammation. Transl Res. 2016;167(1):7-34. Soares MP, Marguti I, Cunha A, Larsen R. Immunoregulatory effects of HO-1: how does it work? Curr Opin Pharmacol. 2009; 9(4):482-489. Wagener FA, Volk HD, Willis D, et al. Different faces of the heme-heme oxygenase system in inflammation. Pharmacol Rev. 2003;55(3):551-571. Santos-Silva T, Mukhopadhyay A, Seixas JD, Bernardes GJ, Romao CC, Romao MJ. Towards improved therapeutic CORMs: understanding the reactivity of CORM-3 with proteins. Curr Med Chem. 2011; 18(22):3361-3366. Jozkowicz A, Dulak J. Effects of protoporphyrins on production of nitric oxide and expression of vascular endothelial growth factor in vascular smooth muscle cells and macrophages. Acta Biochim Pol. 2003;50(1):69-79. Figueiredo RT, Fernandez PL, Mourao-Sa DS, et al. Characterization of heme as acti-

1037

E. Godefroy et al.

31.

32.

33.

34.

35.

1038

vator of Toll-like receptor 4. J Biol Chem. 2007;282(28):20221-20229. Ghosh S, Adisa OA, Chappa P, et al. Extracellular hemin crisis triggers acute chest syndrome in sickle mice. J Clin Invest. 2013;123(11):4809-4820. Lin S, Yin Q, Zhong Q, et al. Heme activates TLR4-mediated inflammatory injury via MyD88/TRIF signaling pathway in intracerebral hemorrhage. J Neuroinflammation. 2012;9:46. Mogami H, Kishore AH, Shi H, Keller PW, Akgul Y, Word RA. Fetal fibronectin signaling induces matrix metalloproteases and cyclooxygenase-2 (COX-2) in amnion cells and preterm birth in mice. J Biol Chem. 2013;288(3):1953-1966. Stein MF, Lang S, Winkler TH, et al. Multiple interferon regulatory factor and NF-kappaB sites cooperate in mediating cell-type- and maturation-specific activation of the human CD83 promoter in dendritic cells. Mol Cell Biol. 2013;33(7):13311344. Gabrilovich D. Mechanisms and functional significance of tumour-induced dendriticcell defects. Nat Rev Immunol. 2004;4(12):941-952.

36. Serafini P, Borrello I, Bronte V. Myeloid suppressor cells in cancer: recruitment, phenotype, properties, and mechanisms of immune suppression. Semin Cancer Biol. 2006;16(1):53-65. 37. Solito S, Pinton L, Damuzzo V, Mandruzzato S. Highlights on molecular mechanisms of MDSC-mediated immune suppression: paving the way for new working hypotheses. Immunol Invest. 2012;41(6-7):722-737. 38. Nickel RS, Hendrickson JE, Fasano RM, et al. Impact of red blood cell alloimmunization on sickle cell disease mortality: a case series. Transfusion. 2016;56(1):107-114. 39. Fujimoto Y, Tu L, Miller AS, et al. CD83 expression influences CD4+ T cell development in the thymus. Cell. 2002;108(6):755767. 40. Garcia-Martinez LF, Appleby MW, Staehling-Hampton K, et al. A novel mutation in CD83 results in the development of a unique population of CD4+ T cells. J Immunol. 2004;173(5):2995-3001. 41. Kruse M, Rosorius O, Kratzer F, et al. Inhibition of CD83 cell surface expression during dendritic cell maturation by interference with nuclear export of CD83 mRNA. J

Exp Med. 2000;191(9):1581-1590. 42. Al-Huseini LM, Aw Yeang HX, Hamdam JM, et al. Heme oxygenase-1 regulates dendritic cell function through modulation of p38 MAPK-CREB/ATF1 signaling. J Biol Chem. 2014;289(23):16442-16451. 43. Wang XM, Kim HP, Nakahira K, Ryter SW, Choi AM. The heme oxygenase-1/carbon monoxide pathway suppresses TLR4 signaling by regulating the interaction of TLR4 with caveolin-1. J Immunol. 2009; 182(6):3809-3818. 44. Seldon MP, Silva G, Pejanovic N, et al. Heme oxygenase-1 inhibits the expression of adhesion molecules associated with endothelial cell activation via inhibition of NF-kappaB RelA phosphorylation at serine 276. J Immunol. 2007;179(11):7840-7851. 45. Ardeshna KM, Pizzey AR, Devereux S, Khwaja A. The PI3 kinase, p38 SAP kinase, and NF-kappaB signal transduction pathways are involved in the survival and maturation of lipopolysaccharide-stimulated human monocyte-derived dendritic cells. Blood. 2000;96(3):1039-1046. 46. Prazma CM, Tedder TF. Dendritic cell CD83: a therapeutic target or innocent bystander? Immunol Lett. 2008;115(1):1-8.

haematologica | 2016; 101(9)

ARTICLE

Platelet Biology & Its Disorders

Immune thrombocytopenia in adults: a prospective cohort study of clinical features and predictors of outcome

Lamiae Grimaldi-Bensouda,1 Clémentine Nordon,1 Marc Michel,2 Jean-François Viallard,3 Daniel Adoue,4 Nadine Magy-Bertrand,5 Jean-Marc Durand,6 Philippe Quittet,7 Olivier Fain,8 Bernard Bonnotte,9 Anne-Sophie Morin,10 Nathalie Morel,11 Nathalie Costedoat-Chalumeau,12 Brigitte Pan-Petesch,13 Mehdi Khellaf,14 Antoinette Perlat,15 Karim Sacre,16 François Lefrere,17 Lucien Abenhaim,1 and Bertrand Godeau18 for the PGRx-ITP Study Group*

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2016 Volume 101(9):1039-1045

LASER, Paris; 2Centre de Référence des Cytopénies Auto-Immunes de l'Adulte, Service de Médecine Interne, CHU Henri Mondor, Créteil; 3Service de Médecine Interne, Hôpital Haut-Lévêque, Pessac; Université Bordeaux 2, Bordeaux; 4IUCT Oncopôle, Toulouse; 5 Service de Médecine Interne, CHU Jean Minjoz, Besançon; 6Service de Médecine Interne, Hôpital de la Timone, Marseille; 7Département d’Hématologie Clinique, Hôpital St-Eloi, Montpellier; 8Service de Médecine Interne, Hôpital Saint Antoine; Hôpitaux Universitaires de l'Est Parisien, AP-HP Université Paris 6, Paris; 9Service de Médecine Interne et Immunologie Clinique, INSERM 1098, CHU Bocage Central, Dijon; 10Service de Médecine Interne CHU Jean Verdier, Université Paris 13, Assistance PubliqueHôpitaux de Paris (AP-HP), Bondy; 11Service de Médecine Interne, Hôpital Cochin, Paris; 12 Service de Médecine Interne, Hôpital Cochin, Paris; 13Service Hématologie, CHU Morvan 29609 Brest Cedex; 14Service des Urgences, Centre Hospitalier Henri Mondor, Assistance Publique-Hôpitaux de Paris, Université Paris-Est Créteil, Créteil; 15Service de Médecine Interne, CHU Rennes; 16Université Paris-Diderot, Paris; Assistance Publique Hôpitaux de Paris; INSERUM U1149, Paris; 17Service de Biothérapie, Hôpital Necker, APHP, Paris; and 18Centre de Référence des Cytopénies Auto-Immunes de l'Adulte, Service de Médecine Interne, CHU Henri Mondor, Créteil, France 1

*Contributing members of the PGRx-ITP Study Group are listed in the Online Supplementary Appendix.

ABSTRACT

his prospective observational cohort study aimed to explore the clinical features of incident immune thrombocytopenia in adults and predictors of outcome, while determining if a family history of autoimmune disorder is a risk factor for immune thrombocytopenia. All adults, 18 years of age or older, recently diagnosed with immune thrombocytopenia were consecutively recruited across 21 hospital centers in France. Data were collected at diagnosis and after 12 months. Predictors of chronicity at 12 months were explored using logistic regression models. The association between family history of autoimmune disorder and the risk of developing immune thrombocytopenia was explored using a conditional logistic regression model after matching each case to 10 controls. One hundred and forty-three patients were included: 63% female, mean age 48 years old (Standard Deviation=19), and 84% presented with bleeding symptoms. Median platelet count was 10x109/L. Initial treatment was required in 82% of patients. After 12 months, only 37% of patients not subject to disease-modifying interventions achieved cure. The sole possible predictor of chronicity at 12 months was a higher platelet count at baseline [Odds Ratio 1.03; 95%CI: 1.00, 1.06]. No association was found between outcome and any of the following features: age, sex, presence of either bleeding symptoms or antinuclear antibodies at diagnosis. Likewise, family history of autoimmune disorder was not associated with incident immune thrombocytopenia. Immune thrombocytopenia in adults has been shown to progress to a chronic form in the majority of patients. A lower platelet count could be indicative of a more favorable outcome. haematologica | 2016; 101(9)

Correspondence: clementine.nordon@la-ser.com

Received: March 18, 2016. Accepted: May 24, 2016. Pre-published: May 26, 2016. doi:10.3324/haematol.2016.146373

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1039

1039

L. Grimaldi-Bensouda et al.

Introduction

tory of autoimmune disorder in first-degree relatives could constitute a risk factor for developing ITP.

Immune thrombocytopenia (ITP) is an autoimmune disorder mediated by platelet antibodies thought to accelerate platelet destruction while inhibiting also their production,1 resulting in low platelet counts with potentially spontaneous bruising, petechial rash, mucosal bleeding or even life-threatening hemorrhage. ITP affects children and adults, with an incidence rate for the latter estimated between 2.8 and 3.9 per 100,000 person-years in Europe,2-6 and namely 2.9/100,000 in France.6 The onset of ITP is frequently insidious and low platelet counts often last beyond six months.7 A recent retrospective study based on administrative registers reported that about two-thirds of adult ITP patients are likely to develop a chronic form of the disease.6 The risk factors for chronicity were mainly explored in children8-10 and rarely in adults.11 Regarding the genetic risk, a few studies reported clusters of ITP incidence within families,12,13 but it is unknown whether a family history of autoimmune disorder can be a risk factor for developing ITP. In France, a nationwide prospective cohort of adult patients presenting with a newly diagnosed episode of ITP was constituted primarily to explore the association between exposure to common vaccines and risk of developing ITP.14 In this context, the present study aimed at describing the clinical features of adult ITP and its evolution over a 12-month period and exploring the baseline predictors of chronicity. We also explored whether a his-

Methods Source of data This was a prospective observational cohort study using data from the Pharmacoepidemiologic General Research eXtension (PGRx) information system which is a set of population-based registries including case-patients (cases) with specific diseases recruited by their specialist physician and referent-patients (controls) routinely recruited by their general practitioner (GP).14,15 Inclusion and exclusion criteria are similar for cases and controls, except for the disease of interest. Medical information is collected by specialists (for cases) and GPs (for controls). All patients undergo the same standardized telephone interview to collect information on family medical history, lifestyle and use of medication.

Study population Participating physicians in the PGRx-ITP registry were requested to consecutively include cases of ITP. All the patients met the following criteria: 1) age between 18 and 79 years; 2) diagnosis of primary ITP according to international consensus;16 3) delay between the first symptoms of ITP and inclusion of less than 365 days; 4) normal physical examination, except for bleeding symptoms; 5) domiciled in continental France; 6) able to understand and read French; and 7) agreement to participate. A strict procedure was used to confirm diagnosis (see Table 1 and Online

Table 1. Definitions used for isolated thrombocytopenia, secondary immune thrombocytopenia (ITP), incident ITP and outcome.

Condition Isolated thrombocytopenia

Secondary ITP

Incident ITP

Recovery at 12 months

Chronic ITP

1040

Definition

Exclusion/inclusion criteria

• platelet count <100×10 /L at baseline • in the absence of abnormality at physical examination, aside from bleeding • titration of human immunodeficiency virus (HIV) or hepatitis C virus (HCV) either not performed or returning a positive result • and/or any of the following co-morbidities: lymphoproliferative disorder, definite antiphospholipid syndrome,17 common variable immunodeficiency, thyroid disease, and overt systemic lupus erythematosus

Patients with platelet count ≥100×109/L at baseline were excluded

All patients diagnosed with secondary ITP were excluded

• presence of thrombocytopenia (<100×109/L) upon diagnosis • with a reference platelet count >150×109/L in the 12 months preceding diagnosis, even in the absence of bleeding symptoms upon diagnosis

In the absence of a reference platelet count >150×109/L in the 12 months preceding diagnosis, only patients with both acute bleeding symptoms and a platelet count ≤50×109/L were not excluded. Also, patients showing a delay between the first symptoms and inclusion of more than 365 days were excluded.

• platelet count ≥100×109/L • in the absence of ongoing treatment for ITP in the previous 8 weeks

Patients showing a platelet count ≥100×109/L at 12 months but who had been treated for ITP in the previous 8 weeks, were subject to a new evaluation at 14 months and considered cured only if showing a sustained response without treatment between 12 and 14 months.

Platelet count <100×109/L at 12 months, regardless of treatment. haematologica | 2016; 101(9)

Outcome of immune thrombocytopenia in adults

Supplementary Appendix). In order to determine whether a history of autoimmune disorder in first-degree relatives constituted a risk factor for developing ITP or not, up to 10 controls with no lifetime history of ITP were identified from the PGRx-General Practice registry and matched to each case according to sex, age (±1 year), and inclusion date (±5 years).

Measures All patient data were collected through dedicated case report forms (CRFs), similar in cases and controls regarding socio-demographics, personal medical history and family history of autoimmune disorder (in first-degree relatives): multiple sclerosis, lupus erythematosus, rheumatoid arthritis, Crohn disease, chronic ulcerative colitis, Hashimoto thyroiditis or Graves-Basedow disease. In cases, data were also collected on the ITP at baseline and after 12 months of follow up (Online Supplementary Appendix). The time elapsing between diagnosis and outcome measure at 12 months was preferred, given that first symptoms are often more difficult to specify. Recovery and chronicity were defined as in Table 1.

Statistical analysis Statistical analysis was performed with SAS software version 9.18 and all comparative tests were two-sided with a type-1 error set at α=0.05. Bleeding symptoms were categorized into: a) no bleeding; b) cutaneous bleeding alone; and c) severe bleeding defined as mucocutaneous and/or visceral bleeding (epistaxis, or bleeding of the mucous membrane, or visceral bleeding). The titration of antinuclear antibodies (ANA) was considered positive if the titer was more than 1/80. Further details are provided in the Online Supplementary Appendix.

Ethics approval This research was approved by the consultative committee for non-interventional research (Comité Consultatif sur le Traitement de l'Information en Matière de Recherche) and the French Data Protection Authority (Commission Nationale de l'Informatique et des Libertés) was informed that data were being collected for this purpose. All patients included in the PRGx database provided informed consent.

Results Over a 28-month period, 188 patients diagnosed with ITP were consecutively included in the PGRx-ITP registry by hematologists and internists from 21 hospitals across France. Of these, 18 patients were excluded once diagnostic procedures had been completed and another 27 patients were excluded as a result of failure to confirm diagnosis of ITP at three months; this left 143 patients for analysis.

Characteristics of ITP at onset A median of 35 days elapsed between the first symptoms and inclusion in the study, and 35 (24.5%) patients were recruited at 90 days or later after the onset of ITP symptoms. Table 2 and Figure 1 show the characteristics of the 143 patients and incident ITPs upon diagnosis. Median age was 50 years and on average there were more female than male, with a female/male ratio of 1.7 overall which dropped to 1 in patients over 50 years of age. Information on family history of autoimmune disorder was available in 109 (76.2%) patients. In 23 (16.1%) patients, bleeding symptoms were absent and the diagnosis was made fortuitously after a routine blood test. Visceral bleeding was present in 7 (4.9%) patients and none of the patients presented with intracerebral hemorrhage. The median platelet count was 10x109/L, and 99 (69.2%) patients showed values less than 20x109/L at presentation. A bone marrow aspiration was performed in 111 (77.6%) patients, namely in 93.5% of patients over 60 years of age (n=43). ANA was tested in 136 (95.1%) patients. A “watch and wait” strategy was decided for 18 (12.6%) patients. Among patients initiating treatment, 61 (48.8%) were administered corticosteroids and intravenous immunoglobulins (IVIg), 50 (40.0%) corticosteroids alone, and 14 (11.2%) IVIg alone. In 7 patients, additional therapy was necessary (vinblastine, vincristine or platelet transfusion). Table 3 provides comparisons among 35 patients testing

Figure 1. Distribution of age and sex in 143 adults diagnosed with a first lifetime episode of immune thrombocytopenia.

haematologica | 2016; 101(9)

1041

L. Grimaldi-Bensouda et al.

positive for ANA (titer >1/80) and 101 patients with a negative ANA test. The two groups differed, albeit non-significantly, in terms of bleeding symptoms at diagnosis, the absence of such symptoms being more frequent in patients testing positive for ANA. The presence of a family history of autoimmune disorder was twice as frequent in patients with a positive ANA test, yet not statistically significant. There was no difference in other characteristics (age, sex, and platelet count) between groups.

Twelve-month outcome A total of 28 (19.6%) patients were lost to follow up at twelve months after ITP had been first diagnosed. Among the 115 patients with data available at one year, 58 (50.4%) showed progression to chronic ITP while 57 recovered: 43 (37.4%) did so spontaneously (i.e. with no disease-modifying treatment) and 14 (12.2%) achieved cure with a disease-modifying treatment: either rituximab alone (n=12) or rituximab plus splenectomy (n=2). Table 4 provides the results from univariate logistic regression models exploring the baseline factors associated with 12-month outcome in patients showing progression to chronic ITP (n=58) and patients with spontaneous recovery (n=43). The presence of a higher platelet count or the absence of severe bleeding at baseline was found to be associated with increased odds of chronicity at 12 months. No difference was found among patients in terms of sex, age and testing positive for ANA. A bivariate model was fitted to explore the independent predictive effect of platelet counts (continuous variable) and severe bleeding (severe vs. no bleeding or cutaneous bleeding alone) at baseline. The model revealed that severe bleeding was not associated with chronicity [Odds Ratio (OR) 0.47; 95%CI: 0.19, 1.19; P=0.11], whereas a higher platelet count remained associated with increased odds of chronicity, despite not reaching the 0.05 significance level (OR 1.03; 95%CI: 1.00, 1.06; P=0.08). The interpretation of this result needs to take into account that "platelet counts" have been used as a continuous variable: for every 10x109/L increase in the platelet counts (e.g. from 10x109/L to 20x109/L, or from 20x109/L to 30x109/L), the equivalent increase in terms of odds of chronicity is about 34%. All results were similar when exploring the odds of chronicity versus recovery, which also included patients who recovered after disease-modifying treatment (data not shown).

Family history of autoimmune disorder A familial history of autoimmune disorder was found in 9 (8.3%) out of the 109 case-patients explored. Table 5 provides the results from conditional logistic regression models comparing 109 case-patients to 913 controls matched on sex, age, and inclusion date. Overall, having a first-degree relative with a history of autoimmune disorder was not associated with increased odds of developing ITP. No association was found in either the subgroup of ITP patients with a positive ANA test at baseline or the subgroup of ITP patients progressing to a chronic form of the disease. Table 2. Characteristics and initial medical management of 143 adult patients diagnosed with a first lifetime episode of immune thrombocytopenia (ITP).

Patients with an incident ITP (N=143) Median delay (in days) between 1st symptoms and inclusion [min, max] Clinical and biological features Mean age (years) [SD]1 Female (overall) 18-49 years (n=71) 50-79 years (n=72) Family history of autoimmune disorder2 (n=109) Rapid onset of symptoms Bleeding symptoms No bleeding symptoms Cutaneous bleeding only Severe bleeding3 Mean platelet count (x109/L) [SD]1 Positive ANA test (>1/80) (n=136) Medical management “Watch and wait” strategy Initiation of a treatment Including corticosteroids Including intravenous immunoglobulin Including other treatment4

35 [2 – 343] n (%) 47.8 [18.6] 90 (62.9%) 53 (74.6%) 37 (51.4%) 9 (8.3%) 117 (81.8%) 23 (16.1%) 40 (28.0%) 80 (55.9%) 16.0 [17.2] 35 (25.7%) n (%) 18 (12.6%) 125 (87.4%) 111 (77.6%) 75 (52.5%) 7 (4.9%)

Mean and Standard Deviation (SD) are described for parameters assumed to be normally-distributed; 2family history of autoimmune disorder included the following in firstdegree relatives: lupus erythematosus, rheumatoid arthritis, multiple sclerosis, Crohn’s disease, chronic ulcerative colitis, autoimmune thyroiditis; 3severe bleeding included epistaxis and/or mucous membranes and visceral bleeding (gastrointestinal, cerebral, gynecological, etc.); 4other treatment included platelet transfusion, dapsone, vincristine, or vinblastine. ANA: antinuclear antibodies. 1

Table 3. Comparison of patients testing positive (>1/80) or negative for antinuclear antibodies (ANA) at baseline in 136 patients with a newly diagnosed episode of immune thrombocytopenia.

Female sex Mean age at diagnosis [SD]2 Presence of bleeding symptoms Mean platelet count3 [SD] Family history of AID4

Positive ANA (N=35) n (%)

Negative ANA (N=101) n (%)

25 (71.4%) 48.9 [22.3] 26 (74.3%) 18.1 [16.9] n=30 4 (13.3%)

61 (60.4%) 46.8 [17.3] 88 (87.1%) 15.2 [16.9] n=73 4 (5.5%)

0.2460 0.5686 0.0808 0.3752 0.1895

P-values from Pearson’s c2 test or Fisher’s exact test (categorical variables) and Mann-Whitney test (age, platelet count), as appropriate; 2Mean and Standard Deviation (SD) are described for parameters assumed to be normally-distributed; 3expressed in 109/L; 4AID: autoimmune disorder, including multiple sclerosis, lupus erythematosus, rheumatoid arthritis, Crohn’s disease, chronic ulcerative colitis, and autoimmune thyroiditis.

1042

haematologica | 2016; 101(9)

Outcome of immune thrombocytopenia in adults

Discussion Main results Up till now, epidemiology data on ITP have been based on retrospective studies and/or administrative registers.6,19 Here we report for the first time results based on a prospective nationwide cohort of adults with incident ITP studied consecutively, using stringent diagnostic criteria, and followed up for at least one year. This one-year follow-up study provides interesting insights into characteristics of ITP as well as the risk factors predicting which patients will evolve to chronic forms of the disease. First, our results agree with earlier findings that ITP in adults is more frequently found in females, with a female/male ratio of 1.7 overall and a female/male ratio of 2.8 for patients between the 18 and 49 years of age. Both ratios agree with previous results.20,21 Regarding the clinical symptoms of ITP, only 16% of patients with ITP showed no signs of bleeding, a figure that is twice as low as that reported in the PARC-ITP Registry observational study.21 This discrepancy in results may be explained by the procedure used in our study to check diagnosis. In the present study, patients with no bleeding symptoms and no record of a normal platelet count in the preceding year were excluded, thus over-estimating the frequency of bleeding symptoms at diagnosis. It is interesting to note that only 4.9% of patients presented with visceral bleeding at baseline and that no patient had a fatal hemorrhage, thus confirming that ITP is rarely associated with life-threatening bleeding events at diagnosis.22 Second, our results show that a large majority of adults with newly diagnosed ITP are likely to develop a chronic form of the disease after one year. In our sample, only 37% of patients â&#x20AC;&#x153;spontaneouslyâ&#x20AC;? recovered, i.e. without the need for disease-modifying treatment such as rituximab or splenectomy. This confirms previous data suggesting that ITP is more likely to follow a chronic course in affected adults, compared to children in whom cure is achieved within several weeks in most patients.10 In this context, the ability to predict the course of newly diagnosed ITP in adults is crucial for both patients and physicians, allowing for a potential reduction in healthcare

costs, patient anxiety, and less need for aggressive initial treatment, such as splenectomy or immunosuppressant drugs. Our results confirm that sex and age are not risk factors for chronicity.23 Likewise, no association was found between bleeding symptoms at baseline and chronicity. Interestingly, a higher platelet count at baseline was found to be associated with increased odds of chronicity at 12 months, independent of confounders. However, it did not reach the 5% level of significance and further studies are needed to confirm such an association, which was also found when considering all patients recovering from ITP, regardless of disease-modifying treatment. Although of interest, this predictive association is insufficiently robust to be considered for clinical practice or to be taken into account by physicians when tailoring treatment decisions to individual patients. In our opinion, clinicians would only be able to reliably predict progression to chronic status by using a combination of clinical and biological factors. We previously reported that the presence of platelet antibodies detected by a MAIPA assay could help in predicting chronicity in ITP.11 However, the poor sensitivity of this test and its limited availability in many countries restricts its potential use. In the present study, the detection of platelet antibodies using a MAIPA assay was not systematically performed across all the participant hospitals, and, therefore, this measure was not taken into account, and other clinical and biological predictive factors should be identified. Third, the present findings provide further insight into the frequency and prognostic value of autoimmune markers, particularly ANA at ITP onset, which have so far proved controversial.24,25 ANA has been found to be a possible predictor of chronicity in childhood ITP.25 In adults, this is most likely not the case. In a previous retrospective study focused on adults, we reported that the presence of autoimmune markers did not correlate with presenting features, response to treatments, or long-term outcome of ITP.26 These conflicting results explain why assays for ANA are not routinely recommended by an international panel of experts.27 In our sample, assays for ANA were performed at baseline in nearly all of the participants and the test was found positive (titer>1/80) in more than 25%

Table 4. Baseline factors associated with the 12-month outcome, in adults recently diagnosed with an immune thrombocytopenia (ITP); patients who recovered without any disease-modifying drug are compared to patients with chronic ITP; results of univariate logistic regression models providing Odds Ratio (OR) and 95% Confidence Intervals (95%CI).

Female sex (vs. male) Mean age [SD]1 Bleeding symptoms No bleeding symptoms (=ref) Cutaneous bleeding only Severe bleeding2 Mean platelet count3 [SD]1 ANA4 dosage negative (=ref) Positive

Recovery without disease-modifying drug N=43 n (%)

Chronic ITP N=58 n (%)

OR [95%CI]

27 (62.8%) 47.0 [19.7]

39 (67.2%) 45.4 [16.6]

1.22 [0.53-2.78] 1.00 [0.97-1.02]

0.64 0.66

5 (11.6%) 10 (23.3%) 28 (65.1%) 10.7 [11.4] 27 (62.8%) 13 (30.2%)

14 (24.1%) 22 (37.9%) 22 (37.9%) 20.2 [19.7] 38 (65.5%) 19 (32.8%)

1 0.79 [0.22-2.79] 0.28 [0.09-0.90] 1.04 [1.01-1.07] 1 1.04 [0.44-2.46]

0.71 0.032 0.010 0.93

Mean and Standard Deviation (SD) are described for parameters assumed to be normally-distributed; 2severe bleeding included epistaxis and/or mucous membranes and visceral bleeding (gastrointestinal, cerebral, gynecological); 3expressed in 109/L; 4ANA: antinuclear antibodies. 1

haematologica | 2016; 101(9)

1043

L. Grimaldi-Bensouda et al. Table 5. Association between family history of autoimmune disorder and risk of developing an immune thrombocytopenia (ITP) in adulthood; results of univariate conditional logistic regression models providing Odds Ratio (OR) and 95% Confidence Intervals (95%CI).

Case-patients1 with a newly diagnosed ITP N (%) In all patients Family history of AID3 In patients with positive ANA4 (titre>1/80) Family history of AID In patients with chronic ITP at 12-month Family history of AID

Controls2 with no lifetime ITP N (%)

109 9 (8.3%) 30 4 (13.3%) 50 4 (8.0%)

913 103 (11.3%) 241 24 (10.0%) 372 40 (10.8%)

OR [95%CI]

0.83 [0.40 - 1.71]

0.614

1.43 [0.46 - 4.43]

0.535

0.87 [0.29 - 2.59]

0.798

Cases were adult patients with a first lifetime episode of ITP recruited in specialized centers; 2controls were adults with no lifetime history of ITP and recruited in a populationbased cohort; cases and controls are matched for age, sex and index date; 3auto-immune disorders (AID) included lupus erythematosus, rheumatoid arthritis, multiple sclerosis, Crohn’s disease, chronic ulcerative colitis, autoimmune thyroiditis; 4ANA: antinuclear antibodies.

of patients, a finding that is in line with previous reports.24,25 In patients with a positive ANA titer, the absence of bleeding symptoms was significantly more frequent, and unsurprisingly, the proportion of females was higher than that of males, and the presence of autoimmune disorder history in first-degree relatives was twice as frequent, although this was not significant, probably because of low sample size and insufficient statistical power. Unexpectedly, a positive ANA titer was not associated with ITP evolution. These results suggest that routine ANA assays appear non-mandatory in newly diagnosed ITP in adults. However, taking into account that the follow-up period for our study was limited to one year, we cannot exclude the possibility that the presence of ANA at the time of diagnosis may be a risk factor for late development of systemic lupus erythematosus. Finally, using a case-control design, we explored whether a family history of autoimmune disorder was a risk factor of developing ITP in adulthood. Our results suggest that this is not the case.

Study limitations The study setting used the entire network of French reference centers, which are highly specialized centers for ITP. Thus, more severe patients could have been included. However, there is no reason to believe that this selection

References 1. Cines DB, Bussel JB, Liebman HA, Luning Prak ET. The ITP syndrome: pathogenic and clinical diversity. Blood. 2009;113(26):6511-6521. 2. Abrahamson PE, Hall SA, Feudjo-Tepie M, Mitrani-Gold FS, Logie J. The incidence of idiopathic thrombocytopenic purpura among adults: a population-based study and literature review. Eur J Haematol. 2009;83(2):83-89. 3. Schoonen WM, Kucera G, Coalson J, et al. Epidemiology of immune thrombocytopenic purpura in the General Practice Research Database. Br J Haematol. 2009;145(2):235-244. 4. Michel M. Immune thrombocytopenic pur-

1044

7. 8.

bias confounded the predictive value of patients’ characteristics or modified the results regarding family history of autoimmune disorder. Due to strict eligibility criteria for the analyses, the population used for the analyses consisted of 143 out of 188 adult patients initially identified and recruited. The choice to select patients with confirmed diagnosis of primary and incident ITP has consequently lowered the statistical power for some analyses, in particular regarding any association between ANA positivity and patients’ characteristics.

Conclusion The study confirms that ITP is more common in women than men, and is frequently associated with cutaneous bleeding, even if life-threatening bleeding is rare. Initial treatment is required in more than 80% of patients and only 37% of patients achieve cure at one year of follow up with no need for disease-modifying treatment, indicating that ITP is a chronic disease in adults with a less favorable outcome compared to children. The predictors of chronicity were explored prospectively for the first time in adults, suggesting that higher platelet counts at diagnosis are negatively correlated with the risk of chronicity. However, further research is needed to confirm this finding and potentially to identify other predictive factors for chronicity at the time of diagnosis.

pura: epidemiology and implications for patients. Eur J Haematol Suppl. 2009;(71):37. Bennett D, Hodgson ME, Shukla A, Logie JW. Prevalence of diagnosed adult immune thrombocytopenia in the United Kingdom. Adv Ther. 2011;28(12):1096-1104. Moulis G, Palmaro A, Montastruc JL, Godeau B, Lapeyre-Mestre M, Sailler L. Epidemiology of incident immune thrombocytopenia: a nationwide populationbased study in France. Blood. 2014;124(22):3308-3315. Fogarty PF, Segal JB. The epidemiology of immune thrombocytopenic purpura. Curr Opin Hematol. 2007;14(5):515-519. Zeller B, Rajantie J, Hedlund-Treutiger I, et al; NOPHO ITP. Childhood idiopathic thrombocytopenic purpura in the Nordic

countries: epidemiology and predictors of chronic disease. Acta Paediatr. 2005;94 (2):178-184. 9. Bansal D, Bhamare TA, Trehan A, Ahluwalia J, Varma N, Marwaha RK. Outcome of chronic idiopathic thrombocytopenic purpura in children. Blood. 2014;124(22):3295-307. 10. Heitink-Pollé KM, Nijsten J, Boonacker CW, de Haas M, Bruin MC. Clinical and laboratory predictors of chronic immune thrombocytopenia in children: a systematic review and meta-analysis. Blood. 2014;124(22):3295-3307. 11. Grimaldi D, Canouï-Poitrine F, Croisille L, et al. Antiplatelet antibodies detected by the MAIPA assay in newly diagnosed immune thrombocytopenia are associated with chronic outcome and higher risk of bleeding.

haematologica | 2016; 101(9)

Outcome of immune thrombocytopenia in adults Ann Hematol. 2014;93(2):309-315. 12. Patel AP. Idiopathic autoimmune thrombocytopenia and neutropenia in siblings. Eur J Haematol. 2002;69(2):120-121. 13. Rischewski JR, Imbach P, Paulussen M, Kühne T. Idiopathic thrombocytopenic purpura (ITP): is there a genetic predisposition? Pediatr Blood Cancer. 2006;47(5 Suppl):678-680. 14. Grimaldi-Bensouda L, Michel M, Aubrun E et al; PGRx Immune Thrombocytopenia Study Group. A case-control study to assess the risk of immune thrombocytopenia associated with vaccines. Blood. 2012;120(25):4938-4944. 15. Grimaldi-Bensouda L, Guillemot D, Godeau B, et al.; PGRx-AID Study Group. Autoimmune disorders and quadrivalent human papillomavirus vaccination of young female subjects. J Intern Med. 2014;275(4):398-408. 16. Rodeghiero F1, Stasi R, Gernsheimer T, et al. Standardization of terminology, definitions and outcome criteria in immune thrombocytopenic purpura of adults and children: report from an international working group. Blood. 2009;113(11):2386-2393. 17. Miyakis S, Lockshin MD, Atsumi T, et al.

haematologica | 2016; 101(9)

18. 19.

20.

21.

22.

International consensus statement on an update of the classification criteria for definite antiphospholipid syndrome (APS). J Thromb Haemost. 2006;4(2):295-306. The SAS Institute Inc., SAS software: version 9.3. 2008: Cary, NC, USA. Vianelli N, Valdrè L, Fiacchini M, et al. Long-term follow-up of idiopathic thrombocytopenic purpura in 310 patients. Haematologica. 2001;86(5):504-509. Frederiksen, H, Schmidt K. The incidence of idiopathic thrombocytopenic purpura in adults increases with age. Blood. 1999; 94(3):909-913. Kühne T, Berchtold W, Michaels LA, et al; Intercontinental Cooperative ITP Study Group. Newly diagnosed immune thrombocytopenia in children and adults: a comparative prospective observational registry of the Intercontinental Cooperative Immune Thrombocytopenia Study Group. Haematologica. 2011;96(12):1831-1837. Michel M, Suzan F, Adoue D, et al. Management of immune thrombocytopenia in adults: a population-based analysis of the French hospital discharge database from 2009 to 2012. Br J Haematol. 2015;170(2):218-222.

23. Andrès E, Mecili M, Fothergill H, Zimmer J, Vogel T, Maloisel F. Gender-related analysis of the clinical presentation, treatment response and outcome in patients with immune thrombocytopenia. Presse Med. 2012;41(9 Pt 1):e426-431. 24. Abbasi SY, Milhem M, Zaru L. A positive antinuclear antibody test predicts for a poor response to initial steroid therapy in adults with idiopathic thrombocytopenic purpura. Ann Hematol. 2008;87(6):459-462. 25. Altintas A, Ozel A, Okur N, et al. Prevalence and clinical significance of elevated antinuclear antibody test in children and adult patients with idiopathic thrombocytopenic purpura. J Thromb Thrombolysis. 2007;24(2):163-168. 26. Vantelon JM, Godeau B, André C, Bierling P. Screening for autoimmune markers is unnecessary during follow-up of adults with autoimmune thrombocytopenic purpura and no autoimmune markers at onset. Thromb Haemost. 2000;83(1):42-45. 27. 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.

1045

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Coagulation & Its Disorders

Ferrata Storti Foundation

Joint effects of cancer and variants in the factor 5 gene on the risk of venous thromboembolism

Olga V. Gran,1 Erin N. Smith,2,3,4 Sigrid K. Brækkan,1,5 Hilde Jensvoll,1,5 Terry Solomon,6 Kristian Hindberg,1 Tom Wilsgaard,7 Frits R. Rosendaal,1,8 Kelly A. Frazer,1,2 and John-Bjarne Hansen1,5

K.G. Jebsen Thrombosis Research and Expertise Center (TREC), Department of Clinical Medicine, UiT -The Arctic University of Norway, Tromsø, Norway; 2Department of Pediatrics and Rady’s Children’s Hospital, University of California, San Diego, La Jolla, CA, USA; 3 Moores Cancer Center, University of California, San Diego, La Jolla, CA, USA; 4Institute for Genomic Medicine, University of California, San Diego, La Jolla, CA, USA; 5Division of Internal Medicine, University Hospital of North Norway, Tromsø, Norway; 6Biomedical Sciences Graduate Program, University of California, San Diego, La Jolla, CA, USA; 7 Department of Community Medicine, UiT - The Arctic University of Norway, Tromsø, Norway; and 8Department of Clinical Epidemiology, Leiden University Medical Center, The Netherlands 1

Haematologica 2016 Volume 101(9):1046-1053

ABSTRACT

Correspondence: olga.gran@uit.no

Received: April 5, 2016. Accepted: June 10, 2016. Pre-published: June 16, 2016. doi:10.3324/haematol.2016.147405

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1046

1046

enous thromboembolism occurs frequently in cancer patients. Two variants in the factor 5 gene (F5), rs6025 encoding for the factor V Leiden mutation R506Q, and rs4524 encoding K858R, have been found to be associated with venous thromboembolism. We assessed the joint effect of active cancer and these two F5 variants on venous thromboembolism risk in a case-cohort study. Cases with a first venous thromboembolism (n=609) and a randomly selected age-weighted cohort (n=1,691) were sampled from the general population in Tromsø, Norway. Venous thromboembolism was classified as cancerrelated if it occurred in the period 6 months before to 2 years after a diagnosis of cancer. Active cancer was associated with an 8.9-fold higher risk of venous thromboembolism (95% CI 7.2-10.9). The risk of cancer-related venous thromboembolism was 16.7-fold (95% CI 9.9-28.0) higher in subjects heterozygous for rs6025 compared with non-carriers of this variant without active cancer. In subjects with active cancer the risk of venous thromboembolism was 15.9-fold higher (95% CI 9.1-27.9) in those with one risk allele at rs4524, and 21.1-fold (95% CI 12.4-35.8) higher in those with two risk alleles compared with non-carriers without active cancer. A synergistic interaction was observed between active cancer and factor V Leiden (relative excess risk due to interaction 7.0; 95% CI 0.5-14.4) and rs4524 (relative excess risk due to interaction 15.0; 95% CI 7.5 -29.2). The incidence of venous thromboembolism during the initial 6 months following a diagnosis of cancer was particularly high in subjects with risk alleles at these loci. This implies that the combination of cancer and F5 variants synergistically increases venous thromboembolism risk.

Introduction Venous thromboembolism (VTE), consisting of deep vein thrombosis (DVT) and pulmonary embolism (PE), is a common and potentially fatal condition.1 The overall incidence is 1-2 per 1,000 person-years in adults and increases sharply with age.2,3 The association between VTE and cancer is well established and dates back to the 19th century when Bouillaud4 and later Trousseau5 described a relationship between certain malignancies, hypercoagulability and thrombosis. Cancer is associated with a 4- to 7-fold increased risk of VTE, and it is estimated that 15% of cancer patients haematologica | 2016; 101(9)

F5 variants and cancer-related VTE

develop symptomatic VTE during the course of their disease.6-8 The risk of death in cancer patients with VTE is two times greater than in those with cancer alone, and cancer-related VTE is the second leading cause of death in cancer patients.9-11 Family and twin studies have shown that 60% of the VTE risk can be attributed to genetic factors.12,13 In the last two decades, knowledge in this field has expanded and several single nucleotide polymorphisms have been identified that, if present, can affect an individual’s risk of VTE.14 These polymorphisms are single base-pair variations that have the potential to alter the function (e.g. factor V Leiden) and plasma levels (e.g. prothrombin 20210A) of proteins and are predominately located in or near genes encoding for proteins in the coagulation or fibrinolytic pathways.15 Activated factor V (FV) acts as an important cofactor in the coagulation cascade by facilitating conversion of prothrombin to thrombin and by promoting degradation of activated factor VIII together with activated protein C (APC) and protein S.16,17 Variations in the factor 5 gene (F5) commonly result in attenuated down-regulation of activated FV by APC.18 The rs4524 single nucleotide polymorphism (169542517T>C) is the result of a missense mutation (lysine to arginine) at position 2684 in the B-domain of F5 which is thought to result in APC resistance.19 A missense mutation (arginine to glutamine) at position 506 results in the well-known factor V Leiden (FVL) (1691G>A, rs6025) single nucleotide polymorphism. FVL is pro-thrombotic by two mechanisms: its APC-resistant properties and its abnormal factor VIII degradation by APC.18 Cohort studies have found that among white Americans and Europeans approximately 5-8% of the population are heterozygous for FVL.20-23 In recent years, major advances have been made in understanding the role of genetic factors in the risk of VTE. However, the majority of genetic studies have excluded individuals with cancer-related thrombosis, and the few studies that have been performed on inherited risk factors for VTE (e.g. FVL, prothrombin mutation) in cancer patients have shown inconsistent and conflicting results, likely because of insufficient statistical power due to small populations of patients.24-27 As a consequence, knowledge on the biological interaction between cancer and inherited risk factors for VTE is limited. The aim of the present study was to investigate the joint effect of inherited risk variants of F5, FVL and rs4524, and active cancer on the absolute and relative risks of VTE in a population-based case-cohort study.

From the date of inclusion in the Tromsø 4 (1994/95) or Tromsø 6 (2007/08) survey until the end of follow-up on December 31, 2012, there were 660 incident VTE events. The individuals who had these events were cases in our study. A sub-cohort of ageweighted subjects (n=1,793) was randomly selected from the fourth and sixth surveys of the Tromsø study (n=29,128). Subjects with a previous history of cancer (n=152) or with missing values for FVL (n=5) and F5 rs4524 (n=13) variants were excluded. Our final case-cohort therefore consisted of a total of 2,300 subjects: 609 VTE cases and 1,691 subjects in the sub-cohort. The constitution of the case-cohort study is summarized in Figure 1. This study was approved by the Regional Committee for Medical and Health Research Ethics in Northern Norway (REC North) and all participants provided informed written consent to participation.

Baseline measurements Baseline measures are described in the Online Supplementary Methods.

Sequenom genotyping and quality control Genotyping methods are described in the Online Supplementary Methods.

Cancer assessment Cancer assessment is described in the Online Supplementary Methods.

Definition of active cancer Temporal proximity to cancer diagnosis is a strong predictor of VTE risk. Our study and previous studies have found that up to 50% of cancer-related VTE events occur in the 2.5-year period starting from 6 months prior to a diagnosis of cancer until 2 years after the diagnosis of cancer.29-31 The risk of VTE by time since can-

Methods Study population Participants were recruited from the fourth (n=27,158) and sixth (n=12,984) surveys of the Tromsø Study, a single-center population-based cohort study, conducted in 1994-1995 and 2007-2008, respectively. All (Tromsø 4) or parts (Tromsø 6) of the population of the Tromsø municipality over the age of 24 were invited to participate in these surveys. The attendance rate was high, with 77% of the eligible population participating in Tromsø 4 and 66% participating in Tromsø 6. A total of 29,128 subjects aged 25 to 97 years participated in at least one of the studies. A detailed description of the Tromsø Study has previously been published elsewhere.28 The registration and validation of VTE is described in the Online Supplementary Methods. haematologica | 2016; 101(9)

Figure 1. Flowchart illustrating the composition of the case–cohort study.

1047

O.V. Gran et al.

cer diagnosis in our cohort is illustrated in Figure 2. For the purpose of our study, a VTE event was classified as related to active cancer if it occurred within 6 months prior to a diagnosis of cancer until 2 years following the cancer diagnosis. Extending the observation period of cancer increases the likelihood of dilution due to inclusion of VTE not necessarily caused by cancer. Following the period of active cancer, subjects with cancer were classified as having a previous cancer, as the risk of VTE remains marginally increased even after the period of active cancer.

cohort (n=29,128) as a basis for the calculation of absolute risks. Methods for assessing synergism between the F5 variants and active cancer on VTE risk are described in detail in the Online Supplementary Methods. We used Kaplan-Meier curves to estimate cumulative incidence to illustrate time to VTE in subjects with active cancer and the presence of risk alleles. The Fine-Gray model32 was applied for sensitivity analysis to account for mortality as a competing event.

Statistical analysis

Results

The statistical analysis was carried out using STATA version 13.0 (Stata Corporation, College Station, TX, USA). Cox proportional hazards regression models were used to calculate age- and sex-adjusted hazard ratios (HR) with 95% confidence interval (CI) for VTE across categories of cancer (no, active and previous) and gene variants. Subjects were considered heterozygous if one risk allele was present at the locus of interest and homozygous if two risk alleles were present. Cancer-free subjects without the risk alleles were used as the reference group. In addition to calculating hazard ratios for VTE, sub-group analyses were performed separately for DVT and PE. The proportional hazards assumption was confirmed by the use of the Schoenfeld global test. Cancer was entered as a time-varying co-variate in the model. The data were split with respect to the date of cancer diagnosis to distinguish those related to active and previous cancer. Hence, subjects who developed cancer during follow-up contributed with “non-exposed” person-time from the baseline inclusion date to 6 months before cancer diagnosis, with “active cancer exposed” person-time during the active cancer period (-6 to 24 months around the date of cancer diagnosis), and with “previous cancer exposed” person-time in the period >24 months following a diagnosis of cancer. We used the number of person-years from the original

Table 1. Baseline characteristics of the study population according to cancer status.

Subjects Age (years) Sex (males) BMI (kg/m2) Daily smoking Physical activity Self-reported DM Self-reported CVD WBC count (109/L) Platelet count (109/L) VTE DVT PE F5 rs6025_t (FVL)* 1 risk allele 2 risk alleles F5 rs4524_t* 1 risk allele 2 risk alleles

Entire Case-Cohort

Previous Cancer

Active Cancer

2300 58 ±14 46.9 (1079) 26.0 ±4.2 34.7 (799) 22.3 (510) 3.1 (71) 11.7 (269) 7.0 ±1.8 253 ±63 609 343 266 0.04 188 3 0.73 899 1233

201 61 ±10 49.8 (100) 25.9 ±3.9 37.8 (76) 23.9 (48) 3.0 (6) 9.5 (19) 7.0 ±1.7 252 ±60 8.7 (53) 8.5 (29) 9.0 (24) 0.04 15 0 0.69 83 98

461 61 ±11 49.7 (229) 25.7 ±4.1 42.7 (197) 19.6 (90) 3.9 (18) 10.8 (50) 7.2 ±1.8 252 ±62 19.2 (117) 19.5 (67) 18.8 (50) 0.04 39 0 0.72 190 235

Values are numbers or percentages with numbers or means ±SD in parentheses. Active cancer: period from 6 months before a cancer diagnosis until 2 years after; Previous cancer: 2 years or more following a cancer diagnosis. BMI: body mass index; DM: diabetes mellitus; CVD: cardiovascular disease; WBC: white blood cell count; *allele frequency.

1048

There were 461 subjects diagnosed with cancer during a median of 12.9 years follow-up. The distribution of cancer by cancer site is presented in Online Supplementary Table S1. The mean age at cancer diagnosis was 69 years. The baseline characteristics of subjects according to cancer status are presented in Table 1. The median age of all subjects was 59 years. The proportion of current smokers was higher in the cancer population (41.2% versus 34.7%). Of the 609 subjects with VTE, 343 (56.3%) had DVT and 266 (43.7%) had PE; 117 of the VTE events were cancer-related. A VTE event was classified as related to active cancer if it occurred within 6 months prior to the diagnosis of cancer until 2 years following a cancer diagnosis. There were 53 events in the previous cancer group and the remaining 439 events occurred in the non-cancer group. Allele frequencies for FVL and F5 rs4524 are presented in Table 1. The FVL variant was heterozygous in 188 of 2,300 subjects (8.2%) and homozygous in three (0.1%). Similar to a previously cited reference population (allele frequency 0.05),33 the overall allele frequency of FVL was 0.042 in our cohort. The allele frequency was substantially higher in cases than in the sub-cohort (0.073 versus 0.031, respectively). The allele frequency of the F5 rs4524 variant was 0.732 in our population, which is similar to that in a reference population of European ancestry (0.736).34 The frequency of this allele was also higher in VTE cases than in the sub-cohort (0.775 versus 0.732, respectively). There were 899 heterozygous individuals (39.1%) and 1,233

Figure 2. Hazard ratios for VTE at 6-month intervals starting from 1 year before cancer diagnosis.

haematologica | 2016; 101(9)

F5 variants and cancer-related VTE

homozygous individuals (53.6%). With the exception of FVL and urological cancers (allele frequency 0.104, P=0.009) and FVL and pancreatic cancer (allele frequency 0.167, P=0.001), which has been previously described,35 the allele frequencies of FVL and F5 rs4524 in subjects with various different types of cancer were similar to those in non-cancer subjects. Active cancer was associated with an overall 8.9-fold (95% CI 7.2-10.9) higher risk of VTE, with an age- and sex-adjusted hazard ratio of 8.5 (95% CI 6.9-10.5). In subjects with active cancer, the risk of DVT was increased 9.4-fold (95% CI 7.1-12.3) and the risk of PE 8.3-fold (95% CI 6.1-11.4), when compared to the risks in subjects without active cancer. Based on person-year information from the full cohort, the incidence rate of VTE was 1.6 per 1,000 person-years in the total population and 13.6 per 1,000 person-years in subjects with active cancer. The incidence rate of VTE was 29.4 per 1,000 person-years in subjects with FVL and active cancer, and 12.5 and 16.7 per 1,000 person-years in subjects with active cancer and one or two risk alleles at F5 rs4524, respectively. The risk estimates for VTE by categories of FVL and active cancer are shown in Table 2. Among the subjects with active cancer, 39 (8.5%) were heterozygous for FVL; there were no homozygous individuals. Among the noncancer subjects, the risk of VTE was increased 2.1-fold (95% CI 1.6-2.7) in the presence of one FVL risk allele and 3.3-fold (95% CI 0.8-13.2) in the presence of two risk alleles. The risk of an active cancer-related VTE was considerably higher in subjects heterozygous for FVL (HR 16.7, 95% CI 9.9-28.0) than in non-carriers without active cancer. The risk of VTE in heterozygous subjects was increased 3.4-fold (95% CI 1.7-6.8) in the group with previous cancer. Heterozygous individuals with active cancer

had a 25.3-fold (95% CI 14.1-45.4) higher risk of DVT and 7-fold (95% CI 2.2-22.0) higher risk of PE. When non-carriers with active cancer were used as the reference group, there was a 1.9-fold (95% CI 1.1-3.3) and a 2.7-fold (95% CI 1.5-5.1) higher risk of VTE and DVT, respectively, in those with one FVL risk allele and active cancer (Table 3). When adjusting for the F5 rs4524 single nucleotide polymorphism in the model, the risk estimates were essentially the same (data not shown). The additive interaction (relative excess risk due to interaction, RERI) and attributable proportion due to interaction (AP) between FVL and active cancer on VTE risk are shown in Table 4. We observed a supra-additive interaction between FVL and active cancer for VTE risk (RERI; 7.0 95% CI 0.5-14.4), demonstrating a synergistic effect, as the joint effect of active cancer and FVL was stronger than the sum of the individual effects. The AP was 0.42, demonstrating that 42% of the cases in the combined group could be attributed to the interaction between the two exposures. This synergistic effect was driven mainly by the synergistic interaction on DVT risk, as a positive additive interaction was not seen for PE in subjects heterozygous for FVL with active cancer. An interaction was not, however, observed on a multiplicative scale, when fitting the interaction terms into the Cox regression model (HR 0.9, 95% CI 0.5-1.7). In cancer-free subjects, the risk of VTE in individuals with the F5 rs4524 variant was similar to that of subjects with FVL with a 1.9-fold (95% CI 1.2-3.1) and 2.3-fold (95% CI 1.4-3.8) higher VTE risk in subjects with one and two risk alleles, respectively, compared to non-carriers (Table 2). In the presence of active cancer, the risk of VTE increased from 15.9-fold (95% CI 9.1-27.9) in F5 rs4524 heterozygous subjects to 21.1-fold (95% 12.4-35.8) in homozygous subjects. The risk of DVT was higher across

Table 2. Age- and sex-adjusted hazard ratios for VTE according to categories of F5 rs6025 (FVL) and F5 rs4524 risk alleles and cancer.

Risk Alleles

VTE HR (95% CI)

DVT HR (95% CI)

PE HR (95% CI)

0 1 2 0 1 2 0 1 2

374 63 2 46 7 0 102 15 0

1.0 2.1 (1.6-2.7) 3.3 (0.8-13.2) 2.4 (1.8-3.2) 3.4 (1.7-6.8) 8.6 (6.9-10.8) 16.7 (9.9-28.0) -

198 47 2 26 3 0 55 12 0

1.0 2.9 (2.1-4.0) 6.4 (1.6-25.9) 2.7 (1.8-4.2) 2.8 (0.9-8.7) 9.3 (6.8-12.5) 25.3 (14.1-45.4) -

176 16 0 20 4 0 47 3 0

1.0 1.5 (0.7-1.9) 1.8 (1.1-2.8) 3.3 (1.2-8.8) 7.9 (5.7-11.0) 7.0 (2.2-22.0) -

0 1 2 0 1 2 0 1 2

17 160 262 3 24 26 3 44 70

1.0 1.9 (1.2-3.1) 2.3 (1.4-3.8) 2.7 (0.8-9.3) 5.0 (2.7-9.3) 4.3 (2.3-7.9) 4.8 (1.4-16.5) 15.9 (9.1-27.9) 21.1 (12.4-35.8)

7 93 147 2 13 14 2 26 39

1.0 2.7 (1.2-5.8) 3.1 (1.5-6.6) 4.9 (1.0-23.6) 7.5 (3.0-19.0) 6.5 (2.6-16.1) 8.5 (1.8-41.0) 23.9 (10.4-55.3) 29.7 (13.3-66.6)

10 67 115 1 11 12 1 18 31

1.0 1.4 (0.7-2.6) 1.7 (0.9-3.3) 1.4 (0.2-10.7) 3.3 (1.4-7.8) 2.9 (1.2-6.7) 2.5 (0.3-19.2) 10.4 (4.8-22.6) 15.1 (7.4-30.9)

FVL No cancer

Previous cancer

Active cancer

F5 rs4524 No cancer

Previous cancer

Active cancer

Active cancer: period from 6 months before a cancer diagnosis until 2 years after: Previous cancer: 2 years or more following a cancer diagnosis. N: number of events (VTE, DVT, PE) in each category.

haematologica | 2016; 101(9)

1049

O.V. Gran et al.

all categories of gene variants and cancer with the hazard ratios increasing to 23.9-fold (95% CI 10.4-55.3) in subjects with one risk allele and to 29.7-fold (95% CI 13.366.6) in those with two risk alleles. In subjects with a previous cancer, the risk of a VTE was 5-fold (95% CI 2.7-9.3) higher in F5 rs4524 heterozygous subjects and 4.3-fold (95% CI 2.3-7.9) higher in homozygous subjects. The risk estimates were 4.4-fold (95% CI 1.4-13.9) and 3.5-fold (95% CI 0.8-14.5) higher for VTE and DVT, respectively, in homozygous subjects with active cancer when non-carriers with active cancer were used as the reference group (Table 3). When adjusting for FVL, the risk estimates remained essentially similar (data not shown). We observed a supra-additive interaction (RERI; 15 95% CI 7.14-22.81) between active cancer and the single nucleotide polymorphism in F5 rs4524, demonstrating a strong synergistic effect, which was supported by an AP value of 0.71 (Table 4). Interaction on a multiplicative scale was not significant (HR 2.0, 95% CI 0.6-7.1). Based on the entire TromsĂ¸ Study cohort (our source population for the case-cohort), we estimated that 4.2% of the subjects with cancer developed a VTE during the active cancer period. The cumulative incidence of VTE during the active cancer period in subjects without and with risk alleles is shown in Figure 3. There was a notable increase in the cumulative incidence of VTE during the initial 6 months following a cancer diagnosis, with a substantially steeper slope in the incidence curve for carriers com-

Table 3. Age- and sex-adjusted hazard ratios for VTE by categories of FVL and F5 rs4524 risk alleles among those with active cancer.

Risk Alleles Venous thromboembolism 0 1 2 Deep vein thrombosis 0 1 2 Pulmonary embolism 0 1 2

FVL HR (95% CI)

F5 rs4524 HR (95% CI)

1.0 1.9 (1.1 â&#x20AC;&#x201C; 3.3) -

1.0 3.3 (1.0-10.7) 4.4 (1.4-13.9)

1.0 2.7 (1.5-5.1) -

1.0 2.8 (0.7-11.9) 3.5 (0.8-14.5)

1.0 0.9 (0.3-2.9) -

1.0 4.2 (0.6-31.8) 6.1 (0.8-45.1)

pared to that of non-carriers of the risk alleles (Figure 3, panels A and B). The majority of the VTE events occurred during the first 6 months after a cancer diagnosis. The cumulative incidence of VTE was 10.3% in the period 0-6 months before the cancer diagnosis, and increased to 29.4% at 6 months after the cancer diagnosis for those heterozygous for FVL. Likewise, the cumulative incidence of VTE increased from 6.42% in the period 0-6 months before the cancer diagnosis to 20.2% at 6 months after the cancer diagnosis for those homozygous for F5 rs4524. The cumulative incidence of VTE at the end of the active cancer period was notably higher for individuals with one FVL allele and individuals with one or two F5 rs4524 alleles compared to those without a mutation. Finally, we tested the impact of competing risk by death on our risk estimates for VTE by applying the Fine-Gray model. The hazard ratios and sub-distribution hazard ratios (SHR), as estimates of the relative risk of overall VTE, were nearly identical when comparing the presence of risk alleles in subjects with active cancer to that in subjects with the wild-type allele and active cancer: HR 1.9 (95% CI 1.1-3.3) versus SHR 1.9 (95% CI 0.9-3.9) for FVL, and HR 3.5 (95% CI 0.8-11.9) versus SHR 3.9 (95% CI 1.113.2) for F5 rs4524.

HR: hazard ratio; CI: confidence interval.

Table 4. Additive interaction between F5 variants and active cancer.

Relative excess risk due to interaction (RERI)

Proportion due to interaction (AP)

7.0 (95 % CI (0.5-14.4) 14.1 (95% CI 2.83-26.48) -1.4 (95% CI -8.39-8.37)

0.42 0.56 -0.2

15.0 (95% CI 7.14-22.81) 19.10 (95% CI 4.65-60.77) 11.90 (95% CI 6.15-42.61)

0.71 0.64 0.78

FVL VTE DVT PE

F5 r s4524 VTE DVT PE

Relative excess risk by interaction >0 indicates a positive additive interaction. Proportion due to interaction (AP) >0 indicates a positive additive interaction.

1050

Figure 4. Cumulative incidence of VTE in the presence of F5 rs6025 (FVL) and F5 rs4524 risk alleles during the active cancer period. (A) Kaplan-Meier estimates of cumulative incidence in the presence of F5 rs6025 (FVL) risk alleles during the active cancer period. (B) Kaplan-Meier estimates of cumulative incidence in the presences of F5 rs4524 risk alleles during the active cancer period.

haematologica | 2016; 101(9)

F5 variants and cancer-related VTE

Discussion We found that both F5 variants (rs4524 and FVL) were associated with a higher risk of VTE, and substantially so in individuals with active cancer. On an additive scale, there was a synergistic effect of active cancer and the F5 rs4524 and FVL variants on the risk of VTE. The risk of VTE increased for each additional risk allele. The incidence of VTE was highest during the first 6 months following the cancer diagnosis, which is likely due to cancer treatment-related factors, such as surgery, chemotherapy, acute infections, and central venous catheters. Previous studies determined that the presence of F5 rs4524 risk alleles is associated with a moderately higher risk (26-67%) of DVT in middle-aged populations,36 of VTE in post-menopausal women,37 and of VTE in the antenatal period.38 The risk of VTE increased weakly in an allele-dependent manner and risk estimates were higher in post-menopausal women with additional risk factors.37 Accordingly, we found that F5 rs4524 in individuals without cancer was associated with a moderate allele-dependent increased risk of VTE. To date, there have been, to the best of our knowledge, no other studies exploring the impact of F5 rs4524 on cancer-related VTE. In our study, active cancer had a synergistic effect with F5 rs4524 on the risk of VTE. The combined effect was stronger for DVT than for PE. The genetic risk difference between DVT and PE appears to follow the same pattern as has been previously described for FVL.39 We observed that the presence of F5 rs4524 had a particularly strong impact on the incidence of VTE during the first 6 months after the diagnosis of cancer. Active cancer displayed synergism, on an additive scale, with FVL on the VTE risk, an effect mainly driven by the impact of the interaction on DVT risk. Previously, few studies had investigated the role of FVL in cancer-related VTE. Initial results from small cancer cohorts, including 75-353 patients, reported either no40,41 or a 2- to 4.4-fold higher risk24,42 of VTE in cancer patients with FVL compared to those without the FVL mutation. In a large population-based VTE case-control study with cancer diagnoses registered during 5 years prior to the inclusion date, Blom and colleagues found a 2.4-fold and 12-fold higher VTE risk in individuals with FVL and cancer compared to those with cancer without the mutation, and those without cancer and the mutation, respectively.25 The latter findings support our observation of a joint effect of cancer and FVL on VTE risk. In a larger cohort of 982 cancer patients, Pabinger and colleagues reported a 2.0-fold (95% CI 1.03.97) higher risk of cancer-related VTE among subjects with FVL.43 In agreement with our findings, they described that the presence of the FVL had a particular strong impact on the incidence of VTE during the first 6 months after the cancer diagnosis. Interactions can be statistical or biological. A statistical interaction refers to the departure from the underlying form of a statistical model, and for regression analysis, it is normally assessed by entering a product term into the regression model. A biological interaction means that two causes are both required to precipitate disease; i.e. that the two causes are component causes in the same causal model (the effect of one is biologically dependent on the presence of another). A biological interaction is not dependent on the underlying statistical model as it always refers to departure from additivity.44,45 The underlying haematologica | 2016; 101(9)

mechanisms for the synergistic effect of cancer and mutations at sites rs6025 and rs4524 in the F5 gene on VTE risk may involve factors related to APC resistance. Both FVL and F5 rs4524 have been reported to exhibit their prothrombotic actions by attenuated down-regulation of activated FV by APC.19 Previous studies have also found an increased prevalence of acquired APC resistance in patients with cancer.46-48 It is licit to assume that two sources of APC resistance (i.e. acquired and inherited) in cancer patients would greatly increase the risk of a VTE, similarly to the synergistic effect of oral contraceptive use in individuals with FVL, as both result in a poor response to APC.49,50 Our findings of a pronounced increased risk of VTE during the first 6 months after cancer diagnosis in subjects with risk alleles fit well with the thrombosis potential model.51 This model illustrates how individual risk factors alone may be insufficient to trigger VTE. Inherited risk factors, such as FVL and rs4524, alone only mildly increased the VTE risk. Cancer itself is a strong provoking factor for VTE, and in the presence of inherited risk factors the thrombosis potential is further increased. Additional treatment-related provoking factors such as surgery, chemotherapy, and central venous catheters along with treatment-related complications including acute infections and immobilization, further elevate the risk of VTE. Accordingly, VTE occurs when the thrombosis potential exceeds the thrombosis threshold as a consequence of additive or synergistic effects of accumulated risk factors. These aggregated conditions may explain why, in individuals with these risk alleles, there is such a considerable increase in the number of VTE events in the initial months following a cancer diagnosis. According to a recent Cochrane review, thromboprophylaxis with low molecular weight heparin given to ambulatory cancer patients receiving chemotherapy lowered the incidence of symptomatic VTE by 47% at the expense of a non-significant increase in major bleeding (30%).52 Even though cancer patients are at a high risk of VTE, and cancer patients with VTE have shortened life expectancy,8,10 current international guidelines do not recommend medical thromboprophylaxis for ambulatory cancer patients without additional risk factors due to an uncertain benefit to harm (i.e. bleeding risk) ratio.53,54 It is, therefore, very important to recognize individuals at high risk of cancer-related VTE in order to identify those patients who would benefit from prophylaxis. Current risk prediction models,55,56 are based on cancer localization and laboratory parameters influenced by several factors (i.e. infection/inflammation, surgery, various medications, and dehydration). A validation study revealed that these risk prediction models may have limited clinical utility because of their low potential to identify patients in the high-risk category (12% of the entire cancer cohort study population) and inadequate capacity to predict VTE in the high-risk subjects, as only 7% of the high-risk subjects developed a VTE over 2.5 months.55 As risk alleles for VTE in the F5 gene are common, exhibit synergistic effects with active cancer on VTE risk, and particularly discriminate patients at risk during the first 6 months after cancer diagnosis, these may be attractive candidates to pursue in future research on prediction models of VTE risk in cancer patients. A major strength of our study is its prospective design with subjects recruited from a general population. The high participation rate in the TromsĂ¸ study and the broad 1051

O.V. Gran et al.

age range formed a cohort that is representative of the general population and also eliminated selection bias in the sub-cohort. The long duration of follow-up enabled us to capture a large quantity of cancer events in the study population. Furthermore, all VTE and cancer events were objectively confirmed and systematically validated. The main limitation of our study was insufficient power for subgroup analysis of VTE (i.e. PE) in individuals with active cancer, demonstrated by the wide confidence intervals for our risk estimates in these categories. Secondly, information regarding cancer treatment modalities was unfortunately not available. Such data could have provid-

References 1. Heit JA. Venous thromboembolism: disease burden, outcomes and risk factors. J Thromb Haemost. 2005;3(8):1611-1617. 2. Naess IA, Christiansen SC, Romundstad P, Cannegieter SC, Rosendaal FR, Hammerstrom J. Incidence and mortality of venous thrombosis: a population-based study. J Thromb Haemost. 2007;5(4):692699. 3. Cushman M, Tsai AW, White RH, et al. Deep vein thrombosis and pulmonary embolism in two cohorts: the longitudinal investigation of thromboembolism etiology. Am J Med. 2004;117(1):19-25. 4. Bouillaud S. De l’obliteration des veines et de son influence sur la formation des hydropisies partielles: considerations sur l’hydropisie passive et generale. Arch Gen Med. 1823;1:188-204. 5. Trousseau A. CJt. Lectures on clinical medicine: Delivered at the Hotel-Dieu, Paris. The New Syndenham Society 1872. 6. Agnelli G, Verso M. Thrombosis and cancer: clinical relevance of a dangerous liaison. Haematologica. 2005;90(2):154-156. 7. Heit JA, Silverstein MD, Mohr DN, et al. Risk factors for deep vein thrombosis and pulmonary embolism: a population-based case-control study. Arch Intern Med. 2000;160(6):809-815. 8. Timp JF, Braekkan SK, Versteeg HH, Cannegieter SC. Epidemiology of cancerassociated venous thrombosis. Blood. 2013;122(10):1712-1723. 9. Sorensen HT, Mellemkjaer L, Olsen JH, Baron JA. Prognosis of cancers associated with venous thromboembolism. N Engl J Med. 2000;343(25):1846-1850. 10. Khorana AA, Francis CW, Culakova E, Kuderer NM, Lyman GH. Thromboembolism is a leading cause of death in cancer patients receiving outpatient chemotherapy. J Thromb Haemost. 2007;5(3):632-634. 11. Monreal M, Trujillo-Santos J. Lessons from VTE registries: the RIETE experience. Best Pract Res Clin Haematol. 2009;22(1):25-33. 12. Larsen TB, Sorensen HT, Skytthe A, Johnsen SP, Vaupel JW, Christensen K. Major genetic susceptibility for venous thromboembolism in men: a study of Danish twins. Epidemiology. 2003;14(3): 328-332. 13. Souto JC, Almasy L, Borrell M, et al. Genetic susceptibility to thrombosis and its relationship to physiological risk factors: the GAIT study. Genetic Analysis of Idiopathic Thrombophilia. Am J Hum

1052

ed additional insight into the relationship between genes and treatment-related risk factors. In conclusion, there is a synergistic effect of the FVL and rs4524 single nucleotide polymorphisms and active cancer on the risk of VTE. Future studies should address whether information on the presence of these polymorphisms can improve prediction of VTE in cancer patients. Acknowledgments The K.G. Jebsen Thrombosis Research and Expertise Center is supported by an independent grant from the K.G. Jebsen foundation.

Genet. 2000;67(6):1452-1459. 14. Morange PE, Tregouet DA. Current knowledge on the genetics of incident venous thrombosis. J Thromb Haemost. 2013;11 (Suppl 1):111-121. 15. Heit JA, Cunningham JM, Petterson TM, Armasu SM, Rider DN, de Andrade M. Genetic variation within the anticoagulant, procoagulant, fibrinolytic and innate immunity pathways as risk factors for venous thromboembolism. J Thromb Haemost. 2011;9(6):1133-1142. 16. Mann KG, Kalafatis M. Factor V: a combination of Dr Jekyll and Mr Hyde. Blood. 2003;101(1):20-30. 17. Nicolaes GA, Dahlback B. Factor V and thrombotic disease: description of a Janusfaced protein. Arterioscler Thromb Vasc Biol. 2002;22(4):530-538. 18. Vos HL. Inherited defects of coagulation factor V: the thrombotic side. J Thromb Haemost. 2006;4(1):35-40. 19. Kostka H, Siegert G, Schwarz T, et al. Frequency of polymorphisms in the Bdomain of factor V gene in APC-resistant patients. Thromb Res. 2000;99(6):539-547. 20. Severinsen MT, Overvad K, Johnsen SP, et al. Genetic susceptibility, smoking, obesity and risk of venous thromboembolism. Br J Haematol. 2010;149(2):273-279. 21. Kujovich JL. Factor V Leiden thrombophilia. Genet Med. 2011;13(1):1-16. 22. Sode BF, Allin KH, Dahl M, Gyntelberg F, Nordestgaard BG. Risk of venous thromboembolism and myocardial infarction associated with factor V Leiden and prothrombin mutations and blood type. CMAJ. 2013;185(5):E229-237. 23. Rees DC. The population genetics of factor V Leiden (Arg506Gln). Br J Haematol. 1996;95(4):579-586. 24. Pihusch R, Danzl G, Scholz M, et al. Impact of thrombophilic gene mutations on thrombosis risk in patients with gastrointestinal carcinoma. Cancer. 2002;94(12): 3120-3126. 25. Blom JW, Doggen CM, Osanto S, Rosendaal FR. Malignancies, prothrombotic mutations, and the risk of venous thrombosis. JAMA. 2005;293(6):715-722. 26. Kennedy M, Andreescu ACM, Greenblatt MS, et al. Factor V Leiden, prothrombin 20210A and the risk of venous thrombosis among cancer patients. Br J Haematol. 2005;128(3):386-388. 27. Eroglu A, Ulu A, Cam R, Kurtman C, Akar N. Prevalence of factor V 1691 G-A (Leiden) and prothrombin G20210A polymorphisms and the risk of venous thrombosis among cancer patients. J Thromb

Thrombolysis. 2007;23(1):31-34. 28. Jacobsen BK, Eggen AE, Mathiesen EB, Wilsgaard T, Njølstad I. Cohort profile: the Tromsø study. Int J Epidemiol. 2011;41(4): 961-967. 29. White RH, Chew HK, Zhou H, et al. INcidence of venous thromboembolism in the year before the diagnosis of cancer in 528 693 adults. Arch Intern Med. 2005;165(15):1782-1787. 30. Walker AJ, Card TR, West J, Crooks C, Grainge MJ. Incidence of venous thromboembolism in patients with cancer - a cohort study using linked United Kingdom databases. Eur J Cancer. 2013;49(6):14041413. 31. Blix K, Severinsen M, Braekkan S, et al. Cancer-related venous thromboembolism in the general population: results from the Scandinavian Thrombosis and Cancer (STAC) study. J Thromb Haemost. 2015; 13(Suppl 2):549. 32. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(446):496509. 33. Morange PE, Tregouet DA. Lessons from genome-wide association studies in venous thrombosis. J Thromb Haemost. 2011;9 (Suppl 1):258-264. 34. Germain M, Chasman Daniel I, de Haan H, et al. Meta-analysis of 65,734 Individuals Identifies TSPAN15 and SLC44A2 as two susceptibility loci for venous thromboembolism. Am J Hum Genet. 2015;96(4):532542. 35. Mozsik G, Rumi G, Domotor A, et al. Involvement of serum retinoids and Leiden mutation in patients with esophageal, gastric, liver, pancreatic, and colorectal cancers in Hungary. World J Gastroenterol. 2005;11(48):7646-7650. 36. Bezemer ID, Bare LA, Doggen CJ, et al. Gene variants associated with deep vein thrombosis. JAMA. 2008;299(11):13061314. 37. Smith NL, Hindorff LA, Heckbert SR, et al. Association of genetic variations with nonfatal venous thrombosis in postmenopausal women. JAMA. 2007;297(5):489-498. 38. Dahm AE, Tiscia G, Holmgren A, et al. Genetic variations in the annexin A5 gene and the risk of pregnancy-related venous thrombosis. J Thromb Haemost. 2015;13 (3):409-413. 39. Corral J, Roldán V, Vicente V. Deep venous thrombosis or pulmonary embolism and factor V Leiden: enigma or paradox. Haematologica. 2010;95(6):863-866. 40. Ravin AJ, Edwards RP, A Krohn M, Kelley

haematologica | 2016; 101(9)

F5 variants and cancer-related VTE

41.

42.

43.

44. 45.

JR, Christopherson WA, Roberts JM. The factor V Leiden mutation and the risk of venous thromboembolism in gynecologic oncology patients. Obstet Gynecol. 2002;100(6):1285-1289. Ramacciotti E, Wolosker N, Puech-Leao P, et al. Prevalence of factor V Leiden, FII G20210A, FXIII Val34Leu and MTHFR C677T polymorphisms in cancer patients with and without venous thrombosis. Thromb Res. 2003;109(4):171-174. Otterson GA, Monahan BP, Harold N, Steinberg SM, Frame JN, Kaye FJ. Clinical significance of the FV:Q506 mutation in unselected oncology patients. Am J Med. 1996;101(4):406-412. Pabinger I, Ay C, Dunkler D, et al. Factor V Leiden mutation increases the risk for venous thromboembolism in cancer patients â&#x20AC;&#x201C; results from the Vienna Cancer And Thrombosis Study (CATS). J Thromb Haemost. 2014;13(1):17-22. Rothman KJ, Greenland S, Walker AM. Concepts of Interaction. Am J Epidemiol. 1980;112(4):467-470. Knol MJ, van der Tweel I, Grobbee DE, Numans ME, Geerlings MI. Estimating interaction on an additive scale between continuous determinants in a logistic

haematologica | 2016; 101(9)

46.

47.

48.

49.

50.

51.

regression model. Int J Epidemiol. 2007;36(5): 1111-1118. Haim N, Lanir N, Hoffman R, Haim A, Tsalik M, Brenner B. Acquired activated protein C resistance is common in cancer patients and is associated with venous thromboembolism. Am J Med. 2001;110 (2):91-96. Green D, Maliekel K, Sushko E, Akhtar R, Soff GA. Activated-protein-C resistance in cancer patients. Haemostasis. 1997;27(3):112-118. De Lucia D, De Vita F, Orditura M, et al. Hypercoagulable state in patients with advanced gastrointestinal cancer: evidence for an acquired resistance to activated protein C. Tumori. 1997;83(6):948-952. Vandenbroucke JP, Koster T, Rosendaal FR, BriĂŤt E, Reitsma PH, Bertina RM. Increased risk of venous thrombosis in oral-contraceptive users who are carriers of factor V Leiden mutation. Lancet. 1994;344(8935): 1453-1457. Hellgren M, Svensson PJ, Dahlback B. Resistance to activated protein C as a basis for venous thromboembolism associated with pregnancy and oral contraceptives. Am J Obstet Gynecol. 1995;173(1):210-213. Rosendaal FR. Venous thrombosis: a multi-

52.

53.

54.

55.

56.

causal disease. Lancet. 1999;353(9159): 1167-1173. Di Nisio M, Porreca E, Otten HM, Rutjes AW. Primary prophylaxis for venous thromboembolism in ambulatory cancer patients receiving chemotherapy. Cochrane Database Syst Rev. 2014;(8):CD008500. Kahn SR, Lim W, Dunn AS, et al. Prevention of VTE in nonsurgical patients: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest. 2012;141(Suppl 2):e195S-226S. Lyman GH, Khorana AA, Kuderer NM, et al. Venous thromboembolism prophylaxis and treatment in patients with cancer: American Society of Clinical Oncology clinical practice guideline update. J Clin Oncol. 2013; 31(17):2189-2204. Khorana AA, Kuderer NM, Culakova E, Lyman GH, Francis CW. Development and validation of a predictive model for chemotherapy-associated thrombosis. Blood. 2008;111(10):4902-4907. Ay C, Dunkler D, Marosi C, et al. Prediction of venous thromboembolism in cancer patients. Blood. 2010;116(24):53775382.

1053

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Myeloproliferative Disorders

Ferrata Storti Foundation

A novel role for nuclear factor-erythroid 2 in erythroid maturation by modulation of mitochondrial autophagy

Monika Gothwal,1 Julius Wehrle,1 Konrad Aumann,2 Vanessa Zimmermann,1 Albert GrĂźnder,1 and Heike L. Pahl1 Clinic for Internal Medicine I, Division of Molecular Hematology; and 2Department of Pathology, Medical Center - University of Freiburg, Faculty of Medicine, University of Freiburg, Germany

Haematologica 2016 Volume 101(9):1054-1064

ABSTRACT

Correspondence: pahl@uni-freiburg.de

Received: June 26, 2015. Accepted: June 10, 2016. Pre-published: June 16, 2016.

e have recently demonstrated that the transcription factor nuclear factor-erythroid 2, which is critical for erythroid maturation and globin gene expression, plays an important role in the pathophysiology of myeloproliferative neoplasms. Myeloproliferative neoplasm patients display elevated levels of nuclear factor-erythroid 2 and transgenic mice overexpressing the transcription factor develop myeloproliferative neoplasm, albeit, surprisingly without erythrocytosis. Nuclear factor-erythroid 2 transgenic mice show both a reticulocytosis and a concomitant increase in iron deposits in the spleen, suggesting both enhanced erythrocyte production and increased red blood cell destruction. We therefore hypothesized that elevated nuclear factor-erythroid 2 levels may lead to increased erythrocyte destruction by interfering with organelle clearance during erythroid maturation. We have previously shown that nuclear factor-erythroid 2 overexpression delays erythroid maturation of human hematopoietic stem cells. Here we report that increased nuclear factor-erythroid 2 levels also impede murine maturation by retarding mitochondrial depolarization and delaying mitochondrial elimination. In addition, ribosome autophagy is delayed in transgenics. We demonstrate that the autophagy genes NIX and ULK1 are direct novel nuclear factor-erythroid 2 target genes, as these loci are bound by nuclear factor-erythroid 2 in chromatin immunoprecipitation assays. Moreover, Nix and Ulk1 expression is increased in transgenic mice and in granulocytes from polycythemia vera patients. This is the first report implying a role for nuclear factorerythroid 2 in erythroid maturation by affecting autophagy.

doi:10.3324/haematol.2015.132589

Introduction Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1054

ÂŠ2016 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights reserved to the Ferrata Storti Foundation. Copies of articles are allowed for personal or internal use. Permission in writing from the publisher is required for any other use.

1054

The transcription factor nuclear factor-erythroid 2 (NF-E2) plays an essential role in erythroid maturation, and is a critical regulator of globin gene expression.1 Patients with polycythemia vera (PV), whose lead symptom is erythrocytosis, show significantly elevated NF-E2 levels both in progenitor cells and in mature compartments.2,3 Recently, we have shown that NF-E2 overexpression in a transgenic (tg) mouse model leads to a myeloproliferative neoplasm (MPN) phenotype characterized by thrombocytosis, leukocytosis, the expansion of stem and progenitor compartments and erythropoietin-independent colony formation.2 In addition, NF-E2 transgenic mice presented with elevated reticulocyte counts, indicating an increased erythroid drive. At the same time, however, tg mice show an increase in iron deposits in the spleen as well as elevated lactate dehydrogenase levels, strongly suggesting augmented red blood cell (RBC) destruction.2 Taken together, these two observations could explain the absence of polycythemia in NF-E2 tg mice. Erythroid maturation is a complex multistep process in which the erythroid progenitor cell becomes highly specialized for the transport of oxygen. This entails a vast increase in hemoglobin synthesis followed by the elimination of internal haematologica | 2016; 101(9)

NF-E2 regulates mitophagy

organelles as well as the cell nucleus.4 Elimination of organelles like ribosomes and mitochondria, called autophagy in general, or more specifically ribophagy and mitophagy, respectively, is an essential step in the maturation of reticulocytes to erythrocytes.5 Autophagy is an organized process wherein cellular components and organelles are targeted to lysosomes for degradation.6,7 Autophagy occurs either during homeostasis, where the degradation of damaged or dysfunctional mitochondria is essential to protect cells from reactive oxygen species-mediated damage.8 Alternatively, as mentioned above, programmed, maturation-induced autophagy must occur during terminal erythroid differentiation. Signals triggering the elimination of damaged mitochondria during homeostasis include the oxidation of mitochondrial lipids9 and the loss of mitochondrial membrane potential.10,11 In contrast, the signals and molecular mechanisms mediating mitophagy during the course of normal erythrocyte development are incompletely understood. However, impaired mitophagy has been shown to lead to increased red cell destruction.12,13 Given our observation of increased red cell destruction in NF-E2 overexpressing mice, we hypothesized that this transcription factor may play a novel role in mediating mitochondrial and ribosomal clearance during erythroid maturation.

Methods Mice The establishment of the VAV-HA-hNF-E2 transgenic animals has been described elsewhere.2 Mice of two independent founder lines were analyzed as detailed in Kaufmann et al.2 All animal experiments were performed in compliance with the German animal protection law. The mice were housed and handled in accordance with good animal practice as defined by FELASA and the national animal welfare body GV-SOLAS. The animal welfare committees of the University of Freiburg as well as the local authorities approved all animal experiments.

Ex vivo maturation of reticulocytes and staining Cultures were set up according to Zhang et al.14 Briefly, heparinized whole blood was diluted in maturation medium and cultured at 37Â°C until staining. The cells were stained with 100nM MitoTracker green dye (MTG) (M7514, Invitrogen), Ter119-PE (22155234, ImmunoTools) and CD71-APC (C355, Leinco Technologies) antibodies. Alternatively, cells were stained with thiazole orange (TO) (390062, Sigma-Aldrich), in PBS (2 Îźg/ml) and Ter119 and CD71 antibodies. JC-1 (65-0851-38, eBioscience) staining was done according to Cossarizza et al.15

Statistical analysis Statistical analysis was done using GraphPad Prism software. P-values were determined by either one-tailed or two-tailed Studentâ&#x20AC;&#x2122;s t-tests.

Results NF-E2 overexpression in a tg mouse model led to a significant increase in reticulocyte counts and in the red cell distribution width,2 suggesting an enhanced erythroid drive and increased red cell production. Surprisingly, however, NF-E2 tg mice have a normal hematocrit and normal RBC counts in the peripheral blood (PB).2 We observed increased iron deposits in splenic histiocytes in NF-E2 tg haematologica | 2016; 101(9)

mice, indicating increased red cell destruction,2 which may counteract the increased RBC production, explaining the lack of polycythemia. However, although we expected increased RBC destruction to lead to an increased RBC turnover and a shorter RBC half-life, in fact, NF-E2 tg mice displayed a decreased RBC turnover, manifested by a significantly increased half-life in the PB (Online Supplementary Figure S1). We therefore examined the effect of elevated NF-E2 levels on RBC pathophysiology. Because of the increase in PB reticulocyte counts2 in the absence of increased RBC turnover, we hypothesized that elevated NF-E2 levels perturb the kinetics of RBC maturation by impeding mitochondrial autophagy. As reticulocytes eliminate internal organelles and mature, they progressively lose CD71 expression and gain Ter119 expression.16 We therefore compared the reticulocyte maturation status of NF-E2 tg mice and wild-type (wt) littermates by FACS staining for CD71, Ter119 as well as for mitochondria and ribosome content. Ter119 and CD71 staining distinguishes four stages of erythroid maturation,16 the least mature cells (Ter119low/CD71high), gain Ter119 expression to become Ter119+CD71high. Maturing erythrocytes subsequently lose CD71 expression to become Ter119+/CD71med, and finally Ter119+/CD71low cells. Compared to wt littermates, NF-E2 tg mice display a significant increase in the percentage of immature erythrocytes in both PB and bone marrow (BM), as indicated by the elevated proportion of Ter119low/CD71high, Ter119+/CD71high and Ter119+/CD71med populations (Figure 1A-C). Taken together, these data demonstrate a delay in erythrocyte maturation and an accumulation of immature cell populations in the presence of elevated NF-E2 levels. Since the elimination of mitochondria is a crucial process during erythrocyte maturation, the presence of mitochondria was analyzed in the CD71 and Ter119 positive cell populations. CD71 or Ter119 staining was combined with MTG, a permeable dye that accumulates in active mitochondria.17 NF-E2 tg mice display a significantly increased population of MTG+/CD71+ double positive cells compared to wt controls (Figure 1D,E). A similar trend is seen for the MTG+/Ter119+ population (Figure 1F, G), although the latter did not reach statistical significance. Comparison of the absolute numbers also shows that these two erythroid populations are increased in the tg mice (Online Supplementary Figure S2). Most importantly, within the immature populations (Ter119+/CD71high and Ter119+/CD71med), NF-E2 tg mice show a significant increase in the proportion of MTGhigh cells (Online Supplementary Figure S3A and S3B), which represent cells carrying the largest number of mitochondria. In order to determine whether cells on average contain more mitochondria per cell, we determined the mean fluorescence intensity (MFI) for MTG in each of the subpopulations (Online Supplementary Figure S3C). Indeed, the MFI for MTG is significantly increased, demonstrating a higher number of mitochondria per cell in NF-E2 tg mice. Increased NF-E2 expression thus impedes mitochondrial elimination in maturing erythrocytes, leading to the accumulation of larger numbers of mitochondria containing red cells, with a higher number of mitochondria per cell in the PB of NF-E2 tg mice. In a wt adult mouse, the PB contains around 3% reticulocytes, a proportion insufficient for ex vivo maturation analysis. NF-E2 tg and wt mice were therefore treated 1055

M. Gothwal et al.

with phenylhydrazine (PHZ), which induces hemolytic anemia leading to increased reticulocyte production.18 The experimental strategy is detailed in Figure 2A. PB cells isolated after PHZ treatment were cultured and matured ex vivo for 3 days and analyzed. Histological analysis verified that the cells analyzed represent reticulocytes (Online

Supplementary Figure S4). Mice were additionally followed up to analyze the recovery of their hematocrit values, as shown in the Online Supplementary Figure S5. Similar to the untreated mice described above, the NF-E2 tg PHZ treated mice displayed a significantly increased percentage of immature Ter119+/CD71high and

Figure 1. Increased retention of mitochondria in CD71 and Ter119 positive erythrocytes in NF-E2 tg mice. Whole blood or bone marrow from wt and tg animals, as indicated, was stained with CD71 and Ter119 antibodies as well as MitoTracker green (MTG) dye. (A, D, F) Representative FACS analysis and gating strategy for one wt and one tg animal for (A) CD71 and Ter119, (D) MTG and CD71 and (F) MTG and Ter119. (B, C) Percentages of erythroid populations in (B) PB and (C) bone marrow of NF-E2 tg and wt mice. (E, G) Percentages of (E) MTG+/CD71+ and (G) MTG+/Ter119+ stained cells in NF-E2 tg and wt mice. n= 12 wt, n= 21 tg (15 strain “9”, 6 strain “39”). Histograms show mean and SEM. *P<0.05. Statistical significance was calculated by Student’s t-tests.

1056

haematologica | 2016; 101(9)

NF-E2 regulates mitophagy

Ter119+/CD71med populations compared to controls, and thus a lower percentage of Ter119+/CD71low population (Figure 2B,C). Cells from NF-E2 tg mice thus contained a higher percentage of immature cells at the onset (day 0) of ex vivo maturation. After 3 days in culture, reticulocytes from both the tg and the wt mice had undergone matura-

tion, becoming predominantly Ter119+/CD71low. However, the NF-E2 tg cultures still retained a significantly higher proportion of immature cells, both Ter119+/CD71high and Ter119+/CD71med (Figure 2B,C). NF-E2 overexpressing mice thus retain their less mature phenotype despite following PHZ treatment and ex vivo maturation. Following

Figure 2. Delayed clearance of mitochondria from newly formed reticulocytes in NF-E2 tg mice. (A) Experimental procedure: mice were treated with phenylhydrazine (PHZ) to induce reticulocytosis, PB obtained on day 7 and matured ex vivo for 3 days. (B, D) Representative FACS analysis for one wt and one tg animal for day 0 and day 3 staining for (B) CD71 and Ter119, (D) MTG. Gating strategy was as detailed in Figure 1A. (C, E) Percentage of stained cells on day 0 and day 3 of ex vivo maturation in NF-E2 tg and wt mice: (C) Ter119+/CD71high, CD71med and CD71low, n= 7 wt, n= 8 tg (5 strain “9”, 3 strain “39”); (E) MTG+, n= 8 wt, n= 8 tg (5 strain “9”, 3 strain “39”). Histograms show mean and SEM. **P<0.01, ***P<0.001. Statistical significance was calculated by Student’s t-tests.

haematologica | 2016; 101(9)

1057

M. Gothwal et al.

PHZ treatment, NF-E2 tg mice again display a higher proportion of mitochondria retaining cells, on both day 0 and day 3 of culture (Figure 2D,E). In order to distinguish whether mitochondria are aberrantly retained in more mature populations or whether the increase in mitochondria simply reflects the excess of immature cells, reticulocytes in culture were stained with CD71, Ter119 and MTG. On day 0 of ex vivo culture NF-E2 tg mice contain a significantly increased proportion and absolute number of immature cells retaining mitochondria (MTG+/CD71high) (Figure 3A,B). During culture, these cells mature, giving rise to a MTG+/CD71low population on day 3 (Figure 3A,B). Nonetheless, after 3 days in culture, NF-E2 tg mice still retain a significantly higher percentage of the MTG+/CD71low population as well as a higher absolute number of MTG+/CD71low cells compared to wt controls.

Similarly, NF-E2 tg mice contain a higher percentage and a higher absolute number of more mature RBCs with mitochondria (MTG+/Ter119+) (Figure 3C,D). Analysis of defined subpopulations showed that before and after ex vivo maturation, Ter119+/CD71low cells overexpressing NF-E2 contain more mitochondria retaining cells than their wt littermate controls (Online Supplementary Figure S3D and S3E). While ex vivo culture decreases the proportion of cells containing mitochondria in both genotypes, nonetheless, on day 3, NF-E2 overexpressing Ter119+/CD71low cells still contain significantly more MTGhigh cells than wt controls (Online Supplementary Figure S3E, right panel). In order to determine whether these Ter119+/CD71low MTGhigh cells are only more abundant, or also aberrantly retain mitochondria, we determined the MFI for MTG,

Figure 3. Delayed clearance of mitochondria from Ter119 and CD71 positive newly formed reticulocytes in NF-E2 tg mice. (A, C) Representative FACS analysis and gating strategy for one wt and one tg animal for day 0 and day 3 staining for (A) MTG and CD71, (C) MTG and Ter119. (B, D) Percentage and absolute numbers per 10.000 measured events (per 3 μl whole blood) of stained cells in NF-E2 tg and wt mice on day 0 and day 3 of ex vivo maturation: (B) MTG+/CD71high (day 0) and MTG+/CD71low (day 3), n= 8 wt, n= 7 tg (4 strain “9”, 3 strain “39”); (D) MTG+/Ter119+, n= 8 wt, n= 7 tg (4 strain “9”, 3 strain “39”). Histograms show mean and SEM. *P<0.05, **P<0.01. Statistical significance was calculated by Student’s t-tests.

1058

haematologica | 2016; 101(9)

NF-E2 regulates mitophagy

which reflects the amount of mitochondria per cell (Online Supplementary Figure S3F). The MFI for MTG was significantly increased on day 0 demonstrating that, on average, mature Ter119+/CD71low cells in NF-E2 tg mice contain a higher number of mitochondria per cell. Hence, it is not the presence of increased numbers of immature cells but rather the aberrant retention of mitochondria within phenotypically mature cells that leads to the increased number of MTG+ cells in NF-E2 tg mice. Interestingly, PHZ altered the characteristics of PB reticulocytes. Under steady state conditions, NF-E2 tg mice display increased mitochondrial retention in the less mature populations (Online Supplementary Figure S3A and S3B). In contrast, under erythropoietic stress, following PHZ treatment, immature cells of both genotypes show large proportions of MTGhigh cells (60-75% MTGhigh in Ter119+CD71med, Online Ter119+/CD71high and Supplementary Figure S3D and S3E, compared to 15–30% MTGhigh in Ter119+/CD71high and Ter119+/CD71med, Online Supplementary Figure S3A and S3B). The increased retention of mitochondria in NF-E2 tg animals following PHZ treatment is seen in the more mature, Ter119+/CD71low

population (Online Supplementary Figure S3D and S3E). Both analyses (native state and PHZ) thus clearly show that elevated NF-E2 levels cause a delay in in vivo and ex vivo maturation of reticulocytes. At the same time, NFE2 tg erythrocytes are capable of exuding their mitochondria appropriately, as the proportion of Ter119+/CD71low cells remaining positive for MTG on day 3 is less than 5% (Online Supplementary Figure S3E, right panel). Mitochondrial transmembrane potential plays an important role in mitochondrial autophagy.10,19 Reticulocytes from mice deficient in the mitochondrial outer membrane protein NIX fail to eliminate mitochondria in culture. However, they correctly eliminate their organelles when treated with a depolarizing agent.13,20 We hypothesized that the higher percentage of mitochondria retaining RBCs in NF-E2 tg mice results from a higher mitochondrial membrane potential in these cells. In order to test this hypothesis, reticulocytes were stained with JC-1, a cationic dye that accumulates in mitochondria15 and emits fluorescence at 540nm. The intensity of emission correlates with the membrane potential, hence, depolarization decreases JC-1 fluorescence.21

JC-1

MFI (JC-1)

haematologica | 2016; 101(9)

Figure 4. Analysis of mitochondrial membrane potential during maturation of newly formed reticulocytes. PB of PHZ treated mice (as shown in Figure 2A) was stained with JC-1 and analyzed at the indicated time points. (A) Representative JC-1 staining for one wt and one tg animal. Red line: staining for 0 hour control; Blue line: staining after culture for the indicated time points. (B) Mean fluorescence intensity (MFI) of JC-1 staining during ex vivo maturation of newly formed reticulocytes from NF-E2 tg and wt mice. n= 3 wt, n= 3 tg (3 strain “39”). (C) Percentage of MTG+/Ter119+ population of reticulocytes from wt and tg mice cultured in the presence or absence of the depolarizing agents FCCP and ABT-737 for 4 hours. n= 12 wt, n= 12 tg (2 strain “9”, 10 strain “39”). The percentages were determined according to the gating strategy shown in Figure 3C. (D) Determination of mean fluorescence intensity (MFI) for MTG in the MTG+/Ter119+ cells of populations shown in Figure 4C. Histograms show mean and SEM. *P<0.05. Statistical significance was calculated by Student’s t-tests.

1059

M. Gothwal et al.

Reticulocytes from PHZ treated mice were cultured ex vivo and stained with JC-1 at several time points. At culture initiation, no difference in mitochondrial membrane potential between tg and wt mice was observed (Figure 4A,B), indicating that elevated NF-E2 levels do not alter the mitochondrial membrane potential. At subsequent time points, JC-1 fluorescence declines in both the tg and

wt reticulocytes (Figure 4A,B). However, the mean fluorescence intensity (MFI) of JC-1 staining in NF-E2 tg reticulocytes remains significantly higher at all measured time points, indicating an aberrant retention of mitochondrial membrane potential during maturation (Figure 4B). This could prevent mitochondrial elimination early during maturation, providing a mechanistic explanation for the

Figure 5. Ribosomal elimination is delayed during maturation of newly formed reticulocytes. PB of PHZ treated mice (as shown in Figure 2A) was stained with thiazole orange (TO) to determine ribosomal content in the maturing reticulocytes. (A, C) Representative FACS analysis and illustration of the gating strategy for one wt and one tg animal for day 0 and day 3 staining for (A) TO and CD71, (C) TO and Ter119. Insert: path of maturation of reticulocytes in culture is shown. (B, D) Percentage of stained cells on day 0 and day 3 of ex vivo maturation in NF-E2 tg and wt mice: (B) TO+/high/CD71high, n= 7 wt, n= 5 tg (3 strain “9”, 2 strain “39”) and (D) TO+/Ter119+, n= 7 wt, n= 6 tg (3 strain “9”, 3 strain “39”). (E) Determination of mean fluorescence intensity (MFI) for TO in the subpopulations determined by CD71 and Ter119 staining (as in Figure 2B) on days 0 and 3. Histograms show mean and SEM. *P<0.05, **P<0.01. Statistical significance was calculated by Student’s t-tests.

1060

haematologica | 2016; 101(9)

NF-E2 regulates mitophagy

observed ex vivo delay in mitochondrial elimination of cells with elevated NF-E2 levels. The mitochondrial membrane undergoes depolarization in the presence of depolarizing agents like FCCP and ABT737,13,22 which leads to mitochondrial elimination. We therefore investigated whether treatment with depolarizing agents would normalize mitochondrial elimination in

NF-E2 tg reticulocytes. Reticulocytes were cultured to undergo ex vivo maturation in the presence or absence of the depolarizing agents FCCP and ABT-737 followed by MTG/Ter119 staining for the detection of mitochondria. Again, untreated reticulocytes from NF-E2 tg mice showed a significantly higher proportion of MTG+/Ter119+ cells (Figure 4C). However, the treatment

Figure 6. NF-E2 regulates the expression of key mitophagy genes. (A) In silico analysis of NFE2 binding sites on the NIX and ULK1 gene (UCSC Genome Browser). Blue peaks represent the locus predicted to be bound by NF-E2. Horizontal arrows under the schematic represent primer locations for PCR after ChIP. (B) HEL cell lysates were chromatin immunoprecipitated with antibodies against NF-E2 or an IgG control. PCR was performed with primers flanking the predicted binding sites and the control site, as represented in (A). (C, D) RNA was isolated either from bone marrow cells (C) or from Ter119+/CD71+ sorted cells from PB (D) of wt and NF-E2 tg mice and subjected to qRT-PCR for Nix, Ulk1 and Atg7. Expression levels were normalized to B2m expression. (E) PB granulocytes from HC and PV were analysed by western blotting for NF-E2, NIX and ULK1 protein expression. Quantification and normalization to the loading control GAPDH is depicted underneath. (F) UKE-1, SET-2 and HEL cells were transduced with an empty LeGO-iG lentivirus (“empty”) or with viruses carrying a scrambled, ineffective shRNA (“scr”) or an shRNA directed against NF-E2 (“NF-E2”),23 as indicated. Infected cells were sorted by FACS and protein extracts interrogated by western blotting for expression of the indicated proteins. An arrowhead points to the 75 kDa NIX protein. *P<0.05. Statistical significance was calculated by Student’s t-tests.

haematologica | 2016; 101(9)

1061

M. Gothwal et al.

of NF-E2 tg cells with either FCCP or ABT-737 for 4 hours significantly reduced the number of mitochondria retaining cells, eliminating the difference between the two genotypes and normalizing mitochondrial retention in NFE2 tg reticulocytes to wt levels (Figure 4C). In order to determine whether depolarizing agents also normalized the number of mitochondria per cell, we determined the MFI for MTG (Figure 4D). The MFI for MTG was significantly increased in NF-E2 tg MTG+/Ter119+ cells in the absence of depolarizing agents, again demonstrating the increased number of mitochondria per cell in NF-E2 tg cells, but this difference disappeared in the presence of either FCCP or ABT-737. (Figure 4D). Thus, membrane depolarization overcomes the abnormal delay in mitochondrial elimination effected by elevated NF-E2 levels. Besides mitochondria, the elimination of ribosomes is also essential during RBC maturation.5 In order to assess whether elevated NF-E2 levels specifically delay mitochondrial elimination or also impede the exudation of ribosomes, reticulocytes undergoing ex vivo maturation were stained with thiazole orange (TO), a ribonucleic acid (RNA) binding dye, which, by nature of the vast amount of ribosomal RNA (rRNA) in ribosomes, specifically stains these organelles. In combination with CD71, TO delineated three populations: immature TO+/high/CD71high, maturing finally becoming mature to TO+/CD71low+neg, TOâ&#x20AC;&#x201C;/CD71low+neg cells (Figure 5A, insert). On day 0, NF-E2 tg mice showed a significantly increased percentage of both the highly immature TO+/high/CD71high population (Figure 5A,B) and the more mature TO+/Ter119+ cells (Figure 5C,D), evidence that elevated NF-E2 levels also lead to an aberrant retention of ribosomes. While the percentage of ribosome containing cells decreased during ex vivo maturation in both the wt and the tg animals, NFE2 tg mice nonetheless retained a significantly elevated proportion of ribosome retaining reticulocytes on day 3 (Figure 5B,D). Again the question arises as to whether the number of ribosomes per cell is increased. We therefore determined the MFI for TO in all subpopulations determined by CD71/Ter119 staining (Figure 5E). Indeed, the MFI for TO was significantly increased in the less mature subpopulations in NF-E2 tg cells compared to controls, and in Ter119+/CD71med a significant increase remained after 3 days of ex vivo culture (Figure 5E). Hence, at the same level of maturation, NF-E2 tg cells retain significantly more ribosomes per cell, demonstrating a defect in ribophagy. We hypothesized that the mechanism by which elevated levels of NF-E2 delay autophagy is by directly modulating the expression of autophagy genes. We therefore initially analyzed NF-E2 binding to autophagy genes by mining in silico ChIP-seq data. In human erythroid K562 cells, NF-E2 bound sites in the genes for NIX and ULK1 (Figure 6A). We verified the in silico data by performing ChIP in human erythroleukemia (HEL) cells. Indeed, NF-E2 binds both a distal site in the NIX locus as well as a site within the ULK1 gene. Control regions within both genes as well as the housekeeping gene myogenin were not bound (Figure 6B). These data demonstrate that these two autophagy genes constitute novel NF-E2 target genes. This observation is corroborated by analysis of gene expression in two large data sets, one from a collection of 200 AML patients, the second from purified erythroid precursor cells (Online Supplementary Figure S6). In both cases, NF-E2 1062

expression correlated highly significantly with the expression of both NIX and ULK1 (P=0.000002 and P=0.000009 and P=0.001 and P=0.00003, respectively, while NF-E2 and ATG7 expression did not correlate, Online Supplementary Figure S6), again suggesting that NF-E2 levels regulate expression of these two autophagy genes. An increase in NF-E2 levels alone is sufficient to increase expression of both Nix and Ulk1, as NF-E2 tg mice show significant elevation of Nix and Ulk1 expression in the BM (Figure 6C) as well as in sorted CD71+/Ter119+ cells from PB (Figure 6D). Hence, NF-E2 levels directly affect the autophagy process by altering the expression of critical components. Because elevated NF-E2 levels in mice are sufficient to increase Nix and Ulk1 expression and PV patients present with NF-E2 overexpression, we determined NIX and ULK1 protein levels in purified primary granulocytes from PV patients and healthy controls (Figure 6E). Expression of both autophagy proteins is significantly increased in PV patients, confirming the fidelity of our mouse model and suggesting alterations in red cell physiology in PV patients. We determined whether NF-E2 is required for ULK1 and NIX expression by silencing the expression of the transcription factor using a short hairpin RNA (shRNA) against NF-E223 as well as a scrambled control. Three different myeloid cell lines, UKE-1, SET-2 and HEL, all carrying the JAK2V617F mutation prevalent in PV patients, were used. While decreasing NF-E2 expression did not alter ULK1 levels (data not shown), it caused a strong decrease in NIX expression in all three cell lines, particularly of the 75 kDa protein overexpressed in PV patients (Figure 6E,F, arrowheads). These data demonstrate that NF-E2 is required for optimal NIX expression, and show that this autophagy gene is a novel NF-E2 target gene. Interestingly, PV patients display reticulocytosis, similar to NF-E2 tg mice.24,25 In contrast, the RBC life span has not been reported to be altered in PV patients,26 but these studies were conducted in 1951, and very subtle differences, such as the ones we report (Online Supplementary Figure S1), may not have been evident. Elevated NF-E2 levels thus cause a delay in both autophagy processes, and the exclusion of mitochondria and ribosomes, an observation that provides a mechanistic explanation for the two observed RBC abnormalities in NF-E2 tg mice. Firstly, delayed autophagy can account for the significant increase in the half-life of NF-E2 tg RBCs (Online Supplementary Figure S1), secondly, abnormal, mitochondria-retaining RBCs may be subject to enhanced elimination in the spleen, explaining the increased splenic RBC destruction observed in NF-E2 tg mice.

Discussion Programmed removal of internal organelles from maturing reticulocytes constituted an essential step during terminal erythroid differentiation. Several studies have focused on the mechanisms of mitophagy (removal of mitochondria) and ribophagy (removal of ribosomes) in developing erythrocytes. The Bcl2 related mitochondrial outer membrane protein NIX plays an important role in mitochondrial elimination during reticulocyte maturation.13,20 In reticulocytes from Nixâ&#x20AC;&#x201C;/â&#x20AC;&#x201C; mice, mitochondria fail to enter the autophagosomes. However, the autophagy process itself remains haematologica | 2016; 101(9)

NF-E2 regulates mitophagy

functional, as the addition of depolarizing agents can correct the defect and force mitophagy.13,20 NIX has been shown to bind LC3/GABARAP proteins on autophagosomes, tethering the outer mitochondrial membrane to the autophagosomes. NIX thus functions as a mitophagy receptor, targeting superfluous mitochondria to the autophagosome for degradation.27 Several other proteins have been implicated in mitophagy during erythroid differentiation, including ULK1 and ATG7.28,29 Mice deficient in the kinase ULK1 display a defect in reticulocyte maturation. Ulk1–/– mice show impaired clearance of both mitochondria and ribosomes.28 Again, the depolarizing agent carbonyl cyanide m-chlorophenylhydrazone (CCCP) was able to rescue mitophagy in ULK1 deficient mice. Since ULK1 functions downstream of mitochondrial depolarization,30 these data suggest the presence of ULK1-independent pathways of mitochondrial clearance in RBCs. ATG7 is an E1-activating enzyme essential for the ubiquitin-like conjugation of autophagosomal proteins to each other and to phosphatidylethanolamine and, ultimately, for autophagosome formation.31 ATG7 deficient reticulocytes show impaired mitochondrial clearance.29 Interestingly, even during ex vivo maturation, ATG7 deficient reticulocytes do not depolarize. Since ATG7 is required for autophagosome formation, these data suggest that mitochondrial depolarization is a consequence, rather than a cause, of autophagosome formation. However, a transient depolarization, which is undetectable in ex vivo cultures, may normally precede and cause autophagosome formation. Alternatively, novel, unidentified signals may trigger or contribute to autophagy in maturing reticulocytes. Overexpression of the transcription factor NF-E2 in our transgenic mouse model led to a significant increase in the number of erythroid precursor cells in the BM.2 Furthermore, NF-E2 tg mice present with both reticulocytosis and a concomitant increase in red cell destruction in the spleen, manifested by increased iron deposits. As a result, NF-E2 tg mice do not display polycythemia, since the increased RBC destruction appears to outweigh the increased red cell production. Nonetheless, the absence of polycythemia in our mice appears paradoxical given the observed NF-E2 overexpression in more than 90% of patients with polcythemia vera. We therefore sought to determine the molecular effect of elevated NF-E2 levels on red cell physiology. We hypothesized that elevated NF-E2 levels may perturb the kinetics of erythroid differentiation, leading either to an accelerated or to a delayed maturation. To date, a possible role for NF-E2 in programmed organelle clearance during erythroid differentiation has not been investigated. Therefore, we sought to analyze the role of NF-E2 in erythroid maturation, mitophagy and ribophagy using our NF-E2 overexpressing transgenic mouse model. We first assessed whether the NF-E2 tg reticulocytes displayed a maturation defect. Indeed, increased NF-E2 levels led to an excess of immature erythroid cells in the PB and a delay in their ex vivo maturation. This maturation delay can explain both the observed reticulocytosis and the increased RBC half-life, as both observations result from immature cells remaining in the peripheral circulation for a longer time. Delayed erythroid maturation has previously been observed by elevating NF-E2 levels in human CD34+ hematopoietic stem cells.32 Our current data thus haematologica | 2016; 101(9)

corroborates the importance of correct NF-E2 levels for physiological erythroid maturation. Delayed maturation can result from inefficient or absent elimination of internal organelles from maturing reticulocytes. Our analyses showed a significantly increased proportion of cells retaining mitochondria and ribosomes in both the CD71 and the Ter119 positive populations in NF-E2 tg mice, indicating a defect in organelle elimination. The measurement of mitochondrial membrane potential revealed significantly increased potential in NF-E2 tg mice compared to wt controls, even after 48 hours of ex vivo culture, indicating that membrane depolarization is inefficient or completely absent (Figure 4B). As in NIX and ULK1-deficient mice,13,28 the addition of depolarizing agents rescued the defect (Figure 4C), demonstrating that in the presence of elevated NF-E2 levels, the autophagy process itself remains functional. The mitophagy defect observed in NF-E2 tg mice is less severe than that of Nix–/– mice,13,20 as the organelles are retained in more than 70% of Ter119+ NIX deficient cells after ex vivo maturation, compared to the 20% retention in NF-E2 tg mice (Figure 3D). Our observations are more similar to those in Atg7–/– mice, which also show a delay in maturation but undergo organelle clearance if given enough time during ex vivo maturation.29 The relatively mild autophagy defect in NF-E2 tg mice is exacerbated during stress erythropoiesis following PHZ treatment (compare Figure 1 and Figure 2). Stress erythropoiesis occurs in the spleen and liver, rather than in the BM, where steady state erythropoiesis takes place. Moreover, stress erythropoiesis utilizes a population of stress erythroid progenitor cells, which are distinct from the BM, steady state erythroid progenitors.33 The effect of NF-E2 overexpression appears to be more pronounced in these stress erythroid progenitor cells. We investigated the mechanism underlying the maturation defect. Two critical autophagy genes, NIX and ULK1 are directly regulated by NF-E2 (Figure 6B). Increased NF-E2 activity in tg mice led to a significant increase in Nix and Ulk1 expression in the BM (Figure 6C) as well as in sorted CD71+/Ter119+ cells from PB (Figure 6D). Randhawa and colleagues have recently published in silico data that likewise suggests an important role for NF-E2 in ULK1 regulation, albeit this paper does not contain functional data.34 Ours is the first report demonstrating a role for NF-E2 in erythroid maturation by affecting autophagy genes. Because NIX and ULK1 interact with partner proteins in the autophagy process,35 their increased expression may disrupt complex formation. If the levels of interacting proteins are not raised concomitantly, ineffective complexes may be formed. We recently described that, surprisingly for a transcription factor, NF-E2 is found predominantly in the cytoplasm of maturing, nucleated erythrocytic cells.36 Only 10% of all erythroid cells, mainly the early erythroblasts, showed nuclear NF-E2 staining. So far, no function has been attributed to the cytoplasmically located NF-E2 protein. Our data presented herein support the hypothesis that cytoplasmically located NF-E2 may have a function in the elimination of internal organelles. The mild autophagy defect observed in our mice may be due to the requirement for additional proteins in the formation of larger complexes to exert this function. The elevation of NF-E2 levels in isolation may not completely disrupt the function of this complex. 1063

M. Gothwal et al.

References 1. Armstrong JA, Emerson BM. NF-E2 disrupts chromatin structure at human betaglobin locus control region hypersensitive site 2 in vitro. Mol Cell Biol. 1996;16(10):5634-5644. 2. Kaufmann K, GrĂźnder A, Hadlich T, et al. A novel murine model of myeloproliferative disorders generated by overexpression of the transcription factor NF-E2. J Exp Med. 2012;209(1):35-50. 3. Goerttler PS, Kreutz C, Donauer J, et al. Gene expression profiling in polycythaemia vera: overexpression of transcription factor NF-E2. Br J Haematol. 2005;129(1):138-150. 4. Chasis JA, Mohandas N. Erythroblastic islands: niches for erythropoiesis. Blood. 2008;112(3):470-478. 5. Koury MJ, Koury ST, Kopsombut P, Bondurant MC. In vitro maturation of nascent reticulocytes to erythrocytes. Blood. 2005;105(5):2168-2174. 6. Nakatogawa H, Suzuki K, Kamada Y, Ohsumi Y. Dynamics and diversity in autophagy mechanisms: lessons from yeast. Nat Rev Mol Cell Biol. 2009;10(7): 458-467. 7. Yang Z, Klionsky DJ. Mammalian autophagy: core molecular machinery and signaling regulation. Curr Opin Cell Biol. 2010;22(2):124-131. 8. Alexeyev MF, Ledoux SP, Wilson GL. Mitochondrial DNA and aging. Clin Sci (Lond). 2004;107(4):355-364. 9. Kissova I, Deffieu M, Samokhvalov V, et al. Lipid oxidation and autophagy in yeast. Free Radic Biol Med. 2006;41(11):16551661. 10. Elmore SP, Qian T, Grissom SF, Lemasters JJ. The mitochondrial permeability transition initiates autophagy in rat hepatocytes. FASEB J. 2001;15(12):2286-2287. 11. Narendra D, Tanaka A, Suen DF, Youle RJ. Parkin is recruited selectively to impaired mitochondria and promotes their autophagy. J Cell Biol. 2008;183(5):795-803. 12. Holm TM, Braun A, Trigatti BL, et al. Failure of red blood cell maturation in mice with defects in the high-density lipoprotein receptor SR-BI. Blood. 2002;99(5):18171824. 13. Sandoval H, Thiagarajan P, Dasgupta SK, et al. Essential role for Nix in autophagic maturation of erythroid cells. Nature. 2008;

1064

454(7201):232-235. 14. Zhang J, Kundu M, Ney PA. Mitophagy in mammalian cells: the reticulocyte model. Methods Enzymol. 2009;452:227-245. 15. Cossarizza A, Salvioli S. Flow cytometric analysis of mitochondrial membrane potential using JC-1. Curr Protoc Cytom. 2001;Chapter 9:Unit 9.14. 16. Socolovsky M, Nam H, Fleming MD, Haase V, Brugnara C, Lodish HF. Ineffective erythropoiesis in Stat5a(-/-)5b(-/-) mice due to decreased survival of early erythroblasts. Blood. 2001;98(12):3261-3273. 17. Sutovsky P, Navara CS, Schatten G. Fate of the sperm mitochondria, and the incorporation, conversion, and disassembly of the sperm tail structures during bovine fertilization. Biol Reprod. 1996;55(6):1195-1205. 18. Vannucchi AM, Bianchi L, Cellai C, et al. Accentuated response to phenylhydrazine and erythropoietin in mice genetically impaired for their GATA-1 expression (GATA-1(low) mice). Blood. 2001;97(10): 3040-3050. 19. Priault M, Salin B, Schaeffer J, Vallette FM, di Rago JP, Martinou JC. Impairing the bioenergetic status and the biogenesis of mitochondria triggers mitophagy in yeast. Cell Death Differ. 2005;12(12):1613-1621. 20. Schweers RL, Zhang J, Randall MS, et al. NIX is required for programmed mitochondrial clearance during reticulocyte maturation. Proc Natl Acad Sci USA. 2007;104(49):19500-19505. 21. Reers M, Smiley ST, Mottola-Hartshorn C, Chen A, Lin M, Chen LB. Mitochondrial membrane potential monitored by JC-1 dye. Methods Enzymol. 1995;260:406-417. 22. Gottlieb E, Vander Heiden MG, Thompson CB. Bcl-x(L) prevents the initial decrease in mitochondrial membrane potential and subsequent reactive oxygen species production during tumor necrosis factor alphainduced apoptosis. Mol Cell Biol. 2000; 20(15):5680-5689. 23. Roelz R, Pilz IH, Mutschler M, Pahl HL. Of mice and men: human RNA polymerase III promoter U6 is more efficient than its murine homologue for shRNA expression from a lentiviral vector in both human and murine progenitor cells. Exp Hematol. 2010;38(9):792-797. 24. Anger B, Haug U, Seidler R, Heimpel H. Polycythemia vera. A clinical study of 141 patients. Blut. 1989;59(6):493-500. 25. d'Onofrio G, Tichelli A, Foures C,

26.

27.

28.

29.

30.

31.

32.

33.

34.

35. 36.

Theodorsen L. Indicators of haematopoietic recovery after transplantation: the role of reticulocyte measurements. Clin Lab Haematol. 1996;18 Suppl 1:45-53. Berlin NI, Lawrence JH, Lee HC. The life span of the red blood cell in chronic leukemia and polycythemia. Science. 1951;114(2963):385-387. Novak I, Kirkin V, McEwan DG, et al. Nix is a selective autophagy receptor for mitochondrial clearance. EMBO Rep. 2010; 11(1):45-51. Kundu M, Lindsten T, Yang CY, et al. Ulk1 plays a critical role in the autophagic clearance of mitochondria and ribosomes during reticulocyte maturation. Blood. 2008; 112(4):1493-1502. Zhang J, Randall MS, Loyd MR, et al. Mitochondrial clearance is regulated by Atg7-dependent and -independent mechanisms during reticulocyte maturation. Blood. 2009;114(1):157-164. Itakura E, Kishi-Itakura C, Koyama-Honda I, Mizushima N. Structures containing Atg9A and the ULK1 complex independently target depolarized mitochondria at initial stages of Parkin-mediated mitophagy. J Cell Sci. 2012;125(Pt 6):14881499. Komatsu M, Waguri S, Ueno T, et al. Impairment of starvation-induced and constitutive autophagy in Atg7-deficient mice. J Cell Biol. 2005;169(3):425-434. Mutschler M, Magin AS, Buerge M, et al. NF-E2 overexpression delays erythroid maturation and increases erythrocyte production. Br J Haematol. 2009;146(2):203217. Paulson RF, Shi L, Wu DC. Stress erythropoiesis: new signals and new stress progenitor cells. Curr Opin Hematol. 2011;1 8(3):139-145. Randhawa R, Sehgal M, Singh TR, Duseja A, Changotra H. Unc-51 like kinase 1 (ULK1) in silico analysis for biomarker identification: a vital component of autophagy. Gene. 2015;562(1):40-49. Zhang J, Wu K, Xiao X, et al. Autophagy as a regulatory component of erythropoiesis. Int J Mol Sci. 2015;16(2):4083-4094. Aumann K, Frey AV, May AM, et al. Subcellular mislocalization of the transcription factor NF-E2 in erythroid cells discriminates prefibrotic primary myelofibrosis from essential thrombocythemia. Blood. 2013;122(1):93-99.

haematologica | 2016; 101(9)

ARTICLE

Myeloproliferative DIsorders

Safety and efficacy of ruxolitinib in an open-label, multicenter, single-arm phase 3b expanded-access study in patients with myelofibrosis: a snapshot of 1144 patients in the JUMP trial

Haifa Kathrin Al-Ali,1 Martin Griesshammer,2 Philipp le Coutre,3 Cornelius F. Waller,4 Anna Marina Liberati,5 Philippe Schafhausen,6 Renato Tavares,7 Pilar Giraldo,8 Lynda Foltz,9 Pia Raanani,10 Vikas Gupta,11 Bayane Tannir,12 Julian Perez Ronco,12 Jagannath Ghosh,13 Bruno Martino,14 and Alessandro M. Vannucchi15

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2016 Volume 101(9):1065-1073

University Hospital of Leipzig, Germany; 2Johannes-Wesling Academic Medical Center, University of Hannover Teaching Hospital, Minden, Germany; 3Charité – Universitätsmedizin Berlin, Germany; 4Department of Medicine I: Hematology and Oncology, University Hospital Freiburg, Germany; 5University of Perugia, Azienda Ospedaliera S. Maria, Terni, Italy; 6 Department of Internal Medicine II, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; 7Universidade Federal de Goiania, Goiás, Brazil; 8Miguel Servet University Hospital. IIS Aragon and Centro de Investigación Biomédica en Red de Enfermedades Raras (CIBERER), Zaragoza, Spain; 9St. Paul’s Hospital, University of British Columbia, Vancouver, BC, Canada; 10Rabin Medical Center, Petah Tikva, Israel; 11Princess Margaret Cancer Centre, Toronto, ON, Canada; 12Novartis Pharma AG, Basel, Switzerland; 13 Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA; 14Azienda Ospedaliera Bianchi Melacrino Morelli, Reggio Calabria, Italy; and 15University of Florence, Italy 1

ABSTRACT

UMP is a phase 3b expanded-access trial for patients without access to ruxolitinib outside of a clinical study; it is the largest clinical trial to date in patients with myelofibrosis who have been treated with ruxolitinib. Here, we present safety and efficacy findings from an analysis of 1144 patients with intermediate- or high-risk myelofibrosis, as well as a separate analysis of 163 patients with intermediate-1-risk myelofibrosis – a population of patients not included in the phase 3 COMFORT studies. Consistent with ruxolitinib’s mechanism of action, the most common hematologic adverse events were anemia and thrombocytopenia, but these led to treatment discontinuation in only a few cases. The most common non-hematologic adverse events were primarily grade 1/2 and included diarrhea, pyrexia, fatigue, and asthenia. The rates of infections were low and primarily grade 1/2, and no new or unexpected infections were observed. The majority of patients achieved a ≥50% reduction from baseline in palpable spleen length. Improvements in symptoms were rapid, with approximately half of all patients experiencing clinically significant improvements, as assessed by various quality-of-life questionnaires. The safety and efficacy profile in intermediate-1-risk patients was consistent with that in the overall JUMP population and with that previously reported in intermediate-2- and highrisk patients. Overall, ruxolitinib provided clinically meaningful reductions in spleen length and symptoms in patients with myelofibrosis, including those with intermediate-1-risk disease, with a safety and efficacy profile consistent with that observed in the phase 3 COMFORT studies. This trial was registered as NCT01493414 at ClinicalTrials.gov. Introduction Myelofibrosis (MF) is a chronic myeloproliferative neoplasm characterized by a dysregulated Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway that can present as primary MF or evolve from polycythemia vera or essential thrombocythemia.1-4 Common aspects of the disease include bone marhaematologica | 2016; 101(9)

Correspondence: alah@medizin.uni-leipzig.de

Received: January 29, 2016. Accepted: May 25, 2016. Pre-published: May 31, 2016. doi:10.3324/haematol.2016.143677

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1065

1065

H.K. Al-Ali et al.

row fibrosis, extramedullary hematopoiesis, anemia, splenomegaly, a debilitating symptom burden (e.g., fatigue, weight loss, night sweats, pruritus),4-7 and reduced survival.8 Patients are stratified as having low-, intermediate-1-, intermediate-2-, or high-risk MF by International Prognostic Scoring System (IPSS) criteria, with corresponding median survivals of 11, 8, 4, and 2 years.8 Ruxolitinib is a potent JAK1/JAK2 inhibitor that has demonstrated superiority over placebo9 and best available therapy10 in the pivotal phase 3 COMFORT (Controlled Myelofibrosis study with Oral JAK inhibitor Treatment) studies in patients with intermediate-2- or high-risk MF. Ruxolitinib treatment led to durable improvements in splenomegaly, symptoms, and quality-of-life (QOL) measures as well as improved survival.9-11 Based on these findings, ruxolitinib became the first JAK inhibitor approved for the treatment of MF. Despite the impact of these studies on the treatment landscape in higher-risk MF, not much is known about the efficacy and safety of ruxolitinib in patients with intermediate-1-risk MF – a group that also has significant constitutional symptoms, splenomegaly, and a reduced health-related QOL.8 Findings from ROBUST, a phase 2 study that evaluated ruxolitinib in patients with high-, intermediate-2-, and intermediate-1risk MF, indicated that ruxolitinib treatment results in clinically meaningful reductions in spleen length and symptoms in most patients, including those classified as intermediate-1 risk.12 To allow for collection of additional safety and efficacy data for ruxolitinib and provide an access path to ruxolitinib for patients with MF, the JUMP (JAK Inhibitor RUxolitinib in Myelofibrosis Patients; NCT01493414) study was initiated. JUMP is a phase 3b expanded-access trial for patients in countries without access to ruxolitinib outside of a clinical study and includes those classified as intermediate-1 risk, a population that was not included in the COMFORT trials. Here, we present safety and efficacy findings from an analysis of 1144 patients with MF and 163 patients with intermediate-1-risk MF who started ruxolitinib treatment ≥1 year before the data cut-off.

Methods Eligibility criteria

Eligible patients were aged ≥18 years with a diagnosis of primary or secondary MF by World Health Organization criteria3,13 and classified as IPSS high, intermediate-2, or intermediate-1 risk by the treating investigator.8 Patients with intermediate-1-risk MF were required to have a palpable spleen (≥5 cm from the costal margin). However, IPSS risk status was not recorded until implementation of protocol amendment 2, and most patients did not have IPSS risk status at the time of this data cut. Patients were required to have a baseline platelet count ≥50×109/L; those with a platelet count of 50×109/L to 100×109/L were included through amendments to the protocol.

Study design JUMP is a single-arm, open-label phase 3b expanded-access global study. Patients received starting doses of ruxolitinib based on platelet counts at baseline: 5 mg twice daily (bid; 50×109/L to <100×109/L), 15 mg bid (100 to 200×109/L), or 20 mg bid (>200×109/L). Doses could be increased (up to 25 mg bid if platelet and neutrophil counts were adequate) in 5-mg bid increments (5 mg/day for patients starting ruxolitinib at 5 mg bid) when efficacy 1066

was insufficient. Dose decreases or interruptions were mandatory for safety reasons (e.g., declining platelet counts) and were made according to a protocol-specified dosing regimen (Online Supplementary Methods). Patients were followed up for 28 days after treatment discontinuation. The final analysis will be performed after all patients have completed 24 months of treatment, unless discontinuation criteria are met (Online Supplementary Methods).

Endpoints The primary endpoint was assessment of ruxolitinib safety and tolerability by the frequency, duration, and severity of adverse events. Additional endpoints included the proportion of patients with ≥50% reduction in palpable spleen length, patient-reported outcomes [Functional Assessment of Cancer Therapy Lymphoma (FACT-Lym) total score (TS) and Functional Assessment of Chronic Illness Therapy (FACIT) - Fatigue Scale], progression-free survival, survival without transformation to acute myeloid leukemia (AML-free survival), and overall survival (Online Supplementary Methods).

Study oversight The study was sponsored and designed by Novartis Pharmaceuticals Corporation, approved by the institutional review board at each participating institution, and conducted in accordance with applicable local regulations and the principles of the Declaration of Helsinki. All patients provided written informed consent. Data were analyzed and interpreted by the sponsor’s clinical and statistical teams in collaboration with authors not affiliated with the sponsor. A publication steering committee discussed the data and analyses and helped guide the publication plan for this study. The first author prepared the first draft of the manuscript, with assistance from a medical writer funded by Novartis Pharmaceuticals Corporation, and made the final decision to submit the manuscript for publication. All authors reviewed and amended the manuscript.

Statistical analysis This analysis includes results from patients who started ruxolitinib treatment ≥1 year before the data cutoff. Only descriptive statistics are provided. Changes in spleen length were assessed in patients with baseline and post-baseline assessments. A 100% reduction from baseline in spleen length was defined as non-palpable. Survival assessments were performed at the end of the 28day follow-up period. Progression-free and overall survival were estimated using the Kaplan-Meier method.

Results Patients A total of 2233 patients were enrolled at more than 200 study sites in 26 countries (last patient first visit, 23 December 2014). This analysis includes results for 1144 patients who started ruxolitinib treatment ≥1 year before the data cutoff date (01 January 2014). Patients were from Europe (83.6%), South America (8.5%), North America (4.6%), and other regions (3.5%) (Online Supplementary Table S1). The baseline characteristics of the patients are shown in Table 1. The median age of the patients was 68 years (range, 18-89 years), with a median time since initial diagnosis of 36 months (range, 0.1-433.1 months). Overall, 53.3% of patients were male and 58.8% had primary MF. Most patients completed treatment per protocol (i.e., had access to commercially available drug; 33.5%) or haematologica | 2016; 101(9)

Ruxolitinib for MF: 1144-patient snapshot of JUMP

remained on treatment (35.5%; Online Supplementary Table S2). The most common reasons for treatment discontinuation included adverse events (13.8%), investigator-determined disease progression (7.1%), death (3.8%), consent withdrawal (3.8%), and physician’s decision (1.4%). Specific adverse events that led to discontinuation are shown in Online Supplementary Table S3. The median exposure to ruxolitinib was 11.1 months. The mean total daily dose over time is shown in Figure 1. The median average daily dose was 37.2 mg for patients starting at 20 mg bid (63.6%), 23.4 mg for patients starting at 15 mg bid (32.7%), and 13.3 mg for patients starting at 5 mg bid (1.0%). The majority of patients (65.9%) had a dose modification, and 23.9% had a dose interruption.

Serious adverse events were reported in 32.3% of patients (n=369). Serious adverse events occurring in >1% of patients included anemia (4.5%), pneumonia (3.9%), pyrexia (3.2%), cardiac failure (1.8%), dyspnea (1.7%), abdominal pain (1.6%), gastrointestinal hemorrhage (1.4%), thrombocytopenia (1.0%), cardiac atrial fibrillation (1.0%), and general physical health deterioration (1.0%). There were no reports of progressive multifocal leukoencephalopathy. The most common non-hematologic adverse events (occurring in ≥5% of patients) were primarily grade 1/2 (Table 2) and included diarrhea, pyrexia, fatigue, and asthenia. Rates of non-hematologic grade 3/4 adverse events were low overall (<2%), except the rate of pneu-

Safety Consistent with findings from the COMFORT studies,9,10 the most common hematologic adverse events were anemia (all grades, 56.3%; grade 3/4, 33.0%) and thrombocytopenia (all grades, 42.2%; grade 3/4, 12.5%; Table 2). However, only 2.6% (n=30) and 3.2% (n=37) of patients, respectively, discontinued treatment (Online Supplementary Table S3), indicating that these adverse events were manageable in most patients. Overall, 159 patients (13.9%) received erythropoiesis-stimulating agents as concomitant medication; preliminary results are summarized for these patients. Similar to what has been observed previously,9,10 mean hemoglobin levels decreased from baseline (108 g/L) during the first 8 to 12 weeks of the study but increased to near-baseline levels after week 12 (Figure 2A). Mean platelet counts decreased from baseline (319×109/L) during the first 4 weeks and then remained stable over time (Figure 2B). As expected, the rates of thrombocytopenia were higher in patients who had baseline platelet counts of 100×109/L to 200×109/L and received ruxolitinib at a starting dose of 15 mg bid (all grades, 58.8%; grade 3/4, 22.2%) than in patients who had baseline platelet counts >200×109/L and started treatment at 20 mg bid (all grades, 33.9%; grade 3/4, 7.1%). The rates of anemia were similar in both groups (all grades, 53.2% versus 57.9%; grade 3/4, 32.6% versus 33.1%, respectively).

Table 1. Patients’ baseline characteristics.

All Patients N=1144 Age, median (range), years 68.0 (18.0-89.0) ≥65 years, n (%) 701 (61.3) Male, n (%) 610 (53.3) Time since initial diagnosis, median (range), months 36.0 (0.1-433.1) MF subtype, n (%)a PMF 673 (58.8) PPV-MF 279 (24.4) PET-MF 191 (16.7) Hemoglobin level, median (range), g/Lb 105.0 (44.0-200.0) <100 g/L, n (%) 461 (40.3) 9 c Platelet count, median (range), ×10 /L 256.0 (75.0-1910.0) <100×109/L, n (%) 16 (1.4) 100 to <200×109/L, n (%) 385 (33.7) ≥200×109/L, n (%) 738 (64.5) Prior history of transfusions, n (%) 373 (32.6) Peripheral blasts ≥1%, n (%)d 401 (35.1) e Palpable spleen length, median (range), cm 13.0 (1.0-43.0) MF: myelofibrosis; PET-MF: post–essential thrombocythemia myelofibrosis; PMF: primary myelofibrosis; PPV-MF, post–polycythemia vera myelofibrosis. Note: Patients were classified as intermediate-1, intermediate-2, or high risk per study entry criteria. However, International Prognostic Scoring System risk status was not collected until implementation of protocol amendment 2; 95% of patients (n=1087) had missing risk status information. a n=1143; b n=1140; c n=1139; d n=1059; e n=1072.

Figure 1. Mean total daily dosea,b of ruxolitinib. bid: twice daily; qd: once daily. aIncludes patients with complete drug administration dates (n=1141). b 28 patients started at doses other than 5, 15, or 20 mg bid: 5 mg qd (n=1), 7.5 mg bid (n=1), 15 mg qd (n=6), 10 mg bid (n=16), 20 mg qd (n=4).

haematologica | 2016; 101(9)

1067

H.K. Al-Ali et al.

monia (3.6%), which led to drug discontinuation in six patients (0.5%) (Online Supplementary Table S3). The rates of infections were also low; all-grade infections in ≥1% of patients included nasopharyngitis (6.3%), urinary tract infection (6.0%), pneumonia (5.3%), bronchitis (4.2%), herpes zoster (3.6%), influenza (3.0%), upper respiratory tract infection (2.9%), cystitis (2.5%), gastroenteritis (1.8%), respiratory tract infection (1.8%), and oral herpes (1.6%) (Online Supplementary Table S4). Other infections included tuberculosis in three patients (0.3%) and legionella pneumonia in one patient (0.1%); no hepatitis B reactivation was reported.

Efficacy At weeks 24 and 48, 56.9% and 62.3% of evaluable patients achieved a ≥50% reduction from baseline in palpable spleen length, respectively (Figure 3A); additionally, 23.9% and 18.2% had 25% to <50% reductions at weeks 24 and 48, respectively. At each assessment, at least twothirds of patients had a ≥25% reduction in palpable spleen length, and approximately 80% had a ≥25% reduction after week 4. Most patients (69.0%) experienced a ≥50% reduction in spleen length at any time by week 48 (Figure 3B), and 23.0% had spleens that became non-palpable. The proportion of patients who achieved a ≥50% reduction in spleen length at any time by week 48 was similar for patients with primary or secondary MF (66.9% versus 71.6%, respectively; see the Online Supplementary Results for additional details). At both weeks 24 and 48, ≥50% reductions from baseline in palpable spleen length were achieved by a higher proportion of patients starting treatment at 20 mg bid than by those starting at 15 mg bid (week 24, 61.8% versus 47.3%; week 48, 68.9% versus 48.7%). Likewise, the proportion of patients who achieved a ≥50% reduction from baseline in palpable spleen length at any time was higher in patients who started treatment at 20 mg bid (73.8% versus 60.1%). The median time to the first ≥50% reduction in palpable spleen length was 5.1 weeks (range, 0.1-53.1 weeks). The Kaplan-Meier estimated probability of maintaining a spleen response for 24 weeks and 48 weeks was 93% (95% CI, 91%-95%) and 72% (95% CI, 54%-84%), respectively. The median duration of spleen response was not yet reached; 9.8% of patients had a loss of response (i.e., a return of spleen length to baseline size). Preliminary results suggest that patients who required concomitant erythropoiesis-stimulating agents on treatment also had clinically meaningful spleen responses: 70.1% of evaluable patients (101/144) who received erythropoiesis-stimulating agents achieved a ≥50% reduction from baseline in palpable spleen length at any time. Clinically meaningful improvements in symptoms were seen as early as 4 weeks after the start of treatment and were maintained over time, as evaluated by the FACTLym TS (mean change from baseline was 11.0 at week 4 and 9.4 at week 48; Online Supplementary Figure S2A) and FACIT-Fatigue scale (mean change from baseline was 3.8 at week 4 and 3.0 at week 48; Online Supplementary Figure S2B). Approximately 44% to 46% and 46% to 52% of patients achieved a response (i.e., a minimally important difference at each time point) in the FACT-Lym TS (Figure 4A) and the FACIT-Fatigue scale (Figure 4B), respectively; symptom response rates were not affected by ruxolitinib starting dose. Response rates on the FACT-Lym TS and FACIT-Fatigue scales were similar when evaluated by MF 1068

subtype (Online Supplementary Figures S3 and S4). Similar improvements were observed in other QOL questionnaires, with the proportions of patients achieving a response ranging from 44% to 48% for the FACT-Lym subscale, 40% to 43% for the FACT-Lym Trial Outcome Index, and 39% to 42% for the FACT-General TS. A total of 13 patients developed acute myeloid leukemia during the study or within 28 days following study discontinuation (Table 3). The estimated probability of AML-free survival at 48 weeks was 94% (95% CI, 92%-95%; Table 3). The estimated probability of progression-free survival at 48 weeks was 91% (95% CI, 89%-93%). The estimated probability of overall survival at 48 weeks was 94% (95% CI, 93%-96%) for the overall JUMP population, with 86 reported deaths (Table 3). Investigator-determined causes of death on study, regardless of causality, included myelofibrosis (n=11), cardiac failure (n=7), cardiac arrest (n=7), pneumonia (n=6), septic shock (n=5), multi-organ failure (n=4), transformation to acute myeloid leukemia (n=4), respiratory failure (n=3), sepsis (n=3), cardiorespiratory arrest (n=3), gastrointestinal hemorrhage (n=2), peritonitis (n=2), renal failure (n=1), and general health deterioration (n=2). All other causes were undetermined or reported for one patient each. The survival estimate for patients who received erythropoiesis-stimulating agents was similar to that for the overall population (99%; 95% CI, 96%-100%; deaths, n=10/159). Table 2. Adverse events regardless of study drug relationship (in ≥5% of patients)a.

Preferred Term

Hematologic adverse events Anemia Thrombocytopenia Neutropenia Non-hematologic adverse events Diarrhea Pyrexia Fatigue Asthenia Edema peripheral Headache Dyspnea Abdominal pain Nausea Cough Arthralgia Pain in extremity Nasopharyngitis Urinary tract infection Pruritus Constipation Pneumonia Dizziness Epistaxis

All Patients N=1144 All Grades, n (%) Grade 3/4, n (%) 644 (56.3) 483 (42.2) 63 (5.5)

378 (33.0) 143 (12.5) 44 (3.9)

166 (14.5) 152 (13.3) 148 (12.9) 143 (12.5) 105 (9.2) 105 (9.2) 101 (8.8) 91 (8.0) 82 (7.2) 77 (6.7) 72 (6.3) 72 (6.3) 72 (6.3) 69 (6.0) 68 (5.9) 68 (5.9) 61 (5.3) 61 (5.3) 59 (5.2)

13 (1.1) 16 (1.4) 15 (1.3) 18 (1.6) 8 (0.7) 4 (0.3) 22 (1.9) 14 (1.2) 3 (0.3) 1 (0.1) 8 (0.7) 8 (0.7) 0 13 (1.1) 2 (0.2) 1 (0.1) 41 (3.6) 4 (0.3) 8 (0.7)

Adverse events are reported for the entire treatment period.

haematologica | 2016; 101(9)

Ruxolitinib for MF: 1144-patient snapshot of JUMP

Retreatment after ruxolitinib interruption Overall, 207 of the 1144 patients had a therapy interruption of ≥7 days before restarting treatment. Compared with the overall patient population, this subgroup had a higher proportion of patients with primary MF (68.1% versus 58.8%) and hemoglobin level <100 g/L (50.7% versus 40.3%) and a lower median platelet count at baseline (199.5×109/L versus 256×109/L). Interruptions lasted from 7 to 14 days in 41.1% of patients, 15 to 21 days in 26.6% of patients, and >21 days in 32.4% of patients. Treatment interruptions were mostly due to adverse events (92.3%). Adverse events occurring before treatment interruption were primarily grade 3/4 anemia (34.8%) or thrombocytopenia (33.8%); grade 3/4 adverse events occurring in >2% of patients included decreased platelet count (8.2%), pneumonia (4.4%), neutropenia (4.4%), leukopenia (3.4%), abdominal pain (2.9%), and increased gamma-glutamyltransferase (2.4%). Most patients (67.1%) had only one interruption. There were no reports of a withdrawal effect following treatment interruption, and, although exposure to ruxolitinib was more than three times longer after restarting treatment, the rates of adverse events before interruption and after restarting treatment were similar overall. The median duration of exposure was 1.9

months from baseline to interruption and 6.5 months after restarting treatment. The mean daily dose was 30.5 mg prior to treatment interruption and 19.4 mg between restarting treatment and the end of the follow-up. Despite treatment interruption, 68.2% of patients (133/195) experienced a ≥50% reduction in palpable spleen length at any time. This was achieved by 77 patients (58%) before ruxolitinib interruption and 56 patients (42%) after restarting ruxolitinib. From spleen length at restarting treatment, 24% of patients achieved a further 50% reduction, 9% experienced a spleen length increase ≥50%, and 67% remained stable in between. Clinically meaningful improvements in symptoms, as assessed by the FACT-Lym TS, were observed as early as week 4 (mean change from baseline, 9.6), with a trend toward improvement at week 48 (mean change, 6.1).

Patients with intermediate-1-risk myelofibrosis JUMP included patients classified as intermediate-1-risk, a group of patients not included in the COMFORT studies. Overall, 163 intermediate-1-risk patients who started treatment ≥1 year before data cutoff were identified; because risk status was not initially collected on study, a second, later data cutoff date (01 July 2014) was used. The

Figure 2. Hemoglobin levels and platelet counts over time. (A) Hemoglobin levels over time. Boxes represent the upper to lower quartiles and the bars represent the range; the line is connected through the mean values. (B) Platelet counts over time. Boxes represent the upper to lower quartiles and the bars represent the range; the line is connected through the mean values.

haematologica | 2016; 101(9)

1069

H.K. Al-Ali et al.

median age in this cohort of patients was 62 years (range, 25-79 years), with a median time since diagnosis of 17.9 months (range, 0.2-276 months); the median palpable spleen length at baseline was 12 cm (range, 4-45 cm; Online Supplementary Table S5). Overall, 21.5% and 8.0% of patients had hemoglobin <10 g/dL and platelet counts <100×109/L at baseline, respectively. In general, ruxolitinib was well tolerated by intermediate-1-risk patients. At the time of data cutoff, most patients had completed the study per protocol (4.3%) or remained on treatment (76.1%). Reasons for treatment discontinuation included adverse events (11%), withdrawal of consent (2.5%), disease progression (2.5%), investigator’s decision (1.8%), physician’s decision (1.8%), death (1.2%), and protocol deviation (0.6%). The majority of patients received starting doses of 20 mg bid (65.0%) or 15 mg bid (23.9%), with a median exposure of 14.4 months (range, 0.1-19.1 months) and a median daily dose of 30.0 mg/day (range, 6.3-49.4 mg/day). Overall, 64% of patients had dose modifications and 27% had a dose interruption (23% due to adverse events). In this cohort of patients, ruxolitinib demonstrated an adverse event profile consistent with that previously

reported in intermediate-2- and high-risk patients with MF. The most common hematologic adverse events were anemia (all grades, 54.0%; grade 3/4, 24.5%), thrombocytopenia (all grades, 40.5%; grade 3/4, 11.0%; Online Supplementary Table S6). Anemia and thrombocytopenia led to treatment discontinuation in one patient (0.6%) and three patients (1.8%), respectively. Similar to what was observed in the overall JUMP population, mean hemoglobin levels decreased from baseline (118.5 g/L) during the first 8 to 12 weeks and increased to near-baseline levels after week 24. Mean platelet counts decreased from baseline (309.4×109/L) during the first 4 weeks but then remained stable over time. Rates of non-hematologic grade 3/4 adverse events were low overall (<2%), with the exception of asthenia (2.5%; Online Supplementary Table S6). Adverse events leading to treatment discontinuation were reported for 11% of patients (n=18). These occurred in one patient each, with the exception of pyrexia (n=2; 1.2%). Rates of infections were also low: all-grade infections in ≥5% of patients included herpes zoster (8.0%) and bronchitis (6.1%). Grade 3/4 infections occurring in more than two patients included pneumonia and sepsis (both 1.8%). Hepatitis B was reported in one patient (0.6%;

Figure 3. Spleen response. (A) Evaluable patients with a ≥25% decrease from baseline in palpable spleen length. (B) Best percent change from baseline in palpable spleen length at any time by week 48. Each bar represents data from an individual patient. Max: maximum; min: minimum. a Note: −100% change is defined as non-palpable. bOnly patients with spleen length assessments at baseline and at a post-baseline visit were included in this analysis.

1070

haematologica | 2016; 101(9)

Ruxolitinib for MF: 1144-patient snapshot of JUMP

grade 3/4) and led to treatment discontinuation. No tuberculosis was reported in this cohort. Serious adverse events were reported for 22.1% of patients. Serious adverse events occurring in >1% of patients included esophageal variceal hemorrhage and sepsis (both 1.8%) and bronchopneumonia, cardiac failure, pyrexia, pneumonia, acute renal failure, respiratory failure, syncope, and urinary tract infection (1.2% each). There were no reports of progressive multifocal leukoencephalopathy in this cohort of patients. Overall, there were three deaths among intermediate-1-risk patients. Primary causes of death included cardiac failure, sepsis, and undetermined cause (n=1 each). Similar to patients with higher-risk MF, the majority of intermediate-1-risk patients in JUMP experienced clinically meaningful reductions in spleen size and improvements in disease-related symptoms. At weeks 24 and 48, 63.8% and 60.5% of patients achieved a ≥50% reduction from baseline in palpable spleen length, respectively. Additionally, 19.6% and 21.0% of patients had 25% to <50% reductions at weeks 24 and 48, respectively (Online Supplementary Figure S5A). At each assessment, ≥75% of patients had a ≥25% reduction from baseline in palpable spleen length. By week 72, 77.6% of patients had achieved a ≥50% reduction, including 21% (n=34) with complete resolution of splenomegaly (Online Supplementary Figure S5B). The median time to a ≥50% reduction in palpable spleen length was 4.7 weeks (range, 3.1-60.1 weeks), and the estimated probability of maintaining a response was 91% (95% CI, 83%-95%) at 48 weeks and 88% (95% CI, 79%-94%) at 60 weeks.

Overall, 9.6% of patients had a loss of response. In contrast to what was observed in higher-risk patients, the proportion of patients who achieved ≥50% reductions from baseline in palpable spleen length at week 24 was similar between those starting with a ruxolitinib dose of 20 mg bid and those starting with 15 mg bid (65.2% versus 62.9%, respectively). At week 48, the proportion of patients who achieved a spleen response was slightly higher in those who started treatment at 20 mg bid (66.2% versus 55.9%). However, the proportion of patients who achieved a ≥50% reduction from baseline in palpable spleen length at any time was higher in patients who started treatment at 20 mg bid (81.9% versus 68.4%). Clinically meaningful improvements in symptoms were seen as early as 4 weeks after the start of treatment, as evaluated per the FACT-Lym TS and FACIT-Fatigue scale (Online Supplementary Figure S6). Approximately 30% to 40% and 34% to 47% of patients achieved a response at each time point in the FACT-Lym TS and on the FACITFatigue scale, respectively (Online Supplementary Figure S6). Improvements were also observed on other FACT-Lym scales, with the proportion of patients achieving a response ranging from 30% to 43% for the FACT-Lym subscale, 23% to 31% for the FACT-Lym Trial Outcome Index, and 29% to 36% for the FACT-General TS.

Discussion The JUMP study is the most extensive study in MF and includes the largest cohort of patients treated with ruxoli-

Figure 4. Quality of life responses. (A) Proportion of patients achieving a response in the FACT-Lymphoma total score. FACT: Functional Assessment of Cancer a Response was Therapy. defined as the upper limit of the minimally important difference (FACT-Lym total score, 11.2 points). (B) Proportion of patients achieving a response on the FACIT-Fatigue Scale: FACIT: Functional Assessment of Chronic Illness Therapy. a Response was defined as the upper limit of the minimally important difference (FACITFatigue score, 3 points).

haematologica | 2016; 101(9)

1071

H.K. Al-Ali et al. Table 3. Progression-free, leukemia-free, and overall survival estimates.a

First Event, n (%) Progression-free survivalb Total n of events Disease progression Death AML-free survivalc,d Total n of events Transformation to AML Death Overall survivald Total n of deaths

Patients N=1144

Probability Estimates at 24 Weeks (95% CI)

Probability Estimates at 48 Weeks (95% CI)

0.95 (0.94-0.96)

0.91 (0.89-0.93)

0.97 (0.95-0.98)

0.94 (0.92-0.95)

0.97 (0.96-0.98)

0.94 (0.93-0.96)

126 (11.0) 49 77 96 (8.4) 13 83 86 (7.5)

AML: acute myeloid leukemia. aIncludes events that occurred within 28 days after last dose of study drug. bThe time from first dose to date of progressive disease (by International Working Group for Myeloproliferative Neoplasms Research and Treatment criteria) or death. Median follow-up, 48 weeks. cThe time from first dose to earliest date of either first bone marrow blasts ≥20% or first peripheral blood blasts ≥20% for ≥8 weeks or death. dMedian follow-up, 52 weeks.

tinib reported to date. Compared with the COMFORT studies, the JUMP trial seems to have a slightly older population with a longer evolution of MF prior to study entry and a greater proportion of patients with primary MF. Overall, the tolerability of ruxolitinib in JUMP was similar to that observed in the COMFORT studies.9,10 Consistent with the mechanism of action of ruxolitinib, the most common adverse events were anemia and thrombocytopenia; however, these adverse events were manageable and led to treatment discontinuation in only 2.6% and 3.2% of patients, respectively. Although findings are preliminary, concomitant administration of erythropoiesisstimulating agents was well tolerated and did not have a negative impact on the efficacy of ruxolitinib, as was seen in COMFORT-II.14 The rates of infections in JUMP were low and the infections were primarily grade 1/2; no new or unexpected infections were observed. Additionally, no cases of progressive multifocal leukoencephalopathy were reported. Herpes zoster (3.6%) and influenza (3.0%) were the most common viral infections. No hepatitis B reactivation was reported in the 1144-patient cohort, and one case was reported in the intermediate-1-risk group of patients. Prophylaxis for herpes zoster or other infections should be considered on a case-by-case basis and may vary by region and local risk. Of the 86 deaths in the overall survival analysis, 83 were recorded as deaths in the leukemia-free survival analysis; 13 patients had transformation to acute myeloid leukemia. However, the limited follow-up for survival in this study should be taken into consideration. As observed in other studies with ruxolitinib,9,10,12 the majority of patients in JUMP experienced clinically meaningful reductions in spleen size and improvements in disease-related symptoms. Ruxolitinib was also generally safe and well tolerated and provided meaningful reductions in splenomegaly and symptoms in patients who restarted ruxolitinib after treatment interruption. Consistent with ruxolitinib’s mechanism of action, hematologic adverse events were the primary cause of treatment interruptions. In the 207 patients who restarted treatment after therapy interruption, ruxolitinib provided reductions in palpable spleen length and symptoms prior to and after treatment interruption. After restarting treatment, patients were able to stay on ruxolitinib at a median dose of ≈10 mg bid, and most patients did not 1072

require another interruption. Rates of adverse events did not increase after restarting treatment. Additionally, the discontinuation rate after restarting treatment was comparable to that observed in the overall study population. Of note, there was no evidence of a withdrawal effect after discontinuation of ruxolitinib treatment in this cohort of patients. JUMP also includes patients with intermediate-1-risk MF, a risk group that was not included in the COMFORT studies. Importantly, the IPSS was used to determine prognostic score in this study because at the time the protocol was released in January 2011, the Dynamic IPSS had only recently been published (October 2010). As mentioned previously, the adverse event profile of ruxolitinib in intermediate-1-risk patients was consistent with that previously reported in higher-risk patients.9,10 Rates of non-hematologic adverse events were similar in both groups of patients, although those with higher-risk MF reported higher rates of fatigue (12.9% versus 5.5%). Herpes zoster reactivation was observed in both groups, with intermediate-1–risk patients reporting higher rates of reactivation (8.0% versus 3.6%). Patients with intermediate-1-risk MF achieved clinically meaningful reductions in spleen size and symptom improvement consistent with those seen in intermediate-2- and high-risk patients enrolled in this study. At week 24, slightly more patients with intermediate-1-risk MF had achieved a ≥50% reduction from baseline in palpable spleen length than had patients in the overall population (63.8% versus 56.9%, respectively); the rates were similar at week 48 (60.5% versus 62.3%). Additionally, a similar proportion of patients in each cohort achieved a ≥50% reduction from baseline in palpable spleen length at any time (overall population, 69%; intermediate-1-risk patients, 77.6%). The median time to a spleen response was also similar (4.7 versus 5.1 weeks). Likewise, the proportion of patients who achieved clinically meaningful improvements in symptoms (≈30% to 40%) was within the expected range and consistent with that seen in the overall JUMP population (≈45% to 50%). These findings support those observed in the UK ROBUST study12 as well as those from real-world clinical evidence of ruxolitinib use in patients with lower-risk MF20 and indicate that ruxolitinib is an effective treatment for patients with intermediate-1-risk disease. Overall, findings from the JUMP study confirm the effihaematologica | 2016; 101(9)

Ruxolitinib for MF: 1144-patient snapshot of JUMP

cacy and safety of ruxolitinib in the treatment of patients with MF. Ruxolitinib provided clinically meaningful reductions in spleen size and symptoms, including for those patients with intermediate-1-risk disease, with a safety and efficacy profile consistent with that observed in the phase 3 COMFORT studies. Furthermore, JUMP is a global study conducted in a setting that resembles routine clinical practice. Findings from this study will help guide clinicians in the management of their patients with MF and

References 8. 1. Mesa RA, Verstovsek S, Cervantes F, et al. Primary myelofibrosis (PMF), post polycythemia vera myelofibrosis (post-PV MF), post essential thrombocythemia myelofibrosis (post-ET MF), blast phase PMF (PMFBP): consensus on terminology by the International Working Group for Myelofibrosis Research and Treatment (IWG-MRT). Leuk Res. 2007;31(6):737-740. 2. Swerdlow SH, Campo E, Harrison NL, et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Vol 2. 4th ed. Geneva, Switzerland: World Health Organization; 2008. 3. Vardiman JW, Thiele J, Arber DA, et al. The 2008 revision of the World Health Organization (WHO) classification of myeloid neoplasms and acute leukemia: rationale and important changes. Blood. 2009;114(5):937-951. 4. Tefferi A. Primary myelofibrosis: 2013 update on diagnosis, risk-stratification, and management. Am J Hematol. 2013;88(2): 141-150. 5. Mesa RA, Niblack J, Wadleigh M, et al. The burden of fatigue and quality of life in myeloproliferative disorders (MPDs): an international Internet-based survey of 1179 MPD patients. Cancer. 2007;109(1):68-76. 6. Abdel-Wahab OI, Levine RL. Primary myelofibrosis: update on definition, pathogenesis, and treatment. Annu Rev Med. 2009;60:233-245. 7. Mesa RA, Schwager S, Radia D, et al. The Myelofibrosis Symptom Assessment Form (MFSAF): an evidence-based brief inventory to measure quality of life and symptomatic

haematologica | 2016; 101(9)

10.

11.

12.

13.

14.

may help shape the current treatment paradigm, ultimately maximizing the benefits that patients can derive from treatment. Acknowledgments We thank Catherine Bouard and Raj Vadde, employees of Novartis, for their substantial contributions to the JUMP study. Editorial assistance was provided by Karen Chinchilla, PhD, and was funded by Novartis.

response to treatment in myelofibrosis. Leuk Res. 2009;33(9):1199-1203. Cervantes F, Dupriez B, Pereira A, et al. New prognostic scoring system for primary myelofibrosis based on a study of the International Working Group for Myelofibrosis Research and Treatment. Blood. 2009;113(13):2895-2901. Verstovsek S, Mesa RA, Gotlib J, et al. A double-blind, placebo-controlled trial of ruxolitinib for myelofibrosis. N Engl J Med. 2012;366(9):799-807. Harrison C, Kiladjian JJ, Al-Ali HK, et al. JAK inhibition with ruxolitinib versus best available therapy for myelofibrosis. N Engl J Med. 2012;366(9):787-798. Cervantes F, Kiladjian JJ, Niederwieser D, et al. Long-term efficacy, safety, and survival findings from COMFORT-II, a phase 3 study comparing ruxolitinib with best available therapy for the treatment of myelofibrosis. Blood. 2012;120(21):801. Mead AJ, Milojkovic D, Knapper S, et al. Response to ruxolitinib in patients with intermediate-1-, intermediate-2-, and highrisk myelofibrosis: results of the UK ROBUST trial. Br J Haematol. 2015; 170(1):29-39. Barosi G, Mesa RA, Thiele J, et al. Proposed criteria for the diagnosis of post-polycythemia vera and post-essential thrombocythemia myelofibrosis: a consensus statement from the International Working Group for Myelofibrosis Research and Treatment. Leukemia. 2008;22(2):437-438. McMullin MF, Harrison CN, Niederwieser D, et al. The use of erythropoiesis-stimulating agents with ruxolitinib in patients with myelofibrosis in COMFORT-II: an open-

15.

16.

17.

18.

19.

20.

label, phase 3 study assessing efficacy and safety of ruxolitinib versus best available therapy in the treatment of myelofibrosis. Exp Hematol Oncol. 2015;4:26. Tefferi A, Barosi G, Mesa RA, et al. International Working Group (IWG) consensus criteria for treatment response in myelofibrosis with myeloid metaplasia, for the IWG for Myelofibrosis Research and Treatment (IWG-MRT). Blood. 2006;108(5): 1497-1503. Cella DF, Tulsky DS, Gray G, et al. The Functional Assessment of Cancer Therapy scale: development and validation of the general measure. J Clin Oncol. 1993;11(3): 570-579. Webster K, Cella D, Yost K. The Functional Assessment of Chronic Illness Therapy (FACIT) Measurement System: properties, applications, and interpretation. Health Qual Life Outcomes. 2003;1:79. Carter GC, Liepa AM, Zimmermann AH, Morschhauser F. Validation of the Functional Assessment of Cancer Therapy-Lymphoma (FACT-LYM) in patients with relapsed/refractory mantle cell lymphoma. Blood 2008;112(11):2376 Cella D, Eton DT, Lai JS, Peterman AH, Merkel DE. Combining anchor and distribution-based methods to derive minimal clinically important differences on the Functional Assessment of Cancer Therapy (FACT) anemia and fatigue scales. J Pain Symptom Manage. 2002;24(6):547-561. Davis KL, Cote I, Kaye JA, et al. Real-world assessment of clinical outcomes in patients with lower-risk myelofibrosis receiving treatment with ruxolitinib. Adv Hematol. 2015;2015;848473.

1073

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Acute Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2016 Volume 101(9):1074-1081

Clinical characteristics and prognosis of acute myeloid leukemia associated with DNA-methylation regulatory gene mutations Takeshi Ryotokuji,1* Hiroki Yamaguchi,1* Toshimitsu Ueki,2 Kensuke Usuki,3 Saiko Kurosawa,4 Yutaka Kobayashi,5 Eri Kawata,5 Kenji Tajika,6 Seiji Gomi,6 Junya Kanda,7 Anna Kobayashi,1 Ikuko Omori,1 Atsushi Marumo,1 Yusuke Fujiwara,1 Shunsuke Yui,1 Kazuki Terada,1 Keiko Fukunaga,1 Tsuneaki Hirakawa,1 Kunihito Arai,1 Tomoaki Kitano,1 Fumiko Kosaka,1 Hayato Tamai,1 Kazutaka Nakayama,1 Satoshi Wakita,1 Takahiro Fukuda,4 and Koiti Inokuchi1

1 Department of Hematology, Nippon Medical School, Tokyo; 2Department of Hematology, Nagano Red Cross Hospital; 3Department of Hematology, NTT Medical Center Tokyo; 4 Department of Hematopoietic Stem Cell Transplantation, National Cancer Center Hospital, Tokyo; 5Department of Hematology, Japanese Red Cross Kyoto Daini Hospital; 6Department of Hematology, Yokohama Minami Kyousai Hospital, Kanagawa; and 7Division of Hematology, Saitama Medical Center Jichi Medical University, Japan

*TR and HY contributed equally to this work

ABSTRACT

Correspondence: y-hiroki@fd6.so-net.ne.jp

Received: January 23, 2016. Accepted: May 30, 2016. Pre-published: May 31, 2016. doi:10.3324/haematol.2016.143073

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1074

1074

n recent years, it has been reported that the frequency of DNAmethylation regulatory gene mutations – mutations of the genes that regulate gene expression through DNA methylation – is high in acute myeloid leukemia. The objective of the present study was to elucidate the clinical characteristics and prognosis of acute myeloid leukemia with associated DNA-methylation regulatory gene mutation. We studied 308 patients with acute myeloid leukemia. DNA-methylation regulatory gene mutations were observed in 135 of the 308 cases (43.8%). Acute myeloid leukemia associated with a DNA-methylation regulatory gene mutation was more frequent in older patients (P<0.0001) and in patients with intermediate cytogenetic risk (P<0.0001) accompanied by a high white blood cell count (P=0.0032). DNA-methylation regulatory gene mutation was an unfavorable prognostic factor for overall survival in the whole cohort (P=0.0018), in patients aged ≤70 years, in patients with intermediate cytogenetic risk, and in FLT3-ITD-negative patients (P=0.0409). Among the patients with DNA-methylation regulatory gene mutations, 26.7% were found to have two or more such mutations and prognosis worsened with increasing number of mutations. In multivariate analysis DNA-methylation regulatory gene mutation was an independent unfavorable prognostic factor for overall survival (P=0.0424). However, patients with a DNA-methylation regulatory gene mutation who underwent allogeneic stem cell transplantation in first remission had a significantly better prognosis than those who did not undergo such transplantation (P=0.0254). Our study establishes that DNA-methylation regulatory gene mutation is an important unfavorable prognostic factor in acute myeloid leukemia.

Introduction Acute myeloid leukemia (AML) is a heterogeneous disease whose onset involves a variety of chromosomal abnormalities and gene mutations.1,2 To improve the outcome of AML treatment, it is very important to establish a prognosis from cytogenetic analysis and provide accordingly differentiated treatment.1,2 Standard haematologica | 2016; 101(9)

DNA-methylation regulatory gene mutations in AML

chemotherapy is given to patients with a favorable prognosis according to their cytogenetic profile, whereas allogeneic transplantation in first remission is actively promoted for patients with an unfavorable cytogenetic prognosis.3 Meanwhile, for the approximately 60% of cases that fall into the intermediate prognosis group, clinicians are seeking to provide a clear prognosis and differentiated therapy based on cytogenetic analysis, which have not been available for this group of patients. In the latter half of the 2000s, gene mutations in AML were successively discovered, and attempts have been made to use these to provide a differentiated prognosis. The most important of these gene mutations for differentiated prognosis are FMS-like tyrosine kinase 3 internal tandem duplications (FLT3-ITD), nucleophosmin 1 (NPM1) gene mutation, and CCAAT/enhancer binding protein A (CEBPA) gene mutation.49 In 2008, Richard et al. reported that AML cases that were FLT3-ITD-negative and NPM1-mutation-positive, or that were accompanied by CEBPA biallelic mutation, were associated with a favorable prognosis and that allogeneic hematopoietic stem cell transplantation in first remission was not indicated.9,10 It was also reported that FLT3-ITDpositive AML has a very unfavorable prognosis, which might be improved by allogeneic hematopoietic stem cell transplantation in first remission.10 These findings have been integrated as prognostic factors in the European Leukemia Net (ELN) and National Comprehensive Cancer Network (NCCN) guidelines,12,13 which are becoming widely applied in clinical practice. However, these gene mutations are observed in only around 30% of cases with intermediate cytogenetic prognosis, meaning that differentiation of prognosis is still insufficient. In recent years, the use of next-generation sequencers has facilitated energetic exploration of gene mutations, leading to the discovery of other AML-related gene mutations,14-17 and mutations in the genes that regulate DNA methylation, such as DNA methyltransferase 3 alpha (DNMT3A), Tet methylcytosine dioxygenase 2 (TET2), isocitrate dehydrogenase1 (IDH1), and isocitrate dehydrogenase 2 (IDH2).18-23 It has been suggested that these gene mutations may also have prognostic relevance in some cases of AML, but this view has yet to become established. Moreover, a further series of gene mutations has recently been discovered, so that differentiated prognosis based on gene mutations has actually become more confused.14,16,18,24 The focus of the present study was DNA-methylation regulatory gene mutations (DMRGM), which are mutations of the genes that regulate gene expression through DNA methylation (IDH1, IDH2, DNMT3A, TET2). Studies have been carried out of the various DMRGM and other genes reported as indicating an unfavorable prognosis in AML.18-20,22, 23,25-30 Our group has also reported that AML with DMRGM at onset is associated with a high frequency of FLT3-ITD at relapse and has an unfavorable prognosis.31 However, since comprehensive gene mutation analysis using a next-generation sequencer is costly, there have so far been few reports on DMRGMbased prognostic analysis covering a large number of cases. In particular, there are very few reports on large cohorts in which there is a combined analysis of both TET2 mutation, which is mutually exclusive with IDH1 and IDH2 mutations, and DNMT3A mutation.20 Since IDH1/2 mutations and TET2 mutation are mutually complementary, DMRGM need to be subjected to integrated haematologica | 2016; 101(9)

analysis as a group, but there have so far been no reports of such analysis having been performed. We therefore analyzed the clinical characteristics of a group of AML patients with DMRGM and the prognostic impact of these mutations.

Methods Patients We studied 308 patients with de novo AML (excluding M3) treated at Nippon Medical School Hospital or its affiliated institutions. A comprehensive genetic mutational analysis, as described below, was conducted among patients with â&#x2030;Ľ20% blasts in bone marrow or peripheral blood. The study was conducted in accordance with the Declaration of Helsinki; written informed consent was obtained from the participants, and the patients were analyzed and treated with respect for their welfare and free will. The study protocol was approved by our institutional review board.

Screening for cytogenetic mutations G-band analysis was performed on bone marrow samples obtained from patients at initial presentation. When it was difficult to obtain bone marrow samples, peripheral blood was used instead. For patients suspected of having M2, M3, or M4e AML based on the French-American-British classification, fluorescence in situ hybridization analysis was used to search additionally for RUNX1-RUNX1T1, PML-RARA, and CBFB-MYH11 mutations. The cytogenetic prognosis was then classified in accordance with the system recommended by the ELN.

Screening for molecular genetic abnormalities in all exons of 20 genes and hotspots of eight genes An oligonucleotide library was generated by emulsion polymerase chain reaction using order-made probes designed against the exons of the following 20 genes: TET2 (HGNC:25941), DNMT3A (HGNC:2978), ASXL1 (HGNC:18318), KMT2A (HGNC:7132: the HGNC nomenclature of MLL has recently changed to KMT2A), RUNX1 (HGNC:10471), KIT (HGNC:6342), TP53 (HGNC:11998), PTPN11 (HGNC:9644), GATA2 (HGNC:4171), WT1 (HGNC:12796), STAG2 (HGNC:11355), RAD21 (HGNC:9811), SMC1A (HGNC:11111), SMC3 (HGNC:2468), DAXX (HGNC:2681), BCOR (HGNC:20893), BCORL1 (HGNC:25657), NF1 (HGNC:7765), DDX41 (HGNC:18674), and PHF6 (HGNC:18145). The library was sequenced with the next-generation sequencer Ion PGMTM. With respect to detected mutations, the NCBI and COSMIC databases were used to search for polymorphisms and cancer-related mutations. For newly identified mutations, genetic polymorphisms were checked using Sanger sequencing with remission-stage samples. Genes located in known hot spots (FLT3-ITD and FLT3-TKD, NPM1, IDH1, IDH2, NRAS, and KRAS), those for which probe design for emulsion sequencing was difficult (CEBPA), and those for which analysis with Ion PGMTM was difficult (KMT2A-PTD), were analyzed using previously reported methods.31

Statistical analysis The primary endpoint was overall survival of DMRGM-positive AML patients in the study cohort. A hazard ratio of 0.707 and an overall survival rate of 50% were assumed for the sample size calculation. Recruitment of approximately 300 patients allowed for a power of 80% in detecting a difference of that size with a type I error of 5%. During the study period, there were 80 deaths among the 135 DMRGM-positive AML patients. This provided an 80% 1075

T. Ryotokuji et al.

statistical power to detect a hazard ratio of 0.542 with a significance level (alpha) of 0.05 (two-tailed) regarding overall survival, which was defined as the time interval from the date of diagnosis to the date of death. Relapse-free survival for patients who had achieved complete remission was calculated as the time interval from the date of complete remission to the date of relapse. The c2 test was used to test the association between categorical variables and the presence and absence of mutations. The Fisher exact test was used if the expected frequency of an event was less than five in any cell of a 2 × 2 table. The non-parametric Mann– Whitney U test was used to determine the statistical significance of differences in median values. All statistical tests were twosided. The Kaplan-Meier method and log-rank test were applied to analyze overall survival and relapse-free survival. With respect to prognostic factors, multivariate analysis was conducted with the Cox proportional hazards model. A stepwise backward procedure selection model was used to extract independent events. Events at a significance level of P≤0.20 were analyzed. Statistical analyses were performed using GraphPad Prism (version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) and IBM SPSS Statistics (version 21.0 for Windows, IBM Corp., Armonk, NY, USA), while the power calculation was performed using GraphPad StatMate (version 2.00 for Windows).

Results Patients’ background The average age of the 308 patients studied was 54.1 years (range, 17-86 years). There were 181 males (58.8%) and 127 females (41.2%). Based on cytogenetic analysis, 60 cases (19.5%) were assigned to the favorable risk group, 51 (16.6%) to the unfavorable risk group, and 184 (59.7%) to the intermediate risk group. Overall, 148 (48.1%) had a normal karyotype (Table 1, Online Supplementary Table S1). Gene mutations were observed in 265 cases (86.0%), with the most frequent being NPM1 mutation (88 cases, 28.6%), DNMT3A mutation (71, 23.1%), and FLT3-ITD (65, 21.1%) (Online Supplementary Figure S1, Online Supplementary Table S2). Gene mutations with an incidence of ≤3% were excluded from the analysis as they were too infrequent to be studied as prognostic factors (Online Supplementary Figure S1).

Clinical characteristics of patients with DNA-methylation regulatory gene mutations A DNMT3A mutation was found in 71 cases (23.1%), TET2 mutation in 57 (18.5%), IDH2 mutation in 28

Table 1. Clinical background of the AML patients studied.

All n=308 Age, years Mean Range Sex Male Female White cell count, x109/L Mean Range Cytogenetic risk group Favorable Intermediate Adverse Unknown FAB classification M0 M1 M2 M4 M4e M5 M6 M7 Not determined Induction therapy (IDA/DNR/ACR)+AraC Others(AVVV, VP16+AraC) Unknown Stem cell transplantation First CR Second CR ≥Third CR/disease

54.1 17-86 181 127

58.8 41.2

57.2 0.3-677

DMRGM-positive n=135 %

DMRGM-negative n=173 %

P value

59.0 18-86

50.2 17-82

≺0.0001

72 63

53.3 46.7

75.2 0.3-677

109 64

63.0 37.0

43.1 0.6-483

0.1025

0.0032 ≺0.0001

60 184 51 13

19.5 59.7 16.6 4.2

4 109 16 6

3.0 80.7 11.9 4.4

56 75 35 7

32.4 43.2 20.2 4.0

17 75 115 40 12 33 7 1 8

5.5 24.4 37.3 13.0 3.9 10.7 2.3 0.3 2.6

5 42 39 22 0 21 2 1 3

3.7 31.1 28.9 16.3 0.0 15.6 1.5 0.7 2.2

12 33 76 18 12 12 5 0 5

6.9 19.1 43.9 10.4 6.9 6.9 2.9 0.0 2.9

0.3150 0.0163 0.0089 0.1712 0.0015 0.0246 0.4729 0.4383 1.0000

292 3 13

94.8 1.0 4.2

125 2 8

82.6 1.5 5.9

167 1 5

96.5 0.6 2.9

0.2989

37 13 36

12.0 4.2 11.7

14 4 12

10.4 3.0 8.9

23 9 24

13.3 5.2 13.9

0.4831 0.4015 0.2122

CR: complete remission. IDA: idarubicin; DNR: daunorubicin; ACR: aclarubicin; AraC: cytarabine; AVVV: cytarabine+etoposide+vincristine+vinblastine; VP16: etoposide.

1076

haematologica | 2016; 101(9)

DNA-methylation regulatory gene mutations in AML

(9.1%), and IDH1 mutation in 17 (5.5%). TET2 mutation was more frequent in older patients (average 63.5 years, P<0.0001) (Online Supplementary Table S3). High white blood cell counts were found in patients with mutations of DNMT3A (average 86.6x109/L, P=0.0028) and TET2 (average 80.7x109/L, P=0.0389). Mutations in DNMT3A (P<0.0001), TET2 (P=0.0087), and IDH2 (P=0.0199) were more frequent in the intermediate cytogenetic risk group (Online Supplementary Table S3). A DMRGM was found in 135 cases (43.8%), indicating that this group of gene mutations occurs with very high frequency in AML. There were 99 cases (73.3%) with one DMRGM (DMRGM1), 34 (25.2%) with two (DMRGM2), and two (1.5%) with three (DMRGM3). No cases had four or more DMRGM (Figure 1). Compared to AML without DMRGM (non-DMRGM), AML with DMRGM was more frequent in older patients (DMRGM: average 59.0 years; non-DMRGM: average 50.2 years; P<0.0001), in patients with high white blood cell count (DMRGM: average 75.2 x109/L; non-DMRGM: average 43.1x109/L; P=0.0032), and in the intermediate cytogenetic risk group (DMRGM: 80.7%; non-DMRGM: 43.4%; P<0.0001) (Table 1).

Overlapping gene mutations DNMT3A mutation was frequently present together with FLT3-ITD (P=0.0005), FLT3-TKD (P=0.0200), and mutations of PTPN11 (P=0.0254), NPM1 (P<0.0001), and IDH2 (P=0.0002), but was mutually exclusive with NRAS mutation (P=0.0364) and CEBPA double mutation (CEBPA dm, P=0.0265). TET2 mutation was frequently present together with NPM1 mutation (P=0.0031) and CEBPA monoallelic mutation (P=0.0441), but was mutually exclusive with NRAS mutation (P=0.0229), IDH1 mutation (P=0.0497), and IDH2 mutation (P=0.0380) (Online Supplementary Table S4). DMRGM was frequently present together with FLT3-ITD (P=0.0007) and mutations of PTPN11 (P=0.0190) and NPM1 (P<0.0001), but was mutually exclusive with mutations of NRAS (P=0.0035) and WT1 (P=0.0241) (Online Supplementary Table S4).

Prognostic analysis in all subjects With regards to overall survival, the factors associated

with an unfavorable prognosis were age >70 years (P<0.0001), adverse cytogenetic risk (P<0.0001), FLT3-ITD (P<0.0001), KMT2A-PTD (P=0.0152), DNMT3A mutation (P=0.0017), TET2 mutation (P=0.0043), and TP53 mutation (P<0.0001). In recent years, the development of a nonmyeloablative regimen for hematopoietic stem cell transplantation has made it possible also for patients aged 65 to 69 years to undergo allogeneic hematopoietic stem cell transplantation. In the present study, analysis of prognosis was therefore stratified between patients aged â&#x2030;¤70 and >70 years. Favorable cytogenetic risk (P<0.0001), allogeneic hematopoietic stem cell transplantation (in first or second remission) (P<0.0001), and CEBPA dm (P=0.0282) were associated with a favorable prognosis (Online Supplementary Table S5). As regards relapse-free survival, the factors associated with an unfavorable prognosis were adverse cytogenetic risk (P=0.0210), FLT3-ITD (P<0.0001), TET2 mutation (P=0.0219), and TP53 mutation (P=0.0011), whereas allogeneic hematopoietic stem cell transplantation (in first remission) (P<0.0001) and NRAS mutation (P=0.0142) were associated with a favorable prognosis (Online Supplementary Table S5).

Stratified analysis of FLT3-ITD-negative patients aged below 70 years with intermediate cytogenetic prognosis Age >70 years, adverse cytogenetic risk, and FLT3-ITD were powerful poor prognostic factors. Rates of overall survival and relapse-free survival were therefore analyzed following stratification based on age â&#x2030;¤70 years, intermediate cytogenetic prognosis, and FLT3-ITD negativity. For overall survival, DNMT3A mutation was associated with an unfavorable prognosis (P=0.0429). On the other hand, allogeneic hematopoietic stem cell transplantation (in first or second remission) was associated with a favorable prognosis (P=0.0283) (Online Supplementary Table S5). For relapse-free survival, TP53 mutation was associated with an unfavorable prognosis (P=0.0068), while, allogeneic hematopoietic stem cell transplantation (in first remission) was associated with a favorable prognosis (P=0.0008) (Online Supplementary Table S5).

Figure 1. Frequency and overlap of DMRGM. There were 173 cases of DMRGMnegative AML (DMRGM0), 99 cases with one DMRGM (DMRGM1: 10 cases with IDH1, 13 with IDH2, 35 with DNMT3A, and 41 with TET2), 34 cases with two DMRGM (DMRGM2: 6 with IDH1+DNMT3A, 13 with DH2+DNMT3A, and 15 with DNMT3A+TET2), and two cases with three DMRGM (DMRGM3: 1 with IDH1+IDH2+DNMT3A and 1 with IDH2+DNMT3A+TET2).

haematologica | 2016; 101(9)

1077

T. Ryotokuji et al.

Prognostic impact of DNA-methylation regulatory gene mutations

P=0.0651) (Figure 3A). As far as concerns relapse-free survival, DMRGM2+3 was associated with a poorer prognosis than DMRGM0 (P=0.0244) and DMRGM1 (P=0.0824) (Figure 3B). Additionally, overall and relapse-free survival rates were analyzed following stratification based on age ≤70 years, intermediate cytogenetic risk, and FLT3-ITD negativity. The overall survival rate of patients with DMRGM2+3 was significantly lower than that of patients with DMRGM0 (P=0.0189) (Figure 3C). The relapse-free survival rate of patients with DMRGM2+3 tended to be worse than that of patients with DMRGM1, but not significantly so (P=0.0600) (Figure 3D).

The significance of DMRGM as a prognostic factor was investigated. With regards to overall survival, cases with DMRGM had a significantly poorer prognosis than cases without DMRGM (P=0.0018) (Figure 2A). Furthermore, patients with DMRGM tended to have a higher rate of relapse than patients without DMRGM, but the difference was not statistically significant (Figure 2B). Additionally, overall and relapse-free survival rates were analyzed following stratification based on age ≤70 years, intermediate cytogenetic risk, and FLT3-ITD negativity. With regards to overall survival, cases with DMRGM had a significantly poorer prognosis than cases without DMRGM (P=0.0409) (Figure 2C). However, there was no significant difference in relapse-free survival between patients with and without DMRGM (Figure 2D).

Prognostic significance of DNA-methylation regulatory gene mutation combinations In the present analysis, prognosis was poor for patients with DMRGM, but was found to be poorer still for those with DMRGM2+3. We, therefore, investigated whether the combinations of DMRGM influence prognosis in patients with DMRGM2+3. The 15 patients with DNMT3A mutation accompanied by TET2 mutation (DNMT3A mutation+/TET2 mutation+) were compared with the 20 patients with DNMT3A mutation accompanied by IDH mutation (DNMT3A mutation+/IDH1 or IDH2 mutation+). No clear significant difference was

Prognostic impact of number of DNA-methylation regulatory gene mutations Overall and relapse-free survival rates were studied relative to the number of DMRGM mutations. The greater the number of DMRGM, the poorer the prognosis (DMRGM0 versus DMRGM2+3: P<0.0001; DMRGM1 versus DMRGM2+3: P=0.0244; DMRGM0 versus DMRGM1:

Figure 2. Overall and relapse-free survival rates in AML cases with and without DMRGM. (A) Overall survival rate for all cases. (B) Relapse-free survival rate for all cases. (C) Overall survival rate in FLT3-ITD-negative cases aged ≤ 70 years with intermediate cytogenetic prognosis. (D) Relapse-free survival rate in FLT3-ITD-negative cases aged ≤ 70 years with intermediate cytogenetic prognosis. HR: Hazard ratio.

1078

haematologica | 2016; 101(9)

DNA-methylation regulatory gene mutations in AML

established in overall survival rate (Online Supplementary Figure S2A), but DNMT3A mutation+/TET2 mutation+ was associated with a poorer relapse-free survival rate (P=0.0349) (Online Supplementary Figure S2B).

Multivariate analysis Multivariate analysis (Cox proportional hazard model) via the step-wise method was carried out using the following variables: age >70 years, poor cytogenetic risk, transplantation time (in first or second complete remission), DMRGM, and gene mutation associated with poor prognosis. The following were identified as independent unfavorable prognostic factors for overall survival: age >70 years (P=0.0001); adverse cytogenetic risk (P<0.0001); DMRGM (P=0.0424); FLT3-ITD (P<0.0001); and TP53 mutation (P=0.0003) (Table 2), whereas the independent unfavorable prognostic factors for relapse-free survival were: unfavorable cytogenetic risk (P=0.0005); FLT3-ITD (P<0.0001); TP53 mutation (P=0.0089); and KMT2A-PTD (P=0.0236) (Table 2).

Efficacy of allogeneic transplantation in first remission for patients with DNA-methylation regulatory gene mutation-positive acute myeloid leukemia The efficacy of allogeneic transplantation in first remission for patients with DMRGM aged ≤70 years was examined. In terms of overall survival, patients with DMRGM who underwent allogeneic stem cell transplantation in

first remission [SCT (1st CR)+/DMRGM+] had a significantly more favorable prognosis than those who did not undergo such transplantation [SCT(1st CR)-/DMRGM+] (P=0.0254) (Figure 4A). Likewise, in terms of relapse-free survival, SCT (1st CR)+/DMRGM+ cases had a significantly more favorable prognosis than SCT (1st CR)-/DMRGM+ cases (P=0.0049) (Figure 4B). A similar analysis was undertaken in cases aged ≤70 years, cases with intermediate cytogenetic risk, and FLT3-ITD-negative cases, but as the number of SCT(1st CR)+/DMRGM+ cases was low, at just seven, the efficacy of allogeneic stem cell transplantation in first remission is not shown here.

Discussion It was confirmed that DMRGM are very frequently present in AML, being observed in 135 of the 308 cases (43.8%) studied. A DMRGM was an unfavorable prognostic factor for overall survival in the whole group and in patients aged ≤70 years, in patients with an intermediate cytogenetic risk group, and in FLT3-ITD-negative patients. Allogeneic stem cell transplantation in first remission may improve the prognosis of cases with DMRGM mutation. Until now, the significance of individual DMRGM as prognostic factors in AML has not been clear. Regarding DNMT3A, the most frequent of the DMRGM, Ley et al. reported that DNMT3A R882 mutation and non-R882

Figure 3. Overall and relapse-free survival rates in AML patients with no DMRGM, one DMRGM, and two or more DMRGM (A) Overall survival rate for all cases. (B) Relapse-free survival rate for all cases. (C) Overall survival rate in FLT3-ITD-negative cases aged ≤ 70 years with intermediate cytogenetic prognosis. (D) Relapse-free survival rate in FLT3-ITD-negative cases aged ≤ 70 years with intermediate cytogenetic prognosis. HR: Hazard ratio.

haematologica | 2016; 101(9)

1079

T. Ryotokuji et al.

mutation are both associated with unfavorable prognosis, independently of whether FLT3-ITD is present.18 Meanwhile, Patel et al. reported that patients with intermediate cytogenetic risk who have a DNMT3A mutation do not have an unfavorable prognosis even if FLT3-ITD is not present.24 Because of the expense associated with comprehensive gene mutation analysis using a next-generation sequencer, there have so far been few reports on DMRGM-based prognostic analysis involving a large number of patients. Such studies consist of the one by Patel et al., based on 398 cases, and one by Hou et al., based on 500 cases.24,25 Our analysis covered four DMRGM – IDH1, IDH2, DNMT3A, and TET2 – in 308 cases of AML, a large cohort suggesting reliable results. Moreover, ours is the first study in which AML prognosis was explored with division into groups based on DMRGM. Our study revealed that the greater the number of DMRGM, the poorer the prognosis is. It has been reported that increasing numbers of DMRGM are associated with unfavorable outcome in a number of hematologic malignancies. For example, Papaemmanuil et al. described that leukemia-free survival in patients with myelodysplastic syndromes becomes shorter with increasing number of gene mutations.32 Guglielmelli et al. also reported that, in primary myelofibrosis, the greater the number of mutations of ASXL1, EZH2, SRSF2, and IDH1/2, the poorer the patients’ prognosis is.33 We have also reported recently that three or more gene mutations is an unfavorable prognostic factor in de novo AML.34 According to whole-exon mutation analysis using a next-generation sequencer, there are on average 2.5-5.0 gene mutations per case in AML,14,18,34 and it has been established that the onset of AML requires the combination of a number of gene mutations.16,17 These findings suggest that cases with a large number of gene mutations have a correspondingly high level of genomic instability and point to the strong possibility of mutations in genes not yet examined.34 This is thought to result in a poorer prognosis. Our finding that prognosis in AML worsens with increasing number of DMRGM suggests that to improve prognostic analysis in AML it will be important not only to focus on each individual gene mutation, but also to seek to identify overall genomic instability in individual cases.

In a whole-exon analysis of 12,380 healthy subjects, Genovese et al. found that healthy subjects in whom DMRGM mutations were observed had a high rate of myeloid malignancies in hematopoietic organs a number of years later.35 Shlush et al. demonstrated that, in AML with DNMT3A mutation, the mutation also occurred in hematopoietic stem cells with normal differentiation potential, so that even in cases in which chemotherapy induced remission, the DNMT3A mutation remaining in the stem cells subsequently led the resulting blood cells to proliferate and cause relapse of the leukemia.36 If this finding applies to all DMRGM-positive AML cases, then chemotherapy alone cannot be expected to be effective in DMRGM-positive AML. As indicated by Patel et al., for young patients with DMRGM-positive AML, allogeneic hematopoietic stem cell transplantation in first remission may be the only curative therapy available. The findings above indicate that the prognosis of patients with DMRGM-positive AML is unfavorable, Table 2. Multivariate analysis of prognostic factors.

Hazard ratio P value Overall survival Age>70 years Adverse cytogenetic risk SCT (first or second CR) DMRGM FLT3-ITD RUNX1 TP53 KMT2A-PTD GATA2 Relapse free survival Age>70 years Adverse cytogenetic risk SCT(first remission) FLT3-ITD NRAS RUNX1 TP53 KMT2A-PTD

95% Confidence interval

2.2901 2.7025 0.2917 1.5782 2.6250 0.6524 2.5915 1.5755 0.5523

0.0001 <0.0001 <0.0001 0.0424 <0.0001 0.0918 0.0003 0.0983 0.1717

1.5049－3.4848 1.7472－4.1802 0.1658－0.5133 1.0158－2.4521 1.7393－3.38945 0.3971－1.0719 1.5488－4.3362 0.9191－2.7010 0.2358－1.2939

1.4937 2.6814 0.1466 3.1407 0.5634 0.6602 2.3102 2.2479

0.1443 0.0005 <0.0001 <0.0001 0.0888 0.1802 0.0089 0.0236

0.8716－2.5599 1.5347－4.6848 0.0670－0.3211 2.0494－4.8131 0.2910－1.0909 0.3597－1.2117 1.2341－4.3244 1.1147－4.5330

SCT：stem cell transplantation, CR: complete remission.

Figure 4. Efficacy of allogeneic stem cell transplantation in first remission for DMRGM-positive AML cases aged 70 years or below. (A) Overall survival rate for all cases. (B) Relapse-free survival rate for all cases. HR: hazard ratio; SCT: stem cell transplantation; 1stCR: first complete remission.

1080

haematologica | 2016; 101(9)

DNA-methylation regulatory gene mutations in AML

that the outcome worsens with increasing numbers of such mutations and that these patients should be treated with allogeneic hematopoietic stem cell transplantation in first remission. One of the issues that needs to be considered when going forward with this research is that the subjects in this study were of Japanese ethnicity. So far. there have been no reports on the relation of ethnicity to frequency of gene

References 1. Valk PJ, Verhaak RG, Beijen MA, et al. Prognostically useful gene-expression profiles in acute myeloid leukemia. N Engl J Med. 2004;350(16):1617-1628. 2. Bullinger L, Döhner K, Bair E, et al. Use of gene-expression profiling to identify prognostic subclasses in adult acute myeloid leukemia. N Engl J Med. 2004;350(16): 1605-1616. 3. Schlenk RF, Döhner K, Mack S, et al. Prospective evaluation of allogeneic hematopoietic stem-cell transplantation from matched related and matched unrelated donors in younger adults with high-risk acute myeloid leukemia: German-Austrian trial AMLHD98A. J Clin Oncol. 2010:28 (30):4642-4648. 4. Preudhomme C, Sagot C, Boissel N, et al. Favorable prognostic significance of CEBPA mutations in patients with de novo acute myeloid leukemia: a study from the Acute Leukemia French Association (ALFA). Blood. 2002;100(8):2717-2723. 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. Kottaridis PD, Gale RE, Frew ME, et al. The presence of a FLT3-internal tandem duplication in patients with acute myeloid leukemia (AML) adds important prognostic information to cytogenetic risk group and response to the first cycle of chemotherapy: analysis of 854 patients from the United Kingdom Medical Research Council AML 10 and 12 trials. Blood. 2001;98(6):17521759. 7. Schnittger S, Schoch C, Kern W, et al. Nucleophosmin gene mutations are predictors of favorable prognosis in acute myelogenous leukemia with a normal karyotype. Blood. 2005;106(12):3733-3739. 8. Döhner K, Schlenk RF, Habdank M, et al. Mutant nucleophosmin (NPM1) predicts favorable prognosis in younger adults with acute myeloid leukemia and normal cytogenetics: interaction with other gene mutations. Blood. 2005;106(12):3740-3746. 9. Schlenk RF, Döhner K, Krauter J, et al. Mutation and treatment outcome in cytogenetically normal acute myeloid leukemia. N Engl J Med. 2008;358(18):1909-1918. 10. 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.

haematologica | 2016; 101(9)

mutations and prognosis in AML. Whether the present research findings would be replicated in cohorts of patients of other ethnicities is an area that requires investigation in the future. Moreover, the present study did not include patients treated with azacitidine or other demethylating agents. The efficacy of demethylating agents in DMRGM-positive AML is therefore another area requiring investigation.

2011;117(8):2469-2475. 11. Brunet S, Labopin M, Esteve J, et al. Impact of FLT3 internal tandem duplication on the outcome of related and unrelated hematopoietic transplantation for adult acute myeloid leukemia in first remission: a retrospective analysis. J Clin Oncol. 2012;30(7):735-741. 12. Döhner 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. 13. O'Donnell MR, Abboud CN, Altman J, et al. Acute myeloid leukemia. J Natl Compr Canc Netw. 2012;10(8):984-1021. 14. The Cancer Genome Atlas Research Network. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368 (22):2059-2074. 15. Kihara R, Nagata Y, Kiyoi H, et al. Comprehensive analysis of genetic alterations and their prognostic impacts in adult acute myeloid leukemia patients. Leukemia. 2014;28(8):1586-1595. 16. Döhner H, Weisdorf DJ, Bloomfield CD. Acute myeloid leukemia. N Engl J Med. 2015;373(12):1136-1152. 17. Xie M, Lu C, Wang J, et al. Age-related mutations associated with clonal hematopoietic expansion and malignancies. Nat Med. 2014;20(12):1472-1478. 18. Ley TJ, Ding L, Walter MJ, et al, DNMT3A mutations in acute myeloid leukemia. N Engl J Med. 2010;363(25):2424-2433. 19. Thol F, Damm F, Lüdeking A, et al. Incidence and prognostic influence of DNMT3a mutations in acute myeloid leukemia. J Clin Oncol. 2011;29(21):28892896. 20. Chou WC, Chou SC, Liu CY, et al. TET2 mutation is an unfavorable prognostic factor in acute myeloid leukemia patients with intermediate-risk cytogenetics. Blood. 2011;118(14):3803-3810. 21. Figueroa ME1, Abdel-Wahab O, Lu C, et al. Leukemic IDH1 and IDH2 mutations result in a hypermethylation phenotype, disrupt TET2 function, and impair hematopoietic differentiation. Cancer Cell. 2010;18(6): 553-567. 22. Marcucci G1, Maharry K, Wu YZ, et al. IDH1 and IDH2 gene mutations identify novel molecular subsets within de novo cytogenetically normal acute myeloid leukemia: a Cancer and Leukemia Group B study. J Clin Oncol. 2010;28(14):2348-2355. 23. Boissel N, Nibourel O, Renneville A, et al. Prognostic impact of isocitrate dehydrogenase enzyme isoforms 1 and 2 mutations in acute myeloid leukemia: a study by the Acute Leukemia French Association group.

J Clin Oncol. 2010;28(23):3717-3723. 24. Patel JP, Gönen M, Figueroa ME, et al. Prognostic relevance of integrated genetic profiling in acute myeloid leukemia. N Engl J Med. 2012;366(12):1079-1089. 25. Hou HA, Kuo YY, Liu CY, et al. DNMT3A mutations in acute myeloid leukemia: stability during disease evolution and clinical implications. Blood. 2012;119(2):559-568. 26. Marcucci G, Metzeler KH, Schwind S, et al. Age-related prognostic impact of different types of DNMT3A mutations in adults with primary cytogenetically normal acute myeloid leukemia. J Clin Oncol. 2012;30 (7):742-750. 27. Aslanyan MG, Kroeze LI, Langemeijer SM, et al. Clinical and biological impact of TET2 mutations and expression in younger adult AML patients treated within the EORTC/GIMEMA AML-12 clinical trial. Ann Hematol. 2014;93(8):1401-1412. 28. Ribeiro AF, Pratcorona M, ErpelinckVerschueren C, et al. Mutant DNMT3A: a marker of poor prognosis in acute myeloid leukemia. Blood. 2012;119(24):5824-5831. 29. Gaidzik VI, Schlenk RF, Paschka P, et al. Clinical impact of DNMT3A mutations in younger adult patients with acute myeloid leukemia: results of the AML Study Group (AMLSG). Blood. 2013;121(23):4769-4777. 30. Gaidzik VI, Paschka P, Späth D, et al. TET2 mutations in acute myeloid leukemia (AML): results from a comprehensive genetic and clinical analysis of the AML study group. J Clin Oncol. 2012;30(12):1350-1357. 31. Wakita S, Yamaguchi H, Omori I, et al. Mutations of the epigenetics-modifying gene (DNMT3a, TET2, IDH1/2) at diagnosis may induce FLT3-ITD at relapse in de novo acute myeloid leukemia. Leukemia. 2013;27(5):1044-1052. 32. Papaemmanuil E, Gerstung M, Malcovati L, et al. Clinical and biological implications of driver mutations in myelodysplastic syndromes. Blood. 2013;122(22):3616-3627. 33. Guglielmelli P, Lasho TL, Rotunno G, et al. The number of prognostically detrimental mutations and prognosis in primary myelofibrosis: an international study of 797 patients. Leukemia. 2014;28(9):1804-1810. 34. Wakita S, Yamaguchi H, Ueki T, et al. Complex molecular genetic abnormalities involving three or more genetic mutations are important prognostic factors for acute myeloid leukemia. Leukemia. 2016:30(3): 545-554. 35. Genovese G, Kähler AK, Handsaker RE, et al. Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. N Engl J Med. 2014;371(26):2477-2487. 36. Shlush LI, Zandi S, Mitchell A, et al. Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia. Nature. 2014;506(7488):328-333.

1081

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Acute Lymphoblastic Leukemia

Ferrata Storti Foundation

Haematologica 2016 Volume 101(9):1082-1093

Characterization of leukemias with ETV6-ABL1 fusion

Marketa Zaliova,1 Anthony V. Moorman,2 Giovanni Cazzaniga,3 Martin Stanulla,4 Richard C. Harvey,5 Kathryn G. Roberts,6 Sue L. Heatley,7 Mignon L. Loh,8 Marina Konopleva,9 I-Ming Chen,5 Olga Zimmermannova,1 Claire Schwab,2 Owen Smith,10 Marie-Joelle Mozziconacci,11 Christian Chabannon,12 Myungshin Kim,13 J. H. Frederik Falkenburg,14 Alice Norton,15 Karen Marshall,16 Oskar A. Haas,17 Julia Starkova,1 Jan Stuchly,1 Stephen P. Hunger,18 Deborah White,7 Charles G. Mullighan,6 Cheryl L. Willman,5 Jan Stary,1 Jan Trka,1 and Jan Zuna1

CLIP, Department of Pediatric Hematology and Oncology, 2nd Faculty of Medicine, Charles University and University Hospital Motol, Prague, Czech Republic; 2Leukaemia Research Cytogenetics Group, Northern Institute for Cancer Research, Newcastle University, Newcastle upon Tyne, UK; 3Centro Ricerca Tettamanti, Clinica Pediatrica, Università di Milano-Bicocca, Fondazione MBBM/Ospedale San Gerardo, Monza, Italy; 4Pediatric Hematology and Oncology, Hannover Medical School, Germany; 5University of New Mexico Cancer Center, Albuquerque, NM, USA; 6Department of Pathology, St. Jude Children's Research Hospital, Memphis, TN, USA; 7South Australia Health and Medical Research Institute, Adelaide, Australia; 8Department of Pediatrics, Hematology-Oncology, Benioff Children’s Hospital, and the Helen Diller Family Comprehensive Cancer Center, University of California, San Francisco, CA, USA; 9Department of Leukemia, The University of Texas M.D. Anderson Cancer Center, Houston, TX, USA; 10Department of Haematology, Our Lady’s Children’s Hospital, Dublin, Ireland; 11Department of Cancer Biology, Institut Paoli Calmettes, Marseille, France; 12 Department of Hematology, Institut Paoli Calmettes, Marseille, France; 13Department of Laboratory Medicine, Seoul St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea; 14Department of Hematology, Leiden University Medical Center, The Netherlands; 15Birmingham Children’s Hospital, NHS Foundation Trust, UK; 16 Department of Cytogenetics, Leicester Royal Infirmary NHS Trust, UK; 17St. Anna Children's Hospital, Childrens Cancer Research Institute, Vienna, Austria; and 18Department of Pediatrics and the Center for Childhood Cancer Research, Children’s Hospital of Philadelphia and the University of Pennsylvania Perelman School of Medicine, PA, USA 1

ABSTRACT

Correspondence: marketa.zaliova@lfmotol.cuni.cz or jan.zuna@lfmotol.cuni.cz

Received: February 12, 2016 Accepted: May 18, 2016 Pre-published: May 26, 2016. doi:10.3324/haematol.2016.144345

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1082

1082

o characterize the incidence, clinical features and genetics of ETV6-ABL1 leukemias, representing targetable kinase-activating lesions, we analyzed 44 new and published cases of ETV6-ABL1-positive hematologic malignancies [22 cases of acute lymphoblastic leukemia (13 children, 9 adults) and 22 myeloid malignancies (18 myeloproliferative neoplasms, 4 acute myeloid leukemias)]. The presence of the ETV6-ABL1 fusion was ascertained by cytogenetics, fluorescence in-situ hybridization, reverse transcriptase-polymerase chain reaction and RNA sequencing. Genomic and gene expression profiling was performed by single nucleotide polymorphism and expression arrays. Systematic screening of more than 4,500 cases revealed that in acute lymphoblastic leukemia ETV6-ABL1 is rare in childhood (0.17% cases) and slightly more common in adults (0.38%). There is no systematic screening of myeloproliferative neoplasms; however, the number of ETV6-ABL1-positive cases and the relative incidence of acute lymphoblastic leukemia and myeloproliferative neoplasms suggest that in adulthood ETV6-ABL1 is more common in BCR-ABL1-negative chronic myeloid leukemia-like myeloproliferations than in acute lymphoblastic leukemia. The genomic profile of ETV6-ABL1 acute lymphoblastic leukemia resembled that of BCR-ABL1 and BCR-ABL1like cases with 80% of patients having concurrent CDKN2A/B and IKZF1 deletions. In the gene expression profiling all the ETV6-ABL1-positive samples clustered in close vicinity to BCR-ABL1 cases. All but one of the cases of ETV6ABL1 acute lymphoblastic leukemia were classified as BCR-ABL1-like by a standardized assay. Over 60% of patients died, irrespectively of the disease or age subgroup examined. In conclusion, ETV6-ABL1 fusion occurs in both lymphoid and myeloid leukemias; the genomic profile and clinical behavior resemble BCR-ABL1-positive malignancies, including the unfavorable prognosis, particularly of acute leukemias. The poor outcome suggests that treatment with tyrosine kinase inhibitors should be considered for patients with this fusion.

haematologica | 2016; 101(9)

ETV6-ABL1-positive leukemias

Introduction ETV6-ABL1 (TEL-ABL) fusion is a rare but recurrent genetic aberration found in hematologic malignancies. Given the orientation of ETV6 (12p13) and ABL1 (9q34) an in-frame fusion cannot be produced by a simple balanced translocation. In fact, the fusion results from a complex rearrangement involving a translocation and inversion or an insertion of ETV6 into chromosomal band 9q34 or ABL1 into 12p13. Alternative splicing generates two fusion transcripts - type A and B without and with ETV6 exon 5, respectively. Both result in constitutive chimeric tyrosine kinase activity analogous to BCR-ABL1 fusion.1 Likewise the effect of ETV6-ABL1 on cellular proliferation, cell survival and transforming capacity mirrors that seen in cases with BCR-ABL1.2,3 However, unlike BCR-ABL1, ETV6ABL1 does not cause acute leukemia in mice; instead it induces a chronic myeloproliferation similar to BCRABL1-induced chronic myeloid leukemia.4 In humans, the oncogenic potential of ETV6-ABL1 is similar to that of BCR-ABL1 and both fusions can be found in a spectrum of malignancies including myeloproliferative neoplasms (MPN), acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML). There is renewed interest in ETV6-ABL1 since the discovery of a “BCR-ABL1-like” (or “Ph-like”) gene expression profile (GEP) among B-cell precursor (BCP) ALL cases lacking an established chromosomal abnormality (so-called B-other ALL). The BCR-ABL1-like GEP resembles the BCR-ABL1 GEP because both are driven by the activation of kinase signaling.5-8 In vitro studies suggest that many of these kinase activating aberrations, including ETV6-ABL1 fusion, are sensitive to specific tyrosine kinase inhibitors (TKI).7,9,10 In addition, there are case reports of patients responding to TKI treatment indicating that these aberrations represent a promising and relevant therapeutic target especially given the reported unfavorable prognosis of BCR-ABL1-like ALL.57,11-14 However, the biology of this heterogeneous group of abnormalities is not fully understood and there is evidence that the prognosis of patients depends on the type of kinase activating lesion and the presence of cooperating aberrations such as IKZF1 deletions and JAK2 mutations.7 Further studies, based on patients with specific gene fusions, are therefore warranted. Here we present new data from 20 ETV6-ABL1 patients and thoroughly scrutinized existing data from other patients with this fusion to evaluate different detection methodologies, estimate frequency, describe clinical features and outcome, and characterize the profile of copy number aberrations and gene expression of this distinct molecular subtype.

Methods Patients The cohort consisted of 44 patients with an ETV6-ABL1 fusion and comprised newly identified cases (n=9), published cases with additional new data (n=11)7,8,15-20 and cases with re-examined published data (n=24).1,18,21-39 Standard diagnostics, including molecular genetics, karyotyping and fluorescence in-situ hybridization (FISH) were performed according to the standard practice of the local diagnostic laboratories. Basic clinical/outcome data were haematologica | 2016; 101(9)

collected from treating centers. Detailed diagnostic procedures have been published for existing cases. For the newly described patients, diagnostic and treatment procedures and protocols were approved by local Institutional Review Boards. Informed consent was obtained in accordance with the Declaration of Helsinki. Primary samples from the first and second MPN in lymphoid blast crisis of case 34-a-MPN used in this study were provided by the Institute Paoli Calmettes Tumor Bank (Marseille, France). The primary childhood leukemia sample of case 09-ch-ALL was provided by the Leukaemia and Lymphoma Research Childhood Leukaemia Cell Bank working with teams in the Bristol Genetics Laboratory (Southmead Hospital, Bristol), Molecular Biology Laboratory (Royal Hospital for Sick Children, Glasgow), Molecular Haematology Laboratory (Royal London Hospital, London), and Molecular Genetics Service and Sheffield Children’s Hospital, Sheffield) in the UK.

Detection of transcript variants The expression of alternative splice variants (type A/B) was detected by end-point reverse transcriptase polymerase chain reaction (RT-PCR)20 and quantitative RT-PCR (qRT-PCR) analyses, utilizing forward primers annealing to ETV6 exon 5 (type B-specific: 5’- GCCCATCAACCTCTCTCATCG -3’) or ETV6 exon 4/ABL1 exon 2 junction (type A-specific: 5’- CAGAACCATGAAGAAGAAGCCC -3’). For the schematic representation of transcript variants and position of primers see Online Supplementary Figure S1.

Single nucleotide polymorphism array Copy number alterations (CNA) were determined by single nucleotide polymorphism array (HumanOmni Express BeadChip, Illumina, San Diego, USA). DNA from bone marrow or peripheral blood cells was extracted using standard methods. DNA was labeled and hybridized according to the recommended Infinium HD assay Ultra protocol from Illumina; scanning was performed using an Illumina iScan System. GenomeStudio Software v2011.1 (Illumina) was used for genotype calling and quality control. DNA CNA and regions of uniparental disomy (UPD) were analyzed using the CNV Partition 2.4.4 algorithm plug-in within the GenomeStudio followed by visual inspection in the Illumina Chromosome Browser (display option of the GenomeStudio) and manual correction. Identified CNA/UPD were mapped against human genome assembly GRCh37/hg19. Deletions corresponding to somatic rearrangements of the immunoglobulin and T-cell receptor gene loci and CNA/UPD seen in remission samples (or those that overlapped with copy number polymorphisms listed in the Database of Genomic Variants) were excluded from further analysis. To identify recurrently affected regions, the genome was segmented into the minimal segments using genomic coordinates for the start and end of all detected CNA/UPD from all analyzed samples. Segments affected by the same type of aberration (loss or gain or UPD) in more than two cases were called recurrently affected regions (Online Supplementary Figure S2). For this analysis continuous CNA/UPD in a particular sample could be segmented into several neighboring regions when they overlapped in the different sets of samples. 1083

M. Zaliova et al.

Multiplex ligation-dependent probe amplification Multiplex ligation-dependent probe amplification assays (P335-A3 ALL-IKZF1, MRC-Holland, The Netherlands) were performed according to the manufacturer’s instruction. Fragments were analyzed on an Applied Biosystems 3730 DNA Analyzer (Applied Biosystems, Carlsbad, CA, USA) and data analyzed using Peak Scanner v1.0 and Coffalyser v9.4 software. Relative copy number was calculated after intra-sample normalization against control fragments and inter-sample normalization against control samples (healthy blood donor DNA). Probes with a ratio of 1±0.3 were assigned a normal copy number of 2, probes with a ratio of less than 0.7 and less than 0.3 were considered heterozygously and homozygously deleted, respectively, while probes with a ratio of more than 1.3 were considered amplified.

Whole genome gene expression profiling and the BCR-ABL1-like signature RNA was isolated from pretreatment diagnostic ALL samples and analyzed on Affymetrix Human Genome GeneChip® U133_Plus_2.0 arrays. CEL files from the previously published 1181 cases7 (including 2 ETV6-ABL1positive patients reported in this study: 06-ch-ALL and 14a-ALL) were normalized together with another ten ETV6ABL1-positive samples described here. Besides the 12 ETV6-ABL1-positive samples the cohort for gene expression profiling consisted of 664 children with National Cancer Institute–classified high-risk BCP-ALL, 348 adoles-

cents (16 to 20 years old), and 167 young adults (21 to 39 years old). The cohort included 84 BCR-ABL1-positive ALL and 209 BCR-ABL1-like ALL. Expression data were generated using the default RMA conditions of the Expression Console 1.4.1 software (Affymetrix). The clustering was performed using MATLAB R2015a (MathWorks) with Euclidean distance and weighted linkage using 257 probe sets predicting BCR-ABL1-like status according to predictive analysis of microarrays.7 Maximum ranges for coloring high (red) and low (green) expression were set to 10-fold the median. Information for BCR-ABL1-like and BCR-ABL1 status was obtained from previously published data. The BCR-ABL1-like cluster was deduced from the expression patterns of its signature genes.7 The BCR-ABL1-like signature was determined using customized Taqman low density arrays at the South Australian Health & Medical Research Institute.19

Results Disease subtype, frequency, demographics and clinical features Of the 44 ETV6-ABL1-positive patients considered in this study, half were diagnosed with ALL [n=22; including one manifesting as B-lymphoblastic lymphoma (13-chLBL)], while the other half had chronic myeloid leukemia-

Figure 1. Malignancy subtypes, age and gender distribution of the ETV6-ABL1positive cases. In childhood, all cases were diagnosed as ALL (albeit one originally manifested as B-lymphoblastic lymphoma). Nine childhood ALL were of BCP and two of T-cell origin. Notably, there were two infants in this group; however, most of the childhood ALL were diagnosed in the preschool/early school age (5-10 years). ALL is dominant until the fourth decade and then it is replaced by MPN, which account for more than 70% of cases in the age categories over 35 years. Overall, the male to female ratio is 1.9 : 1 (28:15). While gender distribution is almost even in ALL (M:F 12:10), patients with MPN were predominantly male (M:F 12:5) and all four AML cases were diagnosed in males. *The gender of one case of MPN was not reported.

1084

haematologica | 2016; 101(9)

ETV6-ABL1-positive leukemias

like MPN (n=18) or AML (n=4) (Figure 1; Table 1). A high diagnostic white blood cell count (>50x109/L) was observed in 10/12 children and 3/7 adults with ALL. Interestingly, MPN cases presented with lower white blood cell counts (median 36x109/L with only 1/16 patients having a count >100x109/L) than those reported for patients with chronic myeloid leukemia.40 Eosinophilia, suggested as a hallmark of ETV6-ABL1 leukemia,28 was present in all MPN and AML cases but only 4/13 ALL cases had an elevated absolute eosinophil count (>0.5x109/L). Routine screening of newly diagnosed childhood ALL patients for ETV6-ABL1 was started in 2003 and 2007 in the Czech Republic and Italy, respectively. By the end of 2014, 730 and 2795 cases had been screened by PCR. This prospective screening coupled with two smaller studies41,42 allows us to estimate the frequency of ETV6-ABL1 in childhood ALL at 0.17% (6/3610) which corresponds to ~1-2% of BCR-ABL1-like cases. Meta-analysis of the data from four studies of adult ALL cases showed a frequency of 0.38% (4/1060).18,20,39,42 One screening study in AML found a single case (1/1197) which coupled with the fact that only four ETV6-ABL1-positive AML cases have been published so far suggests that ETV6-ABL1 is very rare in AML.18 Estimating the incidence of ETV6-ABL1 in MPN is difficult because no systematic screening data are available.

Identification and location of the ETV6-ABL1 fusion Among the 22 cases of ETV6-ABL1-positive ALL, the fusion was originally identified by PCR screening (n=10), cytogenetics/FISH (n=6), whole transcriptome sequencing (RNAseq, n=3) or another approach (n=3). Among MPN/AML cases the majority of the fusions were originally detected by cytogenetics/FISH (16/21) and only a minority by PCR (n=2) or other techniques (n=3) (Table 1). In 14 cases, we re-analyzed samples to determine ETV6ABL1 splice variants. We detected both types (A and B, with type B expressed at significantly higher levels) in all but one case (patient 18-a-ALL in whom only type B was detected, possibly due to poor quality RNA) (data not shown). A total of 28 cases were tested using commercial ETV6 and/or ABL1 FISH probes. Abnormal signal patterns indicative of ETV6 or ABL1 rearrangement were observed in only 19/28 (67%) cases tested with BCR-ABL1 probes and 10/19 (53%) with ETV6 probes. A further eight ETV6 or ABL1 aberrant signals were detected using specific BAC/PAC/YAC/cosmid FISH probes (7 ETV6, 1 ABL1) (Table 1 and Online Supplementary Table S1). In 30 patients the cytogenetic/FISH result enabled localization of ETV6ABL1 to a particular chromosome (Online Supplementary Figure S3). In most cases (18/30) the fusion was present on chromosome 12 (10/10 BCP-ALL, 7/16 MPN, 1/3 AML) whereas it was located on chromosome 9 in nine cases (7/16 MPN, 1/1 T-ALL, 1/3 AML). In three cases with more complex rearrangements involving a third chromosome the fusion signal was found on chromosome 8 (AML), 16 (MPN) and 17 (MPN).

Copy number aberration profile of ETV6-ABL1-positive leukemia We assessed the CNA profile of 22 samples from 18 ETV6-ABL1 patients by high-density single nucleotide polymorphism array further refined by multiplex ligahaematologica | 2016; 101(9)

tion-dependent probe amplification analysis in seven cases. The cohort comprised 15 patients with ALL and three with MPN (Figure 2; Table 1). Three MPN samples were taken at lymphoid blast crisis (LBC) and two in chronic phase (CP). Among ALL and MPN-LBC samples, we found an average of 12 regions of CNA or UPD per case (range, 2 to 29 alterations). Deletions were more common than gains (206:19). No CNA or UPD abnormalities were detected in two MPN-CP samples (Online Supplementary Tables S2 and S3). Genome segmentation based on the overlap of all CNA identified 74 genomic regions which were affected in at least two patients. The majority of these regions are neighboring loci on chromosomes 7 and 9 (Online Supplementary Figure S1). The most prevalent aberrations detected by single nucleotide polymorphism array in ALL/MPN-LBC were deletions of CDKN2A/CDKN2B (16/17; 94%, although in one MPN case the deletion was apparent only at the second LBC) and deletions of IKZF1 (15/17; 88%), followed by PAX5 (8/17; 47%) and BTG1 (7/17; 41%) deletions (although again in the same MPN case the PAX5 deletion was detected only at the second LBC). In eight patients an aberration of the ABL1 gene was detected (gain of entire ABL1 in 4 cases, partial ABL1 gain in 3 and partial deletion in 1 case) and three patients had partial deletions of ETV6 exons 6-8 (associated with ABL1 gain in all 3 cases). Taking into account the affected exons, these partial deletions/gains seem to result from the primary unbalanced genomic rearrangement creating the fusion gene rather than being secondary aberrations. Other recurrent aberrations occurring in two to five patients included deletions of SLX4IP, ATP10A, BTLA, CD200 and RB1 genes. Overall, the frequency of the aberrations is similar to that in BCR-ABL1-positive and BCR-ABL1-like (Ph-like) ALL. Notably, while the two MPN-CP samples had no detectable CNA, the MPN cases analyzed in LBC had the same CNA pattern as ALL cases (Figure 2).

Gene expression profile and BCR/ABL1-like status Expression profiling was performed on 12 samples from ten patients: diagnostic samples from seven ALL cases along with relapse/LBC samples from two cases of BCPALL and one MPN case. We compared the GEP of these ETV6-ABL1 samples with those of a previously published cohort of 1179 BCP-ALL cases tested on the same platform at the University of New Mexico.7 Overall, with the 257 probe sets predicting BCR-ABL1-like status according to predictive analysis of microarrays,7 the GEP of all the ETV6-ABL1 cases was consistent with the signature produced from BCR-ABL1 cases; furthermore, in an unsupervised analysis, all ETV6-ABL1 samples clustered within the BCR-ABL1-like cluster and in close vicinity to BCRABL1 cases (unlike other ABL1 and ABL2 fusion genes) (Figure 3). Moreover, we analyzed ETV6-ABL1 cases by custom Taqman low-density array19 to determine whether they would be classified as BCR-ABL1-like also when using a method which, unlike GEP, is applicable in routine diagnostics. Eleven of 12 cases (9 BCP-ALL, 1 T-ALL and 1 MPN-LBC) tested positive for the BCR-ABL1-like signature indicating that these cases do indeed share considerable similarity with other BCR-ABL1-like cases. The single negative sample came from a child with BCP-ALL (11-chALL) (Table 1). 1085

M. Zaliova et al. Table 1. Methods and results of the performed diagnostic/research genetic tests.

Case ID

Cancer subtype

ETV6/ABL1 identification

Type of transcript

01-ch-ALL 02-ch-ALL

BCP-ALL BCP-ALL

PCR PCR

A, B * A, B *

03-ch-ALL 04-ch-ALL

BCP-ALL T-ALL

other cytogenetics

A A, B

05-ch-ALL 06-ch-ALL 07-ch-ALL 08-ch-ALL 09-ch-ALL 10-ch-ALL

BCP-ALL BCP-ALL BCP-ALL BCP-ALL BCP-ALL BCP-ALL

A, B * B n.r. A, B * A, B * A, B *

11-ch-ALL 12-ch-ALL 13-ch-LBL 14-a-ALL 15-a-ALL

BCP-ALL T-ALL B-LBL BCP-ALL BCP-ALL

PCR RNAseq RNAseq other cytogenetics SNP array/ PCR ° cytogenetics PCR PCR RNAseq PCR

16-a-ALL

BCP-ALL

PCR

A, B

17-a-ALL

BCP-ALL

cytogenetics

A, B *

18-a-ALL

BCP-ALL

cytogenetics

19-a-ALL 20-a-ALL 21-a-ALL 22-a-ALL

BCP-ALL BCP-ALL BCP-ALL BCP-ALL

PCR PCR PCR cytogenetics

A, B * A A B

23-a-AML 24-a-AML 25-a-AML 26-a-AML

AML-M2 AML-M1 AML AML-M6

cytogenetics cytogenetics PCR other

A B B B

27-a-MPN 28-a-MPN 29-a-MPN 30-a-MPN 31-a-MPN 32-a-MPN 33-a-MPN 34-a-MPN

MPN MPN MPN MPN MPN MPN MPN MPN

SNP cytogenetics cytogenetics cytogenetics cytogenetics n.r. other cytogenetics

A, B B n.r. n.r. A, B B A, B A, B *

35-a-MPN 36-a-MPN 37-a-MPN

MPN MPN MPN

cytogenetics cytogenetics PCR

A, B n.r. A, B

38-a-MPN 39-a-MPN 40-a-MPN 41-a-MPN

MPN MPN MPN-B-LBC MPN-EMT-LBC

cytogenetics cytogenetics cytogenetics cytogenetics

A, B * n.r. n.r. A, B

A, B * A, B * A, B * n.r. A, B

Simplified cytogenetics

ABL1 FISH

46,XX n.r. 46,XY,inv(1),t(1;9),del(1), normal t(9;12),t(1;9),t(1;9;12) n.r. n.r. 47,XXYc,del(6), normal ins(9;12),inv(9) 46,XY abnormal 46,XY,ins(12;9) n.r. 46,XX,t(9;12) n.r. no karyotype n.r. 46,XX.ish ins(12;9) abnormal no karyotype normal

ETV6 FISH

Fusion Ph-like ALL SNP array location (Chr) signature * *¥

n.r. normal

n.r. 12

pos(TLDA) pos(TLDA)

C,P C,I

n.r. abnormal %

n.r. 9

n.a. n.a.

normal n.r. n.r. n.r. normal n.r.

12 12 n.r. n.r. 12 n.r.

pos(TLDA) † n.a. pos (LDA) & pos(TLDA) pos(TLDA) pos(TLDA)

C,I,B C,I,P n.a. C,I,B C,I,B C,I,P,B

12 n.r. n.r. n.r. 12

neg(TLDA) pos(TLDA) pos(TLDA) † n.a. n.a.

I C C,I,B C,I,P,B C,I

n.a.

pos(TLDA)£

C,I,P

n.a.

12 n.r. n.r. n.r.

n.a. n.a. n.a. pos(TLDA)

C,I n.a. n.a. C,I,P

8 9 12 n.r.

n.a. n.a. n.a. n.a.

9 9 n.r. 9 9 12 n.r. 12

n.a. n.a. n.a. n.a. n.a. n.a. n.a. pos(TLDA)$

n.a. n.a. n.a. n.a. n.a. n.a. n.a. C2,I1,P2,B1

9 9 12

n.a. n.a. n.a.

16 12 12 9

n.a. n.a. n.a. n.a.

no n.a. C,I,P n.a.

46,XY,?ins(12;9) abnormal normal 46,XY n.r. n.r. 46,XY n.r. n.r. 46,XY n.r. n.r. 46,XX,del(9), abnormal normal t(9;10) 47,XX,+5. ish abnormal abnormal t(9;12),inv(12) 45,XY,t(1;9), abnormal % abnormal % t(9;13),-13,t(16;22) 46,XY,t(2;9), abnormal normal t(2;8),ins(12;9) 46,XX,t(8;9;12) abnormal normal 46,XX,t(9;12) n.r. n.r. 46,XY,t(9;12) n.r. n.r. complex rearrangements; n.r. n.r. t(9;12) 46,XY,t(8;12) abnormal abnormal % 46,XY,t(9;12) normal abnormal % 46,XY abnormal normal 45,t(9;12;14), n.r. n.r. complex rearrangements 46,XX normal abnormal 46,XY,t(12;14) n.r. n.r. 46,XY,t(9;12) abnormal abnormal 46,XX,t(9;12) normal abnormal (m) 46,XX,t(9;12) abnormal (i) abnormal n.r. n.r. n.r. 46,XY normal n.r. 44,X,-Y,-7,?add(9), abnormal normal del(9),add(12) 46,XY,del(6),t(9;12) normal abnormal % 46,XX,ins(9;12) abnormal n.r. 51,XX,t(5;9),add(17),del(5), abnormal abnormal § (+8,+11,+12,+18,+19) 46,XY,t(5;12;16) abnormal abnormal % 46,XY,ins(12;9) abnormal abnormal (i) 45,XY.isht(9;9),t(9;12) n.r. n.r. 47,XY,t(9;12),del(12), normal abnormal (m) t(7;14),+19

continued on the next page

1086

haematologica | 2016; 101(9)

ETV6-ABL1-positive leukemias continued from the previous page

42-a-MPN 43-a-MPN

MPN-MBC MPN-MBC

cytogenetics cytogenetics

B A, B

44-a-MPN

MPN-B-LBC

cytogenetics

45,XY,-7,t(9;12) abnormal 49,XY,+11,t(9;12), abnormal +der(12)t(9;12),+19,t(1;22) 46,XY,t(12;17) abnormal

abnormal % abnormal §

12 12

n.a. n.a.

abnormal (m)

n.a.

Figure 2. Overview and frequency of selected recurrent copy number aberrations in 22 samples (17 ALL, 3 MPN-LBC, 2 MPN-CP) from 18 ETV6-ABL1-positive cases. The most prevalent aberrations detected by the SNP array in ALL/MPN-LBC were deletions of CDKN2A/CDKN2B and deletions of IKZF1 followed by PAX5 (8/17; 47%) and BTG1 (7/17; 41%) deletions. CDKN2A and CDKN2B as well as BTLA and CD200 genes were co-deleted in all cases, thus the pairs are merged into one column. Deletion of ATP10A was recently shown to be a recurrent aberration in BCP-ALL, associated with decreased 10-year event-free survival56. Gray indicates loss, black indicates a gain, deleted exons within IKZF1, ABL1 and ETV6 genes are indicated by numerals (for IKZF1 white numerals depict losses resulting in a dominant negative IKZF1 isoform or complete loss of IKZF1 expression, black numerals depict haploinsufficiency with the potential exception of combined 2-3/4-7 loss which can affect either a single allele or both gene alleles and thus can result in either haploinsufficiency or complete loss). For frequencies of the selected recurrent copy number aberrations in published ALL studies see Online Supplementary Figure S4. a: adult; ch: childhood; LBC: lymphoid blast crisis of MPN; CP: chronic phase of MPN. $In these cases some of the findings from the SNP array were further refined by multiplex ligation-dependent probe amplification. #ABL1 deletions were not found recurrently in the present study, thus, only ABL1 gains were considered for CNA frequency. * The relatively high frequency of ETV6 deletions in ETV6-ABL1-positive cases compared to that in BCR-ABL1-positive ALL probably does not reflect frequency of genuine secondary ETV6 loss but rather results from ETV6 loss during primary genomic rearrangement.

Clinical characteristics, treatment and outcome Response to treatment is the most important predictor of outcome in ALL.43-45 Treatment response is determined differently per protocol but is usually measured after 1 or 2 weeks of treatment with corticosteroids (“prednisone haematologica | 2016; 101(9)

response”) and again at the end of induction therapy. In this series, both children with a poor prednisone response died whereas three of four patients with a good prednisone response remain alive after >3 years (Table 2). Residual disease was assessed at the end of induction in 1087

M. Zaliova et al.

Figure 3. Unsupervised hierarchical clustering of ETV6-ABL1 cases with a large cohort of BCP-ALL. Twelve ETV6-ABL1 cases clustered with a cohort of 1179 BCPALL cases using data from RMA normalized U133_Plus_2.0 arrays. The 1191 samples are shown in columns, while the 257 BCR-ABL1-like predicting probe sets (described previously7) are in rows. The yellow box highlights the BCR-ABL1-like cluster. BCR-ABL1, ETV6-ABL1 and other ABL1 and ABL2 fusion (NUP214-ABL1, SNX2-ABL1, ZMIZ1-ABL1, RSCD1-ABL1, RSCD1-ABL2, PAG1-ABL2, ZC3HAV1-ABL2) status is shown with black, red and magenta bars, respectively, across the top of the heatmap. All other Ph-like cases are shown with blue bars using the status defined in their original report.7 The order of samples in the Ph-like cluster is enlarged on the top (with the same color coding). The order of the ETV6-ABL1 samples as shown in the figure, from left to right is 34-a-MPN (1st LBC)/08-ch-ALL/10ch-ALL/09-ch-ALL/06-ch-ALL/14-a-ALL/11-ch-ALL/12-ch-ALL/13-ch-LBL (1st relapse)/05-ch-ALL (2nd relapse)/05-ch-ALL (1st relapse)/34-a-MPN (2nd LBC).

ten children; four cases had no or low levels of minimal residual disease (<10-4) whereas six cases had higher levels of residual disease (>10-4). Of the four patients with lowrisk minimal residual disease, three are alive >4 years after diagnosis whereas one patient died after a bone marrow relapse. One infant case with a borderline level of minimal residual disease (3x10-4) is alive after a central nervous system relapse. Out of the five patients with minimal residual disease levels â&#x2030;Ľ10-3 at the end of induction therapy, three relapsed and died while the other two are in first remission, but only 19 and 28 months after diagnosis (Table 2). Of the nine adults with ALL, seven died, one relapsed with no further outcome data and one was in short-term remission after stem cell transplantation. The survival rate of the ALL patients is 46% in children (6/13 alive) and 13% in adults (1/8 alive). Three of the four AML patients failed to achieve a complete remission and hence died within a few months. The overall survival of ETV6-ABL1 MPN cases is 50% (9/18 alive) but the risk of death is more than double for patients diagnosed in blast crisis. Importantly, of the 23 patients with ETV6-ABL1-positive malignancies in whom the cause of death could be evaluated, 19 (>80%) died of disease progression or relapse. Although TKI were not used uniformly to treat this 1088

cohort of patients, 17 patients were given a TKI at some point during their treatment. The two children with ALL who received imatinib during their frontline treatment remain in first remission 46 and 57 months after initial diagnosis even though one experienced an isolated central nervous system relapse after 20 months. Two children were treated with TKI after relapse but both died shortly afterwards. In adult ALL, one patient (aged 81) was treated with TKI within frontline treatment and died after relapse (Table 2). A total of five MPN patients diagnosed in CP received imatinib during frontline therapy and all are still alive. However, two cases did switch to nilotinib because of disease progression. In comparison only three out of eight MPN patients diagnosed in CP who did not receive a TKI are still alive. Two patients received a TKI at disease progression only but both subsequently died. Five of the MPN patients were diagnosed in blast crisis (Table 2). Four of these patients received a TKI during frontline treatment and the fifth patient only after relapse. Four of these MPN patients with blast crisis died, three within 1 year. One patient who received a TKI during frontline therapy remains alive after 12 months but has cytogenetically detectable residual leukemia. haematologica | 2016; 101(9)

haematologica | 2016; 101(9)

BCP-ALL

T-ALL

B-LBL

BCP-ALL

AML-M2

AML-M1

07-ch-ALL

09-ch-ALL

10-ch-ALL

11-ch-ALL

12-ch-ALL

13-ch-LBL

14-a-ALL

15-a-ALL

16-a-ALL

17-a-ALL

18-a-ALL

19-a-ALL

20-a-ALL

21-a-ALL

22-a-ALL

23-a-AML

24-a-AML

BCP-ALL

06-ch-ALL

08-ch-ALL

T-ALL

BCP-ALL

03-ch-ALL

BCP-ALL

02-ch-ALL

05-ch-ALL

BCP-ALL

Case ID

04-ch-ALL

Cancer subtypeÂ°

01-ch-ALL

Age at Dx (years)

Sex

WBC (x109/L)

183

n.r.

129

227

167

284

282

322

184

204

565

158

Eosinophilia

Treatment CHT (MR)

n.r.

Interfant 06 (LR)

Interfant 06 + TKI

TKI

UKALL 2003 (sch B->C)+TKI

AIEOP-BFM ALL 2000 (SR)

AALL1131 (B, VHR)

AALL0232

UKALL 2011 (sch A->C) no

yes

n.r.

Ara-C/IDA/ETO

GIMEMA LAM99

IND (D/V/M)/ PRED+TKI

IND (V/P/D/L)

n.r.

UKALL12

yes

n.r.

MUD

SCT

DASA

n.r.

HRD

yes

n.r.

aSCT

no PBSCT

V/P/D/L+MTX i.th.+Ara-C no

IND + CONS CHT

modif CVAD+ CONS/i.th. CHT

hyper-CVAD + rituximab no

AIEOP LNH 97

n.r. AIEOP-BFM ALL 2009 (HR) no

yes* AIEOP-BFM ALL 2000 (SR) no

n.r.

yes AIEOP-BFM ALL 2000 (HR) no

yes

n.r.

yes*

no 1x10-3

3x10-4

MRD EOI neg

6x10-2

1x10-3

9x10-5

neg 2x10-2

1x10-3

n.r./n.a.

1x10-1

n.r./n.a. n.r./n.a.

n.r./n.a.

n.r./n.a. n.r./n.a.

PGR

n.r./n.a. n.r./n.a.

PPR

n.r./n.a.

PGR

n.r./n.a. <5x10-5

PGR

n.r./n.a.

PPR

n.r./n.a. n.r./n.a.

PGR

Prednisone response

Complete remission no

yes

n.r.

yes

Recurrence/ progression n.a.

yes

n.a.

n.r.

yes

n.a.

yes

Type (time)

n.r.

n.a. n.r.

n.a.

ALL BFM IM, DASA REZ2002 (S2) + TKI

AIEOP ALL REC2003 (S2)

n.a.

BM+CNS (9M)

n.a.

n.r.

BM+CNS (6M)

BM+CNS (4M)

n.r.

BM (14M)

CNS (13M)

EM (43M)

BM+MED (9M)

n.a.

CNS (12M)

n.a.

HD-DEXA/ VCR + TKI

n.a.

n.r.

FLAG

V/P/D

n.r.

DepoCyt

AIEOP ALL REC2003 (A)

paliative

n.a.

n.r.

n.a.

PONA

n.a.

n.r.

n.a.

BM (25M) ALL BFM REZ2002 (S3) no

n.r. (15M)

n.a.

CNS (8M)

CNS (20M)

First recurrence/progression Treatment

First manifestation

TKI

Table 2. Demographics, clinical characteristics and outcome of ETV6/ABL1-positive cases.

SCT n.a.

n.a.

n.r.

aSCT

n.a.

UCB

MUD#

Additional recurrence(s)/ progression(s) n.a.

n.a.

yes

n.a.

n.r.

yes

yes&

n.a.

yes$

n.a.

yes

Outcome

7,19

p.s.

7,8

p.s

Reference(s) continued on the next page

died (6M)

alive (>20M)

died (11M)

died (3M)

alive (>24M)

died (18M)

died (13M)

died (11M)

n.r.

died (15M)

died (20M)

died (73M)

died (11M)

alive (CR>19M)

alive (CR>60M)

alive (CR>57M)

alive (CR>72M)

alive (CR>28M)

died (18M)

died (32M)

died (22M)

died (3M)

died (15M)

alive (CR>46M)

ETV6-ABL1-positive leukemias

1089

MPN

MPN-B-LBC

37-a-MPN

38-a-MPN

39-a-MPN

40-a-MPN

WBC (x109/L)

238

n.r.

Eosinophilia

n.r.

yes

n.r.

yes

n.r.

yes

TKI

Treatment TKI

HU + TKI

LD-Ara-C/ETO/ MTX+TKI

hyper-CVAD+TKI

BFM-93

n.r.

TKI

LD-Ara-C

TKI

IFNa

HU/IFNa

n.r.

TKI

HU/IFNa

HU + TKI

n.r.

TKI

n.r.

DASA

NILO

IND (MTX/ETO/Ara-C) no

yes

n.r.

yes

n.r.

SCT n.r./n.a. n.r./n.a.

MRD EOI

n.r./n.a. n.r./n.a.

Complete remission yes

yes

n.r.

yes

n.r.

yes

n.a.

Recurrence/ progression no

yes

n.a.

yes

n.a.

Type (time) n.a.

MBC

MBC (9M)

n.a.

MPN (37M)

MBC (12M)

n.a.

MPN (18M)

MPN (12M)

B-LBC (19Y)

n.a.

MPN (6M)

n.a.

First recurrence/progression

n.a.

FLAG

n.r.

n.a.

TKI

n.a.

TKI

GC/TG

EWALL + TKI

n.a.

TKI

n.a.

n.r.

n.a.

DASA, NILO

n.a.

NILO

n.a.

SCT n.a.

n.a.

Additional recurrence(s)/ progression(s) no

yes

n.a.

yes&

n.a.

died (4M)

Outcome alive (CR>12M)

died (6M)

died (8M)

died (11M)

died (64M)

died (13M)

alive (>41M)

died (12M)

alive (CR>36M)

died (13M)

died (240M)

alive (CR>69M)

died (3M)

alive (CR>22M)

alive (CR>6M)

alive (CR>60M)

alive (CR>36M)

alive (CR>7M)

died (3M)

p.s.

Reference(s)

BCP-ALL: B-cell precursor acute lymphoblastic leukemia; T-ALL: T-lineage acute lymphoblastic leukemia; AML: acute myeloid leukemia; MPN: myeloproliferative neoplasm; EM: extramedullary; B-LBL: B-lymphoblastic lymphoma; B-LBC: B-lymphoid blast crisis; T-LBC: T-lymphoid blast crisis; MBC: myeloid blast crisis; a: adult; ch: childhood; Dx: diagnosis; n.r.: not reported; n.a.: not applicable; m: male; f: female; WBC: white blood cell count; TKI: tyrosine-kinase inhibitor; LR: low risk; SR: standard risk; MR: medium risk; HR: high risk; VHR: very high risk; sch: schedule; CHT: chemotherapy, IND: induction; CONS: consolidation; i.th.: intrathecal; VCR: vincristine; PRED: prednisone; HD-DEXA: high-dose dexamethasone; MTX: methotrexate; Ara-C: cytosine arabinoside; IDA: idarubicin; ETO: etoposide; HU: hydroxyurea; IFNa: interferone alfa; LD-Ara-C: low dose Ara-C; CG: glucocorticoids; TG: thioguanine; D/V/M: high-dose dexamethasone + pulses of vincristine + intrathecal methotrexate; V/P/D/L: vincristine + prednisone + daunorubicin + L-asparaginase; FLAG: fludarabine + high-dose cytarabine + granulocyte colony-stimulating factor; hyper-CVAD: cyclophosphamide + vincristine + doxorubicin + dexamethasone + methotrexate + cytarabine; modif CVAD: cytarabine + vincristine + doxorubicin + daunorubicin + dexamethasone; IM: imatinib mesylate; DASA, dasatinib; NILO: nilotinib; PONA: ponatinib; SCT: stem cell transplantation; MUD: matched unrelated donor; PBSCT: allogeneic peripheral blood SCT; HRD: haplo-identical related donor; UCB: umbilical cord blood; aSCT: autologous SCT; PPR: prednisone poor response; PGR: prednisone good response; MRD-EOI: minimal residual disease at the end of induction; CNS: central nervous system; BM: bone marrow; MED: mediastinum; M: month; p.s.: present study. Â° The 1st disease manifestation *Only relative eosinophilia due to high WBC (1% eosinophils in differential cell count). #From the same donor as in the 1st SCT. $TKI used in the treatment of the 2nd relapse. &After the 1st recurrence and 2nd complete remission cases 13-ch-LBL and 34-a-MPN experienced 2nd and 3rd recurrences of BCP-ALL character.

MPN-B-LBC

MPN

36-a-MPN

44-a-MPN

MPN

35-a-MPN

n.r.

MPN-MBC

MPN

34-a-MPN

43-a-MPN

MPN

33-a-MPN

MPN

32-a-MPN

MPN-MBC

MPN

31-a-MPN

42-a-MPN

MPN

30-a-MPN

MPN

29-a-MPN

41-a-MPN MPN-EM-T-LBC 31

MPN

28-a-MPN

MPN

27-a-MPN

Age at Dx (years)

Sex

AML

AML-M6

Case ID

26-a-AML

Cancer subtypeÂ°

25-a-AML

Prednisone response

First manifestation Treatment

1090

TKI

continued from the previous page

M. Zaliova et al.

haematologica | 2016; 101(9)

ETV6-ABL1-positive leukemias

Discussion The ETV6-ABL1 fusion is rare but can occur in a range of hematologic malignancies; principally BCP-ALL (sporadically manifesting as lymphoma) and MPN but also TALL and AML. In this respect, ETV6-ABL1-positive malignancies are reminiscent of BCR-ABL1-positive neoplasms.46 An additional similarity lies in the non-random age distribution with acute leukemias dominating in children and young adults and MPN in older adults.46 However, the incidence of ETV6-ABL1 leukemia is considerably lower than that of BCR-ABL1 leukemia, accounting for <1% of cases of ALL at any age. An accurate estimate of the frequency of ETV6-ABL1 in MPN remains elusive as there has not been any systematic screening. However, all cases of BCR-ABL1-negative chronic myeloid leukemialike MPN should be screened for this fusion as it might account for a significant proportion of this relatively rare condition. Given that at least three chromosomal breaks are required to generate an in-frame fusion, the genomic rearrangement is not uniform across cases and typically involves cryptic insertions. Our study suggests that insertions of ABL1 into the ETV6 locus are favored in BCP-ALL (10/10 patients) whereas in other cases the location of the fusion was evenly distributed - we can only speculate about the biological basis and relevance of this finding or whether it is simply a coincidence. The complex and heterogeneous nature of the rearrangement, coupled with its rarity, makes routine detection challenging. Routine karyotyping is usually inconclusive and FISH with either BCR-ABL1 or ETV6-RUNX1 probes (including specific “break-apart” probes) can miss the fusion – the cryptic insertions are often too small to generate a visible signal split. Indeed, insertions of part of the ABL1 gene into the ETV6 locus on 12p were frequently missed with the ETV6 probes (9/13 cases), and, vice versa, insertions of the 5’ part of the ETV6 gene into the ABL1 locus on 9q can be missed with the ABL1 specific probes (6/8 cases) (Online Supplementary Figure S3). Thus, only a combination of ETV6 and ABL1 probes reliably identifies the fusion by FISH. There are two types of ETV6-ABL1 transcript, caused by alternative splicing, but their expression was not consistently analyzed in previous studies. We detected both transcripts in all but one of the samples we analyzed, with the type B variant, including ETV6 exon 5, expressed at higher levels. These observations do not agree with those of previous studies which reported four cases expressing only the type A transcript. However, in three cases the type B fusion could have been missed due to the primers used for RT-PCR.33,39 In the other case, the absence of ETV6 exon 5 in the expressed transcript was shown to result from direct disruption of this exon by a fusion-generating DNA break.28 A chimeric kinase without ETV6 exon 5 lacks the GRB2 binding site and has attenuated oncogenic potential analogous to BCR-ABL1 deficient for GRB2 interaction.28,47,48 The predominance of the type B transcript does, therefore, have a biological basis. The recently reported complex network of kinase lesions in ALL makes RNA sequencing a useful additional diagnostic tool and in three cases in this series the ETV6-ABL1 type B fusion was identified this way. Some clinical study groups have opted to pre-select BCP-ALL for BCR-ABL1-like cases prior to screening for specific fusions.7,8 However, we present an haematologica | 2016; 101(9)

ETV6-ABL1 BCP-ALL case classified as BCR-ABL1-likenegative by Taqman low density array analysis (despite clustering with other ETV6-ABL1 and BCR-ABL1 cases on GEP) and a T-ALL case with ETV6-ABL1 classified as BCRABL1-like positive. Thus although methods to evaluate the BCR-ABL1-like signature were developed on expression data from BCP-ALL, they might also have the potential to identify T-ALL cases with kinase-activating lesions. This potential does, however, need to be evaluated further. Genomic and gene expression profiling demonstrates the similarity of ETV6-ABL1 and BCR-ABL1 cases. Besides the involvement of ABL1, these cases have a similar profile of secondary aberrations including frequent deletions of IKZF1, CDKN2A/B, PAX5, BTG1 and RB1 which distinguish them from other ALL.49-51 We have previously shown the prenatal origin of ETV6-ABL1 in a 5-year old child with ALL20 demonstrating the need for cooperating mutations to induce an overt disease. The two most frequent secondary aberrations in ETV6-ABL1 ALL (CDKN2A/B and IKZF1 loss) have both been shown to cooperate with BCR-ABL1 during leukemogenesis in mice and likely represent cooperating lesions also in ETV6-ABL1 leukemia.52,53 These and other CNA recurrently found in our cohort (ATP10, BTLA/CD200) have been associated with an unfavorable prognosis in ALL overall and specifically in BCRABL1 and BCR-ABL1-like ALL.6,13,54-57 Interestingly, we found no secondary aberrations in MPN-CP samples but progression to LBC was accompanied by the gain of a profile similar to that of ETV6-ABL1 ALL. Most studies show that the BCR-ABL1-like subgroup has an inferior prognosis.5-7 The final outcome was shown to depend significantly on the particular primary lesion (with childhood ALL bearing ABL-class aberrations having 5-year event-free survival and overall survival rates of 50% and 68%, respectively) and also on the presence of secondary aberrations (e.g. IKZF1 deletions).7 A recent study showed that, on current minimal residual disease-based protocols, the prognosis of BCR-ABL1-like patients treated with standard chemotherapy and stem cell transplantation is not inferior to that of patients with other ALL; however, only one out of 40 analyzed BCR-ABL1-like patients in this study bore an ABL gene fusion (NUP214-ABL1) and thus the results are not directly applicable to ETV6-ABL1 cases.58 Our study confirms the poor outcome associated with ETV6-ABL1 in ALL and AML; however, we did observe long-term survivors in childhood ALL among patients with a good initial response. The number of patients in this study was small but there is corroborating evidence in the literature from patients with other kinase fusions58 to support the notion that a TKI can be useful for treating slowresponding or high-risk ALL. Again among a small number of cases of MPN there was a trend towards improved outcome for CP patients treated upfront with a TKI (5/5 patients alive while 5/8 MPN-CP patients without TKI treatment died). In line with increased genomic complexity during LBC, patients who presented at this advanced stage did not survive despite TKI therapy. These observations are consistent with recent data in adult BCR-ABL1 ALL which showed that the effect of a TKI is optimal when it is administered early during frontline therapy.59 In conclusion, the ETV6-ABL1 fusion is a rare and complex rearrangement occurring in a variety of hematologic malignancies but especially ALL and MPN. The diagnosis is often incidental and only targeted RT-PCR screening 1091

M. Zaliova et al.

(e.g. in a multiplex setting with primers for ETV6-RUNX1 and BCR-ABL1 fusions, also detecting ETV6-ABL1 cases), a combination of both ETV6 and ABL1 FISH probes or RNAseq will reliably detect all cases. ETV6-ABL1 leukemias resemble BCR-ABL1 leukemias in many aspects including expression profile and secondary genetic aberrations. The outcome of ETV6-ABL1 ALL is often poor but in childhood ALL with an excellent early treatment response, continuous remissions seem to be achievable with current therapy without TKI. There is increasing evidence that among poor responding childhood ALL cases, as well as in adult acute leukemias and in MPN, the poor outcome can be abrogated by the use of a TKI which should be administered early, before progression of the

References 1. Golub TR, Goga A, Barker GF, et al. Oligomerization of the ABL tyrosine kinase by the Ets protein TEL in human leukemia. Mol Cell Biol. 1996;16(8):4107-4116. 2. Hannemann JR, McManus DM, Kabarowski JH, Wiedemann LM. Haemopoietic transformation by the TEL/ABL oncogene. Br J Haematol. 1998;102(2):475-485. 3. Okuda K, Golub TR, Gilliland DG, Griffin JD. p210BCR/ABL, p190BCR/ABL, and TEL/ABL activate similar signal transduction pathways in hematopoietic cell lines. Oncogene. 1996;13(6):1147-1152. 4. Million RP, Aster J, Gilliland DG, Van Etten RA. The Tel-Abl (ETV6-Abl) tyrosine kinase, product of complex (9;12) translocations in human leukemia, induces distinct myeloproliferative disease in mice. Blood. 2002;99(12):4568-4577. 5. Den Boer ML, van Slegtenhorst M, De Menezes RX, et al. A subtype of childhood acute lymphoblastic leukaemia with poor treatment outcome: a genome-wide classification study. Lancet Oncol. 2009;10(2):125134. 6. Mullighan CG, Su X, Zhang J, et al. Deletion of IKZF1 and prognosis in acute lymphoblastic leukemia. N Engl J Med. 2009;360(5):470-480. 7. Roberts KG, Li Y, Payne-Turner D, et al. Targetable kinase-activating lesions in Phlike acute lymphoblastic leukemia. N Engl J Med. 2014;371(11):1005-1015. 8. Roberts KG, Morin RD, Zhang J, et al. Genetic alterations activating kinase and cytokine receptor signaling in high-risk acute lymphoblastic leukemia. Cancer Cell. 2012;22(2):153-166. 9. Carroll M, Ohno-Jones S, Tamura S, et al. CGP 57148, a tyrosine kinase inhibitor, inhibits the growth of cells expressing BCRABL, TEL-ABL, and TEL-PDGFR fusion proteins. Blood. 1997;90(12):4947-4952. 10. Okuda K, Weisberg E, Gilliland DG, Griffin JD. ARG tyrosine kinase activity is inhibited by STI571. Blood. 2001;97(8):2440-2448. 11. Harvey RC, Mullighan CG, Wang X, et al. Identification of novel cluster groups in pediatric high-risk B-precursor acute lymphoblastic leukemia with gene expression profiling: correlation with genome-wide DNA copy number alterations, clinical characteristics, and outcome. Blood. 2010;116 (23):4874-4884.

1092

disease, and thus depends on timely diagnostics. Future screening algorithms should include appropriate measures to detect ETV6-ABL1 so that the most appropriate treatment strategy can be determined. Acknowledgments The authors would like to thank the Institute Paoli Calmettes Tumor Bank (Marseille, France) and Leukaemia and Lymphoma Research Childhood Leukaemia Cell Bank for providing cell samples. The work was supported by grants IGA-MZ NT/13170-4, GAUK 694414 and AZV 15-30626A and by grants from the National Institutes of Health (USA) CA98543 (COG Chair's grant), CA98413 (COG Statistical Center), and CA114766 (COG Specimen Banking).

12. Lengline E, Beldjord K, Dombret H, Soulier J, Boissel N, Clappier E. Successful tyrosine kinase inhibitor therapy in a refractory B-cell precursor acute lymphoblastic leukemia with EBF1-PDGFRB fusion. Haematologica. 2013;98(11):e146-148. 13. van der Veer A, Waanders E, Pieters R, et al. Independent prognostic value of BCRABL1-like signature and IKZF1 deletion, but not high CRLF2 expression, in children with B-cell precursor ALL. Blood. 2013;122(15):2622-2629. 14. Weston BW, Hayden MA, Roberts KG, et al. Tyrosine kinase inhibitor therapy induces remission in a patient with refractory EBF1PDGFRB-positive acute lymphoblastic leukemia. J Clin Oncol. 2013;31(25):e413416. 15. Baeumler J, Szuhai K, Falkenburg JH, van Schie ML, Ottmann OG, Nijmeijer BA. Establishment and cytogenetic characterization of a human acute lymphoblastic leukemia cell line (ALL-VG) with ETV6/ABL1 rearrangement. Cancer Genet Cytogenet. 2008;185(1):37-42. 16. Malone A, Langabeer S, O'Marcaigh A, Storey L, Bacon CL, Smith OP. A doctor(s) dilemma: ETV6-ABL1 positive acute lymphoblastic leukaemia. Br J Haematol. 2010;151(1):101-102. 17. Mozziconacci MJ, Sainty D, Chabannon C. A fifteen-year cytogenetic remission following interferon treatment in a patient with an indolent ETV6-ABL positive myeloproliferative syndrome. Am J Hematol. 2007;82(7):688-689. 18. Park J, Kim M, Lim J, et al. Variant of ETV6/ABL1 gene is associated with leukemia phenotype. Acta Haematol. 2013;129(2):78-82. 19. Yeung DT, Moulton DJ, Heatley SL, et al. Relapse of BCR-ABL1-like ALL mediated by the ABL1 kinase domain mutation T315I following initial response to dasatinib treatment. Leukemia. 2015;29(1):230-232. 20. Zuna J, Zaliova M, Muzikova K, et al. Acute leukemias with ETV6/ABL1 (TEL/ABL) fusion: poor prognosis and prenatal origin. Genes Chromosomes Cancer. 2010;49(10):873-884. 21. Andreasson P, Johansson B, Carlsson M, et al. BCR/ABL-negative chronic myeloid leukemia with ETV6/ABL fusion. Genes Chromosomes Cancer. 1997;20(3):299-304. 22. Barbouti A, Ahlgren T, Johansson B, et al. Clinical and genetic studies of ETV6/ABL1positive chronic myeloid leukaemia in blast crisis treated with imatinib mesylate. Br J Haematol. 2003;122(1):85-93.

23. Brunel V, Sainty D, Carbuccia N, et al. A TEL/ABL fusion gene on chromosome 12p13 in a case of Ph-, BCR- atypical CML. Leukemia. 1996;10:2003. 24. Gancheva K, Virchis A, Howard-Reeves J, et al. Myeloproliferative neoplasm with ETV6ABL1 fusion: a case report and literature review. Mol Cytogenet. 2013;6(1):39. 25. Kawamata N, Dashti A, Lu D, et al. Chronic phase of ETV6-ABL1 positive CML responds to imatinib. Genes Chromosomes Cancer. 2008;47(10):919-921. 26. Kelly JC, Shahbazi N, Scheerle J, et al. Insertion (12;9)(p13;q34q34): a cryptic rearrangement involving ABL1/ETV6 fusion in a patient with Philadelphia-negative chronic myeloid leukemia. Cancer Genet Cytogenet. 2009;192(1):36-39. 27. Keung YK, Beaty M, Steward W, Jackle B, Pettnati M. Chronic myelocytic leukemia with eosinophilia, t(9;12)(q34;p13), and ETV6-ABL gene rearrangement: case report and review of the literature. Cancer Genet Cytogenet. 2002;138(2):139-142. 28. La Starza R, Trubia M, Testoni N, et al. Clonal eosinophils are a morphologic hallmark of ETV6/ABL1 positive acute myeloid leukemia. Haematologica. 2002;87(8):789794. 29. Lin H, Guo JQ, Andreeff M, Arlinghaus RB. Detection of dual TEL-ABL transcripts and a Tel-Abl protein containing phosphotyrosine in a chronic myeloid leukemia patient. Leukemia. 2002;16(2):294-297. 30. Meyer-Monard S, Muhlematter D, Streit A, et al. Broad molecular screening of an unclassifiable myeloproliferative disorder reveals an unexpected ETV6/ABL1 fusion transcript. Leukemia. 2005;19(6):1096-1099. 31. Nand R, Bryke C, Kroft SH, Divgi A, Bredeson C, Atallah E. Myeloproliferative disorder with eosinophilia and ETV6-ABL gene rearrangement: efficacy of second-generation tyrosine kinase inhibitors. Leuk Res. 2009;33(8):1144-1146. 32. O'Brien SG, Vieira SA, Connors S, et al. Transient response to imatinib mesylate (STI571) in a patient with the ETV6-ABL t(9;12) translocation. Blood. 2002;99(9): 3465-3467. 33. Papadopoulos P, Ridge SA, Boucher CA, Stocking C, Wiedemann LM. The novel activation of ABL by fusion to an ets-related gene, TEL. Cancer Res. 1995;55(1):34-38. 34. Perna F, Abdel-Wahab O, Levine RL, Jhanwar SC, Imada K, Nimer SD. ETV6ABL1-positive "chronic myeloid leukemia": clinical and molecular response to tyrosine kinase inhibition. Haematologica.

haematologica | 2016; 101(9)

ETV6-ABL1-positive leukemias

2011;96(2):342-343. 35. Song JS, Shin SY, Lee ST, Kim HJ, Kim SH. A cryptic ETV6/ABL1 rearrangement represents a unique fluorescence in situ hybridization signal pattern in a patient with B acute lymphoblastic leukemia. Ann Lab Med. 2014;34(6):475-477. 36. Tirado CA, Sebastian S, Moore JO, Gong JZ, Goodman BK. Molecular and cytogenetic characterization of a novel rearrangement involving chromosomes 9, 12, and 17 resulting in ETV6 (TEL) and ABL fusion. Cancer Genet Cytogenet. 2005;157(1):74-77. 37. Van Limbergen H, Beverloo HB, van Drunen E, et al. Molecular cytogenetic and clinical findings in ETV6/ABL1-positive leukemia. Genes Chromosomes Cancer. 2001;30(3): 274-282. 38. Yamamoto K, Yakushijin K, Nakamachi Y, et al. Extramedullary T-lymphoid blast crisis of an ETV6/ABL1-positive myeloproliferative neoplasm with t(9;12)(q34;p13) and t(7;14)(p13;q11.2). Ann Hematol. 2014;93(8): 1435-1438. 39. Zhou MH, Gao L, Jing Y, et al. Detection of ETV6 gene rearrangements in adult acute lymphoblastic leukemia. Ann Hematol. 2012;91(8):1235-1243. 40. Hehlmann R, Heimpel H, Hasford J, et al. Randomized comparison of interferon-alpha with busulfan and hydroxyurea in chronic myelogenous leukemia. The German CML Study Group. Blood. 1994;84(12):4064-4077. 41. Choi HJ, Kim HR, Shin MG, et al. Spectra of chromosomal aberrations in 325 leukemia patients and implications for the development of new molecular detection systems. J Korean Med Sci. 2011;26(7):886-892. 42. Janssen JW, Ridge SA, Papadopoulos P, et al. The fusion of TEL and ABL in human acute lymphoblastic leukaemia is a rare event. Br J Haematol. 1995;90(1):222-224. 43. Miller DR, Coccia PF, Bleyer WA, et al. Early response to induction therapy as a predictor

haematologica | 2016; 101(9)

44.

45.

46.

47.

48.

49.

50.

51.

of disease-free survival and late recurrence of childhood acute lymphoblastic leukemia: a report from the Childrens Cancer Study Group. J Clin Oncol. 1989;7(12):1807-1815. Moricke A, Zimmermann M, Reiter A, et al. Long-term results of five consecutive trials in childhood acute lymphoblastic leukemia performed by the ALL-BFM study group from 1981 to 2000. Leukemia. 2010;24(2): 265-284. van Dongen JJ, Seriu T, Panzer-Grumayer ER, et al. Prognostic value of minimal residual disease in acute lymphoblastic leukaemia in childhood. Lancet. 1998;352(9142):1731-1738. Pane F, Intrieri M, Quintarelli C, Izzo B, Muccioli GC, Salvatore F. BCR/ABL genes and leukemic phenotype: from molecular mechanisms to clinical correlations. Oncogene. 2002;21(56):8652-8667. Million RP, Harakawa N, Roumiantsev S, Varticovski L, Van Etten RA. A direct binding site for Grb2 contributes to transformation and leukemogenesis by the Tel-Abl (ETV6-Abl) tyrosine kinase. Mol Cell Biol. 2004;24(11):4685-4695. Million RP, Van Etten RA. The Grb2 binding site is required for the induction of chronic myeloid leukemia-like disease in mice by the Bcr/Abl tyrosine kinase. Blood. 2000;96(2):664-670. Moorman AV, Enshaei A, Schwab C, et al. A novel integrated cytogenetic and genomic classification refines risk stratification in pediatric acute lymphoblastic leukemia. Blood. 2014;124(9):1434-1444. Mullighan CG, Goorha S, Radtke I, et al. Genome-wide analysis of genetic alterations in acute lymphoblastic leukaemia. Nature. 2007;446(7137):758-764. Mullighan CG, Miller CB, Radtke I, et al. BCR-ABL1 lymphoblastic leukaemia is characterized by the deletion of Ikaros. Nature. 2008;453(7191):110-114.

52. Virely C, Moulin S, Cobaleda C, et al. Haploinsufficiency of the IKZF1 (IKAROS) tumor suppressor gene cooperates with BCR-ABL in a transgenic model of acute lymphoblastic leukemia. Leukemia. 2010;24(6):1200-1204. 53. Williams RT, Roussel MF, Sherr CJ. Arf gene loss enhances oncogenicity and limits imatinib response in mouse models of Bcr-Ablinduced acute lymphoblastic leukemia. Proc Natl Acad Sci USA. 2006;103(17):6688-6693. 54. Ghazavi F, Clappier E, Lammens T, et al. CD200/BTLA deletions in pediatric precursor B-cell acute lymphoblastic leukemia treated according to the EORTC-CLG 58951 protocol. Haematologica. 2015;100(10): 1311-1319. 55. Iacobucci I, Ferrari A, Lonetti A, et al. CDKN2A/B alterations impair prognosis in adult BCR-ABL1-positive acute lymphoblastic leukemia patients. Clin Cancer Res. 2011;17(23):7413-7423. 56. Olsson L, Castor A, Behrendtz M, et al. Deletions of IKZF1 and SPRED1 are associated with poor prognosis in a population-based series of pediatric B-cell precursor acute lymphoblastic leukemia diagnosed between 1992 and 2011. Leukemia. 2014;28(2):302-310. 57. van der Veer A, Zaliova M, Mottadelli F, et al. IKZF1 status as a prognostic feature in BCR-ABL1-positive childhood ALL. Blood. 2014;123(11):1691-1698. 58. Roberts KG, Pei D, Campana D, et al. Outcomes of children with BCR-ABL1-like acute lymphoblastic leukemia treated with risk-directed therapy based on the levels of minimal residual disease. J Clin Oncol. 2014;32(27):3012-3020. 59. Fielding AK, Rowe JM, Buck G, et al. UKALLXII/ECOG2993: addition of imatinib to a standard treatment regimen enhances long-term outcomes in Philadelphia positive acute lymphoblastic leukemia. Blood. 2014;123(6):843-850.

1093

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Hodgkin Lymphoma

Ferrata Storti Foundation

Haematologica 2016 Volume 101(9):1094-1101

Detection and prognostic value of recurrent exportin 1 mutations in tumor and cell-free circulating DNA of patients with classical Hodgkin lymphoma Vincent Camus,1,2 Aspasia Stamatoullas,1,2 Sylvain Mareschal,2 Pierre-Julien Viailly,2 Nasrin Sarafan-Vasseur,3 Elodie Bohers,2 Sydney Dubois,2 Jean Michel Picquenot,2,4 Philippe Ruminy,2 Catherine Maingonnat,2 Philippe Bertrand,2 Marie Cornic,4 Valérie Tallon-Simon,6 Stéphanie Becker,7 Liana Veresezan,4 Thierry Frebourg,3 Pierre Vera,7 Christian Bastard,2,5 Hervé Tilly,1,2 and Fabrice Jardin1,2

1 Department of Hematology, Centre Henri Becquerel; 2INSERM U918, Centre Henri Becquerel, University of Rouen; 3INSERM U1079, University of Rouen; 4Department of Pathology, Centre Henri Becquerel; 5Department of Genetic Oncology, Centre Henri Becquerel; 6Clinical Research Unit, Centre Henri Becquerel; and 7Department of Nuclear Medicine and Radiology, Centre Henri Becquerel and QuantIF (Litis EA4108 – FR CNRS 3638), Rouen, France

ABSTRACT

Correspondence: fabrice.jardin@chb.unicancer.fr

Received: February 25, 2016. Accepted: June 1, 2016. Pre-published: June 13, 2016. doi:10.3324/haematol.2016.145102

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1094

1094

lassical Hodgkin lymphoma is one of the most common lymphomas and shares clinical and genetic features with primary mediastinal B-cell lymphoma. In this retrospective study, we analyzed the recurrent hotspot mutation of the exportin 1 (XPO1, p.E571K) gene, previously identified in primary mediastinal B-cell lymphoma, in biopsies and plasma circulating cell-free DNA from patients with classical Hodgkin lymphoma using a highly sensitive digital PCR technique. A total of 94 patients were included in the present study. This widely expressed XPO1 E571K mutation is present in one quarter of classical Hodgkin lymphoma patients (24.2%). Mutated and wildtype classical Hodgkin lymphomas were similar regarding the main clinical features. Patients with a detectable XPO1 mutation at the end of treatment displayed a tendency toward shorter progression-free survival, as compared to patients with undetectable mutation in plasma cell-free DNA (2-year progression-free survival: 57.1%, 95% confidence interval: 30.1-100% versus 2-year progression-free survival: 90.5%, 95% confidence interval: 78.8-100%, respectively, P=0.0601). To conclude, the detection of the XPO1 E571K mutation in biopsy and plasma cellfree DNA by digital PCR may be used as a novel biomarker in classical Hodgkin lymphoma for both diagnosis and minimal residual disease, and pinpoints a crucial role of XPO1 in classical Hodgkin lymphoma pathogenesis. The detection of somatic mutation in the plasma cell-free DNA of patients represents a major technological advance in the context of liquid biopsies and noninvasive management of classical Hodgkin lymphoma. Introduction Hodgkin lymphoma (HL) is one of the most frequent lymphomas in the world, accounting for 10% of newly diagnosed lymphomas and 1% of all cancers, with an incidence of about 3 cases per 100,000 people per year in western countries.1 HL has been classified into classical HL (cHL), which accounts for 95% of all cases, and the rare nodular lymphocyte-predominant HL, which is considered to be a distinct entity.2 The majority of cHL cases are diagnosed in 20-30 year old patients, with a second smaller peak in adults older than 55 years of age.3 cHL is further subdivided haematologica | 2016; 101(9)

XPO1 mutations in classical Hodgkin Lymphoma

into four subtypes: nodular sclerosis (60% of cases), mixed cellularity (30% of cases), lymphocyte-depleted and lymphocyte-rich HL.4 cHL patients have greatly benefited from multi-agent chemotherapy and improved radiation techniques, and 65-90% of patients can achieve diseasefree survival after five years, depending on stage and clinical risk factors.5 Those with a rapid response to initial treatment have the best outcomes and may benefit from truncated, less-toxic treatment regimens.6 Nevertheless, approximately 20-25% of patients will ultimately experience either primary refractoriness to chemotherapy (within 3 months of doxorubicin-based chemotherapy), early disease relapse (within 12 months after the end of first-line treatment) or late disease relapse,7 underlying the need to understand the mechanisms involved and to identify predictive biomarkers. The singularity of cHL is that tumor cells, designated as Hodgkin and Reed-Sternberg cells (HRS cells), usually account for only about 0.1-2% of cells in the tissue.8 The scarcity of HRS cells, embedded in an extensive inflammatory infiltrate, hampers their molecular analysis and thus the genomic landscape of cHL remains largely unknown. Constitutive activation of the NF-κB pathway in HRS cells has been demonstrated by targeted analyses using laser capture microdissection (LCM)9,10 or cell sorting by flow cytometry.11 Genomic gains of REL, encoding an NF-κB factor, are present in about 30% of cases,12,13 and β-2microglobulin (B2M) is frequently mutated in cHL, which is strongly associated with the nodular sclerosis subtype and better overall survival.11 We recently detected an unexpected recurrent point mutation of XPO1 (exportin 1, also known as CRM1: chromosome region maintenance 1) located in exon 15 (c.1711G>A), leading to the Glu571Lys (p.E571K) missense substitution in relapsed/refractory (R/R) primary mediastinal large B-cell lymphoma (PMBL) patients included in the LYSA LNH03 trial program.14,15 PMBL is well known to share several genetic features with cHL, including mutations in SOCS1,16 STAT617 and PTPN.16 XPO1 is a member of the importin-β superfamily of nuclear export receptors (also termed karyopherins) that mediates the translocation of numerous RNAs and cellular regulatory proteins, including tumor suppressor proteins (TSPs) such as p53, BRCA1, Survivin, NPM, APC and FOXO. The exportin 1 hydrophobic groove binds to the leucine-rich nuclear export signal (NES) domain of its cargo proteins. Importantly, selective inhibitors of nuclear export (SINE), a new class of small molecule inhibitors, have been shown to effectively target XPO1 and retain TSPs in the nucleus, and are currently being evaluated in phase 1 and 2 clinical trials for various cancer types.18,19 This XPO1 E571K mutation has been detected in approximately 25% of PMBL cases but at a lower frequency (1/10) in sorted HRS cells.11 Because of the scarcity of HRS cells in biopsies of patients with cHL, finding recurrent mutations could be easier in the plasma cell-free DNA (cfDNA) of these patients, with potentially fewer heterogeneity issues than tumor tissue testing.20 Furthermore, the concept of "liquid biopsy" was recently highlighted in a series of diffuse large B-cell lymphoma (DLBCL) patients, for whom high-throughput sequencing of a panel of target genes was performed, with the successful detection of somatic variants both in the tumor and in the plasma.21 Tumor circulating cfDNA has also been detected in the plasma of cHL patients.22 Notably, Oki et al.23 successfully haematologica | 2016; 101(9)

detected tumor-specific immunoglobulin gene segments in blood, indicating that the principle of liquid biopsy may also be relevant for cHL patients. We recently designed a highly sensitive and specific probe-based digital polymerase chain reaction (dPCR) assay for the detection of XPO1 E571K somatic mutations in plasma cfDNA24 that could be used as a tool to detect minimal residual disease (MRD). In this study, we investigated the prevalence and clinical relevance of XPO1 E571K in cHL cases and demonstrated that this recurrent mutation was detectable in both tumor and plasma, indicating that XPO1 mutations represent a new genetic biomarker, useful at the time of diagnosis or as a MRD marker.

Methods Patients We retrospectively considered adult patients treated for cHL at the Henri Becquerel Center (Rouen, France) between 2009 and 2015 using available frozen tumor DNA and formalin-fixed, paraffin-embedded (FFPE) samples. According to these inclusion criteria, 94 patients were included in the present analysis (FFPE samples: n=13; frozen tumor samples, n=81). Among these 94 patients, 50 were previously included in a biological monocentric prospective trial that aimed to assess the kinetics of cytokines (N° RCB 2009-A01117-50). These 50 patients (“cytokines” cohort) were included in this trial between 2010 and 2012 and had serial EDTA plasma samples, obtained from blood collection (Online Supplementary Methods) concomitant with the diagnostic biopsy and at the end of chemotherapy/radiotherapy treatment. All experiments were performed in accordance with the Declaration of Helsinki and the study was approved by our internal review board (N° 1601B). All patients gave informed consent for specimen collection, clinical data collection and biomarker analysis.

dPCR experiments A chip-based digital PCR (dPCR) platform (QuantStudio3D® Digital PCR system; Thermo Fisher Scientific, Carlsbad, CA, USA) was used for mutation detection. Mutation analysis with dPCR was based on a 5’-exonuclease assay using TaqMan® MGB probes targeting the XPO1 E571K mutation. dPCR was performed on DNA from tumor tissue and circulating cfDNA from plasma specimens at diagnosis and at the end of treatment. The sequences of our designed Custom TaqMan® MGB probes and primers are listed in the Online Supplementary Table S1. To consolidate the results obtained using this dPCR platform, we also analyzed all samples using a distinct droplet-based dPCR (ddPCRTM) platform (Qx200® droplet digital PCR system, Biorad laboratories, Hercules, CA, USA) using the same XPO1 dPCR assay, according to the manufacturer's instructions. PCR cycling conditions and reagent compositions are described in the Online Supplementary Table S2. To assess the specificity of our dPCR assay with the ddPCR platform, allele burden was measured in twenty preamplified cfDNA samples from DLBCL control blood samples previously found negative for the mutation by next-generation sequencing (NGS). In this study, 0.1% was considered to be the relevant threshold25 to discriminate positive versus negative samples for XPO1 E571K by dPCR (Online Supplementary Table S3).

Ion torrent personal genome machine™ (PGM) sequencing In the aim to validate results obtained by dPCR technologies, NGS experiments were performed using an Ion Torrent Personal 1095

V. Camus et al.

Genome Machine™ (PGM, Thermo Fisher Scientific, Carlsbad, CA, USA). Libraries were prepared from 10 ng of genomic DNA with primers from our previously reported Lymphopanel21,26 targeting the XPO1 E571K nucleotide variant (Online Supplementary Methods).

Response evaluation and PET-CT Responses after treatment were determined according to the International Workshop Group Response (Cheson) criteria.27 PET results were reported using the Deauville 5-point scale (Online Supplementary Methods).

Statistical analysis The relationships between clinical and molecular parameters were assessed using non-parametric tests: Mann-Whitney U test, Kendall’s tau correlation coefficient or the associated test or Fisher’s exact test when appropriate. Results of the comparative tests were considered statistically significant if P<0.05. All statistical analyses were performed using R software v3.0.2.

Results Mutations of XPO1 in cHL at the time of initial diagnosis The main clinical features of the patients are summarized in Table 1. Patients were considered mutated if they had an XPO1 E571K mutation in their biopsy extracted DNA. The mutation was found at a frequency of 24.2% in biopsy extracted DNA by dPCR (Table 1). In this series of 94 cHL patients, the main clinical features at the time of diagnosis, including age, sex ratio, stage and mediastinal involvement were similar between mutated and wild-type cases (Table 1). The concordance of the XPO1 E571K mutation dPCR results within the 50 biopsy/ plasma DNA pairs was highly significant (P=0.0179 Fisher’s exact test, Online Supplementary Table S4). PGM targeted sequencing analysis, carried out on the still available 48 DNA samples extracted from the “cytokines” cohort’s biopsies, confirmed dPCR results in 40/42 (NI=6) cHL (P<0.0001,

Table 1. Comparison of XPO1 WT and MT clinical characteristics.

Characteristics Number of patients Men (n=) Women (n=) Age (years, median) Age (SD) ECOG - 0-1 ECOG - 2-4 1st line : 2-4 ABVD + RT 1st line : 6-8 ABVD 1st line : BEACOPP based regimen (6 cycles) 1st line : MOPP based regimen Mediastinum involvement No mediastinum involvement Tumor size - Median (cm) Tumor size - SD Histology - Mixed cellularity cHL Histology - Nodular Sclerosis cHL Histology - Lymphocyte-rich cHL Histology - Unclassifiable Ann Arbor Stage 1-2 Ann Arbor stage 3-4 Hasenclever - 0-2 IPS Hasenclever - 3-7 IPS Hasenclever - NA Number of deaths End of treatment evaluation - CR End of treatment evaluation - RD End of treatment evaluation - NA Relapse PET C2 NEGATIVE (n=) PET C2 POSITIVE (n=) PET C2 - NA (n=)

All patients

XPO1 E571K MT

94 53 41 32 14.3 89 5 41 23 27 3 37 57 5 3.72 11 78 1 4 43 51 20 24 50 4 69 15 10 26 44 16 34

22 (24.2%) 14 8 31 11 22 0 12 3 7 0 12 10 6 3.52 5 16 0 1 13 9 4 4 14 1 18 3 1 6 10 7 5

69 (75.8%, NA=3) 38 31 34 15 65 4 28 19 19 3 24 45 5 3.86 5 60 1 3 28 41 16 19 34 3 48 12 9 20 32 9 28

0.6219 0.472 0.569 0.371

0.134 0.397 0.113

0.147 1

1 0.748

1 0.197

SD: stable disease; CR: complete remission; RD: refractory disease; PET C2: positron emission tomography after 2 cycles of chemotherapy; NA: not available; ECOG: Eastern Cooperative Oncology Group.

1096

haematologica | 2016; 101(9)

XPO1 mutations in classical Hodgkin Lymphoma

Cell lines analysis We also analyzed 6 HL-derived cell lines by dPCR and PGM : KM-H2 (DSMZ: ACC 8), HDLM-2 (ACC 17), L540 (ACC 72), L-428 (ACC 197), L-1236 (ACC 530), and U-H01 (ACC 626), and we found the XPO1 E571K mutation in the L-1236 and U-H01 cell lines, with VAF of 18.5% and 85%, respectively (Online Supplementary Figure S4). The karyotype of the L-1236 and U-H01 cell lines revealed amplification of the 2p15 - 2p16 region where XPO1 is located (chr2:61703069-61767418:2p15 region). Our data are in favor of duplication of the mutated XPO1 allele in the U-H01 cell line and of the wild-type allele in the L-1236 cell line. We did not find any other XPO1 variants within exons 15 to 18.

Response rate and survival analysis The complete remission rate at the end of treatment (85.7% versus 80%, P=0.537) was similar in MT and WT patients. There was no statistically different rate of posihaematologica | 2016; 101(9)

Kendall’s tau=0.462

P=0.0387

VAF (cfDNA)

Fisher’s exact test, Online Supplementary Table S5 and S6) and VAF were strongly correlated (Kendall’s tau coefficient=0.751, P=0.000193, Online Supplementary Figure S1). Digital PCR generates more copies (analysis points) of the sequence of interest than PGM (mean 16000 points by dPCR versus mean 4800 reads by PGM, data not shown), so the accuracy of the VAF might be lower by PGM than by digital PCR. The cases of discrepancies between PGM and digital PCR results involved FFPE DNA samples whose amplification by PGM is more difficult, with a lower sequencing depth. The XPO1 mutation was also investigated by dPCR in DNA from peripheral blood mononuclear cells (PBMC) of 4 mutated patients, and was not found in those samples (n=0/4, data not shown). The E571K XPO1 mutation VAF in biopsy extracted DNA ranged from 0.13% to 4.78% (median 0.435%). The VAF in plasmatic cfDNA ranged from 0.13% to 14.74% (median 0.24%, Online Supplementary Table S7). The correlation between cfDNA VAF and tumor DNA VAF was low but significant (Kendall’s tau=0.462, P=0.0387, Figure 1). This highlights the fact that, although HRS cells in the tumor are very sparse, the mutation frequencies seem to be correctly represented in the plasma. There was a trend toward higher cfDNA concentration (ng/mL of plasma) in advanced stage disease compared to localized disease (P=0.0894, Online Supplementary Figure S2) but no correlation between cfDNA concentration and XPO1 mutation VAF (P=0.735, Online Supplementary Figure S3). XPO1 mutations were found in both mixed cellularity and nodular sclerosis subtypes, indicating that the mutation is not a feature of a specific histological subtype. No XPO1 mutations were detected in the only lymphocyterich cHL case of the cohort (Table 1). Of note, 5 patients displayed the XPO1 mutation exclusively in the tumor DNA biopsy (Online Supplementary Table S4); 3 of these 5 patients had a localized stage II disease with a small tumor size (4-6 cm diameter), whereas the remaining 2 patients had advanced stage disease (1 stage III, 1 stage IV). Furthermore, 8 patients had an XPO1 mutation detectable only in the cfDNA. These results illustrate that considering only tumor biopsy DNA may lead to an underestimation of the mutation prevalence, and suggest that joint analysis of cfDNA and biopsy extracted DNA is more suitable to detect XPO1 mutations in cHL.

VAF (biopsy) Figure 1. Correlation between XPO1 E571K VAF in biopsies and cfDNA.

tivity of interim PET-CT after two cycles (C2) of chemotherapy (PET-2) in MT patients as compared with WT patients (41.2% versus 21.95% P=0.197). Concerning survival analysis, with a median follow-up of 34.5 months, overall survival (OS) and progression-free survival (PFS) were similar between MT and WT patients with a 2year OS and PFS probability of 95% [CI 95%: 85.9-100%] and 82.2% [CI 95%:67.8-99.7%], respectively, in MT patients, and of 98.1% [CI 95%: 94.5-100%] and 64.8% [CI 95%: 52.9-79.5%], respectively, in WT patients (P=0.637 and 0.248 for OS and PFS, respectively, Online Supplementary Figure S5). Of note, PET-2-positive patients had a shorter 2y-PFS, as compared to PET-2-negative patients (2y-PFS=62.5%, CI 95%:42.8-91.4% versus 2yPFS=86.1%, CI 95 %:75.5-98.2%, respectively, P=0.00159, Online Supplementary Figure S6).

cfDNA as a biomarker for MRD At the end of treatment plasma cfDNA was analyzed for 28 patients; the remaining 66 patients had either no available plasma sample at the end of treatment (n=46), or no XPO1 mutation at diagnosis in either the plasma or the biopsy (n=20). Figure 2 schematically shows the variation of the XPO1 E571K mutation VAF in the plasma cfDNA of the 28 evaluable patients between diagnosis and completion of treatment, with PET-CT at C2 and the end of treatment. For the majority of patients (n = 20), the cfDNA VAF decreased between diagnosis and the completion of treatment. The mutation became undetectable (using the threshold of 0.1%) in plasma for 16 patients. Five patients had plasma cfDNA with undetectable XPO1 E571K mutation at diagnosis and at the end of treatment (“negative” patients, Figure 2), but were mutated in biopsy extracted DNA. Patients with a detectable XPO1 mutation at the end of treatment displayed a trend toward shorter 2y-PFS, as compared to patients with undetectable mutation in plasma cell-free DNA (2y-PFS=57.1%, CI 95%:30.1-100% versus 2y-PFS=90.5%, CI 95%:78.8-100%, respectively, P=0.0601, Figure 3). Furthermore, patients with increasing cfDNA VAF at the end of treatment (n=3/28) seem to dis1097

V. Camus et al.

play shorter OS and PFS, but the sample size is too low to conclude this with certainty (2y-PFS=33.3%, CI 95%:6.7100% versus 2y-PFS=90%, CI 95%:77.8-100%, respectively, P=0.0439, data not shown). Of note, in 2/3 cases, PET was negative at the time of the detection of an MRD increase. In the full cohort of 94 patients, 26 patients experienced relapse during follow-up. Of the 7 patients with a positive plasma at the end of treatment, 4 (57%) relapsed during follow-up (Figure 3), which suggests that the assessment of plasma might be more sensitive than PET because only 1 patient had a positive PET at the end of treatment. Finally, 1 patient (patient NÂ°42) had an XPO1 E571K mutation at diagnosis in both biopsy and plasma samples as well as in cfDNA at the end of treatment (VAF 4.04%, 0.91% and 0.19%, respectively, Online Supplementary Figure S7). The mutation was also detected in plasma cfDNA 3 months after the end of treatment (VAF 0.12%) and in the relapse biopsy 10 months after the end of firstline ABVD chemotherapy (VAF relapse 2%, Figure 4). Unfortunately, we were unable to extract plasma cfDNA at relapse for this patient.

Discussion Herein we describe for the first time the existence of a widely present XPO1 E571K somatic mutation in a large cohort of patients with cHL, as identified by dPCR and

Diagnosis > EoT,

1098

P=0.000366

confirmed by NGS experiments. This observation is novel and could add new information on driver events and tumorigenesis in this disease. In total, 24.2% of our patients with cHL harbored the XPO1 E571K mutation. It is remarkable that 29% of all XPO1 mutations were found only in the plasma but not in the tumor because of the well-known tumor cell sparsity in HL. This indicates that cfDNA assessment might play an important role to define somatic mutations in this disease at the time of diagnosis. The detection of this mutation in the plasma cfDNA of patients represents a major technological progress in the onset of liquid biopsies in Hodgkin lymphoma. There is a pathological overlap between PMBL and, specifically, cHL of the nodular sclerosis subtype; whether both might derive from thymic B cells is currently an area of discussion.28 The fact that the XPO1 E571K mutation is so enriched in both of these subtypes reinforces the hypothesis of a common origin and a strong oncogenic role of this gene. Missense substitutions targeting XPO1 have also previously been reported at a low frequency (<5%) in chronic lymphocytic leukemia (CLL) and esophageal squamous cell carcinoma (ESCC), suggesting that these XPO1 mutations may also play a role in several oncogenic processes.29â&#x20AC;&#x201C;31 Importantly, alternative XPO1 variants, including those located at the hotspot,14,15,32,33 were not assessed by our dPCR approaches and we may have underestimated the rate of XPO1 mutations in cHL. Nevertheless, we did not detect those alternative XPO1 variants by NGS on cHL biopsies. To date, the role of

Figure 2. Plasmatic cfDNA XPO1 E571K mutation VAF at diagnosis and after treatment completion (after 6-8 cycles of chemotherapy for advanced stage disease, or after 3-4 cycles of chemotherapy and radiotherapy for localized stage disease), with interim and the end of treatment (EoT) PET-CT results. The interim PET at C2 is represented at the base of the arrow and the end of treatment PET is represented at the tip of the arrow. The arrows show the variations of the XPO1 mutation VAF in the plasma cfDNA between diagnosis and the end of treatment. The dotted line represents the detection limit (0.001).

haematologica | 2016; 101(9)

XPO1 mutations in classical Hodgkin Lymphoma

P=0.091

P=0.0601

Figure 3. OS and PFS according to the XPO1 E571K cfDNA results at the end of treatment (EoT). Negative: patients with undetectable XPO1 mutation, Positive: patients with detectable XPO1 E571K mutation in cfDNA at EoT.

XPO1 and the impact of the highly selected E571K mutation in the pathogenesis of cHL remain totally unknown. Several proteins known to play a major role in cHL oncogenesis, including STAT1/STAT6, FOXO1 or CIITA are also identified as cargo proteins.34â&#x20AC;&#x201C;36 Whether the XPO1 mutations may interfere with the nucleus/cytoplasm shuttle of these proteins remains to be determined. Genomic analyses of HL have been hampered until recently by the scarcity of HRS cells that are tremendously thinned out (to 1%) by a very large number of robust, nonmalignant infiltrating inflammatory cells that include macrophages, reactive lymphocytes, plasma cells and fibroblasts.4 Reactive infiltrating cells, especially T cells, usually make up at least 99% of the cells in the tumor mass,37 which may explain the low XPO1 E571K VAF observed in our series of cHL patients. We cannot affirm that the mutation in XPO1 is indeed present in HRS cells and not in some bystander cells, but the presence of the mutation in two Hodgkin cell lines and its absence in PBMC are in favor of the presence of this XPO1 mutation in HRS tumor cells specifically. The XPO1 E571K mutation in the L-1236 cell line was first detected by wholeexome sequencing (WES).38 In this particular disease, highly sensitive techniques like dPCR and targeted NGS are essential to highlight low frequency mutations. High-throughput techniques such as low-coverage whole genome sequencing,39 sequencing of circulating cfDNA22 or targeted exome sequencing of isolated HRS cells11 have recently helped to investigate genetic lesions underlying cHL, but sample sizes were very low. WES-based experiments of sorted Reed-Sternberg (RS) cells11,40,41 recently identified new point mutations in cHL cases, including several mutations already reported in PMBL, such as CIITA, SOCS1, STAT6 and B2M mutations, as well as one case harboring the E571K XPO1 mutation, but this study was restricted to 10 cases,11 with an insufficient median depth of sequencing (48X), which is potentially limiting for the discovery of genomic alterations. A recent NGS study on circulating cfDNA of 9 nodular sclerosis HL cases showed genomic imbalances in HRS haematologica | 2016; 101(9)

cells that can be identified at diagnosis with rapid normalization of circulating cfDNA profiles upon therapy initiation, suggesting a potential role for circulating cfDNA profiling in early response monitoring.22 Although cHL is curable even in advanced stages, up to 20% of patients will not be cured or achieve remission with initial therapy,7 and overall more than 25% of cHL patients will ultimately relapse, warranting the identification of a specific and trackable biomarker. In the present study, we performed highly sensitive analysis of a large cohort of 94 cHL patients and showed a recurrent XPO1 E571 somatic point mutation that could be considered as a new biomarker in approximately one quarter of cHL patients. Nevertheless, the applicability as a widely used biomarker would be low since three quarters of the patients do not have a XPO1 E571K mutation and of the positive 25%, an additional 26% are negative for XPO1 mutations in the plasma. In line with this aspect we have identified a trend toward unfavorable prognostic impact in terms of PFS in patients with detectable XPO1 E571K mutation in plasma cfDNA at the end of treatment, which could prove to be statistically significant in a larger cohort. We observed that 57% of the patients who ultimately relapsed were positive for XPO1 mutations in the plasma after the end of therapy. In fact, XPO1 assessment seems to be more sensitive than PET-CT since only 1 patient had a positive PET at the end of treatment. These results suggest that the clearance of the XPO1 mutation in plasma cfDNA may represent a new prognostic marker for mutated patients. A study in a larger prospective cohort is warranted and already planned to draw definitive conclusions. In accordance with this observation, it has been reported that pretreatment levels of plasma Epstein-Barr virus (EBV) DNA, as determined by quantitative real-time polymerase chain reaction (QRTPCR), are associated with inferior outcomes among a large cohort of patients with previously untreated, advancedstage Hodgkin lymphoma.42 Another interesting potential biomarker is the combination of serum CD163 and tumorspecific TARC proteins that predicted disease response in 1099

V. Camus et al.

Figure 4. Monitoring XPO1 mutations in biopsy and cfDNA of a refractory cHL patient (Patient N° 42) throughout her clinical history. The dotted line represents the detection limit (0.001).

a small cohort of 47 patients with HL.43 These results still require validation, and the usefulness of monitoring the MRD by XPO1 somatic mutation detection in cfDNA, compared with other potential biomarkers, has to be assessed. One of the limitations of our study, in addition to its retrospective nature with possible selection bias, is the absence of plasma available at the time of diagnosis for 44 patients which potentially led to underestimating the prevalence of the mutation in these patients, for whom only biopsies were analyzed. The few discrepancies between the detection of the mutation in the plasma and the tumor can be explained partly by the poor quality of some of our biopsies with tumor cell scarcity, potentially rendering the mutation detection in biopsy extracted DNA impossible, and also by the absence of tumor DNA release in plasma by certain tumors and the short half-life (10-15 minutes) of circulating DNA in plasma.44 Although additional XPO1 mutations were found in the plasma compared to the tumor, the sensitivity of the test remains low, with 26% of XPO1 mutations missed in the plasma at diagnosis. Another possibility for the lack of detection of this mutation in our cHL cases is the potential subclonality of the XPO1 E571K mutation. This point warrants further research. In addition, it has been previously described that plasma cfDNA concentrations do not uniformly correlate

References 1. Diehl V, Thomas RK, Re D. Part II: Hodgkin’s lymphoma--diagnosis and treatment. Lancet Oncol. 2004;5(1):19-26. 2. Campo E, Swerdlow SH, Harris NL, et al. The 2008 WHO classification of lymphoid neoplasms and beyond: evolving concepts and practical applications. Blood. 2011; 117(19):5019-032. 3. Siegel RL, Miller KD, Jemal A. Cancer sta-

1100

with disease stage in solid tumors and biological mechanisms that underlined tumor cfDNA amount release are incompletely understood.45,46 The results presented here are limited to patients with histologically confirmed cHL and may not be generalizable to HIV-associated HL, nodular lymphocyte-predominant HL or HL developing in the post-transplant setting. Finally, selective inhibitors of nuclear export (SINE), have been shown to effectively target XPO1 by retaining TSPs in the nucleus.47 Given the frequency of XPO1 E571 mutations in cHL, suggesting a crucial role of XPO1 in cHL pathophysiology, further investigation concerning the impact of XPO1 E571 mutations in response to SINE is currently being investigated.48 To conclude, we identify herein a widely spread XPO1 E571K mutation that is present in about one quarter of cHL patients. The presence of the XPO1 E571K mutation in plasma cfDNA may serve as a novel biomarker in cHL. These data have to be confirmed in a large dedicated prospective study, and it remains to be established whether this mutation adds new relevant value as compared to PET-scans and whether it can be targeted by SINE compounds for the treatment of cHL. Acknowledgments The authors would like to thank all the patients who provided samples for the study.

tistics, 2015. CA Cancer J Clin. 2015;65 (1):5-29. 4. Swerdlow SH, Campo E, Harris NL, et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues, Fourth Edition. Lyon: IARC 2008. 5. Hasenclever D, Diehl V. A prognostic score for advanced Hodgkin’s disease. International Prognostic Factors Project on Advanced Hodgkin’s Disease. N Engl J Med. 1998;339(21):1506-1514. 6. Hutchings M, Loft A, Hansen M, et al. FDG-

PET after two cycles of chemotherapy predicts treatment failure and progression-free survival in Hodgkin lymphoma. Blood. 2006;107(1):52-59. 7. Canellos GP, Rosenberg SA, Friedberg JW, Lister TA, Devita VT. Treatment of Hodgkin lymphoma: a 50-year perspective. J Clin Oncol. 2014;32(3):163-168. 8. Schmitz R, Stanelle J, Hansmann M-L, Küppers R. Pathogenesis of classical and lymphocyte-predominant Hodgkin lymphoma. Annu Rev Pathol. 2009;4:151-714.

haematologica | 2016; 101(9)

XPO1 mutations in classical Hodgkin Lymphoma 9. Slovak ML, Bedell V, Hsu Y-H, et al. Molecular karyotypes of Hodgkin and ReedSternberg cells at disease onset reveal distinct copy number alterations in chemosensitive versus refractory Hodgkin lymphoma. Clin Cancer Res. 2011; 17(10):3443-3454. 10. Schumacher MA, Schmitz R, Brune V, et al. Mutations in the genes coding for the NF- B regulating factors I B and A20 are uncommon in nodular lymphocyte-predominant Hodgkin’s lymphoma. Haematologica. 2010;95(1):153-157. 11. Reichel J, Chadburn A, Rubinstein PG, et al. Flow sorting and exome sequencing reveal the oncogenome of primary Hodgkin and Reed-Sternberg cells. Blood. 2015; 125(7):1061-1072. 12. Martín-Subero JI, Gesk S, Harder L, et al. Recurrent involvement of the REL and BCL11A loci in classical Hodgkin lymphoma. Blood. 2002;99(4):1474-1477. 13. Joos S, Menz CK, Wrobel G, et al. Classical Hodgkin lymphoma is characterized by recurrent copy number gains of the short arm of chromosome 2. Blood. 2002; 99(4):1381-1387. 14. Mareschal S, Dubois S, Viailly P-J, et al. Whole exome sequencing of relapsed/refractory patients expands the repertoire of somatic mutations in diffuse large B-cell lymphoma. Genes Chromosomes Cancer. 2015; 55(3):251-267. 15. Dubois S, Viailly PJ, Mareschal S, et al. Next Generation Sequencing in Diffuse Large B Cell Lymphoma Highlights Molecular Divergence and Therapeutic Opportunities: a LYSA Study. Clin Cancer Res. 2016 Jan 27 [Epub ahead of print]. 16. Steidl C, Gascoyne RD. The molecular pathogenesis of primary mediastinal large Bcell lymphoma. Blood. 2011; 118(10):26592669. 17. Ritz O, Guiter C, Castellano F, et al. Recurrent mutations of the STAT6 DNA binding domain in primary mediastinal Bcell lymphoma. Blood. 2009; 114(6):12361242. 18. Etchin J, Sanda T, Mansour MR, et al. KPT330 inhibitor of CRM1 (XPO1)-mediated nuclear export has selective anti-leukaemic activity in preclinical models of T-cell acute lymphoblastic leukaemia and acute myeloid leukaemia. Br J Haematol. 2013; 161(1):117127. 19. Lapalombella R, Sun Q, Williams K, et al. Selective inhibitors of nuclear export show that CRM1/XPO1 is a target in chronic lymphocytic leukemia. Blood. 2012; 120(23): 4621-4634. 20. Diaz LA, Bardelli A. Liquid biopsies: genotyping circulating tumor DNA. J Clin Oncol. 2014;32(6):579–586. 21. Bohers E, Viailly PJ, Dubois S, et al. Somatic mutations of cell-free circulating DNA detected by Next Generation Sequencing

haematologica | 2016; 101(9)

22.

23.

24.

25. 26.

27. 28.

29.

30.

31.

32.

33.

34.

reflect the genetic changes in both Germinal Center B-Cell like and Activated B-Cell like Diffuse Large B-Cell Lymphoma tumors at the time of diagnosis. Haematologica. 2015;100(7):280-284. Vandenberghe P, Wlodarska I, Tousseyn T, et al. Non-invasive detection of genomic imbalances in Hodgkin/Reed-Sternberg cells in early and advanced stage Hodgkin’s lymphoma by sequencing of circulating cell-free DNA: a technical proof-of-principle study. Lancet Haematol. 2015; 2(2):e55-65. Oki Y, Neelapu SS, Fanale M, et al. Detection of classical Hodgkin lymphoma specific sequence in peripheral blood using a next-generation sequencing approach. Br J Haematol. 2015;169(5):689-693. Camus V, Sarafan-Vasseur N, Bohers E, et al. Digital PCR for quantification of recurrent and potentially actionable somatic mutations in circulating free DNA from patients with diffuse large B-cell lymphoma. Leuk Lymphoma. 2016;1-9. Armbruster DA, Pry T. Limit of Blank, Limit of Detection and Limit of Quantitation. Clin Biochem Rev. 2008; 29(Suppl 1):S49-52. Dubois S, Mareschal S, Picquenot J-M, et al. Immunohistochemical and genomic profiles of diffuse large B-cell lymphomas: Implications for targeted EZH2 inhibitor therapy? Oncotarget. 2015;6(18):16712-16724. Cheson BD, Pfistner B, Juweid ME, et al. Revised response criteria for malignant lymphoma. J Clin Oncol. 2007; 25(5):579-586. Traverse-Glehen A, Pittaluga S, Gaulard P, et al. Mediastinal gray zone lymphoma: the missing link between classic Hodgkin’s lymphoma and mediastinal large B-cell lymphoma. Am J Surg Pathol. 2005; 29(11): 1411-21. Lin D-C, Hao J-J, Nagata Y, et al. Genomic and molecular characterization of esophageal squamous cell carcinoma. Nat Genet. 2014;46(5):467-473. Jeromin S, Weissmann S, Haferlach C, et al. SF3B1 mutations correlated to cytogenetics and mutations in NOTCH1, FBXW7, MYD88, XPO1 and TP53 in 1160 untreated CLL patients. Leukemia. 2014;28(1):108-117. Bond J, Bergon A, Durand A, et al. Cryptic XPO1-MLLT10 translocation is associated with HOXA locus deregulation in T-ALL. Blood. 2014;124(19):3023-3025. Guieze R, Robbe P, Clifford R, et al. Presence of multiple recurrent mutations confers poor trial outcome of relapsed/refractory CLL. Blood. 2015;126(18):2110-2117. Puente XS, Pinyol M, Quesada V, et al. Whole-genome sequencing identifies recurrent mutations in chronic lymphocytic leukaemia. Nature. 2011;475(7354):101-105. Chiu E, Gold T, Fettig V, LeVasseur MT, Cressman DE. Identification of a Nuclear Export Sequence in the MHC CIITA. J Immunol. 2015;194(12):6102-6111.

35. Xie L, Ushmorov A, Leithauser F, et al. FOXO1 is a tumor suppressor in classical Hodgkin lymphoma. Blood. 2012;119(15): 3503-3511. 36. Hao Y, Chapuy B, Monti S, et al. Selective JAK2 inhibition specifically decreases Hodgkin lymphoma and mediastinal large Bcell lymphoma growth in vitro and in vivo. Clin Cancer Res. 2014; 20(10):2674-2683. 37. Küppers R, Engert A, Hansmann M-L. Hodgkin lymphoma. J Clin Invest. 2012; 122(10):3439–3447. 38. Liu Y, Abdul Razak FR, Terpstra M, et al. The mutational landscape of Hodgkin lymphoma cell lines determined by wholeexome sequencing. Leukemia. 2014;28(11): 2248–2251. 39. Salipante SJ, Adey A, Thomas A, et al. Recurrent somatic loss of TNFRSF14 in classical Hodgkin lymphoma. Genes Chromosomes Cancer. 2015; 55(3):278-287. 40. Weniger MA, Melzner I, Menz CK, et al. Mutations of the tumor suppressor gene SOCS-1 in classical Hodgkin lymphoma are frequent and associated with nuclear phospho-STAT5 accumulation. Oncogene. 2006;25(18):2679-2684. 41. Ritz O, Rommel K, Dorsch K, et al. STAT6mediated BCL6 repression in primary mediastinal B-cell lymphoma (PMBL). Oncotarget. 2013;4(7):1093-1102. 42. Kanakry JA, Li H, Gellert LL, et al. Plasma Epstein-Barr virus DNA predicts outcome in advanced Hodgkin lymphoma: correlative analysis from a large North American cooperative group trial. Blood. 2013;1 21(18): 3547-3553. 43. Jones K, Vari F, Keane C, et al. Serum CD163 and TARC as disease response biomarkers in classical Hodgkin lymphoma. Clin Cancer Res. 2013;19(3):731-742. 44. Elshimali Y, Khaddour H, Sarkissyan M, Wu Y, Vadgama J. The Clinical Utilization of Circulating Cell Free DNA (CCFDNA) in Blood of Cancer Patients. Int J Mol Sci. 2013;14(9):18925-18958. 45. Beau-Faller M, Gaub MP, Schneider A, et al. Plasma DNA microsatellite panel as sensitive and tumor-specific marker in lung cancer patients. Int J Cancer J Int Cancer. 2003; 105(3):361-370. 46. Szpechcinski A, Chorostowska-Wynimko J, Struniawski R, et al. Cell-free DNA levels in plasma of patients with non-small-cell lung cancer and inflammatory lung disease. Br J Cancer. 2015;113(3):476-483. 47. Lapalombella R, Sun Q, Williams K, et al. Selective inhibitors of nuclear export show that CRM1/XPO1 is a target in chronic lymphocytic leukemia. Blood. 2012;120(23): 4621-4634. 48. Jardin F, Pujals A, Pelletier L, et al. Recurrent Mutations of the Exportin 1 Gene (XPO1) in Primary Mediastinal B-Cell Lymphoma: a LYSA Study. Blood. 2015;126(23):129.

1101

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Plasma Cell Disorders

Ferrata Storti Foundation

Haematologica 2016 Volume 101(9):1102-1109

Immunoparesis status in immunoglobulin light chain amyloidosis at diagnosis affects response and survival by regimen type Eli Muchtar,1 Angela Dispenzieri,1 Shaji K. Kumar,1 David Dingli,1 Martha Q. Lacy,1 Francis K. Buadi,1 Suzanne R. Hayman,1 Prashant Kapoor,1 Nelson Leung,1,2 Rajshekhar Chakraborty,1,3 Stephen Russell,1 John A. Lust,1 Yi Lin,1 Ronald S. Go,1 Steven Zeldenrust,1 Robert A. Kyle,1 S. Vincent Rajkumar,1 and Morie A. Gertz1

1 Division of Hematology, Mayo Clinic, Rochester; 2Division of Nephrology and Hypertension, Mayo Clinic, Rochester; and 3Hospitalist services, Essentia Health St. Joseph's Hospital, Brainerd, MN, USA

ABSTRACT

C Correspondence: gertz.morie@mayo.edu

Received: March 29, 2016. Accepted: June 10, 2016. Pre-published: June 16, 2016. doi:10.3324/haematol.2016.147041

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1102

linical tools to guide in the appropriate treatment selection in immunoglobulin light chain (AL) amyloidosis are not well developed. We evaluated the response and outcome for various regimens at first-line treatment (n=681) and first progression (n=240) stratified by the immunoparesis status at diagnosis. Immunoparesis was assessed by the average relative difference of the uninvolved immunoglobulins, classifying patients into a negative average relative difference (i.e. significant immunoparesis) or a positive average relative difference (no/modest immunoparesis). Treatment was categorized as autologous stem cell transplant and four non-transplant regimens (melphalan-based; bortezomib-based, immunomodulatory drug-based and dexamethasone alone). Patients with significant immunoparesis who underwent stem cell transplant had a significantly lower rate of very good partial response or better response (58%), progression-free survival (median 30 months) and overall survival (108 months), compared to those without significant immunoparesis (80%, 127 months, median not reached, respectively; P<0.001 for all comparisons). Among the nontransplant regimens, melphalan resulted in an unfavorable progressionfree survival (11 vs. 27 months; P<0.001) and overall survival (30 vs. 74 months; P=0.001) in patients with significant immunoparesis compared to those without significant immunoparesis. In contrast, no significant difference in outcomes between the immunoparesis groups was seen for those treated with bortezomib or immunomodulatory drugs. At first progression, immunoparesis status did not impact response or survival of any regimen. Melphalan at first-line provided poorer outcomes for patients with significant immunoparesis, while bortezomib or immunomodulatory drugs were more likely to overcome the adverse prognosis associated with significant immunoparesis. Introduction

1102

AL amyloidosis is a disorder with considerable therapeutic challenges. The disease can produce profound organ dysfunction which may limit the intensity of the delivered treatment, and therefore treatment efficacy is impaired. Autologous stem cell transplantation (ASCT) is considered the treatment of choice, when applicable, given its superior long-term survival.1,2 However, for non-ASCT eligible patients, therapy selection based on baseline disease characteristics has not been well-developed. The lack of guidelines for treatment selection at diagnosis may lead to the selection of inappropriate treatment that would lead to irreversible organ damage and a shorter survival. Recently, we have reported that significant immunoparesis is an adverse proghaematologica | 2016; 101(9)

Immunoparesis impact on therapy in AL amyloidosis

nostic factor in newly diagnosed AL amyloidosis.3 We proposed that immunoparesis assessment can be carried out by two methods. First, qualitatively, considering the number of uninvolved immunoglobulins below their lower limit of normal (LLN) and second, quantitatively, in which the average relative difference (ARD) of the uninvolved immunoglobulins from their respective lower limits is assessed. It was found that patients with significant immunoparesis (represented by either a reduction of all the uninvolved immunoglobulins below the LLN or by a negative ARD value) had a higher rate of treatment failure and reduced progression-free survival (PFS) and overall survival (OS) compared to patients with no or only moderate immunoparesis. The quantitative assessment was, however, shown to have a better discriminate power for survival. In this study we explored the results of selected treatment regimens at induction and first progression and analyzed these outcomes based on the ARD status at diagnosis.

Methods Six hundred and eighty-one patients (n=681) with systemic AL amyloidosis seen at the Mayo Clinic (Rochester, MN, USA) within 90 days of diagnosis between January 1, 2005 and August 31, 2015 were included. Data were extracted from a prospectively maintained database. The median follow-up of the surviving patients is 53 months (range 3-135 months). All patients gave written informed consent to have their medical records reviewed. The Mayo Clinic Foundation Institutional Review Board (IRB) approved this study. Patients were excluded if they had prior treatment, amyloidosis associated with a lymphoproliferative disorder, an incidental positive bone marrow and/or fat aspirate without an amyloid-specific syndrome or a localized disease. Three hundred and seventeen patients who met inclusion criteria but lacked response evaluation

data were not included in this study for the following reasons: no treatment/less than one cycle of treatment (n=58), lack of sufficient laboratory data for response evaluation, and/or early death rendering them inevaluable for response (n=259). The diagnosis of AL amyloidosis was based on a tissue specimen positive for Congo red staining and with green birefringence under polarized light, followed by typing with immunohistochemistry, immunofluorescence, or mass spectrometry. All patients had immunoglobulin measurements before treatment. For immunoparesis assessment we used a quantitative measure, which utilizes the ARD of the uninvolved immunoglobulins. ARD was calculated as the mean value of the relative difference of the uninvolved immunoglobulins from their lower limit of normal (See the Online Supplementary File for a sample calculation). Based on this method, patients were stratified to those with a negative ARD value (i.e. significant immunoparesis) and those with a positive ARD value (i.e. no immunoparesis or modest immunoparesis only). Treatment regimens at induction and first progression were grouped into autologous stem cell transplant (ASCT) and nontransplant regimens. The latter category includes the following regimen categories: melphalan-based regimen, bortezomib-based regimen (which includes 4 patients at first progression treated with other proteasome inhibitors) immunomodulatory drugs (IMiD)-based regimen (thalidomide, lenalidomide or pomalidomide), and dexamethasone alone. Eligibility criteria for ASCT at our center have been previously described.4 Two-hundred and ninety four patients (43% of the study population) progressed during follow-up. Of these, 51 patients (17%) progressed but were not treated and 3 additional patients had treatment of an unknown type. Therefore, 240 patients are evaluable for response and survival at first progression. Details of the specific regimens used at first-line of treatment and at first progression can be viewed in the Online Supplementary Material. For assignment of ARD group at first progression, we applied the baseline ARD groups at diagnosis, which represents an intrinsic feature of the disease not influenced by treatment.

Table 1. Baseline characteristics of all patients and by immunoparesis status at diagnosis.

Characteristic Age, years Median (range) Male sex, N (%) Organ involved, N (%) Median (range) >2 2004 Mayo AL amyloidosis stage, (n=652) N (%) I II III 2012 Mayo AL amyloidosis revised stage, (n=661) N (%) I II III IV Treatment categories ASCT Melphalan-based regimens Bortezomib-based regimens IMiD-based regimens Dexamethasone alone

All cohort (N=681)

Positive ARD (N=441)

Negative ARD (N=240)

62 (26-89) 422 (62%)

62 (26-89) 282 (64%)

63 (35-88) 140 (58%)

0.15 0.32

2 (1-4) 51 (21%)

0.22

2 (1-4) 161 (24%)

2 (1-4) 112 (25%)

165 (25%) 283 (44%) 204 (31%)

117 (28%) 176 (41%) 133 (31%)

48 (21%) 107 (47%) 71 (32%)

186 (28%) 166 (25%) 165 (25%) 144 (22%)

138 (32%) 102 (24%) 110 (25%) 82 (19%)

48 (21%) 64 (28%) 55 (24%) 62 (27%)

289 (42%) 220 (32%) 131 (19%) 31 (5%) 10 (2%)

191 (43%) 145 (33%) 86 (19.5%) 17 (4%) 2 (0.5%)

98 (41%) 75 (31%) 45 (19%) 14 (6%) 8 (3%)

0.17

0.007

0.04

ARD: average relative difference; ASCT: autologous stem cell transplantation; IMiD: immunomodulatory drug.

haematologica | 2016; 101(9)

1103

E. Muchtar et al. The Pearson c2 test and the Kruskal-Wallis test were used to ascertain differences between nominal and continuous variables, respectively. Evaluation of response, PFS and OS were performed in accordance with consensus criteria,5 the Kaplan-Meier method was used to measure survival analysis. Univariate Cox proportional regression analysis was used to examine the relationship between ARD groups, regimen types and survival outcome. P values less than 0.05 were considered significant. All statistical analyses were performed on JMP software (SAS, Cary, NC, USA).

underwent ASCT were younger (median age 59), with fewer organs involved (median one organ) and lower 2004 Mayo stage III (15%) and 2012 revised stages III-IV (25%), compared to patients treated with non-transplant regimens (median age 65; median involved organs 2; 2004 Mayo stage III 44%; 2012 revised stage III-IV 64%; all comparisons P<0.001). In patients who received nontransplant regimens, baseline characteristics were balanced between regimens in both immunoparesis groups (data not shown).

Results

Hematological response to first-line treatment by immunoparesis status

Baseline characteristics

The rate of very good partial response (VGPR) or better (â&#x2030;ĽVGPR) was 48% in patients with a negative ARD compared to 70% in patients with a positive ARD (P<0.001). While the rate of VGPR was similar between groups (25% and 26%, respectively), fewer patients in the negative ARD group reached a complete response (CR) compared to those with a positive ARD (23% vs. 44%). The rates of partial response and no response were 30% and 22% vs. 19% and 11%, respectively. The rate of VGPR or better response by treatment categories can be seen in Figure 1. When comparing the â&#x2030;ĽVGPR rate for each regimen between patients with a negative ARD to those with a positive ARD, the difference between groups was seen in those receiving ASCT (58% vs. 80%; P<0.001), a melphalan-based regimen (37% vs. 57%; P=0.006) and a bortezomib-based regimen (47% vs. 76%; P=0.001). However, no difference between groups was found in the IMiD-based regimen category (57% vs. 53%; P=0.81) or dexamethasone alone (25% vs. 50%; P=0.5)

The median age of the 681 patients was 62 years [range 26-89]; 62% of patients were male. The median number of organs involved was 2 (range 1-4). The 2004 Mayo stage6 and the 2012 revised Mayo stage7 are listed in Table 1.

First line treatment Induction treatment categories are listed in Table 1. A melphalan-based regimen was given to 32% of patients, bortezomib-based regimens to 19% of patients, an IMiDbased regimen to 5% of patients and dexamethasone alone for 2% of patients. Autologous stem cell transplant (ASCT) was performed in 32% of patients without prior induction, while an additional 10% of patients proceeded to ASCT following induction treatment making a total of 42% undergoing ASCT as a first-line treatment. There was no difference in treatment categories distribution, including ASCT, melphalan-based, bortezomib-based and IMiD-based regimens between immunoparesis groups (P=0.66), but not dexamethasone alone. Patients who

P<0.001 P=0.001

Rate of VGPR or better (%)

P<0.001

P=0.006

P=0.81 P=0.5

Figure 1. Rate of very good partial response (VGPR) or better by regimen type and immunoparesis status at first-line treatment. ASCT: autologous stem cell transplant; ARD: average relative difference; IMiD: immunomodulatory drug.

1104

haematologica | 2016; 101(9)

Immunoparesis impact on therapy in AL amyloidosis

Organ response by immunoparesis status Organ response, (i.e. response in at least one involved organ) was seen in 57% of patients, and was significantly lower in patients with a negative ARD compared to those with a positive ARD (48% vs. 62%, respectively; P<0.001) (Figure 2). No difference in the achievement of organ response between the two immunoparesis groups was seen in those treated with ASCT (66% vs. 75%; P=0.12), a bortezomib-based regimen (48% vs. 56%; P=0.41), an IMiD-based regimen (54% vs. 43%; P=0.56) and dexamethasone alone (14% vs. 0%; P=0.59). However, for patients treated with a melphalan-based regimen, those with a negative ARD were less likely to achieve an organ response compared to those with a positive ARD (28% vs. 52%; P<0.001).

Effect of immunoparesis status on progression-free survival and overall survival by regimen type Patients with a negative ARD, indicating significant immunoparesis, had a shorter PFS (median 16 months) compared to patients with a positive ARD [49 months, P<0.001; Hazard ratio (HR) 1.9, (95% confidence interval (CI) 1.5-2.3)]. The comparison of PFS between immunoparesis groups based on the given regimen demonstrated a significant difference in favor of the positive ARD group and was maintained in those undergoing ASCT [30 vs. 127 months, respectively, P<0.001; HR 2.3 (95% CI 1.6-3.2)] and for those treated with a melphalan-based regimen [11 vs. 27 months, P<0.001; HR 1.9 (95% CI 1.4-2.6)]. In contrast, no significant difference in PFS was noted between the negative and positive ARD groups in those treated with a bortezomib-based regimen [14 vs. 23 months, P=0.13; HR 1.4 (95% CI 0.9-2.3)], an IMiD-based regimen [20 vs. 13 months, P=0.84; HR 0.9 (95% CI 0.4-2)] or dexamethasone alone [5 vs. 6 months, P=0.66; HR 0.7 (95% CI 0.1-5)] (Figure 3).

Patients with a negative ARD had a shorter OS (median 66 months) compared to patients with a positive ARD [median 127 months, P<0.001; HR 1.6, (95% CI 1.3-2)]. In patients who underwent ASCT, inferior OS was seen in patients with a negative ARD (median 108 months) compared to patients with a positive ARD [median not reached, P=0.01; HR 1.9, (95% CI 1.1-3)]. Similarly, patients with a negative ARD treated with a melphalanbased regimen had an inferior OS compared to patients with a positive ARD treated with the same regimen type [30 vs. 74 months, P=0.001; HR 1.7, (95% CI 1.2-2.5)]. However, no significant difference in OS between immunoparesis groups was observed for a bortezomibbased regimen [57 vs. 41 months, P=0.61; HR 1.2, (95% CI 0.6-2.0)], an IMiD-based regimen [32 vs. 16 months, P=0.4; HR 0.7, (95% CI 0.3-1.6)] or dexamethasone alone [20 months vs. median not reached, P=0.57; HR 1.8, (95% CI 0.3-34)] (Figure 4). Of note, patients with a positive ARD treated with an IMiD-based regimen had a significantly shorter OS compared to other non-transplant regimens in this group (P=0.01).

Treatment at first progression The progression rate in patients with a negative ARD was 53% compared to 39% in patients with a positive ARD (P<0.001). The median time to first progression was 14 months, shorter in patients with a negative ARD (12 months) compared to patients with a positive ARD (17 months; P=0.01). The characteristics of patients at first progression are listed in Table 2. The most common regimen used at first progression was bortezomib-based (53%), followed by IMiD-based (25%), melphalan-based (14%), ASCT (5%) and dexamethasone alone (3%). A different treatment distribution was noted at first progression by baseline immunoparesis status. Patients with a baseline negative ARD were more likely to receive a mel-

P=0.12 P<0.001 P=0.41

P=0.56

P<0.001

P=0.59

Figure 2. Rate of organ response by regimen type and immunoparesis status at first-line treatment. ASCT: autologous stem cell transplant; ARD: average relative difference; IMiD: immunomodulatory drug.

haematologica | 2016; 101(9)

1105

E. Muchtar et al.

phalanâ&#x20AC;&#x201C;based (18%) or an IMiD-based regimen (33%) compared to patients with a positive ARD (11% and 20%, respectively). In contrast, patients with a positive ARD were more likely to receive a bortezomib-based regimen (59%) compared to those with a negative ARD (44%). ASCT was used at a similar rate between groups (4% in

the negative ARD groups and 6% in the positive ARD group) (P for all comparisons=0.02). Moreover, patients receiving a melphalan-based regimen at first progression were less likely to receive ASCT at first-line (30%) compared to other non-transplant regimens at first progression (49%; P=0.04), with no difference between ARD groups.

Table 2. Characteristics of patients at first progression by immunoparesis status at diagnosis.

Characteristic

All patients (N=240)

Age, years Median (range) 61 (36-83) Male sex, N (%) 154 (64%) Organ involved, N (%) Median (range) 2 (1-4) >2 63 (26%) 2004 Mayo AL amyloidosis stage, (n=231) N (%) I 50 (22%) II 125 (54%) III 56 (24%) 2012 Mayo AL amyloidosis revised stage, (n=234) N (%) I 61 (26%) II 73 (31%) III 62 (27%) IV 38 (16%) Treatment categories ASCT 12 (5%) Melphalan-based regimens 33 (14%) Bortezomib-based regimens 127 (53%) IMiD-based regimens 61 (25%) Dexamethasone alone 7 (3%)

Positive ARD at baseline (N=138)

Negative ARD at baseline (N=102)

61 (36-83) 88 (64%)

62 (38-82) 66 (65%)

0.12 0.88

2 (1-4) 41 (30%)

1 (1-4) 22 (22%)

0.003 0.15

32 (24%) 73 (55%) 28 (21%)

18 (18%) 52 (53%) 28 (29%)

0.33

41 (30%) 40 (30%) 34 (25%) 20 (15%)

20 (20%) 33 (33%) 28 (28%) 18 (18%)

8 (6%) 15 (11%) 82 (59%) 27 (20%) 6 (4%)

4 (4%) 18 (18%) 45 (44%) 34 (33%) 1 (1%)

0.36

0.02

ARD: average relative difference; ASCT: autologous stem cell transplantation; IMiD: immunomodulatory drug.

Figure 3. Progression-free survival by regimen type and immunoparesis status at first-line treatment. ARD: average relative difference A. Autologous stem cell transplantation (ASCT) B. Melphalan-based regimen C. Bortezomib-based regimen D. IMiD-based regimen.

1106

haematologica | 2016; 101(9)

Immunoparesis impact on therapy in AL amyloidosis

Response at first progression The rate of VGPR or better at first progression did not differ between the immunoparesis groups. Fifty-seven percent of patients with a baseline negative ARD achieved ≥VGPR compared to 56% in patients with a positive baseline ARD (P=0.97). No difference was seen in the depth of response between groups (CR 19% vs. 15%; VGPR 23% vs. 28%, respectively; P=0.32). Organ response also did not differ between negative and positive ARD groups (42% vs. 40%; P=0.73). The comparison of the rate of ≥VGPR by regimens between patients with a negative ARD to those with a positive ARD can be viewed in Figure 5. ASCT at first progression was utilized in 12 patients, and yielded the highest VGPR or better response rate (100% in patients with a negative ARD and 86% in patients with a positive ARD; P=0.46). As for the non-transplant regimens, no difference in the rate of ≥VGPR between the negative and positive ARD groups was noted for any regimen (melphalan-based regimen 29% vs. 43%, P=0.43; bortezomib-based 74% vs. 60%, P=0.14; IMiD-based 50% vs. 50%, P=1.0; dexamethasone alone 0% vs. 33%, P=0.39).

for any regimen at first progression [ASCT median not reached in both, P=0.23; HR not estimable); melphalanbased median 8 months vs. 13 months , P=0.36; HR 1.5 (95% CI 0.6-3.5); bortezomib-based 27 vs. 22 months, P=0.64; HR 0.9 (95% CI 0.5-1.5); IMiD-based 17 months both, P=0.59; HR 0.8 (95% CI 0.4-1.7); dexamethasone alone 3 vs. 7 months, P=0.17; HR 5.5 (95% CI 0.2-138)]. OS from first progression was similar between groups [median 42 months in those with a negative ARD compared to 47 months in those with a positive ARD, P=0.65; HR 1.1 (95%CI 0.8-1.5)] (Figure 6B). No significant difference in OS was seen between the negative and positive ARD groups for any given regimen at first progression [ASCT median not reached in both groups, P=0.18; HR not estimable; melphalan-based median 31 months vs. 40 months, P=0.9; HR 0.9 (95%CI 0.4-2.5); bortezomibbased median not reached vs. 72 months, P=0.42; HR 0.8 (95%CI 0.4-1.4); IMiD-based 54 months vs. 68 months, P=0.95; HR 1 (95%CI 0.5-2.2); dexamethasone alone 5 vs. 37 months, P=0.17; HR 5.5 (95%CI 0.2-138)].

Discussion Survival from first progression PFS from first progression was comparable between those with a negative ARD and those with a positive ARD (median 17 vs. 20 months, respectively, P=0.95; HR 1 95% CI 0.7-1.4) (Figure 6A). No significant difference in PFS was seen between the negative and positive ARD groups

This study provides data on the differential response to various chemotherapeutic regimens in AL amyloidosis at first-line and first progression stratified by the immunoparesis status at diagnosis. For newly diagnosed patients, ASCT provided the best outcomes within each immuno-

Figure 4. Overall survival by regimen type and immunoparesis status at first-line treatment. ARD: average relative difference. A. Autologous stem cell transplantation (ASCT) B. Melphalan-based regimen C. Bortezomib-based regimen D. IMiD-based regimen.

haematologica | 2016; 101(9)

1107

E. Muchtar et al. P=0.46

Rate of VGPR or better (%)

P=0.14

P=0.97 P=1.0 P=0.43 P=0.39

Figure 5. Rate of ≥VGPR or better by regimen type and immunoparesis status at first progression. ASCT: autologous stem cell transplant; ARD: average relative difference; IMiD: immunomodulatory drug.

paresis group, but with a greater benefit seen in patients without significant immunoparesis (i.e. positive ARD) compared to those with significant immunoparesis (i.e. negative ARD). For non-transplant first-line regimens, a comparison between immunoparesis groups shows that while the advantage for those with a positive ARD was seen with a melphalan-based regimen, this advantage disappeared with bortezomib or IMiD-based regimens. The response and outcome at first progression were comparable between immunoparesis groups, suggesting neutralization of the adverse prognostic effect of immunosuppression at progression. However, when the data was analyzed by regimen, melphalan-based and dexamethasone alone generally produced poorer response and survival compared to other regimens, possibly reflecting selection bias. ASCT provided an improved survival rate in each immunoparesis group. The advantage of ASCT was seen as a higher rate of ≥VGPR as well as a longer PFS and OS. ASCT is, however, applicable only in a fraction of patients, which carry a more favorable prognosis, as reflected by younger age, a median of one involved organ and lower risk-stratified stage. In this study, ASCT (with or without prior induction) was performed in over 40% of patients, and reflects a referral bias. It is important to note that although patients with significant immunoparesis were shown to have more cardiac involvement, less renal involvement and a higher tumor burden (dFLC, bone marrow plasma cell percentage),3 they were as likely to proceed to ASCT as those without significant immunoparesis (41% vs. 43%, respectively). Even with ASCT, patients with a negative ARD had a lower ≥VGPR rate, PFS and OS compared to patients with a positive ARD, although organ 1108

response was achieved at a similar rate (66% vs. 75%, respectively). It appears that ASCT is the treatment of choice regardless of immunoparesis status, but response and response duration in patients with significant immunoparesis are lower than in those without significant immunoparesis. Exploration of the treatment options in the non-transplant regimens reveals that patients with significant immunoparesis had a poorer response to melphalan-based regimens. This was reflected by a low rate of ≥VGPR (which represents the therapeutic endpoint in AL amyloidosis),7 as well as significantly lower organ response rate. In comparison, patients lacking significant immunoparesis treated with similar regimens had higher hematological (57%) and organ response (53%) rates. Moreover, a PFS and OS advantage in favor of patients without significant immunoparesis was seen in those treated with high-dose melphalan or low-intensity melphalan, but not for bortezomib or IMiDs. This finding suggests that in ASCT ineligible patients, melphalan has a greater impact in those without significant immunoparesis, while those with significant immunoparesis are less likely to benefit from melphalan. The reason for this is unclear, but might reflect different disease biology based on immunoparesis status. Melphalan, an alkylating agent is, unlike bortezomib and IMiDs, genotoxic.8 As such, it has the potential to impact DNA integrity and accelerate progression of a genomically unstable plasma cell clone. At progression, melphalan therapy produced the poorest results, which also supports this hypothesis. However, this clearly needs further investigation. While IMiDs produced a relatively good response rate and survival in patients with significant immunoparesis, haematologica | 2016; 101(9)

Immunoparesis impact on therapy in AL amyloidosis

Figure 6. Survival curves at first progression by baseline immunoparesis status, ARD: average relative difference. A. Progression-free survival from progression. B. Overall survival from progression.

References 1. D'Souza A, Dispenzieri A, Wirk B, et al. Improved Outcomes After Autologous Hematopoietic Cell Transplantation for Light Chain Amyloidosis: A Center for International Blood and Marrow Transplant Research Study. J Clin Oncol. 2015; 33(32):3741-3749. 2. Cordes S, Dispenzieri A, Lacy MQ, et al. Ten-year survival after autologous stem cell transplantation for immunoglobulin light chain amyloidosis. Cancer. 2012; 118(24):6105-6109. 3. Muchtar E, Dispenzieri A, Kumar SK, et al. Immunoparesis in newly diagnosed AL amyloidosis is a marker for response and survival. Leukemia. 2016. [Epub ahead of print].10.1038/leu.2016.140

haematologica | 2016; 101(9)

they were associated with a reduced OS in patients without significant immunoparesis. IMiDs in AL amyloidosis are generally not well tolerated and produce modest benefit,9,10 and therefore are not considered as first-line treatment for most patients. Based on the data presented, the selection of an IMiD as first-line should be discouraged in those without significant immunoparesis but can be considered for those with significant immunoparesis. The number of patients treated with IMiDs in this study was small, so this conclusion must be taken with caution and should be confirmed by other studies. When dexamethasone alone was utilized, the results were generally poor. Patients at first progression were still able to achieve â&#x2030;ĽVGPR in over 50% of patients in both groups, but had a shorter duration of response with a second-line of therapy. PFS and OS were comparable between patients with or without significant immunoparesis, independent of the salvage regimen. However, patients treated with melphalan-based regimens had poorer response, PFS and OS in immunoparesis groups. This may reflect a bias, as patients treated with low-dose melphalan at first progression were less likely to have been ASCT-eligible at diagnosis and carry a poorer prognosis. In conclusion, first-line ASCT provides the best response and survival in patients with AL amyloidosis, irrespective of immunoparesis status. Better results for ASCT, however, were seen in those without significant immunoparesis. For non-transplant regimens, bortezomib and IMiDs were more likely to overcome the poorer prognosis associated with significant immunoparesis, while low-dose melphalan was associated with the least benefit for patients with significant immunoparesis. These findings should be assessed in prospective studies.

. 4. Dispenzieri A, Buadi F, Kumar SK, et al. Treatment of Immunoglobulin Light Chain Amyloidosis: Mayo Stratification of Myeloma and Risk-Adapted Therapy (mSMART) Consensus Statement. Mayo Clin Proc. 2015;90(8):1054-1081. 5. Comenzo RL, Reece D, Palladini G, et al. Consensus guidelines for the conduct and reporting of clinical trials in systemic lightchain amyloidosis. Leukemia. 2012; 26(11):2317-2325. 6. Dispenzieri A, Gertz MA, Kyle RA, et al. Prognostication of survival using cardiac troponins and N-terminal pro-brain natriuretic peptide in patients with primary systemic amyloidosis undergoing peripheral blood stem cell transplantation. Blood. 2004;104(6):1881-1887. 7. Kumar S, Dispenzieri A, Lacy MQ, et al.

Revised prognostic staging system for light chain amyloidosis incorporating cardiac biomarkers and serum free light chain measurements. J Clin Oncol. 2012; 30(9):989-995. 8. Ranaldi R, Palma S, Tanzarella C, Lascialfari A, Cinelli S, Pacchierotti F. Effect of p53 haploinsufficiency on melphalan-induced genotoxic effects in mouse bone marrow and peripheral blood. Mutat Res. 2007; 615(1-2):57-65. 9. Dispenzieri A, Lacy MQ, Rajkumar SV, et al. Poor tolerance to high doses of thalidomide in patients with primary systemic amyloidosis. Amyloid. 2003;10(4):257-261. 10. Sanchorawala V, Wright DG, Rosenzweig M, et al. Lenalidomide and dexamethasone in the treatment of AL amyloidosis: results of a phase 2 trial. Blood. 2007;109(2):492496.

1109

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Acute Lymphoblastic Leukemia

Ferrata Storti Foundation

Geriatric assessment in multiple myeloma patients: validation of the International Myeloma Working Group (IMWG) score and comparison with other common comorbidity scores

Monika Engelhardt,1 Sandra Maria Dold,1 Gabriele Ihorst,2 Alexander Zober,1 Mandy MĂśller,1 Heike Reinhardt,1 Stefanie Hieke,3 Martin Schumacher,3 and Ralph WĂ¤sch1

Department of Medicine I, Hematology, Oncology & Stem Cell Transplantation, Medical Center, Faculty of Medicine, University of Freiburg; 2Clinical Trials Unit, Medical Center, Faculty of Medicine, University of Freiburg; and 3Institute for Medical Biometry and Statistics, Medical Center, Faculty of Medicine, University of Freiburg, Germany

Haematologica 2016 Volume 101(9):1110-1119

ABSTRACT

Correspondence: monika.engelhardt@uniklinik-freiburg.de

Received: April 20, 2016. Accepted: June 10, 2016. Pre-published: June 16, 2016. doi:10.3324/haematol.2016.148189

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1110

1110

his first validation of the International Myeloma Working Group geriatric assessment in 125 newly diagnosed multiple myeloma patients was performed using the International Myeloma Working Group score based on age, the Charlson Comorbidity Index and cognitive and physical conditions (Activities of Daily Living / Instrumental Activities of Daily Living) to classify patients as fit, intermediate-fit or frail. We verified the International Myeloma Working Group score's impact on outcome, and whether additional tools complement it. Since our prior analyses determined renal, lung and Karnofsky performance impairment as multivariate risks, and the inclusion of frailty, age and cytogenetics complements this, we included the revised myeloma comorbidity index, the Charlson Comorbidity Index, the Hematopoietic Cell Transplantation-Comorbidity Index and the Kaplan-Feinstein Index in this assessment. Multivariate analysis confirmed cytogenetics, Activities of Daily Living, Instrumental Activities of Daily Living and the Charlson Comorbidity Index as risks: 3-year overall survival for fit, intermediate-fit and frail patients was 91%, 77% and 47%, respectively. Using the Charlson Comorbidity Index, the Hematopoietic Cell Transplantation-Comorbidity Index, the KaplanFeinstein Index and the revised Myeloma Comorbidity Index allowed us to define fit and frail patients with distinct progression-free and overall survival rates, with the most pronounced differences evidenced via the International Myeloma Working Group score, the Charlson Comorbidity Index and the revised Myeloma Comorbidity Index. Since the Charlson Comorbidity Index is included in the International Myeloma Working Group score, we propose the latter and the revised Myeloma Comorbidity Index for future frailty measurements. Both are useful instruments for identifying myeloma patients with a geriatric risk profile and have a strong prognostic value for functional decline and overall survival. The study was registered as: (clinicaltrials.gov Identifier: 00003686).

Introduction Treatment concepts and survival of multiple myeloma (MM) patients have substantially changed due to our better understanding of the disease, novel risk-adapted therapies and improved supportive care measures.1-3 MM typically affects elderly patients haematologica | 2016; 101(9)

Geriatric assessment validation in myeloma

whose prognosis varies widely and remains more unfavorable than in younger patients. This is shown to be related to a higher frequency of treatment discontinuation and nonhematological adverse events.3-8 Moreover, elderly and frail patients are less frequently included in clinical trials and may receive fewer novel agents.2-4,7-10 This typically occurs because multimorbidity and the interaction of various medications can complicate the treatment of patients, limit their physical condition and impair survival.7,8,10 However, the global population is aging rapidly and the increasing number of elderly patients demands reliable tools to assess their vulnerability as expressed in chronic conditions and limitations in daily activities. Novel risk scores can either rely on MM tumor burden, as postulated with the combined use of the International Staging System (ISS), lactate dehydrogenase (LDH) and high-risk cytogenetics,5,11 and/or the functional condition of patients, which is assessed worldwide.12-16 A functional or geriatric assessment (GA) offers the possible advantage of guiding therapeutic decisions and may prove essential when accounting for treatment compatibility, drug-induced side effects, and mortality.2,14,17,18 This has been acknowledged as relevant, along with competent clinical judgment, apart from those risks generated through the myeloma itself. GA tools have been postulated to be valuable in different cancers. However, most are not myeloma-specific (such as the Charlson Comorbidity Index [CCI], the Hematopoietic Cell Transplantation-Comorbidity Index [HCT-CI] or the Kaplan-Feinstein [KF] Index). It has not yet been determined which of these is most applicable to myeloma. The International Myeloma Working Group (IMWG) demonstrated, in a pooled analysis of 869 newly diagnosed MM patients, that their IMWG score defined fit, intermediate-fit and frail patients and predicted the risk of mortality. This IMWG score combines age, Activities of Daily Living (ADL), Instrumental Activities of Daily Living (IADL) and CCI. The authors proposed the IMWG score as an additional tool for clinical evaluation, cross-comparison of clinical trials and for the measurement of frailty in designing future trials.4 Indeed, prospective randomized studies constitute a good basis for the development of prognostic scores, since they meet all requirements postulated to be important.2,6-8 However, they bear the additional challenge that patients therein are selected according to strict inclusion criteria. Therefore, an internal and external validation of postulated prognostic scores in unselected patient cohorts is necessary.7,8,10,19 Since the IMWG score was tested, but not validated, the authors encouraged others to substantiate their findings.4,6 In particular, "real world" patients were urged to be assessed, since the IMWG data was based on clinical trial patients who were stringently treated within trial protocols, where the frailest patients were excluded.6 Thus, it was of relevance to assess the external validity of the IMWG score in order to obtain confirmation in "real-world" patients from population-based registries and prospective analyses.6 Splitting the IMWG cohort by the investigators into a test and validation cohort would have resulted in an internal validation, again bearing the limitation of the results being solely based on clinical trial participants.19 Herein, we prospectively and carefully assessed the IMWG score's impact on clinical outcome, and are the first who have thoroughly validated the IMWG baseline GA in a well-characterized external cohort. Since our prior analyses demonstrated that multivariate risk factors in MM haematologica | 2016; 101(9)

patients include impaired renal function, lung function and Karnofsky performance status (KPS),2,7,8,10,20 and that with a revised Myeloma Comorbidity Index (R-MCI), the inclusion of frailty, age and cytogenetics improves its prediction of fit, intermediate-fit and frail patients,21 we included the R-MCI, CCI, HCT-CI and KF Index in this analysis. Our intention was validation, as diligently performed herein, and not the improvement of the IMWG score.

Methods Patient population and study design In analogy to the IMWG analysis,4 we performed a baseline GA in 125 consecutive patients with MM at the time of initial diagnosis and first presentation at our center. Patients received standard antimyeloma treatment according to the institutional MM pathway and current recommendations.2,22 As seen with the IMWG cohort (a.] EMN-01 trial: Rd vs. MPR or CPR, b.] 26866138MMY2069 trial: VP vs. VCP or VMP, and c.] IST-CAR506 trial: carfilzomib (Cd)), therapy regimens differed, but were those of current antimyeloma treatment: induction mostly consisted of VCD (bortezomib, cyclophosphamide, dexamethasone): patients ineligible for autologous stem cell transplantation (ASCT) received 9 cycles of VCD, and for medically fit patients up to the age of 70 years old, ASCT with VCD induction was performed. If feasible, patients were included in prospective multicenter trials such as the Clarion study, where instead of VCD, patients were randomized to VMP (bortezomib, melphalan, prednisone) or carfilzomib plus MP (melphalan, prednisone), or in the DSMM XII, XIII, XIV trials (n=49, 40%). Briefly, patients in the DSMM XII and XIV trials received RAD (lenalidomide, adriamycin, dexamethasone)23,24 induction, or RAD vs. VRD (bortezomib, lenalidomide, dexamethasone),23,25 respectively. Patients in the DSMM XIII trial were randomized to lenalidomide and dexamethasone (Rd) vs. Rd with ASCT.26 The diversity of different induction regimens in the IMWG cohort vs. ours were seven vs. six, respectively. As treatment differed in both the IMWG and our cohort, both analyses adjusted their univariate and multivariate models for known prognostic factors (ISS, cytogenetics and therapy). The analysis was carried out according to the guidelines of the Declaration of Helsinki and Good Clinical Practice. All patients gave their written informed consent for institutional-initiated research studies and analyses of clinical outcome studies conforming to the institutional review board guidelines. The study was registered as follows: (clinicaltrials.gov identifier 00003868). The primary objectives of this analysis were to recapitulate the IMWG score in our MM cohort, to assess additional GA tools to predict fit vs. frail patients and how these geriatric parameters predict overall survival (OS). The secondary objectives included the impact of the IMWG score as compared to the R-MCI, CCI, HCTCI and KF Index, and to assess their value for OS and progressionfree survival (PFS).

Assessment The GA consisted of 6 tools: the Katz ADL, the Lawton IADL, CCI, HCT-CI, KF Index and R-MCI as described.4,9,18,21,22 The comorbidities assessed in the R-MCI are depicted in the Online Supplementary Table S1, and the CCI, HCT-CI, KF Index and RMCI in the Online Supplementary Table S2. Online Supplementary Table S1 defines 13 comorbidity factors as mildly, moderately or severely impaired based on the Common Terminology Criteria for Adverse Events (CTCAE) 4.0 and included: renal impairment, 1111

M. Engelhardt et al.

lung and KPS impairment, cardiac, liver or gastrointestinal disease, disability, frailty, infection, thromboembolic events, peripheral neuropathy, pain, and secondary malignancies. In addition, age and cytogenetics were assessed: del(17p13), del(13q14), t(4;14), t(14;16); t(14;20), hypodiploidy, c-Myc and chromosome 1 aberrations were defined as unfavorable, and t(11;14), hyperdiploidy and a normal karyotype as favorable cytogenetics. Genetic abnormalities were detected by fluorescence in situ hybridization (FISH). Renal function was determined via estimated glomerular filtration rate ([eGFR] by MDRD), and lung disease via a lung function test. Pulmonary obstruction and/or restriction were distinguished with the aid of parameters such as forced expiratory volume in one second (FEV1), the Tiffeneau-Pinelli

index (FEV1/FVC) and total lung capacity (TLC). Respiratory insufficiency was detected through oxygen and carbon dioxide levels in arterial blood gas analysis. Pulmonary obstruction was graded through the impairment of the FEV1: a FEV1 of ≥80% was scored as mild, <80-50% as moderate, and <50% as severe. The KPS was defined as normal (100%), mildly (90%), moderately (80%) or more substantially impaired (≤70%). Frailty and disability were assessed in order to obtain a more precise determination of the physical condition of patients. The Fried definition was utilized for frailty: this takes into account the added presence of weakness, poor endurance, low physical activity, and slow gait speed.27,28 Patient characteristics included age, ISS and treatment (Table 1).

Table 1. Baseline patient characteristics of University Clinic Freiburg (UKF) and IMWG cohorts.

UKF cohort (n=125) % of patients Median (IQR/range) Age, years ≤65 66-74 ≥75 ≥80 Creatinine, mg/dl <2 ≥2 Missing ECOG PS 0 1 2 3 ISS I II III Chromosomal aberrations Favorable Unfavorable Missing ADL >4 ≤4 IADL >5 ≤5 CCI <2 ≥2 HCT-CI <2 ≥2 KF <2 ≥2 R-MCI 0-3 4-6 7-9 Therapy Lenalidomide-containing regimens PI-containing regimens

IMWG cohort (n=869) % of patients Median (IQR/range)

63 (56-71) 59 26 15 3

74 (70-78) 2 52 46 19

1 (0.80-1.40)

0.98 (0.80-1.22)

85 15 0

92 5 3

22 50 26 2

30 46 19 2

28 34 38

28 42 31

51 32 17

38 24 17* 4 (4-5)

48 52

6 (5-6) 86 14

8 (6.4-8) 85 15

8 (6-8) 82 18

2 (1-3) 35 65

0 (0-1) 83 17

2 (0-3)

n.d.

1 (1-2)

n.d.

5 (4-6)

n.d.

49 51 62 38 22 65 13 68 32

76 24

UKF: University Clinic Freiburg = medical center cohort of patients of the UKF site; IMWG: International Myeloma Working Group4; IQR, interquartile range. Unfavorable defined as t(4;14) or t(14;16) or del(17p13); * 21% still missing (∑ 79%); ECOG PS: Eastern Cooperative Oncology Group Performance Status; ADL: Activities of Daily Living; IADL: Instrumental Activities of Daily Living; CCI: Charlson Comorbidity Index, scored without lymphoma/myeloma; ISS: International Staging System; HCT-CI: Hematopoietic Cell Transplantation-Comorbidity Index; KF: Kaplan-Feinstein Comorbidity Index; R-MCI: revised Myeloma Comorbidity Index; n.d.: not done; PI: proteasome inhibitor.

1112

haematologica | 2016; 101(9)

Geriatric assessment validation in myeloma

Statistical analysis Data were analyzed using SAS 9.2 (SAS Institute Inc., NC, USA). OS was calculated from the date of first presentation at our center until the date of death from any cause, while PFS was calculated from the date of first presentation until the date of progression, relapse or death from any cause. Observations, where the event of interest did not occur, were censored at the time the patient was last seen alive/without a documented event, or at the latest on June 1st, 2015. OS and PFS rates were estimated using the Kaplan-Meier method and compared using the log-rank test. The IMWG score was assessed and compared to the R-MCI, HCT-CI, CCI and KF Index, evaluating the prognostic role on OS in our cohort with Cox regression models (Table 2). Results were presented as estimated hazard ratios (HRs) with two-sided 95% confidence intervals (CI) and corresponding P-values (Table 2). Cox regression models obtained and displayed in the analysis of the IMWG data were repeated using our data (Tables 2-4) in order to compare the IMWG score and R-MCI, as well as other internationally renowned, but MM unspecific comorbidity scores, such as the CCI, HCT-CI and KF Index (Tables 3 and 4). Multivariate risks, on which the IMWG score is based,4 were used to score fit, intermediate-fit and frail patients (Table 2). This IMWG score was also applied in order to compare fit vs. frail patients defined by the use of the R-MCI, CCI, HCT-CI and KF Index (Table 3), and via univariate and multivariate models evaluating the prognostic value of the scores, alone and adjusted for known prognostic factors (ISS, cytogenetics, therapy; Table 4): 125 patients with 28 OS events (deaths) and 64 PFS events (death or disease progression) were included in the analyses. Our main

results relied on univariate Cox models, namely the presentation of OS and PFS comparisons according to the fitness scores of patients (Table 4), and on multivariate analyses to adjust for 3 additional known prognostic factors, specifically ISS, chromosomal abnormalities and therapy. (Table 4).

Results Patient characteristics The analysis included 125 consecutive, prospectively assessed MM patients. The median follow-up was 28 months (interquartile range [IQR] 22-33). The median age was 63 years of age, 26% of patients were 66-74 years old and 15% older than 75 years of age (Table 1), which is typical for tertiary centres.7,8,10,21,22,29 This was in contrast to the IMWG cohort,4 where 46% of patients were older than 75 years of age (Table 1), albeit various other patient characteristics were comparable, e.g. renal function in both showed a median creatinine of 1mg/dl, the Eastern Cooperative Oncology Group performance status (ECOG PS) 0-1 was similarly distributed with 72% and 76%, respectively, patients had mostly ISS II/III stages in 72% and 73%, respectively, and the frequencies of unfavorable, favorable and missing chromosomal aberrations appeared similar. Moreover, the median IADL score in both cohorts was uncompromised at 8 (Table 1). Since the IMWG cohort consisted exclusively of clinical trial patients4 and ours of consecutive "real world" patients,

Table 2. Final Cox regression model of University Clinic Freiburg (UKF) and IMWG cohorts.

Age, years ≤75 76-80 >80 ISS I II III Chromosomal aberrations Favorable Unfavorable Missing ADL >4 ≤4 IADL >5 ≤5 CCI <2 ≥2 Therapy PI-containing regimens Lenalidomide-containing regimens

HR (95% CI)

UKF cohort (n=125) P

1 1.72 (0.53-5.62)

0.369

1 0.80 (0.25-2.54) 1.30 (0.50-3.77)

HR (95% CI)

IMWG cohort (n=869) P

Score*

1 1.13 (0,76-1,69) 2.40 (1.56-3.71)

0.549 <0.001

0 1 2

0.704 0.628

1 2.37 (1.38-4.09) 3.21 (1.85-5.58)

0.002 <0.001

1 2.51 (1.11-5.69) 0.51 (0.10-2.50)

0.027 0.403

1 1.79 (1.23-2.60) 1.13 (0.69-1.83)

0.002 0.036

1 2.37 (1.05-5.35)

0.039

1 1.67 (1.08-2.56)

0.020

0 1

1 5.65 (1.99-16.01)

0.001

1 1.43 (0.96-2.14)

0.078

0 1

1 3.59 (1.17-11.03)

0.026

1 1.37 (0.92-2.05)

0.125

0 1

1 0.56 (0.24-1.28)

0.165

1 0.74 (0.5.-1.11)

0.142

UKF: University Clinic Freiburg = medical center cohort of patients of the UKF site; IMWG: International Myeloma Working Group. Unfavorable defined as t(4;14) or t(14;16) or del(17p13); CI: confidence interval; HR: hazard ratio; ADL: Activities of Daily Living; IADL: Instrumental Activities of Daily Living; CCI: Charlson Comorbidity Index, scored without lymphoma/myeloma and age; ISS: International staging System; PI: proteasome inhibitor. *The IMWG score was based on final Cox regression model results as described4; we therefore used this IMWG score likewise with identical weights.

haematologica | 2016; 101(9)

1113

M. Engelhardt et al.

there were some differences: the number of patients with substantial renal impairment (creatinine >2 mg/dl) was 15% in our cohort (IMWG cohort: 5%), our patients showed an ECOG PS of 2-3 in 28% (IMWG: 21%), ISS III frequencies were higher at 38% (IMWG 31%), and unfavorable cytogenetics in 32% of our patients were higher as compared to 24% in the IMWG cohort. Unfavorable, favorable and missing cytogenetics in the IMWG cohort were 24%, 38% and 17%, respectively, vs. 32%, 51% and 17% in ours, respectively, suggesting with the IMWG cytogenetic data in 79% of patients that another 21% were missing (17+21%=38%). Supporting the characteristics of our "real world" patients vs. the IMWG cohort, our median ADL

score was lower with 4 vs. 6, respectively, and the CCI was higher with 2 vs. 0, respectively (Table 1). In agreement with these findings, the median HCT-CI, KF Index and RMCI in our patients - not objectives of the IMWG analysis - were 2, 1 and 5, respectively, and thus reflected a typical, moderately impaired patient cohort.2,7,8,10,22,29

Identification of prognostic variables in the Cox regression model The impact of advanced age, functional decline on ADL and IADL, ISS, cytogenetics, therapy and the presence of comorbidities leading to a worsening of OS was investigated in a multivariate Cox regression model (Table 2). The

Progression-free survival

P=0.0956

Progression-free survival

P<0.0001

P=0.0331

P=0.7858

Progression-free survival

P=0.3822

1114

Figure 1. Kaplan-Meier estimates for progression-free survival (PFS) (n=125) according to different comorbidity scores [P-values; log-rank test]. (A) PFS in our patients divided into fit, intermediate-fit and frail patients showed group differences between both fitter patient groups and frail patients using the IMWG score. (B) PFS according to the R-MCI revealed better group distinctions between fit, intermediate-fit and frail patents. (C-E) PFS according to CCI risk groups again showed significant difference, whereas these were undetectable with the use of both HCT-CI (D) and Kaplan-Feinstein (E).

haematologica | 2016; 101(9)

Geriatric assessment validation in myeloma

final prognostic model obtained after backward selection in the IMWG data was applied to ours. The results were compared to the IMWG results. Age <80 years showed no significant impact on OS (HR 1.72; P=0.369), which is in agreement with the IMWG score,4 whereas unfavorable cytogenetics (HR 2.51; P=0.027), ADL ≤4 (HR 2.37; P=0.039), IADL ≤5 (HR 5.65; P=0.001) and CCI ≥2 (HR 3.59; P=0.026) significantly reduced OS, the latter three being in accordance with IMWG results (Table 2). Both IADL and CCI in the IMWG cohort increased the HR for impaired OS to 1.43 and 1.37, respectively, but failed to reach significance.4 In our patients, the IADL and CCI revealed a higher HR and reached significance for the CCI, which was likely related to our median CCI of 2 as opposed to 0 in the IMWG trial cohort. Therefore, more CCI-relevant comorbidities were present in our cohort vs. those within the IMWG study4 (Tables 1 and 2). Using the 4 IMWG risk factors: age, impaired ADL, IADL and CCI to define 3 risk groups, patients were stratified into fit (score=0), intermediate-fit (score=1) and frail (score ≥2) patients, displaying substantially different OS and PFS (Tables 3 and 4).

Comorbidity scores and OS and PFS at 3 years for both the University Clinic Freiburg (UKF) and IMWG cohorts According to the proposed IMWG score that determined fit, intermediate-fit and frail patients with an additive total score of 0, 1 and 2, respectively,4 we had fewer fit (18% vs. 39%), similar intermediate-fit (34% vs. 31%) and more frail patients (48% vs. 30%) than in the IMWG cohort (Table 3): similar to the IMWG data, our 3-year OS was 91% for fit, 77% for intermediate-fit (HR 1.77; 95% CI 0.36-8.75; P=0.487) and 47% for frail patients (HR 5.80; 95% CI 1.3524.96; P=0.018; Tables 3 and 4). In the multivariate analysis we also confirmed that, when adjusted for staging and the treatment administered, frailty profiles and comorbidity scores were associated with shorter OS (Table 4). By applying the IMWG score, the 3-year PFS in our cohort was 43% for fit, 25% for intermediate-fit (HR 1.19; 95% CI 0.56-2.55; P=0.648) and 22% for frail patients (HR

1.90; 95% CI 0.94-3.86; P=0.075; Tables 3 and 4). PFS for fit patients was comparable to the IMWG data, albeit lower for intermediate-fit and frail patients, which was likely related to more patients with comorbidities, higher CCI and lower ADL in our cohort (Tables 1 and 3). Using the 4 other comorbidity scores, namely R-MCI, CCI, HCT-CI and KF Index, also allowed division of fit and frail patients based on previously proposed cutoffs, with substantially different OS and PFS (Table 3). Of note, the proportion of patients via elevated HCT-CI ≥2, KF Index ≥2, CCI ≥2 and R-MCI ≥4, differed substantially, with 38%, 51%, 65% and 78%, respectively (Table 3). Nevertheless, all 4 scores revealed PFS and OS group differences between fit and frail patients, the most pronounced being observed with the use of the IMWG, CCI and R-MCI (Figures 1 and 2).

Univariate and multivariate analysis of the impact of frailty on OS and PFS for the University Clinic Freiburg (UKF) and IMWG cohorts In Table 3, 3-year OS and PFS rates are presented with accompanying 95% CI. These comparisons refer to one single time point, where CIs are large due to smaller observation numbers at the end of the observation period. In Table 3, we intended to provide additional descriptive information to Figures 1 and 2, and to provide comparability to 3year OS and PFS rates reported by the IMWG.4 The proper way to compare groups with respect to OS and PFS over the complete observation period is via Cox models, as presented in Table 4. HRs via multivariate analysis were associated with shorter OS and PFS using the IMWG score, as well as when R-MCI, CCI, HCT-CI and the KF Index were applied. These frailty scores reached significance for OS with the use of the IMWG score for frail vs. both fit and intermediate-fit patients, although the two fittest patient groups clustered together. Similarly, significant OS differences between fit and frail patients were notable with the use of both the R-MCI and CCI. For PFS, the IMWG score showed differences for frail patients vs. the two fitter

Table 3. Comorbidity scores and related rates of OS and PFS at 3 years of University Clinic Freiburg (UKF) and IMWG cohort.

Score IMWG score 0 1 >2 CCI <2 ≥2 HCT-CI <2 ≥2 Kaplan-Feinstein (KF) <2 ≥2 R-MCI 0-3 4-6 7-9

UKF cohort (n=125) No. of Patients % (95% CI) Patient status (%) OS PFS

IMWG cohort (n=869) No. of Patients % (95% CI) (%) OS PFS

Fit Intermediate-fit Frail

22 (18) 43 (34) 60 (48)

91 (78-100) 77 (55-95) 47 (26-67)

43 (17-70) 25 (1-49) 22 (2-43)

340 (39) 269 (31) 260 (30)

84 (78-89) 76 (67-82) 57 (45-68)

48 (41-56) 41 (32-49) 33 (25-41)

Fit Frail

44 (35) 81 (65)

89 (79-99) 52 (34-69)

44 (24-63) 13 (0-33)

n.d.

Fit Frail

61 (62) 64 (38)

70 (46-93) 60 (44-76)

34 (18-51) 27 (9-45)

n.d.

Fit Frail

78 (49) 47 (51)

68 (51-86) 62 (44-80)

38 (24-53) 13 (0-33)

n.d.

n.d

Fit Intermediate-fit Frail

27 (22) 82 (66) 16 (12)

89 (75-100) 68 (54-82) 20 (0-51)

25 (9-41) -

n.d.

UKF: University Clinic Freiburg = medical center cohort of patients of the UKF site; IMWG, International Myeloma Working Group; CI: confidence interval; CCI: Charlson Comorbidity Index; HCT-CI: Hematopoietic Cell Transplantation-Comorbidity Index; R-MCI: revised Myeloma Comorbidity Index; n.d.: not done; OS: overall survival; PFS: progression-free survival.

haematologica | 2016; 101(9)

1115

M. Engelhardt et al.

patient groups, however this difference did not reach significance in our cohort. Significant PFS differences via CCI (<2 vs. ≥2) and R-MCI (≤3 vs. ≥4) were also observed (Table 4, Figures 1 and 2). These frailty scores continued to be associated with shorter OS and PFS when adjusted for staging and the treatment administered, again with the most pronounced group differences being those between frail vs. fitter patients when these were determined using the IMWG score, CCI and R-MCI (Table 4).

yet the overall prognosis depends on a variety of diseaseand host-related risks.2,18,21,22,29 As individualized treatment aims to balance therapy options against toxic side effects, to preserve quality of life and to further improve survival, there is an obvious rationale for incorporating risk assessment tools into the management of predominantly older patients. This supports the inclusion of additional objective, quantifiable and reproducible information on individual patients beyond the clinical judgment of physicians. Furthermore, this seems relevant in order to avoid applying less effective therapies to older yet fit patients.2,7,16 Therefore, cancer experts have recognized that the consideration of age alone is insufficient for choosing adequate therapy strategies. Nevertheless, age cutoffs continue to exist, and unfortunately older patients are often excluded from clinical trials.7,9,10,18,22,24,30–33

Discussion MM management strategies continue to evolve, and in the past few decades survival has improved significantly,

P=0.0021

P=0.0054

P=0.0059

P=0.0991

P=0.2754

1116

Figure 2. Kaplan-Meier estimates for overall survival (OS) (n=125) according to different comorbidity scores [P-values; log-rank test]. (A) Overall survival (OS) according to the IMWG score assessed with our MM patient cohort showed improved OS for patients classified as fit (low-risk) or intermediate-fit (intermediate-risk) vs. those determined frail (high-risk)4. OS for the fit and intermediate-fit group was slightly, but insignificantly different. (B) OS according to the R-MCI showed improved survival of fit (low-risk) vs. intermediate-fit or frail (high-risk) patients. Differences between these 3 groups of fit, intermediate-fit and frail patients were more distinct compared to the IMWG score.4 (C-E) OS according to the CCI,12 dividing patients into two groups with CCI <2 vs. CCI ≥2 comorbidities, revealed significant OS differences, which were more pronounced than fit vs. frail patients assessed via HCT-CI (D) or KaplanFeinstein (E).

haematologica | 2016; 101(9)

Geriatric assessment validation in myeloma

Our prior test and validation analyses have demonstrated the significant influence of impaired organ function, such as renal function and lung function as well as KPS on the PFS and OS of MM patients, thus implementing these factors in a comorbidity tool, named MCI.7,8,10 The more refined use of this MCI, tested and validated in a much larger independent cohort of 801 patients, confirmed the importance of renal function, lung function and KPS, as well as age, frailty and cytogenetics within a refined MCI (R-MCI).21 Other studies have confirmed the relevance of assessing frailty, quality of life and physical activity.18,32,34,35 The assessment of organ function, such as renal and lung impairment, has shown to influence survival rates, treatment toxicity and early death.8,16,36 Moreover, tumor genetics have been reported to relevantly influence the clinical heterogeneity of MM.1,11,14,25,37,38 Although cytogenetics are important for risk appraisal, we and others have demonstrated that physical condition and organ function are likewise crucial.7,8,10,18 We herein assessed the IMWG score4 in an independent external validation cohort. Patient characteristics were com-

parable, with a median creatinine of 1mg/dl, an ECOG PS of 0-1 which was similarly distributed, primarily ISS II/III stages, and an uncompromised median IADL score of 8. However, there were differences: more of our patients had unfavorable cytogenetics, lower ADL and higher CCI, which is in agreement with the IMWG cohort consisting of clinical trial patients only with strict inclusion criteria. Importantly, we confirmed that an age of <80 years may not substantially increase the risk for MM patients, demonstrating that suitable comorbidity tools and a precisely performed GA are helpful. Unfavorable cytogenetics increase the HR for survival (HR 2.51; 95% CI 1.11-5.69), in addition, functional impairment as determined via ADL, IADL or CCI increased the HRs in our cohort, showing higher HRs in all 4 compared to the IMWG cohort (Table 2). Our cohort was likewise split into IMWG patient groups with 0, 1 and >2 risk factors showing comparably different OS and PFS groups of fit vs. intermediate-fit and frail patients; this was, however, also possible to perform via RMCI, CCI, HCT-CI and the KF Index (Table 3). Crude and

Table 4. Univariate and multivariate analysis of the impact of frailty on OS and PFS of University Clinic Freiburg (UKF) and IMWG cohort.

Score

Patient status

Crude IMWG score 0 Fit 1 Intermediate-fit >2 Frail CCI <2 Fit ≥2 Frail HCT-CI <2 Fit ≥2 Frail Kaplan-Feinstein (KF) <2 Fit ≥2 Frail R-MCI 0-3 Fit 4-6 Intermediate-fit 7-9 Frail Adjusted* IMWG score* 0 Fit 1 Intermediate-fit >2 Frail CCI* <2 Fit ≥2 Frail HCT-CI* <2 Fit ≥2 Frail Kaplan-Feinstein (KF)* <2 Fit ≥2 Frail R-MCI* 0-3 Fit 4-6 Intermediate-fit 7-9 Frail

UKF cohort (n=125) OS PFS HR (95% CI) P HR (95% CI)

OS HR (95% CI)

IMWG cohort (n=869) PFS P HR (95% CI) P

1 1.77 (0.36-8.75) 5.80 (1.35-24.96)

0.487 0.018

1 1.19 (0.56-2.55) 1.90 (0.94-3.86)

1 0.648 1.61 (1.02-2.56) 0.075 3.57 (2.37-5.39)

0.042 0.001

1 1.18 (0.91-1.53) 1.68 (1.31-2.15)

.211 .001

1 3.98 (1.38-11.49)

0.011

1 1.79 (1.04-3.11)

0.036

n.d.

1 1.89 (0.89-4.12)

0.105

1 1.07 (0.65-1.76)

0.786

n.d.

n.d

1 1.50 (0.72-3.18)

0.279

1 1.25 (0.76-2.05)

0.384

n.d.

1 3.32 (0.77-14.29) 8.77 (1.82-42.32)

0.107 0.007

1 1.76 (0.86-3.62) 5.90 (2.42-14.39)

0.123 0.0001

n.d.

1 1.45 (0.28-7.64) 6.06 (1.35-27.25)

0.661 0.019

1 1.22 (0.55-2.71) 2.15 (1.03-4.49)

1 0.631 1.37 (0.86-2.18) 0.042 2.88 (1.88-4.40)

0.181 0.001

1 1.08 (0.83-1.40) 1.48 (1.15-1.92)

0.583 0.003

1 3.62 (1.19-10.98)

0.023

1 1.94 (1.09-3.45)

0.024

n.d.

1 1.76 (0.76-4.08)

0.185

1 1.08 (0.64-1.83)

0.779

n.d.

1 1.20 (0.53-2.76)

0.659

1 1.25 (0.72-2.17)

0.428

n.d.

1 3.13 (0.70-13.92) 8.34 (1.69-41.17)

0.134 0.009

1 1.83 (0.87-3.85) 6.26 (2.48-15.80)

0.114 0.0001

n.d.

UKF: University Clinic Freiburg = medical center cohort of patients of the UKF site; IMWG, International Myeloma Working Group, *adjusted for ISS, cytogenetics, therapy; CI: confidence interval; R-MCI: revised Myeloma Comorbidity Index; HCT-CI: Hematopoietic Cell Transplantation-Comorbidity Index; n.d., not done; HR: hazard ratio; OS: overall survival; PFS: progression-free survival.

haematologica | 2016; 101(9)

1117

M. Engelhardt et al.

adjusted HRs of the IMWG score for OS and PFS proved to separate frail, intermediate-fit and fit patients, the latter two of which grouped together. This was confirmed when assessing the IMWG score in a multivariate model adjusting for known prognostic factors (Table 4). Both the CCI and RMCI, divided into fit vs. frail, showed substantially increased HRs for OS and PFS, whereas less pronounced differences for fit vs. frail patients were obvious with the use of both the KF Index and HCT-CI. Therefore, our detailed comparison of both the crude and adjusted IMWG score with others, suggested some to be of particular value in MM, such as the IMWG score, CCI and R-MCI, whereas others, such as the KF Index and HCT-CI were of lesser significance. The CCI in particular has been tested in several clinical settings and has shown its usefulness.9,12,18,34 However, scoring of the CCI has been modified2,4,7-10,18,39 since it is not specific for MM. Why the IMWG has chosen the CCI to complement their IMWG score is therefore a possible choice, albeit the CCI has been validated in diabetes and its value has also been questioned, all the more as the median CCI in the IMWG cohort was extremely low, with 0. Our results, that all comorbidity tools: IMWG score, R-MCI, CCI, HCT-CI and the KF Index divide patients with largely different OS and PFS, even though only the IMWG score, R-MCI and CCI reached significance, are therefore of value. Of note, both the IMWG score and R-MCI performed comparably well, as shown in other scores, where 2 geriatric screening tools (the G8 questionnaire and the Flemish version of the Triage Risk Screening Tool [fTRST]) were compared.35 This is the first validation study that externally confirmed the IMWG score and other relevant comorbidity tools in real life patients, although our cohort was smaller, it nevertheless provides useful information for a validation analysis. A common procedure is to divide a sample number into 2/3 for score construction and 1/3 for internal validation. In the IMWG analysis,4 smaller subgroup analyses were performed, which also resulted in reduced sample sizes. Therefore, we consider our data valuable, supporting the usefulness of the IMWG score (and others) when applied to different populations. Albeit our patients were typical for a university center, they were younger than the IMWG cohort.4 We have previously shown that stage and age migration may occur29 and that older patients are increasingly seen in university centers: in a previous analysis of 816 MM patients, 3-fold increases for >70-year old patients were observed.29 In this analysis we confirmed that the HR for 60 to 70 and >70-year old MM patients increased from 1.72 to 3.46, respectively.29 The reason why the IMWG cohort did not see an age risk in 76 to 80-year old patients was most likely because: a) age is a lesser risk factor than initially presumed - making simple comorbidity assessments as performed with experience as in our centers the more important, and b) because the IMWG cohort included only trial patients that were much fitter, and age is less relevant in those that fulfill all eligibility criteria. Another criticism might be that antimyeloma treatment differed from those used in the IMWG study.4 Since easily assessable risk scores are important to apply independent of treatment, and previous analyses have shown that heterogeneous therapies are not surrogates for comorbidity risks, we consider our analysis to be of equal importance to prior GA analyses, including that of the IMWG. This is even more evident

1118

since we validated and complemented their findings, and all risk scores were assessed as crude and adjusted scores (Table 4).4 The strengths of this analysis were the accurate and prospective assessment of the physical condition of patients with no restriction of validity and information loss based on multicenter data entries.6 Moreover, our cohort reflected typical day to day patients, since ADL and CCI were affected compared to the IMWG cohort, whereas in both the IADL was uncompromised with a median of 8. This suggests this daily activities score to be of lesser importance, and questions the necessity to use both ADL and IADL in a combined risk score. Our thoroughly performed validation of the IMWG score and others, including the R-MCI, was performed within a structured prospective GA and by an experienced group who has been doing these assessments for years.2,7,10,20,21 Moreover, we applied all current state-ofthe-art statistics with our renowned statistical team with the important aim to validate easy to assess risk scores that function to improve MM care.40 Currently, the IMWG score consists of: 1. age (3 age groups), 2. ADL (6 self-care tasks), 3. IADL (8 house-hold tasks), and 4. CCI (18 factors and maximum points of 33, plus 1 point per decade of aging from the age of 50). As a sum risk assessment, this constitutes 3+10+8+18=39 risks, instead of 6 within the R-MCI. Thus, within the IMWG score (including age and CCI), age is scored twice; suggesting by the direct comparison of both IMWG score and RMCI that the former is more challenging than the latter. This can be verified, if the R-MCI is calculated via the website which we designed: www.myelomacomorbidityindex.org. Nevertheless, our intention was pure validation of the IMWG score, which we performed fastidiously. As with the IMWG score, treatment was not modified according to this score. A next step includes prospective randomized clinical studies to design therapeutic approaches with the help of a GA algorithm.2,14,20,41,42 We conclude that both IMWG score and R-MCI are useful instruments in older myeloma patients for identifying those with geriatric risks. The publication of our validation analysis should attract the attention of cancer experts and help in the care of MM patients, the latter being an aim that all of us are enthusiastically persisting in daily. Acknowledgments The authors thank Prof. Dr. Hermann Einsele, University of WĂźrzburg, Germany, Prof. Dr. Pieter Sonneveld, Erasmus MC Rotterdam, The Netherlands, Prof. Dr. Christian Straka, SchĂśn Clinic Starnberger See, Germany, Prof. Dr. Keith Stewart, Mayo Clinic Arizona & Rochester, USA, Prof. Dr. Torben Plesner, Center Lillebaelt University of Southern Denmark, Denmark and Prof. Dr. Justus Duyster, University of Freiburg, Germany, for their significant support, fruitful discussion and valuable suggestions. We are also highly obliged to the three anonymous reviewers for their insightful comments that have inspired us to further improve the manuscript and Dr. Marie Follo for proofreading the manuscript, to Hans Schall and Judith Urban for diligent data management using our MM database, and to Dr. Patrique Wolfrum for IT support. Funding This work is supported by the Deutsche Krebshilfe (grants 1095969 and 111424 [to ME and RW]).

haematologica | 2016; 101(9)

Geriatric assessment validation in myeloma

References 1. Lohr JG, Stojanov P, Carter SL, et al. Widespread genetic heterogeneity in multiple myeloma: implications for targeted therapy. Cancer Cell. 2014;25(1):91-101. 2. Engelhardt M, Terpos E, Kleber M, et al. European Myeloma Network recommendations on the evaluation and treatment of newly diagnosed patients with multiple myeloma. Haematologica. 2014; 99(2):232242. 3. Palumbo A, Bringhen S, Ludwig H, et al. Personalized therapy in multiple myeloma according to patient age and vulnerability: a report of the European Myeloma Network (EMN). Blood. 2011; 118(17):4519-4529. 4. Palumbo A, Bringhen S, Mateos M-V, et al. Geriatric assessment predicts survival and toxicities in elderly myeloma patients: an International Myeloma Working Group report. Blood. 2015;125(13):2068-2074. 5. Palumbo A, Avet-Loiseau H, Oliva S, et al. Revised International Staging System for Multiple Myeloma: A Report From International Myeloma Working Group. J Clin Oncol. 2015;33(26):2863-2869. 6. Mellqvist U-H. New prognostic tools for myeloma. Blood. 2015;125(13):2014-2015. 7. Kleber M, Ihorst G, Gross B, et al. Validation of the Freiburg Comorbidity Index in 466 multiple myeloma patients and combination with the international staging system are highly predictive for outcome. Clin Lymphoma Myeloma Leuk. 2013;13(5):541-551. 8. Kleber M, Ihorst G, Udi J, Koch B, Wäsch R, Engelhardt M. Prognostic risk factor evaluation in patients with relapsed or refractory multiple myeloma receiving lenalidomide treatment: analysis of renal function by eGFR and of additional comorbidities by comorbidity appraisal. Clin Lymphoma Myeloma Leuk. 2012;12(1):38-48. 9. Bringhen S, Mateos MV, Zweegman S, et al. Age and organ damage correlate with poor survival in myeloma patients: metaanalysis of 1435 individual patient data from 4 randomized trials. Haematologica. 2013;98(6):980-987. 10. Kleber M, Ihorst G, Terhorst M, et al. Comorbidity as a prognostic variable in multiple myeloma: comparative evaluation of common comorbidity scores and use of a novel MM-comorbidity score. Blood Cancer J. 2011;1(9):e35. 11. Moreau P, Cavo M, Sonneveld P, et al. Combination of international scoring system 3, high lactate dehydrogenase, and t(4;14) and/or del(17p) identifies patients with multiple myeloma (MM) treated with front-line autologous stem-cell transplantation at high risk of early MM progressionrelated death. J Clin Oncol. 2014; 32(20):2173-2180. 12. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383. 13. Sorror ML. How I assess comorbidities before hematopoietic cell transplantation. Blood. 2013;121(15):2854-63. 14. Palumbo A, Anderson K. Multiple myeloma. N Engl J Med. 2011;364(11):1046-1060. 15. Deschler B, Ihorst G, Platzbecker U, et al. Parameters detected by geriatric and quality of life assessment in 195 older patients with myelodysplastic syndromes and acute

haematologica | 2016; 101(9)

16.

17.

18. 19. 20.

21.

22.

23.

24. 25.

26.

27. 28.

29.

30.

myeloid leukemia are highly predictive for outcome. Haematologica. 2013; 98(2):208216. Offidani M, Corvatta L, Polloni C, et al. Assessment of vulnerability measures and their effect on survival in a real-life population of multiple myeloma patients registered at Marche Region Multiple Myeloma Registry. Clin Lymphoma Myeloma Leuk. 2012;12(6):423-432. Ganna A, Ingelsson E. 5 year mortality predictors in 498,103 UK Biobank participants: a prospective population-based study. Lancet Lond Engl. 2015;386(9993):533–540. Larocca A, Palumbo A. How I treat fragile myeloma patients. Blood. 2015; 126(19):2179-2185. Altman DG, Royston P. What do we mean by validating a prognostic model? Stat Med. 2000;19(4):453-473. Terpos E, Kleber M, Engelhardt M, et al. European Myeloma Network Guidelines for the Management of Multiple Myelomarelated Complications. Haematologica. 2015;100(10):1254-1266. Domm A-S, Hieke S, Ihorst G, et al. Importance and Determinants of Comorbidities, Functional Limitations and Multiple Myeloma (MM)-Specific Risk Factors: Further Development of an Improved and Weighted MM-Risk Score (Freiburg Comorbidity Index [FCI]). Blood. 2014;124(21):733-733. Engelhardt M, Ihorst G, Landgren O, et al. Large registry analysis to accurately define second malignancy rates and risks in a wellcharacterized cohort of 744 consecutive multiple myeloma patients followed-up for 25 years. Haematologica. 2015; 100(10):1340-1349. Knop S, Langer C, Engelhardt M, et al. Response to Lenalidomide, Doxorubicin and Dexamethasone (RAD) in Newly Diagnosed Multiple Myeloma Is Independent of Cytogenetic Risk and Retained after Double Stem Cell Transplant. Blood. 2014;124(21):177-177. Röllig C, Knop S, Bornhäuser M. Multiple myeloma. Lancet Lond Engl. 2015; 385(9983):2197-2208. Richardson PG, Bladé J. The comprehensive clinical management of multiple myeloma and related-plasma cell disorders. Expert Rev Hematol. 2014;7(1):1-3. Straka C, Schaefer-Eckart K, Bassermann F, et al. Lenalidomide with Low-Dose Dexamethasone (Rd) Continuously Versus Rd Induction, Tandem MEL140 with Autologous Transplantation and Lenalidomide Maintenance: Planned Interim Analysis of a Prospective Randomized Trial in Patients 60-75 Years of Age with Multiple Myeloma. Blood. 2014;124(21):3969-3969. Rodriguez-Mañas L, Fried LP. Frailty in the clinical scenario. Lancet Lond Engl. 2015;385(9968):e7-9. Xue Q-L, Walston JD, Fried LP, Beamer BA. Prediction of risk of falling, physical disability, and frailty by rate of decline in grip strength: the women’s health and aging study. Arch Intern Med. 2011;171(12):11191121. Hieke S, Kleber M, König C, Engelhardt M, Schumacher M. Conditional Survival: A Useful Concept to Provide Information on How Prognosis Evolves over Time. Clin Cancer Res. 2015;21(7):1530-1536. Cho H, Klabunde CN, Yabroff KR, et al. Comorbidity-adjusted life expectancy: a new tool to inform recommendations for

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

optimal screening strategies. Ann Intern Med. 2013;159(10):667-676. Wildes TM, Rosko A, Tuchman SA. Multiple myeloma in the older adult: better prospects, more challenges. J Clin Oncol. 2014;32(24):2531-2540. Ludwig H, Sonneveld P, Davies F, et al. European perspective on multiple myeloma treatment strategies in 2014. Oncologist. 2014;19(8):829-844. Auner HW, Szydlo R, Hoek J, et al. Trends in autologous hematopoietic cell transplantation for multiple myeloma in Europe: increased use and improved outcomes in elderly patients in recent years. Bone Marrow Transplant. 2015;50(2):209-215. Maes H, Delforge M. Optimizing quality of life in multiple myeloma patients: current options, challenges and recommendations. Expert Rev Hematol. 2015;8(3):355-366. Kenis C, Decoster L, Van Puyvelde K, et al. Performance of Two Geriatric Screening Tools in Older Patients With Cancer. J Clin Oncol. 2014;32(1):19-26. Augustson BM, Begum G, Dunn JA, et al. Early mortality after diagnosis of multiple myeloma: analysis of patients entered onto the United kingdom Medical Research Council trials between 1980 and 2002-Medical Research Council Adult Leukaemia Working Party. J Clin Oncol. 2005;23(36):9219-9226. Kuiper R, van Duin M, van Vliet MH, et al. Prediction of high- and low-risk multiple myeloma based on gene expression and the International Staging System. Blood. 2015; 126(17):1996-2004. van de Donk NWCJ, Sonneveld P. Diagnosis and risk stratification in multiple myeloma. Hematol Oncol Clin North Am. 2014;28(5):791-813. Zweegman S, Palumbo A, Bringhen S, Sonneveld P. Age and aging in blood disorders: multiple myeloma. Haematologica. 2014;99(7):1133–1137. Ihorst G, Waldschmidt J, Schumacher M, Wäsch R, Engelhardt M. Analysis of survival by tumor response: have we learnt any better? Ann Hematol. 2015; 94(9):1615-1616. Stairmand J, Signal L, Sarfati D, et al. Consideration of comorbidity in treatment decision making in multidisciplinary cancer team meetings: a systematic review. Ann Oncol. 2015;26(7):1325-1332. Decoster L, Van Puyvelde K, Mohile S, et al. Screening tools for multidimensional health problems warranting a geriatric assessment in older cancer patients: an update on SIOG recommendations†. Ann Oncol. 2015; 26(2):288-300. Kristinsson SY, Pfeiffer RM, Björkholm M, Schulman S, Landgren O. Thrombosis is associated with inferior survival in multiple myeloma. Haematologica. 2012;97(10): 1603-1607. Hasskarl J, Ihorst G, De Pasquale D, et al. Association of multiple myeloma with different neoplasms: systematic analysis in consecutive patients with myeloma. Leuk Lymphoma. 2011;52(2):247-259. ElSawy M, Storer BE, Pulsipher MA, et al. Multi-centre validation of the prognostic value of the haematopoietic cell transplantation- specific comorbidity index among recipient of allogeneic haematopoietic cell transplantation. Br J Haematol. 2015; 170(4):574-583. Extermann M. Measuring comorbidity in older cancer patients. Eur J Cancer. 2000; 36(4):453-471.

1119

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Stem Cell Transplantation

Ferrata Storti Foundation

Haematologica 2016 Volume 101(9):1120-1127

Outcomes of unrelated cord blood transplantation in patients with multiple myeloma: a survey on behalf of Eurocord, the Cord Blood Committee of Cellular Therapy and Immunobiology Working Party, and the Chronic Leukemia Working Party of the EBMT Annalisa Paviglianiti,1,2* Erick Xavier,1,2* Annalisa Ruggeri,1,2,3 Patrice Ceballos,4 Eric Deconinck,5 Jan J. Cornelissen,6 Stephanie Nguyen-Quoc,7 Natacha Maillard,8 Guillermo Sanz,9 Pierre-Simon Rohrlich,10 Laurent Garderet,3 Fernanda Volt,1,2 Vanderson Rocha,1,2,11 Nicolaus Kroeger,12 Eliane Gluckman,1,2 Nathalie Fegueux,4* and Mohamad Mohty3,13*

Eurocord, APHP, University Paris Diderot, Hôpital Saint Louis, Paris, France; 2Monacord, Centre Scientifique de Monaco, France; 3Service d’Hématologie et Thérapie Cellulaire, Hôpital Saint Antoine, AP-HP, Paris, France; 4Département d’Hématologie Clinique, CHU Lapeyronie, Montpellier, France; 5Department Hematology, CHU Besancon, France; 6 Department of Hematology, Erasmus MC-Daniel den Hoed Cancer Centre, Rotterdam, the Netherlands; 7Department Hematology, University Paris IV, Hôpital la PitiéSalpêtrière, Paris, France; 8Bone Marrow Transplant Unit Clinical Hematology, Hôpital La Miletrie, Poitiers, France; 9Servicio de Hematologia, Hospital Universitario y Politecnico La Fe, Valencia, Spain; 10Hematologie Clinique, CHU - Hôpital de l`ARCHET I, Nice, France; 11Department Hematology, Churchill Hospital, Oxford, UK; 12Dept. Stem Cell Transplantation, University Hospital of Hamburg, Germany; and 13INSERM, UMRs 938, Paris, France 1

*AP and EX, and NF and MM contributed equally to this work.

ABSTRACT

Correspondence: annalisa.paviglianiti@gmail.com

Received: November 13, 2015. Accepted: May 24, 2016. Pre-published: May 26, 2016. doi:10.3324/haematol.2015.138917

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/101/9/1120

1120

lthough allogeneic stem cell transplantation is not a standard therapy for multiple myeloma, some patients can benefit from this intense therapy. There are few reports on outcomes after umbilical cord blood transplantation in multiple myeloma, and investigation of this procedure is warranted. We retrospectively analyzed 95 patients, 85 with multiple myeloma and 10 with plasma cell leukemia, receiving single or double umbilical cord blood transplantation from 2001 to 2013. Median follow up was 41 months. The majority of patients received a reduced intensity conditioning. The cumulative incidence of neutrophil engraftment was 97%±3% at 60 days, and that of 100-day acute graft-versus-host disease grade II-IV was 41%±5%. Chronic graft-versus-host disease at two years was 22%±4%. Relapse and non-relapse mortality was 47%±5% and 29%±5% at three years, respectively. Three-year progression-free survival and overall survival were 24%±5% and 40%±5%, respectively. Anti-thymocyte globulin was associated with decreased incidence of acute graft-versus-host disease, higher non-relapse mortality, decreased overall and progressionfree survival. Patients with high cytogenetic risk had higher relapse, and worse overall and progression-free survival. In conclusion, umbilical cord blood transplantation is feasible for multiple myeloma patients. Introduction The current standard of care for patients with multiple myeloma (MM) is the use of drugs such as bortezomib, thalidomide and lenalidomide followed by autologous stem cell transplantation (ASCT).1,2 Although this combined treatment has markedly improved prognosis, disease recurrence remains high in MM patients. The Intergroupe Francophone du Myelome3 was the first to demonstrate, in a large randomized trial, the long-term benefits in survival of double ASCT in comparison haematologica | 2016; 101(9)

Cord blood transplantation in multiple myeloma

with single transplants. Since then, other randomized trials4 and registry studies5,6 have shown less recurrence and long-term disease control in patients treated with double ASCT.4,7 However, relapse remains the main reason for treatment failure after ASCT, due to the presence of nondetectable residual disease in the patient, the graft, or both.8 Conversely, allogeneic stem cell transplantation (alloSCT) is a potentially curative alternative, offering a tumor-free graft along with the benefit of a graft-versusMM (GvM) effect.9 Nevertheless, the role of alloSCT in MM patients is still controversial.10-12 Several studies have shown elevated rates of molecular remission after myeloablative conditioning regimen (MAC) alloSC.13 This regimen is characterized by high morbidity and mortality, especially in elderly patients with co-morbidities due to previous treatments, for whom less intensive conditioning regimens are preferable. The use of reduced intensity conditioning (RIC) extends the indication of hematopoietic stem cell transplantation to a higher number of patients who may benefit from an aggressive, but less toxic therapy than MAC, while maintaining the GvM effect. However, the efficacy of RIC on MM outcome remains uncertain. Different studies comparing tandem ASCT to alloSCT/RIC have given discordant results, with some of them failing to demonstrate the benefit of alloSCT/RIC.10,14,15 There is, therefore, a general hesitancy in recommending up-front alloSCT. Consequently, most of the time alloSCT is offered as a salvage therapy postASCT relapse or for refractory disease. Outcomes after matched unrelated donor (MUD) transplantations for MM have been reported to be similar to those with HLA identical siblings; however, outcomes after MUD appear to be associated with higher non-relapse mortality (NRM).16 Little is known of the use of other alternative donors, such as haploidentical or cord blood in patients with MM.17-19 We performed a registry-based study to evaluate risk factors and outcomes of patients undergoing umbilical cord blood transplantation (UCBT) with the aim of analyzing the role of this stem cell source in patients with plasma cell disorders.

Methods Study design, inclusion criteria and data collection This is a retrospective observational registry-based study using Eurocord/EBMT data. Patients over 18 years of age and diagnosed with MM or plasma cell leukemia (PCL) receiving single or double UCBT (dUCBT) between 2001 and 2013 were included. Exclusion criteria were: previous alloSCT, primary amyloidosis without MM, manipulated cord blood, intra-bone injection of cord blood cells or cord blood transplants associated with another stem cell source. All patients gave informed consent for research. The study was conducted in accordance with the Declaration of Helsinki. The Internal Review Board of Eurocord-EBMT approved the study.

End points and definitions The primary end point was progression-free survival (PFS) defined as time from UCBT to progression, relapse or death from any cause, whichever occurred first. Secondary end points were neutrophil and platelet recovery, acute and chronic graft-versushost disease (GvHD), NRM, relapse incidence (RI) and overall survival (OS). OS was defined as time from transplant to death from any cause. Neutrophil (PMN) engraftment was defined as the first haematologica | 2016; 101(9)

of three consecutive days with an absolute neutrophil count of 0.5x109/L or over, without evidence of autologous reconstitution. Platelet (PLT) engraftment was defined as the first date at which an unsupported platelet count of 20x109/L or over for seven consecutive days was achieved. MAC regimen was defined as a regimen containing total body irradiation (TBI) with a dose of more than 6 Gy or a dose of more than 8 mg/kg oral or more than 6.4 mg/kg intravenous busulfan or chemotherapy combination containing more than 10 mg/kg thiotepa. Response to treatment was defined according to standard criteria.20 Chemo-refractory myeloma was defined as progression or non-response within 60 days of last therapy. GvHD was evaluated based on standard criteria.21,22 For dUCBT, human leukocyte antigen (HLA) degree of matching was defined considering the UCB unit with the higher number of disparities with the recipient. High-risk cytogenetic abnormalities included at least one of the following: del17p, t(4;14) or t(14;16) performed by fluorescence in situ hybridization (FISH) or conventional metaphase cytogenetics, according to the policy of each center.

Statistical analysis The probabilities of PFS and OS were estimated using the Kaplan-Meier method and compared with the log rank test. In the case of no event, observations were censored at the time of last follow up. Cumulative incidence (CI) was calculated in a competing risk setting. Death without an event was treated as a competing risk to calculate probabilities of neutrophil and platelet engraftment, acute and chronic GvHD. Death without progression or relapse was considered as competing risk for RI and relapse was the competing event for NRM. P<0.05 was considered statistically significant. All variables found to have P<0.10 in the univariate analysis were included in a Cox model for PFS and OS, or in a Fine and Gray proportional hazard regression model for engraftment, GvHD, NRM and relapse. Analysis was performed with SPSS 19 and SPLUS software.

Results Patients' and transplant characteristics are summarized in Tables 1 and 2. A total of 95 patients with a median follow up of 41.3 (range 3.7-96) months met the inclusion criteria for the study. Median age at UCBT was 53.3 years (range 24.1-69.6) and median body weight was 70 kg (range 48-110). Median time from diagnosis to UCBT was 41.6 months (range 4.6-235.6). Diagnosis was MM for 85 (90%) and PCL for 10 (10%) patients. The immunoglobulin (Ig) subtype was IgG in 39 (46%), IgA in 23 (27%) and IgD in one case. Light chain myeloma accounted for 24% of patients, non-secretory for 2%, and the isotype was unknown in 9 patients. Twelve patients (17%) had chemo-refractory disease. Nearly all patients (96%) received at least one ASCT before UCBT: 26 (30%) received a planned tandem auto-auto and 18 (20%) a tandem procedure which included the current UCBT transplantation. The remaining 45 patients received one or more ASCT, but not as part of a planned tandem procedure. Only 4 patients did not receive a previous ASCT: 3 of them had PCL and received UCBT as first-line therapy in a median time of 5.5 months (range 4-6) from diagnosis; one had MM and received a UCBT after relapse at five years from diagnosis. Cytogenetic analysis was performed and available in 45 patients and was abnormal in 32 of them. The most frequent alteration was del13q (n=17). High-risk abnormalities [del17p or t(4;14)] were present in 1121

A. Paviglianiti et al. Table 2. Transplant characteristics.

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

Value Follow up, median (range) Age at UCBT , median (range) Diagnosis MM PCL Sex Male Female Subtype IgG IgA IgD Lambda or kappa light chain Non secretory Missing n=9 ISS stage I II III Missing n=28 Recipient CMV status Negative Positive Missing n=1 Cytogenetic abnormalities High-risk alterations [del17p, t(4;14)] Other alterations Normal Not performed Missing n=17 Chemosensitivity Chemo-refractory disease Chemosensitive disease Missing=22 Extramedullary disease Yes No Missing=27 Previous autotransplant 0 1 2 3 Missing n=2 Previous tandem auto-auto transplantation Yes No Missing=6 Disease status at UCBT 1st CR 2nd CR VGPR PR SD PD Missing n=4 Exposed to new drugs before transplant Yes No Missing n=6

41.3 mo (3.7-96) 53.3 yrs (24-69.6) 85 (90%) 10 (10%) 51 (54%) 44 (46%) 39 (46%) 23 (27%) 1 (1%) 21 (24%) 2 (2%)

26 (38%) 17 (26%) 24 (36%)

44 (47%) 50 (53%)

11 (14%) 21 (27%) 13 (17%) 33 (42%)

12 (17%) 61 (83%)

13 (18%) 55 (82%)

4 (4%) 46 (50%) 38 (41%) 5 (5%) 26 (29%) 63 (71%)

10 (11%) 10 (11%) 20(22%) 37 (41%) 4 (4%) 10 (11%)

82 (92%) 7 (8%)

MM: multiple myeloma; PCL: plasma cell leukemia; Kg: kilogram; CMV: cytomegalovirus; ISS: international scoring system; CR: complete remission;VGPR: very good partial remission; PR: partial remission; SD: stable disease; PD: progressive disease; mo: months; yrs: years; UCBT: umbilical cord blood transplantation.

1122

Value Type of UCBT sUCBT 36 (38%) dUCBT 59 (62%) Planned tandem auto-UCBT Yes 18 (20%) No 71 (80%) Missing n=6 HLA mismatches 0-1 mismatch 28 (32%) 2 or more mismatches 61 (68%) Missing n=6 Infused TNCX107/Kg, median (range) 3.3 (0.8-7.8) Infused CD34X105/Kg, median (range) 1.25 (0.1-4.5) Transplant year, median (range) 2009 (2001-2013) Time from diagnosis to transplant, median (range) 41.6 mo (4.6-235.6) Conditioning MAC Bu-based 9 (10%) TBI-based 7 (7%) Other 1 (1%) RIC Cy+Flu+TBI 60 (64%) Others 17 (18%) Missing n=1 GvHD prophylaxis CsA+MMF 74 (80%) Others 19 (20%) Missing n=2 ATG use Yes 22 (24%) No 68 (76%) Missing n=5 UCBT: umbilical cord blood transplantation; TNC: total nucleated cell at collection; Kg: kilogram; HLA: human leukocyte antigen; MAC: myeloablative conditioning regimen; RIC: reduced conditioning regimen; TBI: total body irradiation; Cy: cyclophosphamide; Flu: fludarabine; Bu: busulfan; GvHD: graft- versus-host-disease; CsA: cyclosporine; MMF: mycophenolate mofetil; ATG: anti-thymocyte globulin; mo: months.

11 patients. Ten patients were in first complete remission (CR) at UCBT, 10 in second CR, 20 in very good partial response (VGPR), 37 in partial response (PR), 14 in stable or progressive disease, and data were missing for the remaining 4 patients. Eighty-two patients received proteasome inhibitors or immunomodulatory drugs before UCBT. Among 43 patients with available information on HCTI-CI, 21 were reported as HCTI-CI 0, 4 HCTI-CI 1, 13 HCTI-CI 2 and 5 HCTI-CI 3. Fifty-nine patients (62%) received a dUCBT. The majority of patients were conditioned with an RIC regimen (n=77, 82%). The most common conditioning regimen was cyclophosphamide+fludarabine+TBI (2-6 Gy) (64%) and antithymocyte globulin (ATG) was given to 24% of the patients. Cyclosporine A (CSA)+mycophenolate mofetil (MMF) was the most frequent GvHD prophylaxis (80%). The median number of total nucleated cells (TNC) was 4.24x107/kg (range 2.2-7.8) at cryopreservation, and 3.3x107/kg (range 0.8-7.8) at infusion. The median number of cryopreserved and infused CD34+ cells was 1.78x105/kg (range 0.5-6.6) and 1.25x105/kg (range 0.1-4.5), haematologica | 2016; 101(9)

Cord blood transplantation in multiple myeloma

ently associated with higher NRM in the multivariate analysis (HR 3.35, 95%CI: 1.44-7.81; P=0.005). The CI of relapse at three years was 47%±5% (Figure 1). The RI was higher in chemo-refractory MM (75% vs. 45%; P=0.05). Moreover, in multivariate analysis patients with high cytogenetic risk had higher RI (HR 3.83, 95%CI: 1.26-11.61; P=0.018).

Overall survival and progression-free survival

Figure 1. 3-year non-relapse mortality and relapse incidence. Solid line represents relapse incidence; dashed line represents non-relapse mortality incidence.

respectively. The majority of patients (68%) received a graft with 2 HLA mismatches. Among 63 patients with available information on maintenance therapy, 3 were treated with lenalidomide after UCBT. Summaries of the univariate and multivariate analyses for major outcomes are shown in Tables 3 and 4, respectively.

Engraftment and GvHD The CI of 60-day PMN and 180-day PLT engraftment were 97%±3% and 72%±5%, respectively. The median time of PMN and PLT engraftment were 20 (range 7-53) and 33 (range 8-98) days, respectively. Seven patients failed to achieve PMN engraftment; of these, 4 died within a median time of 21 months after UCBT. Three patients who experienced graft failure were alive at last follow up, 2 after an autologous rescue. The CI of 100-day acute GvHD (aGvHD) grade II-IV and grade III-IV were 41%±5% and 16%±4%, respectively. aGvHD was lower in patients who received ATG (18% vs. 48%; P=0.02) and in those who did not receive TBI (15% vs. 49%; P<0.001). The use of ATG was associated with significant lower incidence of aGvHD in multivariate analysis (HR 0.25, 95%CI: 0.08-0.80; P=0.020). The CI of chronic GvHD (cGvHD) at two years was 22%±4%, with a median time of onset of 188 days. Among the 23 patients who experienced cGvHD, 11 were alive at last follow up and 9 were disease free. Extensive cGvHD was observed in 5 patients (5 of 23). Patients who underwent dUCBT had a higher incidence of cGvHD than those receiving sUCBT in univariate analysis (30% vs. 9%; P=0.015).

Non-relapse mortality and relapse incidence The CI of NRM at three years was 29%±5% (Figure 1). Overall, 63 patients died: 30 of relapse and 33 of transplant-related causes (infections, n=16; GvHD, n=5; other causes, n=12). In univariate analysis, ATG use (52% vs. 22%; P=0.004), MAC conditioning (54% vs. 23%; P=0.01) and TBI (51% vs. 22%; P=0.005) were associated with higher incidence of NRM. The use of ATG was independhaematologica | 2016; 101(9)

The median follow up for survivors was 41 months (range 3.7-96). The 3-year probability of PFS and OS was 24%±5% and 40%±5%, respectively (Figures 2 and 3). In univariate analysis, RIC regimen and CsA+MMF as GvHD prophylaxis were associated with improved OS (43% vs. 30%, P=0.05, and 45% vs. 14%, P<0.001, respectively). Conversely, the use of ATG was associated with a decreased survival (10% vs. 46%; P<0.001). The effect of ATG use retained significance in multivariate analysis, with decreased OS (HR 4.03, 95%CI: 2.13-7.64; P<0.001) and PFS (HR=2.73, 95%CI: 1.48-5.05; P=0.001). Moreover, in multivariate analysis patients with high-risk cytogenetic had poorer OS (HR 2.99, 95%CI: 1.31-6.83; P=0.009) and PFS (HR 2.88, 95%CI: 1.26-6.57; P=0.012).

Discussion We conducted a registry-based study with the objective of defining the role of UCBT in patients with plasma cell disorders. AlloSCT is not a standard treatment for patients with MM, and transplantation with alternative stem cell sources, such as UCBT, is even less common. The results of the current study suggest that UCBT is a feasible alternative for MM patients, and that high-risk cytogenetics and the use of ATG are independently associated with worse survival. In the recent guidelines from the American Society for Blood and Marrow Transplantation on the indications for ASCT and alloSCT, the former was considered “standard of care” for MM patients in initial response or in sensitive relapse, and is considered “standard of care with clinical evidence” in refractory MM and PCL. On the other hand, alloSCT is still considered “developmental” for MM in initial response, but for patients in other disease stages or PCL, it is accepted as “standard of care with clinical evidence”.23 These recommendations do not take into account other factors such as age, comorbidities, donor source, and HLA incompatibilities. Overall, results of alloSCT are poor because of the high transplant-related mortality and high risk of relapse. The EBMT reported 3-year OS and PFS of 41% and 21%, respectively, in 229 MM patients who received RIC alloSCT from related and unrelated donors.24 To date, only a few studies on the use of UCBT in MM have been published, and they were mostly isolated cases.17,18 Recently, a more comprehensive survey was reported by the Japanese registry19 in 86 patients with MM, showing 6-year OS and PFS of 15.2% and 13%, respectively. However, it has been shown in several publications that results are not always similar for Japanese and Western populations.25 Moreover, our study differs from the previous publication because it includes both single and double UCBT and uses a different classification for high-risk cytogenetics. In our series, in which 82% of patients received RIC regimen, OS and PFS were 40% and 24% at three years, 1123

A. Paviglianiti et al. Table 3. Univariate analysis of main transplant outcomes.

All patients Sex Male Female Cytogenetics High risk Not high risk Type of graft Single Double Number of HLA disparities 0-1HLA disparities 2 HLA disparities Conditioning regimen RIC MAC Use of TBI No Yes GvHD prophylaxis CsA MMF Others Use of ATG before day 0 No Yes TNCx107/kg â&#x2030;¤3.3 >3.3

(%)

95 95 51 44 45 11 34 95 36 59 89 28 61 94 77 17 94 21 73 90 71 19 91 69 22 93 48 45

aGvHD P

(%)

cGvHD P

(%)

NRM P

(%)

Relapse P

OS (%)

PFS (%)

42% 40%

0.82

16% 29%

0.22

36% 21%

0.07

42% 53%

0.24

36% 43%

0.20

22% 26%

0.32

18% 44%

0.09

0% 25%

0.06

45% 27%

0.21

36% 49%

0.43

18% 47%

0.07

17% 18%

0.27

31% 37%

0.12

9% 30%

0.01

28% 29%

0.94

45% 49%

0.9

39% 40%

0.49

27% 21%

0.80

43% 43%

0.96

14% 26%

0.20

33% 26%

0.66

50% 49%

0.98

40% 37%

0.96

17% 24%

0.52

45% 20%

0.06

24% 14%

0.41

23% 54%

0.01

52% 20%

0.14

43% 30%

0.05

25% 27%

0.20

15% <0.001 49%

19% 24%

0.50

22% 51%

<0.001

32% 52%

0.15

25% 44%

<0.001 17% 25%

0.10

46% 26%

0.11

25% 10%

0.19

24% 53%

<0.001

47% 39%

0.73

45% 14%

<0.001 28% 8%

<0.001

48% 18%

0.02

23% 11%

0.31

22% 52%

<0.001

49% 39%

0.57

46% 10%

<0.001 28% 0%

<0.001

38% 43%

0.60

17% 28%

0.18

34% 25%

0.45

46% 47%

0.66

33% 47%

0.27

20% 28%

0.42

aGvHD: acute graft-versus-host-disease; cGvHD: chronic graft-versus-host-disease; NRM: non-relapse mortality; OS: overall survival; PFS: progression-free survival; HLA: human leukocyte antigen; MAC: myeloablative conditioning regimen; RIC: reduced conditioning regimen; TBI: total body irradiation; ATG: anti-thymocyte globulin; TNC: total nucleated cell infused.

respectively. Although the number of previous therapy lines is not available, 90% of patients were transplanted beyond CR1, indicating that UCBT was not the first-line therapy for these patients. Our results are comparable to those observed in MM patients undergoing RIC-alloSCT with other stem cell sources, not only for PFS and OS, but also NRM (29%).24 We observed a detrimental impact of adverse karyotype in PFS and OS in multivariate analysis. Other authors have previously shown the negative impact of high-risk abnormalities, such as t(4;14), t(14;16), t(14;20) and del17p, on survival outcomes in patients with newly diagnosed MM.26,27 Similar findings were demonstrated in patients receiving front-line ASCT, in which high-risk cytogenetics was associated with worse outcomes28 and unsustained CR at one year.29 The prognostic impact of adverse cytogenetics on alloSCT outcome in MM is not well established. Schilling et al. showed that alloSCT can be beneficial for patients with t(4;14), but not for those with del17p.30 On the contrary, Roos-Weil et al. demonstrated that the increased risk associated with either of these mutations could be overcome with alloSCT.31 More recently, the benefit of alloSCT for patients with MM harboring both t(4;14) and del17p was confirmed in a prospective tandem auto/RIC-alloSCT protocol.32 1124

Contrary to the current results and other previous publicaFigure 2. 3-year overall survival.

haematologica | 2016; 101(9)

Cord blood transplantation in multiple myeloma

Table 4. Multivariate analysis. PFS ATG use vs. no ATG High-risk cytogenetics vs. no high risk Number of mismatch (0-1 vs. 2 or more) Single vs. double UCBT Median year of UCBT (≤2009 vs.>2009) CR1/CR2/VGPR vs. PR/SD/PD Median infused TNCx107/kg (≤3.3 vs. >3.3) OS ATG use vs. no ATG High-risk cytogenetics vs no high risk Number of mismatch (0-1 vs. 2 or more) Single vs. double UCBT Median year of UCBT (≤2009 vs. >2009) CR1/CR2/VGPR vs. PR/SD/PD Median infused TNCx107/kg (≤3.3 vs. >3.3) RI High-risk cytogenetics vs. no high risk Number of mismatch (0-1 vs. 2 or more) Single vs. double UCBT Median year of UCBT (≤2009 vs. >2009) CR1/CR2/VGPR vs. PR/SD/PD Median infused TNCx107/kg (≤3.3 vs. >3.3) ATG use vs. no ATG NRM ATG use vs. no ATG High-risk cytogenetics vs. no high risk Number of mismatch (0-1 vs. 2 or more) Single vs. double UCBT Median year of UCBT (≤2009 vs. >2009) CR1/CR2/VGPR vs. PR/SD/PD Median infused TNCx107/kg (≤3.3 vs. >3.3) Acute GvHD ATG use vs. no ATG High-risk cytogenetics vs. no high risk Number of mismatch (0-1 vs. 2 or more) Single vs. double UCBT Median year of UCBT (≤2009 vs. >2009) CR1/CR2/VGPR vs. PR/SD/PD Median infused TNCx107/kg (≤3.3 vs. >3.3)

95% CI

2.73 2.88 1.07 1.06 1.41 1.64 0.99

1.48-5.05 1.26-6.57 0.58-1.97 0.86-1.32 0.79-2.5 0.96-2.83 0.59-1.68

0.001 0.012 0.83 0.59 0.24 0.72 0.98

4.03 2.99 1.33 1.08 1.54 1.31 1.02

2.13-7.64 1.31-6.83 0.70-2.51 0.85-1.36 0.82-2.88 0.73-2.34 0.57-1.81

<0.001 0.009 0.38 0.55 0.18 0.38 0.95

3.83 0.86 0.98 0.88 1.76 1.13 2.01

1.26-11.61 0.39-1.93 0.75-1.29 0.39-2.03 0.85-3.63 0.58-2.20 0.67-6.05

0.018 0.72 0.88 0.77 0.13 0.72 0.21

3.35 2.06 1.68 1.24 2.47 1.46 0.75

1.44-7.81 0.55-7.6 0.64-4.39 0.87-1.75 1.03-5.95 0.63-3.40 0.31-1.80

0.005 0.28 0.30 0.24 0.04 0.38 0.52

0.24 0.51 0.77 1.00 1.31 1.08 1.24

0.08-0.80 0.12-2.23 0.35-1.70 0.75-1.32 0.63-2.72 0.49-2.38 0.62-2.48

0.020 0.37 0.51 0.98 0.46 0.86 0.54

HR: hazard ratio; CI: confidence interval; NRM: non-relapse mortality; RI: relapse incidence; OS: overall survival; PFS: progression-free survival; GvHD: graft-versus-host disease; UCBT: umbilical cord blood transplantation; CR1/2: first/second complete remission; VGPR: very good partial response; PR: partial remission; SD: stable disease; PD: progressive disease; TNC: total nucleated cells.

tions,27 a recent study on UCBT19 showed no association between high-risk cytogenetic and poor outcomes. A possible explanation for these findings might be the different risk group classification of patients harboring del13q. In our series, ATG use was associated with lower OS, PFS and aGvHD, and with higher NRM. This was also reported, recently, in a larger series of patients with hematologic malignancies undergoing UCBT after RIC regimen.33 As described in previous studies, immunosuppression with ATG is associated with a high incidence of infections.34 In our series, infection was the primary cause of transplant-related deaths among patients who received ATG (n=22). We were unable to identify any significant association between the impact of disease status at UCBT and planned tandem transplantation on MM outcomes, as suggested by the Japanese group.19 The lack of association haematologica | 2016; 101(9)

Figure 3. 3-year progression-free survival.

of these factors on UCBT outcomes may be related to the low number of patients included in the categories of some specific variables or to actual differences between the populations in the different studies. In our series, RI was high (47%), but comparable to results reported with other stem cell sources (bone marrow and peripheral blood stem cell) in previous alloSCT for MM studies. Strategies to prevent relapse after UCBT may include post-transplant consolidation and/or maintenance with immunomodulatory drugs or proteasome inhibitors.2 Thalidomide has been investigated as salvage therapy in 31 patients35 and at low doses with DLI in 18 patients,36 both after alloSCT. Lenalidomide is already used after ASCT, but its use is still controversial after alloHSCT.37 However, it may be beneficial, especially for high-risk MM,38 as it has been demonstrated to improve response rate and to increase T-cell activity.39 Bortezomib has been used in relapsed MM after RIC alloSCT (n=18) with a certain level of toxicity40 and in patients not responding to DLI.41 However, the application of novel agents in the UCBT setting and their potential in intensifying the GvMM effect after transplant ought to be further explored. In fact, there are several ongoing prospective studies (clinicaltrials.gov identifiers: 02440464, 020308280, 01460420, 01131169, 02447055) including anti-myeloma drugs as maintenance therapy early after alloSCT that may improve outcomes of MM patients.42 One study in particular (clinicaltrials.gov identifier: 02440464) will investigate the use of a 2nd-generation anti-myeloma drug (ixazomib) in association with immunosuppressive therapy after alloSCT. Unfortunately, only 3 patients in this series were reported to have received maintenance therapy, therefore we were unable to evaluate such strategies. Despite some limitations intrinsic to the retrospective nature of our study, we have demonstrated that UCBT is a feasible option for MM patients needing alloSCT. 1125

A. Paviglianiti et al.

Furthermore, the clinical applications for UCBT are still evolving. Several methods, such as the combination of cord blood and CD34+ selected haploidentical graft, the addition of mesenchymal stem cells, cord blood intrabone infusion and ex vivo expansion techniques are under investigation to improve engraftment.43-45 However, further studies are needed to determine the potential benefit of these innovative strategies. Also, the use of haploidentical transplantation may deserve to be investigated to determine its applicability in this setting. The place of alloSCT, including UCBT, is still unclear,

References 1. Warren JL, Harlan LC, Stevens J, Little RF, Abel GA. Multiple myeloma treatment transformed: a population-based study of changes in initial management approaches in the United States. J Clin Oncol. 2013;31(16):1984-1989. 2. Lonial S, Boise LH, Kaufman J. How I treat high-risk myeloma. Blood. 2015; 126(13):1536-1543. 3. Attal M, Harousseau JL, Facon T, et al. Single versus double autologous stem-cell transplantation for multiple myeloma. N Engl J Med. 2003;349(26):2495-2502. 4. Cavo M, Tosi P, Zamagni E, et al. Prospective, randomized study of single compared with double autologous stemcell transplantation for multiple myeloma: Bologna 96 clinical study. J Clin Oncol. 2007;25(17):2434-2441. 5. Morris C, Iacobelli S, Brand R, et al. Benefit and timing of second transplantations in multiple myeloma: clinical findings and methodological limitations in a European Group for Blood and Marrow Transplantation registry study. J Clin Oncol. 2004;22(9):1674-1681. 6. Barlogie B, Attal M, Crowley J, et al. Longterm follow-up of autotransplantation trials for multiple myeloma: update of protocols conducted by the intergroupe francophone du myelome, southwest oncology group, and university of arkansas for medical sciences. J Clin Oncol. 2010;28(7):1209-1214. 7. Barlogie B, Tricot GJ, van Rhee F, et al. Long-term outcome results of the first tandem autotransplant trial for multiple myeloma. Br J Haematol. 2006;135(2):158164. 8. Alici E, Bjorkstrand B, Treschow A, et al. Long-term follow-up of gene-marked CD34+ cells after autologous stem cell transplantation for multiple myeloma. Cancer Gene Ther. 2007;14(3):227-232. 9. Mohty M, Boiron JM, Damaj G, et al. Graft-versus-myeloma effect following antithymocyte globulin-based reduced intensity conditioning allogeneic stem cell transplantation. Bone Marrow Transplant. 2004;34(1):77-84. 10. Bruno B, Rotta M, Patriarca F, et al. A comparison of allografting with autografting for newly diagnosed myeloma. New Engl J Med. 2007;356(11):1110-1120. 11. Moreau P. Death of frontline allo-SCT in myeloma. Blood. 2012;119(26):6178-6179. 12. Lokhorst HM, van der Holt B, Cornelissen JJ, et al. Reduced relapse rate in upfront tandem autologous/reduced intensity allogeneic transplantation in multiple myeloma

1126

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

but progress may be expected with a better identification of high-risk criteria, and a co-ordinated sequential approach with new drugs and transplant strategies. Funding MM was supported by educational grants from the â&#x20AC;&#x153;Association for Training, Education and Research in Hematology, Immunology and Transplantationâ&#x20AC;? (ATERHIT). M Mohty would like to thank Prof. Junia V. Melo (University of Adelaide, Australia, and Imperial College, London) for critical reading of this manuscript.

only results in borderline non-significant prolongation of progression free and not of overall survival. Haematologica. 2015; 100(12):e508-510. Corradini P, Cavo M, Lokhorst H, et al. Molecular remission after myeloablative allogeneic stem cell transplantation predicts a better relapse-free survival in patients with multiple myeloma. Blood. 2003;102(5):1927-1929. Rosinol L, Perez-Simon JA, Sureda A, et al. A prospective PETHEMA study of tandem autologous transplantation versus autograft followed by reduced-intensity conditioning allogeneic transplantation in newly diagnosed multiple myeloma. Blood. 2008; 112(9):3591-3593. Gahrton G, Iacobelli S, Bjorkstrand B, et al. Autologous/reduced-intensity allogeneic stem cell transplantation vs autologous transplantation in multiple myeloma: longterm results of the EBMT-NMAM2000 study. Blood. 2013;121(25):5055-5063. Kroger N, Shimoni A, Schilling G, et al. Unrelated stem cell transplantation after reduced intensity conditioning for patients with multiple myeloma relapsing after autologous transplantation. Br J Haematol. 2010;148(2):323-331. Fenk R, Neumann F, Fenk B, et al. Unrelated umbilical cord blood transplantation as salvage treatment for engraftment failure following autologous stem cell transplantation. Leuk Res. 2008;32(7):1157-1159. Kasahara I, Nishio M, Yamamoto S, et al. Cord blood transplantation with a reducedintensity conditioning regimen for patients with relapsed aggressive multiple myeloma after cytoreduction with bortezomib. Int J Hematol. 2009;90(3):413-415. Kawamura K, Takamatsu H, Ikeda T, et al. Cord Blood Transplantation for Multiple Myeloma: A Study from the Multiple Myeloma Working Group of the Japan Society for Hematopoietic Cell Transplantation. Biol Blood Marrow Transplant. 2015;21(7):1291-1298. Durie BG, Harousseau JL, Miguel JS, et al. International uniform response criteria for multiple myeloma. Leukemia. 2006; 20(9):1467-1473. Przepiorka D, Weisdorf D, Martin P, et al. 1994 Consensus Conference on Acute GVHD Grading. Bone Marrow Transplant. 1995;15(6):825-828. Flowers ME, Kansu E, Sullivan KM. Pathophysiology and treatment of graftversus-host disease. Hematol Oncol Clin North Am. 1999;13(5):1091-1112, viii-ix. Majhail NS, Farnia SH, Carpenter PA, et al. Indications for Autologous and Allogeneic

24.

25.

26.

27.

28.

29.

30.

31.

32.

Hematopoietic Cell Transplantation: Guidelines from the American Society for Blood and Marrow Transplantation. Biol Blood Marrow Transplant. 2015; 21(11):1863-1869. Crawley C, Lalancette M, Szydlo R, et al. Outcomes for reduced-intensity allogeneic transplantation for multiple myeloma: an analysis of prognostic factors from the Chronic Leukaemia Working Party of the EBMT. Blood. 2005;105(11):4532-4539. Tanimoto TE, Yamaguchi T, Tanaka Y, et al. Comparative analysis of clinical outcomes after allogeneic bone marrow transplantation versus peripheral blood stem cell transplantation from a related donor in Japanese patients. Br J Haematol. 2004;125(4):480493. Avet-Loiseau H, Attal M, Moreau P, et al. Genetic abnormalities and survival in multiple myeloma: the experience of the Intergroupe Francophone du Myelome. Blood. 2007;109(8):3489-3495. Bergsagel PL, Mateos MV, Gutierrez NC, Rajkumar SV, San Miguel JF. Improving overall survival and overcoming adverse prognosis in the treatment of cytogenetically high-risk multiple myeloma. Blood. 2013;121(6):884-892. Moreau P, Cavo M, Sonneveld P, et al. Combination of international scoring system 3, high lactate dehydrogenase, and t(4;14) and/or del(17p) identifies patients with multiple myeloma (MM) treated with front-line autologous stem-cell transplantation at high risk of early MM progressionrelated death. J Clin Oncol. 2014;32(20): 2173-2180. Paiva B, Gutierrez NC, Rosinol L, et al. High-risk cytogenetics and persistent minimal residual disease by multiparameter flow cytometry predict unsustained complete response after autologous stem cell transplantation in multiple myeloma. Blood. 2012;119(3):687-691. Schilling G, Hansen T, Shimoni A, et al. Impact of genetic abnormalities on survival after allogeneic hematopoietic stem cell transplantation in multiple myeloma. Leukemia. 2008;22(6):1250-1255. Roos-Weil D, Moreau P, Avet-Loiseau H, et al. Impact of genetic abnormalities after allogeneic stem cell transplantation in multiple myeloma: a report of the Societe Francaise de Greffe de Moelle et de Therapie Cellulaire. Haematologica. 2011; 96(10):1504-1511. Kroger N, Badbaran A, Zabelina T, et al. Impact of high-risk cytogenetics and achievement of molecular remission on long-term freedom from disease after autol-

haematologica | 2016; 101(9)

Cord blood transplantation in multiple myeloma

33.

34.

35.

36.

ogous-allogeneic tandem transplantation in patients with multiple myeloma. Biol Blood Marrow Transplant. 2013; 19(3):398-404. Pascal L, Tucunduva L, Ruggeri A, et al. Impact of ATG-containing reduced-intensity conditioning after single- or double-unit allogeneic cord blood transplantation. Blood. 2015;126(8):1027-1032. Brunstein CG, Barker JN, Weisdorf DJ, et al. Umbilical cord blood transplantation after nonmyeloablative conditioning: impact on transplantation outcomes in 110 adults with hematologic disease. Blood. 2007; 110(8):3064-3070. Mohty M, Attal M, Marit G, et al. Thalidomide salvage therapy following allogeneic stem cell transplantation for multiple myeloma: a retrospective study from the Intergroupe Francophone du Myelome (IFM) and the Societe Francaise de Greffe de Moelle et Therapie Cellulaire (SFGMTC). Bone Marrow Transplant. 2005;35(2):165-169. Kroger N, Shimoni A, Zagrivnaja M, et al. Low-dose thalidomide and donor lymphocyte infusion as adoptive immunotherapy

haematologica | 2016; 101(9)

37.

38.

39.

40.

after allogeneic stem cell transplantation in patients with multiple myeloma. Blood. 2004;104(10):3361-3363. Kneppers E, van der Holt B, Kersten MJ, et al. Lenalidomide maintenance after nonmyeloablative allogeneic stem cell transplantation in multiple myeloma is not feasible: results of the HOVON 76 Trial. Blood. 2011;118(9):2413-2419. Alsina M, Becker PS, Zhong X, et al. Lenalidomide maintenance for high-risk multiple myeloma after allogeneic hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2014;20(8): 1183-1189. Wolschke C, Stubig T, Hegenbart U, et al. Postallograft lenalidomide induces strong NK cell-mediated antimyeloma activity and risk for T cell-mediated GvHD: Results from a phase I/II dose-finding study. Exp Hematol. 2013;41(2):134-142 e133. Kroger N, Zabelina T, Ayuk F, et al. Bortezomib after dose-reduced allogeneic stem cell transplantation for multiple myeloma to enhance or maintain remission status. Exp Hematol. 2006;34(6):770-775.

41. van de Donk NW, Kroger N, Hegenbart U, et al. Prognostic factors for donor lymphocyte infusions following non-myeloablative allogeneic stem cell transplantation in multiple myeloma. Bone Marrow Transplant. 2006;37(12):1135-1141. 42. Dhakal B, Vesole DH, Hari PN. Allogeneic stem cell transplantation for multiple myeloma: is there a future? Bone Marrow Transplant. 2016;51(4):492-500. 43. Frassoni F, Gualandi F, Podesta M, et al. Direct intrabone transplant of unrelated cord-blood cells in acute leukaemia: a phase I/II study. Lancet Oncol. 2008; 9(9):831-839. 44. Delaney C, Heimfeld S, Brashem-Stein C, Voorhies H, Manger RL, Bernstein ID. Notch-mediated expansion of human cord blood progenitor cells capable of rapid myeloid reconstitution. Nat Med. 2010;16(2):232-236. 45. Kwon M, Bautista G, Balsalobre P, et al. Haplo-cord transplantation using CD34+ cells from a third-party donor to speed engraftment in high-risk patients with hematologic disorders. Biol Blood Marrow Transplant. 2014;20(12):2015-2022.

1127

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

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

Ancient Greek

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

Scientific Latin

haematologicus (adjective) = related to blood

Scientific Latin

haematologica (adjective, plural and neuter, used as a noun) = hematological subjects

Modern English

The oldest hematology journal, publishing the newest research results. 2015 JCR impact factor = 6.671

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