Haematologica, Volume 102, issue 12 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), Ziggy Kennell (English Editor)

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

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

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

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

Information for readers, authors and subscribers Haematologica (print edition, pISSN 0390-6078, eISSN 1592-8721) publishes peer-reviewed papers on all areas of experimental and clinical hematology. The journal is owned by a non-profit organization, the Ferrata Storti Foundation, and serves the scientific community following the recommendations of the World Association of Medical Editors (www.wame.org) and the International Committee of Medical Journal Editors (www.icmje.org). Haematologica publishes editorials, research articles, review articles, guideline articles and letters. Manuscripts should be prepared according to our guidelines (www.haematologica.org/information-for-authors), and the Uniform Requirements for Manuscripts Submitted to Biomedical Journals, prepared by the International Committee of Medical Journal Editors (www.icmje.org). Manuscripts should be submitted online at http://www.haematologica.org/. Conflict of interests. According to the International Committee of Medical Journal Editors (http://www.icmje.org/#conflicts), “Public trust in the peer review process and the credibility of published articles depend in part on how well conflict of interest is handled during writing, peer review, and editorial decision making”. The ad hoc journal’s policy is reported in detail online (www.haematologica.org/content/policies). Transfer of Copyright and Permission to Reproduce Parts of Published Papers. Authors will grant copyright of their articles to the Ferrata Storti Foundation. No formal permission will be required to reproduce parts (tables or illustrations) of published papers, provided the source is quoted appropriately and reproduction has no commercial intent. Reproductions with commercial intent will require written permission and payment of royalties. Detailed information about subscriptions is available online at www.haematologica.org. Haematologica is an open access journal. Access to the online journal is free. Use of the Haematologica App (available on the App Store and on Google Play) is free. For subscriptions to the printed issue of the journal, please contact: Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, E-mail: info@haematologica.org). Rates of the International edition for the year 2017 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 November 2017.

haematologica calendar of events

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

The American Society of Hematology 59th ASH Annual Meeting and Exposition American Society of Hematology (ASH) Chairs: KC Anderson, A Cuker, L Sehn, SA Armstrong, AS Weyrich December 9-12, 2017 Atlanta, US

EHA-TSH Hematology Tutorial on Acute Leukemias April 28-29, 2018 Istanbul, Turkey

EHA-SWG Scientific Meeting on Integrated Diagnosis Strategies in Oncohematology for the management of cytopenias and leukocytosis Chair: MC Béné February 8-10, 2018 Barcelona, Spain

EHA-SAH Hematology Tutorial on lymphoid Malignancies and Plasma Cell Dyscrasias September 13-14, 2018 Buenos Aires, Argentina

EHA-ISHBT Hematology Tutorial on Lymphoproliferative and Plasma Cell Disorders February 16-18, 2018 Lucknow, India

EHA Hematology Tutorial on Thalassemia May 3-4, 2018 Shiraz, Iran

EHA-SWG Scientific Meeting on Aging and Hematology Chair: D Bron October 12-14, 2018 Location TBC

EuroClonality Workshop: Clonality assessment in Pathology European Scientific foundation for Laboratory Hemato Oncology (ESLHO) Chairs: PJTA Groenen, F Fend, AW Langerak February 19-21, 2018 Nijmegen, The Netherlands EHA-SWG Scientific Meeting on New Molecular Insights and Innovative Management Approaches for Acute Lymphoblastic Leukemia Chair: N Gökbuget April 12-14, 2018 Barcelona, Spain

Calendar of Events updated on November 1, 2017

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

Table of Contents Volume 102, Issue 12: December 2017 Cover Figure Image generated by www.somersault1824.com.

Editorials 1969

Balance your folate or the yin and yang of folate in hematopoiesis Hartmut Geiger

1970

Transplantation for therapy-related, TP53-mutated myelodysplastic syndrome – not because we can, but because we should Corey Cutler

Guideline Article 1972

Ferroportin disease: pathogenesis, diagnosis and treatment Antonello Pietrangelo

Articles Hematopoiesis

1985

Folate dietary insufficiency and folic acid supplementation similarly impair metabolism and compromise hematopoiesis Curtis J. Henry et al.

Red Cell Biology & its Disorders

1995

Hydroxyurea differentially modulates activator and repressors of γ-globin gene in erythroblasts of responsive and non-responsive patients with sickle cell disease in correlation with Index of Hydroxyurea Responsiveness Xingguo Zhu et al.

Hemostasis

2005

Molecular and clinical profile of von Willebrand disease in Spain (PCM-EVW-ES): Comprehensive genetic analysis by next-generation sequencing of 480 patients Nina Borràset al.

Myelodysplastic Syndromes

2015

Pro-inflammatory proteins S100A9 and tumor necrosis factor-α suppress erythropoietin elaboration in myelodysplastic syndromes Thomas Cluzeau et al.

2021

Red cell alloimmunization is associated with development of autoantibodies and increased red cell transfusion requirements in myelodysplastic syndrome Deepak Singhal et al.

2030

Favorable impact of allogeneic stem cell transplantation in patients with therapy-related myelodysplasia regardless of TP53 mutational status Ibrahim Aldoss et al.

Acute Myeloid Leukemia

2039

Circular RNAs of the nucleophosmin (NPM1) gene in acute myeloid leukemia Susanne Hirsch et al.

Haematologica 2017; vol. 102 no. 12 - December 2017 http://www.haematologica.org/

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation 2048

2058

Phenotype in combination with genotype improves outcome prediction in acute myeloid leukemia: a report from Children’s Oncology Group protocol AAML0531 Andrew P. Voigt et al.

Chronic Lymphocytic Leukemia

2069

CD40 signaling instructs chronic lymphocytic leukemia cells to attract monocytes via the CCR2 axis Martijn H.A. van Attekum et al.

Non-Hodgkin Lymphoma

2077

Pattern of somatic mutations in patients with Waldenström macroglobulinemia or IgM monoclonal gammopathy of undetermined significance Marzia Varettoni et al.

2086

Results and conclusions of the European Intergroup EURO-LB02 trial in children and adolescents with lymphoblastic lymphoma Eva Landmann et al.

2097

Dose-adjusted EPOCH chemotherapy for untreated peripheral T-cell lymphomas: a multicenter phase II trial of West-JHOG PTCL0707 Yoshinobu Maeda et al.

2104

Pan-phosphatidylinositol 3-kinase inhibition with buparlisib in patients with relapsed or refractory non-Hodgkin lymphoma Anas Younes et al.

Plasma Cell Disorders

2113

The kinesin spindle protein inhibitor filanesib enhances the activity of pomalidomide and dexamethasone in multiple myeloma Susana Hernández-García et al.

Cell Therapy & Immunotherapy

2125

Steroid treatment of acute graft-versus-host disease grade I: a randomized trial Andrea Bacigalupo et al.

2134

Mixed phenotype acute leukemia: outcomes with allogeneic stem cell transplantation. A retrospective study from the Acute Leukemia Working Party of the EBMT Reinhold Munker et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/102/12.toc

e473

Clinical relevance of silent red blood cell autoantibodies Francesca R. Mauro et al. http://www.haematologica.org/content/102/12/e473

e476

Rac1 functions downstream of miR-142 in regulation of erythropoiesis Natalia Rivkin et al. http://www.haematologica.org/content/102/12/e476

e481

In vitro evidence of complement activation in patients with sickle cell disease Eleni Gavriilaki et al. http://www.haematologica.org/content/102/12/e481

Haematologica 2017; vol. 102 no. 12 - December 2017 http://www.haematologica.org/

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

e483

Catalytically inactive Gla-domainless factor Xa binds to TFPI and restores ex vivo coagulation in hemophilia plasma Atanur Ersayin et al. http://www.haematologica.org/content/102/12/e483

e486

Impact of hospital experience on the quality of tyrosine kinase inhibitor response monitoring and consequence for chronic myeloid leukemia patient survival Inge G.P. Geelen et al. http://www.haematologica.org/content/102/12/e486

e490

High prevalence of relapse in children with Philadelphia-like acute lymphoblastic leukemia despite risk-adapted treatment Susan L. Heatley et al. http://www.haematologica.org/content/102/12/e490

e494

Consolidation treatment with lenalidomide following front-line or salvage chemoimmunotherapy in chronic lymphocytic leukemia Paolo Strati et al. http://www.haematologica.org/content/102/12/e494

e497

Rituximab biosimilar evaluated by network meta-analysis Marco Chiumente et al. http://www.haematologica.org/content/102/12/e497

e499

Acute myeloid leukemia with mutated nucleophosmin 1: an immunogenic acute myeloid leukemia subtype and potential candidate for immune checkpoint inhibition Jochen Greiner et al. http://www.haematologica.org/content/102/12/e499

Comments Comments are available online only at www.haematologica.org/content/102/12.toc

e502

Amotosalen/UVA pathogen inactivation technology reduces platelet activability, induces apoptosis and accelerates clearance Beatrice Hechler et al. http://www.haematologica.org/content/102/12/e502

e504

In response to the comment by Hechler et al.: Amotosalen/UVA pathogen inactivation technology reduces platelet activatability, induces apoptosis and accelerates clearance Simona Stivala et al. http://www.haematologica.org/content/102/12/e504

Case Reports Case Reports are available online only at www.haematologica.org/content/102/12.toc

e506

Self-reverting mutations partially correct the blood phenotype in a Diamond Blackfan anemia patient Parvathy Venugopal et al. http://www.haematologica.org/content/102/12/e506

e510

A t(1;9) translocation involving CSF3R as a novel mechanism in unclassifiable chronic myeloproliferative neoplasm JesĂşs GutiĂŠrrez-Abril et al. http://www.haematologica.org/content/102/12/e510

e514

Co-occurrence of CRLF2-rearranged and Ph+ acute lymphoblastic leukemia: a report of four patients Nitin Jain et al. http://www.haematologica.org/content/102/12/e514

Haematologica 2017; vol. 102 no. 12 - December 2017 http://www.haematologica.org/

EDITORIALS Balance your folate or the yin and yang of folate in hematopoiesis Hartmut Geiger Institute of Molecular Medicine and Aging Research Center, University of Ulm, Germany and Division of Experimental Hematology and Cancer Biology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA E-mail: hartmut.geiger@uni-ulm.de / hartmut.geiger@cchmc.org doi:10.3324/haematol.2017.179838

new study in mice demonstrates that, in general, both low and high levels of dietary folate compromise hematopoiesis, and affect the B-cell progenitor compartment in particular.1 Given that there are substantial fractions of the world population at risk for either excessive (via fortified diets) or very low folate intake, this study might contribute to novel public policy recommendations on the level of folate intake, specifically in the context of fortified diets. There has, of late, been a widespread interest in the role of dietary nutrients in hematopoiesis, including specific types of vitamins and amino acids, as, interestingly, dietary factors seem to play a critical regulatory role in blood cell production as well as in leukemia. For example, high-dose vitamin C treatment induced hematopoietic stem cells (HSCs) to mature, and suppressed the growth of Tet methylcytosine dioxygenase 2 (Tet2) deficient leukemia cancer stem cells from human patients implanted in mice, by promoting DNA methylation.2 It was further demonstrated that in the absence of retinoic acid, a vitamin A metabolite, active HSCs are unable to return to a dormant state, instead maturing into specialized blood cells, and that mice fed with a vitamin A-deficient diet showed a disrupted reentry of HSCs into dormancy following exposure to inflammatory stress stimuli, thus eventually losing their HSCs.3 The Nakauchi laboratory showed that a diet deficient in the essential amino acid valine, resulted in the selective depletion of HSCs in the bone marrow of mice, which allowed for a successful transplant of HSCs without an additional preconditioning regimen.4 Ultimately, it has also been reported that multiple cycles of short-term fasting, and the accompanying restricted uptake of vitamins and essential amino acids, leads to signal transduction changes in HSCs and niche cells which promote stress resistance, self-renewal, and regeneration.5 We can now add to this list a novel yin/yang role of folate in hematopoiesis. Folate is a water-soluble B vitamin (coenzyme) listed as vitamin B9. Generally, folate coenzymes, such as vitamin B9 and vitamin B12, act as acceptors and donors of one-carbon units in a variety of reactions critical to nucleic acids synthesis (including thymine, but also purine bases), amino acids and DNA methylation reactions, and hence epigenetic mechanisms.6 An inadequate low folate status during early pregnancy increases the risk of congenital abnormalities. Notably, neural tube defects have been linked to inadequate levels of folate provided to the fetus in the first trimester of development. In general, due to the molecular role of folate in nucleotide synthesis, systems with an overall high number of dividing cells, such as the hematopoietic system, depend on proper levels of folate intake to provide adequate levels of erythrocytes and leukocytes and to prevent anemia.7 Erythroblasts, for example, require folate for proliferation during their differentiation. Folate further supports the generation of “normal” red blood cells in sickle cell haematologica | 2017; 102(12)

patients.8 The negative role of low folate levels with regard to hematopoiesis has therefore been recognized for quite some time.7,9 Folate is provided naturally by a whole variety of foods, including green leafy vegetables as well as meat, fish, eggs and grains. Folate can also be chemically synthesized. Due to the low dietary intake of folate in multiple countries, and its vital role in preventing multiple congenital defects, folate has been supplemented in at least one major cereal grain product since the late 1990s in order to reduce, for instance, neural tube defects (which it proved to do) in more than 70 countries worldwide, including North America, while in the EU such a fortification with folate is not mandatory. In contrast to the well-known positive benefits of folate supplementation, epidemiological data also support a likely negative impact of high levels of folate intake with respect to diseases and cancer;10 thus, both too low and very high levels of folate in diets might have negative consequences on health. While there are ample experimental studies on the negative consequences of low levels of dietary folate intake, the impact of high levels of folate has not of yet been addressed in great detail. The toxicity of folic acid has, up to now, been universally regarded as being very low, as folate is a water-soluble vitamin and thus an excess will be likely removed from the body through urine, as in the case of vitamin C. Henry et al. have now addressed, in an elegant set of well designed experiments, the extent to which distinct levels of dietary folate influence B cell development and hematopoiesis, via transplantation experiments in particular.1 To this end, they simply fed mice diets deficient in folate or diets that contained high levels of folate for up to 12 months. The diet high in folate resulted in folate serum levels that are within the range which are likewise achieved with fortified diets in human serum. Both low and high levels of folate resulted in impaired hematopoiesis, and primarily affected the B cell compartment. High and low levels of folate further reduced the survival rate of animals exposed to an irradiation dose that was not lethal to control animals. These novel studies ultimately confirm that excessively low and high levels of dietary folic acid negatively impact B cell development and hematopoiesis. Interestingly, changes in folate primarily affected the nucleotide synthesis pathway in hematopoietic cells, while other pathways linked to folate seemed to play only a minor role in affecting hematopoiesis. This suggests that hematopoietic cells might be very sensitive to changes in the level of nucleotides, both too low and too high. This work, although inherently less mechanistic than, for example, pure “genetic” approaches, is nonetheless an important contribution to the field of dietary interventions in hematopoiesis and their influence on metabolites in the system. These results will need to be further tested in additional epidemiological studies, focusing 1969

Editorials

on the high or higher end levels of folate and their connection to the long-term outcome rather than investigating low levels and their connection to disease. Taken together, these data will support the consideration of novel public guidelines on the recommended level of folate that will keep one healthy and strong.

4. 5. 6.

References 1. Henry C, Nemkov T, CasĂĄs-Selves M, et al. Folate dietary insufficiency and folic acid supplementation similarly impair metabolism and compromise hematopoiesis. Haematologica 2017;102(12):19851994. 2. Cimmino L, Dolgalev I, Wang Y, et al. Restoration of TET2 function blocks aberrant self-renewal and leukemia progression. Cell. 2017;170(6):1079-1095.e20. 3. Cabezas-Wallscheid N, Buettner F, Sommerkamp P, et al. Vitamin A-

7. 8. 9. 10.

retinoic acid signaling regulates hematopoietic stem cell dormancy. Cell. 2017;169(5):807-823.e19. Taya Y, Ota Y, Wilkinson AC, et al. Depleting dietary valine permits nonmyeloablative mouse hematopoietic stem cell transplantation. Science. 2016;354(6316):1152-1155. Cheng CW, Adams GB, Perin L, et al. Prolonged fasting reduces IGF1/PKA to promote hematopoietic-stem-cell-based regeneration and reverse immunosuppression. Cell Stem Cell. 2014;14(6):810-823. Yang M, Vousden KH. Serine and one-carbon metabolism in cancer. Nat Rev Cancer. 2016;16(10):650-662. Bills ND, Koury MJ, Clifford AJ, Dessypris EN. Ineffective hematopoiesis in folate-deficient mice. Blood. 1992;79(9):2273-2280. Koury MJ, Ponka P. New insights into erythropoiesis: the roles of folate, vitamin B12, and iron. Annu Rev Nutr. 2004;24:105-131. Salojin KV, Cabrera RM, Sun W, et al. A mouse model of hereditary folate malabsorption: deletion of the PCFT gene leads to systemic folate deficiency. Blood. 2011;117(18):4895-4904. Rycyna KJ, Bacich DJ, O'Keefe DS. Opposing roles of folate in prostate cancer. Urology. 2013;82(6):1197-1203.

Transplantation for therapy-related, TP53-mutated myelodysplastic syndrome â&#x20AC;&#x201C; not because we can, but because we should Corey Cutler Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA E-mail: corey_cutler@dfci.harvard.edu doi:10.3324/haematol.2017.181180

To the man who only has a hammer, everything he encounters begins to look like a nail. Abraham Maslow, and others

ransplant physicians have often been accused of performing allogeneic hematopoietic stem cell transplant (HSCT) in patients for whom no hope of cure, or even meaningful gain, was thought plausible, simply because HSCT was deemed possible. Exemplifying this is the concept of transplantation for therapy-related myelodysplastic syndrome (tMDS), or MDS associated with the most severe of genetic lesions, TP53, for which outcomes have historically been poor. In this issue of Haematologica, Aldoss and colleagues from the City of Hope Medical Center compare the outcomes of patients with tMDS and de novo MDS who underwent allogeneic HSCT, and correlate molecular features with outcome.1 Although the study was limited by small sample size, the authors noted no significant differences in all important post-HSCT clinical outcomes, including survival, between tMDS and de novo MDS patients, even when prior chemoradiotherapy was considered in multivariable models. This analysis therefore suggests that allogeneic transplantation for tMDS should be performed whenever it would be considered in the de novo MDS setting. Perhaps even more importantly, the authors examine the impact of molecular lesions on transplant outcome in a subset of the patients. From the original cohort, 60 tMDS patients underwent a comprehensive molecular analysis: 30% had a TP53 mutation, and the authors found that the presence or absence of a TP53 mutation had no correlation with outcome among these tMDS

1970

patients, although sample sizes were limited. In this context, grouping a heterogeneous subset of patients with genetic changes associated with adverse outcome in the non-transplant setting did carry prognostic information. In patients with any one of five genes (TP53, EZH2, ETV6, RUNX1, ASXL1) associated with adverse risk (48% of molecularly characterized tMDS patients) the authors demonstrated an adverse impact on relapse-free survival (hazard ratio: 1.58). This difference was also not statistically significant due to small sample size. It is worth noting that in this single center, retrospective analysis, outcomes were generally favorable, with 5-year overall survival rates approaching 50%. Numerous studies have now examined the outcomes associated with molecular mutations following allogeneic HSCT (Table 1); however, this is one of the largest to specifically examine tMDS patients and compare their outcomes with those of patients with de novo MDS. Initial studies examining molecular prognostic features demonstrated dismal results for patients with mutated TP53.2 In a study of 87 MDS patients who underwent allogeneic HSCT at our institute, there were no long-term survivors in the subset with TP53 mutations, with the majority of deaths occurring within the first 12 months following allogeneic HSCT. Fifteen of 18 deaths in this group were attributed to disease relapse. While sample sizes were exceedingly small, this result instantly changed the landscape of transplantation, with many centers deciding not to perform HSCT in TP53-mutated MDS patients at all. Other larger series have now reported outcomes of TP53-mutated HSCT patients with slightly more promising results. For example, Yoshizato and colleagues haematologica | 2017; 102(12)

Editorials

Table 1. Outcomes associated with molecular mutations in MDS patients following allogeneic HSCT.

Sample size

Fraction of cases with TP53 mutation

Impact on survival following transplant

Outcome influenced by complex karyotype?

Outcome influenced by variant allele fraction?

Outcome influenced by TP53 mutation type?

20.7% 12.7% 13% 19%

HR 4.22, P< 0.001 HR 1.49 HR 1.82, P=0.022b HR 1.96, P<0.001

Unknowna Yes Not studied No

Not studied Yes No No

Not studied Not studied Not studied Yes

Bejar et al.2 87 Yoshizato et al.3 797 Della Porta et al.4 401 Lindsley et al.5 1514

16/18 TP53 mutated patients had a complex karyotype; bConsidering MDS patients alone. HR: hazard ratio.

demonstrated in a cohort of nearly 800 subjects that while TP53 mutations did adversely influence outcomes, complex karyotype had a strong influence.3 While survival with a complex karyotype and a TP53 mutation was poor, being as low as 10%, for those without a complex karyotype, results were quite good (73% survival at 5 years), although sample sizes were very small (n=12) because a number of patients with unrecognized complex karyotypes were excluded using sensitive sequencingbased copy-number analysis. Della Porta and colleagues studied 401 patients with MDS or acute myeloid leukemia arising from MDS who underwent allogeneic HSCT.4 The presence of a TP53 mutation significantly affected outcomes in both groups of patients, but there was no influence of variant allele fraction on outcomes. Incorporating TP53 (and other) mutation states into the revised International Prognostic Scoring System could improve prognostication. It is notable that in this analysis, the survival rate at 4 years was approximately 30%, but there were no survivors beyond 10 years in the presence of a TP53 mutation, with more than half of deaths occurring prior to the 2-year mark. In the largest analysis of molecular features and their influence on transplantation outcomes, Lindsley et al. examined 1514 subjects, 19% of whom had a TP53 mutation. As in the other series, the presence of a TP53 mutation significantly affected outcome; however, the 5-year survival rate was approximately 20%, suggesting, in contrast to prior reports, that long-term cure and survival are possible, even when an unselected registry population is examined. Importantly, 311 subjects analyzed had tMDS, and a high proportion of these (38%) had TP53 mutations. In contrast to the analysis of Aldoss et al., published in this issue of Haematologica, the presence of a TP53 mutation in tMDS was strongly associated with inferior survival (hazard ratio 1.63; P<0.001) while the outcomes of tMDS patients without a TP53 mutation were similar to those of de novo MDS patients. While novel therapeutics might eventually improve non-transplant outcomes, at present, none is associated with favorable outcomes in TP53-mutated MDS. For example, a novel 10-day decitabine regimen was recently described by the Washington University group. In this experience, 21 subjects with TP53-mutated acute myeloid leukemia or MDS all attained a marrow remission (defined as <5% blasts); however, median survival was only 12.7 months for MDS patients with a TP53 mutation, and fewer than 20% were alive at 2 years.6 While transplantation was not prospectively assigned in haematologica | 2017; 102(12)

that study, transplantation improved outcomes among all patients, and TP53 mutation status did not have a negative impact on transplantation outcomes. Can we rationally incorporate molecular features into prognostic modeling for transplantation outcomes? This question was asked by the GITMO group, who devised a four-category risk score incorporating marrow blasts, cytogenetic risk, responsiveness to chemotherapy and the presence of driver mutations in ASXL1, RUNX1, or TP53.7 While this proof-of-concept risk score was devised from a relatively small number of patients (n=401), a much larger, multinational effort (n>2500) is underway to redesign the revised International Prognostic Scoring System, with incorporation of molecular risk factors analyzed on a common next-generation sequencing platform (R. Bejar, personal communication). Once validated, we can incorporate these risk scores into clinical decision making. Based on the recently published literature and the new analysis by Aldoss and colleagues, we can now conclude that transplantation even in the context of TP53 mutation should be performed in MDS patients, provided these patients are made aware of the negative prognostic impact of the mutation that their tumor cells harbor. Long-term survival can be attained in a minority of patients, and for these select few, the decision to transplant is as simple as hitting the nail on its head.

References 1. Aldoss I, Pham A, Sierra Min Li, et al. Favorable impact of allogeneic stem cell transplantation in patients with therapy-related myelodysplasia regardless of TP53 mutational status. Haematologica 2017;102(12):2030-2038. 2. Bejar R, Stevenson KE, Caughey B, et al. Somatic mutations predict poor outcome in patients with myelodysplastic syndrome after hematopoietic stem-cell transplantation. J Clin Oncol. 2014;32(25):2691-2698. 3. Yoshizato T, Nannya Y, Atsuta Y, et al. Genetic abnormalities in myelodysplasia and secondary acute myeloid leukemia: impact on outcome of stem cell transplantation. Blood. 2017;129(17):23472358. 4. Della Porta MG, GallĂŹ A, Bacigalupo A, et al. Clinical effects of driver somatic mutations on the outcomes of patients with myelodysplastic syndromes treated with allogeneic hematopoietic stem-cell transplantation. J Clin Oncol. 2016 Sep 16. [Epub ahead of print] 5. Lindsley RC, Saber W, Mar BG, et al. Prognostic mutations in myelodysplastic syndrome after stem-cell transplantation. N Engl J Med. 2017;376(6):536-547. 6. Welch JS, Petti AA, Miller CA, et al. TP53 and decitabine in acute myeloid leukemia and myelodysplastic syndromes. N Engl J Med. 2016;375(21):2023-2036. 7. Rossi M, Della Porta MG, Bacigalupo A, et al. Driver somatic mutations and transplantation decision making in patients with myelodysplastic syndrome. Blood. 2016;128(22):53.

1971

REVIEWâ&#x20AC;&#x2C6;ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Ferroportin disease: pathogenesis, diagnosis and treatment Antonello Pietrangelo

Center for Hemochromatosis, Department of Internal Medicine II, University of Modena and Reggio Emilia Policlinico, Modena, Italy

ABSTRACT

Haematologica 2017 Volume 102(12):1972-1984

Correspondence: antonello.pietrangelo@unimore.it

Received: July 10, 2017. Accepted: September 25, 2017. Pre-published: November 3, 2017.

erroportin Disease (FD) is an autosomal dominant hereditary iron loading disorder associated with heterozygote mutations of the ferroportin-1 (FPN) gene. It represents one of the commonest causes of genetic hyperferritinemia, regardless of ethnicity. FPN1 transfers iron from the intestine, macrophages and placenta into the bloodstream. In FD, loss-of-function mutations of FPN1 limit but do not impair iron export in enterocytes, but they do severely affect iron transfer in macrophages. This leads to progressive and preferential iron trapping in tissue macrophages, reduced iron release to serum transferrin (i.e. inappropriately low transferrin saturation) and a tendency towards anemia at menarche or after intense bloodletting. The hallmark of FD is marked iron accumulation in hepatic Kupffer cells. Numerous FD-associated mutations have been reported worldwide, with a few occurring in different populations and some more commonly reported (e.g. Val192del, A77D, and G80S). FPN1 polymorphisms also represent the gene variants most commonly responsible for hyperferritinemia in Africans. Differential diagnosis includes mainly hereditary hemochromatosis, the syndrome commonly due to either HFE or TfR2, HJV, HAMP, and, in rare instances, FPN1 itself. Here, unlike FD, hyperferritinemia associates with high transferrin saturation, iron-spared macrophages, and progressive parenchymal cell iron load. Abdominal magnetic resonance imaging (MRI), the key non-invasive diagnostic tool for the diagnosis of FD, shows the characteristic iron loading SSL triad (spleen, spine and liver). A non-aggressive phlebotomy regimen is recommended, with careful monitoring of transferrin saturation and hemoglobin due to the risk of anemia. Family screening is mandatory since siblings and offspring have a 50% chance of carrying the pathogenic mutation.

doi:10.3324/haematol.2017.170720 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/1972 ÂŠ2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1972

Introduction The name Ferroportin Disease (FD) refers to a clinical entity that differs from all other known forms of hereditary iron overload, including hemochromatosis (HC) [synonymous for hereditary hemochromatosis (HH)], i.e. the syndrome due to either HFE or non-HFE hemochromatosis gene mutations.1 In humans, a number of genetic disorders associate with systemic iron overload (Table 1) while others are caused by iron misdistribution and are associated with the regional accumulation of iron in subcellular compartments (e.g. mitochondria in Friedreich ataxia) or certain cell types and organs (e.g. basal ganglia in neuroferritinopathy) (Table 1). In strict terms, the latter disorders may not all qualify as true iron-overload states, as the total body iron content may not be increased. FD, which today accounts for one of the commonest forms of hereditary iron overload disorder besides HFEhemochromatosis, is characterized by a unique pathogenic basis and clinical presentation and, unlike HC, has been reported worldwide, regardless of ethnicity. Ferroportin Disease (phenotype MIM number 606069, gene/Locus; MIM number 604653; https://www.omim.org/entry/606069?search=ferroportin%20dis-

haematologica | 2017; 102(12)

Ferroportin Disease

ease&highlight=ferroportin%20disease) is due to pathogenic (usually missense) mutations of the ferroportin1 gene (FPN1; SLC40A1) which encodes the only iron exporter so far identified in mammals;2-4 lack-of-function mutations impair the iron-export capability of FPN1, particularly in cells with high iron turnover, such as tissue macrophages. Unlike the mutations causing FD, other rare mutations of FPN1 (such as N144H, C326Y, C326S and C326F),5-8 do not impair protein expression at the cell membrane or its iron export capability, but make FPN1 resistant to the inhibitory effect of hepcidin, the physiological FPN1 inhibitor (see below under Pathogenesis). This causes unchecked iron export activity of FPN1; the resulting clinical disorder is different from FD and indistinguishable from other forms of hereditary HC (Tables 1 and 2).

Definition and classification The OMIM database classifies the two forms of hereditary iron overload due to FPN1 mutations within the same taxonomic category as “hemochromatosis type 4” (https://www.omim.org/entry/606069?search=ferroportin%20disease&highlight=ferroportin%20disease). Similar terminology has then been adopted by Orphanet with the inclusion of two subcategories: hemochromatosis type 4A (referring to classic FD due to lack-of-function FPN1 mutations) and hemochromatosis type 4B (referring to FD due to gain-of-function FPN1 mutations) ( h t t p : / / w w w. o r p h a . n e t / c o n s o r / c g i bin/OC_Exp.php?Lng=EN&Expert=139491). These classifications have been incorporated into recent publications, with some variants.9,10 Disease naming and classification (taxonomy) can vary depending on different criteria, such as pathogenic genes, mechanisms, clinical manifestations, etc. Ideally, disease taxonomy (and names) should also help clinicians to recognize, diagnose, and cure diseases. In this context, the taxonomy adopted by OMIM and Orphanet, by embracing two pathogenically and clinically different disorders caused by mutations in the same gene under the term “hemochromatosis”, may fail to reach those objectives. Over the past decades, the term hemochromatosis has been inconsistently used in the literature and in clinical practice to imprecisely refer to: i) any form of body iron overload; ii) tissue iron overload causing organ damage and disease; iii) genetically determined iron overload; and, recently, iv) HFE-related iron

overload.11 Recent discoveries in the field have shown that, regardless of the underlying genetic defect, a number of hereditary iron loading disorders (i.e. those due to lossof-function mutations of HFE, TfR2, HJV, HAMP and gainof-function mutations of FPN1) belong to the same syndromic entity as they share the pathogenic basis (lack of hepcidin function-activity), biochemical expressivity (high transferrin saturation and high serum ferritin), liver pathology features (iron accumulation in parenchymal cells with iron-spared Kupffer cells until late stage), damage and disease of distinct target organs (liver, heart, endocrine glands, joints), and the therapeutic approach with optimal response to phlebotomy.11 As discussed in the following sections, each individual feature reported above is different in classic FD.1 Therefore, using the term “hemochromatosis” for the classic FD or the term “Ferroportin Disease” for FPN1-associated HC, is misleading, particularly for clinicians, since clinical suspicion, diagnostic strategy and management differ profoundly. Based on these considerations, and on our present understanding of the pathogenesis and clinical manifestations of these disorders, it is proposed that the disorder due to lack-of-function mutations of FPN1 is termed “Ferroportin Disease”, as originally described,1 and the disorder due to gain-of-function FPN1 mutations is termed “FPN-1 associated hemochromatosis” (Table 1). Instead, in analogy with other protein-related classifications (e.g. ferritinopathies; hemoglobinopathies), both disorders due to lack- and gain-of-function mutations of FPN1 may well be included in a broader taxonomic category named “ferroportinopathies”.

Historical aspects In 1996, the HFE hemochromatosis gene, whose C282Y homozygote mutation is responsible for most cases of HH in Caucasians, was identified.12 Soon after, it became apparent that not all hereditary iron overload disorders could be explained by HFE mutations, particularly in Southern Europe, where an active search for other genes linked to genetic iron overload flourished. From 2000 to 2004, all known non-HFE genes associated so far with HC, namely transferrin receptor 2 (TfR2),13 FPN1,5 hepcidin (HAMP),14 and hemojuvelin (HJV)15 were identified. A few years earlier, in 1999, a distinct and somehow unusual phenotype had been reported in a large family with hereditary iron overload from Italy. Selective iron

Table 1. Human hereditary disorders associated with iron overload and iron mis-distribution.

Disorder/cause

Iron overload Pattern of iron accumulation

Hereditary hemochromatosis (HFE-TfR2-, HJV-, HAMP-, FPN1-associated) Ferroportin Disease

Systemic (iron accumulation in parenchymal cells) Systemic (preferential iron accumulation in macrophages) Aceruloplasminemia Systemic Atransferrinemia Systemic DMT-1 deficiency Regional (mainly liver) H-ferritin-related iron overload Systemic Hereditary iron-loading anemias with inefficient Systemic (early iron erythropoiesis accumulation in hepatocytes due to increased iron absorption) haematologica | 2017; 102(12)

Iron mis-distribution Disorder/cause Pattern of iron accumulation X-linked sideroblastic anemias

Systemic (mitochondria)

Friedreich ataxia

Systemic (mitochondria)

Neuroferritinopathy

Regional (brain)

1973

A. Pietrangelo

loading of liver macrophages, hyperferritinemia co-existing with normal-low transferrin saturation, and tendency to anemia after intense phlebotomy were the hallmarks of the disorder.16 In 2001, all affected family members were reported to be heterozygous for a c. 230 C→A substitution resulting in the replacement of alanine 77 with aspartate in FPN1.17 This entity was subsequently named FD.1 On the other hand, FPN1-related HC, due to a gain-offunction mutation of FPN1 (p.N144H), was first reported by Njajou et al. in 2001.5 Yet, it is worth mentioning that the first clinical description of an “autosomal dominant” form of classic HC had been already reported by Eason et al. in a Melanesian kindred in 1990.18 In this same population, Arden et al.19 have later linked the HC phenotype to the NI44T gain-of-function mutation of FPN1.

Ferroportin biology and physiology and FD FPN1, the product of the FPN1 (SLC40A1) gene, transfers iron from the external milieu (i.e. maternal blood or intestinal lumen) and from internal sites of iron storage and recycles it into the bloodstream. In fact, it is highly expressed in liver and spleen macrophages, the luminal site of enterocytes and placental syncytiotrophoblasts.2-4 FPN1 is regulated at different levels by a number of factors, including transcriptionally by heme,20 translationally by the iron-regulatory proteins (IRPs),21 and posttranslationally mainly by hepcidin, the iron hormone. Hepcidin

is produced by the liver in response to iron, inflammation, and a variety of stressors.22-25 Hepcidin binds to the extracellular loop of FPN1 and triggers its ubiquitinylation on lysine residues located in the intracellular domain leading to internalization and degradation in lysosomes.26-28 This mechanism allows a finely-tuned control of iron efflux from enterocytes and macrophages toward the bloodstream when more iron is needed during active erythropoiesis (in this case, hepcidin synthesis is inhibited by erythroid signals), or blood iron must be controlled due to pathogen proliferation/growth or incipient iron overload (here hepcidin synthesis is induced by inflammatory or iron mediators, respectively) (reviewed by Drakesmith, Nemeth and Ganz29). FPN1 topology and membrane organization have long been addressed with controversial results concerning localization of the N- and C-terminal extremities and number of transmembrane segments.30-39 Recently, an inward-open conformation of the transporter has been predicted,34,37 with a cluster of residues lying in the central core of the protein important for iron traffic and consistent with an iron-binding site37 and residues involved in hepcidin binding fully accessible in the outward-open model.37 The inward-open form may represent the resting state of the protein, and the outward-open state as a conformation attainable only in the presence of intracellular iron, i.e. when FPN1 shuttles between the two conforma-

Table 2. Main features of Ferroportin Disease and other hereditary iron overload disorders in humans.

Disorder

Affected gene (symbol / location)

Known or postulated gene product function

I. Ferroportin Disease Solute carrier Iron exporter family 40 (iron-regulated (SLC40A1 / 2q32) transporter), member 1 from cells including macrophages, enterocytes, syncityotrophoblasts II. Hemochromatosis Hemochromatosis Hepcidin regulator gene (HFE / 6p21.3) Transferrin-receptor 2 (TfR2 / 7q22)

Epidemiology

Genetics

Mechanism for Clinical increased cellular onset iron deposits (decade)

Affects Caucasians and non Caucasians

Autosomal dominant

Iron retention due to decreased iron export

Affects Caucasians and non Caucasians Solute carrier Iron exporter Affects Caucasians family 40 (ironfrom cells including and non Caucasians regulated transporter), macrophages, member 1 enterocytes, (SLC40A1 / 2q32) syncityotrophoblasts Hepcidin antimicrobial Degradation Affects Caucasians peptide of ferroportin and non Caucasians (HAMP /19q13.1) and downregulation of iron efflux from cells Hemojuvelin Hepcidin Affects Caucasians (HJV/ 1p21) regulator and non Caucasians III. Aceruloplasminemia Ceruloplasmin Iron efflux Affects Caucasians (CP / 3q23-q25) from cells and non Caucasians IV. A (hypo)transferrinemiaTransferrin (Tf / 3q21) 1974

• Iron transport in the Affects Caucasians bloodstream and non Caucasians

Clinical course

• Liver disease • Marginal anemia

Mild

4th-5th

Affects Caucasians of North European descent

Hepcidin regulator

Any

Main clinical manifestation

Autosomal recessive

Mildsevere

3rd-4th

Autosomal dominant

Autosomal recessive

Increased iron accumulation in parenchymal cells due to increased transferrin and non-transferrin bound iron import

4th-5th

2nd-3rd

Autosomal recessive

Decreased iron efflux

2nd-3rd

Autosomal recessive

Increased iron influx

1st-2nd

•Liver Disease • Arthropathy • Cardiomyopathy • Endocrinopathy • Hypogonadism and cardiomyopathy Severe • Liver disease

• Neurological manifestations • Anemia Anemia

haematologica | 2017; 102(12)

Severe

Ferroportin Disease

tions (Figure 1A). The selective binding of hepcidin to the outer-facing conformation would, therefore, guarantee that FPN1 degradation can occur only when intracellular iron is abundant37 and actively pumped through the channel38 (Figure 1B). Recently, the crystal structures of a bacterial homolog of FPN1, BbFPN, has been resolved in both the outward- and inward-facing states, and a homology model with human FPN1 has been developed.39 According to the Au, FPN1 has 12 TM helices, as previously predicted,32 and is divided into two halves, one forming the N lobe, the other the C lobe connected by a long cytosolic loop, with a central cavity between the lobes that is open towards the extracellular side and not accessible from the intracellular side (Figure 1A). FPN1 undergoes an intra-domain conformational rearrangement during the transport cycle. When hepcidin enters the central cavity

between the N and C lobes, and interacts with the hepcidin-binding site located on the C lobe, it elicits two effects: a) it increases the accessibility of the intracellular loops that harbor the ubiquitination sites to the ubiquitin ligases; and b) it arrests the conformational transition of FPN1 from the outward-facing state to the inward-facing state, inhibiting the access of iron from the cytoplasm to the substrate-binding site within the intracellular gate (Figure 1B).39

Molecular pathogenesis The general pathophysiological basis of FD is well defined and relies on the impaired iron export from the iron storage/recycling site (particularly macrophages) towards the bloodstream. Figure 2 shows the basic iron transport defect of the FD as opposed to FPN1-associated

Figure 1. Biology of ferroportin and postulated pathobiology of Ferroportin Disease (FD). (A) Structure-function relationship of iron-export ferroportin activity.39 (B) Putative mechanisms of hepcidin binding to FPN and its degradation.39 (C) Postulated basis for FD. (Upper panel) In cells undergoing relatively low iron flux, such as enterocytes, the product of the FPN wild-type allele is able to reach the plasma membrane and export iron. For clarity, mutated FPN1 was not depicted at the cell surface: based on previous in vitro work, it has been postulated that some mutant FPN1 can still reach the cell surface and preserve some iron-transport competence, but this is still controversial. (Lower panel) In cells undergoing high iron turnover, such as macrophages, increased requests for iron export impose high demands on FPN traffic leading to a 'traffic jam' within the endocytic/plasmamembrane and degradation compartments and inappropriately low wild-type allele product targeting to the cell membrane.54 (D) Postulated effect of FPN mutations that affect formation of the intracellular gate and access to the iron binding site.39

haematologica | 2017; 102(12)

1975

A. Pietrangelo

HC. In the latter cases, as discussed above, mutations that affect the hepcidin binding site and or FPN1 ubiquitination result in reduced FPN 'sensitivity' to hepcidin, leading to the FPN1-related HC phenotype. This has been nicely exemplified by an informative murine model corresponding at the mutation of the hepcidin binding site.40 In spite of these advances, the molecular pathogenesis of FD has long been elusive. A number of in vitro studies, mostly using over-expressed exogenous wild-type and mutant FPN1 in a variety of cell lines, have investigated FPN1 biology and function and the effect of different FPN1 mutants on protein traffic and iron-transfer capability, although with conflicting results, depending on the cell line or methodology used.30-32,34,35,38,41-47 In this context, it has been actively debated whether FPN1 haplo-insufficiency would explain FD or whether the disease results from a dominant-negative effect. It has been argued that if haplo-insufficiency was the explanation for FD, then nonsense mutations should also result in the disorder; however, so far, the vast majority of reported mutations in FD are missense mutations.48 In addition, a targeted gene deletion in the murine Fpn1 gene has little effect in heterozygous animals,49 whereas the flatiron (ffe) mouse with a missense mutation in Fpn1 that affects its localization and iron export activity when over-expressed in vitro, present a phenotype similar to human patients.50 In studies using exogenous tagged protein in vitro, Fpn1 forms multimers and mutant Fpn1 prevents cell membrane localization of wild-type Fpn1.33,42,50 A multimeric protein, through a dominant-negative effect, would better explain the autosomal dominant trait of FD. However, other studies from different groups have provided experimental evidence to support the opposite conclusion and showed that Fpn1 is a monomer in cultured cells35,51,52 and in vivo.53 More recently, Sabelli et al.,54 using for the first-time cultured macrophages from FD patients, found that endogenous FPN1 shows a similar localization to that in donor macrophages, except for greater accumulation in lysosomes, suggesting a higher degradation rate of mutant FPN1. Unexpectedly, and contrary to previous studies using over-expressed mutant protein in cell lines, FPN1 in FD macrophage circulates in the early endocytic compartment, does not multimerize, it reaches the plasma membrane, is iron-transport competent (although to a lesser extent than normal macrophages), and is promptly internalized and degraded upon exposure to hepcidin. However, when FD macrophages are exposed to large amounts of heme iron, in contrast to donor macrophages, FPN1 can no longer reach the cell surface, leading to marked intracellular iron retention. Based on these observations, a model of FD has been proposed in which FPN1 monomers, in spite of the fact that half proteins are mutated, can still reach the cell surface and export iron in cells that are exposed in vivo to a relatively low flux of iron, such as enterocytes (Figure 1C).54 On the contrary, in cells undergoing high iron turnover in vivo, such as tissue macrophages, sufficient FPN1 is prevented from reaching the plasma membrane, possibly due to a 'traffic jam' in the degradation and/or endocytic cycling pathways. This model is consistent with the clinical manifestation of FD characterized by early iron accumulation in hepatic Kupffer cells and normal transferrin saturation, indicating that mutant FPN1 activity does not limit intestinal iron transfer; the latter becomes critically low in young females at menarche or after aggressive phlebotomy, when high 1976

iron demands for erythropoiesis likely impose increased FPN1 traffic/cycling within tissue macrophages.1,16,17 (See below under Clinical Manifestations and Diagnosis and Treatment). This study did not address the question as to whether mutant or only wild-type FPN1 reaches the plasma membrane and whether mutant-FPN1 is transport competent. Previous studies have not been conclusive. Exogenously expressed p.A77D and p.Val162del FPN1 mutants have been found to be iron-transport incompetent in all studies, but able to reach the cell membrane in some,34,35,38,41,43 and not in others.30,31 The p.G80S FPN1 mutant has been localized at the cell surface in two published studies,43,55 and found iron transport competent in one55 and incompetent in the other.43 According to Taniguchi et al.,39 the mutation sites associated with FD are mainly mapped onto the inter-lobe interface, mostly on the intracellular side, and form the intracellular gate. These mutations would, therefore, destabilize the inter-lobe interactions, thereby affecting the stable formation of the intracellular gate and reducing the iron transport activity of FPN1 (Figure 1D). It is possible that different mutants differently affect iron transport capability of FPN1; while this may be better overcome by the normal allele product in cells with low iron turnover such as enterocytes or hepatocytes, it may be further hampered in cells like macrophages where the additional 'traffic jam' in the endocytic-plasmamembrane compartment will aggravate the basic defect (see above).

Genetics and epidemiology A list of published mutations associated with FD and FPN1-related HC is reported in Table 3.5-8,17,43,56-102 Numerous mutations of the FPN1 gene have been identified so far in probands with the classic FD phenotype of French-Canadian, Melanesian, Thai, Japanese and European heritages. A few common FPN1 mutations have been reported in independent pedigrees, in different countries (e.g. Val192del;56,60,72-79,92 A77D,17,59,60 G80S.43,55,56,61-63 It is now believed that the most frequently reported FPN1 mutations, such as the p.Val162del, are more frequently identified than other SLC40A1 mutations because they have occurred multiple times in isolated populations rather than occurring once and spreading to different populations, as indicated by the identification of a de novo p.Val162del variant in an isolated case of FD.79 FPN1 variants are highly prevalent in African populations. The first prevalent FPN1 variant reported in Africans and Black Americans was the common Q248H polymorphism (p.Gln248His).82,83,100-102 Interestingly, global analysis of variants in the SLC40A1 gene (which includes mutations associated with both the FD and FPN1-associated HH) revealed an allele frequency of 0.0364%, giving a predicted pathogenic genotype carrier rate of 1 in 1373, a figure that approaches the frequency of HFE-HC.103 This was largely due to the relatively high allele frequencies for two SLC40A1 variants (p.Asp270Val84,85 and p.Arg371Trp56) in the African populations; the predicted SLC40A1 pathogenic genotype carrier rate of these two variants is 1 in 197 among the African population.103 The Q248H,101,102 the p.Asp270Val and the p.Arg371Trp and other FPN1 polymorphic variants84 may also predispose to iron overload; but no clear evidence for this has been provided (e.g. lack of functional studies), while the possibility remains that, because of the small sample size, these observations could haematologica | 2017; 102(12)

Ferroportin Disease

Table 3. Disease-associated mutations of the FPN1 gene.

Gene SLC40A1 (RefSeq NM_014585.5, NP_055400.1) A. Ferroportin disease Nucleotide change Amino acid change 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49.

c.134C>A c.205G>A c.206C>T c.212C>T c.214G>T c.230C>A c.238G>A c.239G>T c.252C>G c.262A>G c.263G>C c.386T>C c.454A>T c.469G>A c.469G>T c.470A>C c.470A>G c.473G>T c.474G>T c.484_486del c.521A>T c.532C>G c.533G>A c.539T>C c.542A>T c.546G>T c.553A>G c.554A>C c.610G>A c.689C>A c.695C>A c.698T>C c.744G>T c.797T>C c.800G>A c.809A>T c.968G>T c.1035G>C c.1051A>G c.1111C>T c.1112G>A c.1328C>T c.1402G>A c.1466G>A c.1467A>C c.1468G>A c.1469G>A c.1520A>G c.1681A>G

p.Ala45Glu p.Ala69Thr p.Ala69Val p.Ser71Phe p.Val72Phe p.Ala77Asp p.Gly80Ser p.Gly80Val p.Asp84Glu p.Arg88Gly p.Arg88Thr p.Leu129Pro p.Ile152Phe p.Asp157Asn p.Asp157Tyr p.Asp157Ala p.Asp157Gly p.Trp158Leu p.Trp158Cys p.Val162del p.Asn174Ile p.Arg178Gly p.Arg178Gln p.Ile180Thr p.Asp181Val p.Gln182His p.Asn185Asp p.Asn185Thr p.Gly204Ser p.Thr230Asn p.Ala232Asp p.Leu233Pro p.Gln248His p.Met266Thr p.Gly267Asp p.Asp270Val p.Gly323Val p.Leu345Phe p.Ile351Val p.Arg371Trp p.Arg371Gln p.Pro443Leu p.Gly468Ser p.Arg489Lys p.Arg489Ser p.Gly490Ser p.Gly490Asp p.His507Arg p.Arg561Gly

Type of variation*

Phenotype

Reference

Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Deletion Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense Missense

ferroportin diseasea ferroportin disease ferroportin diseasea ferroportin diseasea ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin diseasea ferroportin diseasea ferroportin diseasea ferroportin disease ferroportin disease ferroportin diseasea ferroportin diseasea ferroportin disease ferroportin disease ferroportin disease ferroportin diseasea ferroportin diseasea ferroportin diseasea ferroportin diseasea ferroportin diseasea ferroportin diseasea ferroportin disease ferroportin disease ferroportin disease ferroportin disease ferroportin diseasea ferroportin diseasea

56 57 56 56 58 17,59,60 43,55,56,61-63 64 91 56, 65 66 67 68 58 69 70 56,69,71 56 47 56,60,72-79,92 61 77 56,65 66 56,64,69 56, 71 56, 80 56 56 69 81 68,69 82,83,100-102 69 64 84,85 71 69 69 56 56 69 86 46 87 56,65,69 88, 69 89 90 continued on the next page

haematologica | 2017; 102(12)

1977

A. Pietrangelo continued from the previous page

B. Ferroportin1-associated hemochromatosis 1.

c.-59_-45del

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

c.190T>A c.190T>C c.430A>C c.430A>G c.431A>C c.718A>G c.977G>A c.977G>C c.977G>T c.1014T>G c.1502A>G c.1510G>A c.1520A>G

p.Tyr64Asn p.Tyr64His p.Asn144His p.Asn144Asp p.Asn144Thr p.Lys240Glu p.Cys326Tyr p.Cys326Ser p.Cys326Phe p.Ser338Arg p.Tyr501Cys p.Asp504Asn p.His507Arg

Deletion (heterozygous state) Missense (heterozygous state) Missense (heterozygous state) Missense (heterozygous state) Missense (heterozygous state) Missense (heterozygous state) Missense (heterozygous state) Missense (heterozygous state) Missense (heterozygous state) Missense (heterozygous state) Missense (heterozygous state) Missense (heterozygous state) Missense (heterozygous state) Missense (heterozygous state)

hemochromatosis

hemochromatosisb hemochromatosis hemochromatosisb hemochromatosisb hemochromatosisb hemochromatosis hemochromatosis hemochromatosisb hemochromatosisb hemochromatosisb hemochromatosis hemochromatosis hemochromatosisb

93 94 5 6 19 95 96, 97 7 8 98 99 69 47

*All reported ferroportin mutations are at the heterozygous state (autosomal dominant trait). aReported data do not allow a classic Ferroportin Disease phenotype to be assigned. bPatients carrying a ferroportin missense mutation in whom a classic hemochromatosis phenotype is confirmed by a liver biopsy.

be attributable to chance or that the polymorphisms identified may be in linkage disequilibrium with other diseasecausing loci. Yet, taken together, the collected data make FPN1 the gene most frequently associated with hereditary hyperferritinemia in Africans.

Clinical manifestations and diagnosis As discussed, FD is caused by loss-of-function mutations in FPN1. These mutations impair iron export, particularly from reticuloendothelial macrophages. The result is iron accumulation in macrophages of the spleen, liver, and bone (reflected by high levels of SF) (Figure 3). At liver histology, parenchymal cells of these organs are largely spared (Figure 3), but discrete hepatocytic iron deposits are also appreciable, due to defective FPN1 activity in hepatocytes, even at early stages.16 Clinical presentation appears heterogeneous, but overall expressivity is milder than classic HC, and the associated liver disease is usually not as severe (Table 2 and Figure 3).1,16,17,56 As occurs in classic forms of HFE HC, also in the FD host factors (menses, blood loss, etc.), co-inheritance of mutations of other irongenes or variants in genes associated to antioxidant defense and organ fibrosis, and associated pathological conditions (metabolic syndrome, viral hepatitis, etc.) may all affect the phenotype. Hypochromic anemia is not uncommon in young menstruating females. Owing to the mild clinical expressivity reported in the literature, doubts have been raised on the penetrance of the genetic defect and the rationale for iron-removal therapy. However, there is limited and usually not detailed clinical information in the published reports; this, and the lack of prospective studies, still hamper our understanding of the actual clinical impact of the disorder. In the only report published so far, 6 members of the pedigree in which FD was first described16 were followed for 11-24 years.104 The proband, aged 83, who had carried an occult HBV infection since the age of 56, developed a liver cancer in a non-cirrhotic liver after discontinuation of a 20-year long phlebotomy program; 2 siblings, who had also interrupted treatment, showed a fibrosis progression. These 1978

clinical data, while of interest, do not allow definite conclusions to be drawn as to a pathogenic link between iron accumulation in FD and liver damage and disease. The hallmark of classic FD is marked iron accumulation in Kupffer cells (Figure 3). Kupffer cells are vital to the production of fibrogenic mediators, to immunological tumor surveillance, and disposal of transformed hepatocytes.105 Selective and massive iron overload may impair these activities and favor fibrogenesis and carcinogenesis. Moreover, as discussed above, hepatocytic iron accumulation also takes place in FD, although to a much lesser extent than in HFE- and non-HFE HC, and the established pro-oxidant damaging activity of iron in parenchymal cells may also contribute to disease progression. Unlike HFE-HC, the pattern of inheritance of FD is autosomal dominant. Therefore, either parent carries the pathogenic mutation of FPN1 and presents with unexplained hyperferritinemia. In addition, the proband carries a 50% risk of having an affected child. The disease must be suspected in any individual with unexplained hyperferritinemia and low-normal transferrin saturation (TS), or non-parenchymal cell siderosis at liver biopsy or liver and spleen iron accumulation at MRI (Table 4). Hyperferritinemia in FD appears very early in life, and unexplained hyperferritinemia with normal TS in a child should prompt MRI evaluation to evaluate iron accumulation in liver, spleen and bone marrow106 (see below). Figure 4 shows a proposed algorithm for the diagnosis of FD. If hyperferritinemia associates with high TS (as confirmed in at least two sequential determinations) but in the absence of anemia, a typical picture of HFE- and non HFEHC (including FPN1-associated HC due to gain-of-function FPN1 mutations) is ruled out a priori. If hyperferritinemia associates with high TS and anemia, a typical picture of hereditary hemoglobinopathies and red cell defects or atransferrinemia, FD is again ruled out (Figure 4). On the contrary, in subjects with increased serum ferritin and low or normal TS, the workup should focus on common causes of secondary hyperferritinemia and other rare causes of hereditary hyperferritinemia to confirm the haematologica | 2017; 102(12)

Ferroportin Disease

Figure 2. The basis for abnormal iron transfer into the bloodstream in Ferroportin Disease as opposed to FPN-associated hereditary hemochromatosis.

diagnosis of FD (Figure 4). First, common causes of hyperferritinemia, such as metabolic disorders, inflammation, cancer, etc., should be considered. If they are not found, or if the hyperferritinemia persists after their treatment, the next step depends on whether or not anemia is present. In the absence of overt anemia, if liver and spleen iron content are increased at MRI or liver biopsy shows prominent Kupffer cell iron load, FD disease should be considered and genetic testing performed for confirmation of diagnosis (Figure 4). Another common cause of hereditary hyperferritinemia with normal TS associated with iron accumulation and anemia is Gaucher disease, usually associated with hepatosplenomegaly, cytopenia, abnormal coagulation, bone disease, and neuropathic manifestations.107 In the absence of body iron accumulation, but in the presence of elevated SF levels and normal TS, autosomal dominant hyperferritinemia with cataract (due to mutations of the iron responsive element in the 5' untranslated region of the L ferritin mRNA108) or without cataract,109 should be considered. If overt anemia is present, but TS is normal/low, aceruloplasminemia should be suspected, a rare autosomal recessive disease due to loss of function haematologica | 2017; 102(12)

mutations in ceruloplasmin (CP) and resulting in iron overload in the liver and pancreas, and progressive neurodegeneration, diabetes and retinal degeneration.110 Brain MRI with typical iron accumulation in basal ganglia and thalamus may help confirm the diagnosis. As mentioned above, another rare genetic disease presenting with hyperferritinemia and anemia is atransferrinemia/ hypotransferrinemia111 which, however, is characterized by increased transferrin saturation due to extremely low serum transferrin levels. Differential diagnosis mainly includes the classic (HFE) and non-classic (TfR2, HAMP, HJV and FPN1) forms of HH, all characterized by early and progressive increase of TS followed by elevation of serum ferritin as iron accumulation increases in parenchymal cells of the liver, pancreas, heart and other organs (Table 2). As discussed, unlike HH, in FD clinical expressivity is milder. Abdominal MRI is a useful non-invasive tool to categorize and diagnose the disorder, as it can differentiate patients with FD, characterized by the SSL triad (spleen, spine, liver) iron retention (Figure 5B), from all other forms of HH, including FPN1-HC, associated with liver iron 1979

A. Pietrangelo

overload but normal spleen and bone marrow iron content (Figure 5D).112

Treatment Venesection is the cornerstone of therapy also in FD, but it may not be tolerated equally in all patients, and low TS with anemia may be rapidly established despite SF still being elevated.1 Macrophage iron overload is very resistant to iron withdrawal in this disorder, even in patients who are apparently well-treated (Figure 5C). Therefore, unlike HH, not only serum ferritin, but especially TS should be carefully monitored during therapy. In addition, therapy should not aim at reaching the usual HH targets for iron depletion (TS below 20%, SF 50 ng/L or slight anemia) but be more conservative. There are no studies on the optimal phlebotomy schedule in FD. In practical

terms, a monthly/bi-monthly phlebotomy session for 1-2 years, depending on the underlying mutation, allows an acceptable state of iron depletion to be reached, while maintenance therapy (usually a phlebotomy session every 4-6 months) should be continued for life. A reasonable target for therapy is an SF level of 100-200 ng/mL. In certain cases, such Ft values may still reflect some iron loading of tissue macrophages (Figure 5C), but the associated clinical risk is negligible. Ideally, the optimal target is the lowest acceptable ferritin level for TS and hemoglobin levels not below the lower limit of normal. The (controversial) dietary restrictions sometimes recommended for patients with HH (avoiding vitamin C or iron-rich or enriched foods) do not apply to FD due to the different pathogenic basis as compared to HH: normal/sufficient enterocyte iron absorption and normal/marginally increased iron

Figure 3. The different stages and outcomes of “iron retention” in Ferroportin Disease versus “iron accumulation” in FPN1-associated hemochromatosis (HC). Liver histology pictures are reproduced with the permission of Sabelli et al.54

Table 4. Suspecting and diagnosing Ferroportin Disease.

Sex Either

Ethnicity

Age, y

Any

10-80

When to suspect Signs • Unexplained hyperferritinemia and normal or inappropriately low transferrin saturation • Isolated hyperferritinemia in father or mother • Sinusoidal (Kupffer cell) iron overload at liver biopsy or spleen (and liver), iron accumulation at MRI in patients with unexplained hyperferritinemia and normal or inappropriately low transferrin saturation

Essential for diagnosis Heterozygosity for FPN mutation and hyperferritinemia with normal or inappropriately low transferrin saturation and Kupffer cell iron overload at liver biopsy

y: years; MRI: magnetic resonance imaging.

1980

haematologica | 2017; 102(12)

Ferroportin Disease

accumulation in parenchymal cells in the FD versus increased iron absorption and marked iron accumulation in parenchymal cells in HH. Iron chelation may be an option in selected cases.79 Siblings of patients with FD, like their offspring, must undergo screening since they have a 50% chance of being susceptible.

Conclusions FPN1 is a multipass membrane iron-exporter that has evolved in mammals to assure sufficient iron delivery from the external milieu and internal sites of iron storage and recycling to the bloodstream, mainly to support the erythron activity. Overall, the FPN1/SLC40A1 gene is essential for humans and total loss (homozygote muta-

tion) of its product is incompatible with life.49 Loss-offunction of one FPN1 allele in humans results in FD, characterized by a preserved intestinal iron export activity but compromised iron export from tissue macrophages. This leads to progressive iron retention in liver, spleen and bone marrow macrophages, resulting in inappropriately low iron delivery to circulating transferrin and marginal ironrestricted erythropoiesis that may result in overt anemia when bone marrow demands are increased (e.g. menarche, aggressive phlebotomy regimen). Gain-of-function mutations of FPN1 preclude the inhibitory activity of hepcidin, thereby leading to unrestricted iron transfer to the bloodstream and causing a rare form of HH. The pathogenic, biochemical and clinical signatures of FD are symmetrical and opposite to HFE and non HFEHH: normal/sufficient enterocyte iron absorption, marked iron accumulation in non parenchymal cells in

Figure 4.Diagnostic algorithm for Ferroportin Disease and hereditary hyperferritinemia. ACD/AI: anemia of chronic disease/anemia of inflammation. *Gaucher disease may present with or without siderosis depending on the disease stage. **Advanced ACD/AI may also present with siderosis at MRI (usually spleen and bone marrow).

haematologica | 2017; 102(12)

1981

A. Pietrangelo A

D Figure 5. Abdominal magnetic resonance imaging (MRI) pattern of Ferroportin Disease (FD). MRI scans. T2*-weighted gradient-echo sequences were used to detect iron accumulation. (A) Normal subject. (B) FD. (C) FD after completion of phlebotomy program (note that excess iron is still detectable in the liver and spine in spite of normal serum ferritin and transferrin saturation levels). (D) Ferroportin-associated hereditary hemochromatosis: iron accumulation involves only the liver and spares the spleen and spine (arrows).

the FD versus increased iron absorption and marked iron accumulation in parenchymal cells in HH; hyperferritinemia with normal/low transferrin saturation in FD versus hyperferritinemia and high transferrin saturation in HH; intolerance to aggressive phlebotomy regimens in FD versus optimal response to intense phlebotomy in HH; mild and benign clinical course in FD versus potentially severe clinical expressivity in HH; vertical hereditary transmission and presentation at each generation of FD versus recessive transmission of most forms of HH (except FPN1-HH).

References 1. Pietrangelo A. The ferroportin disease. Blood Cells Mol Dis. 2004;32(1):131-138. 2. Abboud S, Haile DJ. A novel mammalian iron-regulated protein involved in intracellular iron metabolism. J Biol Chem. 2000;275(26):19906-19912. 3. Donovan A, Brownlie A, Zhou Y, et al. Positional cloning of zebrafish ferroportin1 identifies a conserved vertebrate iron exporter. Nature. 2000;403(6771):776-781. 4. McKie AT, Marciani P, Rolfs A, et al. A novel duodenal iron-regulated transporter, IREG1, implicated in the basolateral transfer of iron to the circulation. Mol Cell. 2000;5(2):299309. 5. Njajou OT, Vaessen N, Joosse M, et al. A mutation in SLC11A3 is associated with autosomal dominant hemochromatosis. Nat Genet. 2001;28(3):213-214. 6. Wallace DF, Clark RM, Harley HA, Subramaniam VN. Autosomal dominant iron overload due to a novel mutation of ferroportin1 associated with parenchymal iron loading and cirrhosis. J Hepatol. 2004;40(4):710-713. 7. Sham RL, Phatak PD, West C, Lee P,

1982

10.

11. 12.

13.

While the molecular pathogenesis of FD is becoming more and more defined, the long-term effect of massive iron retention in tissue macrophages in the setting of chronic inflammatory/infectious or degenerative disorders is still unclear. Today, isolated or unexplained hyperferritinemia represents one of the commonest reasons for referral. Knowing that FD is one of the most frequent genetic causes of hyperferritinemia, regardless of ethnicity, it is important to maintain a high diagnostic suspicion for this disorder.

Andrews C, Beutler E. Autosomal dominant hereditary hemochromatosis associated with a novel ferroportin mutation and unique clinical features. Blood Cells Mol Dis. 2005;34(2):157-161. Chen SR, Yang LQ, Chong YT, et al. Novel gain of function mutation in the SLC40A1 gene associated with hereditary haemochromatosis type 4. Intern Med J. 2015;45(6):672-676. Brissot P, Loreal O. Iron metabolism and related genetic diseases: A cleared land, keeping mysteries. J Hepatol. 2016;64(2): 505-515. Hollerer I, Bachmann A, Muckenthaler MU. Pathophysiological consequences and benefits of HFE mutations: 20 years of research. Haematologica. 2017;102(5):809-817. Pietrangelo A. Hemochromatosis: an endocrine liver disease. Hepatology. 2007;46(4):1291-1301. Feder JN, Gnirke A, Thomas W, et al. A novel MHC class I-like gene is mutated in patients with hereditary haemochromatosis. Nat Genet. 1996;13(4):399-408. Camaschella C, Roetto A, Cali A, et al. The gene TFR2 is mutated in a new type of haemochromatosis mapping to 7q22. Nat Genet. 2000;25(1):14-15.

14. Rivard SR, Lanzara C, Grimard D, et al. Juvenile hemochromatosis locus maps to chromosome 1q in a French Canadian population. Eur J Hum Genet. 2003;11(8):585589. 15. Papanikolaou G, Samuels ME, Ludwig EH, et al. Mutations in HFE2 cause iron overload in chromosome 1q-linked juvenile hemochromatosis. Nat Genet. 2004;36(1): 77-82. 16. Pietrangelo A, Montosi G, Totaro A, et al. Hereditary hemochromatosis in adults without pathogenic mutations in the hemochromatosis gene [see comments]. N Engl J Med. 1999;341(10):725-732. 17. Montosi G, Donovan A, Totaro A, et al. Autosomal-dominant hemochromatosis is associated with a mutation in the ferroportin (SLC11A3) gene. J Clin Invest. 2001;108(4):619-623. 18. Eason RJ, Adams PC, Aston CE, Searle J. Familial iron overload with possible autosomal dominant inheritance. Aust NZ J Med. 1990;20:226-230. 19. Arden KE, Wallace DF, Dixon JL, et al. A novel mutation in ferroportin1 is associated with haemochromatosis in a Solomon Islands patient. Gut. 2003;52(8):1215-1217. 20. Marro S, Chiabrando D, Messana E, et al.

haematologica | 2017; 102(12)

Ferroportin Disease

21.

22.

23.

24.

25.

26.

27.

28.

29. 30.

31.

32. 33. 34.

35.

36.

37.

Heme controls ferroportin1 (FPN1) transcription involving Bach1, Nrf2 and a MARE/ARE sequence motif at position 7007 of the FPN1 promoter. Haematologica. 2010;95(8):1261-1268. Lymboussaki A, Pignatti E, Montosi G, et al. The role of the iron responsive element in the control of ferroportin1/IREG1/MTP1 gene expression. J Hepatol. 2003;39(5):710715. Krause A, Neitz S, Magert HJ, et al. LEAP-1, a novel highly disulfide-bonded human peptide, exhibits antimicrobial activity. FEBS Lett. 2000;480(2-3):147-150. Pigeon C, Ilyin G, Courselaud B, et al. A new mouse liver-specific gene, encoding a protein homologous to human antimicrobial peptide hepcidin, is overexpressed during iron overload. J Biol Chem. 2001;276(11): 7811-7819. Park CH, Valore EV, Waring AJ, Ganz T. Hepcidin, a urinary antimicrobial peptide synthesized in the liver. J Biol Chem. 2001;276(11):7806-7810. Vecchi C, Montosi G, Zhang K, et al. ER stress controls iron metabolism through induction of hepcidin. Science. 2009;325(5942):877-880. Nemeth E, Tuttle MS, Powelson J, et al. Hepcidin regulates cellular iron efflux by binding to ferroportin and inducing its internalization. Science. 2004;306(5704):20902093. Qiao B, Sugianto P, Fung E, et al. Hepcidininduced endocytosis of ferroportin is dependent on ferroportin ubiquitination. Cell Metab. 2012;15(6):918-924. Ross SL, Tran L, Winters A, et al. Molecular Mechanism of Hepcidin-Mediated Ferroportin Internalization Requires Ferroportin Lysines, Not Tyrosines or JAKSTAT. Cell Metab. 2012;15(6):905-917. Drakesmith H, Nemeth E, Ganz T. Ironing out Ferroportin. Cell Metab. 2015;22(5):777787. Schimanski LM, Drakesmith H, Merryweather-Clarke AT, et al. In vitro functional analysis of human ferroportin (FPN) and hemochromatosis-associated FPN mutations. Blood. 2005;105(10):4096-4102. Drakesmith H, Schimanski LM, Ormerod E, et al. Resistance to hepcidin is conferred by hemochromatosis-associated mutations of ferroportin. Blood. 2005;106(3):1092-1097. Liu XB, Yang F, Haile DJ. Functional consequences of ferroportin 1 mutations. Blood Cells Mol Dis. 2005;35(1):33-46. De Domenico I, Ward DM, Musci G, Kaplan J. Evidence for the multimeric structure of ferroportin. Blood. 2007;109(5):2205-2209. Wallace DF, Harris JM, Subramaniam VN. Functional analysis and theoretical modeling of ferroportin reveals clustering of mutations according to phenotype. Am J Physiol Cell Physiol. 2010;298(1):C75-84. Rice AE, Mendez MJ, Hokanson CA, Rees DC, Bjorkman PJ. Investigation of the biophysical and cell biological properties of ferroportin, a multipass integral membrane protein iron exporter. J Mol Biol. 2009;386(3):717-732. Le Gac G, Ka C, Joubrel R, et al. Structurefunction analysis of the human ferroportin iron exporter (SLC40A1): effect of hemochromatosis type 4 disease mutations and identification of critical residues. Hum Mutat. 2013;34(10):1371-1380. Bonaccorsi di Patti MC, Polticelli F, Cece G, et al. A structural model of human ferroportin and of its iron binding site. FEBS J. 2014;281(12):2851-2860.

haematologica | 2017; 102(12)

38. Praschberger R, Schranz M, Griffiths WJ, et al. Impact of D181V and A69T on the function of ferroportin as an iron export pump and hepcidin receptor. Biochim Biophys Acta. 2014;1842(9):1406-1412. 39. Taniguchi R, Kato HE, Font J, et al. Outward- and inward-facing structures of a putative bacterial transition-metal transporter with homology to ferroportin. Nat Commun. 2015;6:8545. 40. Altamura S, Kessler R, Grone HJ, et al. Resistance of ferroportin to hepcidin binding causes exocrine pancreatic failure and fatal iron overload. Cell Metab. 2014;20(2):359367. 41. McGregor JA, Shayeghi M, Vulpe CD, et al. Impaired iron transport activity of ferroportin 1 in hereditary iron overload. J Membr Biol. 2005;206(1):3-7. 42. De Domenico I, Ward DM, Nemeth E, et al. The molecular basis of ferroportin-linked hemochromatosis. Proc Natl Acad Sci USA. 2005;102(25):8955-8960. 43. De Domenico I, McVey Ward D, Nemeth E, et al. Molecular and clinical correlates in iron overload associated with mutations in ferroportin. Haematologica. 2006;91(8):10921095. 44. De Domenico I, Ward DM, Langelier C, et al. The molecular mechanism of hepcidinmediated ferroportin down-regulation. Mol Biol Cell. 2007;18(7):2569-2578. 45. Mayr R, Janecke AR, Schranz M, et al. Ferroportin disease: A systematic metaanalysis of clinical and molecular findings. J Hepatol. 2010;53(5):941-949. 46. Griffiths WJ, Mayr R, McFarlane I, et al. Clinical presentation and molecular pathophysiology of autosomal dominant hemochromatosis caused by a novel ferroportin mutation. Hepatology. 2010;51(3): 788-795. 47. Mayr R, Griffiths WJ, Hermann M, et al. Identification of Mutations in SLC40A1 That Affect Ferroportin Function and Phenotype of Human Ferroportin Iron Overload. Gastroenterology. 2011;140(7): 2056-2063 e2051. 48. Pietrangelo A, Caleffi A, Corradini E. NonHFE hepatic iron overload. Sem Liv Dis. 2011;31(3):302-318. 49. Donovan A, Lima CA, Pinkus JL, et al. The iron exporter ferroportin/Slc40a1 is essential for iron homeostasis. Cell Metab. 2005;1(3):191-200. 50. Zohn IE, De Domenico I, Pollock A, et al. The flatiron mutation in mouse ferroportin acts as a dominant negative to cause ferroportin disease. Blood. 2007;109(10):41744180. 51. Goncalves AS, Muzeau F, Blaybel R, et al. Wild-type and mutant ferroportins do not form oligomers in transfected cells. Biochem J. 2006;396(2):265-275. 52. Schimanski LM, Drakesmith H, Talbott C, et al. Ferroportin: lack of evidence for multimers. Blood Cells Mol Dis. 2008;40(3):360369. 53. Pignatti E, Mascheroni L, Sabelli M, Barelli S, Biffo S, Pietrangelo A. Ferroportin is a monomer in vivo in mice. Blood Cells Mol Dis. 2006;36(1):26-32. 54. Sabelli M, Montosi G, Garuti C, et al. Human macrophage ferroportin biology and the basis for the ferroportin disease. Hepatology. 2017;65(5):1512-1525. 55. McDonald CJ, Wallace DF, Ostini L, Bell SJ, Demediuk B, Subramaniam VN. G80Slinked ferroportin disease: classical ferroportin disease in an Asian family and reclassification of the mutant as iron transport

defective. J Hepatol. 2011;54(3):538-544. 56. Le Lan C, Mosser A, Ropert M, et al. Sex and acquired cofactors determine phenotypes of ferroportin disease. Gastroenterology. 2011;140(4):1199-1207.e1-2. 57. Ferbo L, Manzini PM, Badar S, et al. Detection of a rare mutation in the ferroportin gene through targeted next generation sequencing. Blood Transfus. 2016;14(6):531-534. 58. Pelucchi S, Mariani R, Salvioni A, et al. Novel mutations of the ferroportin gene (SLC40A1): analysis of 56 consecutive patients with unexplained iron overload. Clin Genet. 2008;73(2):171-178. 59. Subramaniam VN, Wallace DF, Dixon JL, Fletcher LM, Crawford DH. Ferroportin disease due to the A77D mutation in Australia. Gut. 2005;54(7):1048-1049. 60. Lim FL, Dooley JS, Roques AW, Grellier L, Dhillon AP, Walker AP. Hepatic iron concentration, fibrosis and response to venesection associated with the A77D and V162del "loss of function" mutations in ferroportin disease. Blood Cells Mol Dis. 2008;40(3):328333. 61. Corradini E, Montosi G, Ferrara F, et al. Lack of enterocyte iron accumulation in the ferroportin disease. Blood Cells Mol Dis. 2005;35(3):315-318. 62. Mougiou A, Pietrangelo A, Caleffi A, Kourakli A, Karakantza M, Zoumbos N. G80S-linked ferroportin disease: the first clinical description in a Greek family. Blood Cells Mol Dis. 2008;41(1):138-139. 63. Wolff F, Bailly B, Gulbis B, Cotton F. Monitoring of hepcidin levels in a patient with G80S-linked ferroportin disease undergoing iron depletion by phlebotomy. Clin Chim Acta. 2014;430:20-21. 64. Cremonesi L, Forni GL, Soriani N, et al. Genetic and clinical heterogeneity of ferroportin disease. Br J Haematol. 2005;131(5): 663-670. 65. Cunat S, Giansily-Blaizot M, Bismuth M, et al. Global sequencing approach for characterizing the molecular background of hereditary iron disorders. Clin Chem. 2007;53(12):2060-2069. 66. Bach V, Remacha A, Altes A, Barcelo MJ, Molina MA, Baiget M. Autosomal dominant hereditary hemochromatosis associated with two novel Ferroportin 1 mutations in Spain. Blood Cells Mol Dis. 2006;36(1):4145. 67. Moreno-Carralero MI, Munoz-Munoz JA, Cuadrado-Grande N, et al. A novel mutation in the SLC40A1 gene associated with reduced iron export in vitro. Am J Hematol. 2014;89(7):689-694. 68. Girelli D, De Domenico I, Bozzini C, et al. Clinical, pathological, and molecular correlates in ferroportin disease: A study of two novel mutations. J Hepatol. 2008;49(4):664671. 69. Callebaut I, Joubrel R, Pissard S, et al. Comprehensive functional annotation of 18 missense mutations found in suspected hemochromatosis type 4 patients. Hum Mol Genet. 2014;23(17):4479-4490. 70. Saja K, Bignell P, Robson K, Provan D. A novel missense mutation c.470 A>C (p.D157A) in the SLC40A1 gene as a cause of ferroportin disease in a family with hyperferritinaemia. Br J Haematol. 2010;149(6):914-916. 71. Hetet G, Devaux I, Soufir N, Grandchamp B, Beaumont C. Molecular analyses of patients with hyperferritinemia and normal serum iron values reveal both L ferritin IRE and 3 new ferroportin (slc11A3) mutations. Blood.

1983

A. Pietrangelo 2003;102(5):1904-1910. 72. Devalia V, Carter K, Walker AP, et al. Autosomal dominant reticuloendothelial iron overload associated with a 3-base pair deletion in the ferroportin 1 gene (SLC11A3). Blood. 2002;100(2):695-697. 73. Wallace DF, Pedersen P, Dixon JL, et al. Novel mutation in ferroportin1 is associated with autosomal dominant hemochromatosis. Blood. 2002;100(2):692-694. 74. Cazzola M, Cremonesi L, Papaioannou M, et al. Genetic hyperferritinaemia and reticuloendothelial iron overload associated with a three base pair deletion in the coding region of the ferroportin gene (SLC11A3). Br J Haematol. 2002;119(2):539-546. 75. Melis MA, Cau M, Congiu R, et al. A mutation in the TMPRSS6 gene, encoding a transmembrane serine protease that suppresses hepcidin production, in familial iron deficiency anemia refractory to oral iron. Haematologica. 2008;93(10):1473-1479. 76. Roetto A, Merryweather-Clarke AT, Daraio F, et al. A valine deletion of ferroportin 1: a common mutation in hemochromastosis type 4. Blood. 2002;100(2):733-734. 77. Speletas M, Kioumi A, Loules G, et al. Analysis of SLC40A1 gene at the mRNA level reveals rapidly the causative mutations in patients with hereditary hemochromatosis type IV. Blood Cells Mol Dis. 2008;40(3):353-359. 78. Zoller H, McFarlane I, Theurl I, et al. Primary iron overload with inappropriate hepcidin expression in V162del ferroportin disease. Hepatology. 2005;42(2):466-472. 79. Unal S, Piperno A, Gumruk F. Iron chelation with deferasirox in a patient with de-novo ferroportin mutation. J Trace Elem Med Biol. 2015;30:1-3. 80. Morris TJ, Litvinova MM, Ralston D, Mattman A, Holmes D, Lockitch G. A novel ferroportin mutation in a Canadian family with autosomal dominant hemochromatosis. Blood Cells Mol Dis. 2005;35(3):309-314. 81. Relvas L, Claro MT, Bento MC, Ribeiro ML. Novel human pathological mutations. Gene symbol: SLC40A1. Disease: haemochromatosis, type 4. Hum Genet. 2009;125 (3):338. 82. Gordeuk VR, Caleffi A, Corradini E, et al. Iron overload in Africans and AfricanAmericans and a common mutation in the SCL40A1 (ferroportin 1) gene small star, filled. Blood Cells Mol Dis. 2003;31(3):299304. 83. Beutler E, Barton JC, Felitti VJ, et al. Ferroportin 1 (SCL40A1) variant associated with iron overload in African-Americans. Blood Cells Mol Dis. 2003;31(3):305-309. 84. Zaahl MG, Merryweather-Clarke AT, Kotze MJ, van der Merwe S, Warnich L, Robson KJ. Analysis of genes implicated in iron regulation in individuals presenting with primary iron overload. Hum Genet. 2004;115(5):409-417. 85. Lee PL, Gaasterland T, Barton JC. Mild iron overload in an African American man with SLC40A1 D270V. Acta Haematol.

1984

2012;128(1):28-32. 86. Lee PL, Gelbart T, West C, Barton JC. SLC40A1 c.1402G-->a results in aberrant splicing, ferroportin truncation after glycine 330, and an autosomal dominant hemochromatosis phenotype. Acta Haematol. 2007;118(4):237-241. 87. Koyama C, Wakusawa S, Hayashi H, et al. A Japanese family with ferroportin disease caused by a novel mutation of SLC40A1 gene: hyperferritinemia associated with a relatively low transferrin saturation of iron. Intern Med. 2005;44(9):990-993. 88. Jouanolle AM, Douabin-Gicquel V, Halimi C, et al. Novel mutation in ferroportin 1 gene is associated with autosomal dominant iron overload. J Hepatol. 2003;39(2):286-289. 89. Yamakawa N, Oe K, Yukawa N, et al. A Novel Phenotype of a Hereditary Hemochromatosis Type 4 with Ferroportin1 Mutation, Presenting with Juvenile Cataracts. Intern Med. 2016;55(18):26972701. 90. Sussman NL, Lee PL, Dries AM, Schwartz MR, Barton JC. Multi-organ iron overload in an African-American man with ALAS2 R452S and SLC40A1 R561G. Acta Haematol. 2008;120(3):168-173. 91. Wallace DF, McDonald CJ, Ostini L, Iser D, Tuckfield A, Subramaniam VN. The dynamics of hepcidin-ferroportin internalization and consequences of a novel ferroportin disease mutation. Am J Hematol. 2017;92(10):1052-1061. 92. Wallace DF, Browett P, Wong P, Kua H, Ameratunga R, Subramaniam VN. Identification of ferroportin disease in the Indian subcontinent. Gut. 2005;54(4):567568. 93. Rivard SR, Lanzara C, Grimard D, et al. Autosomal dominant reticuloendothelial iron overload (HFE type 4) due to a new missense mutation in the FERROPORTIN 1 gene (SLC11A3) in a large French-Canadian family. Haematologica. 2003;88(7):824-826. 94. Raszeja-Wyszomirska J, Caleffi A, Milkiewicz P, Pietrangelo A. Ferroportinrelated haemochromatosis associated with novel Y64H mutation of the SCL40A1 gene. Prz Gastroenterol. 2014;9(5):307-309. 95. Del-Castillo-Rueda A, Moreno-Carralero MI, Alvarez-Sala-Walther LA, et al. Two novel mutations in the SLC40A1 and HFE genes implicated in iron overload in a Spanish man. Eur J Haematol. 2011;86(3): 260-264. 96. Viprakasit V, Merryweather-Clarke AT, Chinthammitr Y, et al. Molecular Diagnosis of the First Ferroportin Mutation (C326Y) in the Far East Causing a Dominant Form of Inherited Iron Overload. Blood. 2004;104 (11):3204. 97. Lok CY, Merryweather-Clarke AT, Viprakasit V, et al. Iron overload in the Asian community. Blood. 2009;114(1):20-25. 98. Wallace DF, Dixon JL, Ramm GA, Anderson GJ, Powell LW, Subramaniam VN. A novel mutation in ferroportin implicated in iron overload. J Hepatol. 2007;46(5):921-926.

99. Letocart E, Le Gac G, Majore S, et al. A novel missense mutation in SLC40A1 results in resistance to hepcidin and confirms the existence of two ferroportin-associated iron overload diseases. Br J Haematol. 2009;147 (3):379-385. 100. Barton JC, Acton RT, Rivers CA, et al. Genotypic and phenotypic heterogeneity of African Americans with primary iron overload. Blood Cells Mol Dis. 2003;31(3):310319. 101. Gordeuk VR, Diaz SF, Onojobi GO, et al. Ferroportin Q248H, dietary iron, and serum ferritin in community African-Americans with low to high alcohol consumption. Alcohol Clin Exp Res. 2008;32(11):19471953. 102. Barton JC, Acton RT, Lee PL, West C. SLC40A1 Q248H allele frequencies and Q248H-associated risk of non-HFE iron overload in persons of sub-Saharan African descent. Blood Cells Mol Dis. 2007;39(2): 206-211. 103. Wallace DF, Subramaniam VN. The global prevalence of HFE and non-HFE hemochromatosis estimated from analysis of nextgeneration sequencing data. Genet Med. 2016;18(6):618-626. 104. Corradini E, Ferrara F, Pollicino T, et al. Disease progression and liver cancer in the ferroportin disease. Gut. 2007;56(7):10301032. 105. Manifold IH, Triger DR, Underwood JC. Kupffer-cell depletion in chronic liver disease: implications for hepatic carcinogenesis. Lancet. 1983;2(8347):431-433. 106. Galicia-Poblet G, Cid-Paris E, Lopez-Andres N, et al. A Pediatric Case Report of Ferroportin Disease. J Pediatr Gastroenterol Nutr. 2015;63(6):e205-e207. 107. Mistry PK, Lopez G, Schiffmann R, Barton NW, Weinreb NJ, Sidransky E. Gaucher disease: Progress and ongoing challenges. Mol Genet Metab. 2017;120(1-2):8-21. 108. Beaumont C, Leneuve P, Devaux I, et al. Mutation in the iron responsive element of the L ferritin mRNA in a family with dominant hyperferritinaemia and cataract. Nat Genet. 1995;11(4):444-446. 109. Kannengiesser C, Jouanolle AM, Hetet G, et al. A new missense mutation in the L ferritin coding sequence associated with elevated levels of glycosylated ferritin in serum and absence of iron overload. Haematologica. 2009;94(3):335-339. 110. Miyajima H, Nishimura Y, Mizoguchi K, Sakamoto M, Shimizu T, Honda N. Familial apoceruloplasmin deficiency associated with blepharospasm and retinal degeneration. Neurology. 1987;37(5):761-767. 111. Heilmeyer L, Keller W, Vivell O, Betke K, Woehler F, Keiderling W. [Congenital atransferrinemia]. Schweiz Med Wochenschr. 1961;91:1203. 112. Pietrangelo A, Corradini E, Ferrara F, et al. Magnetic resonance imaging to identify classic and nonclassic forms of ferroportin disease. Blood Cells Mol Dis. 2006;37(3):192196.

haematologica | 2017; 102(12)

ARTICLE

Hematopoiesis

Folate dietary insufficiency and folic acid supplementation similarly impair metabolism and compromise hematopoiesis

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Curtis J. Henry,1,2,3 Travis Nemkov,1 Matias Casás-Selves,1,4 Ganna Bilousova,5 Vadym Zaberezhnyy,1 Kelly C. Higa,2 Natalie J. Serkova,7 Kirk C. Hansen,1 Angelo D’Alessandro1 and James DeGregori1,2,6,8*

Department of Biochemistry and Molecular Genetics, University of Colorado AMC, Aurora, CO, USA; 2Department of Immunology and Microbiology, University of Colorado AMC, Aurora, CO, USA; 3Current address: Department of Pediatrics, Emory University, Atlanta, GA, USA; 4Current address: Ontario Institute for Cancer Research, Toronto, ON, Canada; 5Department of Dermatology and Charles C. Gates Center for Regenerative Medicine, University of Colorado AMC, Aurora, CO, USA; 6Department of Medicine, Section of Hematology, University of Colorado AMC, Aurora, CO, USA; 7Department of Anesthesiology, University of Colorado AMC, Aurora, CO, USA and 8Department of Pediatrics, Section of Hematology/Oncology, University of Colorado AMC, Aurora, CO, USA 1

Haematologica 2017 Volume 102(12):1985-1994

ABSTRACT

hile dietary folate deficiency is associated with increased risk for birth defects and other diseases, evidence suggests that supplementation with folic acid can contribute to predisposition to some diseases, including immune dysfunction and cancer. Herein, we show that diets supplemented with folic acid both below and above the recommended levels led to significantly altered metabolism in multiple tissues in mice. Surprisingly, both low and excessive dietary folate induced similar metabolic changes, which were particularly evident for nucleotide biosynthetic pathways in B-progenitor cells. Diet-induced metabolic changes in these cells partially phenocopied those observed in mice treated with anti-folate drugs, suggesting that both deficiency and excessive levels of dietary folic acid compromise folate-dependent biosynthetic pathways. Both folate deficiency and excessive dietary folate levels compromise hematopoiesis, resulting in defective cell cycle progression, persistent DNA damage, and impaired production of lymphocytes. These defects reduce the reconstitution potential in transplantation settings and increase radiation-induced mortality. We conclude that excessive folic acid supplementation can metabolically mimic dietary folate insufficiency, leading to similar functional impairment of hematopoiesis.

Correspondence: james.degregori@ucdenver.edu

Received: April 18, 2017. Accepted: September 6, 2017. Pre-published: September 7, 2017. doi:10.3324/haematol.2017.171074

Introduction Given the genetic variability within the human population, divergent lifestyles, vastly variable diets, and inaccurate self-reporting, unambiguous links between diet and disease predisposition have been difficult to establish. Folate, a B vitamin, is an important factor for a number of metabolic pathways, including DNA methylation and the biosynthesis of nucleotides.1 While dietary folate deficiency is a problem in much of the developing world, mandatory folate supplementation of grain products in the USA and Canada since the late 1990s has nearly eliminated dietary folate deficiency in these countries and reduced the rate of neural tube defects.1,4 Folate is important for the synthesis of purines and thymidylate, which are required for mitochondrial and cytosolic adenosine triphosphate (ATP), total nucleotide triphosphate (NTP), and deoxy-NTP (dNTP) production.1 Folate also contributes to the one-carbon/methyl donor pathway, being critical for the production of S-adenosylmethionine (SAM), which is essential for the methylation of DNA, glutathione, and other macromolecules. Importantly, while natural folates in foods are primarily tetrahydrofolates (THF), folic acid (the synthetic oxidized form of folate) is the form primarily used for supplementation, due to its economical synthesis and good bioavailability. haematologica | 2017; 102(12)

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/1985 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1985

C.J. Henry et al.

High level folic acid intake is common in todayâ&#x20AC;&#x2122;s society, given both the supplementation of grains and the common consumption of additional vitamin supplements, energy drinks, and breakfast cereals with added folic acid.2,3 Indeed, many breakfast cereals are fortified at 160175% over reported levels,4,5 and often consumed at well above the suggested serving sizes.6 While the Recommended Dietary Allowance (RDA) for folate is 400 mg/day, folic acid supplementation above the recommended limit of 1000 Âľg/day is not uncommon for women of childbearing age.7 Even higher daily doses, up to 5 mg, can be recommended for pregnant women with certain preconditions, such as obesity, diabetes, MTHFR status, or a history of pregnancies associated with neural tube defects.8,9 Given that supplemented folate is primarily in the form of folic acid, which is not normally present in vivo, and that folic acid has been shown to inhibit at least one key metabolic enzyme,10 it is imperative that we gain a full understanding as to how this supplementation impacts cellular metabolism. Dietary folate levels have been linked to cancer risk in a puzzling way; dietary folate deficiency has been associated with increased risk of some cancers,2 while excessive folic acid supplementation may also be associated with increased cancer risks.2,3 For example, an inverse correlation between folate intake and the risk of colorectal adenocarcinomas have been supported by some studies carried out in both mice and humans.11 In contrast, other trials indicate that folic acid supplementation (1 mg/day) after detection of polyps or in individuals with a history of colorectal adenoma is associated with increased progression to, or recurrence of, adenomas;12-14 a connection further supported by mouse studies.15 Moreover, a reversal in the downward trend of colorectal cancer incidence in the USA and Canada is evident, starting in 1996 and coinciding with the onset of folate supplementation in these countries.3 Another clinical trial showed that supplementation with folic acid plus vitamin B12 increased cancer incidence and all-cause mortality in patients with ischemic heart disease.16 Nonetheless, other studies have failed to observe such associations,17 and differences in supplementation and the myriad of other genetic, dietary and lifestyle complications likely contribute to the lack of clear associations. Both low and high levels of dietary folate have been shown to negatively impact immune function in humans. A study of postmenopausal women describes a bellshaped curve for folate intake and natural killer (NK) cell cytotoxicity,18 with reduced NK cell activity in both low and high intake groups. This study also noted an inverse association between unmetabolized folic acid in plasma and NK cell cytotoxicity, suggesting that free folic acid may negatively impact immune function. Maternal folate supplementation has been shown to associate with increased incidence of allergy-related respiratory impairment in children18 and multi-generational respiratory defects in rats.18 In rats, altering dietary folate levels reduces the percentages of circulating B cells and augments splenic lymphocyte responses to lipopolysaccharide (LPS; particularly in the context of folic acid supplementation).19 Moreover, long-term and multigenerational exposure to folic acid supplementation can exacerbate neural tube defects associated with several different mutations in mice.20 On the other hand, maternal folate supple1986

mentation is associated with a number of positive health outcomes (in humans and rodents), such as reductions of neural tube defects and congenital cardiac defects in children.18 Taken together, these observations suggest that both insufficient and excessive dietary folate can impact multiple tissues in as of yet undefined ways, highlighting our lack of understanding of how alterations in dietary folate levels impact cellular homeostasis. Given that both low and high dietary folate have been associated with various diseases, in the study herein we sought to determine how modulating dietary folate levels impact metabolic, developmental, and physiological processes in hematopoietic progenitor cells. Strikingly, we found that both insufficient and excessive dietary folate levels similarly compromised nucleotide metabolism, leading to functional defects in hematopoietic cells.

Methdos Mice and folate supplementation Mice were fed deficient (FD; 0.1 mg/ kg folic acid), control (CD; 2 mg/ kg folic acid), or supra-folate diets (SD; 10 mg/ kg folic acid). 2mg/kg folic acid complies with the recommendations of the American Institute for Nutrition for rodents.21 The chow was purchased from Research Diets (AIN-76A, except that folic acid levels were varied), and was sterilized by irradiation. All chow was supplemented with the antibiotic succinylsulfathiazole to prevent folate production from gut bacteria.

Mass spectrometry analysis for organ metabolomics Bone marrow (BM) B-cell progenitors were isolated by antiB220 Magnetic Activated Cell Sorting (Miltenyi Biotec) and processed for ultra high performance liquid chromatography mass spectrometry (UHPLC-MS) analysis as described in Online Supplementary Methods.

Untargeted quantitative 1H-nuclear magnetic resonance metabolomics analysis Isolated B-cell progenitors from pooled animals were extracted and nuclear magnetic resonance (NMR) analyses performed on a Bruker 500 MHz spectrometer, as described in Online Supplementary Methods.

Bone marrow transplants For competitive BM transplantation assays shown in Figure 6, whole donor BM from mice on various levels of dietary folate (green fluorescent protein (GFP)-) was mixed at a 3:1 ratio with competitor BM (GFP+) from GFP-expressing mice fed normal folate diets. For competitive BM transplantation assays shown in Figure 6, whole donor BM from mice on various levels of dietary folate was mixed at a 3:1 ratio with green fluorescent protein (GFP)-expressing competitor BM from mice fed normal folate diets.

Flow cytometric analysis and complete blood counts For surface stains: Single-cell suspensions were plated in 96-well round-bottomed plates and washed in fluorescence-activated cell sorting (FACS) buffer [3% fetal bovine serum (FBS) + 1X phosphate buffered saline (PBS) + 2mM ethylenediamine tetraacetic acid (EDTA; v/v)]. After washing, cells were surface stained for 1 hour on ice in 50 ml of antibody solution, and analyzed by flow cytometry to identify the hematopoietic populations of interest. The antibodies used are listed in Online Supplementary Methods.

haematologica | 2017; 102(12)

Unbalanced folate levels impair hematopoiesis

Figure 1. Mice on low and high dietary folate exhibit reduced peripheral leukocyte numbers. (A) BALB/c mice were fed control (CD), folate-deficient (FD) and suprafolate (SD) diets for 4 months and serum was collected and analyzed for the presence of folic acid using a folic acid microbiological test kit. Values represent mean ±SEM of (6 total samples). (B) The weights of BALB/c mice maintained on CD, FD and SD for six months are shown. Weights represent mean ±SEM of 2 independent sets of mice (6 total samples). (C and D) Complete blood counts were performed on peripheral blood collected from BALB/c mice kept on CD, FD and SD folate diets for 2, 4, and 6 months. Values represent mean ±SEM from 5-10 mice/diet at the various time points. All statistical analyses were performed using Student’s t-test relative to CD for each experiment. Student’s t-test was used in (A) and (B) comparing both FD and SD diets to CD. In (C) and (D) statistical analyses were performed using a one-way ANOVA test followed by Tukey’s post-test. *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001.

Complete blood counts Peripheral blood was collected from the lateral tail vein in heparinized microfuge tubes at the indicated time points. Complete blood counts were performed on a Cell-Dyn 1700 (Abbott Laboratories, Abbott Park, IL, USA).

Statistics Unpaired t-tests, the Cox proportional hazards, and one-way ANOVA were used to analyze experiments, with significance indicated by *P<0.05, **P<0.01, ***P<0.001, and **** P<0.0001. All errors shown represent biological replicates (different mice), not technical replicates. Statistical analyses were performed using GraphPad Prism (version 6.07; GraphPad Software). Survival curves were analyzed by the log-rank (Mantel–Cox) test. All results are expressed as mean ± SEM.

Results Mice on low and high dietary folate exhibit reduced peripheral leukocyte numbers In order to determine the effects of having too little or too much dietary folate, Balb/c mice were fed deficient (FD; 0.1 mg/kg), control (CD; 2 mg/kg), or supra (SD; 10 mg/kg) folate diets for 2 to 12 months. We analyzed how these diets affected circulating folate levels, weight, and the representation of peripheral blood cells. The folatedeficient (FD) diet resulted in a ~3-fold decrease in the amount of circulating folate while the supra-folate (SD) diet significantly increased serum folate levels by ~2-fold (Figure 1A). The observed folate levels for control and supra-folate diets in mice are within the range of variances observed in humans (5.9-24.6 µg/L depending on the haematologica | 2017; 102(12)

extent of fortification and supplementation reported for participants from the USA between 2001 and 2004).22 The observed folate levels for the FD diet in mice reflect what is considered folate-deficiency for humans (3 mg/L).23 Given the importance of folate in the production of nucleotides and ATP, complete blood cell counts were performed in order to determine if altering dietary folate levels impacted circulating blood cell populations. Leukocyte numbers were consistently reduced in mice on both FD and SD diets, which was apparent by 4-6 months on the altered diets (Figure 1C). By 6 months on these diets, reductions in peripheral lymphocyte numbers were also evident (Figure 1D). The representation of circulating neutrophils and thrombocytes was not significantly altered by modulating dietary folate levels for up to 6 months (Online Supplementary Figure S1A-S1C); however, hemoglobin and red blood cells were significantly reduced with FD diets in some experiments, but not others (Online Supplementary Figure S1D). Importantly, mice maintained on the FD, SD, and CD diets for up to a year were not distinguishable by observable features including differences in weight (Figure 1B and Online Supplementary Figure S1E), activity, or survival (data not shown). Taken together, these observations reveal that mice on the FD and SD diets maintained normal physical phenotypes, but exhibited reduced circulating levels of leukocytes and lymphocytes.

Control, folate-deficient and supra-folate diets alter cellular homeostasis in distinctive ways Based on the observation that both FD and SD diets reproducibly reduced systemic lymphocyte numbers, we next determined how altering dietary folate levels affected 1987

C.J. Henry et al. A

Figure 2. Both deficient and supra dietary folate levels alter metabolism in B-progenitor cells. (A-H) BALB/c mice were fed control (CD), folate-deficient (FD) and supra-folate (SD) diets for 4 months and B-progenitor cells were isolated using B220+ MACS selection. NMR analysis was performed on these cells in order to determine diet-induced metabolic changes. Values represent mean ÂąSEM from 2 independent experiments (5 pooled mice/group/experiment), which precluded statistical analysis due to the number of technical replicates (n=2).

metabolism in B-lymphocyte progenitor cells, given that this is a highly proliferative population in the BM. Altering dietary folate levels did not significantly change the representation of pro-B, pre-B, and immature B-progenitor cell populations in the BM (Online Supplementary Figure S2), although the ratios of pro-B to pre-B cells trended higher in mice on FD and SD diets. NMR metabolomic analysis of BM B220+ cells revealed that B-cell progenitors from mice on both the FD and SD diets exhibited significantly increased lactate levels (Figure 2A), with reductions in citrate, glutamate and glutamine levels, suggesting that unbalanced dietary folate levels alters central carbon metabolism (Figure 2B-D; Online Supplementary Table S1). Furthermore, levels of total glutathione (a key cellular reducing agent), which is derived from s-adenosyl methionine (SAM), glutamine and glutamate, were significantly reduced in B-progenitors isolated from mice on the FD and SD diets (Figure 2E). Additionally, both FD and SD diets significantly decreased creatine (which can be used to make ATP), total nucleotides, and total adenosine levels in B-cell progenitors (Figures 2F-H). Due to the striking observation that both low and high levels of dietary folate similarly compromise metabolic activity in B-progenitor cells, we further explored metabolic perturbations resulting from altered folate intake using ultra-high pressure liquid chromatography combined with UHPLC-MS. This method employs a highly sensitive technique that allows for the robust and comprehensive detection of metabolites in a high-throughput manner (for full results see Online Supplementary Table S2). We additionally determined if alterations in dietary folate levels induce metabolic perturbations in the heart, liver, and intestinal tissues (Online Supplementary Figure S3 and Online Supplementary Table S2). Unsupervised hierarchical clustering and partially supervised partial least squares1988

discriminant analysis (PLS-DA) of metabolomics data revealed that both the FD and SD diets impacted metabolism in B-cell progenitors, the heart, liver, and intestines, with each altered diet eliciting distinct signatures (Online Supplementary Figures S3B-S3F). PC1 explains the highest percentage of the variance (20-30%) calculated on the basis of metabolic phenotypes of a given sample set. Notably, in all tested tissues PC1 could discriminate between the CD and the FD/SD diets, while the FD and SD diets either overlapped or were partially discriminated along PC2 (explaining < 15% of the variance). Results were even more striking when performing unsupervised hierarchical clustering for B-progenitor cells (Euclidean distance, furthest neighbor â&#x20AC;&#x201C; Online Supplementary Figure S3B), which revealed co-clustering of FD and SD samples (t-test of square distances - P-value = 0.27) and separate clustering of CD samples (t-test of square distances P-valueCD/FD = 0.052 and P-valueCD/SD = 0.016). We performed a Metabolite Set Enrichment Analysis (MSEA) to identify metabolic pathways that were significantly enriched (False discovery rate (FDR) < 0.05) in the tissues from mice on both the FD and SD diets. Results revealed enrichment for purine metabolism and alanine, aspartate, and glutamate metabolism for B-progenitors (Online Supplementary Figure S4); protein biosynthesis for liver; purine metabolism, protein biosynthesis, gluconeogenesis, and the urea cycle for heart; and taurine and hypotaurine metabolism for intestines (Online Supplementary Table S3). These results further highlight that diets deficient in folate or having excess levels of folate alter cellular function in multiple organs, and that insufficient and excess dietary folate have profound and comparable impacts on metabolism. Surprisingly, we did not observe significant changes within the one-carbon metabolism pathway leading to haematologica | 2017; 102(12)

Unbalanced folate levels impair hematopoiesis

Figure 3. Both folate-deficient and supra-folate diets impact nucleotide metabolism in B-progenitor cells. BALB/c mice were fed control (CD), folate-deficient (FD) and supra-folate (SD) diets for 4 months (2 independent experiments with group sizes of 2 and 3 mice). Metabolites altered by changes in dietary folate levels were identified in B-progenitor cells (B220+ cells isolated by MACs selection) using UHPLC-MS (primary data are shown in Online Supplementary Table S2 and B-progenitor cell profiles used for these experiment are shown in Online Supplementary Figure S2). (A) PLS-DA analysis was performed on the data from Online Supplementary Table S2 to determine metabolites associated with nucleotide production in B-progenitor cells from animals on the three diets. (B) MSEA was performed to identify pathways that are affected by altered folic acid diets. Metabolic pathways are graphed based on scores from enrichment analysis (y-axis), which identifies enriched pathways based on the number of associated metabolites that are identified, and topology analysis (x-axis), which is a measure of a centrality for a particular pathway enriched in a given data set. The color of each circle is proportional to the P-value, with red representing relatively low P-value within the data set, and the size of each circle is proportional to the pathway impact value.

SAM, rather, there was a trend towards increased levels of SAM in B-progenitors from mice on the FD and SD diets (Online Supplementary Table S2). Methionine levels, along with most other amino acids, significantly increased in FD and SD B-progenitors (Online Supplementary Figure S4A). Given the significant impact of the FD and SD diets on nucleotide synthesis pathways, as shown using both NMR and MS approaches, we have focused on nucleotide metabolism and its impact on cell fitness.

Folate-deficient and supra-folate diets similarly reduce nucleotide metabolism in B-progenitor cells Given the strikingly similar metabolic alterations induced by FD and SD diets on B-progenitor cells and observed reductions in lymphocyte numbers in peripheral blood, PLS-DA and hierarchical clustering were performed to identify similar pathways in these cells that were affected by altering dietary folate. Consistent with the enrichment for purine metabolism by metabolite set enrichment analysis (MSEA), PLS-DA analysis indicated a highly similar impact of both the FD and SD diets on nucleotide metabolism (Figure 3A). Pathway analysis of metabolites showing the highest loading weights along PC1 in PLSDA analyses revealed that both the FD and SD diets (compared to CD) significantly and similarly impacted purine metabolism, pyrimidine metabolism and amino acid homeostasis (Figure 3B and Online Supplementary Figure S3B), most likely by altering the function of similar enzymes involved in these metabolic pathways (Online Supplementary Figure S4C). Nonetheless, certain metabohaematologica | 2017; 102(12)

lites were differentially altered in B-cell progenitors from mice on the FD and SD diets. In particular, we observed increased levels of saturated and unsaturated fatty acid metabolites in FD B-progenitors, but decreased levels of many of the same metabolites in SD B-progenitors (Online Supplementary Table S2). Various metabolites in the purine and pyrimidine nucleotide synthesis pathways were significantly perturbed by altering dietary folate. Reductions in the levels of inosine, adenylosuccinate, and guanine were evident in B-progenitors from mice on the FD and SD diets relative to those on the CD diet (Figure 4A,B). Moreover, we observed significant reductions in adenosine monophosphate (AMP), adenosine diphosphate (ADP), adenosine triphosphate (ATP), guanosine monophosphate (GMP), guanosine diphosphate (GDP), and Guanosine-5'-triphosphate (GTP; Figure 4B) in B-progenitors from mice on the FD and SD diets. Large reductions in oxidative Pentose Phosphate Pathway intermediates, such as D-Glucono-15-lactone 6-phosphate and 6-phosphogluconate, coupled with increased non-oxidative phase intermediates such as sedoheptulose 1-phosphate and ribose 1-phosphate, were also observed in both FD and SD progenitors (Online Supplementary Table S2). Furthermore, we observed reductions in both oxidized and reduced glutathione (Online Supplementary Figure S5) and in the key second messenger, cyclic-AMP, perhaps as a consequence of reduced ATP availability, which could impact redox control and cell signaling. While levels of deoxyadenosine triphosphate (dATP) were modestly reduced in B-progenitors from SD 1989

C.J. Henry et al. A

Figure 4. Folate-deficient and supra-folate diets compromise purine nucleotide metabolism. (A and B) BALB/c mice were fed control (CD), folate-deficient (FD), and supra-folate (SD) diets for 4 months as described in Figure 3, and specific metabolites in the purine nucleotide synthesis pathway that were altered by changes in dietary folate levels are graphed (data presented in Online Supplementary Table S2). Values represent mean ±SEM of 5 mice/diet from 2 independent experiments with group sizes of 2 and 3 mice. Each metabolite was analyzed using the Student’s t-test with the CD values set as the control. *P<0.05; **P<0.01; ***P<0.001; UMP: uridine monophosphate; IMP; inosine monophosphate; AMP: adenosine monophosphate; ADP: adenosine diphosphate; ATP: adenosine triphosphate; GMP: guanosine monophosphate; GDP: guanosine diphosphate; GTP: guanosine triphosphate

mice, levels of deoxyguanosine triphosphate (dGTP) were reduced by 4.3- and 9.7-fold in FD and SD B-progenitors, respectively (Online Supplementary Table S2), which could contribute to impaired DNA synthesis. Defects were also observed for pyrimidine synthesis, with reductions in uridine monophosphate (UMP), uridine-5'-triphosphate (UTP), and deoxythymidine monophospate (dTMP) in FD and SD B-progenitors (Online Supplementary Figure S6). Since both high and low folate intake result in a similar phenotype, we hypothesize that excess folate may exert negative inhibitory activity on rate-limiting enzymes of folate metabolism, particularly dihydrofolate reductase (DHFR). To test this hypothesis, we asked whether treatment of mice for 5 consecutive days with the DHFR inhibitor methotrexate (MTX), which is routinely used as a chemotherapeutic agent, would phenocopy purine metabolic defects observed in FD and SD mice in the Bprogenitor cell compartment. Treatment of wild-type Balb/c mice (maintained on standard mouse chow with 2 mg/kg folic acid) with MTX resulted in patterns of impaired nucleotide metabolism in B-progenitor cells, which in some cases mirrored those observed for progenitors from mice on the FD and SD diets. Surprisingly, the impacts of MTX treatment on these metabolites were less than for the altered diets (Figure 5A and Online Supplementary Figure S6; for full results see Online Supplementary Table S4). For example, decreases in GMP and AMP observed in B-cell progenitors from mice on the FD and SD diets were recapitulated in B-cell progenitors isolated from MTX treated mice (Figure 5B). 1990

The function of pro B-progenitor cells is impaired in mice fed folate deficient and supra-folate diets We determined the effects of these diets on hematopoiesis in mice. The numbers of early hematopoietic stem and progenitor cells (HSPC) and myeloid progenitor cells in the tibias and femurs did not differ for mice on the CD, FD and SD diets (Online Supplementary Figure S7). Given that the FD and SD diets led to a reduction in the number of circulating lymphocytes and reduced metabolism in B-progenitor cells, we next determined if these diets altered DNA replication in B-progenitors using EdU (5-ethynly 2’-deoxyuridine) incorporation assays (Figure 6A). While the percentage of B-progenitor cells in S-phase did not change as a result of modulating dietary folate levels (Figure 6B), the FD and SD diets significantly reduced the rate of S-phase progression in pro-B-cells (particularly in the case of folate deficiency; Figure 6C) and compromised the efficiency of nucleotide incorporation (Figure 6D). Furthermore, we observed that FD and SD diets promoted persistent DNA damage in B-progenitor cells (Figure 6E).

Both low and high dietary folate impair hematopoietic reconstitution post-irradiation Transgenic mice ubiquitously expressing GFP (GFPTG) were irradiated and transplanted with GFPneg whole bone marrow cells (BMC) isolated from mice fed CD, FD and SD diets, along with competitor GFPTG whole BM (at a 3:1 ratio). Recipient mice were maintained on a normal diet (2 mg/kg folate). While the contribution of donor CD BMC haematologica | 2017; 102(12)

Unbalanced folate levels impair hematopoiesis

Figure 5. Methotrexate reduces purine synthesis in B-progenitor cells. BALB/c mice were treated for 5 consecutive days with vehicle (1X PBS) or methotrexate (MTX; 10 mg/kg) injected intraperitoneally. (A and B) On the day after the last injection, B-progenitor cells were isolated by MACs selections (B220+) and UHPLC-MS was used to determine the metabolic changes in the purine nucleotide synthesis pathway that resulted from treatment with MTX. Values represent mean ÂąSEM of 4 mice/treatment group. Studentâ&#x20AC;&#x2122;s t-test was used to compare metabolic changes induced in B-progenitors isolated from PBS and MTX treated mice. *P<0.05; **P<0.01; ***P<0.001. UMP: uridine monophosphate; IMP; inosine monophosphate; AMP: adenosine monophosphate; ADP: adenosine diphosphate; ATP: adenosine triphosphate; GMP: guanosine monophosphate; GDP: guanosine diphosphate; GTP: guanosine triphosphate.

continued to increase in recipient mice for over 6 months, we observed a significant and steady decline in hematopoietic contributions to peripheral blood cells from donor FD and SD BMC (Figure 7A). Upon sacrifice of recipient mice at 7 months post-transplant, clear reductions in contributions towards B-cell, myeloid, and multipotent progenitor populations from transplanted FD and SD BMC were evident (Figure 7B-D). While the decline in hematopoietic contribution from FD and SD donor BMC could result from altered methylation states, no significant total DNA methylation differences were observed in Bprogenitor cells isolated from FD or SD mice (Figure 6F), consistent with the lack of effects on one-carbon metabolism. Since FD and SD diets did not significantly alter the number of long-term hematopoietic stem cells (Online Supplementary Figure S7A-S7C) or common myeloid progenitor cells (Online Supplementary Figure S7D-S7F), it is unlikely that the declining contributions of FD and SD BMC to hematopoiesis in recipient mice resulted from the transplantation of unequal numbers of HSCP. Therefore, transplanted FD and SD HSCP are likely compromised, leading to reduced hematopoietic reconstitution potential prior to or after seeding of the BM microenvironment of CD recipient mice. Given these observations, we asked whether altered folate diets would impact survival post-irradiation for mice. Mice maintained on CD, FD and SD diets for 4 months were sublethally irradiated (5 Gy) and monitored for survival post-irradiation. Surprisingly, all mice fed low and high levels of dietary folate succumbed to irradiationhaematologica | 2017; 102(12)

induced complications within two weeks post-irradiation and required sacrifice, whereas only one CD mouse was removed from the study due to signs of morbidity (Figure 7E). The increased mortality for mice on the FD and SD diets post-irradiation likely, or at least in part, could result from reduced ability to restore hematopoiesis post-irradiation. Nonetheless, since the mechanism of death was not determined, impacts of altered dietary folate on other organs (i.e., the intestines) could contribute to enhanced radiation sensitivity.

Discussion Data presented herein support a model whereby impairment of folate-dependent metabolism due to diets both with high and low folic acid dietary supplementation leads to hematopoietic defects (Figure 7F). As previously reviewed,2,28,24 while folic acid supplementation has been shown to be beneficial in reducing neural tube defects in newborns, folic acid supplementation may also be associated with numerous health problems in humans. These include respiratory disorders, cancers, autism spectrum disorders, and multiple sclerosis (although cause and effect relationships were not established, and some studies fail to find such associations). A number of these associations have been tested and substantiated with rodent models. Dietary deficiency in folates is also linked to a number of disorders, including cancer and neural tube defects. However, our understanding of how both insufficient and 1991

C.J. Henry et al. A

Figure 6. Alterations in dietary folate levels lead to functional impairments in B-progenitor cells. BALB/c mice were fed control (CD), folate-deficient (FD), and suprafolate (SD) diets for 5 months and in vivo EdU incorporation was determined in B-progenitor cell populations (A). The percentage of EdU positive B-progenitor cells (B) and the normalized X-mean and Y-mean mean fluorescence intensity (MFI) were calculated (C and D). The normalized X-mean and Y-mean MFI serve as proxies for progression through S-phase and the efficiency of nucleotide incorporation, respectively. The normalized X-mean MFI of EdU+ populations was calculated using the following formula: [(x-mean MFI of EdU+ cells) - (x-mean MFI of the G1 population)]. The normalized Y-mean MFI of EdU+ populations was calculated using the following formula: [(y-mean MFI of EdU+ cells) - (y-mean MFI of EdU– cells)]. (E) Bone marrow derived pro B-progenitor cells were identified using the gating strategy defined in (A) and intracellular staining was performed in order to determine phospho-γH2AX levels in this population using flow cytometric analysis. (F) For the DNA methylation assay, DNA was extracted using DNeasy Blood and Tissue Kit (Qiagen) from MACS-sorted B220+ cells (Miltenyi Biotec). Methylation levels were analyzed using MethylFlash™ Global DNA Methylation (5-mC) ELISA Easy Kit (Epigentek) per manufacturer's instructions. One representative experiment out of two is shown. Controls provided by the kit as well as 5-azacytidine-treated KG-1 cells as a hypomethylated DNA control were used (data not shown). Values in (A-E) represent mean ±SEM of 5 samples/ group. All statistical analyses were performed using a one-way ANOVA followed by a Tukey’s post-test in order to compare the effects of all diets to each other.

excess folate could contribute to reductions in human health is limited. Data presented herein demonstrate that both low and high levels of dietary folate compromise metabolism in multiple organs, with significant defects manifesting in the B-progenitor compartment. Although folate is required in multiple biosynthetic pathways (including for nucleotides, glutathione, SAM and amino acids), we primarily observed deficiencies in the nucleotide synthesis pathways for B-progenitors isolated from mice fed FD and SD diets, as demonstrated using both NMR and UHPLCmass spectrometry. The impact of the FD and SD diets on reduced nucleotide synthesis in B-progenitor cells likely contributed to the S-phase proliferative defects observed using EdU analysis. The observation that both low and high levels of dietary folate promote similar metabolic defects in hematopoietic cells is surprising. Nonetheless, based on mathematical modeling, Ulrich and colleagues have suggested that adverse effects of high folate diets might be explained by the ability of key enzymes, including in the purine and pyrimidine synthetic pathways, to be inhibited by products of folate metabolism.2,25 For example, thymidylate synthase (TS) uses methylene-tetrahydrofolate (MTHF) to 1992

make dTMP from deoxyuridine monophosphate (dUMP), producing dihydrofolate (DHF), which must be recycled by DHFR to regenerate tetrahydrofuran (THF). Importantly, DHF can inhibit TS activity.2,25 Notably, supplemented folic acid requires reduction by DHFR (mostly in the liver) to DHF, but this reaction is more than 1000X slower relative to the reduction of DHF to THF, and DHFR exhibits variable activity in the livers of different people.10 Accordingly, folic acid inhibits DHFR reduction of DHF.10 A possible explanation for the metabolic defects observed in SD B-progenitors is that excessive folic acid may inhibit DHFR, thus impairing reduced folate dependent pathways. Indeed, conversion of folic acid to the reduced folates DHF and THF presents saturation kinetics (additional substrate cannot be processed effectively), and folic acid consumption above ~200 µg/day is estimated to overwhelm an individual’s ability to convert folic acid to THFs, leading to unmetabolized folic acid in the circulation.6,26 Notably, in a study of postmenopausal women from the USA, circulating folic acid was found in 78% of participants,22 and both mandated fortification and further supplementation were shown to have increased circulating folic acid levels in the Framingham Offspring Cohort.27 Thus, albeit speculative, the negative impact of high folate haematologica | 2017; 102(12)

Unbalanced folate levels impair hematopoiesis

Figure 7. Alterations in dietary folate levels lead to reduced hematopoietic reconstitution potential and increased mortality following irradiation treatment. Two month old BALB/c GFP-expressing transgenic mice were lethally irradiated (10 Gy) and transplanted with GFP-negative whole BMC isolated from mice fed control (CD), folate-deficient (FD) and supra-folate (SD) diets for two months along with competitor BMC expressing GFP at a ratio of 3:1 (GFPPos: GFPNeg). (A) The contribution of the donor BMC to hematopoiesis was assessed over 7 months by monitoring the percentage of GFPNeg cells in the peripheral blood. The number of donor-derived B-progenitor cells (B), hematopoietic stem and progenitor cells (C), and myeloid progenitor cells (D) was determined by flow cytometric analysis. Values in (A-D) represent ÂąSEM of 4- 5 mice/ group. (E) BALB/c mice fed CD, FD, and SD diets for 4 months were sublethally irradiated (5 Gy) and survival monitored post-irradiation. The Kaplan-Meier curves reflect results pooled from 2 experiments with a total of 4 mice/ group. Statistical analyses in (A-D) were performed using a one-way ANOVA followed by a Tukeyâ&#x20AC;&#x2122;s post-test in order to compare the effects of all diets to each other, and in (E) the log-rank (Mantel-Cox) test was used to determine statistical significance. **P<0.01; ****P<0.0001. GFP: green fluorescent protein; BM: bone marrow; BMT: bone marrow transplant.

diets may not result from folates per se, but from supplementation with folic acid (as opposed to reduced folates);28,29 however, further experimentation is required to test this hypothesis. Together with these previous studies, the results presented herein suggest that studies examining the impact of dietary folate on disease risk need to consider the source and type of folate. We observed that both folate deficiency and supra-levels of dietary folate lead to reduced hematopoietic reconstitution potential in competitive bone marrow transplant (BMT) experiments and increased death post-irradiation, consistent with observed cell cycle defects in hematopoietic progenitors. Our BMT studies could indicate that the fitness defects apparent in hematopoietic progenitors are somatically heritable, in that they manifest in recipient mice that are maintained on normal diets. Still, given that hematopoietic progenitors in FD and SD mice appear to be deficient in their ability to reconstitute hematopoiesis post-irradiation, reduced competition in recipient mice could in part relate to an immediate failure in HSPC to reconstitute the host, independent of the impact of the altered diets on DNA methylation. The paradoxical association of diets both low and high in folates with increased cancers has been ascribed to different effects of folate on cancer initiation and on the growth of preexisting tumors.2,3 Folate insufficiency is haematologica | 2017; 102(12)

thought to enhance cancer initiation by increasing the misincorporation of dUTP in DNA, leading to oncogenic mutations. For the latter, excessive folate is thought to fuel the growth of pre-existing tumors. However, we show that both insufficient and excessive folic acid are associated with impaired DNA synthesis. Our data suggest that common effects of insufficient or excessively supplemented folate on disease could have a common cause â&#x20AC;&#x201C; impairments in folate-dependent metabolism. While it may seem counterintuitive that impairing pathways essential for cell proliferation (such as nucleotide synthesis) would increase the risk of cancer, these impairments would be expected to enhance selection for oncogenic events that are adaptive in this context.30 In addition, increased levels of DNA damage could increase the frequency of potentially oncogenic events on which such altered selection could act. Given our results that diets high in folic acid can, over time, have substantial negative impacts on tissue and progenitor cell metabolism and fitness, mirroring those of dietary folate deficiency, excessive folic acid consumption by many Americans could have unappreciated negative impacts on their health. There are clear indications that combined mandatory and voluntary folic acid supplementation are vastly exceeding the targeted level,2,6,31,32 with the intake of folic acid for most Americans well above the 1993

C.J. Henry et al.

RDA. Indeed, 23% of the population of the USA (including 43% of children) were considered high in serum folates in the NHANES 1999-2000 study,23 and the diets of many Americans are fortified at >1000 mg/day. With a substantial fraction of the North American population at risk for excessive folate intake, and much of the rest of the world potentially deficient in folate, the studies presented herein should stimulate parallel research in humans that could have significant public policy implications. Funding These studies were supported by grants to C.J.H. (NIH grants

References 1. Depeint F, Bruce WR, Shangari N, Mehta R, O'Brien PJ. Mitochondrial function and toxicity: role of B vitamins on the one-carbon transfer pathways. Chem Biol Interact. 2006 Oct 27;163(1-2):113-132. 2. Ulrich CM, Potter JD. Folate supplementation: too much of a good thing? Cancer Epidemiol Biomarkers Prev. 2006;15(2):189193. 3. Mason JB, Dickstein A, Jacques PF, et al. A temporal association between folic acid fortification and an increase in colorectal cancer rates may be illuminating important biological principles: a hypothesis. Cancer Epidemiol Biomarkers Prev. 2007;16(7):1325-1329. 4. Rader JI, Weaver CM, Angyal G. Total folate in enriched cereal-grain products in the United States following fortification. Food Chemistry. 2000;70(3):275-289. 5. Whittaker P, Tufaro PR, Rader JI. Iron and folate in fortified cereals. J Am Coll Nutr. 2001;20(3):247-254. 6. Quinlivan EP, Gregory JF, 3rd. Effect of food fortification on folic acid intake in the United States. Am J Clin Nutr. 2003;77(1):221-225. 7. Hoyo C, Murtha AP, Schildkraut JM, et al. Folic acid supplementation before and during pregnancy in the Newborn Epigenetics STudy (NEST). BMC Public Health. 2011;11(1):1-8. 8. Laanpere M, Altmae S, Stavreus-Evers A, Nilsson TK, Yngve A, Salumets A. Folatemediated one-carbon metabolism and its effect on female fertility and pregnancy viability. Nutr Rev. 2010;68(2):99-113. 9. Barua S, Kuizon S, Junaid MA. Folic acid supplementation in pregnancy and implications in health and disease. J Biomed Sci. 2014;21:77. 10. Bailey SW, Ayling JE. The extremely slow and variable activity of dihydrofolate reductase in human liver and its implications for high folic acid intake. Proc Natl Acad Sci USA. 2009;106(36):15424-15429.

1994

K01CA160798) and to J.D. (American Institute for Cancer Research Grants 05B088 and 209227, and NIH grant R21CA137262). The Metabolomics and Flow Cytometry Shared Resources are supported by NCRR CCTSI/5UL1-RR025780 and by Cancer Center Support Grant P30-CA046934. Acknowledgments We would also like to thank Drs. Lee Niswander and Juliette Petersen of the University of Colorado Anschutz Medical Campus for their careful review of the manuscript. We also thank Drs. Amber Marean and Lee Niswander for the measurements of serum folate in Figure 1A.

11. Kim YI. Folate and carcinogenesis: evidence, mechanisms, and implications. J Nutr Biochem. 1999;10:66-88. 12. Ulrich CM, Potter JD. Folate and cancer-timing is everything. JAMA. 2007; 297(21):2408-2409. 13. Cole BF, Baron JA, Sandler RS, et al. Folic acid for the prevention of colorectal adenomas: a randomized clinical trial. JAMA. 2007;297(21):2351-2359. 14. Kim YI. Folate, colorectal carcinogenesis, and DNA methylation: lessons from animal studies. Environ Mol Mutagen. 2004; 44(1):10-25. 15. Lawrance AK, Deng L, Rozen R. Methylenetetrahydrofolate reductase deficiency and low dietary folate reduce tumorigenesis in Apc min/+ mice. Gut. 2009;58(6):805-811. 16. Ebbing M, Bonaa KH, Nygard O, et al. Cancer incidence and mortality after treatment with folic acid and vitamin B12. JAMA. 2009;302(19):2119-2126. 17. Wien TN, Pike E, Wisloff T, Staff A, Smeland S, Klemp M. Cancer risk with folic acid supplements: a systematic review and meta-analysis. BMJ Open. 2012;2(1):e000653. 18. Burdge GC, Lillycrop KA. Folic acid supplementation in pregnancy: Are there devils in the detail? Br J Nutr. 2012;108(11):19241930. 19. Partearroyo T, Ubeda N, Montero A, Achon M, Varela-Moreiras G. Vitamin B(12) and folic acid imbalance modifies NK cytotoxicity, lymphocytes B and lymphoprolipheration in aged rats. Nutrients. 2013; 5(12):4836-4848. 20. Marean A, Graf A, Zhang Y, Niswander L. Folic acid supplementation can adversely affect murine neural tube closure and embryonic survival. Hum Mol Genet. 2011; 20(18):3678-3683. 21. Reeves PG, Nielsen FH, Fahey GC, Jr. AIN93 purified diets for laboratory rodents: final report of the American Institute of Nutrition ad hoc writing committee on the reformulation of the AIN-76A rodent diet. J Nutr. 1993;123(11):1939-1951.

22. Yeung L, Yang Q, Berry RJ. Contributions of total daily intake of folic acid to serum folate concentrations. JAMA. 2008; 300(21):2486-2487. 23. Pfeiffer CM, Caudill SP, Gunter EW, Osterloh J, Sampson EJ. Biochemical indicators of B vitamin status in the US population after folic acid fortification: results from the National Health and Nutrition Examination Survey 1999-2000. Am J Clin Nutr. 2005;82(2):442-450. 24. Shorter KR, Felder MR, Vrana PB. Consequences of dietary methyl donor supplements: Is more always better? Prog Biophys Mol Biol. 2015;118(1-2):14-20. 25. Nijhout HF, Reed MC, Budu P, Ulrich CM. A mathematical model of the folate cycle: new insights into folate homeostasis. J Biol Chem. 2004;279(53):55008-55016. 26. Kelly P, McPartlin J, Goggins M, Weir DG, Scott JM. Unmetabolized folic acid in serum: acute studies in subjects consuming fortified food and supplements. Am J Clin Nutr. 1997;65(6):1790-1795. 27. Kalmbach RD, Choumenkovitch SF, Troen AM, D'Agostino R, Jacques PF, Selhub J. Circulating folic acid in plasma: relation to folic acid fortification. Am J Clin Nutr. 2008;88(3):763-768. 28. Scaglione F, Panzavolta G. Folate, folic acid and 5-methyltetrahydrofolate are not the same thing. Xenobiotica. 2014;44(5):480488. 29. Obeid R, Holzgreve W, Pietrzik K. Is 5methyltetrahydrofolate an alternative to folic acid for the prevention of neural tube defects? J Perinat Med. 2013;41(5):469-483. 30. Rozhok AI, DeGregori J. Toward an evolutionary model of cancer: Considering the mechanisms that govern the fate of somatic mutations. Proc Natl Acad Sci USA. 2015;112(29):8914-8921. 31. Shane B. Folate fortification: enough already? Am J Clin Nutr. 2003;77(1):8-9. 32. Choumenkovitch SF, Selhub J, Wilson PW, Rader JI, Rosenberg IH, Jacques PF. Folic acid intake from fortification in United States exceeds predictions. J Nutr. 2002; 132(9):2792-2798.

haematologica | 2017; 102(12)

ARTICLE

Red Cell Biology & its Disorders

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Xingguo Zhu,1* Tianxiang Hu,1* Meng Hsuan Ho,1,2 Yongchao Wang,1,3 Miao Yu,4 Niren Patel,5 Wenhu Pi,1,6 Jeong-Hyeon Choi,4,7 Hongyan Xu,7 Vadivel Ganapathy,1,8 Ferdane Kutlar,5 Abdullah Kutlar5 and Dorothy Tuan1

1 Department of Biochemistry and Molecular Biology, Augusta University, GA; 2School of Dentistry, Meharry Medical College, Nashville, TN; 3Department of Pharmacology and Nutritional Sciences, University of Kentucky, Lexington, KY; 4Georgia Cancer Research Center, Augusta University, GA; 5Division of Hematology/Oncology, Augusta University, GA; 6Department of Radiation Oncology, Indiana University School of Medicine, Indianapolis, IN; 7Department of Biostatistics, Augusta University, GA and 8Department of Cell Biology and Biochemistry, Texas Tech University Health Sciences Center, Lubbock, TX, USA

*XZ and TH contributed equally to this work.

Haematologica 2017 Volume 102(12):1995-2004

ABSTRACT

ydroxyurea (HU), the first of two drugs approved by the US Food and Drug Administration for treating patients with sickle cell disease (SCD), produces anti-sickling effect by re-activating fetal γ-globin gene to enhance production of fetal hemoglobin. However, approximately 30% of the patients do not respond to HU therapy. The molecular basis of non-responsiveness to HU is not clearly understood. To address this question, we examined HU-induced changes in the RNA and protein levels of transcription factors NF-Y, GATA-1, -2, BCL11A, TR4, MYB and NF-E4 that assemble the γ-globin promoter complex and regulate transcription of γ-globin gene. In erythroblasts cultured from peripheral blood CD34+ cells of patients with SCD, we found that HU-induced changes in the protein but not the RNA levels of activator GATA-2 and repressors GATA-1, BCL11A and TR4 correlated with HU-induced changes in fetal hemoglobin (HbF) levels in the peripheral blood of HU high and low responders. However, HU did not significantly induce changes in the protein or RNA levels of activators NF-Y and NF-E4. Based on HU-induced changes in the protein levels of GATA-2, 1 and BCL11A, we calculated an Index of Hydroxyurea Responsiveness (IndexHU-3). Compared to the HU-induced fold changes in the individual transcription factor protein levels, the numerical values of IndexHU3 statistically correlated best with the HU-induced peripheral blood HbF levels of the patients. Thus, IndexHU-3 can serve as an appropriate indicator for inherent HU responsiveness of patients with SCD.

Correspondence: dtuanlo@augusta.edu

Received: July 1, 2017. Accepted: September 29, 2017. Pre-published: September 29, 2017. doi:10.3324/haematol.2017.175646 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/1995 ©2017 Ferrata Storti Foundation

Introduction Sickle cell disease (SCD) is a common, genetic disorder of adult b-hemoglobin, which affects millions of people of diverse racial groups worldwide, including approximately 100,000 Americans, mostly of African descent. Hydroxyurea (HU) is the first of two US Food & Drug Administration (FDA)-approved drugs for treating SCD. In contrast to the recently approved Endari (L-glutamine), HU is shown to ameliorate the SCD symptoms by re-activating the fetal γ-globin gene to produce fetal hemoglobin (HbF) with anti-sickling activity,1-10 although HU also provides beneficial effects in decreasing adhesion of sickle erythrocytes to vascular endothelial cells, thus reducing complications of vaso-occlusion and infarction.11,12 However, approximately 30% of SCD patients do not respond to HU therapy in haematologica | 2017; 102(12)

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

1995

X. Zhu et al. increasing HbF levels to ameliorate the SCD symptoms.3-10 The molecular basis of HU non-responsiveness is largely unknown. The fetal γ-globin gene is silenced in adult erythroid cells but can be re-activated through mechanisms that include the signal-transduction pathway.13 Thus, the cGMP pathway provides a potential mechanism of γ-globin gene reactivation by HU: HU and/or the nitric oxide generated by HU binds to and activates soluble guanylyl cyclase to synthesize cGMP;14,15 cGMP in turn activates cGMP-dependent protein kinase PKG to phosphorylate and activate p38 MAPK,16,17 whose downstream targets ultimately impinge on the γ-globin promoter to activate synthesis of γ-globin mRNA and HbF to produce anti-sickling effect.13,18 However, the nuclear targets of the HU-induced signaling pathway, the transcription factors (TFs) that bind to γ-globin promoter and activate transcription of γ-globin gene, have not been clearly identified. A number of TFs bind to the proximal γ-globin promoter and regulate transcription of γ-globin gene. These TFs could be the ultimate nuclear targets of HU in re-activating γ-globin gene in adult erythroid cells. For example, NF-Y binds to the tandem CCAAT motifs in the γ-globin promoter to serve as a pioneering TF in recruiting other TFs to assemble the proximal γ-globin promoter complex and activate transcription of γ-globin gene (Figure 1).19-21 CoupTFII and dimeric TR2/TR4 compete with NF-Y for binding to DNA motifs overlapping the distal CCAAT box and repress γ-globin gene;22-25 GATA-1, and -2 bind to the GATA motif in γ-globin proximal promoter to respectively repress and activate γ-globin gene21,26,27 NF-E4/CP2 dimer binds to its cognate DNA motif near the TATA box to activate γ-globin gene28 (Figure 1). In addition, BCL11A and MYB are involved in γ-globin gene regulation, since their genetic variants are associated with elevation of HbF levels.29,30 BCL11A can bind to DNA motifs distal to the γ-globin promoter and act over distance to indirectly repress transcription of γ-globin gene,31,32 although BCL11A as well as MYB also binds directly to the γ-globin promoter to repress γ-globin gene (Figure 1).21,33,34 Thus, the inactive γ-globin promoter in adult erythroid cells can bind both a repressor hub of BCL11A/GATA1/CoupTFII/TR2/TR4 and an activator hub of NFY/GATA-2/NF-E4 (Figure 1).21 The poised state of the γ-globin promoter suggests that pharmacological compounds including HU can modulate the levels of the TFs in the activator and repressor hubs to re-activate the silenced γ-globin gene in adult erythroid cells. Here, we report that, in erythroblasts cultured ex vivo for ten days from peripheral blood CD34+ cells of HU high responsive SCD patients, HU increased the protein level of activator GATA-2 and drastically decreased the protein levels of repressors GATA-1, BCL11A, TR4 and MYB to activate transcription of γ-globin gene and synthesis of HbF to produce corresponding anti-sickling effect. In cultured erythroblasts of HU low/non-responders, such HUinduced differential changes in protein levels of the key activator and repressors were not observed. Our findings indicated that HU-induced changes in protein levels, but not RNA levels, of key TFs in the γ-globin promoter complex were strong modulators of HU responsiveness of the SCD patients. Thus, IndexHU-3, based on combined, HUinduced changes in protein levels of GATA-2, -1 and BCL11A, could serve as a strong indicator for inherent HU responsiveness of the SCD patients. 1996

Methods Isolation of CD34+ cells from peripheral blood samples and ex vivo culture of CD34+ cells in the presence or absence of HU CD34+ cells were isolated from peripheral blood (30 mL) of homozygous HbS/HbS SCD patients seen at the pediatric and adult sickle cell clinics at Medical College of Georgia (MCG), Augusta University, USA and from aphoresed mononuclear cells of normal donors obtained anonymously from MCG Tumor Cell Bank, using protocol approved by the institutional review board (HAC #1009064). The isolated CD34+ cells were cultured to days 10-12 in appropriate medium as described,21,35 in the absence or presence of 50 µM HU, which was the lowest concentration of HU that activated γ-globin mRNA to a high level (see Online Supplementary Methods and Online Supplementary Figure S1).

RNA and protein analyses by qRT-PCR, RNA-seq and Western blots The numerical values of HU-induced fold changes, the +HU/HU values, in the RNA and protein of each TF were calculated as the ratios of the levels of the RNA or protein normalized to the RNA and protein levels of b-actin in cells cultured with HU over the normalized levels of the respective RNA and protein in cells cultured without HU (Online Supplementary Methods).

Calculation of IndexHU IndexHU-3 was calculated according to the formula: IndexHU3= (FcGATA-2)/(FcGATA-1)x(FcBCL11A), where Fc was the fold change of the respective TFs induced by HU, i.e. the +HU/-HU ratio obtained as shown above. IndexHU-4 was calculated with (FcTR4) included in the denominator and IndexHU-5 with both (FcTR4) and (FcMYB) included in the denominator.

Statistical analysis Correlation between HU-induced changes in RNA and protein levels of the TFs and HU-induced HbF levels was performed using Prism 5 linear regression analysis program. Pearson correlation coefficient was calculated between peripheral blood HbF levels and IndexHUs. A 2-sample t-test was performed to compare HUinduced changes in each of the TFs and to compare the IndexHUs between high- and low/non-responders. Paired t-test was used to analyze equivalence of HU-induced changes in HbF levels in cultured erythroblasts and in peripheral blood of the SCD patients. The statistical tests were two-sided at 0.05 significance levels with SAS 9.4.0 (SAS InstituteInc., Cary, NC, USA).

Hypoxia condition to induce sickling of SCD erythrocytes Day 12 erythrocytes were isolated from ex vivo cultured day 12 cells (erythroblasts mixed with erythrocytes) of peripheral blood CD34+ cells by Ficoll-Paque gradient as described (Online Supplementary Methods). The day 12 erythrocytes were deoxygenated in a hypoxia chamber (Coy Laboratory Products) at 2% oxygen for 6 hours. The deoxygenated cells were fixed with 3.7% formaldehyde on ice for 15 minutes in the hypoxia chamber before removal to atmosphere for microscopic observation and imaging (EVOS fluorescent cell imaging system).

Results HU increased HbF levels and produced anti-sickling effect in ex vivo cultured erythroid cells In order to unambiguously analyze the molecular basis haematologica | 2017; 102(12)

Hydroxyurea on protein levels of transcription factors

Figure 1. Molecular assembly of the key transcription factors (TFs) in proximal γ-globin promoter complex. (A) Sequence of the proximal γ-globin promoter. Underlined bases: DNA motifs that bind transcription factors as marked. Numbers in parentheses: first base positions in the motifs relative to the transcription start site of γ-globin mRNA at +1 located 25 bases 3’ of the TATA box. The MYB binding site CAATG at -18133 was not shown. (B) Proximal γ-globin promoter complex. Blue ribbon: promoter DNA containing transcription activator-binding motifs (red bars) and repressor-binding motifs (light green bars), which respectively bind transcription activators, NF-Y, GATA-2 and NF-E4, marked in red and transcription repressors, BCL11A, GATA-1, CoupTFII and TR2/TR4, marked in green. Blue rectangle with angled arrow: γ-globin gene and transcriptional direction of γ-globin mRNA. NF-Y, composed of subunits YA, YB and YC, binds to each of the tandem CCAAT motifs and bends the DNA by approximately 70° to form the pocket to recruit and interact with other TFs in assembling the proximal promoter complex (adapted from Zhu et al.21). MYB protein was not shown, as its interaction with TF partner(s) in the proximal γ-globin promoter complex was not known.

of HU-responsiveness of SCD patients, we focused on two groups of patients with widely separated peripheral blood HbF levels induced by HU therapy: the HU low/non-responsive patients with post HU HbF levels of ≤10% and the HU high responsive patients with post HU HbF levels of 20-30%3,9 (Online Supplementary Table S1). We isolated CD34+ cells from peripheral blood samples of the SCD patients and cultured the CD34+ cells in erythroid differentiation medium for ten days in the presence or absence of HU. To validate these ex vivo cultured day 10 erythroblasts as an appropriate cell system for dissecting the molecular basis of HU responsiveness, we compared the HbF levels induced by HU in cultured day 10 erythroblasts and in peripheral blood of the SCD patients. We found that in day 10 erythroblasts of HU high responders SCD 01 and 02 and low responder SCD 04, HU increased HbF levels by 2.5-, 2.1- and 2-fold, from 8% to 20%, 14% to 29% and 4% to 8%, respectively (Figure 2A). In the peripheral blood of these 3 SCD patients, the clinical records showed that HU at maximum tolerated dose (MTD) increased HbF levels by 2.1-, 2.7- and 2-fold, from 12% to 25%, 11% to 30% and 3% to 6%, respectively (Table 1 and Online Supplementary Table S1). The comparable fold changes in HbF levels induced by HU in cultured erythroblasts and in initial HU trials at maximum tolerated dose (MTD) in peripheral blood of both HU responsive and low/non-responsive SCD patients (see Online Supplementary Table S2 for paired t-tests) indicated that successive transfusions in SCD04 and other low/nonresponsive patients on blood exchange (Online Supplementary Table S1), which could blunt subsequent HbF induction by HU in the patients, did not exert lasting haematologica | 2017; 102(12)

genetic effects on patient CD34+ cells to suppress HUinduced HbF levels in cultured day 10 erythroblasts. Thus, the ex vivo cultured patient erythroblasts could serve as an appropriate cell system for designing bioassays to dissect the in vivo molecular basis of HU responsiveness of the SCD patients. Since the HU low responsive SCD04 patient required blood exchange from normal donors to ameliorate the SCD symptoms, HbA expressed by donor erythrocytes, in addition to patient HbS, was detected by HPLC in the exchanged peripheral blood of the patient (Figure 2A, right 2nd panel). However, the donor HbA was not detected by HPLC in the cultured day 10 erythroblasts (Figure 2A, right 3rd and 4th panels). This was expected, since the transfused, donor blood did not contain nucleated progenitor cells including CD34+ cells, which were removed prior to transfusion. Thus, the day 10 erythroblasts were derived only from the patient CD34+ cells and expressed only HbS. In HPLC analysis of HbF levels in patient erythrocytes after blood exchange, HbF% was calculated as HbF/HbF+HbS without including HbA (see Online Supplementary Methods), since HbA was expressed in separate donor erythrocytes. However, the calculated HbF% of 6% for SCD04 (Figure 2A, right 2nd panel) could be overestimated slightly, since the HbF peak in HPLC contained also approximately 0.2-0.5% of HbF contributed by the normal donor erythrocytes (Table 1). Hypoxia chamber assay of cultured day 12 erythrocytes showed that HU treatment reduced sickled erythrocytes of HU high responder, SCD 02, from 38% to 20%, and HU low responder, SCD 04, from 80% to 58% (Figure 2B). Thus, HU-induced increase in HbF levels correlated with 1997

X. Zhu et al.

HU-induced reduction in sickling of the erythrocytes (Figure 2C). The cell sickling assays together with the HPLC analysis (Figure 2A) showed that cultured patient erythroblasts provided an appropriate ex vivo cell system for dissecting the molecular basis of HU-responsiveness of the SCD patients.

HU slowed down the cell cycle but did not delay ex vivo erythropoiesis of cultured patient erythroblasts As HU is an inhibitor of DNA synthesis,37 it could slow

down cell division and delay ex vivo erythropoiesis.2,38 Thus, it could be argued that the HU treated day 10 erythroblasts, as compared to the untreated day 10 erythroblasts, contained more abundant earlier stage erythroblasts, which expressed higher levels of Îł-globin gene and thus higher HbF levels. Therefore, HU-induced HbF production did not directly involve TF-mediated nuclear events on transcription of the Îł-globin gene. To investigate this possibility, we analyzed by FACS the day 10 cells stained with antibodies against erythroid dif-

Figure 2. Hydroxyurea (HU) increased fetal hemoglobin (HbF) levels and produced anti-sickling effect in cultured erythrocytes of sickle cell disease (SCD) patients. (A) HPLC analysis of HbF levels in day 10 erythroblasts (Day 10E) and HbFPB in peripheral blood (P.blood) of SCD patients. (-HU) and (+HU): day 10 erythroblasts cultured without and with HU, or patient peripheral blood obtained before HU therapy and after HU and blood exchange (Bex) therapies. HbF, HbS and HbA elution peaks were as marked. y-axis: absorption units (AU) at 410 nm of the eluted hemoglobins; x-axis: time in minutes when the hemoglobins were eluted from the HPLC column. (B) Day 12 erythrocytes (Day 12e) of HU high and low responders, SCD02 and 04, respectively, cultured without and with HU, and subjected to hypoxia to induce cell sickling. Images of peripheral blood erythrocytes of SCD04 without and with HU therapy were similar to those of cultured erythroblasts (data not shown). (C) Percentages of sickled erythrocytes [among 400 counted cells in images in (2B)] plotted against HU-induced HbF levels of Day 10 erythroblasts from the same patients in (2A). (D) FACS analysis of Day 10 erythroblasts cultured without and with HU from HU high responders SCD 01, 02, and low responders SCD04, 14. Cells were stained with erythroid markers CD71 and CD235a, respectively.

1998

haematologica | 2017; 102(12)

Hydroxyurea on protein levels of transcription factors

ferentiation markers transferrin receptor and glycophorinA to determine if HU-treated day 10 cells contained more abundant early stage erythroblasts. We found that day 10 cells cultured in the presence or absence of HU showed similar staining patterns for both HU high and low responders (Figure 2D and Online Supplementary Figure S2), indicating that the HU-treated erythroblasts did

not contain a higher proportion of earlier erythroid progenitors. However, the cell numbers of the day 10 erythroblasts grown in the presence of HU were 1/2 to 1/4 of those grown in the absence of HU for both HU high and low responsive patients. This indicated that HU slowed down the cell culture by 1-2 cell divisions without apparently delaying the ex vivo erythropoiesis of the day 10 ery-

Table 1. Hydroxyurea (HU)-induced changes in RNA and protein levels of transcription factors (TFs), γ-globin (γ-glob) and fetal hemoglobin (HbF) levels in day 10 erythroblasts of 18 sickle cell disease (SCD) patients (SCD #1-18) and 3 normal donors (Normal #1-3).

SCD

NFY

GATA2

GATA1

BCL

4.7 2.86 4.4 4.1 2.9 3.7 2.47 1.21 3.88 1.42 2.67 1.74 4.1 0.99 2.9 1.9 2.7 2.24 2.5 1.0 1.96 1.05 3.25 0.87 3.49 0.76 4.8 0.42 3.39 2.64 0.54

1.3 0.195 0.9 0.97 0.172 0.97 1.04 0.25 1.38 0.18 0.89 0.21 1.2 0.53 1.2 0.53 0.7 3.2 0.89 0.52 1.0 0.78 1.2 0.95 1.21 0.4 1.16 0.48 0.61 0.57 0.77

0.9 0.175 0.83 0.65 0.166 0.86 0.75 0.10 0.61 0.08 0.78 0.10 1.2 0.61 1.1 0.49 0.8 0.34 0.78 0.68 0.95 0.58 0.7 0.99 0.72 0.4 0.5 0.65 0.66 0.4 1.17

NFY

GATA2

GATA1

BCL

1.10 1.05 0.97 1.05 1.14

2.78 0.67 3.79 1.7 1.87 1.06 1.53 2.56 0.91

1.06 0.41 0.74 0.93 1.38 1.04 0.82 0.96 0.78

0.79 0.95 0.82 0.64 0.87 1.13 0.9 0.98 1.47

1.91 1.65 1.79 1.1 1.63 1.6

0.98 1.0 0.86 1.07 0.9 0.91

1.13 1.23 1.34 1.06 0.65 0.62

1 RNA Protein 2 RNAa RNAb Protein 8 RNAa RNAb Protein 9 RNAa Protein 10 RNAa Protein 3 RNAa Protein 4 RNAa Protein 5 RNAa Protein 6 RNAa Protein 7 RNAa Protein 11 RNAa Protein 12 RNAa Protein 13 RNAa Protein 14 RNAa RNAb Protein

Normal 15 RNAa Protein 16 RNAa Protein 17 RNAa Protein 18 RNAa RNAb Protein

0.93 1.1 1.11 0.93 1.13 1.11 0.99 1.10

0.94 0.95 0.95 1.12

0.86

TR4

MYB

0.8

1.38

0.87 1.03 0.23 0.5 0.14 0.87 0.18

0.57 0.44 0.29 0.53 0.09 0.61 0.2

NFE4

γ-glob 2.8 3.51 2.5 2.09

0.95 0.90 1.1 1.17 0.95 0.82

2.0 2.44 1.81 2.38 2.94 2.53 3.86 1.8

HbF%

HbF%PB

IndexHU-3

8/20

12/25

83.8

14/29

11/30

101.6

5.3/16.4

6.2/17

6/18

7/22

98.6

12/23.6

4.2/9

4.5/8.8

3.0

4/8

3/6

7.3

2.2/5.3

1.5/4.9

2.1

4/10

2.8

5.1/9.6

4.7/11.5

2.3

3/4

3.2/3.1

0.9

1.2/2.9

4.7

2.1/3.5

1.4

1.25 1.2 0.83 1.16 0.96 0.95

0.63 0.2 1.1 1.0

1.0 0.93 0.99 0.98

1.87 1.23 1.51 1.06 1.75 1.63 0.46

0.98

0.85

TR4

MYB

0.53 0.51

NFE4

γ-glob

HbF%

1.1/1.8

0.6

HbF%PB

IndexHU-3

1.4/2.9

1.7

2.6/3.8

2.9

0.7/1.0

0.9

2.1/3.7

0.8

0.2

1.3

0.3

1.0

0.5

5.5

1.12 1.16 1.10 0.95 0.75 0.63 0.86

1.27 0.58 0.7 0.49

1.36 1.89 1.16

0.91 0.24

0.75 0.39

1.41 1.28 1.1

Normal 1 RNAa Protein 2 RNAa Protein 3 RNAa Protein

0.98 1.0

RNAa and RNAb: hydroxyurea (HU)-induced changes in RNA levels in total cellular RNAs determined by qRT-PCR and RNA-seq, respectively. Protein: HU-induced changes in protein levels determined by Western blots. Numbers: HU-induced fold changes, +HU/-HU, in RNAs and proteins. HbF%PB : fetal hemoglobin (HbF) levels in peripheral blood (PB) of the sickle cell disease (SCD) patients; two numbers separated by a slash: HbF levels in day 10 erythroblasts before and after HU treatment or HbF levels in peripheral blood of SCD patients pre- and post-HU treatment at maximum tolerated dose (MTD) recorded in the clinic. IndexHU-3: quantitative estimates of HU responsiveness calculated from HU-induced changes in the protein levels of GATA2, GATA-1 and BCL11A. CoupTFII was not analyzed as its RNA was not detected by RNA-seq on day 10 erythroblasts cultured either with or without HU (Online Supplementary File 2), nor was TR2, as TR4 with lower RNA level than TR2 RNA level (Online Supplementary File 2) appeared to be the limiting partner in the TR2/TR4 heterodimer. See Online Supplementary Files S1 and S2 for data and calculations.

haematologica | 2017; 102(12)

1999

X. Zhu et al.

throblasts. One explanation could be that, in our cell culture condition, HU was added on day 4, by which time the erythroid differentiation program of the cultured cells could have already been set up. Therefore, HU did not appreciably delay the differentiation of the cultured erythroblasts, indicating that the cultured erythroblasts were appropriate for subsequent studies to dissect the molecular basis of HU responsiveness of the SCD patients.

HU-induced changes in protein but not RNA levels of key TFs correlated with HU-induced HbF levels in cultured erythroblasts of SCD patients To determine the HU-induced changes in the RNA and protein levels of activators NF-Y, GATA-2 and NF-E4 and repressors GATA-1, BCL11A, TR4 and MYB that assemble the γ-globin proximal promoter complex (Figure 1), we used qRT-PCR, RNA-seq and Western blots to analyze the day 10 erythroblasts cultured with or without HU. To achieve statistical significance for the bioassays, we analyzed a total of 18 homozygous SCD patients (7 pediatric and 11 adult patients) among whom 5 were HU high responders (SCD 01, 02, 08, 09 and 10) and 13 were HU low responders (Online Supplementary Table S1). Although HU low/non-responsive patients comprise approximately 30% of the SCD patients, approximately 70% of the blood samples examined in this study were from low/non-responsive patients, because blood samples in large volumes of 30 mL required for the bioassays were more readily available from HU low/non-responsive patients undergoing blood exchange (see Online Supplementary Methods). RNA analysis by qRT-PCR was performed for all 18 SCD patients and 3 normal donors; genome-wide RNAseq analysis was performed for HU high responders (SCD 02 and 8), and HU low responders (SCD 14 and 18). The results showed that HU significantly and universally increased GATA-2 RNA levels by 200-500% in day 10 erythroblasts of both HU high and low responders and the normal donors (Table 1, Figure 3A, and Online Supplementary Files 1 and 2), in agreement with an earlier report.39 In contrast, HU mildly decreased or increased by ≤50% the RNA levels of repressors GATA-1, BCL11A, TR4 and MYB in erythroblasts of all the SCD patients and normal donors (Table 1 and Online Supplementary Files 1 and 2). These results were consistent with earlier reports that HU treatment modulates the RNA levels of BCL11A and MYB.40,41 However, CoupTFII RNA was not detected by either RT-PCR or RNA-seq (Online Supplementary File 2) in day 10 erythroblasts and was not further analyzed. The HU-induced changes in TF RNA levels did not consistently correspond to HU-induced changes in protein levels of the respective TFs. Thus, despite the HUinduced, universal increase in GATA-2 RNA levels in all SCD patient erythroblasts, HU increased GATA-2 protein level only in HU high responders SCD 01, 02, 08-10, whereas HU appeared to randomly decrease or increase GATA-2 protein level in the HU low responders (Table 1 and Figure 3B). Similarly, despite the HU-induced mild changes in RNA levels of repressors GATA-1, BCL11A, TR4 and MYB in the 18 SCD patient erythroblasts, HU drastically decreased by 70-80% the protein levels of these repressors only in high responders SCD 01, 02, 08, 09 and 10; but HU did not reduce or mildly reduced by ≤50% the repressor protein levels in the erythroblasts of HU low/non-responders SCD 03-7 and 11-18, with the 2000

exception that MYB protein level was significantly reduced by ≥50% also in HU low responders SCD 06 and 18 (Table 1 and Online Supplementary File S1). Statistical analysis by 2-sample t-tests confirmed that HU differentially modulated the protein levels, but not the RNA levels, of the repressors in HU high versus HU low responders; in contrast, in both HU high and low responders, HU did not induce significant changes in either the protein or the RNA levels of NF-Y and NF-E4, indicating that these two activator TFs did not mediate HU responsiveness of the SCD patients (Online Supplementary Table S3). In HU high responders, HU drastically decreased the repressor protein levels and mildly increased the protein level of activator GATA-2, which in combination lead to activation of γ-globin RNA transcription, synthesis of γ-globin protein and HbF (Table 1, and Online Supplementary Table S3 and Online Supplementary Figure S3A), and consequently production of anti-sickling effect in both the cultured erythroblasts and peripheral blood of the SCD patients (Figure 2B and C). In HU low responders, HU mildly decreased the repressor protein levels and inconsistently increased GATA-2 protein level, which in combination did not sufficiently activate γ-globin gene and synthesis of HbF to produce significant anti-sickling effect in these patients. We next used scatter plots to statistically and graphically correlate the HU-induced changes in protein and RNA levels of the TFs with HU-induced changes in HbF levels in peripheral blood of the 18 SCD patients. Regression analysis showed that HU-induced changes in GATA-1 and BCL11A protein levels, but not RNA levels, significantly correlated with HU-induced changes in peripheral blood HbF levels of the patients, while the correlation between GATA-2 protein levels and HbF levels was less significant (Figure 4A and B). The negative slopes of the correlation graphs of GATA-1 and BCL11A (higher HU-induced repressor protein levels correlating with lower HUinduced HbF levels) were consistent with GATA-1 and BCL11A being repressors of γ-globin gene; the positive slope of the GATA-2 graph was consistent with GATA-2 being an activator of γ-globin gene. Scatter plots also indicated that the HU-induced changes in TR4 and MYB protein levels correlated less strongly with the HU-induced changes in HbF levels, as compared to HU-induced changes in GATA-1 and BCL11A protein levels (compare Online Supplementary Figure S3B and C with F and G). In contrast, HU-induced changes in GATA-2 protein level correlated weakly but NF-Y and NF-E4 protein levels correlated not at all with the HU-induced changes in HbF levels (Online Supplementary Figure S3D and E). Thus, the protein levels of repressors GATA-1 and BCL11A were strong modulators and the protein level of activator GATA-2 was a weak modulator of the HU responsiveness of the SCD patients.

Index of HU responsiveness, IndexHU-3, calculated from HU-induced changes in protein levels of GATA -1, BCL11A and GATA-2, as a numerical indicator for HU responsiveness of SCD patients To quantify the cumulative effects of HU-induced changes in the protein levels of the transcription activator and repressors, we calculated the IndexHU-3 according to the formula: IndexHU-3= (FcGATA-2)/(FcGATA1)x(FcBCL11A), where Fc was the HU-induced fold changes in the protein levels of GATA-1, -2 and BCL11A haematologica | 2017; 102(12)

Hydroxyurea on protein levels of transcription factors

(see Methods section). The underlying rationale for the formulation was that HU-induced changes in activator protein level in the numerator together with HU-induced decrease in the repressor protein levels in the denominator would produce numerical values of IndexHUs quantitatively correlating with HU-induced changes in HbF levels and thus HU responsiveness of the SCD patients. Indeed, IndexHU-3s had numerical values of 40-100 for HU high responders and â&#x2030;¤10 for HU low/non-responders (Table 1). Correlation analysis by scatter plots showed that IndexHU-3 calculated from HU-induced protein levels, but not RNA levels, correlated strongly with the HUinduced HbF levels in peripheral blood of the SCD patients, with R2=0.9, as compared to the correlation between HU-induced changes in the protein levels of the individual TFs, with R2=0.3-0.6 (Figure 4B and C). Thus, IndexHU-3 could serve as a quantitative indicator/predictor for the inherent HU responsiveness or non-responsiveness of the SCD patients (Online Supplementary Table S3). As HU also induced a drastic reduction in the protein levels of TR4 and MYB in HU high responders, we calcu-

lated IndexHU-4 and -5 to include HU-induced (FcTR4) and (FcMYB) in the denominator. However, plotting IndexHU-3, -4 and -5 with respect to HU high versus HU low responders showed that IndexHU-4 and -5 did not improve the power to distinguish between HU high and low responders (Figure 4D). Thus, IndexHU-3 could serve as a reliable indicator to predict HU responsiveness of the SCD patients.

Discussion In this study, we investigated the molecular basis of HU responsiveness of SCD patients to ascertain the underlying mechanism(s) of why approximately 30% of SCD patients could not respond to HU therapy in enhancing HbF levels to produce an anti-sickling effect and to ameliorate the SCD symptoms. We first validated the appropriateness of the ex vivo cultured patient erythroblasts for the bioassays to dissect the in vivo molecular basis of HU responsiveness. We found that HU similarly induced HbF

Figure 3. Hydroxyurea (HU)-induced changes in RNA and protein levels of GATA-2, -1 and BCL11A in day 10 erythroblasts of HU high and low/non-responsive sickle cell disease (SCD) patients. (A) Total cellular RNAs isolated from day 10 patient erythroblasts cultured without and with HU and analyzed by qRT-PCR. For SCD 02, 04 and 14, the RT-PCR results shown were means of technical triplicates from two independent RNA preparations normalized with respect to the RNA level of bactin; for SCD 01, the RT-PCR results were means of technical triplicates. y-axis: the +HU/-HU ratios of normalized RNA levels of genes in cells treated with HU over the normalized RNA levels of the same genes in control cells not treated with HU, which were set at 1. Numbers in parentheses: numerical values of +HU/-HU ratios of each of the genes. (B) Western blots of protein levels of GATA-1, -2 and BCL11A in day 10 erythroblasts cultured without and with HU, (-) and (+), respectively, from the same SCD patients as in (A). Numbers underneath the blots: +HU/-HU ratios of the protein levels of the TFs normalized with respect to the protein level of bactin in erythroblasts treated with HU over the normalized protein levels of the same TFs in control erythroblasts without HU. HbF% and HbF%PB: Fetal hemoglobin (HbF) levels in day 10 erythroblasts determined by HPLC and in peripheral blood of the SCD patients obtained from the clinic; ND: not done; IndexHU-3: Index of HU responsiveness calculated from HU-induced fold changes in protein levels of GATA-2, -1 and BCL11A.

haematologica | 2017; 102(12)

2001

X. Zhu et al.

levels in cultured patient erythroblasts as in peripheral blood of the SCD patients on HU and/or blood exchange therapies (Table 1 and Figure 2). It has recently been reported that, in erythroblasts cultured from peripheral blood CD34+ cells of a group of SCD and b-thalassemia patients who were initially not on HU therapy but were subsequently put on prospective HU therapy, HU-induced changes in HbF levels either before or after HU therapy are similar.42 This finding, together with our results, indicates that HU and blood exchange therapies did not exert lasting genetic effects on bone marrow CD34+ cells of the patients to significantly change HU-induced HbF levels in patient erythroblasts cultured from the CD34+ cells. Protein and RNA analyses of the cultured patient erythroblasts showed that HU-induced changes in the protein levels of repressors GATA-1 and BCL11A were strong modulators, and activator GATA-2 a weak modulator of HU-induced HbF levels, and hence HU responsiveness of the SCD patients. In HU low/non-responsive patients with post-HU HbF levels ≤10%, HU did not drastically decrease the protein levels of the repressor TFs nor consistently increase the protein level of activator GATA-2 (Table 1 and Figure 3) to sufficiently activate transcription of γ-globin RNA and synthesis of HbF to produce significant anti-sickling effect in cultured erythroblasts and

peripheral blood (Figure 2). Since HU-induced changes in the RNA levels of the key TFs did not correlate at all with HU-induced peripheral blood HbF levels (Figure 4A), HUinduced changes in the RNA levels of the TFs could not serve as appropriate indicators of HU responsiveness of the SCD patients. IndexHU-3, calculated from combined HU-induced changes in the protein levels of GATA-2, -1 and BCL11A, correlated strongly (R2=0.9) with HU-induced peripheral blood HbF levels of the patients and, therefore, could serve as a strong indicator of HU responsiveness (Figure 4B and C, and Online Supplementary Table S3). It has been shown recently that HU-induced fold changes in γ-globin RNA levels in cultured erythroblasts of a group of SCD and b-thalassemia patients are the best indicator so far to predict HU responsiveness for these patients.42 Indeed, HU-induced fold changes in γ-globin RNA levels correlated with HUinduced γ-globin protein and HbF levels in cultured patient erythroblasts (Figure 2 and Online Supplementary Figure S3A) and showed a significant difference between HU responsive and non-responsive patients (Online Supplementary Table S3). However, the HU-induced fold changes in γ-globin RNA levels were in a narrow range of 1.2-1.9 for HU low/non-responders and 2-2.8 for HU high responders (Table 1 and Figure 3). IndexHU-3 with numerical values of

Figure 4. Statistical correlation analysis by scatter plots of hydroxyurea (HU)-induced fold changes in RNA and protein levels of GATA-2, -1 and BCL11A and of Index HU with respect to HU-induced changes in peripheral blood HbF levels of the 18 sickle cell disease (SCD) patients. (A and B) Scatter plots of HU-induced fold changes in RNA and protein levels of BCL11A, GATA-1 and -2 with respect to HU-induced peripheral blood HbF levels of the 18 SCD patients. (C) Scatter plots of IndexHU-3s calculated from HU-induced fold changes in the RNA and protein levels of GATA-2, -1 and BCL11A plotted against HU-induced peripheral blood HbF levels of the 18 SCD patients. (D) IndexHU-3, -4 and -5 calculated from the HU-induced changes in protein levels of 3, 4 and 5 TFs in 6 SCD patients. SCD 08, 09 and 10: HU high responders; SCD 06, 07 and 18: HU low responders. y-axis: numerical values of the respective IndexHUs; P-values: significance of the separation in the numeric values of IndexHU-3, -4 and -5 of the HU high versus low responsive groups.

2002

haematologica | 2017; 102(12)

Hydroxyurea on protein levels of transcription factors

40-100 for high responders and <10 for low/non-responders (Table 1 and Figure 3), therefore, provided a much wider numeric range for more accurate assessment of HU responsiveness of the SCD patients. Hydroxyurea could modulate the protein levels of the key TFs by modulating translational efficiency and/or stability of the TF proteins. Thus, genetic variations in HU low/non-responders, such as quantitative trait loci (QTL) identified by single nucleotide polymorphisms (SNPs) to associate with HU response in SCD patients,43 could impair critical steps in the HU-mediated protein translation and degradation pathways of the key TFs, resulting in low HU responsiveness of the patients. Recent studies on translational initiation and ribosome profiling show that the translation efficiency of key erythroid mRNAs, including BCL11A and γ-globin mRNAs, is dynamically controlled during erythropoiesis and could be subject to regulation by HU.44,45 In addition, HU through regulating specific miRNA levels,41,46 could differentially block or unblock translation of the activator and repressor TF proteins. During erythropoiesis, GATA-1 protein has been shown to be post-transcriptionally phosphorylated and subsequently degraded through the ubiquitin-proteasome pathway.47-49 These HU-downstream pathways that regu-

References 1. Platt O, Orkin S, Dover G, Beardsley P, Miller B, Nathan D. Hydroxyurea Enhances Fetal Hemoglobin Production in Sickle Cell Anemia. J Clin Invest. 1984;74(2):652-656. 2. Veith R, Galanello R, Papayannopoulou T, Stamatoyannopoulos G. Stimulation of Fcell production in patients with sickle-cell anemia treated with cytarabine or hydroxyurea. N Engl J Med. 1985;313(25):15711575. 3. Noguchi CT, Rodgers GP, Serjeant G, Schechter AN. Levels of fetal hemoglobin necessary for treatment of sickle cell disease. N Engl J Med. 1988;318(2):96-99. 4. Rodgers G, Dover G, Noguchi C, Schechter A, Nienhuis A. Hematologic responses of patients with sickle cell disease to treatment with hydroxyurea. N Engl J Med. 1990;322(15):1037-1045. 5. Goldberg MA, Brugnara C, Dover GJ, Schapira L, Charache S, Bunn HF. Treatment of sickle cell anemia with hydroxyurea and erythropoietin. N Engl J Med. 1990;323(6):366-372. 6. Steinberg MH, Lu ZH, Barton FB, Terrin ML, Charache S, Dover GJ. Fetal hemoglobin in sickle cell anemia: determinants of response to hydroxyurea. Blood. 1997;89(3):1078-1088. 7. Charache S. Mechanism of action of hydroxyurea in the management of sickle cell anemia in adults. Semin Hematol. 1997;34(3):15-21. 8. Platt OS. Hydroxyurea for the treatment of sickle cell anemia. N Engl J Med. 2008;358(13):1362-1369. 9. Steinberg MH, McCarthy WF, Castro O, et al. The risks and benefits of long-term use of hydroxyurea in sickle cell anemia: A 17.5 year follow-up. Am J Hematol. 2010; 85(6):403-408.

haematologica | 2017; 102(12)

late protein synthesis and stability of key TFs in the γ-globin promoter complex, as well as other γ-globin modulators such as lysine specific demethylase 1 (LSD1) and GPC1 that could regulate γ-globin through pathways independent of HU,13,50 may provide targets for designing new SCD drugs to ameliorate the SCD symptoms of HU low/non-responsive patients. Acknowledgments The authors would like to thank T. Horne and Drs. R. Vega, C. Neunert and B. Pace for blood samples of pediatric SCD patients, and B. Claire and N. Barrett for blood samples of adult SCD patients, Dr. R. Bollag for exchanged blood samples of SCD patients from the MCG Blood Bank and for aphoresed, nucleated peripheral blood cells of normal donors from the MCG Tumor Cell Bank, Dr. S. Jane for antibody to NF-E4, THJ Huisman Hemoglobinopathy Laboratory for HPLC analysis of HbF levels, and Drs. C. Noguchi and A. Schechter for critical reading of the manuscript and insightful comments and suggestions. Funding The work was supported by P20MD003383 from National Institute on Minority Health and Health Disparities.

10. McGann PT, Ware RE. Hydroxyurea for sickle cell anemia: what have we learned and what questions still remain? Curr Opin Hematol. 2011;18(3):158-165. 11. Hillery CA, Du MC, Wang WC, Scott JP. Hydroxyurea therapy decreases the in vitro adhesion of sickle erythrocytes to thrombospondin and laminin. Br J Haematol. 2000;109(2):322-327. 12. Haynes J Jr, Obiako B, Hester RB, Baliga BS, Stevens T. Hydroxyurea attenuates activated neutrophil-mediated sickle erythrocyte membrane phosphatidylserine exposure and adhesion to pulmonary vascular endothelium. Am J Physiol Heart Circ Physiol. 2008;294(1):H379-385. 13. Mabaera R, West RJ, Conine SJ, et al. A cell stress signaling model of fetal hemoglobin induction: what doesn't kill red blood cells may make them stronger. Exp Hematol. 2008;36(9):1057-1072. 14. Cokic VP, Smith RD, Beleslin-Cokic BB, et al. Hydroxyurea induces fetal hemoglobin by the nitric oxide-dependent activation of soluble guanylyl cyclase. J Clin Invest. 2003;111(2):231-239. 15. Cokic VP, Andric SA, Stojilkovic SS, Noguchi CT, Schechter AN. Hydroxyurea nitrosylates and activates soluble guanylyl cyclase in human erythroid cells. Blood. 2008;111(3):1117-1123. 16. Browning DD, McShane MP, Marty C, Ye RD. Nitric oxide activation of p38 mitogenactivated protein kinase in 293T fibroblasts requires cGMP-dependent protein kinase. J Biol Chem. 2000;275(4):2811-2816. 17. Ikuta T, Ausenda S, Cappellini MD. Mechanism for fetal globin gene expression: role of the soluble guanylate cyclasecGMP-dependent protein kinase pathway. Proc Natl Acad Sci USA. 2001;98(4):18471852. 18. Ramakrishnan V, Pace BS. Regulation of gamma-globin gene expression involves signaling through the p38 MAPK/CREB1

19.

20. 21.

22.

23.

24.

25.

26.

27.

pathway. Blood Cells Mol Dis. 2011;47(1):12-22. Liberati C, Ronchi A, Lievens P, Ottolenghi S, Mantovani R. NF-Y organizes the γ-globin CCAAT boxes region. J Biol Chem. 1998;273(27):16880-16889. Duan Z, Stamatoyannopoulos G, Li Q. Role of NF-Y in regulation of γ-globin gene. Mol Cell Biol. 2001;21(9):3083-3095. Zhu X, Wang Y, Pi W, Liu H, Wickrema A, Tuan D. NF-Y recruits both transcription activator and repressor to modulate tissueand developmental stage-specific expression of human gamma-globin gene. PLoS One. 2012;7(10):e47175. Filipe A, Li Q, Deveaux S, et al. Regulation of embryonic/fetal globin genes by nuclear hormone receptors: a novel perspective on hemoglobin switching. EMBO J. 1999;18(3):687-697. Liberati C, Cera MR, Secco P, et al. Cooperation and competition between the binding of COUP-TFII and NF-Y on human epsilon- and gamma-globin gene promoters. J Biol Chem. 2001;276(45):4170041709. Tanabe O, McPhee D, Kobayashi S, et al. Embryonic and fetal beta-globin gene repression by the orphan nuclear receptors, TR2 and TR4. EMBO J. 2007;26(9):22952306. Aerbajinai W, Zhu J, Kumkhaek C, Chin K, Rodgers GP. SCF induces gamma-globin gene expression by regulating downstream transcription factor COUP-TFII. Blood. 2009;114(1):187-194. Ikonomi P, Noguchi CT, Miller W, Kassahun H, Hardison R, Schechter AN. Levels of GATA-1/GATA-2 transcription factors modulate expression of embryonic and fetal hemoglobins. Gene. 2000; 261(2):277-287. Liu LR, Du ZW, Zhao HL, et al. T to C substitution at -175 or -173 of the gamma-globin promoter affects GATA-1 and Oct-1

2003

X. Zhu et al.

28.

29.

30.

31.

32.

33.

34.

35.

2004

binding in vitro differently but can independently reproduce the hereditary persistence of fetal hemoglobin phenotype in transgenic mice. J Biol Chem. 2005;280(9):7452-7459. Zhou W, Clouston DR, Wang X, et al. Induction of human fetal globin gene expression by a novel erythroid factor, NFE4. Mol Cell Biol. 2000;20(20):7662-7672. Jiang J, Best S, Menzel S, et al. cMYB is involved in the regulation of fetal hemoglobin production in adults. Blood. 2006;108(3):1077-1083. Menzel S, Garner C, Gut I, et al. A QTL influencing F cell production maps to a gene encoding a zinc-finger protein on chromosome 2p15. Nat Genetics. 2007; 39(10):1197-1199. Jawaid K, Wahlberg K, Thein S, Best S. Binding patterns of BCL11A in the globin and GATA1 loci and characterization of the BCL11A fetal hemoglobin locus. Blood Cells Mol Dis. 2010;45(2):140-146. Xu J, Sankaran VG, Ni M, et al. Transcriptional silencing of gamma-globin by BCL11A involves long-range interactions and cooperation with SOX6. Genes Dev. 2010;24(8):783-798. Kuroyanagi Y1, Kaneko Y, Muta K, et al. cAMP differentially regulates gamma-globin gene expression in erythroleukemic cells and primary erythroblasts through cMyb expression. Biochem Biophys Res Commun. 2006;344(3):1038-1047. Chen Z, Luo H, Steinberg M, Chui D. BCL11A represses HBG transcription in K562 cells. Blood Cells Mol Dis. 2009;42(2):144-149. Kang JA, Zhou Y, Weis TL, et al.

36. 37.

38.

39.

40.

41.

42.

Osteopontin regulates actin cytoskeleton and contributes to cell proliferation in primary erythroblasts. J Biol Chem. 2008; 283(11):6997-7006. Kutlar, F. Diagnostic approach to hemoglobinopathies. Hemoglobin. 2007;31(2):243250. Hurta RA, Wright JA. Amplification of the genes for both components of ribonucleotide reductase in hydroxyurea resistant mammalian cells. Biochem Biophys Res Commun. 1990;167(1):258-264. Yang YM, Pace B, Kitchens D, Ballas SK, Shah A, Baliga B. S. BFU-E colony growth in response to hydroxyurea: correlation between in vitro and in vivo fetal hemoglobin induction. Am J Hematol. 1997;56(4):252-258. Wang M, Tang DC, Liu W, et al. Hydroxyurea exerts bi-modal dose-dependent effects on erythropoiesis in human cultured erythroid cells via distinct pathways. Br J Haematol. 2002; 119(4):10981105. Grieco A, Billett H, Green NS, Driscoll MC, Bouhassira EE. Variation in Îł-globin expression before and after induction with hydroxyurea associated with BCL11A, KLF1 and TAL1. PLoS One. 2015; 10(6):e0129431. Pule GD, Mowla S, Novitzky N, Wonkam A. Hydroxyurea down-regulates BCL11A, KLF-1 and MYB through miRNA-mediated actions to induce gamma-globin expression: implications for new therapeutic approaches of sickle cell disease. Clin Transl Med. 2016;5(1):1-15. Sclafani S, Pecoraro A, Agrigento V, et al. Study on Hydroxyurea Response in

43.

44.

45.

46.

47.

48.

49.

50.

Hemoglobinopathies Patients Using Genetic Markers and Liquid Erythroid Cultures. Hematol Rep. 2016;8(4):56-60. Ma Q, Wyszynski DF, Farrell J, et al. Fetal hemoglobin in sickle cell anemia: genetic determinants of response to hydroxyurea. Pharmacogenomics J. 2007;7(6):386-394. Hahn CK, Lowrey CH Induction of fetal hemoglobin through enhanced translation efficiency of Îł-globin mRNA. Blood. 2014; 124(17):2730-2734. Alvarez-Dominguez JR, Zhang X, Hu W. Widespread and dynamic translational control of red blood cell development. Blood. 2017;129(5):619-629. Walker A, Steward S, Howard TA, et al. Epigenetic and molecular profiles of erythroid cells after hydroxyurea treatment in sickle cell anemia. Blood. 2011; 118(20):5664-5670. Towatari M, Ciro M, Ottolenghi S, Tsuzuki S, Enver T. Involvement of mitogen-activated protein kinase in the cytokine-regulated phosphorylation of transcription factor GATA-1. Hematol J. 2004;5(3):262-272. Hernandez-Hernandez A, Ray P, Litos G, et al. Acetylation and MAPK phosphorylation cooperate to regulate the degradation of active GATA-1. EMBO J. 2006;25(14):32643274. de Thonel A, Vandekerckhove J, Lanneau D, et al. HSP27 controls GATA-1 protein level during erythroid cell differentiation. Blood. 2010;116(1):85-96. Cui S, Lim KC, Shi L, et al. The LSD1 inhibitor RN-1 induces fetal hemoglobin synthesis and reduces disease pathology in sickle cell mice. Blood. 2015;126(3):386396.

haematologica | 2017; 102(12)

ARTICLE

Hemostasis

Molecular and clinical profile of von Willebrand disease in Spain (PCM-EVW-ES): comprehensive genetic analysis by next-generation sequencing of 480 patients

Nina Borràs,1,2 Javier Batlle,3 Almudena Pérez-Rodríguez,3 María Fernanda López-Fernández,3 Ángela Rodríguez-Trillo,3 Esther Lourés,3 Ana Rosa Cid,4 Santiago Bonanad,4 Noelia Cabrera,4 Andrés Moret,4 Rafael Parra,1,5 María Eva Mingot-Castellano,6 Ignacia Balda,7 Carme Altisent,5 Rocío Pérez-Montes,8 Rosa María Fisac,9 Gemma Iruín,10 Sonia Herrero,11 Inmaculada Soto,12 Beatriz de Rueda,13 Víctor Jiménez-Yuste,14 Nieves Alonso,15 Dolores Vilariño,16 Olga Arija,17 Rosa Campos,18 María José Paloma,19 Nuria Bermejo,20 Rubén Berrueco,21 José Mateo,22 Karmele Arribalzaga,23 Pascual Marco,24 Ángeles Palomo,6 Lizheidy Sarmiento,25 Belén Iñigo,26 María del Mar Nieto,27 Rosa Vidal,28 María Paz Martínez,29 Reyes Aguinaco,30 Jesús María César,31 María Ferreiro,32 Javier García-Frade,33 Ana María Rodríguez-Huerta,34 Jorge Cuesta,35 Ramón Rodríguez-González,36 Faustino García-Candel,37 Rosa Cornudella,38 Carlos Aguilar,39 Francisco Vidal*1,2,40 and Irene Corrales*1,2 Banc de Sang i Teixits, Barcelona; 2Vall d’Hebron Research Institute, Universitat Autònoma de Barcelona (VHIR-UAB); 3Complexo Hospitalario Universitario A Coruña, INIBIC; 4Hospital Universitario y Politécnico La Fe, Valencia; 5Hospital Universitari Vall d’Hebron, Barcelona; 6Hospital Regional Universitario de Málaga; 7Hospital Universitario Dr. Negrín, Las Palmas de Gran Canaria; 8Hospital Universitario Marqués de Valdecilla, Santander; 9Salud Castilla y León, Segovia; 10Hospital Universitario Cruces, Barakaldo; 11 Hospital Universitario de Guadalajara; 12Hospital Universitario Central de Asturias, Oviedo; 13Hospital Universitario Miguel Servet, Zaragoza; 14Hospital Universitario La Paz, Madrid; 15Hospital Infanta Cristina, Badajoz; 16Complejo Hospitalario Universitario Santiago de Compostela; 17Hospital Universitario Lucus Augusti, Lugo; 18Hospital Jerez de la Frontera, Cádiz; 19Hospital Virgen del Camino, Pamplona; 20Hospital San Pedro de Alcántara, Cáceres; 21Hospital Sant Joan de Deu, Barcelona; 22Hospital de la Santa Creu i Sant Pau, Barcelona; 23Hospital Universitario Fundación de Alcorcón, Madrid; 24Hospital General de Alicante; 25Hospital Universitario 12 de Octubre, Madrid; 26Hospital Clínico San Carlos, Madrid; 27Complejo Hospitalario de Jaén; 28Fundación Jiménez Díaz, Madrid; 29 Hospital Nuestra Señora de Sonsoles, Ávila; 30Hospital Universitario de Tarragona Joan XXIII; 31Hospital Ramón y Cajal, Madrid; 32Hospital Montecelo, Pontevedra; 33 Hospital del Río Hortega, Valladolid; 34Hospital Gregorio Marañón, Madrid; 35Hospital Virgen de la Salud, Toledo; 36Hospital Severo Ochoa, Madrid; 37Hospital Clínico Universitario Virgen de la Arrixaca, Murcia; 38Hospital Clínico Universitario Lozano Blesa, Zaragoza; 39Hospital Santa Bárbara, Soria and 40CIBER de Enfermedades Cardiovasculares, Spain

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(12):2005-2014

*IC and FV contributed equally to this work

Correspondence: icorrales@bst.cat or fvidal@bst.cat

Received: March 16, 2017. Accepted: September 20, 2017. Pre-published: September 29, 2017.

ABSTRACT

olecular diagnosis of patients with von Willebrand disease is pending in most populations due to the complexity and high cost of conventional molecular analyses. The need for molecular and clinical characterization of von Willebrand disease in Spain prompted the creation of a multicenter project (PCM-EVW-ES) that resulted in the largest prospective cohort study of patients with all types of von Willebrand disease. Molecular analysis of relevant regions of the VWF, including intronic and promoter regions, was achieved in the 556 individuals recruited via the development of a simple, innovative, relatively low-cost protocol based on microfluidic technology and next-generation sequencing. A total of 704 variants (237 different) were identified along VWF, 155 of which had not been previously recorded in the international mutation database. The potential pathogenic effect of these variants was assessed by in silico analysis. Furthermore, four short tandem repeats were analyzed in order to evaluate the ancestral origin of recurrent mutations. The outcome of genetic analysis allowed for the reclassification of 110 patients, identification of 37 asymptomatic carriers (important for genetic counseling) and re-inclusion of 43 patients previously excluded by phenotyping results. In total, 480 patients were haematologica | 2017; 102(12)

doi:10.3324/haematol.2017.168765 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2005 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2005

N. Borràs et al.

definitively diagnosed. Candidate mutations were identified in all patients except 13 type 1 von Willebrand disease, yielding a high genotype-phenotype correlation. Our data reinforce the capital importance and usefulness of genetics in von Willebrand disease diagnostics. The progressive implementation of molecular study as the first-line test for routine diagnosis of this condition will lead to increasingly more personalized and effective care for this patient population. clinicaltrials.gov Identifier: 02869074.

Introduction Von Willebrand disease (VWD) is the most common bleeding disorder, with a reported incidence of 0.01% to 1%.1,2 The condition is caused by abnormalities related to the von Willebrand factor protein (VWF), which has an important role in primary hemostasis.3,4 A diagnostic algorithm based on the patient’s clinical history and the results of laboratory testing to assess VWF levels and functionality has enabled the classification of VWD into quantitative (type 1 and type 3) and qualitative (type 2) abnormalities.1 The first-line tests used to classify VWD include the combination of VWF: antigen (Ag), VWF: ristocetin cofactor (RCo), factor VIII:C (FVIII:C), and the VWF:RCo/VWF:Ag ratio, whereas second-line tests include ristocetin-induced platelet aggregation (RIPA), VWF:FVIIIB, VWF: collagen binding (CB), VWF propeptide (pp), and multimer analysis, which are especially useful to differentiate between the type 2 VWD subtypes (2A, 2B, 2M, and 2N). It is essential to correctly identify these subtypes because their treatment differs. Nonetheless, this can be a challenging task as clinical and laboratory phenotypes are very heterogeneous,5 and some tests, such as RIPA and multimeric analysis, are not available in all laboratories.6 Furthermore, there is often a lack of clinical information, and the available laboratory tests are known to have intrinsic limitations.1,7,8 In addition, sequential application of these analyses makes phenotypic VWD diagnosis increasingly more costly. The VWF glycoprotein is encoded by VWF gene, a large (178 kb comprising 52 exons) and very polymorphic gene (>100 coding single-nucleotide polymorphisms (cSNPs)) with a highly homologous partial pseudogene (VWFP; exons 23-34),9 characteristics that make specific amplification and sequencing of VWF difficult. Thus, molecular analysis of VWF is a challenge. Although genetic analysis is considered a valuable tool to support the diagnosis of VWD,10,11 classic Sanger sequencing of the entire VWF is too costly for general use in all patients. Hence, this method has been applied, with slight variations, only to specific exons of VWF, depending on the VWD subtype. Exon 28 is sequenced for 2A, 2B, and 2M patients, and in type 2A, if a mutation is not found in exon 28, gene study is extended to exons 11-17 and 50-52. In 2N cases, exons 18 to 20 are sequenced, corresponding to the FVIII-binging site. Finally, analysis of the complete coding sequence is required in type 1 and type 3 patients, as potential mutations may be spread all along VWF.12 Nonetheless, the advent of next-generation sequencing (NGS) technology is making molecular diagnosis of VWD by complete VWF sequencing faster and progressively less costly,11 in particular when the new “desktop instruments” and simplified, optimized technologies for library preparation are used.13 Based on the results of a previous Spanish survey elucidating the difficulties of diagnosing VWD,7 a multicenter, prospective project (PCM-EVW-ES, Molecular and Clinical Profile of von Willebrand Disease in Spain) was designed to centrally characterize a large multicenter cohort of 2006

VWD patients, with inclusion of NGS molecular analysis. As described in our previous report on this project,14 this new technology enabled us to undertake a nationwide molecular epidemiologic study that supported a new scenario for VWD diagnostics, with genetic analysis being an indispensable first-line tool. Herein, we provide an in-depth description of the technical aspects of NGS-based molecular characterization of VWF in the 556 individuals studied, complete details of the identified variants, and a description of the results of a genotypic-phenotypic correlation study in patients diagnosed with all types of VWD.

Methods Patients Molecular studies were performed on samples from 556 individuals with locally diagnosed VWD recruited in 38 Spanish hospitals participating in the PCM-EVW-ES project.7,14 The inclusion criteria were one or more of the following: 1) VWF:Ag, VWF:RCo and/or VWF:CB ≤30% on at least two occasions regardless of the blood group, 2) detection of multimeric abnormalities, 3) evidence of decreased VWF:FVIIIB level in isolated FVIII deficiency, 4) presence of VWF candidate mutations, and 5) RIPA at a low ristocetin concentration.14 Bleeding score (BS) and central laboratory phenotypic assessment are available in Online Supplementary Methods. The study was performed according to the Declaration of Helsinki, was approved by the local Research Ethics Committee, and all participants provided written informed consent.

VWF Access Array amplification and sequencing We designed 61 pairs of oligonucleotides (Accession NG_029217) capable of amplifying exons 1 to 52, intronic flanking regions, and approximately 1300 bp of the promoter region that were described in our previous report.14 The 48.48 Access Array integrated fluidic circuit (IFC; Fluidigm, San Francisco, CA, USA) is a nanofluidic chip that allows 2304 different polymerase chain reactions (PCRs) in a final volume of 30 nL by combining 48 samples with 48 primer pairs. As 61 primer pairs are needed for total VWF amplification, 12 multiplex PCRs were created, which allowed 2928 reactions per chip. The Access Array was processed following the manufacturers’ recommendations (Online Supplementary Methods and Online Supplementary Table S1). The outcome was a pool of all VWF amplicons from the same sample, plus a distinctive short sequence identification label that acted as a barcode, incorporated into each set of PCRs during the amplification step. The final pools of up to 192 samples (four Access Array fusion libraries) were then combined and sequenced in a MiSeq platform (Illumina, San Diego, CA, USA; Online Supplementary Figure S1).

Data analysis and identification of genetic variants Barcoded sequences were demultiplexed and analyzed individually. The NGS pipeline output, paired sequence files (FASTQ format), was used as input for analysis with the CLC Genomic Workbench software 8.0.2 (Qiagen, Hilden, Germany). After varihaematologica | 2017; 102(12)

Genetic analysis by NGS of PCM-EVW-ES

ant calling, the resulting files (VCF) were used as input for VariantStudio 2.2.1 (Illumina). The result of this step is the identification of potential pathogenic variants and filtering of the polymorphisms described to date in the SNP (dbSNP) and 1000 Genomes databases (Figure 1). The specific analytical parameters used are described in Online Supplementary Methods. Selected mutations detected were validated/confirmed by Sanger sequencing.10 The nomenclature and criteria used to establish variant pathogenicity is also provided in Online Supplementary Methods.

labeled, up to 384 samples can be pooled and sequenced in a single run. Fifteen Access Array and 5 MiSeq 500cycle runs were needed to process the 556 samples (Figure 1, Online Supplementary Results). The output quality parameters obtained for each run and the resulting mean values are shown in Online Supplementary Table S2. In total, 25620 bp of VWF were sequenced in all samples, with homogeneous coverage for nearly all 61 amplicons (Online Supplementary Figure S2). Regions with low or no coverage were completed by Sanger sequencing.10

In silico analysis In silico prediction to evaluate the functional effects of putative pathogenic variants was performed using the Alamut Visual v.2.6.1 software (Interactive Biosoftware, Rouen, France). Missense prediction tools included PolyPhen-2, the Sorting Intolerant From Tolerant (SIFT), Mutation Taster, Mutation Assessor and Provean, whereas predictive tools for synonymous and splice site candidate mutations included GeneSplicer, Splicing Sequences Finder (SSF), Human Splicing Finder (HSF), Neural Network (NN) Splice and Maximum Entropy Modeling (MaxEnt). Upstream VWF variants were localized, visualized by the Integrative Genomics Viewer software, and compared in parallel with datasets from the Encyclopedia of DNA Elements (ENCODE) project to identify potential regulatory regions.15

Multiplex ligation-dependent probe amplification (MLPA) VWF deletions/duplications were detected by MLPA using the SALSA MLPA P011 and P012 VWF kits (version B2; MRCâ&#x20AC;&#x201C; Holland, Amsterdam, The Netherlands), as previously described.14,16 This method was applied to 5 patients whose genotype did not correlate with the phenotype

Microsatellite analysis A multiplex fluorescent PCR described by our group,17 comprising 3 VWF intragenic tetranucleotide short tandem repeats (STR1, STR-2, and STR-3) and 1 dinucleotide repeat in the promoter region (WPA) was applied to genomic DNA samples.

Genotype-phenotype correlation The correlation between genotype and phenotype was assessed by experts from the central laboratories of the PCM-EVW-ES who contrasted the results of the phenotypic test panel and the genetic analysis on the basis of the effect and localization of mutations and previous descriptions in the literature and/or databases.

Genotype-phenotype correlation and classification Of the 556 individuals recruited in the PCM-EVW-ES database, 442 had confirmed VWD based on central phenotype characterization. Mutation analysis by NGS was performed in all 556 individuals, and MLPA was additionally used in five (03012, 05011, 10011, 23010, 25005). Two large deletions were identified: c.1157_5620del in patient 03012 (type 2A) due to homologous recombination between short sequences of introns 10 and 32, and c.(1945+1_1946-1)_(7437+1_7438-1)del in the type 3 patient, 25005 (exact breakpoints pending characterization). Following mutation analysis, 105 were reclassified to a different subtype, five were excluded and 43 patients were reincluded (Figure 2) due to the presence of a candidate mutation in VWF: 19 type 3 VWD carriers, one type 2N VWD carrier, four type 1 VWD with borderline levels, nine type 1H VWD, three type 2M VWD with collagen binding mutations, and seven patients with uncertain classification. All these data, based on phenotyping and genotyping results, leads to a final diagnosis of VWD in 480 patients from 280 families (mean 1.7 members per family; range 1-21) who met the criteria for inclusion in the VWF registry (Figure 3). A total of 704 variants were identified, 237 were different, and 155 had not been described in the VWF European Association for Haemophilia and Allied Disorders (EAHAD) Coagulation Factor Variant Database compiled in the Leiden Open Variation Database (LOVD; EAHAD-VWD-LOVD).18 The patientsâ&#x20AC;&#x2122; phenotypic and molecular data were categorized according to the VWD type and used to determine genotype-phenotype correlations, which were established in 94.6% of subjects (Figure 4). Distribution of the families by VWD subtypes is depicted in Figure 5.

Molecular epidemiology per type of VWD Results Nanofluidic VWF amplification and NGS output The selected VWF regions were amplified with the Fluidigm Access Array. As each library is distinctively

VWD type 1 was diagnosed in 159 patients. In total, 216 variants were identified in 146 patients (91.8%; Online Supplementary Table S3). Eighty-five different variants (54 undescribed in EAHAD-VWD-LOVD)18 were scattered over VWF, being more frequent in the D3 and D4 domains (45.9%). The five most recurrent mutations

Table 1. VWF recurrent mutations from 480 patients included in PCM-EVW-ES, classified by VWD types.

haematologica | 2017; 102(12)

2007

N. BorrĂ s et al.

Figure 1. Flowchart depicting the molecular analysis and identification of potential mutations in individuals enrolled in the study. VWF from 48 patients were simultaneously amplified in a 48.48 Access-Array. Alignment, variant calling and annotation of each variant identified was performed by the CLC Genomic Workbench software. This analysis allowed selection of mutations/variants aligned against VWF (shown in gray) and elimination of those aligned against VWFP (shown in black). Variant filtering was performed by the Variant Studio software. *MAF<0.01 for all variants except three (p.Arg854Gln, p.Arg924Gln, p.Pro2063Ser). VWF: von Willebrand factor gene; VWFP: von Willebrand factor pseudogene; dbSNP: dbSNP database; 1000G: 1000 Genomes Project; ExAC: Exome Aggregation Consortium; MAF: minor allele frequency; NGS: next-generation sequencing.

(Table 1) had been previously reported in type 1 VWD, except for p.Leu1733Pro, with a deleterious in silico score of 4. Patients with the p.Arg1205His, p.Pro1824His, or p.Leu1733Pro mutations showed the lowest levels of VWF:Ag and VWF:RCo (mean 9.1 and 7, respectively). In 30 patients subclassified as VWD 1H,14,19 the most frequent mutation was p.Arg924Gln (six patients), previously described as a mutation and a polymorphism (Online Supplementary Results). With the exception of 13 patients with no detected variant (mean VWF:Ag=31%; range 16-48; 78% blood group O) and patient 01030 who carried the classic p.Arg854Gln 2N mutation, a good phenotype-genotype correlation was established in 145 patients. Forty-two patients classified as type 3 VWD (mean VWF:Ag=1.2%; mean BS=12.6) and 26 as carriers (mean VWF:Ag=55%; mean BS=1.7) were present in 51 different families (Online Supplementary Table S4). In total, 60 different potential mutations were identified throughout VWF, 2008

and 34 had not been previously described in EAHADVWD-LOVD,18 of which 12 were mutations causing null alleles. Genotyping was clear and correlated with the clinical and laboratory data in all patients except three: patients 05011 and 10011 with a mutation (p.Cys1946Phe and c.5455+1G>A, respectively) in heterozygous state, and patient 10006 with a homozygous nonsense mutation (p.Gln2470Ter) and normal laboratory levels (pending new analysis from a freshly obtained sample). Excluding these three cases, the 74.4% of type 3 patients had mutations leading to incontrovertible null alleles in both chromosomes (Online Supplementary Figure S3A). Type 2A VWD was finally diagnosed in 111 patients (detailed classification is disclosed in Online Supplementary Results). In total, 158 variants were identified (Online Supplementary Table S5) and most were clustered in the A1 and A2 domains (38 and 57 potential mutations, respectively) encoded by exon 28. Ninety-eight patients (88%) showed a mutation classified as 2A in EAHAD-VWDhaematologica | 2017; 102(12)

Genetic analysis by NGS of PCM-EVW-ES

Figure 2. Classification of VWD patients based on central phenotypic diagnosis and final assignment according to the genotype-phenotype correlation. The dataset highlights the number of patients reincluded or reclassified after molecular study, from the initial classification based on central phenotypic results (shown in black) to a definite, refined classification based on molecular data analysis together with phenotypic results (shown in gray). *Patients diagnosed initially as type 1 VWD and reclassified to type 2M VWD due to the presence of collagen binding mutations. †Patient with uncertain classification reclassified to type 3 VWD due to the presence of nonsense mutation in homozygous state. This is a discrepant case since laboratory levels do not correlate and it is pending a new analysis from a freshly obtained sample. Out: number of patients removed from this subtype; To: final definite classification; In: number of patients reclassified to this subtype; From: previous phenotypic classification; UC: uncertain classification; AVWS: acquired von Willebrand Syndrome.

LOVD,18 and the remaining 13 patients had novel candidate mutations in one of the 2A-associated domains (D2, D3, A1, A2, and CK). Hence, an unflawed phenotypegenotype correlation was found in all except four patients: patient 42010, who showed a previously described 2A mutation (p.Arg1527Gln)20 with healthy compatible VWF:Ag and VWF:RCo levels (pending new analysis of a freshly obtained sample), patient 03012, who had a large in-frame deletion affecting residue 386-1873 (domains D2A3), although this type of mutation seldom explains a 2A phenotype21 and additional studies are needed to investigate the molecular origin, and patients 02071 and 02072, who had a nearly normal multimeric pattern and VWF:RCo/VWF:Ag>0.7 due to the heterozygous missense mutations, p.Arg976Cys (described as 2A/IIE)22 and p.Pro2063Ser. Thirty-four patients were classified as type 2A/2M VWD (Figure 3 and Online Supplementary Table S6). Remarkably, all families save three presented one of the widely described p.Arg1315Cys and p.Arg1374Cys mutations, located in the A1 domain.23,24 These two controversial mutations were difficult to classify in the light of previous reports and were assigned to 2A/2M, mainly because of their abnormal VWF multimers in medium-resolution gels.25 The new p.Arg763Ser potential mutation, involved in proteolytic processing of the VWF precursor, was found in heterozygous state in two families with symptomatic patients (BS >4), a smeary multimeric pattern, and VWF:FVIIIB <0.7. Of note, mutations in the haematologica | 2017; 102(12)

same amino acid (p.Arg763Gly) have been described in type 2N, but heterozygous patients for this mutation were asymptomatic and classified as 2N carriers.26,27 The remaining family showed p.Cys2491Arg, described as type 3, which involves loss of a cysteine essential for VWF secretion, but these patients were classified as 2A/2M based on their multimeric pattern. Taking these data together, a complete genotype-phenotype correlation could be established for all 2A/2M patients. Thirty-five patients were diagnosed as having type 2B VWD (detailed classification is disclosed in Online Supplementary Results and Online Supplementary Table S7). Interestingly, among the 16 residues associated with type 2B VWD, one out of five of them—p.Pro1266, p.Arg1306, p.Arg1308, p.Val1316 or p.Pro1337—was found to be affected. A good phenotype-genotype correlation could be established for all patients, as most showed a loss of highmolecular-weight multimers and discordance between VWF:Ag and VWF:RCo levels (mean ratio=0.51; range 0.19-1.1) and a classical type 2B mutation. A normal multimeric pattern and VWF:RCo/VWF:Ag>0.7 were observed in only two families. As expected, the classical p.Pro1266Leu and p.Pro1266Gln molecular defects were found, both being responsible for type 2B Malmö/New York VWD.28,29 Thirty-nine patients were classified as type 2M VWD (Online Supplementary Table S8). Twenty-one different missense variants were found, ten undescribed in EAHADVWD-LOVD.18 Of interest, in eight patients with a high 2009

N. BorrĂ s et al.

Figure 3. Summary of the final diagnosis of patients in terms VWD type. On the basis of phenotype data, 98 of the 556 recruited individuals did not meet any inclusion criteria and were initially excluded. Following the molecular analysis, 43 of these patients were reincluded due to the presence of a candidate mutation in VWF. The remaining 55 patients did not meet any inclusion criteria. Moreover, molecular analysis prompted the exclusion of 21 additional individuals: eight HA patients, five HA carriers, and eight patients finally diagnosed as AVWS and confirmed by the absence of mutations in VWF or GP1BA. Of note, 280 families were finally included (shown in black) and nine of them had members with different VWD types. In those particular cases, the family may be counted more than once; that is, within each VWD type where a family member was classified. VWD: von Willebrand disease.

VWF:RCo/VWF:Ag ratio, collagen-binding mutations were identified: p.Arg1399His, p.Ser1731Thr, and the novel p.Arg1395Trp.30 All patients showed mutations in the 2M-associated domains A1 and A3, with the exception of three patients who had mutations in other domains, and patient 27022 (carrying p.Ser1731Thr, a collagen I/III-binding mutation, but with VWF:CB=41), considered to have a discrepant genotype-phenotype correlation. A perfect phenotype-genotype correlation was established in all type 2N VWD patients (Online Supplementary Table S9). Two classical 2N mutations were present in all patients: p.Arg816Trp in exon 19, and p.Arg854Gln in exon 20, the latter seen in 75% of cases. Different combinations of mutation types were found (Online Supplementary Figure S3B): five patients homozygous for a 2N mutation, and four patients compound heterozygous in trans for p.Arg854Gln and a nonsense mutation or p.Gln895His, reported to cause low VWF expression at the messenger ribonucleic acid (mRNA) level.31 In 14 patients the classification was uncertain (Online Supplementary Table S10) because phenotyping or genotyping data were insufficient or inconclusive to indubitably establish the classification. This occurred in eight patients with novel variants of unknown significance (VUS) who had a normal or nearly normal phenotype, and six patients with an abnormal multimeric pattern, VWF:RCo/VWF:Ag <0.7, and a novel candidate mutation. 2010

In these cases, it was difficult to distinguish between 2A or 2B VWD since RIPA assay was not assessed.

Molecular epidemiology per type of mutation A total of 237 different variants were identified in our cohort (Online Supplementary Figure S4A) and a potential ancestral origin was established in 19 of them. All variants, classified by type, are listed in Online Supplementary Tables S11-S17. In total, 119 different missense candidate mutations were identified, scattered over all domains of VWF (33 of 52 exons). Sixty were not included in EAHAD-VWDLOVD18 (Online Supplementary Figure S4B) and the mean in silico score in this group was 3.50 (standard deviation (SD) 1.60). The remaining 59 mutations, previously found in VWD patients, had a similar score of 3.54 (SD 1.62). Twenty-three different nonsense mutations were found, 12 unrecorded in EAHAD-VWD-LOVD.18 As expected, nonsense mutations predominantly occurred in type 3 VWD (41 patients). However, we also found this type of mutation in nine type 1 VWD patients, two type 2N, and one type 2A. Of particular note, several mutations were detected in different VWD types. For example, p.Arg324Ter was found in types 3 and 2N, and p.Gln840Ter, p.Gln1311Ter and p.Gln2470Ter were found in types 1 and 3. Five new inframe candidate mutations (two insertions and three deletions) causing different types of VWD were identified. Thirteen potential splicing haematologica | 2017; 102(12)

Genetic analysis by NGS of PCM-EVW-ES

Figure 4. Schematic representation of phenotype/genotype correlations by VWD type. The correlation between genotype and phenotype in each patient was based on the concordance between the results of the phenotypic test panel and the results of the genetic analysis.

mutations were found, eight undescribed in EAHADVWD-LOVD.18 Their deleterious effect was assessed by five in silico tools: 11 had a predicted effect on splicing, but two of them, those farthest from the exon to be precise, were predicted to have no significant impact on splicing. Ten different upstream variants were found in 20 families with all types of VWD. Only c.-1896C>T and c.-1873A>G have been previously described,32,33 and these were found in type 2A/2M and 2M, respectively. According to their genomic localization, all but the three most upstream (c.3151T>G, c.-2692C>T, c.-2627C>T) were mapped in regions with well-defined or potential regulatory elements. Sixteen synonymous and 51 intronic variants were identified, but only 5 and 3, respectively, were recorded in EAHAD-VWD-LOVD.18 Although it is difficult to demonstrate, any nucleotide change can be considered a potential candidate to alter splicing, especially when an intronic or synonymous variant appears next to an intron/exon junction.34,35

Discussion This report describes the molecular study of VWF in the largest prospective multicenter cohort of VWD patients to date, made possible by the development and successful application of NGS technology. The method presented herein has also been used in a smaller Portuguese cohort, with a similar diagnostic yield.16 Heretofore, VWF genetic analysis in large cohorts has been performed by Sanger sequencing.32,33,36,37 In the study herein, considerable time and cost savings were attained by combining two highthroughput technologies, nanofluidics and NGS. The hands-on time and instrument run times for sequencing the 556 patients initially enrolled can be measured in days or weeks rather than months, whereas the cost per sample in our setting was less than $70, 10-fold lower than Sanger sequencing. The method developed for VWF sequencing is fast and cost-effective for large patient groups; hence, it would seem inappropriate for routine diagnostic laboratories that accumulate small numbers of patients. Nonetheless, NGS haematologica | 2017; 102(12)

can also be applied to individual samples by using the gene panel approach.38 In our opinion, the debate about whether genetic analysis is appropriate for all types of VWD12 will become obsolete in the light of this new technological scenario. It is reasonable to believe that NGS will be progressively incorporated in routine VWD diagnostics. Centralized comprehensive population studies, although logistically complex, would be the best way to obtain a clear picture of this disease and investigate connections between the multifactorial parameters that influence the diagnosis and etiology.12 In this study, 704 variants (237 unique) were compiled (Online Supplementary Tables S11-S17), whereas the total number of entries in EAHAD-VWD-LOVD18 is about 1,200 (708 unique), an indication of the effectiveness of this approach for the genetic study of large cohorts. In contrast to other studies, the present project included patients with all VWD types and used an identical protocol (whole VWF sequencing) for their genetic analysis. The mutation detection rate was high compared with that of previous studies, being close to 100% in some VWD types (Figure 3). Candidate mutations were identified in 91.8% of type 1 individuals versus the 65-82% reported in previous studies,32,33,36,39 a result that may be related to the limiting recruitment of patients with VWF level <30%. With regard to the most recurrent mutations identified (Table 1), the Vicenza mutation (p.Arg1205His) proved to be the most common in our type 1 patients, in contrast to previous studies where p.Tyr1584Cys was prevalent.32,33,40 In type 3 VWD, p.Gln1311Ter41 was one of the most frequent mutations detected, as was also seen in the Portuguese population study,16 whereas in patients from northern Europe, p.Arg2535Ter is predominant. As to type 2, the most common mutations found were similar to those reported in previous cohort studies. Perhaps more important than elucidation of the molecular epidemiology of VWD were the data obtained by NGS, that can resolve many of the drawbacks and limitations of phenotyping. First, molecular characterization of VWD patients enables accurate classification: 27.5% of our patients (153/556) were re-diagnosed after the initial 2011

N. BorrĂ s et al.

Figure 5. Classification of the 280 families included according to VWD type. Numbers in parentheses refer to the total of families included in each VWD type. Of note, nine of the 280 families included in the PCM-EVW-ES presented a mixed phenotype (families in which some members had different VWD subtypes). UC: patients with an uncertain classification.

phenotypic study, 43 reincluded, five patients excluded, and 105 reclassified. Of note, 68 patients with uncertain classification were definitively classified by NGS, as in the case of 25 type 2A patients and ten type 2B missing previous RIPA data. Furthermore, four patients whose phenotype pointed to VWD type 1 were found to have 2B MalmĂś or collagen-binding mutations (Figure 2), the latter being undetectable by standard laboratory tests unless specific binding capacity for collagen types III, IV and VI is performed. Accurate classification is particularly relevant for genetic counseling and detection of carriers in VWD types with autosomal recessive inheritance. We identified 11 asymptomatic carriers 2N and 26 type 3 with borderline plasmatic levels, who would likely have been excluded if the diagnosis had been based only on clinical and laboratory data. Knowledge of the molecular defect also led to diagnostic reassessment of some type 1 patients who could be considered carriers of type 3 mutations (e.g., patient 06024 with the p.Arg2434Ter mutation). Second, genetic study showed that 37.5% of patients in our cohort had more than one variation, a finding essential to unravelling the potential contribution of different variants to the final phenotype. The situation of two 2M patients is paradigmatic in this regard: 44003, with the p.Gly1415Asp mutation and very low FVIII levels (11%) due to a 2N mutation in heterozygous state, and 03024, who had the p.Val1409Phe plus the p.Arg1399His mutation in trans, which explains the reduction in collagenbinding affinity in this patient. Third, genetic analysis in large cohorts could shed light on differentiating a variant previously described as a mutation or polymorphism, such as p.Pro1162Leu caused by the c.3485C>T substitution, described as a polymorphism in African Americans.42 Conversely, in our cohort p.Pro1162Leu was caused by c.3485_3486delinsTG (predicted by in silico analysis with effect on splicing), leading to type 3 phenotype in homozygous state.16 These data suggest that different variants resulting in identical amino acid changes may have different consequences at the transcriptional level. Additionally, regarding the controversial 2012

p.Pro2063Ser variant, a recent report by Kasatkar et al.43 has provided evidence that p.Pro2063Ser in homozygous state causes type 3 VWD. Hence, five patients in our cohort heterozygous for p.Pro2063Ser were classified as type 3 carriers. Fourth, genetic studies can offer valuable data regarding the molecular pathophysiological mechanisms responsible for the symptoms observed, such as defects in the structure, intracellular transport, or secretion of VWF. The location of new mutations in certain domains or specific amino acids helps to predict their potential effect, particularly if previous in vitro studies are performed. Examples of this are the new mutations p.Arg273Pro, located in the propeptide and related with formation of disulphidelinked multimers,44 p.Arg763Ser, located in the VWF propeptide cleavage site,26 and p.Arg1395Trp, which affects an essential amino acid for collagen IV binding.30 In addition, although this approach cannot resolve some discrepant cases and those in which no candidate mutation is found (3.1% of our cohort), exome and genome studies by NGS are useful to identify modifying mutations in other genes45 and reveal structural variations undetectable by MLPA.46 It is also important to point out that the technique has provided molecular data for more than 100 cSNPs described in VWF, which will be further analyzed to determine their potential influence on VWF:Ag, FVIII:C and bleeding in our cohort, as has been done in some healthy populations.42,47 If some of these SNPs are found to contribute to the disease, they could be redefined as hemorrhagic risk polymorphisms (similar to the thrombotic risk polymorphisms). The final obstacle to an accurate diagnosis is the uncertain pathogenicity of novel variations. In silico analysis is considered a suitable supporting tool for genetic diagnosis, and we analyzed all variants using algorithms recommended in specialized guidelines.48 However, the predictive capacity of these programs remains modest: some well-established deleterious mutations have a lower overall score (e.g., p.Leu1580Pro, score=2) than some polymorphisms (e.g., p.Cys325Phe, predicted to be deleterious by haematologica | 2017; 102(12)

Genetic analysis by NGS of PCM-EVW-ES

SIFT and PolyPhen).49 Therefore, in order to unequivocally determine the potential deleterious effect of new variants, in vitro functional studies remain essential.50 Herein, we present the largest prospective study to date of a cohort of VWD including an exhaustive description of their molecular epidemiology. This data will improve knowledge about the molecular mechanisms that contribute to the complexity of disease diagnosis and will help to elucidate the relationship between bleeding parameters and the patients’ laboratory and genetic profiles. The approach used implies a change in the diagnostic paradigm of VWD, which is also occurring in other genetic diseases. The cost reduction and simplification of genetic analysis by NGS will likely lead to the use of molecular study as an additional first-line routine diagnos-

References 1. Sadler JE, Budde U, Eikenboom JC, et al. Update on the pathophysiology and classification of von Willebrand disease: a report of the Subcommittee on von Willebrand Factor. J Thromb Haemost. 2006;4(10): 2103-2114. 2. Rodeghiero F, Castaman G, Dini E. Epidemiological investigation of the prevalence of von Willebrand's disease. Blood. 1987;69(2):454-459. 3. Ruggeri ZM. Von Willebrand factor, platelets and endothelial cell interactions. J Thromb Haemost. 2003;1(7):1335-1342. 4. Vlot AJ, Koppelman SJ, Meijers JC, et al. Kinetics of factor VIII-von Willebrand factor association. Blood. 1996;87(5):18091816. 5. Castaman G, Hillarp A, Goodeve A. Laboratory aspects of von Willebrand disease: test repertoire and options for activity assays and genetic analysis. Haemophilia. 2014;20 Suppl 4:65-70. 6. Hamilton A, Ozelo M, Leggo J, et al. Frequency of platelet type versus type 2B von Willebrand disease. An international registry-based study. Thromb Haemost. 2011;105(3):501-508. 7. Batlle J, Pérez-Rodríguez A, Costa-Pinto J, Lourés E, Rodriguez-Trillo A, LópezFernández MF. Diagnosis and management of von Willebrand disease in Spain. Semin Haemost Thromb 2011; 37(5):503-510. Semin Haemost Thromb. 2011;37(5):503510. 8. Flood VH. New insights into genotype and phenotype of VWD. Hematology Am Soc Hematol Educ Program. 2014;2014(1):531535. 9. Mancuso DJ, Tuley EA, Westfield LA, et al. Human von Willebrand factor gene and pseudogene: structural analysis and differentiation by polymerase chain reaction. Biochemistry. 1991;30(1):253-269. 10. Corrales I, Ramirez L, Altisent C, Parra R, Vidal F. Rapid molecular diagnosis of von Willebrand disease by direct sequencing. Detection of 12 novel putative mutations in VWF gene. Thromb Haemost. 2009;101(3):570-576. 11. Corrales I, Catarino S, Ayats J, et al. Highthroughput molecular diagnosis of von Willebrand disease by next generation sequencing methods. Haematologica. 2012; 97(7):1003-1007. 12. Goodeve AC. The genetic basis of von

haematologica | 2017; 102(12)

13. 14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

tic test, with the ultimate aim of providing more personalized and effective care for this patient population. Funding We are indebted to Baxalta US Inc., now a part of Shire, for its support of the PCM-EVW-ES (H13-000845 Grant). This work was also supported by the Spanish Ministerio de Economiá y Competitividad (MINECO)-Instituto de Salud Carlos III (ISCIII) (PI12/01494, PI15/01643 and RD12/0042/0053). We thank Technoclone GmbH, Austria, Labclinics, SA, Spain for donating the VWF:CB (type VI collagen) kit. We are very grateful for the kind collaboration of the participating patients and their families. CIBERCV is an initiative of ISCIII cofinanced by Fondo Europeo de Desarrollo Regional (FEDER) a way to build Europe.

Willebrand disease. Blood Rev. 2010;24(3): 123-134. Desai A, Jere A. Next-generation sequencing: ready for the clinics? Clin Genet. 2012;81(6):503-510. Batlle J, Perez-Rodriguez A, Corrales I, et al. Molecular and clinical profile of von Willebrand disease in Spain (PCM-EVWES): Proposal for a new diagnostic paradigm. Thromb Haemost. 2016;115(1):4050. Consortium EP, Bernstein BE, Birney E, et al. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489(7414):57-74. Fidalgo T, Salvado R, Corrales I, et al. Genotype-phenotype correlation in a cohort of Portuguese patients comprising the entire spectrum of VWD types: impact of NGS. Thromb Haemost. 2016;116(1):1731. Vidal F, Julia A, Altisent C, Puig L, Gallardo D. Von Willebrand gene tracking by singletube automated fluorescent analysis of four short tandem repeat polymorphisms. Thromb Haemost. 2005;93(5):976-981. European Association for Haemophilia and Allied Disorders (EAHAD). Coagulation Factor Variant Databases. October 01, 2010 ed. Montgomery RR, Christopherson P, Bellissimo DB, et al. The complete type I VWD cohort of the Zimmerman Program for the molecular and clinical biology of VWD - phenotypic assignment, mutation frequency, and bleeding assessment. Blood. 2013;122(21):332. Ahmad F, Jan R, Kannan M, et al. Characterisation of mutations and molecular studies of type 2 von Willebrand disease. Thromb Haemost. 2013;109(1):39-46. Haberichter SL, Allmann AM, Jozwiak MA, Montgomery RR, Gill JC. Genetic alteration of the D2 domain abolishes von Willebrand factor multimerization and trafficking into storage. J Thromb Haemost. 2009;7(4):641-650. Schneppenheim R, Michiels JJ, Obser T, et al. A cluster of mutations in the D3 domain of von Willebrand factor correlates with a distinct subgroup of von Willebrand disease: type 2A/IIE. Blood. 2010;115(23): 4894-4901. Penas N, Perez-Rodriguez A, Torea JH, et al. von Willebrand disease R1374C: type 2A or 2M? A challenge to the revised classification. High frequency in the northwest

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

of Spain (Galicia). Am J Hematol. 2005;80(3):188-196. Batlle J, Perez-Rodriguez A, Franqueira MD, Lopez-Fernandez MF. Type 2M von Willebrand disease: a variant of type 2A? J Thromb Haemost. 2008;6(2):388-390. Gadisseur A, van der Planken M, Schroyens W, Berneman Z, Michiels JJ. Dominant von Willebrand disease type 2M and 2U are variable expressions of one distinct disease entity caused by loss-of-function mutations in the A1 domain of the von Willebrand factor gene. Acta Haematol. 2009;121(23):145-153. Hilbert L, Nurden P, Caron C, et al. Type 2N von Willebrand disease due to compound heterozygosity for R854Q and a novel R763G mutation at the cleavage site of von Willebrand factor propeptide. Thromb Haemost. 2006;96(3):290-294. van den Biggelaar M, Meijer AB, Voorberg J, Mertens K. Intracellular cotrafficking of factor VIII and von Willebrand factor type 2N variants to storage organelles. Blood. 2009;113(13):3102-3109. Weiss HJ, Sussman, II. A new von Willebrand variant (type I, New York): increased ristocetin-induced platelet aggregation and plasma von Willebrand factor containing the full range of multimers. Blood. 1986;68(1):149-156. Weiss HJ. Type 2B von Willebrand disease and related disorders of patients with increased ristocetin-induced platelet aggregation: what they tell us about the role of von Willebrand factor in hemostasis. J Thromb Haemost. 2004;2(11):2055-2056. Flood VH, Schlauderaff AC, Haberichter SL, et al. Crucial role for the VWF A1 domain in binding to type IV collagen. Blood. 2015;125(14):2297-2304. Cabrera N, Casana P, Cid AR, Haya S, Moret A, Aznar JA. Novel missense mutation c.2685G>C (p.Q895H) in VWF gene associated with very low levels of VWF mRNA. Ann Hematol. 2009;88(3):245-247. Goodeve AC, Eikenboom J, Castaman G, et al. Phenotype and genotype of a cohort of families historically diagnosed with type 1 von Willebrand disease in the European study, Molecular and Clinical Markers for the Diagnosis and Management of Type 1 von Willebrand Disease (MCMDM-1VWD). Blood. 2007; 109(1):112-121. James PD, Notley C, Hegadorn C, et al. The mutational spectrum of type 1 von

2013

N. Borràs et al.

34.

35.

36.

37.

38.

39.

2014

Willebrand disease: Results from a Canadian cohort study. Blood. 2007; 109(1):145-154. Yadegari H, Biswas A, Akhter MS, et al. Intron retention resulting from a silent mutation in the VWF gene that structurally influences the 5' splice site. Blood. 2016; 128(17):2144-2152. Sauna ZE, Kimchi-Sarfaty C. Understanding the contribution of synonymous mutations to human disease. Nat Rev Genet. 2011;12(10):683-691. Cumming A, Grundy P, Keeney S, et al. An investigation of the von Willebrand factor genotype in UK patients diagnosed to have type 1 von Willebrand disease. Thromb Haemost. 2006;96(5):630-641. Veyradier A, Boisseau P, Fressinaud E, et al. A laboratory phenotype/genotype correlation of 1167 French patients from 670 families with von Willebrand disease: a new epidemiologic picture. Medicine (Baltimore). 2016;95(11):e3038. Borràs N, Corrales I, Ramírez L, Altisent C, Parra R, Vidal F. Diseño, optimitzación y validación de un panel de secuenciación de 23 genes como herramienta de diagnóstico e investigación de las coagulopatías congénitas. XXXI Congreso Nacional de la Sociedad Española de Trombosis y Hemostasia; 2015 October; Valencia, Spain: Thrombosis and Haemostasis; 2015. p. 95. Flood VH, Christopherson PA, Gill JC, et al. Clinical and laboratory variability in a

40.

41.

42.

43.

44.

45.

cohort of patients diagnosed with type 1 VWD in the United States. Blood. 2016; 127(20):2481-2488. Bowen DJ, Collins PW, Lester W, et al. The prevalence of the cysteine1584 variant of von Willebrand factor is increased in type 1 von Willebrand disease: co-segregation with increased susceptibility to ADAMTS13 proteolysis but not clinical phenotype. Br J Haematol. 2005; 128(6):830-836. Casana P, Martinez F, Haya S, Lorenzo JI, Espinos C, Aznar JA. Q1311X: a novel nonsense mutation of putative ancient origin in the von Willebrand factor gene. Br J Haematol. 2000;111(2):552-555. Bellissimo DB, Christopherson PA, Flood VH, et al. VWF mutations and new sequence variations identified in healthy controls are more frequent in the AfricanAmerican population. Blood. 2012; 119(9):2135-2140. Kasatkar P, Ghosh K, Shetty S. A common founder mutation p.P2063S in exon 36 of VWF in 11 unrelated Indian von Willebrand disease (VWD) families. Ann Hematol. 2013;92(8):1147-1148. Allen S, Abuzenadah AM, Hinks J, et al. A novel von Willebrand disease-causing mutation (Arg273Trp) in the von Willebrand factor propeptide that results in defective multimerization and secretion. Blood. 2000;96(2):560-568. Souto JC, Almasy L, Soria JM, et al.

46.

47.

48.

49.

50.

Genome-wide linkage analysis of von Willebrand factor plasma levels: results from the GAIT project. Thromb Haemost. 2003;89(3):468-474. Suzuki T, Tsurusaki Y, Nakashima M, et al. Precise detection of chromosomal translocation or inversion breakpoints by wholegenome sequencing. J Hum Genet. 2014; 59(12):649-654. Johnsen JM, Auer PL, Morrison AC, et al. Common and rare von Willebrand factor (VWF) coding variants, VWF levels, and factor VIII levels in African Americans: the NHLBI Exome Sequencing Project. Blood. 2013;122(4):590-597. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 2015;17(5):405-424. Wang QY, Song J, Gibbs RA, Boerwinkle E, Dong JF, Yu FL. Characterizing polymorphisms and allelic diversity of von Willebrand factor gene in the 1000 Genomes. J Thromb Haemost. 2013; 11(2):261-269. Corrales I, Ramirez L, Altisent C, Parra R, Vidal F. The study of the effect of splicing mutations in von Willebrand factor using RNA isolated from patients' platelets and leukocytes. J Thromb Haemost. 2011; 9(4):679-688.

haematologica | 2017; 102(12)

ARTICLE

Myelodysplastic Syndromes

Pro-inflammatory proteins S100A9 and tumor necrosis factor-α suppress erythropoietin elaboration in myelodysplastic syndromes

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Thomas Cluzeau,1,2,3 Kathy L. McGraw,1 Brittany Irvine,1 Erico Masala,4 Lionel Ades,3,5 Ashley A. Basiorka,6 Jaroslaw Maciejewski,7 Patrick Auberger,2 Sheng Wei,1 Pierre Fenaux,3,5 Valeria Santini4 and Alan List1 1 H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, USA; 2Cote d’Azur University, INSERM U1065, Centre Méditerranéen de Medecine Moléculaire, Nice, France; 3Groupe Français des Myélodysplasies, Paris, France; 4Hematology Unit, AOU Careggi, Firenze, Italy: 5Senior Hematology Unit, Saint Louis Hospital, Paris, France; 6H. Lee Moffitt Cancer Center and Research Institute and the Cancer Biology Ph.D. Program, University of South Florida, Tampa, FL, USA and 7Cleveland Clinic, Taussig Cancer Institute, Cleveland, OH, USA

Haematologica 2017 Volume 102(12):2015-2020

ABSTRACT

ccumulating evidence implicates innate immune activation in the pathobiology of myelodysplastic syndromes. A key myeloidrelated inflammatory protein, S100A9, serves as a Toll-like receptor ligand regulating tumor necrosis factor-α and interleukin-1b production. The role of myelodysplastic syndrome-related inflammatory proteins in endogenous erythropoietin regulation and response to erythroidstimulating agents or lenalidomide has not been investigated. The HepG2 hepatoma cell line was used to investigate in vitro erythropoietin elaboration. Serum samples collected from 311 patients with myelodysplastic syndrome were investigated (125 prior to treatment with erythroid-stimulating agents and 186 prior to lenalidomide therapy). Serum concentrations of S100A9, S100A8, tumor necrosis factor-α, interleukin1b and erythropoietin were analyzed by enzyme-linked immunosorbent assay. Using erythropoietin-producing HepG2 cells, we show that S100A9, tumor necrosis factor-α and interleukin-1b suppress transcription and cellular elaboration of erythropoietin. Pre-incubation with lenalidomide significantly diminished suppression of erythropoietin production by S100A9 or tumor necrosis factor-α. Moreover, in peripheral blood mononuclear cells from patients with myelodysplastic syndromes, lenalidomide significantly reduced steady-state S100A9 generation (P=0.01) and lipopolysaccharide-induced tumor necrosis factor-α elaboration (P=0.002). Enzyme-linked immunosorbent assays of serum from 316 patients with non-del(5q) myelodysplastic syndromes demonstrated a significant inverse correlation between tumor necrosis factor-α and erythropoietin concentrations (P=0.006), and between S100A9 and erythropoietin (P=0.01). Moreover, baseline serum tumor necrosis factor-α concentration was significantly higher in responders to erythroid-stimulating agents (P=0.03), whereas lenalidomide responders had significantly lower tumor necrosis factor-α and higher S100A9 serum concentrations (P=0.03). These findings suggest that S100A9 and its nuclear factorkB transcriptional target, tumor necrosis factor-α, directly suppress erythropoietin elaboration in myelodysplastic syndromes. These cytokines may serve as rational biomarkers of response to lenalidomide and erythroid-stimulating agent treatments. Therapeutic strategies that either neutralize or suppress S100A9 may improve erythropoiesis in patients with myelodysplastic syndromes. haematologica | 2017; 102(12)

Correspondence: cluzeau.t@chu-nice.fr

Received: October 24, 2016. Accepted: September 28, 2017. Pre-published: October 5, 2017. doi:10.3324/haematol.2016.158857 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2015 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2015

T. Cluzeau et al.

Introduction Ineffective erythropoiesis in patients with myelodysplastic syndromes (MDS) derives from both intrinsic abnormalities affecting the response to erythropoietin and extrinsic pressures on the inflammatory bone marrow microenvironment.1 A number of inflammatory cytokines such as tumor necrosis factor-α (TNFα), interleukin (IL)1b, IL-6 and others, are generated in excess in MDS and adversely influence hematopoietic stem and progenitor cell survival.2 Moreover, in a subset of MDS patients, endogenous erythropoietin production is deficient, further compromising erythropoietic potential.3 Accumulating evidence implicates innate immune activation in the physiopathology of MDS and the accompanying inflammatory microenvironment.1,2,4,5 Bone marrow plasma concentrations of the pro-inflammatory, danger-associated molecular pattern (DAMP) protein, S100A9, are profoundly increased in lower-risk MDS, which serves as a catalyst directing myeloid-derived suppressor cell expansion.6 S100A9 is a ligand for CD33 and the Toll-like receptor (TLR)-4 which, through nuclear factor-kB (NF-kB) activation, regulates the transcription and cellular elaboration of inflammatory cytokines such as TNFα and IL-1b.7 The latter cytokines have been shown to suppress erythropoietin elaboration and have been implicated in the suppression of endogenous erythropoietin production8 in patients with anemia of chronic inflammation.9 The involvement of 100A8/S100A9 in del(5q) MDS has already been described.10 Erythropoietin-stimulating agents (ESA) and lenalidomide are efficient treatments used in lower-risk myelodysplastic syndromes.11-13 To date, the role of inflammatory parameters in the regulation of endogenous erythropoietin production and response to erythropoietic treatments in MDS has not been investigated. Here we show the importance of these inflammatory cytokines as key biological determinants of endogenous erythropoietin production and response to ESA and lenalidomide treatments in patients with non-del (5q) MDS.

Methods Reagents and antibodies Recombinant S100A9 was generated as previously described.6 TNFα, IL-1b and lipopolysaccharide were purchased from SigmaAldrich (Saint Louis, MO, USA). Lenalidomide was purchased from Fisher Scientific (Pittsburgh, PA, USA). A CD33 chimera was constructed as described elsewhere.6,14 NF-kB and Rho GDI antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Lamin A/C was purchased from Cell signaling (Saint Quentin, France). BMS344541 was purchased from Tocris Bioscience (Bristol, UK).

Cell culture HepG2 cells, acquired from the American Type Culture Collection (ATCC, Manassas, VA, USA), were grown in Eagle’s minimum essential medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin solution. Cells were maintained at 37°C under 5% CO2.

(MBL, Nagano, Japan). Quantitative measurements of human TNFα and IL-1b were made using the Human TNFα ELISA Kit and Human IL-1b ELISA Kit, respectively (Life Technologies, Carlsbad, CA, USA). Human erythropoietin in patients’ serum and HepG2 cell line supernatants was quantified using a Human Erythropoietin ELISA Kit (Stemcell Technologies, Vancouver, BC, Canada). All measurements were performed in duplicate.

Real-time quantitative polymerase chain reaction RNA was isolated using the RNAeasy Mini Kit (Qiagen, Valencia, CA, USA) followed by iScript cDNA synthesis (Bio-Rad, Hercules, CA, USA) and amplification using iQ SYBR Green Supermix (Bio-Rad, Herculed, CA, USA). The relative level of gene expression for each experimental sample was calculated using the ΔΔCt method. Untreated cells were the experimental control and the housekeeping gene GAPDH was the endogenous control.

Western blot analysis After treatment for 24 h, cells were harvested and lysed in 1X RIPA buffer supplemented with protease and phosphatase inhibitors for classical western blotting. For the nuclear extraction, cells were lysed in ice with buffer A, then pelleted. After removal of supernatant (cytoplasmic fraction), pellets were lysed in ice with buffer B and pelleted (nuclear fraction) (Nuclear Extraction Kit, Abcam, Cambridge, USA). Lysates were pelleted and 50 µg of protein were resolved by sodium dodecylsulfate polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes. The membranes were blocked for 30 min in 5% nonfat dry milk solution in PBST (phosphate-buffered saline with 0.1% Tween 20) and incubated with the indicated antibodies. Membranes were developed using ECL according to the manufacturer’s protocol (GE Healthcare, Little Chalfont, UK). Densitometry analysis was performed using Image J Software.

Patients and serum samples Serum samples for ELISA analysis were collected from four centers (Taussig Cancer Institute, Cleveland, USA; AOU Careggi, University of Florence, Italy; Saint Louis Hospital, Paris, France; H. Lee Moffitt Cancer Center, Tampa, FL, USA). Peripheral blood mononuclear cells were collected from patients at Moffitt Cancer Center. All patients had provided consent to Institutional Review Board, or equivalent, approved protocols in hematology clinics at each center, and the Eastern Cooperative Oncology Group (ECOG) E2905 trial (www.clinicaltrial.gov NCT00843882). All routine clinical and biological data were available.

Statistical analysis Data are expressed as mean ± standard error for continuous variables, or percentage of total for non-continuous variables. Spearman correlation, Mann-Whitney and Jonckheere-Terpstra tests were used for analysis of continuous variables. The chisquare test was used for analysis of non-continuous variables. Differences between the results of comparative tests were considered statistically significant if the two-sided P-value was less than 0.05. All statistical analyses were performed using SPSS v.22 software (IBM SPSS Statistics).

Results

Enzyme-linked immunosorbent assays

Inflammatory proteins suppress erythropoietin production by HepG2 cells

Human S100A9/MRP14 in patients’ serum and supernatants of the HepG2 cell line was quantified using a CircuLex S100A9/MRP14 enzyme-linked immunosorbent assay (ELISA) Kit

Hepatoma HepG2 cells, which produce erythropoietin under basal conditions,15 were treated with varied concentrations of each of the inflammatory proteins, TNFα, IL-1b

2016

haematologica | 2017; 102(12)

S100A9 and TNFα suppress Epo production in MDS

or S100A9. After 24 h exposure, we observed a concentration-dependent reduction in erythropoietin elaboration (Figure 1). At a concentration of 10 ng/mL, TNFα yielded a 40% reduction in erythropoietin elaboration (Figure 1A), compared to a 60% reduction following incubation with IL-1b (Figure 1B). Concentrations of 10 to 20 mg/mL of S100A9 completely suppressed erythropoietin elaboration, while 1 mg/mL yielded a 95% reduction (Figure 1C). For subsequent experiments, concentrations of 10 ng/mL of TNFα, 10 ng/mL of IL-1b and 1 mg/mL of S100A9 were employed.

kB activation in response to inflammatory cytokine stimulation in lymphocytes and other cell lineages.18,19 To determine whether lenalidomide can modulate suppression of erythropoietin production by inflammatory proteins, we treated HepG2 cells with 1 mM lenalidomide or vehicle control for 30 min prior to exposure to TNFα, IL1b or S100A9. Lenalidomide significantly, but incomplete-

Lenalidomide mitigates suppression of erythropoietin production by S100A9 and tumor necrosis factor-α The transcription factor NF-kB is activated by TLR ligands and inflammatory cytokines such as TNFα which, like GATA2, is implicated in transcriptional suppression of the erythropoietin transcript.16 HepG2 cells are known to express the S100A9 receptor, TLR4, on the plasma membrane.17 Lenalidomide has been reported to suppress NF-

Figure 1. Effects of tumor necrosis factor-α, interleukin-1b and S100A9 on erythropoietin elaboration in the HepG2 cell line. HepG2 cells were stimulated for 24 h with the indicated concentrations of (A) TNFα, (B) IL-1b and (C) S100A9 and erythropoietin (EPO) elaboration was determined by ELISA.

haematologica | 2017; 102(12)

Figure 2. Effect of lenalidomide on erythropoietin elaboration in the HepG2 cell line. (A) HepG2 cells were treated with 1 mM lenalidomide for 30 min prior to addition of S100A9 (1 mg/mL), TNFα (10 ng/mL) or IL-1b (10 ng/mL). Supernatant erythropoietin (EPO) concentration was determined by ELISA; (B) HepG2 cells were treated with 0.1, 1 or 10 mM lenalidomide. Supernatant TNFα concentration was determined by ELISA; (C) HepG2 cells were treated with 1 mM lenalidomide 30 min prior to addition of S100A9 (1 mg/mL) or TNFα (10 ng/mL). Quantitative polymerase chain reaction for IL10 mRNA expression was performed 24 h after treatment. GAPDH was used as an endogenous control and results are expressed as IL10 mRNA relative expression. (D) HepG2 cells were treated with 1 mM lenalidomide 30 min before addition of S100A9 (1 mg/mL) or TNFα (10 ng/mL). NF-kB protein was visualized by western blot 24 h after treatment. Rho GDI and lamin A/C were used as the loading controls for cytoplasmic and nuclear fractions, respectively. *mean P≤0.05.

2017

T. Cluzeau et al.

ly, reversed suppression of erythropoietin production by S100A9 (91% suppression by S100A9 versus 65% after lenalidomide pre-incubation, P=0.04). Following treatment with TNFα, lenalidomide pre-treatment abrogated cytokine suppression of erythropoietin elaboration (TNFα, 67% suppression versus lenalidomide pre-incubation, 18%; P=0.05). Lenalidomide had no effect on IL-1bdirected suppression of erythropoietin elaboration (Figure 2A). We also found that BMS-344541, a specific inhibitor of NF-kB, abrogated cytokine suppression of erythropoietin elaboration with the same conditions (Figure 2A). Moreover, lenalidomide decreased basal TNFα release after lenalidomide stimulation (Figure 2B). IL-10 is known to be an anti-inflammatory cytokine secreted by immune cells.20 We performed quantitative polymerase chain reaction analysis to determine the effects of the inflammatory proteins and lenalidomide on IL10 gene transcription. Pre-incubation of each inflammatory cytokine with lenalidomide significantly increased IL10 mRNA expression compared to S100A9 or TNFα treatment alone (Figure 2C). Lenalidomide had no modulatory effect on IL10 following IL-1b treatment (data not shown). Finally, NF-kB is a key transcription factor involved in S100A9 and TNFα receptor signaling and the transcriptional suppression of erythropoietin mRNA in erythropoietin-producing cells which is modulated by lenalidomide.21-23 We showed that NF-kB targets, including TNFα and IL-10, were regulated by lenalidomide. We therefore performed cytoplasmic and nuclear NF-kB western blotting in HepG2 cells to discern the effect of lenalidomide on inflammatory protein signaling. As demonstrated by nuclear localization after lenalidomide pre-incubation,

Figure 3. Effect of lenalidomide on S100A9 and tumor necrosis factor-α production in peripheral blood mononuclear cells from patients with non-del(5q) myelodysplastic syndrome. (A) Frozen peripheral blood mononuclear cells (PBMC) from lower-risk MDS patients (n=7) were treated with 1 mM lenalidomide for 24 h, before analysis of supernatant S100A9 concentration by ELISA. Results are expressed relative to untreated cells. (B) PBMC from lower-risk MDS patients (n=7) were treated with 1 mM lenalidomide 30 min prior to addition of lipopolysaccharide (LPS) (1 mg/mL): 24 h after stimulation, TNFα ELISA was performed on the supernatants. Results are expressed as a percentage relative to LPS treatment alone. *mean P≤0.05.

2018

active NF-kB was significantly reduced following treatment with S100A9 or TNFα, indicating that lenalidomide suppressed NF-kB activation (Figure 2D). To confirm the results observed in HepG2 cells, we investigated the effects of lenalidomide on steady-state production of S100A9 by peripheral blood mononuclear cells isolated from patients with non-del(5q) MDS (n=7). ELISA showed a significant reduction in S100A9 elaboration after 24 h exposure to lenalidomide (P=0.01) (Figure 3A). Similarly, pre-incubation of MDS peripheral blood mononuclear cells with lenalidomide significantly reduced TNFα production induced by lipopolysaccharide (P=0.002). These findings indicate that lenalidomide-modulated S100A9 and TNFα suppression of erythropoietin elaboration is NF-kB-dependent.

Relationship between inflammatory proteins and endogenous erythropoietin concentration in patients with myelodysplastic syndromes To validate the regulatory role of inflammatory proteins on erythropoietin elaboration in vivo, we assessed the relationships between serum concentrations of various inflammatory cytokines and erythropoietin in MDS patients with symptomatic anemia. Serum samples from 316 patients with non-del(5q) MDS were analyzed. The median age of the patients was 74.7 years (range, 41-94). Distribution of International Prognostic Scoring System (IPSS) categories was low, intermediate-1, intermediate-2 and high risk in 38%, 50%, 10% and 2% of patients, respectively; whereas 24%, 38%, 22%, 13% and 3% of patients were very low, low, intermediate, high and very high risk according to the revised IPSS (IPSS-R) (Table 1). Serum concentrations of erythropoietin, S100A9, S100A8, TNFα and IL-1b were assessed by ELISA. The serum S100A9 concentration was significantly higher in patients with lower-risk MDS than in those with higherrisk MDS [12,226 pg/mL (range, 0-228,880) versus 240 pg/mL (range, 0-43,858), respectively; P=0.001). No significant differences were observed in TNFα and IL-1b concentrations according to IPSS or IPSS-R category (data not shown). There was a statistically significant negative correlation between TNFα and erythropoietin concentrations (r= – Table 1. Patients’ demographics and disease characteristics.

Median age (range) Sex ratio (F/M) IPSS (%) Low Intermediate 1 Intermediate 2 High IPSS-R (%) Very low Low Intermediate High Very high

Global cohort (n=316)

Prior to ESA treatment (n=159)

Prior to lenalidomide ± ESA treatment (n=159)

74.7 (41-94) 212/104

74.8 (41-94) 105/54

74.0 (48-85) 110/49

38 50 10 2 24 38 22 13 3

39 47 12 2 23 39 21 14 3

30 70 0 0 33 17 50 0 0

haematologica | 2017; 102(12)

S100A9 and TNFα suppress Epo production in MDS

0.164, P=0.006), and between S100A9 and erythropoietin concentrations (r= –0.148, P=0.01), whereas there was no discernible relationship between IL-1b and erythropoietin concentrations (Table 2A and Online Supplementary Figure S1). As expected, we also found significant positive correlations between the concentrations of the inflammatory protein S100A9 and TNFα (r=0.294, P<0001), S100A9 and IL-1b (r=0.180, P=0.002), as well as IL-1b and TNFα (r=0.262, P<0.001) (Table 2B). These findings support the notion that S100A9 and TNFα suppress renal erythropoietin elaboration and the endocrine response to anemia in non-del(5q) MDS.

Relationships between inflammatory protein concentrations and response to treatment with erythropoietic agents Within the cohort of patients with non-del(5q) MDS, 159 were studied prior to ESA treatment and 159 prior to lenalidomide ± erythropoietin treatment (Table 1). ESA responders had significantly higher serum TNFα concentrations than non-responders (8.37 pg/mL versus 3.79 pg/mL, respectively; P=0.03) with a corresponding significantly lower erythropoietin concentration in ESA responders (36 mU/mL versus 113 mU/mL, respectively; P<0.0001). Erythroid response rate was 43% versus 55% in patients with low versus high TNFα concentration, respectively (Figure 4A). There was no significant relationship between S100A9 serum concentration and erythropoietin response (data not shown). Nevertheless, erythropoietin concentration was a better biomarker of response to ESA than was TNFα concentration even after adjustment for erythropoietin concentration. Finally, among patients treated with lenalidomide or lenalidomide ± erythropoietin, responding patients had significantly lower serum TNFα concentrations (P=0.02) while there was no relationship with S100A9 concentration (P=0.21). Considering responses to lenalidomide, we observed a significant difference in erythroid response rate (62% versus 12% for patients with low versus high S100A9 serum concentration, respectively; P=0.03) (Figure 4B).

mation and malignancy.24 Indeed, TNFα antagonists have been shown to improve anemia in several inflammatory disorders.25 The pathogenesis of ineffective erythropoiesis in MDS, in particular, is multifactorial, including abnormalities inherent to the neoplastic clone as well as the inflammatory bone marrow microenvironment.26 Inflammatory cytokines such as IL-1b, interferon-γ, transforming growth factor-b and TNFα directly inhibit erythroid progenitor colony-forming capacity in vitro and impair iron turnover.2729 Moreover, both IL-1b and TNFα suppress erythropoietin gene expression and protein secretion in a NF-kB-dependent fashion,30 factors which have been implicated in the disproportionately low endogenous erythropoietin production in response to the anemia of inflammation. In a subset of lower-risk MDS patients who are responsive to treatment with recombinant erythropoietin, renal erythropoietin production is suppressed with a corresponding reduction in serum erythropoietin concentration. The precise physiological events that impair erythropoietin production in lower-risk MDS do, however, remain unexplored. Our investigations show that S100A9, a myeloidderived inflammatory protein produced in excess in MDS, directly suppresses erythropoietin transcription and elaboration in HepG2 hepatoma cells, analogous to the actions of TNFα.6 Moreover, S100A9 serves as a key coordinator of the inflammatory response by inducing the secretion of TNFα, IL-6, IL-8, and IL-1b via TLR4-dependent activation

Discussion Pro-inflammatory cytokines have long been implicated as key effectors of anemia in disorders of chronic inflam-

Table 2. (A) Correlations between concentrations of inflammatory proteins and erythropoietin (EPO) in patients’ serum. (B) Relationships between inflammatory proteins. A.

Inflammatory parameters

EPO (r ; P-value)

S100A9 TNFα IL-1b

(-0.148 ; 0.01) (-0.164 ; 0.006) (-0.003 ; 0.96)

S100A9 S100A9 TNFα IL-1b

TNFα (0.294 ; <0.0001)

(0.294 ; <0.0001) (0.180 ; 0.002)

haematologica | 2017; 102(12)

(0.262 ; <0.0001)

Figure 4. Relationship between serum concentrations of inflammatory proteins and response to erythropoiesis-stimulating agents or lenalidomide treatment. (A) Correlation between serum concentrations of inflammatory proteins and response to ESA or (B) lenalidomide treatment.

2019

T. Cluzeau et al.

of NF-kB.30,31 Our findings support the notion that these inflammatory cytokines similarly suppress erythropoietin production in vivo in lower-risk MDS patients. Serum erythropoietin concentration was inversely related to S100A9 and TNFα concentrations. Furthermore, serum TNFα concentration was significantly higher in patients responding to treatment with recombinant erythropoietin than in nonresponders (P=0.03). Previous investigations showed that higher serum TNFα concentration predicted resistance to ESA: our data do, therefore, need to be confirmed in a larger study.32,33 Of particular interest, lenalidomide suppressed nuclear translocation of NF-kB to mitigate the suppression of erythropoietin production in HepG2 cells by both S100A9 and TNFα. The ability of lenalidomide to modulate cytokine activation may not only reduce progenitor cell injury, but may also relieve repression of renal erythropoietin elaboration and, therefore, the endocrine response to anemia in MDS. We found that the serum concentration

References 1. Ganan-Gomez I, Wei Y, Starczynowski DT, et al. Deregulation of innate immune and inflammatory signaling in myelodysplastic syndromes. Leukemia. 2015;29(7): 1458-1469. 2. Yang L, Qian Y, Eksioglu E, Epling-Burnette PK, Wei S. The inflammatory microenvironment in MDS. Cell Mol Life Sci. 2015;72(10):1959-1966. 3. Valent P. Low erythropoietin production as non-oncogenic co-factor contributing to disease-manifestation in low-risk MDS: a hypothesis supported by unique case reports. Leuk Res. 2008;32(9):1333-1337. 4. Varney ME, Melgar K, Niederkorn M, Smith MA, Barreyro L, Starczynowski DT. Deconstructing innate immune signaling in myelodysplastic syndromes. Exp Hematol. 2015;43(8):587-598. 5. Boiko JR, Borghesi L. Hematopoiesis sculpted by pathogens: Toll-like receptors and inflammatory mediators directly activate stem cells. Cytokine. 2012;57(1):1-8. 6. Chen X, Eksioglu EA, Zhou J, et al. Induction of myelodysplasia by myeloidderived suppressor cells. J Clin Invest. 2013;123(11):4595-4611. 7. Riva M, Kallberg E, Bjork P, et al. Induction of nuclear factor-kappaB responses by the S100A9 protein is Toll-like receptor-4-dependent. Immunology. 2012;137(2):172-182. 8. Faquin WC, Schneider TJ, Goldberg MA. Effect of inflammatory cytokines on hypoxia-induced erythropoietin production. Blood. 1992;79(8):1987-1994. 9. Bertero MT, Caligaris-Cappio F. Anemia of chronic disorders in systemic autoimmune diseases. Haematologica. 1997;82(3):375381. 10. Schneider RK, Schenone M, Ferreira MV, et al. Rps14 haploinsufficiency causes a block in erythroid differentiation mediated by S100A8 and S100A9. Nat Med. 2016;22(3):288-297. 11. Toma A, Kosmider O, Chevret S, et al. Lenalidomide with or without erythropoietin in transfusion-dependent erythropoiesis-stimulating agent-refractory lowerrisk MDS without 5q deletion. Leukemia. 2016;30(4):897-905. 12. Park S, Fenaux P, Greenberg P, et al. Efficacy and safety of darbepoetin alpha in patients

2020

13.

14.

15.

16.

17.

18. 19.

20.

21.

22.

23.

of TNFα was significantly lower and serum concentration of S100A9 was higher in lenalidomide-responsive patients (P=0.03). Together, these findings indicate that S100A9 and its transcriptional target, TNFα, directly suppress erythropoietin elaboration and endocrine response to anemia in MDS and may be useful biomarkers of response to treatment with lenalidomide or recombinant erythropoietin, meriting further investigation. More importantly, our findings suggest that therapeutic strategies that either neutralize or suppress S100A9 may improve erythropoiesis in lower-risk MDS by suppressing inflammatory cytokine generation and restoring endocrine erythropoietin response to anemia. Acknowledgments Funding for this study was provided by the Foundation Nuovo Soldati, Phillippe Foundation and “Les amis de la faculte de medecine de Nice” (to TC).

with myelodysplastic syndromes: a systematic review and meta-analysis. Br J Haematol. 2016;174(5):730-747. Santini V, Almeida A, Giagounidis A, et al. Randomized phase III Study of lenalidomide versus placebo in RBC transfusiondependent patients with lower-risk nondel(5q) myelodysplastic syndromes and ineligible for or refractory to erythropoiesis-stimulating agents. J Clin Oncol. 2016;34(25):2988-2996. Cannon JP, O'Driscoll M, Litman GW. Construction, expression, and purification of chimeric protein reagents based on immunoglobulin Fc regions. Methods Mol Biol. 2011;748:51-67. Porwol T, Ehleben W, Zierold K, Fandrey J, Acker H. The influence of nickel and cobalt on putative members of the oxygen-sensing pathway of erythropoietin-producing HepG2 cells. Eur J Biochem. 1998;256(1): 16-23. La Ferla K, Reimann C, Jelkmann W, Hellwig-Burgel T. Inhibition of erythropoietin gene expression signaling involves the transcription factors GATA-2 and NFkappaB. FASEB J. 2002;16(13):1811-1813. Hsiao CC, Chen PH, Cheng CI, et al. Tolllike receptor-4 is a target for suppression of proliferation and chemoresistance in HepG2 hepatoblastoma cells. Cancer Lett. 2015;368(1):144-152. Crane E, List A. Immunomodulatory drugs. Cancer Invest. 2005;23(7):625-634. Galili N, Raza A. Immunomodulatory drugs in myelodysplastic syndromes. Expert Opin Investig Drugs. 2006;15(7):805-813. Hori S, Nomura T, Sakaguchi S. Control of regulatory T cell development by the transcription factor Foxp3. Science. 2003;299(5609):1057-1061. Zuckerman SH, Evans GF, Guthrie L. Transcriptional and post-transcriptional mechanisms involved in the differential expression of LPS-induced IL-1 and TNF mRNA. Immunology. 1991;73(4):460-465. Szenajch J, Wcislo G, Jeong JY, Szczylik C, Feldman L. The role of erythropoietin and its receptor in growth, survival and therapeutic response of human tumor cells. From clinic to bench - a critical review. Biochim Biophys Acta. 2010;1806(1):82-95. Xu K, Geczy CL. IFN-gamma and TNF regulate macrophage expression of the

24.

25.

26. 27.

28.

29.

30.

31.

32.

33.

chemotactic S100 protein S100A8. J Immunol. 2000;164(9):4916-4923. Morceau F, Dicato M, Diederich M. Proinflammatory cytokine-mediated anemia: regarding molecular mechanisms of erythropoiesis. Mediators Inflamm. 2009;2009: 405016. Corrado A, Di Bello V, d'Onofrio F, Maruotti N, Cantatore FP. Anti-TNF-alpha effects on anemia in rheumatoid and psoriatic arthritis. Int J Immunopathol Pharmacol. 2017;30(3):302-307. Calado RT. Immunologic aspects of hypoplastic myelodysplastic syndrome. Semin Oncol. 2011;38(5):667-672. Wang CQ, Udupa KB, Lipschitz DA. Interferon-gamma exerts its negative regulatory effect primarily on the earliest stages of murine erythroid progenitor cell development. J Cell Physiol. 1995;162(1):134138. Felli N, Pedini F, Zeuner A, et al. Multiple members of the TNF superfamily contribute to IFN-gamma-mediated inhibition of erythropoiesis. J Immunol. 2005;175(3): 1464-1472. Taniguchi S, Dai CH, Price JO, Krantz SB. Interferon gamma downregulates stem cell factor and erythropoietin receptors but not insulin-like growth factor-I receptors in human erythroid colony-forming cells. Blood. 1997;90(6):2244-2252. Simard JC, Cesaro A, Chapeton-Montes J, et al. S100A8 and S100A9 induce cytokine expression and regulate the NLRP3 inflammasome via ROS-dependent activation of NF-kappaB(1.). PLoS One. 2013;8(8): e72138. Chernov AV, Dolkas J, Hoang K, et al. The calcium-binding proteins S100A8 and S100A9 initiate the early inflammatory program in injured peripheral nerves. J Biol Chem. 2015;290(18):11771-11784. Musto P, Matera R, Minervini MM, et al. Low serum levels of tumor necrosis factor and interleukin-1 beta in myelodysplastic syndromes responsive to recombinant erythropoietin. Haematologica. 1994;79(3): 265-268. Stasi R, Brunetti M, Bussa S, et al. Serum levels of tumour necrosis factor-alpha predict response to recombinant human erythropoietin in patients with myelodysplastic syndrome. Clin Lab Haematol. 1997;19(3):197-201.

haematologica | 2017; 102(12)

ARTICLE

Myelodysplastic syndromes

Red cell alloimmunization is associated with development of autoantibodies and increased red cell transfusion requirements in myelodysplastic syndrome

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Deepak Singhal,1,2,3 Monika M. Kutyna,2# Rakchha Chhetri,2# Li Yan A. Wee,2# Sophia Hague,4 Lakshmi Nath,5 Shriram V. Nath,5,6 Romi Sinha,7 Nicholas Wickham,8 Ian D. Lewis,1,2,3 David M. Ross,1,2,3,9,10 Peter G. Bardy,1,2,3 Luen Bik To,1,2,3 John Reynolds,11## Erica M. Wood,11## David J. Roxby4,9## and Devendra K. Hiwase1,2,3,10

Cancer Centre, Royal Adelaide Hospital, Adelaide; 2Haematology Department, SA Pathology, Adelaide; 3School of Medicine, University of Adelaide; 4Transfusion Medicine, SA Pathology, Adelaide; 5Haematology, Clinpath Laboratories, Adelaide; 6Adelaide Haematology Centre, Ashford Specialist Centre, Adelaide; 7Blood, Organ and Tissue Programs, Public Health & Clinical Systems, Department of Health, Adelaide; 8Adelaide Cancer Centre, Ashford Specialist Centre, Adelaide; 9Haematology & Genetic Pathology, Flinders University, Bedford Park; 10Cancer Research, Cancer Theme, South Australian Health and Medical Research Institute, Adelaide and 11Department of Epidemiology and Preventive Medicine, Faculty of Medicine, Nursing and Health Sciences, Monash University, Melbourne, Australia 1

Haematologica 2017 Volume 102(12):2021-2029

MMK, RC and LYAW contributed equally to this manuscript.

JR, EMW and DJR contributed equally to this manuscript.

ABSTRACT

p to 90% of patients with a myelodysplastic syndrome require red blood cell transfusion; nevertheless, comprehensive data on red cell alloimmunization in such patients are limited. This study evaluates the incidence and clinical impact of red cell alloimmunization in a large cohort of patients with myelodysplastic syndrome registered in the statewide South Australian-MDS registry. The median age of the 817 patients studied was 73 years, and 66% were male. The cumulative incidence of alloimmunization was 11%. Disease-modifying therapy was associated with a lower risk of alloimmunization while alloimmunization was significantly higher in patients with a revised International Prognostic Scoring System classification of Very Low, Low or Intermediate risk compared to those with a High or Very High risk (P=0.03). Alloantibodies were most commonly directed against antigens in the Rh (54%) and Kell (24%) systems. Multiple alloantibodies were present in 49% of alloimmunized patients. Although 73% of alloimmunized patients developed alloantibodies during the period in which they received their first 20 red cell units, the total number of units transfused was significantly higher in alloimmunized patients than in non-alloimmunized patients (90±100 versus 30±52; P<0.0001). In individual patients, red cell transfusion intensity increased significantly following alloimmunization (2.8±1.3 versus 4.1±2.0; P<0.0001). A significantly higher proportion of alloimmunized patients than non-alloimmunized patients had detectable autoantibodies (65% versus 18%; P<0.0001) and the majority of autoantibodies were detected within a short period of alloimmunization. In conclusion, this study characterizes alloimmunization in a large cohort of patients with myelodysplastic syndrome and demonstrates a signficant increase in red cell transfusion requirements following alloimmunization, most probably due to development of additional alloantibodies and autoantibodies, resulting in subclinical/clinical hemolysis. Strategies to mitigate alloimmunization risk are critical for optimizing red cell transfusion support.

haematologica | 2017; 102(12)

Correspondence: devendra.hiwase@sa.gov.au

Received: July 3, 2017. Accepted: September 27, 2017. Pre-published: October 5, 2017. doi:10.3324/haematol.2017.175752 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2021 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2021

D. Singhal et al.

Introduction Myelodysplastic syndromes (MDS) are clonal diseases characterized by peripheral cytopenias, ineffective hematopoiesis and an increased risk of leukemic transformation.1 They are among the most commonly diagnosed myeloid malignancies2 with the median age of affected individuals at diagnosis being 72 years.3 Management options include disease-modifying therapies and supportive measures such as red blood cell (RBC) and platelet transfusions, antimicrobials and growth factors.4,5 Transfusion support remains a cornerstone of management for most MDS patients. Up to 90% of patients require RBC transfusions during the course of their disease6 and 30-45% become dependent on RBC transfusions.7,8 Importantly, RBC transfusion dependency is associated with a significantly worse survival independently of the revised International Prognostic Scoring System (IPSS-R) risk score. MDS patients are at high risk of developing transfusion-associated complications such as iron overload and acute or delayed hemolytic transfusion reactions, resulting in significant morbidity and mortality.9 True morbidity and mortality burden from RBC alloimmunization is likely higher than reported to hemovigilance programs.10,11 Alloimmunization may also drive RBC autoantibody formation and subclinical/serological hemolytic transfusion reactions.12 Although RBC autoantibody formation after alloimmunization can occur in any transfused patient, reported rates are much higher in transfused patients with thalassemia or sickle cell disease with a cumulative incidence of 6-10%.13-16 In a study of 717 patients with autoantibodies, 200 (28%) patients had both autoantibodies and alloantibodies, and the majority were detected simultaneously and were induced by transfusion.17 We have observed clinically that some MDS patients have increased RBC transfusion requirement following development of alloimmunization. For the transfusion service, identifying and characterizing allo- and autoantibodies can be time-consuming, laborious and expensive and can cause difficulties and delay in finding compatible units. Despite the high prevalence of MDS, data on alloimmunization in chronically transfused MDS patients mostly reflect experience from single centers with limited numbers of patients and often relatively short follow-up.18-20 Highly variable alloimmunization rates have been reported.18-20 Larger studies are required to better understand alloimmunization in MDS, including the complex interplay between disease- and patient-related factors, and transfusion burden. This study characterizes RBC alloimmunization in a large series of well-annotated patients with MDS followed in a state-wide MDS registry.

Methods The South Australian MDS Registry has been described previously.7 Briefly, it is a comprehensive, state-wide database of adult patients with MDS, MDS/myeloproliferative neoplasm overlap syndrome, acute myeloid leukemia (<30% blasts) and therapy-related myeloid neoplasm from six participating hospitals across the public and private sectors (Online Supplementary Methods). Institutional ethics committee approval was obtained 2022

from all participating institutions and procedures were performed in accordance with the revised Helsinki Declaration. Demographic, clinical, laboratory and treatment (including transfusions) details of patients diagnosed between 1990 and 2015 enrolled in the registry and with at least 6 months of follow-up were analyzed. Diseasemodifying therapies included azacitidine, lenalidomide, intensive chemotherapy and allogeneic hematopoietic stem cell transplantation. RBC transfusion dependency was defined as the requirement of at least one RBC unit every 8 weeks over a 4-month period.7,8 As some patients received RBC transfusions before their MDS diagnosis had been established, we assessed the serial blood counts, clinical profile and RBC transfusion requirements of all patients before and after MDS diagnosis. RBC units transfused before the diagnosis of MDS were considered MDS-related if a patient was transfused to alleviate persistent or progressive anemia due to MDS. RBC units transfused before the MDS diagnosis for other causes, such as gastrointestinal bleeding, surgery or trauma, were considered unrelated. To minimize the influence of clinical variables such as infection, bleeding, disease-modifying therapies and invasive procedures, we evaluated transfusion intensity (number of RBC units transfused per month) in patients requiring regular RBC transfusion before and after alloimmunization during the entire study period, and over a fixed period of 8 months (4 months before and 4 months after first documentation of alloimmunization). Patients who received disease-modifying therapies, died or progressed to acute myeloid leukemia within 4 months of alloimmunization were excluded, as these variables would influence RBC transfusion intensity. Patients who developed alloantibodies before MDS-related transfusion, developed alloantibodies after only one or two episodes of RBC transfusion, received only intermittent RBC transfusions, or did not receive further RBC transfusion after alloantibodies had been detected were also excluded from this analysis as RBC transfusion intensity could not be calculated accurately in these patients. Laboratory data included patients’ ABO/Rh type, antibody screening results, direct antiglobulin test and alloantibody and autoantibody specificities (where specificity was documented). Data concerning transfusion reactions were obtained from the hospitals’ transfusion records (Online Supplementary Methods). A delayed serological transfusion reaction was considered to have occurred if all the following criteria were satisfied: (i) a new antibody was detected; (ii) there was a new positive direct antiglobulin test, (iii) RBC elution identified the presence of the same antibody that was identified in the serum; and (iv) phenotyping of the patient’s RBC demonstrated mixed-field typing or negativity for the antigen towards which the alloantibody in the patient’s serum/eluate was directed. A delayed hemolytic transfusion reaction was defined as having occurred when a patient with a delayed serological transfusion reaction had clinical evidence of hemolysis.21,22 The cumulative incidence of alloimmunization was analyzed by competing-risks regression using the Fine and Gray method. Factors associated with RBC alloantibody formation were investigated using random survival forest, recursive partitioning and competing risk regression analyses (Online Supplementary Methods). haematologica | 2017; 102(12)

RBC alloimmunization in MDS patients

Results

Table 1. Clinical features of patients included in the analysis.

Variable

Patient and clinical characteristics The clinical and laboratory records of 836 MDS patients were reviewed. Nineteen patients were ineligible for further analysis because of inadequate follow up and/or incomplete transfusion history. The median age of the 817 patients eligible for analysis was 73 years (range, 19-98 years); 536 (66%) patients were male. MDS-multilineage dysplasia, MDS with excess blasts-1 and MDS with excess blasts-2 were the most frequent subtypes (Table 1). The majority of patients received supportive care alone (605; 74%) while 204 (25%) patients received diseasemodifying therapies (Table 1). According to IPSS-R score, 457 (56%) patients were classified in the Very Low, Low and Intermediate risk groups, while 179 (22%) patients were classified in the High and Very High risk groups, with significant differences in survival and cumulative incidence of RBC transfusion dependency between the groups (Table 1; Online Supplementary Figure S1A,B). The 132 (16%) patients with therapy-related myeloid neoplasm or proliferative MDS/ myeloproliferative neoplasm overlap were not eligible for IPSS-R calculations. IPSS-R could not be assessed in 49 (6%) patients because of missing data or failed metaphase cytogenetics.

Incidence of red blood cell alloimmunization During the study 695 (85%) patients received at least one unit of RBC, and 98 (12%) patients developed 175 alloantibodies. Of these, seven patients developed antibodies before their first documented RBC transfusion and 11 patients developed alloantibodies following MDS-unrelated RBC transfusion before the diagnosis of MDS (range, 0.23 to 146 months prior) (Online Supplementary Table S1). The remaining 80 patients (including six patients who were also transfused before MDS diagnosis) developed alloantibodies following MDS-related RBC transfusion (Online Supplementary Table SI). Thus, the cumulative incidence of RBC alloimmunization with death as a competing risk was 11% at 50 months following the first MDS-related RBC transfu-

Number of patients (%)

Total 817 Male/female 536 (66%) / 281 (34%) Age; median (range) 73 (19-98) years MDS subtype (WHO 2016 classification) MDS-SLD 51 (6.2%) MDS-MLD 186 (22.8%) MDS-RS-SLD 36 (4.4%) MDS-RS-MLD 37 (4.5%) MDS-EB-1 89 (10.9%) MDS-EB-2 83 (10.2%) MDS with isolated del(5q) 17 (2.1%) Hypoplastic MDS 13 (1.6%) MDS/MPN-U 18 (2.2%) MDS/MPN RS-T 7 (1.0%) MDS-U 6 (0.7%) CMML-0 42 (5.1%) CMML-1 31 (3.8%) CMML-2 20 (2.4%) AML (<30% blasts) 55 (6.7%) T-MN 126 (15.4%) IPSS-R categories Very low 118 (14.4%) Low 216 (26.4%) Intermediate 123 (15.1%) High 91 (11.1%) Very high 88 (10.8%) Not applicable 132 (16.2%) Missing data 49 (6.0%) Treatment for MDS Disease-modifying therapy 204 (25.0%) Supportive care 605 (74.0%) Data missing 8 (1%) The revised International Prognostic Scoring System (IPSS-R) is not applicable to cases of T-MN, proliferative CMML and MDS/MPN overlap syndrome (n=132). In 49 (6.0%) cases, minimal data for calculating IPSS-R were not available. MDS-SLD: MDS with single lineage dysplasia; MDS-MLD: MDS with multilineage dysplasia; MDS-RSSLD: MDS with ring sideroblasts with single lineage dysplasia; MDS-RS-MLD: MDS with ring sideroblasts with multilineage dysplasia; MDS-EB-1: MDS with excess blasts-1; MDS-EB-2: MDS with excess blasts-2; MDS/MPN-U: MDS/myeloproliferative neoplasms, unclassifiable; MDS/MPN-RS-T: MDS/MPN with ring sideroblasts and thrombocytosis; MDS-U: MDS, unclassifiable; CMML: chronic myelomonocytic leukemia; AML: acute myeloid leukemia; T-MN: therapy-related myeloid neoplasms.

Figure 1. Cumulative incidence of red blood cell alloimmunization and time to development of alloantibodies. (A) Probability of alloimmunization with death as competing risk by months following first RBC transfusion; (B) 73% of alloimmunized patients developed alloantibodies within the period of transfusion of their first 20 units of RBC; (C) 50% of alloimmunized patients developed alloantibodies within 6 months after their first RBC transfusion.

haematologica | 2017; 102(12)

2023

D. Singhal et al.

sion (Figure 1A). Importantly, 73% and 50% of alloimmunized patients developed alloantibodies following transfusions of fewer than 20 RBC units (Figure 1B; Online Supplementary Figure S1C) and during the first 6 months following commencement of RBC transfusions (Figure 1C), respectively. Of the 98 patients who developed RBC alloantibodies, 50 (51%) patients developed one alloantibody, while 48 (49%) developed multiple alloantibodies, including four patients who developed four antibodies each, and one patient who developed six antibodies in total. Fourteen and three patients developed two and three antibodies simultaneously, respectively (Table 2 and Online Supplementary Table S1). Rh (54%) and Kell (24%) system antibodies were the most frequent, followed by Kidd (7.5%) and Lutheran (5.7%) system antibodies (Table 2 and Online Supplementary Table S1). Within the Rh group, anti-E (46%) was the most frequent, followed by anti-C (17%) and anti-D (17%) (Table 2). Despite our state-wide policy of providing RhD-compatible RBC transfusions, 16 RhD-negative patients developed anti-D. These patientsâ&#x20AC;&#x2122; transfusion records were reviewed for the period prior to anti-D alloimmunization. Anti-D was detected following RhD-positive platelet transfusions in eight patients (one to two units of platelets), and three patients received RhD-positive RBC units due to a clinical emergency. In three female patients, the anti-D alloimmunization was most probably pregnancy-related. For the remaining two patients, we could find no record of RhD-positive platelet or RBC transfusion. Administration of anti-D Rh immunoglobulin remains a possibility, but we could not find any record of it in the participating institutions.

Autoantibody formation following alloimmunization During the study period, 327 (40%) patients had a direct antiglobulin test performed and the test was positive at least once in 157 (48%) patients. Alloimmunized patients had much higher rates of positive direct antiglobulin test (84% versus 33%, P<0.0001; Online Supplementary Figure S1D) and reactive eluates (80% versus 18%, P<0.0001, Figure 2A) than non-alloimmunized patients. RBC eluates showed pan-agglutination due to non-specific autoantibody, an alloantibody, a combination of an allo- and autoantibody, and non-reactivity in 52%, 15%, 13%, and 19% of alloimmunized patients, respectively. Thus, a significantly higher proportion of alloimmunized patients had detectable autoantibody compared to non-alloimmunized patients (65% versus 18%; P<0.0001) (Figure 2A). Circulating free autoantibody was also detected in 31/51 (61%) of alloimmunized patients. In the 88% of alloimmunized patients who developed autoantibodies, these were detected either at the time of alloimmunization or within the 5 months preceding or following alloimmunization (Online Supplementary Figure S1E). In two cases in which autoantibody was detected 74 and 18 months prior to alloantibody detection, the patients were diagnosed with warm autoimmune hemolytic anemia on the background of non-Hodgkin lymphoma or chronic lymphocytic leukemia. In one female patient, the autoantibody was detected at the time that chronic lymphocytic leukemia was diagnosed, 107 months after the alloantibody had first been detected. In this case, the alloantibody was most probably related to a previous pregnancy. 2024

Table 2. Alloantibody specificities.

Alloantibody specificity(ies) Single alloantibody: E K Kp(a) Lu(a) C c D Jk(a) M C(w) Le(a) Low incidence antigen Two alloantibodies: C+D E+c E + Jk(a) E + Yt(b) Jk(a) + E Jk(a) + Lu(a) c + Kp(a) D+C C+E D+E K + Kp(a) E+K C + Lu(a) Co(b) + D K+C C + Kp(a) E + Kp(a) D + Jk(a) K+S K + HLA Le(a) + HLA Fy(a) + K Three or more alloantibodies: D + Jk(a) + E E + Fy(b) + Jk(a) Jk(a) + Co(b) + Cs(a) E+K+c D + K + Kp(a) E + C(w) + S C(w) + E + K C(w) + K + Lu(a) D+E+C C + D + Fy(a) C(w) + E + c E+M+S E + Lu(a) + c K+C+E E + c + Jk(a) E + K + Kp(a) D + E + Jk(b) + C E + Kp(a) + K + c K + Kp(a) + C(w) + Lu(a) E+K+D+C E + c + Kp(a) + Jk(a) + S + C(w)

N. of patients affected (%) (total n=98) 50 (51%) 16 (16%) 10 (10%) 5 (5.1%) 5 (5.1%) 3 (3.1%) 1 (1%) 3 (3.1%) 3 (3.1%) 2 (2%) 1 (1%) 1 (1%) 1 (1%) 26 (27%) 3 (3.1%) 3 (3.1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1(1%) 1(1%) 1 (1%) 22 (23%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 2 (2%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) 1 (1%) haematologica | 2017; 102(12)

RBC alloimmunization in MDS patients

Clinical consequences of antibody formation Two patients developed significant hemolysis after alloimmunization. A 51-year old male patient developed an alloantibody resulting in life-threatening delayed hemolytic transfusion reactions, cardiac arrest, multiorgan failure and a prolonged stay in the Intensive Care Unit (Online Supplementary Figure S2A-C). Another 83-year old male patient developed hemolysis after 6 months of RBC transfusion. Investigations revealed alloantibody, panagglutinating autoantibody and autoimmune hemolysis, which responded to steroids (Online Supplementary Figure S2D-F). Of the 79 alloimmunized cases with a positive direct antiglobulin test, 22 (28%) satisfied the definition of delayed serological transfusion reaction but were not reported to the transfusion services as having had such a reaction or a delayed hemolytic transfusion reaction. Other transfusion reactions were generally infrequent, mild in nature and limited to febrile non-hemolytic transfusion reactions or allergic reactions. Fifteen alloimmunized patients had 18 episodes of transfusion reactions including 13 febrile non-hemolytic transfusion reactions and five allergic reactions.

Alloimmunization increases red blood cell transfusion intensity The total number of RBC units transfused in alloimmunized patients was significantly higher than in non-alloimmunized patients (90±100 versus 30±52; P<0.0001) (Figure 2B). In alloimmunized patients the total number of RBC transfused after alloimmunization was significantly higher than the number before alloimmunization (P<0.0001)

(Figure 2C). RBC transfusion intensity was compared in 33 individual patients eligible for analysis before and after alloimmunization (Online Supplementary Figure S3). RBC transfusion intensity was significantly higher following documentation of alloimmunization than prior to alloantibody formation (2.8±1.3 versus 4.1±2.0 units per month; P<0.0001) (Figure 2D). To further minimize the impact of other variables, we also compared RBC transfusion intensity over a period of 8 months (4 months before and after detection of alloimmunization). In this analysis, RBC transfusion intensity was also significantly higher following alloimmunization (3.6±2.2 versus 4.1±1.8 units per month; P=0.01) (Figure 2E). Reticulocyte response, as expected, was poor. In these patients, there was no clinically significant different variation in platelet and neutrophil counts during the period of analysis (data not shown). Lactate dehydrogenase and bilirubin levels transiently increased in some patients after alloimmunization (data not shown). Haptoglobin results were not available.

Risk factors for alloimmunization To identify potential predictors of alloimmunization 676 patients, including 80 who developed alloantibodies following MDS-related RBC transfusions, were included in random survival forest and recursive partitioning analyses. The number of RBC units transfused before alloantibody formation was the most important predictor of alloimmunization, followed by RBC transfusion dependency status, treatment type, and age (Figure 3A-C). A classification tree using recursive partitioning suggested that 46% (39/84) patients who were dependent on RBC trans-

Figure 2. Alloimmunization is associated with autoantibody formation and increased red blood cell transfusion requirement. (A) Autoantibodies were detected in a significantly higher number of alloimmunized patients than in non-alloimmunized ones (65% vs. 18%; P<0.0001) (B) The total number of RBC units transfused was significantly higher in alloimmunized patients than in non-alloimmunized patients (P<0.0001) (C) In alloimmunized patients, the total number of RBC units transfused was significantly higher after alloimmunization (D) RBC transfusion intensity was significantly higher following alloimmunization during the whole study period (E) RBC transfusion intensity compared over 8 months (4 months before and 4 months after alloimmunization) also confirmed that RBC transfusion intensity increases significantly following alloimmunization.

haematologica | 2017; 102(12)

2025

D. Singhal et al.

fusions developed an alloantibody within the period they were given their first 20 units of RBC (hazard ratio 5; node 7). In the random survival forest analysis, the predicted error rate was 26%, indicating a 74% chance of correctly predicting alloimmunization risk with a tree constructed from the five factors shown in Figure 3A,B. As 73% and 50% of alloimmunized patients developed alloantibodies within the initial 20 units of RBC and within 6 months after their first RBC transfusion, we performed landmark analyses using a competing risk regression model both at baseline and at 6 months following the start of RBC transfusions. For the baseline analysis age, sex, WHO subtype, IPSS-R risk groups and type of treatment were included. For the 6-month analysis numbers of RBC units transfused within 6 months and RBC transfusion dependency status at 6 months were included, in addition to all the baseline factors.

Table 3. Competing risk regression analysis for alloimmunization.

Variables

Baseline Hazard ratio P value

6-month landmark analysis Hazard ratio P value

Treatment 0.248 0.0013 0.236 0.019 IPSS-R* 0.628 0.14 0.273 0.035 Age at diagnosis 1.02 0.15 1.02 0.16 Sex 1.35 0.26 1.48 0.32 MDS-subtype 1.57 0.26 2.49 0.09 RBC-transfusion 0.87 0.7 dependency Number of RBC 0.997 0.78 units transfused by 6 months IPSS-R*: Very Low, Low and Intermediate risk groups were grouped together while High and Very High risk were grouped together.

Figure 3. Recursive partition and random forest analysis to identify patients at higher risk of alloimmunization. (A-B) The random forest analysis predicted an error rate of 26%, thus indicating a 74% chance of correctly predicting alloimmunization risk. RBC units transfused prior to alloantibody, RBC-TD status and treatment type are major predictors of alloimmunization (C) Recursive partition analysis incorporating these variables produced a classification tree demonstrating that the majority of alloimmunized patients developed antibody within the first 20 units of RBC transfusion. Alloimmunization risk was highest in RBC-TD patients within the initial 20 units of RBC (hazard ratio 5; node 7) (D) Cumulative incidence of alloimmunization was significantly lower in patients treated with DMT (E) Cumulative incidence of alloimmunization was significantly higher in RBC-TD IPSS-R Very Low, Low, intermediate risk groups compared to RBC-TD High and Very High risk groups, while alloimmunization was significantly lower in RBC-TI patients in both groups. In Figure 3C: top, middle and bottom values in each node (box) indicate hazard ratio (HR), number of cases developing alloantibody divided by number of cases in that group and percentage of total cases, respectively. For example in node 7, the HR of developing an alloantibody was 5 (top number), 39/84 cases developed alloantibodies and this node represents 12% of the total cases. IPSS-R categories are Very Low (VL), Low (L), Intermediate (I), High (H), and Very High (VH). T-MN: therapy-related myeloid neoplasm; MDS/MPN overlap myelodysplastic/myeloproliferative neoplasm overlap syndrome. DMT: disease-modifying therapy; BSC: best supporting care; RBC-TD: RBC transfusion dependency; RBC-TI: RBC transfusion independent.

2026

haematologica | 2017; 102(12)

RBC alloimmunization in MDS patients

At the baseline and 6-month landmark analyses, alloimmunization risk was significantly lower in patients treated with disease-modifying therapies (hazard ratio 0.24; P=0.0013) (Table 3). The cumulative incidences of alloimmunization at 50 months (4% versus 14%) and 100 months (5% versus 15%) were significantly lower in patients treated with disease-modifying therapies than in those given supportive care only (P<0.001) (Figure 3D) and appeared to be unrelated to number of RBC transfusions, as patients treated with disease-modifying therapies received significantly more RBC units prior to alloantibody detection than patients treated with supportive care (42.5±44 versus 35±54.6; P<0.0001) (Online Supplementary Figure S4A). Within the group given disease-modifying therapies, the rate of alloimmunization was significantly higher in patients treated with azacitidine/lenalidomide than in those treated with intensive chemotherapy and/or allogeneic hematopoietic stem cell transplantation (2.9% versus 0% at 12 months, P=0.02) (Online Supplementary Figure S4B). At the 6-month landmark, IPSS-R groups also predicted alloimmunization; the cumulative incidence of alloimmunization was significantly higher in patients in the combined IPSS-R Very Low, Low and Intermediate risk groups than in those in the combined High and Very High risk groups (P=0.03). The cumulative incidence of alloimmunization was significantly higher in RBC transfusion-dependent patients in IPSS-R Very Low, Low and Intermediate risk groups than in RBC transfusion-dependent patients in IPSS-R High and Very High risk groups. The incidence of alloimmunization was significantly lower in RBC-transfusion-independent groups (Figure 3E).

Discussion Post-transfusion RBC alloimmunization rates vary from 2.5-3.3% for surgical patients to 9-13% in patients with hematologic malignancies.18,23-25 In a large study of more than 21,000 previously non-transfused patients who received RBC transfusions without extended matching, alloantibodies were detected in 2.2% of all transfused patients with a cumulative alloimmunization incidence of 7.7% after 40 units.26 In MDS, highly variable alloimmunization rates have been reported, ranging from 15 to 59%,18-20,27-30 which may reflect small cohorts of patients, inconsistent inclusion criteria and variable follow-up periods. Of these, studies with smaller numbers of patients reported higher alloimmunization rates of 44 to 57%,18,20,30 while a study of 272 patients reported an alloimmunization rate of only 15%.19 This is similar to the 11% cumulative incidence of alloimmunization in our study, which, to the best of our knowledge, is the largest of its type and, crucially, was also able to distinguish between alloimmunization due to MDS-related and unrelated RBC transfusions. Importantly, this is the first study demonstrating a significant increase in RBC transfusion requirements following alloimmunization in MDS patients. In our study 76% of alloimmunized patients developed antibodies against antigens in the Rh and Kell systems, similar to the 62% reported in MDS by Sanz et al.,19 and consistent with observations made in studies of patients with sickle cell disease and thalassemia31,32 and medical patients.26 Differences in immunogenic RBC antigens between donors and recipients also play a role in alloimmunization. These disparities are unlikely to be a major haematologica | 2017; 102(12)

contributor to alloimmunization in our cohort of patients with MDS as the vast majority of the recipients and donors in our cohort were Caucasian. The life expectancy of some higher risk MDS patients is short and, overall, only 11% of transfused MDS patients developed alloantibodies. It is, however, of considerable interest from clinical and cost-effectiveness standpoints to identify the patients at highest risk of RBC alloimmunization, because they would be the ones to benefit most from a policy of extended antigen-matched RBC transfusions. Although the number of RBC units transfused increases the risk of alloimmunization,19 RBC transfusion requirement is dynamic. We found that 73% of patients developing alloantibodies did so within the period of receiving their first 20 units of RBC and 50% of patients within 6 months of their first RBC transfusion. Hence, it is critical to identify patient- and disease-related factors that will differentiate between “responders” and “non-responders” to RBC antigens. In our study, disease-modifying therapies predicted alloimmunization risk at both the baseline and 6-month landmark analyses. Interestingly, the cumulative incidence of alloimmunization was significantly lower in patients treated with intensive chemotherapy and/or allogeneic hematopoietic stem cell transplantation compared to that in patients treated with azacitidine/lenalidomide, possibly due to the greater degree of immunosuppression. Lower alloimmunization rates in IPSS-R High and Very High risk groups compared to Very Low, Low and Intermediate risk groups could be due to the shorter median overall survival, larger proportion of patients requiring disease-modifying therapies, and greater degree of immunosuppression in higher risk groups. Within each group, alloimmunization risk was significantly higher among the RBC transfusion-dependent group compared to the transfusion-independent group, while alloimmunization rate was similarly low in RBC transfusion-independent regardless of risk and treatment assignment. The number of regulatory T cells, known to inhibit alloimmunization, is significantly lower in IPSS low risk patients than in IPSS high risk patients.33,34 This study focused on the clinical implications of alloimmunization, but alloimmunization also leads to increased laboratory workloads and poses the challenge of securing appropriate RBC units in a timely fashion. The clinical consequences of alloimmunization in our study included at least two cases of severe delayed hemolytic transfusion reaction, 22 cases of delayed serological transfusion reaction, and increased RBC transfusion requirements. RBC transfusion requirement increased following alloimmunization, most likely due to the development of additional alloantibodies and autoantibodies, resulting in subclinical serological hemolytic transfusion reactions and/or autoimmune hemolysis. Notably, autoantibodies were detected before, simultaneously or within a short time after alloimmunization. This suggests that alloimmunization drives autoantibody formation. Young et al. detected autoantibodies in 121/2618 (4.6%) individuals with a positive direct or indirect antiglobulin test.12 Interestingly, 41 (34%) of these individuals had both alloantibodies and autoantibodies, and at least 34% of cases developed RBC autoantibodies after previous blood transfusion and in association with alloimmunization.12 Similarly, other studies reported that 8% to 25% of multiply transfused patients with sickle cell dis2027

D. Singhal et al.

ease16 or thalassemia13 have IgG autoantibodies, mostly associated with alloimmunization. RBC autoantibody formation has also been described in both animal and human experimental models of RBC transfusion.15 The pathophysiological mechanisms are not yet fully understood.35 The implication of autoantibodies is two-fold. Firstly, they pose challenges to the transfusion laboratory. Resolution of these complex cases translates into a heavy workload, delay in provision of transfusions and increased cost. Secondly, autoantibodies can be pathological, causing shortened RBC survival and autoimmune hemolysis, which may be severe.13 In MDS patients, hemolysis assessment can be complicated by higher (disease-related) baseline lactate dehydrogenase and poor reticulocyte response due to dyserythropoiesis. Since RBC transfusion can also influence lactate dehydrogenase, haptoglobin and bilirubin levels, a high degree of clinical suspicion is required. The integration of RBC genotyping can minimize or even potentially eliminate labor-intensive serological testing and provide better matched RBC units. Investigators at the Wisconsin Blood Center genotyped 42

References 1. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016;127(20):2391-2405. 2. Brunning RD, Orazi A, Germing U, et al. Myelodysplastic syndromes/neoplasms, overview. In: Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, Vardiman JW, editors. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissue. 4th ed. Lyon: International Agency for Research on Cancer; 2008. p. 88-93. 3. Visser O, Trama A, Maynadie M, et al. Incidence, survival and prevalence of myeloid malignancies in Europe. Eur J Cancer. 2012;48(17):3257-3266. 4. Sekeres MA, Cutler C. How we treat higher-risk myelodysplastic syndromes. Blood. 2014;123(6):829-836. 5. Fenaux P, Ades L. How we treat lower-risk myelodysplastic syndromes. Blood. 2013; 121(21):4280-4286. 6. Hellstrom-Lindberg E. Management of anemia associated with myelodysplastic syndrome. Semin Hematol. 2005;42(2 Suppl 1):S10-13. 7. Hiwase DK, Singhal D, Strupp C, et al. Dynamic assessment of RBC-transfusion dependency improves the prognostic value of the revised-IPSS in MDS patients. Am J Hematol. 2017;92(6):508-514. 8. Malcovati L, Germing U, Kuendgen A, et al. Time-dependent prognostic scoring system for predicting survival and leukemic evolution in myelodysplastic syndromes. J Clin Oncol. 2007;25(23):3503-3510. 9. FDA. Fatalities Reported to FDA Following Blood Collection and Transfusion. US Department of Health and Human Services. [cited 2017 August 20]. Available from: https://www.fda.gov/BiologicsBloodVaccin es/SafetyAvailability/ReportaProblem/Tran sfusionDonationFatalities.

2028

blood group antigens in 43,066 blood donors and were able to provide antigen-negative RBC for more than 94% of requests. There were no cases of acute hemolytic transfusion reaction, delayed hemolytic transfusion reaction, alloimmunization or other adverse reactions.36 In summary, this large registry-based study shows that RBC alloimmunization is not uncommon in MDS, occurs early after commencing transfusion, and has important clinical consequences, including an association with increased RBC transfusion requirements. Provision of extended phenotype-matched RBC units from initiation of a transfusion program for MDS patients can minimize alloimmunization30 and its complications and may be of benefit. Acknowledgments The authors would like to thank the Royal Adelaide Hospital Research Fund, Contributing Haematologistsâ&#x20AC;&#x2122; Committee, Royal Adelaide Hospital and Novartis Pharmaceuticals Australia Pty Limited for research funding support for the SAMDS Registry.

10. Telen MJ, Afenyi-Annan A, Garrett ME, et al. Alloimmunization in sickle cell disease: changing antibody specificities and association with chronic pain and decreased survival. Transfusion. 2015;55(6pt2):13781387. 11. 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. 12. Young PP, Uzieblo A, Trulock E, et al. Autoantibody formation after alloimmunization: are blood transfusions a risk factor for autoimmune hemolytic anemia? Transfusion. 2004;44(1):67-72. 13. Singer ST, Wu V, Mignacca R, et al. Alloimmunization and erythrocyte autoimmunization in transfusion-dependent thalassemia patients of predominantly Asian descent. Blood. 2000;96(10):3369-3373. 14. Castellino SM, Combs MR, Zimmerman SA, et al. Erythrocyte autoantibodies in paediatric patients with sickle cell disease receiving transfusion therapy: frequency, characteristics and significance. Br J Haematol. 1999;104(1):189-194. 15. Garratty G. Autoantibodies induced by blood transfusion. Transfusion. 2004;44 (1):5-9. 16. Aygun B, Padmanabhan S, Paley C, et al. Clinical significance of RBC alloantibodies and autoantibodies in sickle cell patients who received transfusions. Transfusion. 2002;42(1):37-43. 17. Ahrens N, Pruss A, Kahne A, et al. Coexistence of autoantibodies and alloantibodies to red blood cells due to blood transfusion. Transfusion. 2007;47(5):813-816. 18. Stiegler G, Sperr W, Lorber C, et al. Red cell antibodies in frequently transfused patients with myelodysplastic syndrome. Ann Hematol. 2001;80(6):330-333. 19. Sanz C, Nomdedeu M, Belkaid M, et al. Red blood cell alloimmunization in transfused patients with myelodysplastic syndrome or chronic myelomonocytic leukemia. Transfusion. 2013;53(4):710-715. 20. Novaretti MC, Sopelete CR, Velloso ER, et

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

al. Immunohematological findings in myelodysplastic syndrome. Acta Haematol. 2001;105(1):1-6. Winters JL, Richa EM, Bryant SC, et al. Polyethylene glycol antiglobulin tube versus gel microcolumn: influence on the incidence of delayed hemolytic transfusion reactions and delayed serologic transfusion reactions. Transfusion. 2010;50(7):1444-1452. Vamvakas EC, Pineda AA, Reisner R, et al. The differentiation of delayed hemolytic and delayed serologic transfusion reactions: incidence and predictors of hemolysis. Transfusion. 1995;35(1):26-32. Schonewille H, Haak HL, van Zijl AM. Alloimmunization after blood transfusion in patients with hematologic and oncologic diseases. Transfusion. 1999;39(7):763-771. Heddle NM, Soutar RL, O'Hoski PL, et al. A prospective study to determine the frequency and clinical significance of alloimmunization post-transfusion. Br J Haematol. 1995;91(4):1000-1005. Evers D, Zwaginga JJ, Tijmensen J, et al. Treatments for hematological malignancies in contrast to those for solid cancers are associated with reduced red cell alloimmunization. Haematologica. 2016;102(1):5259. Evers D, Middelburg RA, de Haas M, et al. Red-blood-cell alloimmunisation in relation to antigens' exposure and their immunogenicity: a cohort study. Lancet Haematol. 2016;3(6):e284-e292. Guelsin GA, Rodrigues C, Visentainer JE, et al. Molecular matching for Rh and K reduces red blood cell alloimmunisation in patients with myelodysplastic syndrome. Blood Transfus. 2015;13(1):53-58. Ortiz S, Orero MT, Javier K, et al. Impact of azacitidine on red blood cell alloimmunisation in myelodysplastic syndrome. Blood Transfus. 2017;15(5):472-477. Rozovski U, Ben-Tal O, Kirgner I, et al. Increased incidence of red blood cell alloantibodies in myelodysplastic syndrome. Isr Med Assoc J. 2015;17(10): 624-627. Lin Y, Saskin A, Wells RA, et al. Prophylactic

haematologica | 2017; 102(12)

RBC alloimmunization in MDS patients

RhCE and Kell antigen matching: impact on alloimmunization in transfusion-dependent patients with myelodysplastic syndromes. Vox Sang. 2017;112(1):79-86. 31. Matteocci A, Pierelli L. Red blood cell alloimmunization in sickle cell disease and in thalassaemia: current status, future perspectives and potential role of molecular typing. Vox Sang. 2014;106(3):197-208. 32. Rosse WF, Gallagher D, Kinney TR, et al. Transfusion and alloimmunization in sick-

haematologica | 2017; 102(12)

le cell disease. The Cooperative Study of Sickle Cell Disease. Blood. 1990;76(7): 1431-1437. 33. Kordasti SY, Afzali B, Lim Z, et al. IL-17-producing CD4(+) T cells, pro-inflammatory cytokines and apoptosis are increased in low risk myelodysplastic syndrome. Br J Haematol. 2009;145(1):64-72. 34. Kotsianidis I, Bouchliou I, Nakou E, et al. Kinetics, function and bone marrow trafficking of CD4+CD25+FOXP3+ regulatory T

cells in myelodysplastic syndromes (MDS). Leukemia. 2009;23(3):510-518. 35. Kaminski ER, Hows JM, Goldman JM, et al. Lymphocytes from multi-transfused patients exhibit cytotoxicity against autologous cells. Br J Haematol. 1992;81(1):23-26. 36. Flegel WA, Gottschall JL, Denomme GA. Integration of red cell genotyping into the blood supply chain: a population-based study. Lancet Haematol. 2015;2(7):e282e289.

2029

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Myelodysplastic syndromes

Ferrata Storti Foundation

Haematologica 2017 Volume 102(12):2030-2038

Favorable impact of allogeneic stem cell transplantation in patients with therapy-related myelodysplasia regardless of TP53 mutational status

Ibrahim Aldoss,1* Anh Pham,1* Sierra Min Li,2 Ketevan Gendzekhadze,3 Michelle Afkhami,4 Milhan Telatar,4 Hao Hong,4 Abbas Padeganeh,5 Victoria Bedell,5 Thai Cao,1 Samer K Khaled,1 Monzr M Al Malki,1 Amandeep Salhotra,1 Haris Ali,1 Ahmed Aribi,1 Joycelynne Palmer,2 Patricia Aoun,4 Ricardo Spielberger,1 Anthony S Stein,1 David Snyder,1 Margaret R O'Donnell,1 Joyce Murata-Collins,5 David Senitzer,3 Dennis Weisenburger,4 Stephen J Forman,1 Vinod Pullarkat,1 Guido Marcucci,1 Raju Pillai4# and Ryotaro Nakamura1# Department of Hematology and Hematopoietic Cell Transplantation, Gehr Family Center for Leukemia Research, City of Hope; 2Department of Information Sciences, Division of Biostatistics, City of Hope; 3HLA Laboratory, City of Hope; 4Department of Pathology, City of Hope, and 5Department of Cytogenetics, City of Hope, Duarte, CA, USA 1

*IA and AP contributed equally to this manuscript as first authors #RP and RN contributed equally to this manuscript as senior authors

ABSTRACT

Correspondence: rnakamura@coh.org

Received: June 9, 2017. Accepted: September 20, 2017. Pre-published: September 29, 2017. doi:10.3324/haematol.2017.172544 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2030 ÂŠ2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2030

herapy-related myelodysplastic syndrome is a long-term complication of cancer treatment in patients receiving cytotoxic therapy, characterized by high-risk genetics and poor outcomes. Allogeneic hematopoietic cell transplantation is the only potential cure for this disease, but the prognostic impact of pre-transplant genetics and clinical features has not yet been fully characterized. We report here the genetic and clinical characteristics and outcomes of a relatively large cohort of patients with therapy-related myelodysplastic syndrome (n=67) who underwent allogeneic transplantation, comparing these patients to similarly treated patients with de novo disease (n=199). The 5-year overall survival was not different between patients with therapy-related and de novo disease (49.9% versus 53.9%; P=0.61) despite a higher proportion of individuals with an Intermediate-2/High International Prognostic Scoring System classification (59.7% versus 43.7%; P=0.003) and high-risk karyotypes (61.2% versus 30.7%; P<0.01) among the patients with therapyrelated disease. In mutational analysis, TP53 alteration was the most common abnormality in patients with therapy-related disease (n=18: 30%). Interestingly, the presence of mutations in TP53 or in any other of the high-risk genes (EZH2, ETV6, RUNX1, ASXL1: n=29: 48%) did not significantly affect either overall survival or relapse-free survival. Allogeneic stem-cell transplantation is, therefore, a curative treatment for patients with therapy-related myelodysplastic syndrome, conferring a similar long-term survival to that of patients with de novo disease despite higher-risk features. While TP53 alteration was the most common mutation in therapy-related myelodysplastic syndrome, the finding was not detrimental in our case-series. Introduction Therapy-related myelodysplastic syndrome (t-MDS) is a well-recognized clonal hematopoietic disorder occurring as a late complication following exposure to genotoxic chemotherapy and/or radiation therapy.1 Based on the 2016 World Health Organization (WHO) classification of myeloid neoplasms and acute haematologica | 2017; 102(12)

t-MDS: HCT outcomes and molecular features

leukemia, t-MDS is recognized as part of therapy-related myeloid neoplasms.2,3 With recent advances in the management of early malignancies and prevalent use of adjuvant chemotherapy, the incidence of t-MDS seems to be increasing and the condition is becoming an increasing concern for cancer survivors. t-MDS has an aggressive clinical course with generally dismal outcomes often accompanied by high-risk genetic features.4-7 Patients with t-MDS are usually elderly, have a poor performance status, and frequently manifest residual toxicity from prior therapies. Limiting factors, including adverse disease biology and poor clinical phenotype, commonly result in suboptimal responses to conventional chemotherapy, which consequently lead to low median survival rates for these patients.4,6-8 Allogeneic hematopoietic cell transplantation (HCT) is the only curative option for t-MDS, but to date, the longterm survival rate following this strategy has been relatively low with an excess of treatment-related mortality.9-12 With regards to pathogenesis, t-MDS is thought to be either secondary to genomic alterations induced by cytotoxic therapy, or to arise via outgrowth of pre-existing pre-leukemic myeloid clones after exposure to cytotoxic therapy.13,14 Patients with t-MDS usually present with a TP53 mutation,5,15-17 which is known to be associated with the aggressive disease course. While this mutation is less common in patients with de novo MDS and acute myeloid leukemia,5,16,18-22 the post-transplant relapse rate has been noted to be higher among patients with acute myeloid leukemia/MDS carrying the mutation.5,20,21 Bejar et al. reported no survivors beyond 5 years after allogeneic HCT among 18 patients with a TP53 mutation.20 In cohorts of mixed therapy-related and de novo MDS, correlations between several other mutations (ASXL1, RUNX1, EZH2, ETV6, TET2, and DNMT3A) and inferior transplant outcomes have been demonstrated.19-21 In one study, mutations in any of the five high-risk genes (TP53, EZH2, ETV6, RUNX1 or ASXL1) predicted lower overall survival for MDS patients, independent of other clinical factors.19 However, it is largely unknown whether the presence or absence of these mutations has a similar adverse predictive or prognostic impact when only patients with t-MDS who undergo allogeneic HCT are considered. Here, we report the outcome of one of the largest molecularly characterized cohorts of t-MDS patients who underwent allogeneic HCT.

Methods Study population A total of 266 consecutive patients with a diagnosis of MDS who received an allogeneic transplant from a matched sibling or unrelated donor at the City of Hope between January 2000 and October 2014 were analyzed. Patients with chronic myelomonocytic leukemia, pancytopenia and/or dysplasia associated with paroxysmal nocturnal hemoglobinuria or aplastic anemia, and those who underwent cord blood or haploidentical transplantation were excluded. The diagnosis of t-MDS was based solely on a patient’s medical history of prior exposure to any cytotoxic chemotherapy and/or radiotherapy administered for prior malignant or non-malignant conditions. The study was approved by the City of Hope Institutional Review Board. haematologica | 2017; 102(12)

Data collection Prior medical history, the patient’s demographic information, cytogenetic and molecular data, prior treatments including hypomethylating agents and transplant outcomes were collected through the institution’s electronic medical records, chart reviews and the blood and marrow transplant program database. An International Prognostic Scoring System (IPSS) score was generated for each patient, based on the percentage of myeloblasts in the marrow at the time of diagnosis, number of cytopenias, and risk classification of cytogenetics.

DNA samples for next-generation sequencing and microarrays To include granulocytes, DNA was extracted from whole blood samples, following the manufacturer’s recommendations (Qiagen, Germantown, MD, USA). DNA samples were submitted for initial HLA-typing after the diagnosis of MDS and before HCT, based on the assumption that circulating myeloid cells in MDS patients are clonal and carry the same mutational profile of MDS. All patients had ≥20% circulating myeloid lineage cells at the time of the sample collection.

Next-generation sequencing library preparation and bioinformatics analysis Next-generation sequencing (NGS) libraries were prepared from genomic DNA (40 ng) using the SureSelect target enrichment system (Agilent Technologies Inc.) after transposase-based fragmentation and adapter ligation. The adapter-ligated library was amplified by polymerase chain reaction and quality control was performed for sizing and concentration. Target regions were captured using a customized SureSelect library (Agilent Technologies) for all coding exons plus ten flanking bases of 72 genes (Online Supplementary Table S1). After hybridization of 750 ng of adapter-ligated library with biotin-labeled probes that are specific to target regions, the dual-index tag was added during post-capture polymerase chain reaction amplification. The amplified captured libraries were quality-controlled using a high sensitivity DNA Bioanalyzer kit (Agilent Technologies Inc.) then pooled and sequenced using Miseq V2 Reagent Kit/300 cycles with 150 bp paired-end sequencing. Alignment of sequence reads to the human genome (GRCh37/hg19), variant calling and annotation were performed independently using two software applications – CLCBiomedical Workbench (CLC Bio, Aarhus, Denmark) and NextGENE (Softgenetics, State Collage, PA, USA). Annotated variants were processed using previously published criteria.23,24 Synonymous variants, variants located >2 bp outside protein-coding regions, polymorphisms present in >1% in population databases including ExaC, Exome Variant Server and the 1000 Genomes Project, and variants with <30X coverage were filtered. The remaining variants were evaluated using tumor-specific databases (COSMIC, cBioportal), information retrieved from literature, sequence conservation, and in silico prediction algorithms, including SIFT, Polyphen-2, and FATHMM, for clinical significance.

Microarray methods Cytogenomic microarray analysis was performed using the Affymetrix CytoScan HD platform, which consists of more than 2.6 million oligonucleotide probes across the genome including ~1.9 million unique non-polymorphic probes and 750,000 single nucleotide polymorphisms. Genomic linear positions in this microarray are given relative to GRCh37/hg19. The data were Initially analyzed with the Affymetrix Chromosome Analysis Suite. All pathogenic, likely pathogenic, large regions of copy neutral loss/absence of heterozygosity (LOH/AOH) and variants of uncertain significance are reported in Online Supplementary Table 2031

I. Aldoss et al.

S2. Common copy number variations (CNV) or regions of LOH/AOH observed in the general population were deemed benign. Truly balanced rearrangements, some forms of polyploidy, low-level mosaicism, and point mutations were not detectable using this assay.

Definitions of outcomes Overall survival was defined as the time from the day of transplantation to death from any cause. Patients who were alive at their last follow-up were censored. Death from causes other than relapse was considered non-relapse mortality. Relapse was defined as the time to onset of recurrent t-MDS, determined by morphological evidence in bone marrow or extramedullary sites. Relapse-free survival was defined as the time to relapse or death from any cause, whichever came first. Acute and chronic graft-versus-host disease were graded according to previously published criteria.25,26

Statistical analysis The patients’ characteristics and disease- and transplant-related variables were summarized with descriptive statistics. Overall survival was computed using the Kaplan-Meier method. When calculating non-relapse mortality, relapse was counted as a competing risk, i.e. only patients who did not relapse and died were counted as an event. For patients who relapsed but were alive, the last contact was used as the latest follow-up. Similarly, the cumulative incidence of relapse was calculated with death before relapse as a competing risk factor. Gray method was used when calculating cumulative incidence of relapse and non-relapse mortality. Cox proportional hazard models were used for univariate and multivariate analyses and the hazard ratios are reported with a 95% confidence interval (CI). Stepwise variable selection with a backward Akaika information criterion was used for selecting variables. The full set of variables included age at transplant, sex combination of the donor and recipient, race, period of transplantation, time from diagnosis to transplantation, conditioning regimen, HLA transplant match, graft type, cytomegalovirus status for the donor and recipient and MDS classification. Due to correlations among t-MDS karyotype, marrow blast percentage before HCT and IPSS score (≤1 versus >1), these features were included in the model one at a time after model selection.

Results Patients’ characteristics Of the 266 patients included in this analysis, 67 were classified as having t-MDS and 199 as having de novo MDS. The clinical characteristics of all included patients, and differences between t-MDS and de novo MDS are reported in Table 1. When compared to the de novo MDS cohort, t-MDS cases less frequently presented with refractory anemia with excess blasts-1/2 (39.1% versus 47.0%; P=0.004), had higher cytogenetic risk (P=0.00003), lower percentage of blasts at the time of HCT (P=0.03), higher IPSS score at diagnosis (P=0.03) and more frequently received reduced intensity conditioning (94.0% versus 70.4%; P=0.00003). There was no significant difference in median age at the time of HCT (P=0.80), gender (P=0.06), transplant era (P=0.14), time from diagnosis to HCT (P=0.49), donor type (P=0.72), graft source (P=0.17), graftversus-host disease prophylaxis regimen (P=0.10) or donorrecipient cytomegalovirus status (P=0.10) between patients with t-MDS and those with de novo MDS. 2032

Table 1. Characteristics of the patients and their transplants.

Characteristics of patients

t-MDS

de novo MDS

P-value

Number of patients Median age at transplantation Gender Female Male Oncologic history Hematologic Carcinoma Sarcoma Other Treatment history Chemotherapy alone Radiation alone Chemotherapy/radiation Period of the transplant 2000-2004 2005-2009 2010-2014 Time from diagnosis to transplant ≤1 year 1-3 years 3 + years Missing MDS classification RCUD/RARS/RCMD/5q-syndrome RAEB-1/RAEB-2 MDS-U Other Missing MDS cytogenetics Good Intermediate Poor Unknown Missing Blast percentage before HCT ≤5% 5%-10% >10% IPSS score at diagnosis 0 or 1 >1 Missing Conditioning regimen Myeloablative Total body irradiation-based Chemotherapy-based Reduced intensity Fludarabine/melphalan Others Donor source Identical sibling Matched unrelated Mismatched sibling Mismatched unrelated

67 56.03

199 54.51

0.80

35 (52.24%) 32 (47.76%)

77 (38.70%) 122 (61.31%)

0.06

43 (64.18%) 18 (26.87%) 2 (2.99%) 4 (5.97%)

N/A

37 (55.22%) 4 (5.97%) 26 (38.81%)

N/A

14 (20.90%) 18 (26.87%) 35 (50.24%)

42 (21.11%) 78 (39.20%) 79 (39.70%)

0.14

45 (67.16%) 10 (14.93%) 6 (8.96%) 6 (8.96%)

126 (63.32%) 45 (22.61%) 22 (11.06%) 6 (3.02%)

0.49 (missing excluded)

14 (21.88%) 23 (39.06%) 20 (26.56%) 4 (4.69%) 6 (7.81%)

63 (30.00%) 89 (47.00%) 23 (12.00%) 10 (5.00%) 14 (6.00%)

0.004* (unknown/ missing excluded)

15 (22.39%) 11 (16.42%) 41 (61.19%) 0 (0.00%) 0 (0.00%)

99 (49.75%) 2.84e-05**** 32 (16.08%) (unknown/ 61 (30.65%) missing 2 (1.01%) excluded) 5 (2.51%)

48 (71.64%) 11 (16.42%) 8 (11.94%)

108 (50.24%) 40 (20.10%) 51 (25.63%)

0.03*

26 (38.81%) 40 (59.70%) 1 (1.49%)

106 (53.27%) 87 (43.72%) 6 (3.02%)

0.03 (missing excluded)

4 (5.97%) 1 3 63 (94.03%) 61 2

59 (29.65%) 2.55e-05**** 12 47 140 (70.35%) 126 14

25 (37.31%) 25 (37.31%) 1 (1.49%) 16 (23.88%)

84 (42.21%) 67 (33.67%) 1 (0.50%) 47 (23.62%)

0.72 (HLA mismatched sibling excluded)

4 (5.97%) 63 (94.03%)

25 (12.56%) 174 (87.44%)

0.17

Graft source Bone marrow Peripheral stem cells

continued on the next page

haematologica | 2017; 102(12)

t-MDS: HCT outcomes and molecular features

continued from the previous page

GvHD prophylaxis With sirolimus Other Missing GvHD maximum grade Missing I II-IV None CMV status: donor->recipient Negative-> negative Negative-> positive Positive-> negative Positive-> positive Missing

A 50 (70.31%) 17 (29.69%) 0 (0.00%)

121 (59.50%) 70 (36.00%) 8 (4.50%)

0.10 (unknown/ missing excluded)

1 (1.49%) 15 (22.39%) 26 (38.81%) 25 (37.31%)

4 (2.01%) 28 (14.07%) 99 (47.75%) 68 (34.17%)

0.28 (missing excluded)

5 (7.46%) 30 (44.78%) 6 (8.96%) 25 (37.31%) 1 (1.49%)

23 (11.56%) 56 (28.14%) 25 (12.56%) 93 (46.73%) 2 (1.01%)

0.10 (unknown/ missing excluded)

t-MDS: therapy-related myelodysplastic syndrome; RCUD: refractory cytopenia with unilineage dysplasia; RARS: refractory anemia with ringed sideroblasts; RCMD: refractory cytopenia with multilineage dysplasia; RAEB: refractory anemia with excess blasts; MDS-U: unclassifiable MDS; HCT: hematopoietic cell transplantation; IPSS: International Prognostic Scoring System; GvHD: graft-versus-host disease; CMV: cytomegalovirus.

Allogeneic hematopoietic cell transplantation outcomes After a median follow-up of 4.8 years (range, 0.5-15.8) for surviving patients, the 5-year overall survival for the entire cohort was 52.8% (95% CI: 46.2-59.4%). There were no significant differences in the 5-year overall survival (49.9% versus 53.9%; P=0.61), relapse-free survival (47.2% versus 49.5%; P=0.68), non-relapse mortality (30.2% versus 26.8%; P=0.48) and relapse rate (22.6% versus 23.7%; P=0.81) between patients with t-MDS and those with de novo MDS (Figure 1A-C). There was also no significant difference in the incidence and severity of acute graft-versus-host disease between t-MDS and de novo MDS patients (grade II-IV: 38.8% versus 47.8%; P=0.28).

Prognostic factors for survival On univariate analysis for patients with t-MDS, more recent era of HCT (2005-2009 and 2010-2014) was predictive of longer survival (HR=0.41; P=0.05) when compared with transplantation in the earlier era (2000-2004) (HR=0.32; P=0.01), while blast percentage at the time of HCT predicted inferior survival (HR=2.78; P=0.05). Within the group with de novo MDS, male patients and younger patients showed improved survival, whereas those with a poor-risk karyotype and those receiving stem cells from an unrelated donor had a worse survival. In the multivariate model, prior cytotoxic therapy before MDS diagnosis (t-MDS) did not affect survival (P=0.7) in the whole cohort. For t-MDS (n=67), older age was associated with a trend toward a lower overall survival (HR: 1.04 for each year; P=0.06). Among t-MDS patients who underwent allogeneic HCT, compared with patients transplanted between 2000-2004, those transplanted during more recent periods had statistically superior overall survival (for 2005-2009: HR=0.27; P=0.02 and for 2010-2014: HR=0.21; P=0.002) and relapse-free survival (for 2005-2009: HR=0.28; P=0.16 and for 2010-2014: HR=0.20; P=0.002). Karyotypes, IPSS score and percentage of marrow blasts before allogeneic HCT were not haematologica | 2017; 102(12)

Figure 1. Transplant outcomes for patients with therapy-related (solid line) and de novo myelodysplastic syndrome (dotted line) following allogeneic hematopoietic cell transplantation. (A) Overall survival, (B) relapse free survival, (C) cumulative incidence of relapse and non-relapse mortality (NRM).

independently associated with overall survival or relapsefree survival (Table 2). For de novo MDS (n=199), older age (HR=1.03 for each year; P=0.002), unrelated donor (HR=1.84; P=0.01) and IPSS Intermediate-2/High classification (HR=1.51; P=0.06) were the only independent predictors of overall survival (Table 2). Younger age (HR=1.02 for each year; P=0.006), bone marrow as the graft source (HR=2.03; P=0.02) and Intermediate-2/High IPSS score (HR=1.56; P=0.03) were independently associated with shorter relapse-free survival. 2033

I. Aldoss et al.

Somatic mutations in therapy-related myelodysplasia We focused our molecular analysis on t-MDS patients only. Of the 67 t-MDS patients, 60 had available pre-HCT DNA samples, of which 43 (72%) had at least one detectable gene mutation. Among patients with detectable mutations, the median number of mutations per case was 2 (range, 1-6) (Figure 2). The most common mutated gene was TP53, which was present in 18 (30%) cases. Other more common mutations (observed in ≥4 cases) were RUNX1 (12%), TET2 (8%), U2AF1 (8%), ASXL1 (8%), DNMT3A (7%) and SETBP1 (7%). Of 18 cases with somatic mutations involving the TP53 gene, five (28%) had multiple distinct TP53 mutations (two mutations: n=4; three mutations: n=1). Most TP53 mutations were localized at the DNA binding domain (Online Supplementary Figure S1). Additionally, non-TP53 mutations (U2AF1, RUNX1, ASXL1, TET2, and STAG2) were observed in 33% of TP53-mutated cases (Online Supplementary Table S2). TP53 mutations were more frequently associated with complex and/or monosomal karyotype compared to TP53-wild type cases (78% versus 38%; P=0.02). No statistically significant differences in age, sex, prior therapy, prior malignancies, latency from prior diagnosis to MDS diagnosis, marrow blasts, number of cytopenias or IPSS score were observed between patients with mutated TP53 and those with the wild-type gene (Table 3). There was

also no difference in the number of patients who had received hypomethylating agents before allogeneic HCT between the TP53-mutated and TP53 wild-type groups (39% versus 36%). Interestingly, a TP53 mutation did not adversely affect post-transplant overall survival (HR=1.12; P=0.79) or relapse-free survival (HR=1.42; P=0.37) (Figure 3A,B). The 3-year overall survival and relapse-free survival rates for TP53-mutated t-MDS cases were 51.3% and 41.2%, respectively. The presence of more than one TP53 mutation also did not affect either overall survival or relapse-free survival (Online Supplementary Figure S2A,B). Investigating previously proposed high-risk genes (TP53, EZH2, ETV6, RUNX1 and ASXL1) in MDS,19 we identified that at least one of these genes was mutated in 29 (48%) of our t-MDS cases. t-MDS cases with high-risk mutations were more common after hematologic malignancies (P=0.005) and had additional mutations (55% versus 23%; P=0.02) compared to t-MDS with non-high-risk mutations, but no differences were observed in cytogenetics (P=0.15) or other features (Table 3). Contrary to previous reports,19 the presence of one of the high-risk mutations did not adversely influence overall survival

Table 2. Multivariate analysis for overall survival and relapse-free survival.

HR (95% CI) P-value Overall survival t-MDS (n=67) Age at transplantation 1.04 (1.00, 1.07) Period of the transplant 2000-2004 1.00 2005-2009 0.27 (0.09, 0.77) 2010-2014 0.21 (0.08, 0.58) Time of diagnosis to transplant ≤1 year 1.00 1-3 years 0.31 (0.07, 1.48) 3 + years 2.26 (0.66, 7.76) IPSS score at diagnosis ≤1 (Low, Int-1) 1.00 >1 (Int-2, High) 0.57 (0.22, 1.44) De novo (n=199) Age at transplantation 1.03 (1.01, 1.05) Donor source Identical sibling 1.00 Matched unrelated 1.84 (1.13, 3.00) Mismatched unrelated 1.36 (0.76, 2.45) Graft source Peripheral stem cells 1.00 Bone marrow 1.81 (0.99, 3.32) IPSS score at diagnosis ≤1 (Low, Int-1) 1.00 >1 (Int-2, High) 1.51 (0.99, 2.30) CMV status: donor->recipient Negative-> negative N/A Negative-> positive Positive-> negative Positive-> positive

HR (95% CI) P-value Relapse-free survival

0.060

1.04 (1.00, 1.08)

0.045

0.015* 0.002**

1.00 0.28 (0.10, 0.79) 0.20 (0.07, 0.55)

0.16 0.002**

0.143 0.197

1.00 0.42 (0.11, 1.61) 2.05 (0.60, 6.98)

0.207 0.250

0.233

1.00 0.59 (0.23, 1.47)

0.255

0.002**

1.02 (1.01, 1.04)

0.006**

N/A

0.053

1.00 2.03 (1.13, 3.66)

0.018*

0.055

1.00 1.56 (1.04, 2.33)

0.031*

N/A

1.00 1.11 (0.57, 2.17) 0.64 (0.28, 1.45) 0.76 (0.40, 1.46)

0.014* 0.300

0.754 0.284 0.408

t-MDS: therapy-related myelodysplastic syndrome; IPSS: International Prognostic Scoring System; CMV: cytomegalovirus.

2034

Figure 2. The distribution and frequency of mutations detected by next-generation sequencing in 43 out of 60 patients with therapy-related myelodysplastic syndrome with detected mutation(s).

haematologica | 2017; 102(12)

t-MDS: HCT outcomes and molecular features

(HR=1.29; P=0.52) or relapse-free survival (HR=1.58; P=0.22) among t-MDS patients who underwent allogeneic HCT in our cohort (Figure 4A,B). While there was a trend toward worse relapse-free survival in patients carrying high-risk genes, this trend did not reach statistical significance, possibly due to the lack of power of the study.

Cytogenomic studies in therapy-related myelodysplasia Of 34 cases with t-MDS with available pre-HCT genomic DNA, cytogenomic microarray analysis provided novel CNV/LOH findings in 24 cases (71%) (Online Supplementary Table S3). These novel findings included the identification of additional abnormalities or modified/clarified previously reported observations. More precisely, 15/34 cases (44%) had LOH not otherwise detectable by conventional cytogenetics or fluorescence in situ hybridization (FISH). The most common CNV occurred in

chromosomes 5, 7, 12, 13 and 17, consistent with typical cytogenetic changes reported in MDS. Additional CNV were detected in chromosomes 2, 6, 11, and 18, which were not detectable by standard MDS FISH panels. Importantly, we found four cases of MDS in which TP53 alterations were found only by cytogenomic microarray or FISH and not by NGS. The overall survival of the TP53 mutated and TP53 wild-type groups remained similar when these additional del17p cases were re-classified into the TP53 mutated group (data not shown).

Discussion Here we report one of the largest single center series of t-MDS patients undergoing allogeneic HCT. Our results, in agreement with previous reports,9,12 indicated that allo-

Table 3. Characteristics of t-MDS based on TP53 and high-risk mutations.

No (n=42) Sex Female 25 (60) Male 17 (40) Age, years 53.25[15.41,70.95] Karyotype MK 3 (7) Complex 13 (31) With MK 9 (21) Without MK 4 (10) Others 26 (62) Prior therapy Chemotherapy 22 (52) Radiation 2 (5) Chemotherapy/radiation 18 (43) Prior cancer Hematologic 25 (60) Solid 15 (36) Benign 2 (5) Latency 4.24 [0.48, 33.3] Blast % at diagnosis >5% 29 (69) 5-10% 9 (21) >10% 4 (10) Cytopenia 0 0 (0) 1 7 (17) 2 16 (38) 3 18 (43) Unknown 1 (2) IPSS score â&#x2030;¤1 18 (43) â&#x2030;Ľ1.5 23 (55) Unknown 1 (2) Pre-HCT HMA Yes 15 (36) No 27 (64) Additional mutations Yes 13 (31) No 29 (69)

TP53 mutation Yes (n=18)

P-value

No (n=31)

High risk mutations Yes (n=29)

7 (39) 11 (61) 59.02[32.16,62.14]

0.17

20 (65) 11 (35) 53.91[15.41,70.95]

12 (41) 17 (59) 53.40 [17.27, 65.67]

0.12

1 (6) 13 (72) 9 (50) 4 (22) 4 (22)

0.02*

1 (3) 10 (32) 7 (23) 3 (10) 20 (65)

3 (10) 16 (55) 11 (38) 5 (17) 10 (34)

0.15

14 (78) 1 (6) 3 (17)

0.11

15 (48) 2 (6) 14 (45)

21 (72) 1 (3) 7 (24)

0.14

12 (67) 3 (17) 3 (17) 6.90 [2.13, 18.44]

0.13

16 (52) 14 (45) 1 (3) 4.15 [0.48, 33.3]

22 (76) 3 (10) 4 (14) 6.35 [2.13, 18.44]

0.005**

0.13

0.47

P-value

0.62

0.86

12 (67) 5 (28) 1 (6)

0.90

21 (68) 7 (23) 3 (10)

20 (69) 7 (24) 2 (7)

1.00

1 (6) 4 (22) 3 (17) 10 (56) 0 (0)

0.16

0 4 14 13 0

(0) (13) (45) (42) (0)

1 (3) 7 (24) 5 (17) 15 (52) 1 (3)

0.08

6 (33) 12 (67) 0 (0)

0.57

11 (35) 20 (65) 0 (0)

13 (45) 15 (52) 1 (3)

7 (39) 11 (61)

1.00

13 (42) 18 (58)

9 (31) 20 (69)

6 (33) 12 (66)

1.00

7 (23) 24 (77)

16 (55) 13 (45)

0.44

0.43

0.02

MK: monosomal karyotype; IPSS: International Prognostic Scoring System; HCT: hematopoietic cell transplantation; HMA: hypomethylating agents. haematologica | 2017; 102(12)

2035

I. Aldoss et al.

geneic HCT is a curative therapeutic option for patients with t-MDS, and approximately half of transplanted patients could achieve long-term survival and perhaps a cure. Importantly, our study illustrates comparable outcomes between patients with t-MDS and those with de novo MDS despite the former having a greater prevalence of high-risk cytogenetic features and higher IPSS scores. Hence, having been previously exposed to cytotoxic therapy was not associated with worse survival (P=0.7) in our cohort. Interestingly, the non-relapse mortality rate was also not higher in t-MDS patients than in de novo MDS ones, despite the former having a history of prior exposure to cytotoxic therapy. The significance of these results also validated the feasibility of reduced intensity conditioning, which is a critical finding as patients with t-MDS are frequently ineligible for myeloablative conditioning regimens due to comorbidities and prior exposure to cytotoxic therapy. In our cohort, we found that cytogenetics, IPSS score and marrow blast percentage at the time of transplanta-

tion did not adversely affect survival following allogeneic HCT in t-MDS patients, suggesting that allogeneic HCT may attenuate or abrogate the adverse prognostic impact that these factors were found to have in non-transplanted and/or mixed cohorts of MDS patients. Nonetheless, we acknowledge that our cohort may not have had the statistical power to rule out this prognostic association. In our multivariate models for t-MDS patients, younger age and more recent era of allogeneic HCT were independently associated with longer overall survival and longer relapsefree survival. The positive impact of younger age and a more recent transplant were consistent with prior publications.9,12 Using NGS, we showed that the majority of t-MDS cases in our cohort carried somatic mutations involved in multiple mechanisms and pathways including transcription factors (TP53, RUNX1, GATA, ETV6), splicing (U2AF1, SF3B1, ZRSR2), epigenetics (TET2, ASXL1, DNMT3A, EZH2, IDH), kinase signaling (PTPN11, NRAS, CBL) and others (ATM, STAG2), in accordance with pre-

Figure 3. Transplant outcomes for patients with therapy-related myelodysplastic syndrome with (dashed line) and without (solid line) TP53 mutation who underwent allogeneic hematopoietic cell transplantation. (A) Overall survival, and (B) relapse-free survival.

Figure 4. Transplant outcomes for patients with therapy-related myelodysplastic syndrome with (dashed line) and without (solid line) one of high-risk mutations who underwent allogeneic hematopoietic cell transplantation. (A) Overall survival, and (B) relapse-free survival.

2036

haematologica | 2017; 102(12)

t-MDS: HCT outcomes and molecular features

vious reports.5,15,25 Furthermore, a TP53 mutation was the most common molecular alteration among t-MDS patients in this cohort, consistent with prior reports.5,16,17 Comprehensive analysis of TP53 function requires characterization of single nucleotide variations, small indels, copy number alterations and LOH of the 17p locus. Our results demonstrated that 27% of t-MDS cases positive by NGS for TP53 alterations had more than one mutation. Although we could not confirm whether these mutations were present in cis or trans, it is likely that they were compound heterozygotes. CNV/LOH was examined by cytogenomic microarray and FISH in 76% of cases assessed by NGS. We found that 10% of cases that were negative for TP53 alterations by NGS were positive for CNV/LOH by cytogenomic microarray or FISH studies and that 11% of cases with TP53 mutations detected by NGS had concurrent CNV/LOH TP53 abnormalities. In our study, however, TP53 mutation did not adversely affect the overall survival of transplanted t-MDS patients, regardless of whether they had received hypomethylating agents prior to allogeneic HCT, or had more than one TP53 mutation. These findings are clinically very relevant, since previous reports showed a very limited survival (3-year overall survival of 20%)5 for TP53-mutated MDS patients, including those undergoing transplantation.5,16,18,20,22 In these studies, the majority of TP53-mutated patients succumbed to either relapse or non-relapse mortality following allogeneic HCT,20,21 thereby leading to the conclusion that alternative therapies other than such transplants should be explored. Furthermore, while we did observe the occurrence of TP53 mutations and/or at least one other mutated gene among those incorporated in a recent high-risk five-gene panel for MDS19 in at least half of our patients, these molecular aberrations did not adversely affect transplant outcomes. Thus, we concluded that allogeneic HCT should be considered as the primary approach for t-MDS, regardless of the patientâ&#x20AC;&#x2122;s genetic features including, but not limited to, TP53 mutation. While we could not identify definitive factors explaining the difference between the outcomes of our cohort and patients in previous studies, it is possible that the difference in conditioning regimen may have played a role. In a recent study reported by the Dana Farber Cancer Institute,19 the majority of TP53 cases were conditioned with a busulfan/fludarabine reduced intensity conditioning regimen (14/18), while the majority of our TP53 mutated patients (16/18) received fludarabine/melphalan as their reduced intensity conditioning in combination with tacrolimus/sirolimus-based graft-versus-host disease prophylaxis.27,28 In fact recent retrospective studies comparing busulfan/fludarabine with fludarabine/melphalan as reduced intensity conditioning for HCT in patients with acute leukemia and MDS have demonstrated an association between busulfan/fludarabine and higher cumulative incidence of relapse and lower progressionfree survival;29-31 while a large MDS study analyzing 1514 patients enrolled in the Center for International Blood and Marrow Transplant Research Repository found that the survival rates of patients with TP53 mutations were similar between those conditioned with myeloablative regimens or reduced intensity regimens.5 In addition, CV/LOH was not assessed in the Dana Farber Cancer

haematologica | 2017; 102(12)

Institute study, which may have affected their analysis. Conventional cytogenetics remains the leading prognostic factor across different prognostic scores in MDS. However, the limitations of classical cytogenetics lie in the difficulty of identifying cryptic structural abnormalities and the inability to identify copy neutral LOH. These abnormalities could consequently lead to inactivation of tumor suppressor genes and activation of oncogenes, which are encountered not uncommonly in myeloid malignancies. Therefore, detection of these abnormalities could potentially be used as prognostic tools, which may improve the classification of MDS scores. In our study, microarray analysis identified additional CNV abnormalities and LOH regions in addition to those that could be found by conventional cytogenetic analysis in 24/34 cases. Of note, several LOH regions detected solely by single nucleotide polymorphism array contained genes previously reported to be mutated in MDS, including FLT3, ASXL1, CBL1, FANCC and TP53. In addition, single nucleotide polymorphism array analysis helped to clarify observations from metaphase cytogenetics in several cases. These findings underscore the significant advantage of using genome-wide single nucleotide polymorphism microarray analyses to identify cryptic pathogenic changes in t-MDS. Abnormalities missed by cytogenomic microarray were either below the detection limit of this assay, or could be attributed to the fact that cytogenomic microarray was conducted on peripheral blood specimens and not the corresponding bone marrow, due to lack of availability. In conclusion, despite limitations inherent to the retrospective nature of the analysis and potential for selection bias related to the decisions regarding prior transplant treatment and timing of transplantation, which were not made uniformly for all t-MDS patients, our study shows that allogeneic HCT is curative for patients with t-MDS. These transplanted patients have outcomes similar to those of transplanted patients with de novo MDS, despite more frequently having cytogenetic, molecular and clinical higher-risk features prior to allogeneic HCT. Furthermore, we demonstrated that there was no increase in non-relapse mortality in t-MDS patients despite prior exposure to cytotoxic therapy. Importantly, although TP53 mutations were highly prevalent in t-MDS patients, they did not adversely affect transplant outcome, leading us to conclude, in contrast to prior reports, that TP53-mutated t-MDS can also be cured with allogeneic HCT. Our study is limited by the lack of TP53 mutation analysis in patients with de novo MDS to serve as a control cohort for genomic comparison; however, this analysis has been conducted by other groups, showing lower rates of TP53 mutations overall in de novo MDS.5,16 The present findings warrant future prospective studies focusing on the heterogeneity of TP53 mutated tMDS cases to identify which subsets could potentially be cured with allogeneic HCT and which cases require alternative novel therapies. Acknowledgments The authors would like to thank Dr. Sally Mokhtari for assistance with editing this manuscript. This work was supported by a Hematological Malignancies Cancer Center Seed Grant, City of Hope.

2037

I. Aldoss et al.

References 1. Larson RA. Therapy-related myeloid neoplasms. Haematologica. 2009;94(4):454-459. 2. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016;127(20):2391-2405. 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. Kayser S, Dohner K, Krauter J, et al. The impact of therapy-related acute myeloid leukemia (AML) on outcome in 2853 adult patients with newly diagnosed AML. Blood. 2011;117(7):2137-2145. 5. Lindsley RC, Saber W, Mar BG, et al. Prognostic mutations in myelodysplastic syndrome after stem-cell transplantation. N Engl J Med. 2017;376(6):536-547. 6. Smith SM, Le Beau MM, Huo D, et al. Clinical-cytogenetic associations in 306 patients with therapy-related myelodysplasia and myeloid leukemia: the University of Chicago series. Blood. 2003;102(1):43-52. 7. Zeidan AM, Al Ali N, Barnard J, et al. Comparison of clinical outcomes and prognostic utility of risk stratification tools in patients with therapy-related vs de novo myelodysplastic syndromes: a report on behalf of the MDS Clinical Research Consortium. Leukemia. 2017;31(6):13911397. 8. Granfeldt Ostgard LS, Medeiros BC, Sengelov H, et al. Epidemiology and clinical significance of secondary and therapy-related acute myeloid leukemia: a national population-based cohort study. J Clin Oncol. 2015;33(31):3641-3649. 9. Kroger N, Brand R, van Biezen A, et al. Risk factors for therapy-related myelodysplastic syndrome and acute myeloid leukemia treated with allogeneic stem cell transplantation. Haematologica. 2009;94(4):542-549. 10. Litzow MR, Tarima S, Perez WS, et al. Allogeneic transplantation for therapy-related myelodysplastic syndrome and acute myeloid leukemia. Blood. 2010;115(9):18501857. 11. Nevill TJ, Hogge DE, Toze CL, et al. Predictors of outcome following myeloabla-

2038

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

tive allo-SCT for therapy-related myelodysplastic syndrome and AML. Bone Marrow Transplant. 2008;42(10):659-666. Yakoub-Agha I, de La Salmoniere P, Ribaud P, et al. Allogeneic bone marrow transplantation for therapy-related myelodysplastic syndrome and acute myeloid leukemia: a long-term study of 70 patients-report of the French society of bone marrow transplantation. J Clin Oncol. 2000;18(5):963-971. Gibson CJ, Lindsley RC, Tchekmedyian V, et al. Clonal hematopoiesis associated with adverse outcomes after autologous stem-cell transplantation for lymphoma. J Clin Oncol. 2017;35(14):1598-1605. Takahashi K, Wang F, Kantarjian H, et al. Preleukaemic clonal haemopoiesis and risk of therapy-related myeloid neoplasms: a case-control study. Lancet Oncol. 2017;18(1):100-111. Ok CY, Patel KP, Garcia-Manero G, et al. Mutational profiling of therapy-related myelodysplastic syndromes and acute myeloid leukemia by next generation sequencing, a comparison with de novo diseases. Leuk Res. 2015;39(3):348-354. Ok CY, Patel KP, Garcia-Manero G, et al. TP53 mutation characteristics in therapyrelated myelodysplastic syndromes and acute myeloid leukemia is similar to de novo diseases. J Hematol Oncol. 2015;8:45. Wong TN, Ramsingh G, Young AL, et al. Role of TP53 mutations in the origin and evolution of therapy-related acute myeloid leukaemia. Nature. 2015;518(7540):552-555. Bally C, Ades L, Renneville A, et al. Prognostic value of TP53 gene mutations in myelodysplastic syndromes and acute myeloid leukemia treated with azacitidine. Leuk Res. 2014;38(7):751-755. Bejar R, Stevenson K, Abdel-Wahab O, et al. Clinical effect of point mutations in myelodysplastic syndromes. N Engl J Med. 2011;364(26):2496-2506. Bejar R, Stevenson KE, Caughey B, et al. Somatic mutations predict poor outcome in patients with myelodysplastic syndrome after hematopoietic stem-cell transplantation. J Clin Oncol. 2014;32(25):2691-2698. Della Porta MG, Galli A, Bacigalupo A, et al. Clinical effects of driver somatic mutations on the outcomes of patients with myelodysplastic syndromes treated with allogeneic hematopoietic stem-cell transplantation. J Clin Oncol. 2016 Sept 6. [Epub ahead of print]

22. Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic classification and prognosis in acute myeloid leukemia. N Engl J Med. 2016;374(23):2209-2221. 23. Li MM, Datto M, Duncavage EJ, et al. Standards and guidelines for the interpretation and reporting of sequence variants in cancer: a joint consensus recommendation of the Association for Molecular Pathology, American Society of Clinical Oncology, and College of American Pathologists. J Mol Diagn. 2017;19(1):4-23. 24. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 2015;17(5):405-424. 25. Bejar R. Implications of molecular genetic diversity in myelodysplastic syndromes. Curr Opin Hematol. 2017;24(2):73-78. 26. Przepiorka D, Weisdorf D, Martin P, et al. 1994 Consensus conference on acute GVHD grading. Bone Marrow Transplant. 1995;15(6):825-828. 27. Nakamura R, Palmer JM, O'Donnell MR, et al. Reduced intensity allogeneic hematopoietic stem cell transplantation for MDS using tacrolimus/sirolimus-based GVHD prophylaxis. Leuk Res. 2012;36(9):1152-1156. 28. Nakamura R, Rodriguez R, Palmer J, et al. Reduced-intensity conditioning for allogeneic hematopoietic stem cell transplantation with fludarabine and melphalan is associated with durable disease control in myelodysplastic syndrome. Bone Marrow Transplant. 2007;40(9):843-850. 29. Damlaj M, Alkhateeb HB, Hefazi M, et al. Fludarabine-busulfan reduced-intensity conditioning in comparison with fludarabinemelphalan is associated with increased relapse risk in spite of pharmacokinetic dosing. Biol Blood Marrow Transplant. 2016;22(8):1431-1439. 30. Kawamura K, Shuichi B, Ishiyama K, et al. Comparison of reduced-intensity conditioning with fludarabine/busulfan and fludarabine/melphalan in patients 50 years or older. Blood. 2016;128(22):3414. 31. Caddell RJ, Ma Z, Dimaggio E, et al. Fludarabine and melphalan results in fewer relapses compared to fludarabine and targeted busulfan in patients receiving reduced intensity conditioning for AML and MDS. Blood. 2016;128(22):3486.

haematologica | 2017; 102(12)

ARTICLE

Acute Myeloid Leukemia

Circular RNAs of the nucleophosmin (NPM1) gene in acute myeloid leukemia

Susanne Hirsch,1 Tamara J. Blätte,1 Sarah Grasedieck,1 Sibylle Cocciardi,1 Arefeh Rouhi,1 Mojca Jongen-Lavrencic,2 Peter Paschka,1 Jan Krönke,1 Verena I. Gaidzik,1 Hartmut Döhner,1 Richard F. Schlenk,1 Florian Kuchenbauer,1 Konstanze Döhner,1 Anna Dolnik1* and Lars Bullinger1* Internal Medicine III, University Hospital Ulm, Germany and 2Department of Hematology, Erasmus University Medical Centre, Rotterdam, the Netherlands

*AD and LB contributed equally to this work

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(12):2039-2047

ABSTRACT

n acute myeloid leukemia, there is growing evidence for splicing pattern deregulation, including differential expression of linear splice isoforms of the commonly mutated gene nucleophosmin (NPM1). In this study, we detect circular RNAs of NPM1 and quantify circRNA hsa_circ_0075001 in a cohort of NPM1 wild-type and mutated acute myeloid leukemia (n=46). Hsa_circ_0075001 expression correlates positively with total NPM1 expression, but is independent of the NPM1 mutational status. High versus low hsa_circ_0075001 expression defines patient subgroups characterized by distinct gene expression patterns, such as lower expression of components of the Toll-like receptor signaling pathway in high hsa_circ_0075001 expression cases. Global evaluation of circRNA expression in sorted healthy hematopoietic controls (n=10) and acute myeloid leukemia (n=10) reveals circRNA transcripts for 47.9% of all highly expressed genes. While circRNA expression correlates globally with parental gene expression, we identify hematopoietic differentiation-associated as well as acute myeloid leukemia subgroupspecific circRNA signatures.

Correspondence: lars.bullinger@uniklinik-ulm.de

Introduction Acute myeloid leukemia (AML) is the most common acute leukemia in adults and despite recent progress in understanding leukemia biology, many patients relapse and ultimately die of the disease.1 While AML is a genetically heterogeneous disease, it has become clear that not only changes at the genome level contribute to AML pathogenesis. During the past few years, next-generation sequencing (NGS) has revealed an accumulation of mutations in genes regulating the splicing process in approximately 10% of AML patients,2 and a large comprehensive NGS study in AML determined AML patients with spliceosome mutations as a clinically relevant distinct AML subgroup.3 This places a novel focus on alternative and aberrant splicing events that seem to play a general role in AML, and an improved understanding of these aberrations might harbor the potential for novel treatment strategies.4 Aberrant splicing events might also impact the large subset of AML patients carrying a mutation in the nucleophosmin (NPM1) gene, which encodes a multifunctional chaperone protein involved in ribosomal biogenesis, apoptosis, and cell proliferation.5 In NPM1-mutated AML patients, an insertion into exon 12 of NPM1 leads to an aberrant localization of the protein in the cytoplasm, which contributes to the leukemogenic phenotype and is considered an AML-defining mutation.6,7 Moreover, there is evidence that both the impairment of NPM1 function due to deletion or dislocation, and the enhancement of NPM1 function due to overexpression, can confer a tumorigenic effect.8,9 As NPM1 bears both proto-oncogene and tumor-suppressor properties, its deregulation by means other than gene mutation might also play a pathogenic role. Recently, it has been shown that, in AML, several alternatively spliced linear isoforms are produced from the NPM1 gene, of which a short R2 variant (NCBI Reference Sequence: NM_001037738.2) was shown to be differentially expressed haematologica | 2017; 102(12)

Received: May 16, 2017. Accepted: September 22, 2017. Pre-published: September 29, 2017. doi:10.3324/haematol.2017.172866 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2039 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2039

S. Hirsch et al.

Figure 1. Characterization of circular NPM1 splice variants. (A) Predicted exon composition of linear and circular NPM1 splice variants, based on information provided in the AceView and circBase databases. Presence of reverse complementary sequences in the flanking introns of a particular circRNA was evaluated and is indicated by red diamonds. (B) A PCR using divergent NPM1-specific primers was performed with cDNA of two AML cell lines and 1 healthy volunteer sample. (C) Sanger sequencing of a KASUMI-1 cell line PCR product and NCBI BLAST alignment confirmed the existence of a backsplice junction and non-canonical exon order of a circular RNA-derived PCR product. (D) Detection of circNPM1 transcripts in leukemia cell lines (n=7) and healthy volunteers (HV, n=3) with backsplice-specific primers. b-actin (ACTB) served as an internal control. IDs according to circBase and length of the expected amplicon are indicated. (E) Exon and intron composition of circNPM1 splice variants detected via Oxford Nanopore long-read sequencing of PCR products.

2040

haematologica | 2017; 102(12)

Circular RNAs of NPM1 in AML

in a large cohort of AML patients.10 Notably, there was also an association between high NPM1 R2 expression and better outcome in CN-AML patients, especially in patients without concomitant FLT3-ITD mutation, thereby providing further evidence that deregulated splicing patterns might contribute to AML pathogenesis. In addition to splicing events affecting protein coding genes, recent transcriptome studies have revealed that more than 60% of the human genome is transcribed and reproducibly detected as RNA transcripts, while only 2% of the genome codes for protein sequences.11 This implies an important biological function and a need for more com-

prehensive analyses of the non-coding transcriptome. In accordance with this, several studies have also pointed to an important role for non-coding RNAs, such as microRNAs (miRNA) and long non-coding RNAs (lncRNA), in AML pathogenesis.12-14 Recently, a new class of non-coding RNAs termed circular RNAs (circRNAs) was discovered.15 CircRNAs are abundant, highly conserved, and their expression is specifically regulated, making them promising candidates for biomarker research.15-17 CircRNAs are formed through a ligation of the 5’ and the 3’ end of a transcript (so-called backsplicing), which leads to a non-canonical order of

Figure 2. Expression of circular NPM1 transcript hsa_circ_0075001 and impacts on gene expression in a cohort of 46 acute myeloid leukemia (AML) patients. Expression of hsa_circ_0075001 and total NPM1 was measured in 46 AML patients by qPCR. All values are normalized to b-actin (ACTB) and are relative to the respective mean expression. (A) Difference in hsa_circ_0075001 expression in AML patients allocated to the high (red) or low (blue) expression group (dichotomized at the median; P<0.001, unpaired t-test). Black lines indicate the 25% percentiles. (B) Expression of hsa_circ_0075001 in relation to total NPM1 in the AML cohort. (C) Principal component analysis of the distribution of global gene expression differences in AML patients allocated to the high (red) or low (blue) hsa_circ_0075001 expression group. (D) Key components of the Toll-like receptor (TLR) signaling pathway were down-regulated in patients with high hsa_circ_0075001 expression (red). The heatmap illustrates gene expression of the top 20 genes of the TLR pathway with low expression colored purple and high expression colored yellow. Hierarchical clustering of AML patients with high (red) or low (blue) hsa_circ_0075001 expression based on these genes. R: Pearson correlation coefficient; R²: coefficient of determination.

haematologica | 2017; 102(12)

2041

S. Hirsch et al.

exons. Different mechanisms have been suggested for the biogenesis of circRNAs, which are able to bring downstream donors into close proximity with upstream splice acceptors, including binding sites for RNA-binding proteins and reverse complementary sequences in flanking introns.18,19 Given the general deregulation in splicing mechanisms in AML, expression of circRNAs might also be impaired in leukemia cells, and altered circRNAs could contribute to leukemogenesis.20 In the current study, based on the relevance of NPM1 in AML, we investigate the expression of circular NPM1 transcripts in both healthy hematopoietic and leukemic cells. In addition to known NPM1 circRNAs (annotated in the database circBase21), we provide evidence for novel variants as well as NPM1 circRNAs differentially expressed in AML, such as hsa_circ_0075001. Quantifying the expression of hsa_circ_0075001 in a cohort of 46 AML patients, we reveal NPM1 mutation-independent expression groups characterized by distinct gene expression profiles. This further suggests a potential functional relevance of circular NPM1 transcripts and adds another level of complexity to the multifaceted gene, NPM1. Via a more comprehensive and unbiased RNA-Seq-based transcriptome analysis, we gain additional insights into the circular RNAome of the hematopoietic system, and determine changes in the circRNA repertoire throughout myeloid differentiation, as well as non-physiological patterns associated with leukemic transformation.

Methods Patient material and healthy controls Acute myeloid leukemia patients were selected from a larger cohort enrolled in the AMLSG_07-04 study (clinicaltrials.gov identifier: 00151242). Informed patient consent was obtained for the study and concomitant scientific investigations, and the research was conducted after approval by the local ethical committee and according to the principles of the Declaration of Helsinki. Cytogenetically normal AML cases were selected for the quantification of hsa_circ_0075001 (n=46) and for RNA-Seq (n=10). For more detailed information please refer to the Online Supplementary Methods and Online Supplementary Table S1. Healthy control samples were collected from 6 individuals after informed consent. For RNA-Seq, bone marrow samples of 3 healthy controls were FACS-sorted as previously described.22

Oxford Nanopore sequencing PCR products for Oxford Nanopore sequencing were generated with outward facing, so-called divergent primers specific for NPM1 (Online Supplementary Table S2), with product sizes ranging from 150bp to 2500bp. 1 mg purified PCR product was subjected to library preparation using the SQK-NSK007 Nanopore Sequencing Kit (v.R9, Oxford Nanopore Technologies, Oxford, UK), following the standard protocol for Amplicon sequencing for the MinIONTM device. 135 ng of end-prepped DNA was subjected to adapter ligation. Oxford Nanopore reads were aligned using BWA-MEM23 and NCBI BLAST.24

Integration of gene expression data The 46 AML patients were dichotomized based on the following criteria: NPM1 mutational status [wild-type (wt) vs. mutated], circNPM1 hsa_circ_0075001 expression (high vs. low, relative to the median expression), or total NPM1 expression (high vs. low). Principal component analysis (PCA) of microarray data was con2042

ducted using the Partek® Genomics Suite® software (Partek, St. Louis, MO, USA). One-way ANOVA was used to examine differences between the respective groups. Differentially expressed genes were defined as log2FC>|0.6| and P<0.05 after controlling the false discovery rate (FDR) using the Benjamini–Hochberg procedure. Pathway analysis was performed using the iPathwayGuide web application (Advaita, Plymouth, MI, USA). Statistical thresholds were set at P<0.05 and log2FC>|0.6|. Microarray data are available at Gene Expression Omnibus (GEO; accession number GSE104099).

RNA-Seq Libraries were prepared from 1 µg of input total RNA using the TruSeq Stranded Total RNA Kit with Ribo-Zero Human/Mouse/Rat (Illumina, San Diego, CA, USA) according to the manufacturer’s instructions. The pooled RNA libraries were sequenced on an Illumina HiSeq2000 with 100bp paired-end reads and an average coverage of 62 (±24)x106 reads per sample. RNASeq data were aligned and quantified using STAR.25 Reads derived from circRNAs mapping to a backsplice junction were identified using an in-house analysis pipeline detecting exons in a shuffled order. A more detailed description of the pipeline is provided in the Online Supplementary Methods. Circular reads were then normalized and transformed to the logarithmic scale using DESeq226 and PCA was performed based on those 500 genes with the highest variance of circRNA expression across all samples. Genes differentially expressing circRNAs were tested for log2FC>|0.6|, and FDR<0.1 was set for correction of multiple comparisons. Gene set enrichment analysis was performed using GSEA and the Molecular Signatures Database (MsigDB).27 To calculate the percentage of genes which also produce circRNAs, the top 10,794 genes with a mean read count over 100 were considered as “highly expressed” and the top 17,755 genes with a mean read count over 10 were considered “markedly expressed” (Online Supplementary Figure S1). Normalized RNA-Seq data are provided in Online Supplementary Tables S3 and S4.

Results Circular NPM1 transcripts in leukemic and healthy cells Several linear and circular transcripts are known to be transcribed from the NPM1 gene, but circRNA biogenesis is only partly understood. An overview of variants as annotated in the databases AceView28 for linear NPM1 transcripts and circBase21 for circRNAs is given in Figure 1A. Since reverse complementary intronic sequences, such as inverted Alu repeats, are thought to be of importance for the formation of the circle,16,29 we investigated the presence of such sequences in introns of the NPM1 gene (ENSG00000181163) using NCBI BLAST.24 Moreover, we assessed a possible correlation of the presence of reverse complementary intronic sequences with exon combinations that take part in backsplicing events. Out of 29 known NPM1 variants, we found reverse complementary sequences in flanking introns of 13 (44.8%) annotated circular transcripts, possibly promoting their biogenesis (Online Supplementary Table S5). However, we also found inverse complementary sequences in introns that are not known to take part in backsplicing events to form a circRNA. Thus, reverse complementary intronic sequences are not the only mechanism promoting circRNA formation. As a first experimental approach, we performed PCR haematologica | 2017; 102(12)

Circular RNAs of NPM1 in AML

with divergent primers in two NPM1 wt (NPM1wt) leukemia cell lines, ME-1 and KASUMI-1, and one sample derived from a healthy volunteer. Primers were designed for exons of NPM1 which are located in the majority of annotated circular NPM1 transcripts. This resulted in multiple PCR products of circular origin in all samples, reflecting various circRNAs produced from the NPM1 gene (Figure 1B). Validation by Sanger sequencing confirmed the specificity for NPM1, the non-canonical exon order, and the backsplice sequence of these circRNAs. A representative example of a previously unknown circNPM1 variant detected in the KASUMI-1 cell line consists of exons 8, 9, 11, and 12 with a backsplice junction located between exons 12 and 8 of NPM1 (Figure 1C). Next, for the specific detection of individual circNPM1 transcripts, backsplice sequence-specific PCR primers were designed, and the screening of cell lines was extended. In total, we analyzed six AML cell lines (including the NPM1 mutant cell line OCI-AML3), one CML cell line in terminal myeloid blast crisis, and 3 samples derived from the peripheral blood mononuclear cell fraction of healthy volunteers, and we were able to specifically amplify 15 circNPM1 variants. While some variants were uniformly expressed among the control and leukemia samples, including hsa_circ_0074997, _0075004, _0075005, _0005341, _0075019 and _0075022, other variants showed lower expression in healthy volunteers compared to leukemia cell lines, such as hsa_circ_0074995, _0074998, _0075001, _0075002 and _0075012 (Figure 1D). To gain insight into the internal structure of circRNAs, we made use of the long-read Oxford Nanopore technology to sequence circular-derived PCR products created with NPM1-specific divergent primers. We detected already annotated circNPM1 variants, including hsa_circ_0074997, _0075000 and _0075004 (Figure 1E). Moreover, we found novel variants that retained complete introns, especially the complete intronic sequences between exons 3 and 4 of NPM1, and between exons 5 and 6. Furthermore, parts of intron 1 were shown to take part in backsplicing events, for example, with exon 6 and parts of intron 2 (Figure 1E).

total NPM1 expression (Pearson correlation coefficient R=0.72) (Figure 2B), raising the question of whether the expression of circRNAs might generally be correlated with that of the respective parental gene. Gene expression profiling for the high versus low hsa_circ_0075001 expression groups identified 2292 differentially expressed (DE) genes between patients using an FDR-corrected P<0.05 and a threshold for log2 fold change (log2FC)>|0.6|. Unsupervised principal component analysis (PCA) illustrates that the hsa_circ_0075001 expression status defined distinct subgroups of patients (Figure 2C). One-way ANOVA confirmed significant differences in gene expression between the high versus low hsa_circ_0075001 expression groups, and a comparative gene expression pathway analysis (statistical thresholds for differential gene expression set at P<0.05 and log2FC>|0.6|) revealed significant differences in the expression of genes involved in the Toll-like receptor (TLR) signaling pathway as an example. This pathway was significantly down-regulated in patients with high hsa_circ_0075001 expression compared to patients with low hsa_circ_0075001 expression (FDR-corrected P<0.0001) (Online Supplementary Table S6). A total of 94 genes were perturbed in this pathway, with the main dysregulated components of the pathway being several TLR genes, such as TLR1, TRL4, TLR5, and TLR7/8, as well as levels of co-receptors like CD14, and downstream adaptor proteins like MYD88 (Figure 2D). Moreover, 185 differentially expressed genes, whose expression was decreased in patients with high hsa_circ_0075001 expression, are known target genes of miR-181, such as Caspase recruitment domain-containing protein 8 (CARD8), Caspase 1 (CASP1) Macrophage scavenger receptor 1 (MSR1) solute carrier family 11 member 1 (SLC11A1), and TLR4. It is known that miR-181 is commonly deregulated in cytogenetically normal AML.30 In contrast, a comparison of AML cases with high versus low total NPM1 expression did not reveal an enrichment of TLR signaling pathway genes among the top 40 pathways, but rather ribosomal protein genes being up-regulated in patients with high total NPM1 expression (FDRcorrected P<0.0001) (Online Supplementary Table S7).

Hsa_circ_0075001 correlates with distinct gene expression patterns

Unbiased evaluation of the circular RNAome in AML

As circNPM1 75001 (circBase ID hsa_circ_0075001) exhibited a highly differential expression pattern in the AML cell lines, it was selected for quantitative screening in a cohort of 46 AML patients in parallel to the expression of total NPM1. No considerable difference could be detected regarding the mean, median, and the range of expression between NPM1wt (n=23) and NPM1 mutated (NPM1mut, n=23) patients (Online Supplementary Figure S2A and B). However, we observed high expression of hsa_circ_0075001 in patients with no or minimal maturation of the predominant blast population [subtypes M0 or M1 according to the French-American-British (FAB) classification system] (Online Supplementary Figure S3). Expression of hsa_circ_0075001 in patients with a more mature blast population (M2, M4 and M5) was significantly lower (median fold change=3; P<0.001, unpaired t-test). Across all AML samples, there was a 4-fold difference in expression (P<0.0001, unpaired t-test) and patients were allocated to a low and a high circNPM1 75001 expression cohort based on the median (Figure 2A). Furthermore, expression of hsa_circ_0075001 correlated strongly with

Given the differential expression of NPM1 circRNA in AML, we aimed for a more global insight into circRNA expression in hematopoietic cells. To this end, we performed ribosomal RNA-depleted RNA-Seq of 10 AML patients (n=5 NPM1mut and n=5 NPM1wt cases), and 10 FACS-sorted healthy control samples [n=4 immature myeloid differentiation stages (myeloblasts and promyelocytes), n=6 more mature myeloid differentiation stages (metamyelocytes and neutrophils)]. Using an in-house script for circRNA identification, in total, we detected circRNAs for 31.7% (n=5635) out of 17,755 markedly expressed genes (mean normalized read count >10 across all 20 RNA-Seq samples) and for 47.9% (n=5173) out of the 10,794 highly expressed genes (mean normalized read count >100) (Online Supplementary Figure S1). While circular expression of many genes was comparable between AML patients and healthy control samples (Pearson correlation coefficient R=0.749) (Figure 3A), some genes particularly showed high circular expression in AML but not in healthy samples, and vice versa. In accordance with our previous finding for NPM1, we

haematologica | 2017; 102(12)

2043

S. Hirsch et al.

Figure 3. Circular RNAome in acute myeloid leukemia (AML) patients and healthy hematopoietic cells. Ribosomal RNA-depleted RNA-Seq of 5 NPM1mut, 5 NPM1wt AML, and 10 healthy hematopoietic control samples. (A) Global circular gene expression in 10 AML patients compared to expression in 10 healthy controls. Circularderived reads were detected for 5694 genes. The x-value of each dot is the mean circular read count for the respective gene in 10 healthy samples; y-value represents the mean circular read count in 10 AML patient samples. y=x is plotted in gray. (B) Mean expression of circRNA transcripts relative to parental gene expression of “markedly expressed” genes in 10 AML patients. The log2 of the mean normalized read counts across all 10 samples is shown. The linear regression function is indicated. (C) Mean expression of circRNA transcripts relative to parental gene expression of “markedly expressed” genes in 10 healthy hematopoietic control samples. (D) CircRNA expression in AML patients compared to healthy control samples. Principal component analysis (PCA) of circRNA expression data of 5 NPM1mut patients (red) and 5 NPM1wt patients (green), and 10 healthy control samples, of which 4 were derived from immature myeloid differentiation stages (blue: myeloblasts and promyelocytes) and 6 from more mature myeloid differentiation stages (purple: metamyelocytes and neutrophils). PCA was performed based on the 500 genes with the highest variance across all samples. (E) Normalized circular read counts in 10 AML patients and 10 healthy control samples are shown for the fms like tyrosine kinase 3 (FLT3) gene which is one of the genes differentially expressing circRNAs. (F) Correlation of circFLT3 expression with parental gene expression in 10 AML patients (red circles) and 10 healthy control samples (blue triangles). Values are shown as normalized read counts. R: Pearson correlation coefficient; R²: coefficient of determination.

2044

haematologica | 2017; 102(12)

Circular RNAs of NPM1 in AML

again observed a tendency towards higher circRNA expression in genes with higher parental gene expression in both leukemia cells (Figure 3B) and healthy control cells (Figure 3C). Nevertheless, there are genes that produce only few or no circRNA transcripts despite high gene expression levels, while others with lower parental gene expression show comparatively high circRNA expression. Furthermore, there was a slight, but distinct difference between AML and healthy control cells with circRNAs for 46.7% of the highly expressed genes in AML and only 43.8% in normal cells (P=0.009, Ď&#x2021;2 test).

AML patients and healthy controls differ in circRNA signature We performed PCA on the circular-derived RNA-Seq data comparing AML with healthy control samples, and at the same time contrasting NPM1mut and NPM1wt patients and healthy myeloid cell samples of more mature myeloblasts and promyelocytes, and more immature metamyelocytes and neutrophils (Figure 3D). This analysis showed that circRNA signatures are associated with myeloid differentiation and are distinct in leukemic cells, which is reflected in the fact that AML samples do not merely resemble immature myeloid cells. Although there was no difference in circNPM1 expression, NPM1mut patients could be distinguished from NPM1wt patients based on their global circRNA expression. In total, differentially expressed circRNAs from 27 genes were found comparing AML to healthy control samples (P<0.05) (Online Supplementary Table S8). Notably, these genes were significantly enriched for the RADMACHER_AML_PROGNOSIS gene set (FDR-corrected P<0.05, GSEA) based on the differential expression of angiopoietin 1 (ANGPT1; log2FC=3.3), UDP-glucose ceramide glucosyltransferase (UGCG; log2FC=-3.5), and fms related tyrosine kinase 3 (FLT3; log2FC=3.4) circRNAs. Among 14 genes with higher circRNA expression in AML (log2FC range 2.1-4.7), the most significantly deregulated circRNAs were produced from genetic suppressor element 1 (GSE1; log2FC=4.3) and FLT3 (see above). Circular-derived FLT3 reads mainly mapped to three different variants: two already annotated variants (hsa_circ_0100163, hsa_circ_0100164) and one novel variant with a backsplice sequence between exons 19 and 16 of FLT3. Since FLT3 is often mutated and constitutively activated in AML,3 we were particularly interested in circFLT3 expression in AML patients compared to healthy cells (Figure 3E). The correlation of total circFLT3 expression with parental gene expression was high in healthy samples (Pearson correlation coefficient R=0.98), whereas this was not observed in AML samples (Pearson correlation coefficient R=0.21) (Figure 3F).

Discussion Here, we investigated circRNA expression of one of the most frequently mutated genes in AML. In agreement with others who detected alternative splicing events in circRNAs,31 we found that there are many different circRNA variants for NPM1 that comprise non-canonical exon and intron sequences. Previously, circRNAs were predicted based on non-canonical exon-exon junctions found by RNA-Seq.15,17 Using a new technology for longread sequencing, Oxford Nanopore, we elucidated the haematologica | 2017; 102(12)

internal structure of several circNPM1 variants for the first time in AML and demonstrated the power of this novel approach. We could show that, in addition to canonical exons, some variants retained full introns. Moreover, we found parts of intronic sequences taking part in backsplicing events. Since these sequences have not been taken into account in many circRNA studies so far, backsplice events between canonical exons and intronic sequences have probably been under-estimated. Besides intronic sequences, we also detected truncated alternative exons of NPM1, which participated in forming the fusion site (truncated versions of exons 2, 5, 7, 8 and 12). In this context, it would be of interest to study the possible effects of splicing mutations in AML on the formation and the internal structure of circRNAs. Moreover, future elucidation of circRNA function will help to understand the potential necessity of the large circRNA variety and their role in AML. Quantification of the circNPM1 variant hsa_circ_0075001 and integration of global gene expression data revealed a distinct hsa_circ_0075001-associated gene expression signature, pointing to a biological relevance of this circRNA. High hsa_circ_0075001 expression was highly associated with a significantly lower expression of genes involved in the TLR signaling pathway. While TLRs form a critical part of the innate immune response, they have recently been implicated in the differentiation of normal hematopoietic cells. Furthermore, TLR1 expression has been linked to leukemic stem cell survival in AML, as well as enhanced TLR1/TLR2 activation in leukemic stem cell differentiation.32-34 This is in line with our observation that high hsa_circ_0075001 expression is associated with low TLR gene expression and a more immature phenotype of the AML blasts. Future functional experiments will be needed to show whether in AML, the TLR pathway deregulation is in part caused by aberrant expression of circRNAs, such as hsa_circ_0075001, or whether the respective circRNA expression changes are in turn secondary to TLR-deregulation, for example, as a marker for differentiation. Interestingly, Marcucci et al. have shown that the expression of TLR signaling pathway genes is associated with prognostically relevant microRNA signatures in AML.35 In particular, the expression levels of miRNA-181 family members were inversely correlated with the expression levels of the TLR signaling pathway genes. In our study, we found that the expression of various known miR-181 target genes was significantly decreased in patients with high hsa_circ_0075001 expression. Since the NPM1 gene contains binding sites for miR-181,36 it would be of great interest to investigate whether circular NPM1 transcripts interact with members of the miR-181 family and thereby influence the expression of genes involved in the TLR signaling pathway. In contrast, the TLR signaling pathway was not among the top 40 pathways associated with total NPM1 expression, although its deregulation could be detected at a weaker level considering the positive correlation of total NPM1 and hsa_circ_75001 expression. However, the fact that the signature is more pronounced in patients with high hsa_circ_75001 expression supports the view that circRNA expression is the decisive factor correlating with TLR pathway deregulation. Among the linear-driven NPM1 pathways, we found that ribosomal protein genes were strongly up-regulated 2045

S. Hirsch et al.

in patients with high total NPM1 expression. Since NPM1 is closely involved in the biogenesis of ribosomes, this finding supports the accuracy and fidelity of our pathway analysis. Moreover, these findings underline the conceivably different functions and impacts of circular and linear NPM1 transcripts. While we could show correlation of circNPM1 variant expression and distinct gene expression patterns, the significance of the respective NPM1 variants is still unknown, and further studies are warranted to study the impact of circular NPM1 RNAs on regulation of gene expression. With regard to circRNA biogenesis, complementary intronic sequences are most likely not the only key mechanism. This is consistent with findings of such sequences in only around half of the circNPM1 variants. Whether circRNA expression is directly linked to the transcription of the linear isoforms of a gene is still an issue of current research and might not be consistent for all genes. The much longer half life of circular transcripts compared to their linear counterparts, as well as impairment of the splicing machinery, might also impact findings. In our study, a positive correlation was found for hsa_circ_0075001 and total NPM1 expression. In line with this, RNA-Seq data revealed a general correlation of circRNA and parental gene expression. However, we also found many exceptions. Not all highly expressed genes produce circRNAs, and those circRNAs with the highest expression level are generally not derived from genes with highest mRNA expression. Moreover, circFLT3 expression did correlate with parental gene expression only in healthy samples, but not in AML cells, thereby further pointing towards a non-random deregulation of circRNAs in AML, independent of mRNA expression.

References 1. Döhner H, Weisdorf DJ, Bloomfield CD. Acute myeloid leukemia. N Engl J Med. 2015;373(12):1136-1152. 2. Dolnik A, Engelmann JC, ScharfenbergerSchmeer M, et al. Commonly altered genomic regions in acute myeloid leukemia are enriched for somatic mutations involved in chromatin remodeling and splicing. Blood. 2012;120(18):e83-e92. 3. Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic Classification and Prognosis in Acute Myeloid Leukemia. N Engl J Med. 2016;374(23):2209-2221. 4. Lee SC-W, Dvinge H, Kim E, et al. Modulation of splicing catalysis for therapeutic targeting of leukemia with mutations in genes encoding spliceosomal proteins. Nat Med. 2016;22(6):672-678. 5. Falini B, Nicoletti I, Bolli N, et al. Translocations and mutations involving the nucleophosmin (NPM1) gene in lymphomas and leukemias. Haematologica. 2007;92(4):519-532. 6. Döhner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129(4):424-447. 7. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health

2046

9. 10.

11. 12.

13.

Circular RNAs have been shown to be tightly regulated in the course of differentiation. Hundreds of circRNAs are, for example, differentially expressed during epithelial-mesenchymal transition,37 and circRNAs are globally up-regulated during neuronal differentiation.38 For the first time, we have shown that the expression of circRNAs, in line with mRNA, miRNA and lncRNA expression, is deregulated in AML, a disease characterized by inadequate differentiation of hematopoietic progenitor cells. We detected AML-associated and differentiation-independent differences between healthy hematopoietic cells and AML cells, and could also distinguish different AML subgroups based on their circRNA expression profile. These differences might in part be explained by the different degree of differentiation of the leukemia-initiating cells, but also by the genomic alterations influencing splicing factors and epigenetic modulators. As a consequence, the determination of the circular RNAome provides additional valuable insights into the biology of a leukemic cell. Furthermore, circRNAs are more stable than their linear mRNA counterparts,39,40 and thus could serve as potential biomarkers for classification and risk stratification in AML. In the future, additional studies will be required: i) to further determine the biological function of circRNA candidates in both healthy and malignant tissue; and ii) to define the prognostic and predictive impact of circRNA signatures in AML. Funding The authors would like to thank the Deutsche Forschungsgemeinschaft (SFB 1074 project B3 to KD and LB and Heisenberg-Professur BU 1339/8-1 to LB) and the Studienstiftung des deutschen Volkes which supported SH for funding this project.haematologica | 2017; 102(12)

Organization (WHO) classification of myeloid neoplasms and acute leukemia. Blood. 2016;127(20):2391-2405. Nozawa Y, Van Belzen N, Van der Made AC, Dinjens WN, Bosman FT. Expression of nucleophosmin/B23 in normal and neoplastic colorectal mucosa. J Pathol. 1996;178(1):48-52. Grisendi S, Mecucci C, Falini B, Pandolfi PP. Nucleophosmin and cancer. Nat Rev Cancer. 2006;6(7):493-505. Zajac M, Dolnik A, Stasiak G, et al. Analysis of NPM1 splice variants reveals differential expression patterns of prognostic value in acute myeloid leukemia. Oncotarget. 2017 Aug 3. [Epub ahead of print] Djebali S, Davis CA, Merkel A, et al. Landscape of transcription in human cells. Nature. 2012;489(7414):101-108. Rücker F, Russ A, Cocciardi S, et al. Altered miRNA and gene expression in acute myeloid leukemia with complex karyotype identify networks of prognostic relevance. Leukemia. 2013;27(2):353-361. Russ AC, Sander S, Lück SC, et al. Integrative nucleophosmin mutation-associated microRNA and gene expression pattern analysis identifies novel microRNAtarget gene interactions in acute myeloid leukemia. Haematologica. 2011; 96(12):1783-1791.

14. Garzon R, Volinia S, Papaioannou D, et al. Expression and prognostic impact of lncRNAs in acute myeloid leukemia. Proc Natl Acad Sci USA. 2014;111(52):1867918684. 15. Salzman J, Gawad C, Wang PL, Lacayo N, Brown PO. Circular RNAs are the predominant transcript isoform from hundreds of human genes in diverse cell types. PloS One. 2012;7(2):e30733. 16. Jeck WR, Sorrentino JA, Wang K, et al. Circular RNAs are abundant, conserved, and associated with ALU repeats. RNA. 2013;19(2):141-157. 17. Memczak S, Jens M, Elefsinioti A, et al. Circular RNAs are a large class of animal RNAs with regulatory potency. Nature. 2013;495(7441):333-338. 18. Ashwal-Fluss R, Meyer M, Pamudurti NR, et al. circRNA biogenesis competes with pre-mRNA splicing. Mol Cell. 2014;56(1):55-66. 19. Ivanov A, Memczak S, Wyler E, et al. Analysis of intron sequences reveals hallmarks of circular RNA biogenesis in animals. Cell Rep. 2015;10(2):170-177. 20. Guarnerio J, Bezzi M, Jeong JC, et al. Oncogenic role of fusion-circRNAs derived from cancer-associated chromosomal translocations. Cell. 2016;165(2):289-302. 21. Glažar P, Papavasileiou P, Rajewsky N. circBase: a database for circular RNAs.

haematologica | 2017; 102(12)

Circular RNAs of NPM1 in AML

RNA. 2014;20(11):1666-1670. 22. Sun SM, Dijkstra MK, Bijkerk AC, et al. Transition of highly specific microRNA expression patterns in association with discrete maturation stages of human granulopoiesis. Br J Haematol. 2011;155(3):395398. 23. Li H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. 2013;arzXiv:1303.3997v2. 24. Johnson M, Zaretskaya I, Raytselis Y, Merezhuk Y, McGinnis S, Madden TL. NCBI BLAST: a better web interface. Nucleic Acids Res. 2008;36(Suppl 2):W5-W9. 25. Dobin A, Davis CA, Schlesinger F, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29(1):15-21. 26. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):1. 27. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA. 2005;102(43):15545-15550. 28. Thierry-Mieg D, Thierry-Mieg J. AceView: a comprehensive cDNA-supported gene

haematologica | 2017; 102(12)

29. 30.

31.

32.

33.

34.

and transcripts annotation. Genome Biol. 2006;7(1):1. Liang D, Wilusz JE. Short intronic repeat sequences facilitate circular RNA production. Genes Dev. 2014;28(20):2233-2247. Schwind S, Maharry K, Radmacher MD, et al. Prognostic significance of expression of a single microRNA, miR-181a, in cytogenetically normal acute myeloid leukemia: a Cancer and Leukemia Group B study. J Clin Oncol. 2010;28(36):5257-5264. Gao Y, Wang J, Zheng Y, Zhang J, Chen S, Zhao F. Comprehensive identification of internal structure and alternative splicing events in circular RNAs. Nat Commun. 2016;7:12060. Nagai Y, Garrett KP, Ohta S, et al. Toll-like receptors on hematopoietic progenitor cells stimulate innate immune system replenishment. Immunity. 2006;24(6):801-812. Okamoto M, Hirai H, Taniguchi K, et al. Toll‐like Receptors (TLRs) are expressed by myeloid leukaemia cell lines, but fail to trigger differentiation in response to the respective TLR ligands. Br J Haematol. 2009;147(4):585-587. Eriksson M, Peña P, Chapellier M, et al. Toll-like Receptor 1 Is a Candidate

35.

36.

37.

38.

39.

40.

Therapeutic Target in Acute Myeloid Leukemia. Blood. 2014;124(21):5782-5782. Marcucci G, Radmacher MD, Maharry K, et al. MicroRNA expression in cytogenetically normal acute myeloid leukemia. N Engl J Med. 2008;358(18):1919-1928. Betel D, Wilson M, Gabow A, Marks DS, Sander C. The microRNA. org resource: targets and expression. Nucleic Acids Res. 2008;36(Suppl 1):D149-D153. Conn SJ, Pillman KA, Toubia J, et al. The RNA binding protein quaking regulates formation of circRNAs. Cell. 2015; 160(6):1125-1134. Rybak-Wolf A, Stottmeister C, Glažar P, et al. Circular RNAs in the mammalian brain are highly abundant, conserved, and dynamically expressed. Mol Cell. 2015; 58(5):870-885. Bahn JH, Zhang Q, Li F, et al. The landscape of microRNA, Piwi-interacting RNA, and circular RNA in human saliva. Clin Chem. 2015;61(1):221-230. Memczak S, Papavasileiou P, Peters O, Rajewsky N. Identification and characterization of circular RNAs as a new class of putative biomarkers in human blood. PloS One. 2015;10(10):e0141214.

2047

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Acute Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2017 Volume 102(12):2048-2057

Anexelekto /MER tyrosine kinase inhibitor ONO-7475 arrests growth and kills FMS-like tyrosine kinase 3-internal tandem duplication mutant acute myeloid leukemia cells by diverse mechanisms Peter P. Ruvolo,1,2 Huaxian Ma,1 Vivian R. Ruvolo,1,2 Xiaorui Zhang,1 Hong Mu,1,2 Wendy Schober,1,2 Ivonne Hernandez,1,2 Miguel Gallardo,1 Joseph D. Khoury,3 Jorge Cortes,1, Michael Andreeff1,2 and Sean M. Post1

1 Department of Leukemia; 2Section of Molecular Hematology and 3Department of Hematopathology, The University of Texas MD Anderson Cancer Center, Houston, TX, USA

ABSTRACT

Correspondence: pruvolo@mdanderson.org

Received: March 13, 2017. Accepted: September 7, 2017. Pre-published: September 14, 2017. doi:10.3324/haematol.2017.168856 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2048

early one-third of patients with acute myeloid leukemia have FMS-like tyrosine kinase 3 mutations and thus have poor survival prospects. Receptor tyrosine kinase anexelekto is critical for FMSlike tyrosine kinase 3 signaling and participates in FMS-like tyrosine kinase 3 inhibitor resistance mechanisms. Thus, strategies targeting anexelekto could prove useful for acute myeloid leukemia therapy. ONO-7475 is an inhibitor with high specificity for anexelekto and MER tyrosine kinase. Herein, we report that ONO-7475 potently arrested growth and induced apoptosis in acute myeloid leukemia with internal tandem duplication mutation of FMS-like tyrosine kinase 3. MER tyrosine kinase-lacking MOLM13 cells were sensitive to ONO-7475, while MER tyrosine kinase expressing OCI-AML3 cells were resistant, suggesting that the drug acts via anexelekto in acute myeloid leukemia cells. Reverse phase protein analysis of ONO-7475 treated cells revealed that cell cycle regulators like cyclin dependent kinase 1, cyclin B1, polo-like kinase 1, and retinoblastoma were suppressed. ONO-7475 suppressed cyclin dependent kinase 1, cyclin B1, polo-like kinase 1 gene expression suggesting that anexelekto may regulate the cell cycle, at least in part, via transcriptional mechanisms. Importantly, ONO-7475 was effective in a human FMS-like tyrosine kinase 3 with internal tandem duplication mutant murine xenograft model. Mice fed a diet containing ONO-7475 exhibited significantly longer survival and, interestingly, blocked leukemia cell infiltration in the liver. In summary, ONO-7475 effectively kills acute myeloid leukemia cells in vitro and in vivo by mechanisms that involve disruption of diverse survival and proliferation pathways. Introduction

ÂŠ2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2048

Anexelekto (AXL) is a receptor tyrosine kinase (RTK) of the TAM (Tyro3, AXL, and MER) family.1-4 Activation of AXL by growth arrest specific 6 (GAS6) induces diverse survival cascades.1-4 Recent studies have revealed that AXL regulates survival signaling in many cancers, including leukemia.1-5 High expression of AXL or GAS6 in AML patients is prognostic for poor survival outcome.6-10 The AXL/GAS6 axis promotes leukemia cell proliferation and chemoresistance.7 AXL is a key regulator of myeloid cell differentiation.2 FMS-like tyrosine kinase (FLT3) regulates leukemia cell differentiation.11-13 FLT3 mediated effects on differentiation may be mediated by AXL. Bone marrow (BM) derived mesenchymal stromal cells (MSC) protect leukemia cells from chemotherapy.14-17 FLT3 mutations involving internal tandem duplication (ITD) or point mutations (e.g., D835) are found in >30% of AML patients and are associated with poor survival.18 Furthermore, though FLT3 inhibitors are in the clinic, these agents have shown limited effectiveness, with haematologica | 2017; 102(12)

Targeting AXL for therapy of FLT3-ITD AML

acquired resistance being a major problem.18-20 AXL positively regulates signaling mediated by mutant FLT3 protein.7-10 Thus, disruption of AXL in cells with mutant FLT3 may disrupt signaling controlled by the mutant kinase. ONO-7475 is a novel TAM inhibitor that targets AXL in the nM range.21 In the report herein, we examine the efficacy of ONO-7475 in AML cells in both in vitro and in vivo models.

Methods Cell lines and cell culture MOLM13 were purchased from DSMZ (Braunschweig, Germany). MV4;11 and HL60 were purchased from ATCC (Manassas, VA, USA). OCI-AML3 were kindly provided by Mark Minden (Ontario Cancer Institute, Toronto, ON, Canada). MOLM13 luciferase (luc)/ green fluorescent protein (gfp) cells were generated using lentiviral transduction. The lentiviral plasmid was generated by cloning the firefly luciferase sequence excised from pGL4.51 (Promega, Madison, WI, USA) into the pCDH-CMV-MCS-EF1-copGFP lentiviral vector (System Biosciences Inc., Mountain View, CA, USA). MOLM13 p53 short hairpin ribonucleic acid (shRNA) cells were previously described.22 BM-MSC were acquired in accordance with regulations and protocols approved by the Investigational Review Board of the University of Texas MD Anderson Cancer Center (MDACC). Informed consent was obtained in accordance with the Declaration of Helsinki.

Reagents ONO-7475, mouse feed containing ONO-7475, and control mouse feed were supplied by Ono Pharmaceutical Co. Ltd. (Osaka, Japan). Cytarabine (AraC) was purchased from LC Laboratories (Woburn, MA, USA). Stock solutions were prepared with dimethyl sulfoxide (DMSO; Sigma-Aldrich, St. Louis, MO, USA).

RPPA Reverse phase protein analysis (RPPA) was performed by the RPPA Core at the University of Texas MDACC as described in the Online Supplementary Methods.

DNA synthesis and cell cycle assay Cells were incubated with vehicle or ONO-7475 and then analyzed for their rates of DNA synthesis and cell cycle status using the Alexa Fluor 444 Click-It EdU Flow Cytometry assay kit from Molecular Probes (Carlsbad, CA, USA) per manufacturerâ&#x20AC;&#x2122;s instructions.

Gene expression analysis Real-time polymerase chain reaction (qRT-PCR) was performed with an ABI 7900HT Fast Real-Time PCR System (Life Technologies, Carlsbad, CA, USA), using TaqMan assays listed in the Online Supplementary Methods, as directed by the manufacturer. RQ Manager 1.2.1 (Life Technologies) was used for data analysis.

Immunoblot analysis Cells were incubated with vehicle or ONO-7475 and then lysed, and total protein was fractionated by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS/PAGE) electrophoresis. Immunoblot analysis was performed with antibodies listed in the Online Supplementary Methods. Signals were detected with Odyssey Infrared Imaging System and quantitated by Odyssey software version 3.0 (both LI-COR Biosciences, Lincoln, NE, USA). Tubulin was used as a loading control. haematologica | 2017; 102(12)

Table 1. Proteins affected by ONO-7475 in MOLM13 and MV4;11 cells as identified by RPPA . Proteins reduced in MOLM13 cells by ONO-7475 (bold indicates reduction in MV4;11 cells as well) ACCA (phospho; pS79) CAV1 CDK1 (Total), CHEK2 (phospho; pT68), CCNB1 EEF2 FASN HSP27 (phospho; pS82) MTOR (phospho; pS2448) NDRG1 (phospho; pT346) MAPK14 (phospho; pT180/pY182) PLK1 (Total) RB1 (phospho; pS807/pS811) RPS6 (phospho; pS235/S236) RPS6 (phospho; pS240/S244) SLC1A5 TFRC Proteins increased in MOLM13 cells by ONO-7475 (bold indicates increase in MV4;11 cells as well) AXL CASP3 H3F3A/B (demethylated K9) H2AX (phospho; pS140)

Pathway analysis String software (String 10.0; website: http://string-db.org)23 was used to determine protein associations.

Human AML xenograft in vivo model Human xenograft experiments were approved by the Institutional Animal Care and Use Committee at the University of Texas MDACC and are described in the Online Supplementary Methods.

Immunohistochemistry Immunohistochemistry (IHC) was performed on paraffin embedded spleen sections using cleaved Caspase 3 (Asp 175) rabbit antibody from Cell Signaling (Danvers, MA, USA) and Vectastain ABC AP kit (Vector Laboratories; Burlingame, CA, USA).

Pathologic analysis Immunohistochemistry (IHC) was performed on paraffin embedded spleen sections using cleaved Caspase 3 (Asp 175) rabbit antibody from Cell Signaling (Danvers, MA, USA) and Vectastain ABC AP kit (Vector Laboratories; Burlingame, CA, USA).

Results AXL inhibitor ONO-7475 kills or growth arrests FLT3-ITD AML cells MOLM13 and MV4;11 (FLT3 ITD) and OCI-AML3 (FLT3 WT) cells were incubated with vehicle (0.1% 2049

P.P. Ruvolo et al.

DMSO) or varying doses of ONO-7475 for 48 hours. Viable cell number and apoptosis were measured by flow cytometry. ONO-7475 slightly reduced OCI-AML3 viable cell number and did not induce apoptosis (Figure 1A, Online Supplementary Figure S1A). FLT3 WT HL60 cells were also resistant to ONO-7475 (data not shown). ONO7475 was effective against MOLM13 and MV4;11 cells, both of which have FLT3-ITD. For these cells, 10 nM ONO-7475 significantly reduced cell viability by >50% (P<0.0001; Figure 1A) and induced apoptosis (P<0.00001; Online Supplementary Figure S1A). Cell based tyrosine kinase inhibition assays revealed ONO-7475 is 210-fold more selective for AXL compared to FLT3 (Online Supplementary Table S1). While IC50 for 48-hour treatment of drug was 0.7 nM for AXL and 1.0 nM for MER, the IC50 for FLT3 was 147 nM (Online Supplementary Table S1). As apoptosis was significantly induced in both MOLM13 and MV4;11 cells after 48-hour treatment with a dose (10 nM; Online Supplementary Figure S1B) of drug well below the 48 hour IC50 of ONO-7475 for FLT3 (147 nM), this result suggested that the drug action is independent of FLT3. After 72 hours, 10 nM ONO-7475 nearly eliminated all viable MOLM13 cells (Figure 1B) with >70% of cells becoming apoptotic (Online Supplementary Figure S1B). BM derived MSC protect leukemia cells from chemotherapy.14-17 To simulate the effect of the BM microenvironment during chemotherapy, we tested the efficacy of ONO-7475 using an in vitro co-culture system. MOLM13 cells in monoculture or in co-culture with MSC were incubated for 72 hours with vehicle or ONO-7475. The drug potently induced apoptosis and nearly eliminated the AML cells in monoculture (Figure 1B, Online Supplementary Figure S1B). While MSC protected the AML cells from the inhibitor, treatment with a higher dose of drug abrogated most of this effect (50 nM ONO-7475, Figure 1B). MOLM13 and MV4;11 cells were incubated with vehicle or varying doses of drug for 24 hours, and their cell cycle status and rates of DNA synthesis were assessed by flow cytometry using FX Cycle Violet staining and EdU incorporation. Low dose ONO-7475 (5 nM) suppressed DNA synthesis in both cell lines by >2-fold (Figure 1C). ONO-7475 potently induced G0/G1 arrest in both cell lines (Online Supplementary Figure S2A,B). AraC as well as p53 reduction augments ONO-7475induced apoptosis in MOLM13 cells. MOLM13 cells were incubated with vehicle (0.2% DMSO), 1 mM AraC alone, ONO-7475 alone (10nM or 50 nM), or a combination of the drugs for 48 hours. While AraC and ONO-7475 were effective as single agents, the combination reduced viable cell numbers (Figure 2A) and induced apoptosis (Online Supplementary Figure S3A) to a greater extent. Both MOLM13 and MV4-11 cells have wild-type p53. To determine whether the drugâ&#x20AC;&#x2122;s mechanism of action involves p53, we incubated MOLM13 control shRNA (GFP as non-human shRNA target) or MOLM13 p53 shRNA cells for 48 hours with varying doses of ONO7475 for 48 hours. Expression of p53 in cells with p53 shRNA was ~30% of that of cells with GFP control shRNA (Figure 2B). Surprisingly, the p53 knockdown cells were more sensitive to the inhibitor than the control cells. A dose of 10 nM ONO-7475 reduced the viability of control cells by ~55% (Figure 2B) without inducing apoptosis (Online Supplementary Figure S2B), whilst, however, reducing the viability of knockdown cells by >70% (Figure 2B) and inducing some apoptosis (~20%; Online Supplementary 2050

Figure S2B). Only at doses >100 nM were the control and knockdown cells similarly sensitive (Figures 2B, Online Supplementary Figure S3B), suggesting a pro-survival role for p53 in this scenario in AXL-regulated cascades.

Inhibition of AXL suppresses survival and proliferation signaling and induces apoptotic proteins in FLT3-ITD but not WT FLT3 AML cells RPPA was performed by the MDACC Proteomics Core

Figure 1. ONO-7475 effectively reduces viable FLT3-ITD AML cells even in the presence of MSC. (A) Cells were incubated with varying doses of ONO-7475 and cell viability was assessed by flow cytometry. (B) MOLM13 cells were co-cultured with MSC, incubated with ONO-7475 and cell viability was assessed by flow cytometry. (C) DNA synthesis and cell cycle were assessed by EdU Click-It assay in vehicle treated cells and cells treated with ONO-7475. Student's t-test was performed against vehicle treated cells (*P<0.05; ***P<0.001). DMSO: dimethyl sulfoxide.

haematologica | 2017; 102(12)

Targeting AXL for therapy of FLT3-ITD AML

to analyze MOLM13, MV4:11, OCI-AML3, and HL60 cells that were treated with vehicle or ONO-7475 for 24 hours. Proteins surveyed by RPPA are listed in Online Supplementary Table S2. Trypan blue staining of collected cells revealed that cells were >95% viable in all cases, except that of MOLM13 cells treated with 100 nM ONO7475, which were ~88% viable. There was no significant change (i.e., 2-fold) in the level of protein observed in either FLT3 WT cell line (OCI-AML3 or HL60; Online Supplementary Figure S4A). Protein changes of at least 2fold were most prominent in MOLM13 cells, with some of the same proteins showing similar changes in MV4;11 cells (Online Supplementary Figure S4A-C; Online Supplementary Figure S5A,B). Proteins suppressed in MOLM13 and MV4;11 cells are listed in Table 1. RPPA identified a number of molecules involved in signal transduction and cell cycle regulation. Only four proteins were induced by ONO-7475 in MOLM13 cells, however, one of these was AXL, which was also induced in MV4;11 cells (Online Supplementary Figure S5A,B; Table 1). Interestingly,

total AXL protein is induced by the drug in both cell lines, suggesting the presence of a feedback loop involving the activity of the protein and its level of expression. Protein association network analysis was performed using STRING 10.0 23 on proteins identified as suppressed in MOLM13 cells (Figure 3A). The analysis suggests strong interconnection between various proteins that are distinctly suppressed by the drug, with only ACACA, NDRG1, and SLC1A5 showing no interconnection. There appears to be strong cross-talk between proteins involved in the cell cycle (Figure 3A). The effects of AXL on PLK1, RB, or CDK1 in leukemia cells have yet to be reported. PLK1 is regulated in part via transcriptional mechanisms.24-27 Figure 3B depicts a model of AXL regulation of PLK1. ONO-7475 suppression of CDK1 and Cyclin B1 could inhibit PLK1 expression via the activation of RB. Immunoblot analysis confirmed the effect of ONO-7475 on RPPA candidate target proteins, including PLK1, p-RB, p-S6 ribosomal protein, p-p38, and AXL (Figure 4A,B). Consistent with RPPA results (Online

haematologica | 2017; 102(12)

Figure 2. ONO-7475 augments AraCinduced killing in FLT3-ITD AML cell lines though suppression of p53 sensitizes cells to the drug. (A) Cells were incubated with varying doses of ONO-7475 and/or 1 mM AraC for 48 hours and cell viability assessed by flow cytometry. (B) MOLM13 control cells or MOLM13 p53 shRNA were incubated with ONO-7475 for 48 hours and then cell viability assessed by flow cytometry. Protein expression of p53 and Tubulin in MOLM13 GFP shRNA and MOLM13 p53 shRNA cells as determined by immunoblot is imbedded in (B). Ratio of p53 to Tubulin relative to GFP shRNA control was determined by densitometry of bands using LiCore imager. Student's t-test was performed against AraC treated (*P<0.05; ***P<0.001) for (A) and against GFP control (**P<0.01; ***P<0.001) for (B). NS: not significant; DMSO: dimethyl sulfoxide; AraC: cytarabine; GFP: green fluorescent protein; shRNA: short hairpin ribonucleic acid.

2051

P.P. Ruvolo et al.

Supplementary Figure S4B,C), immunoblot analysis demonstrated that the AXL inhibitor potently suppressed PLK-1 in both cell lines (Figure 4A). Little, if any, PLK1 was detected when either were treated with 50 nM or 100 nM drug. In addition, as predicted by RPPA (Online Supplementary Figure S5A,B), ONO-7475 induced AXL and blocked RB phosphorylation in both cell lines, even at the lowest dose (Figure 4A). The drug reduced total RB in MOLM13 cells, and, to a lesser extent, in MV4;11 cells (Figure 4A). As shown in Figure 4C, ONO-7475 potently suppressed both Cyclin B1 and CDK1 in both cell lines even at a low dose (10 nM) ONO-7475. Immunoblot validation of the suppression of Cyclin B1, CDK1, PLK1 and

the inhibition of RB phosphorylation supports the model proposed in Figure 3B. In MOLM13 cells, the AXL inhibitor inhibited the phosphorylation of S6, and p38 phosphorylation was reduced due to reduction of total p38 (Figure 4B). Though not detected by RPPA, p-ERK and MCL-1 were examined by immunoblot. AXL is known to induce ERK,2,3,28 and ERK is a positive regulator of MCL-1.29,30 ONO-7475 reduced ERK phosphorylation (~50% at 10 nM; Figure 4B) in MOLM13 cells. ONO-7475 suppressed MCL-1 (>80% at 10 nM) in MOLM13 cells but was less effective in MV4;11 cells (50% reduction of MCL-1 at 100 nM; Figure 4A). AXL inhibition by siRNA inhibits FLT3 phosphoryla-

Figure 3. Proteins identified by RPPA as suppressed by ONO-7475 suggest common pathways and a model of cell cycle blockade involving CDK1/CCNB1, RB, and PLK1. String analysis was performed on proteins suppressed by >2-fold according to RPPA. Network is depicted in (A). A model hypothesizing ONO-7475 blocks CDK1/CCNB1 to inhibit RB phosphorylation resulting in the activation of RB, which in turn suppresses PLK1, is depicted in (B).

2052

haematologica | 2017; 102(12)

Targeting AXL for therapy of FLT3-ITD AML

tion.9 Treatment with ONO-7475 reduced FLT3 phosphorylation in both MOLM13 and MV4;11 cells (Online Supplementary Figure S6). An ONO-7475 dose of 10 nM greatly reduced p-FLT3 in MOLM13 cells and nearly eliminated it in MV4;11 cells, suggesting that AXL inhibition may prevent FLT3 activation (Online Supplementary Figure S6). To determine whether AXL modulated its target genes at the transcriptional level, we incubated MOLM13 and MV4;11 cells with vehicle or drug for 24 hours, then extracted total RNA, reverse transcribed it and performed qRT-PCR to measure the abundance of transcripts for PLK-1, CDK1, CCNB1 and B2M. ONO-7475 reduced the messenger (m)RNA of each cell cycle regulator relative to that of B2M in both cell lines. A dose of 10 nM ONO-7475 reduced PLK-1 mRNA ~60% and 50 nM drug reduced PLK-1 mRNA by ~80% in MOLM13 cells (Figure 5A). Both doses more potently suppressed PLK-1 in MV4;11 cells (Figure 5A). The drug reduced CDK1 (Figure 5B) and CCNB1 (Figure 5C) in both cell lines by at least 60% at 10 nM dose. These data suggest that AXL positively regulates transcription of these cell cycle regulators in FLT3-ITD

AML cells. Moreover, we found that ONO-7475 promoted an increase in the level of the AXL transcript. In both cell lines AXL mRNA relative to that of B2M was increased by ~3.5-fold with 10 nM drug and by >6-fold in MOLM13 cells and >8-fold in MV4;11 cells with 50 nM drug after 24 hours (Online Supplementary Figure S7). These results suggest a feedback mechanism whereby AXL inhibition induces AXL transcription.

ONO-7475 is effective in a murine in vivo xenograft model using MOLM13 cells To test the efficacy of ONO-7475 in a human AML xenograft model, the initial experiment utilized a relatively low dose of drug (i.e., 6 mg/kg) which was introduced into mouse food. The manufacturer estimates 0.004% is roughly 6 mg/kg daily consumption of drug (T. Yasuhiro and T. Yoshizawa, Ono Pharmaceutical Co. LTD, personal communication). MOLM13 luc/gfp were injected into NOD scid Îł (NSG) mice. Mice were given control feed for one week and then split into groups that continued control feed (N=5) or feed with 0.004% ONO-7475 (N=5). Mice given feed with ONO-7475 exhibited significantly

Figure 4. ONO-7475 suppresses various positive regulators of survival and cell cycle in FLT3-ITD AML cell lines. MOLM13 and MV4;11 were incubated with varying doses of ONO-7475 for 24 hours. Immunoblot analysis was performed as described in Methods. Ratio of protein signal to that of Tubulin loading control or to total protein for phospho-protein was determined by densitometry using LiCor imager. Veh: vehicle.

haematologica | 2017; 102(12)

2053

P.P. Ruvolo et al.

(P=0.0047) longer median survival compared to mice given control feed (Figure 6A, Online Supplementary Table S3). In vivo imaging system (IVIS) imaging was performed weekly to determine the leukemic burden. On Day 14, three out of five of the mice given food containing ONO7475 exhibited reduced leukemic burden compared to the mice which were fed control food (Figure 6B). Sternum (for BM) and liver were collected from a moribund mouse from the control group and a moribund mouse from the 0.004% ONO-7475 group and H&E staining was performed (Online Supplementary Figure S8A) As shown in Online Supplementary Figure S8A, there were fewer mitotic bodies (red arrows) and more apoptotic cells (yellow arrows) observed in tissues of the mouse fed drug compared to the mouse fed normal diet. Spleen was collected from these animals for IHC staining of cleaved Caspase 3 to visualize apoptosis (Online Supplementary Figure S8B). There were more cells exhibiting cleaved Caspase 3 in the spleen of the mouse that was fed drug. Next, efficacy of a higher dose (i.e., 0.013%, which is roughly 20 mg/kg daily) of ONO-7475 was tested. MOLM13 luc/gfp cells were introduced into NSG mice. Mice were fed control feed for one week and then split into groups that continued control feed (N=9) or feed containing 0.013% ONO7475 (N=10). Mice fed 0.013% ONO-7475 exhibited significantly (P=0.0062) longer median survival compared to mice given control feed (Online Supplementary Figure S9, Online Supplementary Table S3). Sternum (for BM) and liver were collected when mice became moribund, and H&E staining was performed (Figure 7A). Mitotic bodies and apoptotic cells were visualized in samples. While leukemia cells were present in samples from both control and ONO-7475 fed mice, mitotic bodies (red arrows) were higher in control mice while apoptotic cells (yellow arrows) were more prominent in leukemia cells in BM from mice fed the drug (Figure 7A). To quantitate apoptotic cells, IHC using antibody to cleaved Caspase 3 was performed on spleen sections. More apoptotic cells were observed in spleen sections of mice given drug compared to control mice (Figure 7B). Apoptotic cells and total AML cells were counted and the percentage determined for each set of mice. There were significantly more apoptotic cells in spleen sections of mice fed ONO-7475 compared to mice given control food (>2-fold; P=0.010; Figure 7C). An interesting effect of the drug in the in vivo model involves leukemia infiltration in the liver. Mice fed ONO7475 showed little, if any, AML cells in the liver, whereas AML cells infiltrated the livers in control mice (Figure 7A).

extent, in HL60 cells, but not in OCI-AML3 cells (Online Supplementary Figure S10B). Short term treatment of recombinant human GAS6 (500 ng/ml for one hour) did not induce AXL in the OCI-AML3 cells (Online Supplementary Figure S10B). These data suggest that the effects of the drug in these AML cells are mediated only through AXL. The microenvironment is central to the role of AXL in drug resistance.7-10 In our in vitro model of the leukemia microenvironment, MSC protected MOLM13 cells from the inhibitor, though this effect was overcome with a

Discussion

In this study, we found FLT3 WT OCI-AML3 and HL60 cells were resistant to ONO-7475, while FLT3-ITD cell lines MOLM13 and MV4;11 were sensitive (Figure 1A and Online Supplementary Figure S1A). ONO-7475 induced apoptosis in MOLM13 and MV4;11 cells. At low dose (5 nM), ONO-7475 effectively growth arrests both cell lines (Online Supplementary Figure S2A,B). ONO-7475 is a TAM kinase inhibitor, so a role for MER in ONO-7475-induced effects is possible. MER is aberrantly expressed in AML.31 We tested MER expression in the AML cell lines and found it to be abundant in OCI-AML3, minimal in MV4;11, and undetectable in MOLM13 cells even after ONO-7475 treatment (Online Supplementary Figure S10A). AXL was detectable in MOLM13 cells, and, to a lesser

Figure 5. ONO-7475 suppresses PLK1, CDK1, and CCNB1 in MOLM13 and MV4;11 cells. MOLM13 and MV4;11 were incubated with varying doses of ONO7475 for 24 hours. RNA from cells was extracted and reverse transcribed, and the abundance of the transcripts for PLK-1 (A), CDK1 (B), and CCNB1 (C), and B2M (A-C) were determined by qRT-PCR. Gene expression levels were normalized to B2M. DMSO: dimethyl sulfoxide.

2054

haematologica | 2017; 102(12)

Targeting AXL for therapy of FLT3-ITD AML

higher drug dose (Figure 1B and Online Supplementary Figure S1B). AXL mediates drug resistance involving a number of RTK in diverse cancers.1-5,7,9,10 ONO-7475 suppressed FLT3 phosphorylation (Online Supplementary Figure S6), which is consistent with AXL as a positive regulator of mutant FLT3.9 RPPA revealed that a distinct set of proteins was altered by ONO-7475 in FLT3-ITD but not FLT3 WT cells (Online Supplementary Figure S2A). ONO-7475 reduced p-ERK and MCL-1 in MOLM13 cells even at a low dose (Figures 4A,B). MCL-1 is critical for AML cell survival.31,32 FLT3-ITD induces MCL-1 expression in AML stem cells,32 so perhaps MCL-1 is central for AXL-mediated survival in FLT3 ITD cells. The mechanism by which AXL affects proliferation in FLT3 ITD cells is unclear.10 ONO-7475 suppresses DNA synthesis (Figure 1C, Online Supplementary Figure S2A,B) and potently suppresses PLK1 (Figure 4A and Figure 5A, respectively). As PLK-1 is

required for mitotic entry in cancer cells,24-27 this may account for the blockade of DNA synthesis by the drug. Although ONO-7475 more potently induces apoptosis in MOLM13 cells (Online Supplementary Figure S1B), it acts to a greater extent in MV 4;11 cells by inducing growth arrest (Online Supplementary Figure S2A,B). These findings suggest that the physiologic response to the drug will depend on the nature of the pathways affected. Consistent with a model where ONO-7475 activation of RB suppresses PLK1 to growth arrest cells (Figure 3C), the drug more potently blocked RB phosphorylation and PLK-1 expression in MV4;11 cells compared to MOLM13 cells (Figure 4A). ONO-7475 blocks Cyclin B1 and CDK1 expression, suggesting a possible mechanism by which the drug blocks RB phosphorylation. Wilson and colleagues observed that AXL suppression inhibits CDK1 phosphorylation, but they did not observe a reduction of CDK1 expression.33 In

Figure 6. ONO-7475 prolongs survival of mice bearing MOLM13 leukemia cells. MOLM13 luc/gfp cells were injected into NSG mice, and mice received either control diet or diet containing 0.004% ONO7475. (A) Survival analysis of mice fed 0.004% drug using GraphPad software. (B) Imaging of mice fed control feed or feed with 0.004% ONO-7475 taken at Day 7 and Day 14.

haematologica | 2017; 102(12)

2055

P.P. Ruvolo et al.

the study herein, gene expression of both CDK1 (Figure 5B) and CCNB1 (Figure 5C) were suppressed by the drug as observed by qRT-PCR. The possible regulation by AXL of the CDK1/RB/PLK1 axis is novel and suggests that AXL has an important function in regulating cell cycle in FLT3ITD AML cells. Consistent with a role for AXL in cell cycle regulation, Kariolis et al. found that blockade of AXL in AML cells with AXL decoy receptor MYD1-72 was associated with cell cycle arrest.34 Importantly, MYD1-72 killed FLT3 ITD AML cells or GAS6 exposed FLT3 WT cells, suggesting that AXL is a viable target for AML therapy.34 Currently, a number of agents that can inhibit AXL such as gilteritinib (ASP2215) and BGB324 (R48) are in clinical trials for cancers, including AML.35,36 The results of the trial using gilteritinib as a single agent for therapy for refractory or relapse AML were promising. The agent was well tolerated; 90% of patients exhibited inhibition of FLT3 phosphorylation and 40% achieved some response.36 Preclinical studies have shown that gilterinib is effective in in vitro and in vivo models of AML.37,38 Gilteritinib targets both FLT3 and AXL with IC50 is in the nM range (0.29 nM and 0.7 nM, respectively).35,37 BGB324 is another promising agent that targets AXL.35 A recent study found that epigenetic modifying agents induced AXL, and BGB324 could sensitize resistant cells to a combination of decitabine and vorinostat.39 Thus, AXL inhibition may be beneficial at many levels for AML therapy. ONO-7475 is over 200-fold more selective for AXL compared to FLT3 (Online Supplementary Table S1), consequently, perhaps resistance mechanisms that are observed with other FLT3 inhibitors might be avoided with this agent. We are currently testing ONO-7475 in combination with traditional FLT3 inhibitors to test this possibility. ONO-7475 was able to synergize with AraC (Figure 2A and Online Supplementary Figure S3A), thus, the AXL inhibitor could be combined with standard chemotherapy. We examined if ONO-7475 affected the level of p53 and found little effect (data not shown), albeit we did not examine p53 cellular localization or phosphorylation status. We tested the effect of ONO-7475 on MOLM13 cells with suppressed p53 expression. Interestingly, knockdown of p53 sensitized the cells to low doses of the drug (Figure 2B and Online Supplementary Figure S3B). FLT3 mutations and p53 mutations appear to be mutually exclusive in AML;40 perhaps p53 has some survival function in FLT3 mutated cells, even if the precise mechanism remains unclear. The in vivo study using ONO-7475 in an AML xenograft model with MOLM13 luc/gfp cells in NSG mice suggests the drug is efficacious for AML therapy, at least in FLT3ITD AML. Both low dose and high dose ONO-7475 significantly prolonged the survival of mice bearing MOLM13 cells (Figure 6A and Online Supplementary Figure S8; Online Supplementary Table S2). Increased apoptosis in leukemia cells in treated mice was verified by IHC examining Cleaved Caspase 3 levels. There was >2-fold more apoptotic cells in the spleens of mice fed ONO-7475 compared to mice given control feed (Figures 7B,C). Interestingly, mice on control feed showed moderate to severe infiltration of AML cells in the liver, whereas mice fed high dose ONO-7475 showed minimal presence of MOLM13 cells in the liver (Figure 7A). The in vivo data are encouraging as the drug supports prolonged survival and may have beneficial effects in leukemic infiltration in the liver. In summary, ONO-7475 potently arrests growth and 2056

induces apoptosis of FLT3-ITD AML cells. The drug acts via diverse mechanisms, including suppression of the CDK1/RB/PLK1 axis. Importantly, the drug prolonged mouse survival and suppressed AML cell infiltration in the liver in the AML xenograft model. These combined results suggest that ONO-7475 is efficacious for AML therapy and warrants pursuit for development of the drug in the clinic.

Figure 7. ONO-7475 induces apoptosis of MOLM13 leukemia cells in vivo and inhibits metastasis to the liver. MOLM13 luc/gfp cells were introduced into NSG mice and mice received either control feed or feed containing 0.013% ONO-7475. (A) H&E stain of BM and liver tissue. Mitotic cells (red arrows) and apoptotic cells (yellow arrows) are indicated. (B) Representative IHC using antibody against cleaved Caspase 3 to identify apoptotic cells in spleen. (C) Total apoptotic cells counted in spleen IHC samples probed with antibody against cleaved Caspase 3 from mice fed control diet (N=3) or 0.013% ONO-7475 (N=3). Statistical significance was determined by Student's t-test (*P<0.05).

haematologica | 2017; 102(12)

Targeting AXL for therapy of FLT3-ITD AML

Acknowledgments We thank Tomoko Yasuhiro, Kohei Tanaka, and Toshio Yoshizawa from Ono Pharmaceutical Company Ltd., Osaka, Japan for providing drug and technical advice for these studies.

References 1 Lemke G. Biology of the TAM receptors. Cold Spring Harb. Perspect Biol. 2013;5(11):a009076. 2 Axelrod H, Pienta KJ. Axl as a mediator of cellular growth and survival. Oncotarget. 2014;5(19):8818-8852. 3 Paccez JD, Vogelsang M, Parker MI, Zerbini LF. The receptor tyrosine kinase Axl in cancer: biological functions and therapeutic implications. Int J Cancer. 2014;134(5):1024-1033. 4 Brown M, Black JR, Sharma R, Stebbing J, Pinato DJ. Gene of the month: Axl J Clin Pathol. 2016;69(5):391-397. 5 Brandao L, Migdall-Wilson J, Eisenman K, Graham DK. TAM receptors in leukemia: expression, signaling, and therapeutic implications. Crit Rev Oncog. 2011;16(12):47-63. 6 Rochlitz C, Lohri A, Bacchi M, et al. Axl expression is associated with adverse prognosis and with expression of Bcl-2 and CD34 in de novo acute myeloid leukemia (AML):results from a multicenter trial of the Swiss Group for Clinical Cancer Research (SAKK). Leukemia. 1999; 13(9):1352-1358. 7 Ben-Batalla I, Schultze A, Wroblewski M, et al. Axl, a prognostic and therapeutic target in acute myeloid leukemia mediates paracrine crosstalk of leukemia cells with bone marrow stroma. Blood. 2013; 122(14):2443-2452. 8 Whitman SP, Kohlschmidt J, Maharry K, et al. GAS6 expression identifies high-risk adult AML patients: potential implications for therapy. Leukemia. 2014;28(6):12521258. 9 Park IK, Mishra A, Chandler J, et al. Inhibition of the receptor tyrosine kinase Axl impedes activation of the FLT3 internal tandem duplication in human acute myeloid leukemia: implications for Axl as a potential therapeutic target. Blood. 2013; 121(11):2064-2073. 10 Park IK, Mundy-Bosse B, Whitman SP, et al. Receptor tyrosine kinase Axl is required for resistance of leukemic cells to FLT3-targeted therapy in acute myeloid leukemia. Leukemia. 2015;29(12):2382-2389. 11 Sexauer A, Perl A, Yang X, et al. Terminal myeloid differentiation in vivo is induced by FLT3 inhibition in FLT3/ITD AML. Blood. 2012;120(20):4205-4214. 12 Hirade T, Abe M, Onishi C, et al. Internal tandem duplication of FLT3 deregulates proliferation and differentiation and confers resistance to the FLT3 inhibitor AC220 by Up-regulating RUNX1 expression in hematopoietic cells. Int. J. Hematol. 2016;103(1):95-106. 13 Dormady SP, Zhang XM, Basch RS. Hematopoietic progenitor cells grow on

haematologica | 2017; 102(12)

20 21

Funding This work was supported in part by funds from Ono Pharmaceutical Company Ltd. (PR). MDACC Flow Core (MA) and RPPA Core are funded by the National Institutes of Health (CA-016672).

3T3 fibroblast monolayers that overexpress growth arrest-specific gene-6 (GAS6). Proc Natl Acad Sci USA. 2000;97(22): 12260-12265. Konopleva M, Konoplev S, Hu W, et al. Stroma cells prevent apoptosis of AML cells by upregulation of anti-apoptotic proteins. Leukemia. 2002;16(9):1713-1724. Tabe Y, Konopleva M, Munsell MF, et al. PML-RARa is associated with leptin-receptor induction: the role of mesenchymal stem cell-derived adipocytes in APL cell survival. Blood. 2004;103(5):1815-1822. Konopleva M, Tabe Y, Zeng Z, Andreeff M. Therapeutic targeting of microenvironmental interactions in leukemia: mechanisms and approaches. Drug Resist Updat. 2009;12(4-5):103-113. Konopleva MY, Jordan CT. Leukemia stem cells and microenvironment: biology and therapeutic targeting. J Clin Oncol. 2011;29(5):591-599. Grunewald MR, Levis MJ. FLT3 FLT3 tyrosine kinase inhibition as a paradigm for targeted drug development in acute myeloid leukemia. Semin Hematol. 2015;52(3):193199. Daver N, Cortes J, Ravandi F, et al. Secondary mutations as mediators of resistance to targeted therapy in leukemia. Blood. 2015;125(21):3236-3245. Konig H, Levis M. Targeting FLT3 to treat leukemia. Expert Opin Ther Targets. 2015; 19(1):37-54. Yasuhiro T, Yoshizawa T, Fujikawa F, et al. Development of an Axl/Mer dual inhibitor, ONO-9330547: promising single agent activity in an acute myeloid leukemia (AML) model. Blood. 2014;124(21):999. Kojima K, Kornblau SM, Ruvolo V, et al. Prognostic impact and targeting of CRM1 in acute myeloid leukemia. Blood. 2013; 121(20):4166-4174. Szklarczyk D, Franceschini A, Wyder S, et al. STRING v10: protein-protein interaction networks, integrated over the tree of life. Nucleic Acids Res. 2015;43 (Database Issue):D447-52. Cholewa BD, Liu X, Ahmad N. The role of polo-like kinase 1 in carcinogenesis: cause or consequence? Cancer Res. 2013;73(23):6848-6855. Weng Ng WT, Shin JS, Roberts TL, Wang B, Lee CS. Molecular interactions of polo-like kinase 1 in human cancers. J Clin Pathol. 2016;69(7):557-562. Talati C, Griffiths EA, Wetzler M, Wang ES. Polo-like kinase inhibitors in hematologic malignancies. Crit Rev Oncol Hematol. 2016;98:200-210. Louwen F, Yuan J. Battle of the eternal rivals: restoring functional p53 and inhibiting Polo-like kinase 1 as cancer therapy. Oncotarget. 2013;4(7):958-971. Goruppi S, Ruaro E, Varnum B, Schneider C. Gas6-mediated survival in NIH3T3 cells

activates stress signalling cascade and is independent of Ras. Oncogene. 1999;18 (29):4224-4236. Schubert KM, Duronio V. Distinct roles for extracellular-signal-regulated protein kinase (ERK) mitogen-activated protein kinases and phosphatidylinositol 3-kinase in the regulation of Mcl-1 synthesis. Biochem. J. 2001;356(Pt 2):473-480. Domina AM, Vrana JA, Gregory MA, Hann SR, Craig RW. MCL1 is phosphorylated in the PEST region and stabilized upon ERK activation in viable cells, and at additional sites with cytotoxic okadaic acid or taxol. Oncogene. 2004;23(31):5301-5315. Glaser SP, Lee EF, Trounson E, et al. Antiapoptotic Mcl-1 is essential for the development and sustained growth of acute myeloid leukemia. Genes Dev. 2012; 26(2):120-125. Yoshimoto G, Miyamoto T, JabbarzadehTabrizi S, et al. FLT3-ITD up-regulates MCL-1 to promote survival of stem cells in acute myeloid leukemia via FLT3-ITD-specific STAT5 activation. Blood. 2009; 114(24):5034-5043. Wilson C, Ye X, Pham T, et al. AXL inhibition sensitizes mesenchymal cancer cells to antimitotic drugs. Cancer Res. 2014; 74(20):5878-5890. Kariolis MS, Miao YR, Diep A, et al. Inhibition of the GAS6/AXL pathway augments the efficacy of chemotherapies. J Clin Invest. 2017;127(1):183-198. Huey MG, Minson KA, Earp HS, DeRyckere D, Graham DK. Targeting the TAM receptors in leukemia. Cancers (Basel). 2016;8(11).pii: E101. Perl AE, Altman JK, Cortes J, et al. Selective inhibition of FLT3 by gilteritinib in relapsed or refractory acute myeloid leukaemia: a multicentre, first-in-human, open-label, phase 1-2 study. Lancet Oncol. 2017; 18(8):1061-1075. Mori M, Kaneko N, Ueno Y, et al. Gilteritinib, a FLT3/AXL inhibitor, shows antileukemic activity in mouse models of FLT3 mutated acute myeloid leukemia. Invest New Drugs. 2017 May 17. [Epub ahead of print]. Lee LY, Hernandez D, Rajkhowa T, et al. Preclinical studies of gilteritinib, a nextgeneration FLT3 inhibitor. Blood. 2017; 129(2):257-260. Young CS, Clarke KM, Kettyle LM, Thompson A, Mills KI. DecitabineVorinostat combination treatment in acute myeloid leukemia activates pathways with potential for novel triple therapy. Oncotarget. 2017 May 19. [Epub ahead of print]. Hou HA, Chou WC, Kuo YY, et al. TP53 mutations in de novo acute myeloid leukemia patients: longitudinal follow-ups show the mutation is stable during disease evolution. Blood Cancer J. 2015;5:e331.

2057

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Acute Myeloid Leukemia

Ferrata Storti Foundation

Phenotype in combination with genotype improves outcome prediction in acute myeloid leukemia: a report from Children’s Oncology Group protocol AAML0531

Andrew P. Voigt,1* Lisa Eidenschink Brodersen,1* Todd A. Alonzo,2,3 Robert B. Gerbing,2 Andrew J. Menssen,1 Elisabeth R. Wilson,1 Samir Kahwash,4 Susana C. Raimondi,5 Betsy A. Hirsch,6 Alan S. Gamis,7 Soheil Meshinchi,2,8 Denise A. Wells1 and Michael R. Loken1

Haematologica 2017 Volume 102(12):2058-2068

1 Hematologics, Inc, Seattle, WA; 2Children's Oncology Group, Monrovia, CA; 3University of Southern California, Los Angeles, CA; 4Nationwide Children’s Hospital, Columbus, OH; 5 St. Jude’s Children’s Research Hospital, Memphis, TN; 6University of Minnesota Medical Center, Minneapolis, MN; 7Children’s Mercy Hospitals & Clinics, Kansas City, MO and 8Fred Hutchinson Cancer Research Center, Seattle, WA, USA

*APV and LEB contributed equally to this study

ABSTRACT

doi:10.3324/haematol.2017.169029

iagnostic biomarkers can be used to determine relapse risk in acute myeloid leukemia, and certain genetic aberrancies have prognostic relevance. A diagnostic immunophenotypic expression profile, which quantifies the amounts of distinct gene products, not just their presence or absence, was established in order to improve outcome prediction for patients with acute myeloid leukemia. The immunophenotypic expression profile, which defines each patient’s leukemia as a location in 15-dimensional space, was generated for 769 patients enrolled in the Children’s Oncology Group AAML0531 protocol. Unsupervised hierarchical clustering grouped patients with similar immunophenotypic expression profiles into eleven patient cohorts, demonstrating high associations among phenotype, genotype, morphology, and outcome. Of 95 patients with inv(16), 79% segregated in Cluster A. Of 109 patients with t(8;21), 92% segregated in Clusters A and B. Of 152 patients with 11q23 alterations, 78% segregated in Clusters D, E, F, G, or H. For both inv(16) and 11q23 abnormalities, differential phenotypic expression identified patient groups with different survival characteristics (P<0.05). Clinical outcome analysis revealed that Cluster B (predominantly t(8;21)) was associated with favorable outcome (P<0.001) and Clusters E, G, H, and K were associated with adverse outcomes (P<0.05). Multivariable regression analysis revealed that Clusters E, G, H, and K were independently associated with worse survival (P range <0.001 to 0.008). The Children’s Oncology Group AAML0531 trial: clinicaltrials.gov Identifier: 00372593.

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2058

Introduction

Correspondence: lisa@hematologics.com

Received: March 21, 2017. Accepted: September 6, 2017. Pre-published: September 7, 2017.

©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2058

Acute myeloid leukemia (AML) is a heterogeneous disease affecting multiple lineages of hematopoietic cells. The disease is classified by well-defined cytogenetic or molecular abnormalities, and as one of eight broadly defined morphologic classes, each with a variety of immunophenotypic features.1 Such diverse assessment modalities are difficult to compare, preventing a more comprehensive understanding of the relationships between morphology, genotype, immunophenotype, and outcome in patients with AML. Conventional characterization of leukemic immunophenotypes used for lineage assignment involves calculating the proportion of cells with antigen expression above a defined threshold, but does not quantify the amount of each gene product.2 We recently reported that antigen intensity relationships of normal hematopoietic cell populations are invariant throughout maturation from an uncommitted progenitor cell to a mature blood cell among both pediatric and adult individuals.3,4 The study helped confirm that with a high degree of quality control haematologica | 2017; 102(12)

Phenotype with genotype: a new paradigm for AML

and system stability,3 precise quantification of surface gene product expression can provide a robust basis to assess phenotypic deviations from normal maturation patterns that occur as a result of neoplastic transformation.5 This concept is supported by our recent report of the recurrent multidimensional immunophenotype, RAM, which independently identifies high-risk pediatric AML at diagnosis.6 In this study, we used the complete multidimensional, quantitative leukemic immunophenotype [immunophenotypic expression profile (IEP)] to improve the assessment of the heterogeneity seen in AML. In a study of 769 patients, those with similar global immunophenotypic patterns were grouped together by unsupervised hierarchical clustering. This approach provided a focal point to correlate continuous and categorical variables and determine the relationships among immunophenotype, genotype, morphology, and outcome in a sufficiently large cohort of similarly treated patients. The integration of testing modalities helps identify previously unrecognized patients with poor clinical outcomes, and further clarifies the relationship between a specific genetic event and its effect on the expression of surface gene products.

Methods Patient samples Of 1022 newly diagnosed pediatric patients with de novo AML enrolled on the Children's Oncology Group (COG) protocol AAML0531, 769 satisfied three criteria for the study reported herein: (1) submitting a bone marrow aspirate (N=626, 81%) or peripheral blood specimen (N=143, 19%) (when bone marrow was unavailable) for multidimensional flow cytometry (MDF) at diagnosis, (2) providing consent for correlative biology studies, and (3) MDF analysis showing leukemia comprising >10% of nonerythroid cells. Patients with acute promyelocytic leukemia were not enrolled in the AAML0531 study and those with Down syndrome were excluded from analysis. Details of the AAML0531 protocol have been previously published.7,8 Centrally reviewed cytogenetic data and French–American–British (FAB) classification were available for 97.5% and 86.2% of patients, respectively. The study was approved by the institutional review board (IRB) at the National Cancer Institute and IRBs at each of the 184 enrolling centers. Patients and their families provided informed consent or assent as appropriate. The trial was conducted in accordance with the Declaration of Helsinki.

Risk stratification AAML0531 defined diagnostic risk by cytogenetic or molecular markers. Patients with monosomy 7, deletion 5q, monosomy 5, or FLT3-ITD with a high allelic ratio (>0.4) were classified as highrisk. Patients that had inv(16) (including t(16;16) variants), t(8;21), a CEBPA mutation, or an NPM1 mutation were classified as lowrisk. All other patients with known cytogenetics were allocated to the standard-risk group. Patients with persistence of disease, as identified by morphologic assessment at the end of initial induction therapy, were also stratified to the high-risk group.

Flow cytometric analysis Bone marrow aspirates or peripheral blood samples were drawn in heparin or ethylenediaminetetraacetic acid (EDTA) and submitted for MDF assessment. For correlative biology studies, MDF was performed centrally at Hematologics with a standardized panel of monoclonal antibodies designed to detect measurable residual dishaematologica | 2017; 102(12)

ease with a difference-from-normal approach.7 A comprehensive flow cytometric work up was performed at the contributing institution, but was not reviewed centrally. Specimens were processed as previously described.7

Hierarchical clustering Unsupervised hierarchical clustering of the 769 IEPs was performed with R Studio. A dendrogram was constructed using a Euclidian distance metric and a complete-linkage method without scaling of the IEPs. Morphologic and genetic data were not included in the clustering algorithm and did not influence the dendrogram. Selection of the number of phenotypic clusters was validated with the elbow method by comparing within- and betweencluster variation (Online Supplementary Figure S1).9,10

Mutation screening Genomic DNA was extracted from diagnostic bone marrow specimens by the Puregene® protocol (Gentra Systems, Inc.). CEBPA, FLT3-ITD, WT1, and NPM1 mutations were screened as previously described.11-14 Patients with inv(16) or t(8;21) were further analyzed for coinciding c-KIT mutations.

Morphologic assessment The initial AML diagnosis was made at each contributing institution, and concurrence of the diagnostic morphologic assessment was centrally reviewed. In the central review, subtypes were assigned according to the FAB and World Health Organization (WHO) 2001 classifications15 (Online Supplementary Table S1), as the clinical trial began prior to the release of the 2008 WHO.

Results Phenotypic clustering Diagnostic specimens from 769 patients enrolled in AAML0531 were assessed for quantitative expression of several cell surface markers using a standardized panel of reagents (Figure 1A-D).7,8 The neoplastic cell population from each specimen was identified by using CD45 versus log right-angle light scatter (SSC) gating with WinList (Verity Software House, Topsham, ME, USA), and was subsequently verified with all combinations of reagents (Figure 1E). The log mean fluorescence intensities (MFI) of 12 cell surface antigens as well as the physical parameters forward scatter (FSC) and log SSC were then determined for the identified leukemic cell population. The coefficient of variation (CV) of CD34 expression was also calculated as an independent parameter for each patient’s leukemia, since CD34 has been shown to provide a measure of maturation for neoplastic cells.16,17 Together, these independently quantified characteristics defined the IEP for each patient as a location in a 15-dimensional data space (Figure 1F,G). Of note, the methodology of CD45 vs. SSC gating in defining the IEP precludes analysis of the influence of minor phenotypic (sub)clones on phenotype. Unsupervised hierarchical clustering was performed using the calculated IEPs to segregate patients with similar multidimensional phenotypes into related regions of a dendrogram (Figure 2A). The relative intensities of each antigen assessed were depicted in a blue-to-yellow color gradient (extending over four log units) as a heatmap (Figure 2B). Although the dataset consisted of a heterogeneous collection of 769 unique quantitative diagnostic phenotypes, unsupervised clustering identified groups of patients with similar IEPs. Computational analysis sug2059

A. Voigt et al.

gested that the dataset could be appropriately divided into eleven distinct clusters (Online Supplementary Figure S1) with similar IEPs (Clusters A–K, Figure 2A,B). Comparable phenotypic heterogeneity was observed across specimen types (peripheral blood and bone marrow).

Association between phenotype and morphology Although the current WHO classification of AML is dependent on the molecular and genetic features of leukemia,1 morphologic classification of AML describes lineage and maturational features of the leukemic population.18 To determine the relationship between morphologic subtype and immunophenotype, phenotypic clusters were assessed for co-occurrence of FAB subtypes (Figure 2C, Online Supplementary Table S2). Patients classified as FAB-M0 or M1 (N=22 and N=90, respectively) were scattered throughout the dendrogram and had no identifiable groupings. Patients classified as FAB-M2 (N=161) (blue) segregated in two predominant regions of the dendrogram within Clusters A and B. The majority of patients classified as FAB-M4 (N=165) (green) segregated near the top of Cluster A. Patients classified as FAB-M5 (N=144) (yellow) were identified in a large region of the dendrogram corresponding to Clusters D, E, F, and G. Nine patients classified as FAB-M6 did not segregate together. Patients classified as FAB-M7 (N=30) predominantly segregated to Clusters H and K. These findings suggest that some morphologic groups share similar patterns of expression of gene products. Furthermore, some FAB classes can be subdivided according to phenotypic differences.

Association between phenotype and genotype The underlying cytogenetic and mutational status of each patient was appended to the dendrogram to analyze the association between genotype and phenotype (Figure 2C). Clear relationships between IEPs and underlying genotypes were identified, as many patients with the same genetic abnormality segregated in similar regions of the dendrogram. Each phenotypic cluster (A–K, Figure 2A,B) was analyzed for high-density regions of each genetic abnormality (consisting of at least 9 patients). A genotypic subcluster was assigned for each high-density region identified (Subclusters A-i to K-i, Figure 2D and Online Supplementary Table S3). The major chromosomal abnormalities were highly correlated with IEPs. Of the 95 patients with inv(16), 79% were within Cluster A (Figure 2C). Subcluster analysis revealed that 53% of all inv(16) patients were tightly clustered within the A-ii region and 20% of all inv(16) patients segregated to the A-v region of Cluster A (Figure 2C,D). Patients with inv(16) made up 86% of Subcluster A-ii and 35% of Subcluster A-v. Subclusters A-ii and A-v had similar frequencies of patients with coinciding c-KIT mutations (30% and 26%, respectively). Both subclusters were associated with FAB M4 morphology (89% and 48%, respectively). Patients in Subclusters A-ii and A-v had distinct multidimensional phenotypes (Online Supplementary Figure S2). Of the 109 patients with t(8;21), 92% segregated in Cluster A or B. Strikingly, 70% of the patients with t(8;21) were identified in Subclusters A-iii and B-i (Figure 2C,D).

Figure 1. Overview of immunophenotypic expression profiling (IEP). (A) Diagnostic bone marrow specimens were acquired from each patient enrolled in the COG protocol AAML0531. (B) Then, 200 μL of bone marrow was added to 6 tubes containing (C) Fluorescein Isothiocyanate (FITC)-, Phycoerythrin (PE)-, Peridinin Chlorophyll Protein Complex (PerCP)-, and anti-Allophycocyanin (APC)-conjugated antibodies. (D) Flow cytometry was performed on samples in each tube, and fluorescence measurements, forward light scatter (FSC) and right-angle light scatter (SSC) characteristics were collected for 200,000 events. (E) Flow cytometry results were analyzed by an expert, and leukemic populations were identified by CD45 vs. SSC gating. (F) For cells identified in the leukemia gate, the mean intensity for each parameter (black dot) was computed. Mean fluorescence intensity was utilized as an unaltered quantification of signal. In addition, the coefficient of variation (CV) of CD34 was computed as a metric to assess cellular maturation. (G) Collectively, these 15 quantitative intensities constituted the IEP for each patient.

2060

haematologica | 2017; 102(12)

Phenotype with genotype: a new paradigm for AML

Figure 2. Hierarchical clustering of IEPs. (A) A dendrogram was generated by unsupervised hierarchical clustering of the 769 IEPs. Eleven phenotypic clusters (Aâ&#x20AC;&#x201C; K), selected by minimizing within-cluster variation and maximizing between-cluster variation, were identified for outcome analysis. (B) The IEP of each patient is presented in the form of a heatmap. (C) The morphologic, karyotypic, and mutational profiles of each patient were compared to the IEPs. (D) Genotypic (sub)clusters with associations among IEPs and morphologic, karyotypic, and/or mutational abnormalities were identified for further analysis. (E) Key denoting intensity of the surface gene product expression to color scale and mutational and morphologic classifications. Somatic mutations are denoted in red and those for wild-type patients are denoted in gray. FAB classifications are indicated by color.

haematologica | 2017; 102(12)

2061

A. Voigt et al. Table 1. Comparison of 5-year EFS (95%CI) of patients in individual phenotypic clusters with that of all other patients.

Cluster* Cluster A (N=266) Cluster B (N=77) Cluster C (N=106) Cluster D (N=68) Cluster E (N=52) Cluster F (N=28) Cluster G (N=11) Cluster H (N=81) Cluster I (N=23) Cluster J (N=41) Cluster K (N=16)

EFS of patients in cluster of interest

EFS of all other patients

51% (44%–56%) 69% (57%–78%) 50% (40%–69%) 57% (44%–68%) 39% (26%–52%) 39% (22%–57%) 27% (7%–54%) 28% (18%–38%) 52% (30%–70%) 56% (40%–70%) 19% (5%–40%)

48% (43%–52%) 46% (43%–50%) 48% (45%–52%) 48% (44%–51%) 49% (46%–53%) 49% (45%–53%) 49% (45%–53%) 51% (47%–55%) 49% (45%–52%) 48% (44%–52%) 49% (46%–53%)

0.257 <0.001 0.803 0.149 0.041 0.173 0.027 <0.001 0.583 0.479 0.006

*Statistically significant phenotypic clusters (in comparison with all other patients) are highlighted in blue. EFS: event-free survival; CI: confidence interval.

These two phenotypic groups are largely distinguished by quantitative expression of CD56 (Online Supplementary Figure S3). Subclusters A-iii and B-i predominantly included patients with t(8;21) (85% and 83%, respectively). Further, these subclusters were strongly associated with FAB M2 morphology (79% and 80%, respectively). Interestingly Subcluster A-v, which was associated with inv(16), also included 17 patients with t(8;21) (all of which were inv(16) negative). Of all patients with t(8;21), 16% segregated into Subcluster A-v. The 152 patients with 11q23/MLL (KMT2A) alterations had distinct IEPs. Overall, 78% of all 11q23 patients segregated in Cluster D, E, F, G, or H. Within each cluster, a subcluster was defined to further investigate the clinical and biologic features of patients with MLL translocations. The majority of patients in each subcluster harbored MLL translocations (Subcluster D-i: 66%; Subcluster E-i: 67%; Subcluster F-i: 57%; Subcluster G-i: 82%; and Subcluster H-i: 53%), and each subcluster was strongly associated with FAB M5 morphology (Subcluster D-i: 69%; Subcluster E-i: 78%; Subcluster F-i: 63%; Subcluster G-i: 100%; and Subcluster H-i: 24%). The translocation partners for 11q23 did not appear to be associated with phenotypic heterogeneity (Online Supplementary Figure S4). MLL chromosomal rearrangements by abnormality (e.g., t(9;11) or t(11;19)) could not be subdivided further into more specific immunophenotypic associations. Of the 17 patients with the CBFA2T3–GLIS2 chimeric fusion gene transcript,19,20 59% were identified within Cluster K. Conversely, 63% (10 of 16) of patients within this cluster harbored CBFA2T3–GLIS2 fusions. The IEPs of these patients revealed remarkably consistent bright expression of CD56, dim or negative expression of CD45 and CD38, and a lack of HLA-DR expression, which is consistent with the previously reported RAM phenotype.6 Within this cluster, 54% of patients had FAB M7 morphology. AML-associated somatic mutations also had a strong association with immunophenotype. Patients with CEBPA mutations segregated in several small groups throughout the dendrogram, most prominently in Subcluster A-i and Subcluster C-i (Figure 2C,D). Of 46 patients with CEBPA mutations, 30% were identified 2062

within A-i, and 24% were identified within Subcluster Ci. CEBPA mutations occurred in 48% of patients in Subcluster A-i and 21% of patients in Subcluster C-i. FLT3-ITD mutations were associated with 4 genotypic subclusters, often in combination with other genetic mutations. Overall, 61% of all patients with FLT3-ITD mutations were identified in Subclusters A-iv, A-vi, C-i, and C-ii. In Subcluster A-iv, 65% (11 of 17) of patients with FLT3-ITD mutations also had a WT1 mutation, therefore, 42% of all patients in the dataset had both mutations. In Subcluster C-i, only 16% (4/25) of patients with FLT3-ITD mutations also had a CEBPA mutation; however this accounted for 44% of all patients that had co-existing FLT3-ITD and CEBPA mutations. In Subcluster C-ii, 50% (9 of 18) of patients with FLT3-ITD mutations also had an NPM1 mutation, constituting 43% of all patients in the dataset with both FLT3-ITD and NPM1 mutations.

Associations among phenotype, genotype, and outcome Kaplan–Meier analysis of outcomes was performed to define the 5-year event-free survival (EFS) of patients in different phenotypic clusters (Figure 3). The 5-year EFS of patients in each individual cluster was compared to the EFS of all other patients; statistically significant differences were observed for patients in Clusters B, E, G, H, and K (Table 1). Representations of phenotypes observed for these clusters are shown in Online Supplementary Figures S5-S9. Univariable analysis revealed that 5-year EFS and overall survival (OS) varied among patients in different IEP clusters. Patients in Cluster B had more favorable 5-year EFS and patients within Clusters E, G, H, and K had more adverse OS and EFS than those in other clusters. Patients in Cluster B (who predominantly had t(8;21)) had significantly higher 5-year EFS (69%, CI: 57%–78%) than those in other clusters (46%, CI: 43%–50%; P<0.001). Interestingly, patients in Clusters E, G, H, and K had poor 5-year EFS (19%–39%; Table 1). After adjusting for age and molecular/cytogenetic risk groups, multivariable analysis revealed that patients in Clusters G, H, and K had significantly higher hazard ratios (HRs) for EFS and OS, whereas those in Cluster E had a significantly higher HR haematologica | 2017; 102(12)

Phenotype with genotype: a new paradigm for AML

Figure 3. Kaplan–Meier analysis of 5-year EFS of patients by phenotypic cluster. (A) Curves showing differences in EFS for patients in the 11 IEP clusters. (B) Curves showing phenotypic clusters in which the 5-year EFS for patients was significantly different (P<0.05) from that of patients in other clusters. Although patients in Clusters E and F had identical EFS, Cluster F EFS was not statistically significant due to low sample size.

for OS, but not EFS (Table 2). Cluster B, with a high frequency of t(8;21), showed no additional favorable effect on EFS or OS. A similar outcome analysis was performed on genotypic subclusters to determine whether the combination of phenotypic and genotypic features leads to a more accurate prediction of patient outcomes than genotypic features alone. Patients with inv(16) in Subclusters A-ii and A-v had significantly different outcomes (Figure 4A), which was not further explained by the frequency of corresponding c-KIT mutations (30% vs. 26%, respectively). The 5year EFS for patients with inv(16) with a phenotype corresponding to Subcluster A-v was significantly higher (84%, CI: 57%–94%) than for those with a phenotype corresponding to Subcluster A-ii (54%, CI: 39%–67%; P=0.039). In further analysis of the role of c-KIT mutations in core binding factor (CBF) leukemias, CBF/c-KIT positive patients (N=50) demonstrated no statistically significant differences in EFS (P=0.105) or OS (P=0.192) than CBF/c-KIT negative patients (N=154). In addition, three clusters had sufficient (N>1) patients with CBF AML and c-KIT mutations: Clusters A, B, and H. For each of these clusters, the difference in EFS and OS was assessed between CBF/c-KIT positive vs. CBF/c-KIT negative patients. In Clusters B and H, there was no significant difference in OS or EFS between CBF/c-KIT positive and CBF/c-KIT negative patients. Within Cluster A, CBF/c-KIT positive patients (N=29) had a significantly worse 5-year EFS than CBF/c-KIT negative patients (N=91) (50% +/19% vs. 71% +/- 10%, P=0.046). However, a difference in outcome between CBF/c-KIT patients in Subcluster A-ii vs. A-v was not observed for either OS (A-ii: 71.1%, A-v: 77.8%, P=0.915) or EFS (A-ii: 46.7%, A-v: 55.6%, P=0.680). The outcomes of patients with 11q23 abnormalities also differed by phenotype. Patients with 11q23 within haematologica | 2017; 102(12)

Subcluster D-i or E-i, who were assigned to the standardrisk group at diagnosis, had a higher 5-year EFS (Subcluster D-i: 51%, CI: 36%–64%; Subcluster E-i: 42%, CI: 26%–58%) than those in Subclusters F-i, G-i, or H-i (Subcluster F-i: 25%, CI: 8%–47%; Subcluster G-i: 22%, CI: 3%–51%; Subcluster H-i: 20%, CI: 3%–47%), though this difference was not significant (P=0.063) likely due to low sample size (Figure 4B). However, merging these clusters on the basis of their relationships within the dendrogram revealed two distinct 5-year EFS outcomes (Subclusters D-i+E-i: 47% vs. Subclusters F-i+G-i+H-i: 23%, P=0.006). The subclusters in which patients with 11q23 had poorer outcomes did not have a higher frequency of MLL translocation partners associated with higher risk in other pediatric studies of MLL rearrangements.21,22 However, patients with t(9;11) were overrepresented in Subcluster D-i. Therefore, while phenotype did not further subset high-risk MLL rearrangements, it did further identify patients with t(9;11). Similar outcome comparison for patients with t(8;21) within Subclusters Bi, A-iii, and A-v showed no significant difference in outcome with 5-year EFS of 76% (CI: 64%-88%), 85% (CI: 69%-100%), and 58% (CI: 34%-82%), respectively (P=0.152). Likewise, comparison of patients with FLT3-ITD within Subclusters A-iv, A-vi, C-i, and C-ii revealed no significant difference in 5-year EFS. A specific area of the dendrogram, which primarily comprised Clusters H, I, and J, was void of high-density genotypic subclusters. Although patients in these clusters had several genetic abnormalities, none of the patients with unifying abnormalities grouped together with the combined density and frequency observed in other regions of the dendrogram. The outcomes of patients in Clusters I and J were unremarkable, the absence of patients with inv(16) or t(8;21) is, however, notable. Cluster H was marked by a large cohort size (N=81) and poor patient outcomes. Of note, 86% of patients within 2063

A. Voigt et al.

Cluster H were classified in the low-risk or standard-risk group on the basis of cytogenetic or molecular markers. Strikingly, patients classified in the low-risk group by cytogenetic or molecular markers within Cluster H (N=25) had significantly poorer 5-year EFS (33%) and 5-year OS (66%) than all other favorable-risk patients (N=265) in the study (5-year EFS=72%, P<0.001; OS=84%, P=0.008; Online Supplementary Figure S10A,B). Furthermore, Group H predicts significantly worse EFS and OS for high-risk patients, but only predicts significantly worse OS for standard-risk patients (Online Supplementary Figure S10C-F).

Supervised prediction of cluster and subcluster cohorts Unsupervised hierarchical clustering was employed to discover a previously unknown structure in the dataset, namely the relationship between immunophenotype, genotype, and outcome. To apply these identified relationships to new patients, a supervised boosted decision tree algorithm was constructed to replicate the original unsupervised cluster classifications using only the IEP. The 769 patients were divided into training (N=513, 2/3) and testing (N=256, 1/3) cohorts. This algorithm was applied

Table 2. Univariable and multivariable Cox regression analysis of the phenotypic clusters cohorts by age and cytogenetic or molecular risk classification.

OS from study entry Univariable Cluster groups All other clusters Cluster H Cluster E Cluster K Cluster G Cluster B Age (years) 3–10 0-2 ≥11 Risk Group Standard Low High Karyotype by complexity 0-2 3+ Multivariable Cluster groups All other clusters Cluster H Cluster E Cluster K Cluster G Cluster B Age (years) 3–10 0–2 ≥11 Risk Group Standard Low High Karyotype by complexity 0-2 3+

EFS from study entry

95% CI

532 81 52 16 11 77

1 2.08 1.82 3.03 3.03 0.72

1.50– 2.89 1.19–2.79 1.69–5.46 1.42–6.47 0.45–1.16

<0.001 0.006 <0.001 0.004 0.178

1 1.81 1.55 2.29 2.34 0.56

1.36–2.41 1.07–2.25 1.31–3.99 1.16–4.73 0.37–0.85

<0.001 0.022 0.004 0.018 0.007

231 174 364

1 1.15 1.27

0.82–1.61 0.95–1.68

0.640 0.102

1 1.23 1.12

0.93–1.63 0.89–1.42

0.142 0.336

360 290 108

1 0.32 1.26

0.23–0.43 0.93–1.70

<0.001 0.130

1 0.38 1.31

0.29–0.48 1.01–1.70

<0.001 0.045

612 138

1 1.40

1.05-1.87

0.024

1 1.20

0.93-1.54

0.162

519 77 51 16 11 76

1 2.09 1.85 4.18 3.64 1.20

1.48–2.94 1.18–2.91 2.18–8.00 1.65–8.05 0.73–1.98

<0.001 0.008 <0.001 0.001 0.480

1 1.79 1.41 2.35 2.32 0.87

1.33–2.41 0.95–2.09 1.29–4.27 1.12–4.83 0.56–1.35

0.001 0.087 0.005 0.024 0.539

227 173 358

1 0.72 1.51

0.103 0.005

1 0.88 1.27

0.64–1.21 1.00–1.62

0.428 0.051

360 287 103

1 0.31 1.26

0.22–0.44 0.90–1.76

<0.001 0.173

1 0.40 1.34

0.30–0.52 1.00–1.78

<0.001 0.047

612 138

1 1.58

1.17-2.14

0.003

1 1.24

0.96-1.61

0.106

0.49–1.07 1.13–2.02

*Statistically significant hazard ratios with corresponding P values in bold type. OS: overall survival; EFS: event-free survival; HR: hazard ratio; CI: confidence interval.

2064

haematologica | 2017; 102(12)

Phenotype with genotype: a new paradigm for AML

to the test cohort, and accurately classified 84.0% of patients within an eleven-class prediction setting (average sensitivity =0.824, average specificity =0.982, average F1score =0.841). The sensitivity, specificity, and F1-score of predictions for each cluster in the test cohort are detailed in Online Supplementary Table S4. As patients with inv(16) and 11q23 showed divergent clinical outcomes based on subcluster designations, additional boosted tree-based models were trained to identify inv(16) patients within Subclusters A-ii and A-v and 11q23 patients within Subcluster H-i. Subclusters D-i, E-i, F-i, and G-i completely overlap with Clusters D, E, F, and G, hence no additional boosted decision tree models were trained to identify these subclusters. Patients with inv(16) were partitioned into A-ii and A-v subclusters with an overall accuracy of 92.3% (average sensitivity =0.833, average specificity =0.895, average F1-score =0.800). Patients with 11q23 were partitioned into D-i, E-i, F-i, G-i, and H-i with an overall accuracy of 95.4% (average sensitivity =0.743, average specificity =0.979, average F1-score =0.790). Additional details and performance metrics of subcluster models are provided in Online Supplementary Table S5. Each of the eleven clusters demonstrated a unique pattern of dysregulated surface gene product expression. To characterize these immunophenotypic patterns, boosted decision tree models were trained to distinguish patients in each cluster from all other patients using the IEP. The relative influence of each IEP parameter in generating a correct prediction was quantified, where a high relative influence indicates that a given surface gene product is an important component of a clusterâ&#x20AC;&#x2122;s immunophenotypic expression pattern. As opposed to the evaluation of positive or negative expression of single antigens, the variable importance quantifications highlight the multidimensional nature of surface gene product dysregulation that defines each of the eleven clusters (Figure 5). This data is depicted in Figure 5, where the six most important IEP parameters for each cluster are displayed and each parameter is subsequently colored to illustrate the quantitative

amount of each antigen (or non-antigen variable for SSC and FSC), as compared to the quantitative antigen expression of normal myeloid progenitor cells. For example, the six most important IEP parameters for Cluster A are, in order: CD34, CD56, CD13, HLA-DR, CD33, and CD117. CD34 is the most important parameter and the relative intensity of the antigen is essentially the same as that of normal myeloid progenitor cells. CD56 is the second most important parameter for Cluster A and has increased expression of CD56 compared to normal myeloid progenitor cells (which lack the CD56 antigen). In comparison CD34 is the most important parameter for Cluster J, but due to lack of expression, not presence.

Discussion In this study, we present a novel approach for the diagnostic classification of AML that uses quantitative MDFbased diagnostic classification of AML. This method generates a unique patient-specific profile, which, in combination with the diagnostic karyotype and/or somatic mutations, provides a more robust and precise prognostic tool than that of individual testing modalities. Historically, relationships among immunophenotype, genotype, morphology, and outcome have been loosely correlated,23-27 with phenotypic associations hinging largely on the expression of a single antigen.28,29 Although previous studies have performed clustering analysis of immunophenotypic data to identify small subgroups of patients with poor prognosis,30-32 such studies have not evaluated a sufficiently large cohort of uniformly treated patients. By defining the IEP as a continuous variable, patients with similar global immunophenotypic patterns can be grouped together with hierarchical clustering, thus providing a focal point to correlate continuous and categorical test results. As such, our findings clarify the heterogeneous relationships among phenotype, genotype, morphology at diagnosis, and clinical outcome in pediatric

Figure 4. Kaplanâ&#x20AC;&#x201C;Meier analysis of the differences in 5-year EFS among patients with identical phenotypes in different genotypic subclusters. (A) Patients with inv(16) in Subcluster A-v (green) had a significantly better 5-year EFS than those with inv(16) in Subcluster A-ii (light purple) (P=0.039). (B) Patients with 11q23 in Subclusters D-i, E-i, F-i, G-I, and H-i had heterogeneous 5-year EFS.

haematologica | 2017; 102(12)

2065

A. Voigt et al.

AML. Limiting the study to de novo AML in children and young adults avoids the increased complexity of multiple lineages resulting from the progression of myelodysplastic syndrome to AML in adults. Phenotypic heterogeneity is observed in AML to such an extent that the detailed quantitative gene product expression of each leukemia is unique.33 The observed het-

erogeneity is presumably a result of the accumulation of multiple genetic abnormalities that can occur in myriad combinations. Leukemogenesis disrupts normal hematopoietic development by altering the precise amounts and timing of appearance of surface gene products required for proper maturation. The accumulation of multiple genetic mutations causes a loss of gene product

Figure 5. Relative influence of IEP components in each cluster. A boosted decision tree model was trained to identify patients in each cluster versus all other patients. Variable importance was computed by calculating the mean decrease in the Gini index relative to the maximum decrease in the Gini index.10 The relative influence of the six most important IEP components were plotted for each cluster. In addition, the relative influence of each IEP component is colored in comparison to the intensity of the gene product expression on normal, uncommitted progenitor cells for pediatric patients.3 For example, a blue-colored bar indicates that the average intensity of a surface gene product within a cluster is lower than the average intensity of that same surface gene product in normal pediatric patients. The combination of most influential IEP components provides insight regarding the multidimensional pattern of surface gene products that are expressed within each cluster. Of note, surface gene products need not be aberrantly expressed to have a high relative influence.

2066

haematologica | 2017; 102(12)

Phenotype with genotype: a new paradigm for AML

regulation resulting in a unique quantitative immunophenotype for each individual leukemia. Previous efforts have applied computational algorithms to elucidate genomic (in one case fully genomic) classifications of adult AML, correlating overlapping genotypic profiles with clinical outcome.34-36 It is remarkable that by using immunophenotype as the discriminator of patient cohorts we observe several similarities between the current pediatric study and those (using genomic data as the discriminator) in adult AML. These similarities include: the number of computationally relevant AML subtypes, the high level of specificity with which the t(8;21) and inv(16) cohorts cluster together, and indications of further biologic and prognostic subdivisions within current cytogenetic classifications. Most notably, we observe a similar occurrence of multiple FLT3-ITD subgroups, with a subset exhibiting NPM1 co-mutations, in line with those reported by Papaemmanuil and colleagues.36 Additional commonalities include an observed subset of t(8;21) patients with co-occurring c-KIT mutation, and, to a lesser extent, a subset of patients with overlapping inv(16) and c-KIT mutations. Where a few previous studies have shown the negative impact of c-KIT on OS, relative risk (RR), complete response (CR), and/or EFS for CBF-AML patients,37-39 our results are in agreement with those studies which show no additional prognostic effect of c-KIT on the OS and EFS of CBF-AML patients.40,41 Our study also revealed more diverse subgroups of the MLL fusion patients than previous studies, which is not surprising given the higher prevalence of MLL mutations observed in pediatric AML. Interestingly, immunophenotype alone identifies patient subgroups with adverse clinical outcome. Patients in Clusters G, H, and K had poor 5-year EFS and OS, and both univariable and multivariable Cox regression analyses demonstrated that these phenotypes were independent predictors of poor outcome. Interestingly, Group H had markedly poor outcome and no unifying genetic features, yet a high frequency of patients in the cohort had genetic abnormalities. When comparing patients with favorable-risk cytogenetic/molecular markers in Group H to all other patients with the same favorable-risk markers, those in Group H have significantly worse survival, suggesting that additional uncharacterized mutations captured in the immunophenotype have an adverse effect on patient outcome. We recently reported that the RAM immunophenotype independently identifies a cohort of very young pediatric AML patients with poor response to therapy and adverse outcome.6 Herein we demonstrate that this cohort, which was originally identified by expert analysis, can be reproduced by hierarchical cluster analysis. In addition, 63% of RAM patients were discovered to have the CBFA2T3–

References 1. 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. 2. Clinical and Laboratory Standards Institute. Clinical flow cytometric analysis of neoplas-

haematologica | 2017; 102(12)

GLIS2 chimeric fusion gene, which also indicates a poor prognosis in AML.19,20 Patients with RAM positive status but CBFA2T3-GLIS negative status have equally poor outcome (data not shown), highlighting one context in which a solely genomic approach would preclude identifying all poor-risk patients with these clinical features. Multidimensional phenotypes can also help to further explain the heterogeneous response to therapy seen within conventional cytogenetic classifications. Although patients with inv(16) are considered to be low-risk,1,42 patients with inv(16) in Subcluster A-ii had significantly worse 5-year EFS than those in Subcluster A-v. Patients with inv(16) in Subcluster A-ii had distinct immunophenotypic features from those within Subcluster A-v (Online Supplementary Figure S2), suggesting that additional genetic abnormalities may contribute to the differential expression of gene products and perhaps a more aggressive clinical course. However, both subclusters had a similar prevalence of corresponding c-KIT mutations, indicating that the specific addition of the c-KIT mutation does not explain the observed difference in outcome, as has been reported among pediatric patients with core binding factor previously defined, thus it should be deleted here and left as CBF-AML.40 This finding further supports the fact that the combination of phenotype and genotype can provide a more accurate method to predict the risk of induction failure, relapse or death in these genetically defined lowrisk patients. Our novel approach of clustering diagnostic immunophenotypes facilitates the segregation of patients with potentially hundreds of different genotypes into clinically meaningful cohorts, thereby allowing a more accurate prognostic determination within apparently uniform genetic groupings. As patients with similar genotypes segregated in similar regions of the dendrogram, genetic subclusters with high phenotypic-genotypic associations could be identified. This begins to elucidate the relationship between a genetic hit and its phenotypic consequence and the subsequent impact on clinical outcome. We plan to further validate these findings in COG AAML1031. Funding This work was supported by grants U10CA098543 (Chair’s grant), U10CA098413 (the Statistical Center Grant), U10CA180886 (NCTN Operations Center Grant), and U10CA180899 (NCTN Statistics & Data Center Grant). Acknowledgments The authors would like to thank the patients and families for participating in AAML0531. We also thank Vani Shanker for scientific editing.

tic hematolymphoid cells; approved guideline—Second edition. Wayne, PA: US Food and Drug Administration; 2008. FDA publication no. 7-150. 3. Loken MR, Voigt AP, Eidenschink Brodersen L, et al. Consistent gene product expression #2: antigen intensities on bone marrow cells are invariant between individuals. Cytometry A. 2016;89(11):987-996. 4. Loken MR, Voigt AP, Eidenschink Brodersen

L, Fritschle W, Menssen AJ, Wells DA. Consistent quantitative gene product expression: #3. Invariance with age. Cytometry A. 2016;89(11):997-1000. 5. Loken MR. Residual disease in AML, a target that can move in more than one direction [editorial]. Cytometry B. 2014; 86(1):1517. 6. Eidenschink Brodersen L, Alonzo TA, Menssen AJ, et al. A recurrent immunophe-

2067

A. Voigt et al.

10.

11.

12.

13.

14.

15.

16.

17.

18.

2068

notype at diagnosis independently identifies high-risk pediatric acute myeloid leukemia: a report from Children’s Oncology Group. Leukemia. 2016; 119(10):2077-2080. Loken MR, Alonzo TA, Pardo L, et al. Residual disease detected by multidimensional flow cytometry signifies high relapse risk in patients with de novo acute myeloid leukemia: a report from Children’s Oncology Group. Blood. 2012;120(8):15818. Gamis AS, Alonzo TA, Meshinchi S, et al. Gemtuzumab ozogamicin in children and adolescents with de novo acute myeloid leukemia improves event-free survival by reducing relapse risk: Results from the randomized phase III Children's Oncology Group trial AAML0531. J Clin Oncol. 2014; 32(27):3021-3032. Ketchen DJ, Shook CL. The application of cluster analysis in strategic management research: an analysis and critique. Strat. Mgmt. J. 1996;17(6):441-458. Tibshirani R, James G, Witten D, Hastie T. An introduction to statistical learning- with applications in R. New York, NY: Springer, 2013. Ho PA, Alonzo TA, Gerbing RB, et al. Prevalence and prognostic implications of CEBPA mutations in pediatric acute myeloid leukemia (AML): a report from the Children’s Oncology Group. Blood. 2009;113(26):6558-6566. Meshinchi S, Alonzo TA, Stirewalt DL, et al. Clinical implications of FLT3 mutations in pediatric AML. Blood. 2006; 108(12):36543661. Ho PA, Zeng R, Alonzo TA, et al. Prevalence and prognostic implications of WT1 mutations in pediatric acute myeloid leukemia (AML): a report from the Children’s Oncology Group. Blood. 2010;116(5):702710. Brown P, McIntyre E, Rau R, et al. The incidence and clinical significance of nucleophosmin mutations in childhood AML. Blood. 2007;110(3):979-985. Vardiman JW, Harris NL, Brunning RD. The World Health Organization (WHO) classification of the myeloid neoplasms. Blood. 2002;100(7):2292-2302. Civin CI, Loken MR. Cell surface antigens on human marrow cells: Dissection of hematopoietic development using monoclonal antibodies and multiparameter flow cytometry. Int J Cell Cloning. 1987;5(4):1-6. Loken MR, Terstappen LWMM, Civin CI, Fackler MJ. Flow cytometric characterization of erythroid, lymphoid and monomyeloid lineages in normal human bone marrow. In: Laerum OD, Bjerksnes R., eds. Flow Cytometry in Hematology. New York, NY: Academic Press; 1992:31-42. Bennett JM, Catovsky D, Daniel MT, et al. Proposals for the classification of the acute leukaemias. French-American-British (FAB) Co-operative Group. Br J Haematol. 1976;33(4):451-458.

19. Gruber TA, Larsen Gedman A, Zhang J, et al. An Inv(16)(p13.3q24.3)-encoded CBFA2T3-GLIS2 fusion protein defines an aggressive subtype of pediatric acute megakaryoblastic leukemia. Cancer Cell. 2012;22(5):687-697. 20. Masetti R, Pigazzi M, Togni M, et al. CBFA2T3-GLIS2 fusion transcript is a novel common feature in pediatric, cytogenetically normal AML, not restricted to FAB M7 subtype. Blood. 2013;121(17):3469-3472. 21. Balgobind BV, Raimondi SC, Harbott J, et al. Novel prognostic subgroups in childhood 11q23/MLL-rearranged acute myeloid leukemia: results of an international retrospective study. Blood. 2009; 114(12):24892496. 22. Pigazzi M, Masetti R, Bresolin S, et al. MLL partner genes drive distinct gene expression profiles and genomic alterations in pediatric acute myeloid leukemia: an AIEOP study. Leukemia. 2011;25:560-563. 23. Lin LI, Chen CY, Lin DT, et al. Characterization of CEBPA mutations in AML: most patients with CEBPA mutations have biallelic mutations and show a distinct immunophenotype of the leukemic cells. Clin Cancer Res. 2005;11(4):1372-1379. 24. Hurwitz CA, Raimondi SC, Head D, et al. Distinctive Immunophenotypic features of t(8;21)(q22;22) acute myeloblastic leukemia in children. Blood. 1992;80(12):3182-3188. 25. Paietta E, Andersen J, Gallagher R, et al. The immunophenotype of acute promyelocytic leukemia (APL): an ECOG study. Leukemia. 1994;8(7):1108-1112. 26. Adriaansen HJ, te Boekhorst PAW, Hagemeijer AM, van der Schoot CE, Delwel HR, van Dongen JJ. Acute myeloid leukemia M4 with bone marrow eosinophilia (M4Eo) and inv(16)(p13q22) exhibits a specific immunophenotype with CD2 expression. Blood. 1993;81(11):30433051. 27. Baer MR, Stewart CC, Lawrence D, et al. Acute myeloid leukemia with 11q23 translocations: myelomonocytic immunophenotype by multiparameter flow cytometry. Leukemia. 1998; 12(3):317-325. 28. Ossenkoppele GJ, van de Loosdrecht AA, Schuurhuis GJ. Review of the relevance of aberrant antigen expression by flow cytometry in myeloid neoplasms. Br J Haematol. 2011;153(4):421-436. 29. Zeijlemaker W, Kelder A, Wouters R, et al. Absence of leukaemic CD34+ cells in acute myeloid leukaemia is of high prognostic value: a longstanding controversy deciphered. Br J Haematol. 2015;171(2):227238. 30. Chen CY, Chou WC, Tsay W, et al. Hierarchical cluster analysis of immunophenotype classify AML patients with NPM1 gene mutation into two groups with distinct prognosis. BMC Cancer. 2013;13:107. 31. Hulkkonen J, Vilpo L, Hurme M, Vilpo J. Surface antigen expression in chronic lymphocytic leukemia: clustering analysis,

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

interrelationships and effects of chromosomal abnormalities. Leukemia. 2002;16(2): 178-85. Zangrando A, Luchini A, Buldini B, et al. Immunophenotype signature as a tool to define prognostic subgroups in childhood acute myeloid leukemia. Leukemia. 2006; 20(5):888-891. Terstappen LW, Safford M , Konemann S, et al. Flow cytometric characterization of Acute Myeloid Leukemia. Part II. Phenotypic heterogeneity at diagnosis. Leukemia. 1991;5(9):757-767. Valk P JM, Verhaak R GW, Beijen MA, Erpelinck C AJ, Barjesteh van Waaliwijk van Doorn-Khosrovani S, Boer JM. Prognostically useful gene-expression profiles in Acute Myeloid Leukemia. N Engl J Med. 2004;350(6):1617-1628. Bullinger L, Döhner K, Blair E, Fröhling S, Schlenk RF, Tibshirani R. Use of geneexpression profiling to identify prognostic subclasses in adult acute myeloid leukemia. N Engl J Med. 2004;350(16):1605-1616. Papaemmanuil E, Gerstung M, Bullinger L, Gaidzik VI, Paschka P, Roberts ND. Genomic classification and prognosis in Acute Myeloid Leukemia. N Engl J Med. 2016;374:2209-2221. Shimada A, Taki T, Tabuchi K, et al. KIT mutations, and not FLIT3 internal tandem duplication, are strong associated with a poor prognosis in pediatric acute myeloid leukemia with t(8;21): a study of the Japanese Childhood AML Cooperative study group. Blood. 2006;107(5):1806-1809. Manara E, Bisio V, Masetti R, et al. Corebinding factor acute myeloid leukemia in pediatric patients enrolled in the AIEOP AML 2002/1 trial: screening and prognostic impact of c-KIT mutations. Leukemia. 2014;28:1132-1134. Chen W, Xie H, Wang H, et al. Prognostic significance of KIT mutatins in core-binding factor acute myeloid leukemia: a systematic review and meta-analysis. PLoS ONE. 2016;11(1):e0146614. Pollard JA, Alono TA, Gerbing RB, et al. Prevalence and prognostic significance of KIT mutations in pediatric core binding factor AML enrolled on serial pediatric cooperative trials for de novo AML. Blood. 2010;15(12):2372-2379. Klein K, Kaspers G, Harrison CJ, et al. Clinical impact of additional cytogenetic aberrations, cKIT and RAS mutations, and treatment elements in pediatric r(8;21)AML: results from an international retrospective study by the International BerlinFrankfurt-Münster Study group. J Clin Oncol. 2015;33(36):4247- 4259. Grimwade D, Walker H, Oliver F, et al. The importance of diagnostic cytogenetics on outcome in AML: analysis of 1,612 patients entered into the MRC AML 10 trial. The Medical Research Council Adult and Children’s Leukemia Working Parties. Blood. 1998;92(7):2322-2333.

haematologica | 2017; 102(12)

ARTICLE

Chronic Lymphocytic Leukemia

CD40 signaling instructs chronic lymphocytic leukemia cells to attract monocytes via the CCR2 axis

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Martijn H.A. van Attekum,1,2 Jaco A.C. van Bruggen,1,2 Erik Slinger,1,2 M. Cristina Lebre,2 Emilie Reinen,1,2 Sabina Kersting,3 Eric Eldering3,4 and Arnon P. Kater1,4

Department of Hematology, Academic Medical Center, University of Amsterdam; Department of Experimental Immunology, Academic Medical Center, University of Amsterdam; 3Department of Hematology, Haga Teaching Hospital, The Hague and 4 Lymphoma and Myeloma Center Amsterdam (LYMMCARE), the Netherlands 1 2

JvB and ES contributed equally to this work as second authors, and EE and APK are senior authors

Haematologica 2017 Volume 102(12):2069-2076

ABSTRACT

hronic lymphocytic leukemia (CLL) cells are provided with essential survival and proliferative signals in the lymph node microenvironment. Here, CLL cells engage in various interactions with bystander cells such as T cells and macrophages. Phenotypically distinct types of tumor infiltrating macrophages can either be tumor supportive (M2) or play a role in tumor immune surveillance (M1). Although recent in vitro findings suggest a protective role for macrophages in CLL, the actual balance between these macrophage subsets in CLL lymphoid tissue is still unclear. Furthermore, the mechanism of recruitment of monocytes towards the CLL lymph node is currently unknown. Both questions are addressed in this paper. Immunofluorescence staining of lymph node samples showed macrophage skewing towards an M2 tumor-promoting phenotype. This polarization likely results from CLL-secreted soluble factors, as both patient serum and CLL-conditioned medium recapitulated the skewing effect. Considering that CLL cell cytokine secretion is affected by adjacent T cells, we next studied CLL-mediated monocyte recruitment in the presence or absence of T-cell signals. While unstimulated CLL cells were inactive, T cell-stimulated CLL cells actively recruited monocytes. This correlated with secretion of various chemokines such as C-C-motif-ligand-2,3,4,5,7,24, C-X-C-motif-ligand5,10, and Interleukin-10. We also identified CD40L as the responsible Tcell factor that mediated recruitment, and showed that recruitment critically depended on the C-C-motif-chemokine-receptor-2 axis. These studies show that the shaping of a tumor supportive microenvironment depends on cytokinome alterations (including C-C-motif-ligand-2) that occur after interactions between CLL, T cells and monocytes. Therefore, targeted inhibition of CD40L or C-C-motif-chemokine-receptor-2 may be relevant therapeutic options.

Correspondence: a.p.kater@amc.nl

Received: September 27, 2016. Accepted: September 22, 2017. Pre-published: September 29, 2017. doi:10.3324/haematol.2016.157206 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2069 ÂŠ2017 Ferrata Storti Foundation

Introduction Chronic lymphocytic leukemia (CLL) cells strongly depend on interactions with bystander T cells and monocyte-derived cells (MDCs) within the lymph node (LN) microenvironment for their survival and resistance to therapy.1 The role of LNresiding T cells in the pathogenesis of CLL has gained much attention. It is suggested that interaction of neoplastic B cells with T cells results in skewing of the T-cell compartment towards CD40L-expressing CD4+ T cells.2 These T cells, in turn, induce both CLL cell survival and proliferation via upregulation of several pro-survival molecules as well as increased secretion of cytokines.3,4 The interaction between MDCs and CLL is less well understood, although in vitro experiments show that MDCs, in the form of Nurse-like cells, can induce CLL cell survival5 haematologica | 2017; 102(12)

2069

M.H.A. van Attekum et al.

through C-X-C motif chemokine 12, B-cell activating factor and A proliferation-inducing ligand signaling.5,6 Based on data from different malignancies, there are two subgroups of tumor-associated macrophages (TAMs): 1) M2-like CD68+CD163+/CD206+ macrophages are characterized by an immunosuppressive phenotype, whereas 2 M1-like CD68+CD80+ macrophages display an immunesurveilling phenotype.7 Although there is large intratumoral and intertumoral heterogeneity, it has been suggested that M1 TAMs lead to a better and M2 TAMs lead to a worse prognosis across different tumor types.8 Tumors that are associated with M2 TAMs include breast,9 ovarian,7 and prostate10 cancers, whereas colon carcinoma TAMs are of M1 phenotype.11 With respect to CLL, ex vivo evidence shows that MDCs are present in the LN,12 and it was recently shown that MDCs contribute to CLL progression, as MDC depletion by clodronate treatment in the TCL1 CLL mouse model leads to slower CLL progression.13,14 Whether LN-residing

macrophages in human CLL are indeed of a protective M2 phenotype has, however, not been directly studied. It is also not known whether circulating monocytes can actively be recruited towards the tumor-infiltrated LN. Migration of CLL cells to the LN microenvironment depends on chemotactic gradients through the CXCL12/ CXCR4,15 CXCL13/CXCR516 and CCL19,21/CCR717 axes. Upon interaction with LN-residing cells, such as T cells, CLL cells can alter their secretome,4,18,19 which, in turn, could potentially impact both skewing and migration of other cells, like MDCs. Co-operative or reciprocal signals between the triad formed by CLL cells, T cells, and MDCs could, therefore, critically contribute to the supportive microenvironment for CLL cells. Here, we investigated both the possibly supportive differentiation of MDCs and their recruitment as a result of CLL-secreted cytokines in the context of T-cell signals. We found that CLL-secreted factors were able to differentiate macrophages towards a supporting M2 phenotype.

Figure 1. Chronic lymphocytic leukemia (CLL) cells differentiate monocytes towards an M2 phenotype. (A) Paraffin-embedded lymph node (LN) material from 9 CLL patients (for patients' characteristics see Online Supplementary Table S1) was stained by immunohistochemistry with Hematoxylin & Eosin (HE), CLL markers CD5 and CD20, and T-cell marker CD3. One representative slide is shown. Yellow scale bars correspond to 20 Îźm. More information on image acquisition can be found in the Methods section. (B) The same 9 samples as in (A) were stained by immunofluorescence for pan-macrophage marker CD68 in combination with either M1 marker CD80 (top) or M2 marker CD206 (bottom). One representative slide is shown; stainings of the other slides can be found in Online Supplementary Figure S1A. Yellow scale bars correspond to 20 mm. (C) CD80 and CD206 intensity levels (both red signal) were quantified per macrophage (green signal) for each slide presented in Figure 1B and Online Supplementary Figure S1A using automated image analysis (see Methods section). Per-patient (each line) average macrophage intensity of both CD80 and CD206 are indicated (each dot). The patient presented in Figure 1B is indicated in blue. **P<0.01, paired t-tests. (D) A triple immunofluorescence staining with antibodies directed against T-cell marker CD3, CLL cell marker CD20, and macrophage marker CD68 was performed on four of the slides used for (1A). One representative slide is shown (sample CLL LN 09). Scale bars correspond to 100 mm (top) or 10 mm (bottom). (E) Healthy donor (HD) monocytes were differentiated for 72 hours (h) with IMDM containing 25% CLL serum or 25% pooled HD serum, or with complete medium containing IFN-Y (M1) or IL-4 (M2) as controls (left). In a separate experiment, monocytes were differentiated for 72 h by direct contact with CLL cells in complete medium, or with complete medium containing IFN-Y (M1), IL-4 (M2) or recombinant human (rh)NAMPT as controls (right). Monocyte differentiation was then tested by staining for M1 marker CD80 and M2 markers CD163 and CD206 using flow cytometry. Each bar represents the relative geometrical mean (GeoMean) of the fluorescence signal compared to the control condition and error bars indicate Standard Error of Mean (SEM) of n=22 (left) or n=3 (right) CLL samples.

2070

haematologica | 2017; 102(12)

CCR2-mediated monocyte recruitment by CLL cells

Secondly, T cell/CD40 stimulation of CLL cells induced CLL cells to recruit monocytes; an action which critically depends on CCR2 signaling.

Methods Patients’ samples, stimulation and conditioned medium collection Patient material was obtained from CLL patients, after written informed consent according to the guidelines of the Medical Ethical Committee of the Academic Medical Center, Amsterdam, the Netherlands, in accordance with Declaration of Helsinki protocols. For T-cell stimulation, peripheral blood mononuclear cells (PBMCs) were isolated from either healthy donors (HDs) or from CLL patients using Ficoll gradient purification according to the manufacturer's instructions (Lucron, Dieren, the Netherlands). These PBMCs (either magnetically sorted or not to enrichen the T-cell fraction) were added to CLL cells (in either an allogeneic or autologous fashion, as indicated) in a 1:1 ratio, each at a concentration of 1.0*106 cells/mL. Stimulating antibodies directed against

CD3 (1 mg/mL, clone 1XE, Sanquin, Amsterdam, the Netherlands) and CD28 (3 mg/mL, clone 15E8, Sanquin) were added for T-cell activation. After 72 hours (h), conditioned medium was collected. For stimulation with CD40L, CLL cells were cultured at a concentration of 1.5*106 cells/mL on CD40L transfected NIH-3T3 cells or on mock transfected 3T3 cells as described previously,3 all in IMDM supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA, USA), 100 U/mL penicillin-100 mg/mL streptomycin (Life Technologies, Austin, TX, USA), 2 mM L-glutamine (Life Technologies), and 0.00036% b-mercaptoethanol (Sigma, St. Louis, MO, USA) (IMDM+/+) for 16 h, after which conditioned medium was collected. Cell-free conditioned medium was kept at -80°C until use.

Migration assays Conditioned or control media were diluted 1:2 in chemotaxis medium (PBS with 1% albumin, low endotoxin; Sigma). Monocytes were freshly isolated from HDs after obtaining written informed consent using negative MACS depletion as described previously20 and resuspended in chemotaxis medium. The diluted media were added in the lower chambers of a 5 mm chemotaxis

Figure 2. T cell-stimulated chronic lymphocytic leukemia (CLL) cells secrete monocyte-attracting chemokines. (A) Freshly isolated healthy donor (HD) monocytes were seeded in the upper chambers of a trans-well migration plate to migrate towards conditioned media (cond med) obtained from PBMC samples from CLL patients (for patients' characteristics see Online Supplementary Table S1) that were unstimulated (unstim) or stimulated (stim) for 72 hours (h) by contact with HD PBMC T cells that were activated using α-CD3/α-CD28 antibodies. Next, the amount of migrated monocytes was quantified using DAPI staining. Each dot represents the relative (compared to the unstimulated CLL condition) DAPI signals of 8 different CLL conditioned media or 3 control media in 3 independent experiments using monocytes from 3 different donors and mean±Standard Error of Mean (SEM) are shown. All measurements were performed in triplicate. *P<0.05 in t-tests. (B) CLL cells were stimulated with α-CD3/αCD28 activated T cells or not stimulated for 16 h. RNA from CD5/CD19 FACS sorted CLL cells (>99% purity) was subjected to microarray analysis and tested for differential expression of chemokines involved in monocyte migration.23-26 Dots represent expression levels and mean±SEM are shown for 5 paired CLL samples. **P<0.01, ****P<0.0001 in a two-way ANOVA test with Bonferroni post hoc analysis. (C) Protein levels of chemokines involved in monocyte migration23-26 were determined in three conditioned media that were used to perform the migration assays in (A) by using Luminex. Dots represent protein levels and mean+SEM are shown for 3 CLL conditioned media; *P<0.05, ***P<0.001, ****P<0.0001 in a two-way ANOVA test with Bonferroni post hoc analysis.

haematologica | 2017; 102(12)

2071

M.H.A. van Attekum et al.

assay plate (96 well ChemoTX®, NeuroProbe, Gaithersburg, MA, USA) and 100,000 monocytes were transferred to the upper chamber. After 2 h, chemotaxis was quantified by measuring the DAPI (4,6 diamidino-2-phenylindole) signal of migrated monocytes as described previously.20 When measuring inhibitor effects, both media and monocytes were incubated for 30 minutes (min) on ice with the indicated inhibitors directly prior to the migration assay. The following chemokine receptor inhibitors were used: 1 mg/mL CCR1 inhibitor BX471 (Sigma), 1 mg/mL CCR2 inhibitor INCB3284 (Tocris Bioscience, Bristol, UK), 1mM CCR3 inhibitor SB328437 (Tocris), 1 mM CCR5 inhibitor Maraviroc (Apexbio, Houston, TX, USA), 1mM CXCR4/7 inhibitor Plerixafor (Apexbio), and 0.1 mg/mL IL-10 neutralizing antibody (R&D Systems, Minneapolis, MN, USA).

Measurement of chemokine levels Previously generated microarray profiles4 of purified (>99%) CLL cells stimulated for 16 h with activated T cells (deposited under accession number GSE50572) were normalized and analyzed using the R2 platform (http://r2.amc.nl) and data were extracted using its DataGrabber feature. When testing protein secretion, conditioned media were analyzed for the indicated chemokines by Luminex using the ProcartaPlex 9-plex chemokine immunoassay kit extended with CCL7, CCL24, CXCL5, and IL10 (eBioscience, San Diego, CA, USA) according to the manufacturer’s instructions.

Supplementary methods Information on monocyte isolation and in vitro differentiation, LN material and immunofluorescence, rhCD40L stimulation and

Figure 3. CD40L-stimulated chronic lymphocytic leukemia (CLL) cells attract monocytes as a result of CCR2 axis signaling. (A) Freshly isolated healthy donor (HD) monocytes were seeded in the upper chambers of a trans-well migration plate to migrate towards conditioned media (cond med) obtained from PBMC samples from CLL patients (for patients' characteristics see Online Supplementary Table S1) that were cultured for 16 hours (h) on CD40L-overexpressing (CD40L stim) or parental NIH-3T3 cells (unstim). Next, the amount of migrated monocytes was quantified using DAPI staining. Each dot represents the relative [compared to the unstimulated (unstim) CLL condition] DAPI signals of 12 different CLL conditioned media or 3 control media in 3 independent experiments using monocytes from 3 different donors and mean±Standard Error of Mean (SEM) are shown. All measurements were performed in triplicate. ****P<0.0001 in t-tests. (B) Protein levels of chemokines involved in monocyte migration23-26 were determined in the conditioned media that were used to perform the migration assays in (A) by using Luminex. Dots represent protein levels and mean±SEM are shown for 12 CLL conditioned media; **P<0.01, ***P<0.001, ****P<0.0001 in a two-way ANOVA test with Bonferroni post hoc analysis. (C) Freshly isolated monocytes and conditioned media were pre-incubated for 30 minutes (min) with individual small-molecule inhibitors directed against indicated chemokine receptors, with an IL-10 neutralizing antibody, or a combination of all inhibitors (combi), before performing migration assays as in (A). Each dot represents the relative (compared to the unstimulated CLL condition) DAPI signals obtained in 4 independent experiments using monocytes from 3 different donors and different CLL supernatants; mean±SEM are shown. All measurements were performed in triplicate; *P<0.05, ***P<0.001, in a one-way ANOVA test with Bonferroni post hoc analysis. (D) Monocytes were seeded in the upper chambers of a trans-well migration plate to migrate towards migration medium without or with 10 ng/mL recombinant human CCL2 (rhCCL2 low) or 100 ng/mL rhCCL2 (rhCCL2 high). Next, the amount of migrated monocytes was quantified using DAPI staining. Each dot represents the relative (compared to condition without rhCCL2) DAPI signals of 9 separate read-outs in 3 independent experiments using monocytes from 3 different donors and mean±SEM are shown; **P<0.01, ****P<0.0001 in t-tests.

2072

haematologica | 2017; 102(12)

CCR2-mediated monocyte recruitment by CLL cells

intracellular CCL2 measurements, and statistical analyses can be found in the Online Supplementary Methods.

Results LN-residing macrophages display an M2 phenotype, and both CLL cells and CLL serum induce M2 skewing To study the phenotype of macrophages in the CLL LN, paraffin-embedded LN sections from CLL patients were stained for the pan-macrophage marker CD68 in combination with either the M1 marker CD80 or the M2 marker CD206 using immunofluorescence. The CD80/CD206 fluorescence signal per macrophage (CD68+) was then quantified using an automated cell identification pipeline in CellProfiler. CD68 positive cells were present in all samples tested and were dispersed throughout the CLL-infiltrated LNs (Figure 1A and B). Within the CD68+ cells, a higher CD206 intensity was observed as compared to CD80 (1.89 vs. 1.00 arbitrary units) (Figure 1B and C, and Online Supplementary Figure S1A). In order to visualize spatial organization of both T cells and macrophages within CLL LN, CD3– CD68–CD20 combinatory staining was performed. CD68+ cells could be found scattered throughout the LN, and were always surrounded by CLL cells. No typical configuration of CD3 in relation to CD68 could be detected, although, on occasion, CD68+ cells were in close contact with T cells (Figure 1D). We next studied whether the leukemic cells could account for the observed M2 polarization. First, we tested whether soluble factors present in CLL serum differentiated monocytes towards an M2 phenotype. Freshly isolated monocytes from HDs were incubated with either sera from 22 different CLL patients or pooled serum from HDs, and differentiation status was measured using flow cytometry. IFN-Y (M1) and IL-4 (M2) differentiated monocytes were included for comparison. Both M2 markers CD163 [mean relative Geomean 1.55±Standard Error of Mean (SEM) 0.16] and CD206 (2.14±0.21), but not M1 marker CD80 (1.00±0.11) were increased in CLL serum-

differentiated monocytes compared to HD serum-differentiated monocytes (Figure 1E, left, and Online Supplementary Figure S1B for a representative gating strategy). Notably, no difference between serum from CLL samples from patients with either mutated or unmutated Immunoglubulin Heavy gene, or with low (<20*109) versus high (>100*109/L) leukocyte counts was observed (data not shown). As CLL-serum components resulted in M2 differentiation, we next investigated whether the observed M2 differentiation in the LN was actuated by CLL cells. To this end, HD-isolated monocytes were differentiated for 72 h using CLL cells or positive control NAMPT.21 IFN-Y (M1) and IL-4 (M2) differentiated monocytes were again included as control. We found an upregulation of M2 markers after IL-4 stimulation. In line with the differentiation effect of CLL serum, both CLL cells and NAMPT induced an upregulation of the M2 markers, but not of the M1 marker (Figure 1E, right). Furthermore, the M2 differentiation depended on soluble factors, as conditioned medium from CLL cells likewise induced M2 differentiation (Online Supplementary Figure S1C). Taken together these data indicate that CLL-secreted factors are able to differentiate macrophages towards an M2 phenotype.

T-cell-stimulated CLL cells secrete monocyte-attracting chemokines Next, we investigated whether CLL cells could direct monocyte migration. Using trans-well migration assays, we found no migration of HD monocytes towards supernatants of unstimulated CLL cells (Figure 2A). As both in vitro and ex vivo CLL LN studies strongly suggest active interaction of CLL cells with residential T cells, such as CD40L expressing follicular helper T cells within the LN22 we hypothesized that such interaction could affect CLL cytokine secretion. Therefore, supernatants of CLL cells cultured in direct contact with HD PBMCs that included αCD3/αCD28 activated T cells (Tact) (Online Supplementary Figure S2A) were compared to unstimulated CLL cells for

Figure 4. Model of chronic lymphocytic leukemia (CLL) T-cell macrophage triad in the formation of a supportive tumor microenvironment. Stimulation of CLL cells by activated T-cell-produced CD40L (1) induces them to secrete CCL2 (2), which in turn recruits monocytes towards the lymph node (LN) (3). As a result of CLL-secreted factors, monocytes differentiate towards a tumor supporting M2 phenotype (4). Mo: monocyte; PB: peripheral blood.

haematologica | 2017; 102(12)

2073

M.H.A. van Attekum et al.

induction of monocyte recruitment. Indeed, conditioned medium from CLL cells co-cultured in contact with activated T cells induced migration of monocytes (Figure 2A). To exclude the possibility that migration resulted from a mixed-lymphocyte reaction, we verified that autologous T cells enriched from CLL samples induced similar migration (Online Supplementary Figure S2B). In addition, we tested the migration effect of T cells without CLL cells and found some migration induction of T cells only. This effect likely results from T-cell stimulation of B/CLL cells present in these samples due to contamination, as this effect was reduced when T cells were magnetically enriched (Online Supplementary Figure S2C). To determine the candidate chemokines expressed by stimulated CLL cells that could underlie the recruitment of monocytes, we analyzed our previously generated microarray dataset (GSE50572) of purified CLL cells that were stimulated with Tact.4 Expression of several monocyte-attracting chemokines such as CCL2, 3, 4, 5, 7, CXCL1, 5, 10 and IL1023-26 was up-regulated in CLL cells after contact with Tact (Figure 2B). To measure chemokine secretion by CLL cells, a Luminex assay was performed on three conditioned media used in Figure 2A. All chemokines that were upregulated on the mRNA level were also significantly upregulated on the protein level (Figure 2C).

CD40L-stimulated CLL cells attract monocytes as a result of CCR2 axis signaling As Tact -stimulated CLL cells have highly similar gene expression profiles compared to CD40L-stimulated CLL cells,4 we investigated if CD40L stimulation similarly endows CLL cells with monocyte recruiting capacity. Comparable to the Tact results, supernatants from CD40Lstimulated CLL cells induced migration of monocytes (Figure 3A). These data indicate that a co-operative signal from Tact cells is needed for CLL cells to induce monocyte migration. Furthermore, CD40L appears to be responsible for the Tact -mediated monocyte migration. Notably, by using this T-cell free CD40L system, these data indicate that CLL-derived (rather than Tact -derived) chemokines induce recruitment of monocytes. Secreted proteins in the conditioned media from CD40L-stimulated CLL cells were measured. In line with the Tact data, several monocyte-attracting chemokines such as CCL2, 3, 4, 5, 7, 24, CXCL5, 10 and IL-10 were secreted by the CLL cells after CD40L stimulation (Figure 3B). None of the chemokines tested were detectable in supernatant from CD40L overexpressing NIH-3T3 cells alone (data not shown). To pinpoint which of the up-regulated candidate chemokines was responsible for the migration of monocytes, we applied selective small molecule inhibitors for the relevant chemokine receptors.23-26 Inhibition of CCR2 was sufficient to reduce migration to a background level. There was no additive effect of inhibition of other chemokine receptors, as a combination of the different receptor inhibitors yielded similar inhibition to CCR2 inhibition alone (Figure 3C). In a control experiment, no direct cytotoxic effect of the CCR2 inhibitor was detected after 72 h stimulation of CLL cells (Online Supplementary Figure S3A). Furthermore, supernatants from unstimulated CLL cells in combination with the different chemokine receptor inhibitors showed migration comparable to background (data not shown). We also tested if the CCR2 inhibitor could revert CLL cell-induced M2 differentiation, as observed in Figure 1, but no effect was found (data not 2074

shown). As macrophage activation depends on Bruton Tyrosine Kinase,27 we tested if migration could be reverted by inhibition via ibrutinib, but found no effect of this inhibitor (data not shown). As CCL2 is a potent CCR2 ligand,28 we verified its induction in CD40L-stimulated CLL cells by using (cell free) recombinant CD40L (Online Supplementary Figure S3B). Furthermore, recombinant CCL2 resulted in monocyte migration (Figure 3D). Combined, these data suggest that CD40 signaling is responsible for T cell-mediated monocyte migration by CLL cells and that this migration depends on the CCL2-CCR2 axis.

Discussion It is widely accepted that interactions with local bystander cells in the LN are critical for CLL maintenance.1 Various reports have mechanistically elucidated how bystander cells can support CLL cells, but the active role of CLL cells in shaping this supportive microenvironment is still largely unclear. In this complex interplay between the leukemic and various types of surrounding cells, we functionally addressed two key aspects: the chemo-attraction of monocytes, and the crosstalk between CLL cells and activated T cells herein. Our findings are compatible with a model (Figure 4) in which stimulation by CD40L on T cells in the LN induces CLL cells to secrete several monocyte-attracting chemokines. Of these, we found CCL2 to be the most potent chemo-attractor, suggesting that this chemokine potentially plays an important role in vivo by recruiting monocytes towards the malignant cells in the LN via CCR2. The immuno-fluorescence data suggest that, following engagement with CLL cells in the LN, monocytes undergo skewing towards a tumor-supportive M2 phenotype (see also below). Several reports that studied migration of monocytes in the context of inflammation have concluded that chemoattraction can occur via activation of several different chemokine receptor signaling pathways.23-26 We here identified CCR2 as the receptor most likely to be responsible for monocyte recruitment towards CLL cells. The most potent chemokine that recruits monocytes via the CCR2 receptor is CCL2,28 which indeed recruited monocytes in our experiments (Figure 3D). These data are in line with the recent observation that adoptive transfer of leukemic TCL1-derived splenocytes into recipient mice that are deficient for CCR2 resulted in significantly lower percentages and numbers of monocytes in the spleen.13 Besides its importance in CLL, CCL2 has been shown to recruit monocytes towards primary tumors in prostate cancer. Furthermore, this recruitment resulted in enhanced tumor growth.29 CCR2 antagonist PF-04136309 reduced the number of monocytes and restored chemosensitivity in a pancreas tumor mouse model, indicating the therapeutic potential of CCL2/CCR2 inhibition.30 Our studies suggest that, also in CLL, these inhibitors can be a relevant therapeutic option, although additional in vivo studies are required. In the light of the large number of potential interactions in the CLL LN, it is worth noting that specifically the Tcell co-stimulatory signal CD40L leads to induction of monocyte trafficking. The levels of chemokines secreted by unstimulated CLL cells are insufficient to induce migration above background (Figures 2A and 3A). Although haematologica | 2017; 102(12)

CCR2-mediated monocyte recruitment by CLL cells

CLL cells stimulated by monocyte-derived Nurse-like cells show increased production of CCL3 and CCL4,19 these cytokines apparently play a subordinate role in monocyte recruitment; despite their presence in the conditioned media (Figure 3B), monocyte migration is not prevented by blocking their cognate receptors CCR1 or -5 (Figure 3C). In contrast to the monocyte-attracting effect by CLL cells, it has been shown that bystander cells such as CD3+ or CD68+ cells are unable to produce CCL2 themselves.31 We have previously shown that CD40L accounts for most of the transcriptional effects of T cells on CLL cells4 and, based on our data, CD40L is sufficient to induce CCL2 production and monocyte recruitment. In this context, others have shown that another key T-cell cytokine, IL21, is not essential for CCL2 induction.32 Our observation that the large majority of macrophages in the CLL LN are of an M2 phenotype (Figure 1B and C) strongly suggests initiation of M2 differentiating signaling events once monocytes enter the CLL lymph node environment. Factors that can account for this differentiation include NAMPT21 or High mobility group box 1 (HMGB1)12 secreted by LN-residing CLL cells. We could confirm that addition of NAMPT indeed skews monocytes towards an M2 type (Figure 1E). In addition, T-helper-2 cells that also reside in the LN22 secrete various cytokines that induce M2 differentiation, including IL-4, IL-10, and IL-13. Notably, the production of IL-10 could be complemented by CLL cells that are stimulated by T cells (Figures 2C and 3B). Together these findings indicate that the LN provides an M2-inducing milieu, which likely results in a supportive macrophage phenotype that can induce CLL cell survival and immune suppression.

References 1. Burger JA. Nurture versus nature: the microenvironment in chronic lymphocytic leukemia. Hematology Am Soc Hematol Educ Program. 2011;2011:96-103. 2. Ghia P, Strola G, Granziero L, et al. Chronic lymphocytic leukemia B cells are endowed with the capacity to attract CD4+, CD40L+ T cells by producing CCL22. Eur J Immunol. 2002;32(5):1403-1413. 3. Hallaert DY, Jaspers A, van Noesel CJ, et al. c-Abl kinase inhibitors overcome CD40mediated drug resistance in CLL: implications for therapeutic targeting of chemoresistant niches. Blood. 2008;112(13):51415149. 4. Pascutti MF, Jak M, Tromp JM, et al. IL-21 and CD40L signals from autologous T cells can induce antigen-independent proliferation of CLL cells. Blood. 2013; 122(17):3010-3019. 5. Burger JA, Tsukada N, Burger M, et al. Blood-derived nurse-like cells protect chronic lymphocytic leukemia B cells from spontaneous apoptosis through stromal cell-derived factor-1. Blood. 2000; 96(8):2655-2663. 6. Tsukada N, Burger JA, Zvaifler NJ, Kipps TJ. Distinctive features of "nurselike" cells that differentiate in the context of chronic lymphocytic leukemia. Blood. 2002; 99(3):1030-1037.

haematologica | 2017; 102(12)

Indeed, the association of M2 differentiation and tumor support has been pointed out in several other tumor types.8-11 Functionally, the tumor-promoting effects of M2 macrophages have been attributed to an increased production of direct tumor-promoting cytokines33 and a suppression of the immune response.21 M2 macrophages can, for example, induce a suppression of cytotoxic T cells, as they can up-regulate expression of PD-1 on T cells.21 In addition, they inhibit T-cell proliferation.21 Lastly, M2 macrophages suppress T-cell activation and promote the differentiation towards Treg cells.34 In the light of the recent development of T-cell therapy against CLL neoantigens,35 the subversion of T cells by macrophages is an important point to address. In conclusion, our studies provide insight into several aspects of the complex interactions that take place in the CLL LN, and indicate how the triad of CLL cell, T cell, and macrophage potentially contributes to the shaping of the tumor-microenvironment in CLL. Finally, we identified CCR2 as a potential therapeutic target to interrupt the intercellular interplay. Funding This work was supported by Dutch Cancer Foundation grant number UVA 2011-5097 (APK). Acknowledgments The authors would like to thank all volunteers that donated blood for this study. We furthermore thank Richard Volckmann for his help with the microarray analysis and Steven Pals for providing us with the CLL LN slides.

7. Zhang QW, Liu L, Gong CY, et al. Prognostic significance of tumor-associated macrophages in solid tumor: a meta-analysis of the literature. PLoS One. 2012; 7(12):e50946. 8. Lewis CE, Pollard JW. Distinct role of macrophages in different tumor microenvironments. Cancer Res. 2006;66(2):605-612. 9. Sousa S, Brion R, Lintunen M, et al. Human breast cancer cells educate macrophages toward the M2 activation status. Breast Cancer Res. 2015;17(1):101. 10. Soki FN, Koh AJ, Jones JD, et al. Polarization of prostate cancer-associated macrophages is induced by milk fat globule-EGF factor 8 (MFG-E8)-mediated efferocytosis. J Biol Chem. 2014; 289(35):24560-24572. 11. Ong SM, Tan YC, Beretta O, et al. Macrophages in human colorectal cancer are pro-inflammatory and prime T cells towards an anti-tumour type-1 inflammatory response. Eur J Immunol. 2012; 42(1):89-100. 12. Jia L, Clear A, Liu FT, et al. Extracellular HMGB1 promotes differentiation of nurselike cells in chronic lymphocytic leukemia. Blood. 2014;123(11):1709-1719. 13. Hanna BS, McClanahan F, Yazdanparast H, et al. Depletion of CLL-associated patrolling monocytes and macrophages controls disease development and repairs immune dysfunction in vivo. Leukemia. 2015;30(3):570-579.

14. Galletti G, Scielzo C, Barbaglio F, et al. Targeting Macrophages Sensitizes Chronic Lymphocytic Leukemia to Apoptosis and Inhibits Disease Progression. Cell Rep. 2016;14(7):1748-1760. 15. O'Hayre M, Salanga CL, Kipps TJ, et al. Elucidating the CXCL12/CXCR4 signaling network in chronic lymphocytic leukemia through phosphoproteomics analysis. PLoS One. 2010;5(7):e11716. 16. Burkle A, Niedermeier M, Schmitt-Graff A, et al. Overexpression of the CXCR5 chemokine receptor, and its ligand, CXCL13 in B-cell chronic lymphocytic leukemia. Blood. 2007;110(9):3316-3325. 17. Till KJ, Lin K, Zuzel M, Cawley JC. The chemokine receptor CCR7 and alpha4 integrin are important for migration of chronic lymphocytic leukemia cells into lymph nodes. Blood. 2002;99(8):29772984. 18. Herishanu Y, Perez-Galan P, Liu D, et al. The lymph node microenvironment promotes B-cell receptor signaling, NF-kappaB activation, and tumor proliferation in chronic lymphocytic leukemia. Blood. 2011;117(2):563-574. 19. Burger JA, Quiroga MP, Hartmann E, et al. High-level expression of the T-cell chemokines CCL3 and CCL4 by chronic lymphocytic leukemia B cells in nurselike cell cocultures and after BCR stimulation. Blood. 2009;113(13):3050-3058. 20. Lebre MC, Vergunst CE, Choi IY, et al.

2075

M.H.A. van Attekum et al.

21.

22.

23.

24.

25.

2076

Why CCR2 and CCR5 blockade failed and why CCR1 blockade might still be effective in the treatment of rheumatoid arthritis. PLoS One. 2011;6(7):e21772. Audrito V, Serra S, Brusa D, et al. Extracellular nicotinamide phosphoribosyltransferase (NAMPT) promotes M2 macrophage polarization in chronic lymphocytic leukemia. Blood. 2015; 125(1):111-123. Smit LA, Hallaert DY, Spijker R, et al. Differential Noxa/Mcl-1 balance in peripheral versus lymph node chronic lymphocytic leukemia cells correlates with survival capacity. Blood. 2007;109(4):1660-1668. Kaufmann A, Salentin R, Gemsa D, Sprenger H. Increase of CCR1 and CCR5 expression and enhanced functional response to MIP-1 alpha during differentiation of human monocytes to macrophages. J Leukoc Biol. 2001;69(2):248-252. Nieto JC, Canto E, Zamora C, et al. Selective loss of chemokine receptor expression on leukocytes after cell isolation. PLoS One. 2012;7(3):e31297. Ghorpade A, Xia MQ, Hyman BT, et al. Role of the beta-chemokine receptors

26.

27.

28.

29.

30.

CCR3 and CCR5 in human immunodeficiency virus type 1 infection of monocytes and microglia. J Virol. 1998;72(4):33513361. Thivierge M, Parent JL, Stankova J, RolaPleszczynski M. Modulation of formyl peptide receptor expression by IL-10 in human monocytes and neutrophils. J Immunol. 1999;162(6):3590-3595. Da Roit F, Engelberts PJ, Taylor RP, et al. Ibrutinib interferes with the cell-mediated anti-tumor activities of therapeutic CD20 antibodies: implications for combination therapy. Haematologica. 2015;100(1):7786. White GE, Iqbal AJ, Greaves DR. CC chemokine receptors and chronic inflammation--therapeutic opportunities and pharmacological challenges. Pharmacol Rev. 2013;65(1):47-89. Mizutani K, Sud S, McGregor NA, et al. The chemokine CCL2 increases prostate tumor growth and bone metastasis through macrophage and osteoclast recruitment. Neoplasia. 2009;11(11):1235-1242. Mitchem JB, Brennan DJ, Knolhoff BL, et al. Targeting tumor-infiltrating macrophages

31.

32.

33. 34.

35.

decreases tumor-initiating cells, relieves immunosuppression, and improves chemotherapeutic responses. Cancer Res. 2013;73(3):1128-1141. Burgess M, Cheung C, Chambers L, et al. CCL2 and CXCL2 enhance survival of primary chronic lymphocytic leukemia cells in vitro. Leuk Lymphoma. 2012;53(10):19881998. De Cecco L, Capaia M, Zupo S, et al. Interleukin 21 Controls mRNA and MicroRNA Expression in CD40-Activated Chronic Lymphocytic Leukemia Cells. PLoS One. 2015;10(8):e0134706. Sica A, Allavena P, Mantovani A. Cancer related inflammation: the macrophage connection. Cancer Lett. 2008;267(2):204-215. Jitschin R, Braun M, Buttner M, et al. CLLcells induce IDOhi CD14+HLA-DRlo myeloid-derived suppressor cells that inhibit T-cell responses and promote TRegs. Blood. 2014;124(5):750-760. Rajasagi M, Shukla SA, Fritsch EF, et al. Systematic identification of personal tumor-specific neoantigens in chronic lymphocytic leukemia. Blood. 2014;124(3):453462.

haematologica | 2017; 102(12)

ARTICLE

Non-Hodgkin Lymphoma

Pattern of somatic mutations in patients with Waldenström macroglobulinemia or IgM monoclonal gammopathy of undetermined significance

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Marzia Varettoni,1 Silvia Zibellini,1 Irene Defrancesco,1 Virginia Valeria Ferretti,2 Ettore Rizzo,3 Luca Malcovati1,2, Anna Gallì,1 Matteo Giovanni Della Porta,4 Emanuela Boveri,5 Luca Arcaini,1,2 Chiara Candido,1 Marco Paulli5 and Mario Cazzola1,2

Department of Hematology Oncology, Fondazione IRCCS Policlinico San Matteo, Pavia; Department of Molecular Medicine, University of Pavia; 3enGenome srl, Pavia; 4Cancer Center, IRCCS Humanitas Research Hospital & Humanitas University, Milan, and 5 Anatomic Pathology Section, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy 1 2

Haematologica 2017 Volume 102(12):2077-2085

ABSTRACT

e analyzed MYD88 and CXCR4 mutation status of 260 patients with Waldenström macroglobulinemia or IgM monoclonal gammopathy of undetermined significance using allelespecific real time quantitative polymerase chain reaction and Sanger sequencing, respectively. A subgroup of 119 patients was further studied with next-generation sequencing of 11 target genes (MYD88, CXCR4, ARID1A, KMT2D, NOTCH2, TP53, PRDM1, CD79B, TRAF3, MYBBP1A, and TNFAIP3). MYD88 (L265P) was found at diagnosis in 91% of patients with Waldenström macroglobulinemia and in 60% of patients with IgM monoclonal gammopathy of undetermined significance using allele-specific polymerase chain reaction analysis. MYD88 mutations other than the classical L265P (V217F, S219C and M232T) were found in four cases by next-generation sequencing. Waldenström macroglobulinemia patients with wild-type MYD88 had a distinct clinical phenotype characterized by less bone marrow infiltration (P=0.01) and more frequent extramedullary involvement (P=0.001) compared to patients with mutated MYD88. Patients with wild-type MYD88 did not show additional mutations in the other target genes. CXCR4 mutations were found by Sanger sequencing in 22% of patients with Waldenström macroglobulinemia. With next-generation sequencing, a CXCR4 mutation was detected in 23% of patients with Waldenström macroglobulinemia and 9% of those with IgM monoclonal gammopathy of undetermined significance. Asymptomatic Waldenström macroglobulinemia patients harboring a CXCR4 mutation had a shorter treatment-free survival (51 months) than that of patients with wild-type CXCR4 (median not reached) (P=0.007). Analysis of variant allele frequencies indicated that CXCR4 mutations were present in the dominant clone in the majority of cases. Recurrent somatic mutations of KMT2D were found in 24% of patients with Waldenström macroglobulinemia and 5% of patients with IgM monoclonal gammopathy of undetermined significance and were primarily subclonal.

Introduction Waldenström macroglobulinemia (WM) is a rare lymphoproliferative disorder characterized by the presence of a serum IgM paraprotein associated with infiltration of the bone marrow by lymphoplasmacytic lymphoma.1 Familial aggregation of WM and related B-cell disorders strongly supports a role for genetic factors in the pathogenesis of the disease.2-4 haematologica | 2017; 102(12)

Correspondence: m.varettoni@smatteo.pv.it

Received: May 15, 2017. Accepted: September 28, 2017 Pre-published: October 5, 2017. doi:10.3324/haematol.2017.172718 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2077 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2077

M. Varettoni et al.

Some years ago, using whole genome sequencing Treon et al. identified MYD88 (L265P) as the most common somatic mutation in WM.5 Signaling studies showed that the mutant protein encoded by MYD88 triggers tumor growth through the activation of nuclear factor kappa light-chain enhancer of activated B cells (NF-kB) via two independent pathways, namely IRAK1-4 and BTK.6 The frequent recurrence of one single variant (i.e. L265P) in WM/lymphoplasmacytic lymphoma prompted the design of allele-specific polymerase chain reaction (PCR)-based strategies for the detection of the MYD88 mutation.7-8 Using this approach, the MYD88 (L265P) mutation is detectable in more than 90% of WM patients, while it is rarely expressed in other indolent mature B-cell tumors, such as splenic marginal zone lymphoma or chronic lymphocytic leukemia.7-10 The diagnosis of WM may be preceded by a history of IgM monoclonal gammopathy of undetermined significance (IgM MGUS). Long-term follow-up studies show that patients with IgM MGUS have a risk of progression to WM or to other B-cell lymphoproliferative disorders of approximately 1.5-2% per year.11 Using allele-specific PCR, 50-80% of patients with IgM MGUS were found to harbor the MYD88 (L265P) mutation, suggesting that MYD88 mutation is an early genetic event in the development of WM.7-10,12 In a previous study, we demonstrated that IgM MGUS patients harboring the MYD88 (L265P) mutation have a significantly higher risk of progression to WM or other lymphoproliferative disorder as compared with patients with the wild-type gene, regardless of the size of the serum IgM monoclonal protein.13 The second most common mutations in WM are nonsense or frameshift mutations in the carboxyl-terminal cytoplasmic tail of the CXCR4 gene. CXCR4 is overexpressed by cancer cells in many hematopoietic and solid cancers, but WM is the first human cancer in which somatic CXCR4 mutations have been reported.14 These mutations are similar to germline mutations typical of the WHIM syndrome, an inherited autosomal dominant disorder characterized by warts, hypogammaglobulinemia, infections and myelokathexis.15 In WM, somatic CXCR4 mutations result in impaired internalization of the CXCR4 receptor, leading to constitutive activation of the CXCR4 pathway, AKT and ERK activation and eventually WM cell survival.16-17 Using Sanger sequencing, CXCR4 mutations are detected in approximately 30% of WM patients, almost all of whom also harbor the MYD88 (L265P) mutation. CXCR4 mutations are usually subclonal, supporting the notion that they are acquired after the MYD88 (L265P) mutation in the development of WM.18 From a clinical standpoint, the presence of CXCR4 mutations has been associated with a more aggressive clinical presentation, including higher levels of IgM serum monoclonal protein, a higher incidence of hyperviscosity syndrome and more extensive bone marrow infiltration,19-21 as well as with clinical resistance to Ibrutinib.16-17,22 While there is an increasing body of data about MYD88 and CXCR4 mutations, little is known about the remaining genomic landscape of WM. Using whole genome sequencing, somatic mutations of ARID1A have been found in 17% of patients, while CD79B, KMT2D (formerly known as MLL2), MYBBP1A and TP53 mutations have been described in less than 10% of cases each. The aim of this study was to analyze the pattern of mutations of 11 genes in a cohort of well-annotated WM 2078

or IgM MGUS patients, and to evaluate correlations between somatic mutations and disease phenotype or patients’ outcome. The following genes were studied: MYD88, CXCR4, ARID1A, KMT2D, NOTCH2, TP53, PRDM1, CD79B, TRAF3, MYBBP1A, and TNFAIP3. In this study we demonstrated that MYD88 wild-type patients have a distinct clinical phenotype and do not show additional somatic mutations in the other target genes; furthermore, we demonstrated that CXCR4 mutations in WM patients are associated with an earlier need for treatment. Finally we report for the first time that subclonal KMT2D mutations are highly recurrent in WM patients.

Methods These investigations were approved by the local Ethics Committee. The procedures followed were in accordance with the Helsinki Declaration of 1975, as revised in 2000, and samples were obtained after subjects had provided informed consent.

Patients The study population included 130 patients with WM and 130 patients with IgM MGUS, in whom MYD88 and CXCR4 mutation status was evaluated with allele-specific real-time quantitative PCR (RT-qPCR) and Sanger sequencing, respectively. The methods are described in the Online Supplements. A subgroup of 119 patients (62 with WM and 57 with IgM MGUS) was further studied using next-generation sequencing (NGS) of 11 target genes, selected on the basis of data available in the literature. The patients’ characteristics are reported in Online Supplementary Table S1. The diagnoses of WM and IgM MGUS were made according to the consensus recommendations from the Second International Workshop on WM.1

Sample collection and cell separation Sample collection and cell separation are described in the Online Supplement.

Mutation analysis of target genes using next-generation sequencing Mutation analysis was performed on paired tumor/germline gDNA in order to identify relevant somatic mutations of selected genes. A TruSeq Custom Amplicon panel (TSCA; Illumina, San Diego, CA, USA) targeting complete coding exons and their adjacent splice junctions from the 11 genes was designed using Illumina Design Studio software. The TSCA panel consisted of 249 amplicons, 425 bp in length, for a total of 69 kb targeted DNA. Dual-barcoded TSCA libraries were created from 250 ng of high quality DNA according to the manufacturer’s protocol. Libraries were multiplexed and underwent 2x250-bp paired-end sequencing on a MiSeq sequencing system using the MiSeq Reagent Kit version 3 (Illumina). The resulting average depth of coverage for the 249 amplicons was 1009x. Sequence reads were initially aligned to the human genome (GRCh37/hg19) using the Burrows-Wheeler23 aligner. The Genome Analysis Toolkit24 (www.broadinstitute.org/gatk/) was later used to clean up reads and make alignment data more reliable for the variant calling (GATK data cleanup best practice): single nucleotide variants and small insertions-deletions were identified by Mutect25 and UnifiedGenotyper, respectively. Functionally annotated variants were filtered according to the following criteria: synonymous variants and variants located outside protein coding regions were filtered; polymorphisms described in dbSNP (verhaematologica | 2017; 102(12)

Somatic mutations in WM or IgM MGUS

sion 138) and the 1000 Genomes Project with a population frequency higher than 1% and 0.14%, respectively, were removed; variants with coverage <30X and less than ten supporting reads and variants with an allelic fraction lower than 1% were filtered; the remaining variants, evaluated as candidate somatic mutations, were manually reviewed and tagged as oncogenic using different criteria based on information retrieved from the literature, sequence conservation and in silico prediction of effect.

Statistics Quantitative variables are summarized in terms of median and range. Categorical variables are described by absolute and relative frequencies. The association between two categorical variables was estimated by the Fisher exact test. Quantitative variables were compared between two groups of patients using the Wilcoxon rank-sum test. The correlation between two quantitative variables was tested using the Spearman rho correlation. Overall survival was defined as the time from diagnosis to death from any cause or last follow-up (for censored patients) Treatment-free survival was measured (only in WM patients with mutational status evaluated at diagnosis) from the date of diagnosis to the date of first-line treatment or the last follow-up (for censored patients). Overall survival and treatment-free survival were estimated using the Kaplan-Meier product limit method and survival curves were compared by the log-rank test. The effect of variables on overall survival and treatment-free survival was evaluated using the proportional hazard Cox model. The agreement between NGS and PCR results and between methods was tested by the Cohen kappa coefficient. P-values less than 0.05 were considered statistically significant. All analyses were carried out using Stata 12.1 software (2011).

Results MYD88 mutation status and allele burden as determined by polymerase chain reaction Using allele-specific RT-qPCR, the MYD88 (L265P) mutation was found in 78 of 130 (60%) of patients with IgM MGUS and in 112 of 130 (86%) of those with WM. In WM, 96/106 untreated patients (91%) and 16/24 (67%) of previously treated patients were found to harbor the MYD88 (L265P) mutation. Among 24 previously treated WM patients, seven had wild-type MYD88. Two of them had progressive disease with a bone marrow infiltration of 50% and 70%. The other five MYD88 wild-type patients were in very good partial response and did not have detectable bone marrow infiltration at the time of analysis. Among the patients with an MYD88 mutation, the median allele burden was 1.05 (range, 0.1-18.4) in those with IgM MGUS and 15.4 (range, 0.1-96) in those with WM (P<0.001). Patients with symptomatic WM had a higher MYD88 allele burden as compared with asymptomatic ones (28.5 versus 13; P=0.05) (Figure 1). However, there were overlapping values without a clear cut-off between WM and IgM MGUS patients. The prevalence of MYD88 (L265P) mutation was not significantly higher in CD19+-selected bone marrow samples than in unselected samples either in IgM MGUS patients (62% versus 58%; P= 0.722) or in WM patients (87% versus 85%; P=0.802).

Correlations of MYD88 mutation status and MYD88 allele burden with disease phenotype Compared with WM patients harboring the MYD88 (L265P) mutation, patients with wild-type MYD88 had haematologica | 2017; 102(12)

significantly less bone marrow infiltration (median 18% versus 35%; P=0.01) and higher b2-microglobulin levels (median 3109 mg/L versus 2480 mg/L; P=0.03), borderline lower IgM levels (median 536 mg/dL versus 1170 mg/dL; P=0.07) and platelet counts (median 217 x 109/L versus 260 x 109/L; P=0.06), and more frequent extramedullary involvement (60% versus 12% of patients; P=0.001). No significant difference was found for age, sex or the other clinical characteristics. Among WM patients with the MYD88 (L265P) mutation, the median MYD88 allele burden was significantly higher in patients with hemoglobin levels <10 g/dL (P=0.01), b2-microglobulin levels >3000 mg/L (P<0.001), serum albumin levels <3.5 g/dL (P=0.01), detectable Bence-Jones proteinuria (P=0.04), abnormal free-light chain ratio (P=0.046) or bone marrow infiltration >30% (P=0.01) (Figure 2). No significant correlation was found between MYD88 allele burden and age, sex, platelet count, IgM levels, or presence of extramedullary disease.

CXCR4 mutations in patients with IgM monoclonal gammopathy of undetermined significance and Waldenstrรถm macroglobulinemia Thirty-four of 260 patients (13%) had a CXCR4 mutation. Nonsense and frameshift mutations accounted for 16 (47%) and 18 cases (53%), respectively, C1013G being the most common variant (14/34, 41% of all CXCR4 mutations). The frequency of CXCR4 mutations was 5/130 (4%) in IgM MGUS patients and 29/130 (22%) in WM patients (P<0.001). The prevalence of CXCR4 mutations was not statistically different between CD19+-selected and unsorted bone marrow samples (7% versus 2% in IgM MGUS patients; P=0.188 and 25% versus 19% in WM patients; P=0.519). In WM patients, the presence of a CXCR4 mutation was associated with a significantly higher degree of bone marrow infiltration (50% versus 30%; P=0.04) and higher MYD88 allele burden (24% versus 9%; P=0.01) (Online Supplementary Table S2). Taking into account that the threshold for mutant allele detection by Sanger sequencing is approximately 20%, we analyzed the prevalence of CXCR4 mutations according to this cutoff and found that the rate of CXCR4 mutations was 27% in patients with 20% or more bone marrow infiltration

Figure 1. MYD88 allele burden in patients with IgM monoclonal gammopathy of undetermined significance, asymptomatic and symptomatic Waldenstrรถm macroglobulinemia.

2079

M. Varettoni et al.

and 13% in patients with less than 20% bone marrow infiltration (P=0.179).

Prognostic impact of CXCR4 and MYD88 mutations The median follow-up of the whole series was 43 months (38.9 months for IgM MGUS patients and 50.4 months for WM patients). During the follow-up, three of 130 IgM MGUS patients (2%) and 26 of 81 asymptomatic WM patients (32%) progressed to symptomatic WM requiring treatment. Twenty-five of 260 patients (10%) have died. The median treatment-free survival was significantly shorter in asymptomatic WM patients harboring a CXCR4 mutation at diagnosis (median 51 months) than in those with wild type CXCR4 (median not reached) (P=0.007) (Figure 3). In multivariate analysis, CXCR4 mutation and size of serum protein were independent prognostic factors for progression from asymptomatic to symptomatic WM requiring therapy (Online Supplementary Table S3). MYD88 mutational status did not influence time to first treatment (P=0.19). Treatment-free survival was significantly shorter in asymptomatic WM patients harboring both MYD88 and CXCR4 mutations than in MYD88-mutated/CXCR4wild type patients (P=0.019), confirming that CXCR4 but not MYD88 mutations were associated with an earlier need for treatment (Figure 4). Overall survival was not affected by either MYD88 or CXCR4 mutational status (data not shown).

Somatic mutations identified by next-generation sequencing Overall, 88 of 119 patients studied with NGS (74%) had one or more somatic mutations in the 11 target genes. The median number of mutations was significantly higher in WM patients than in IgM MGUS patients (median 2 versus 1; P<0.001) and in those previously treated than in

untreated ones (median 2 versus 1; P<0.001). MYD88 mutations were found in 80/119 patients (67%), with a median variant allele frequency of 34.2% (range, 3.093.3%). The prevalence of MYD88 mutations determined by NGS was 85% in WM and 47% in IgM MGUS patients (P<0.001) (Figure 5). MYD88 mutations other than the classical L265P (n=76) were found in four patients and were V217F (n=2), S219C (n=1) and M232T (n=1). Fourteen patients who resulted MYD88 (L265P) wild-type by NGS were found to have the mutation by allele-specific RT-qPCR (K coefficient of concordance between NGS and PCR, 65%; P<0.001). Twelve of them had a diagnosis of IgM MGUS and two had WM. Of these latter two patients, one was untreated with predominant extramedullary disease and minimal bone marrow infiltration, the other was a previously treated patient who had achieved a very good partial response and did not have detectable bone marrow infiltration at the time of mutation status analysis. Using NGS a CXCR4 mutation was found in 19 patients (16%) with a median variant allele frequency of 22.5% (range, 4.2-49.8%). The K coefficient of concordance between NGS and Sanger sequencing was 89.8% (P<0.001). The prevalence of CXCR4 mutations was 23% in WM and 9% in IgM MGUS patients (P=0.047) (Figure 5). Of the two patients with a CXCR4 mutation who were MYD88-wild type according to NGS, one was found to harbor the MYD88 (L265P) mutation with a low allele burden (0.34%) using allele-specific RT-qPCR. The comparison of variant allele frequencies of CXCR4 and MYD88 mutations in individual patients indicated that CXCR4 mutations were present in the dominant clone in the majority of cases (Figure 6). NGS allowed the identification of somatic mutations in KMT2D (16% of patients), TP53 (8%), NOTCH2 (7%), PRDM1 (4%), ARID1A (3%), CD79B (3%), and TRAF3

Figure 2. Correlation of MYD88 allele burden with disease phenotype in patients with Waldenstrรถm macroglobulinemia.

2080

haematologica | 2017; 102(12)

Somatic mutations in WM or IgM MGUS

(1%). No mutations were found in MYBBP1A or TNFAIP3. Overall, the prevalence of these mutations was significantly lower in patients with either wild-type MYD88 or wild-type CXCR4 than in those with MYD88 and/or CXCR4 mutations (16% versus 41%; P=0.007) (Figure 5). The prevalence of KMT2D mutations was significantly higher in WM patients than in patients with IgM MGUS (24% versus 5%; P=0.002) while the prevalence of mutations in other genes was not statistically different according to diagnosis (P>0.3 for all comparisons). The analysis of variant allele frequency in patients harboring either a MYD88 or a KMT2D mutation showed that KMT2D mutations were primarily subclonal (Figure 7). TP53 mutations were found in three of 57 patients with IgM MGUS (5%) and six of 62 patients with WM (10%). The variant allele frequency of TP53 mutations was below 3.5% in IgM MGUS patients and untreated WM patients. In the four previously treated WM patients, the variant allele frequencies were 2.9%, 10.1%, 11.2% and 57.8%.

We did not find a statistically significant correlation of KMT2D, TP53, NOTCH2, PRDM1, ARID1A, CD79B, or TRAF3 mutations with overall survival or time to first treatment.

Discussion This study was conducted on two well-balanced subgroups of patients with an established diagnosis of IgM MGUS or WM. The study includes the largest series published so far of IgM MGUS patients screened for somatic mutations of genes with potential relevance to the pathogenesis of WM. Since progression from IgM MGUS to WM is likely a multi-step process, in which multiple genetic hits are required for progression from a pre-benign condition to an overt neoplastic disease, the analysis of mutation pattern in IgM MGUS patients is useful to understand whether a given mutation represents an early

Figure 3. Treatment-free survival of asymptomatic patients with Waldenstrรถm macroglobulinemia according to CXCR4 mutation status.

Figure 4. Treatment-free survival of asymptomatic patients with Waldenstrรถm macroglobulinemia according to MYD88 and CXCR4 mutational status.

haematologica | 2017; 102(12)

2081

M. Varettoni et al.

or late event during the oncogenic process starting from MGUS and ending with WM. In agreement with this model, the high prevalence of MYD88 mutations in IgM MGUS patients (60%) confirms that this mutation is an early event, while the low prevalence of CXCR4 and KMT2D mutations in IgM MGUS (<10%) suggests that these mutations usually occur later.

As expected, the MYD88 (L265P) mutation was confirmed as the most common somatic mutation in WM and IgM MGUS, in agreement with results reported by other groups.5,7-10,12-14 Among the patients with WM, the rate of MYD88 (L265P) mutation was significantly lower in previously treated patients than in untreated ones. The majority of WM patients who resulted MYD88 wild-type

Figure 5. Pattern of mutations observed in patients with (A) IgM monoclonal gammopathy of undetermined significance or (B) Waldenstrรถm macroglobulinemia.

Figure 6. Variant allele frequency of MYD88 and CXCR4 mutations as determined by next-generation sequencing.

2082

haematologica | 2017; 102(12)

Somatic mutations in WM or IgM MGUS

after therapy had been successfully treated with immunochemotherapy. The lower yield of clonal B cells following treatment with B-cell depleting agents may at least in part explain this finding, even though a baseline sample was not available to confirm this hypothesis. Samples studied with RT-qPCR for MYD88 (L265P) included either CD19+-selected or unselected bone marrow mononuclear cells. The rate of MYD88 (L265P) mutation was not significantly higher in sorted samples than in unsorted ones, either in IgM MGUS or in WM. This result was not surprising in WM patients, in line with a recent publication showing a high level of concordance in the rate of MYD88 (L265P) mutation between selected and unselected bone marrow cells,26 while it was less expected in the context of IgM MGUS. In CD19+-selected samples, we did not assess CD19+ clonality with light chain restriction, which would probably have increased the accuracy in detecting the MYD88 (L265P) mutation, especially among patients with IgM MGUS. In fact, using this approach the Salamanca group found the MYD88 (L265P) mutation in 87% of IgM MGUS patients, which is a higher rate than in our study as well as in other published series.9 Finally, we did not specifically evaluate the plasma cell compartment, in which the presence of the MYD88 (L265P) mutation has been previously reported.5,26 These observations indicate that the source of DNA may affect the rate of MYD88 mutations, and underline the need to define the best strategy for testing these mutations, balancing the gain in accuracy with time and costs required for CD19+ sorting and light chain restriction assessment. NGS allowed the identification of mutations other than the classical L265P variant in four patients. Among these, the M232T variant has been recently described in a WM patient,27 whereas the V217F and S219C had been previously reported in diffuse large B-cell lymphoma, activated B-cell subtype.28 Overall, there was a strong concordance between NGS and RT-qPCR findings, and discordant cases were mainly represented by IgM MGUS patients in whom the L265P mutation was found only with RTqPCR. This finding confirms the greater sensitivity of RT-

qPCR over NGS, which becomes evident when the B-cell clone is small. WM patients with wild-type MYD88 showed distinct clinical features, including lower IgM levels, less bone marrow infiltration and more frequent extramedullary disease as compared with patients harboring the MYD88 mutation. MYD88 wild-type patients were almost invariably CXCR4 wild-type and did not have additional mutations in the other genes studied with NGS. The existence of a small subgroup of WM patients not harboring the MYD88 (L265P) mutation has been increasingly recognized in recent studies. These patients seem to have lower response rates to the BTK inhibitor ibrutinib and a poorer outcome as compared with MYD88-mutated patients.19,22 Furthermore, MYD88 wild-type cases show lower expression of genes related to B-cell differentiation and a lower rate of IGH somatic hypermutation, suggesting that they constitute a distinct entity with respect to classical WM, possibly deriving from a B cell in an earlier stage of differentiation.9,29 Allele-specific RT-qPCR allowed the estimation of MYD88 allele burden. Although WM patients usually have higher allele burdens as compared with IgM MGUS patients, we could not identify a threshold that reliably distinguished these two conditions. In WM patients, the MYD88 allele burden was found to be a good surrogate marker of clinical disease burden. Although clinical variables associated with a higher MYD88 allele burden include some of the prognostic factors of the WM International Prognostic Scoring System (hemoglobin, albumin, b2-microglobulin levels), a higher MYD88 allele burden was not associated with worse survival. In this study we analyzed CXCR4 mutation status by means of Sanger sequencing and NGS. The threshold for mutant allele detection by Sanger sequencing is approximately 20% and the rate of CXCR4 mutations was twofold higher in patients with a bone marrow infiltration of 20% or more than in patients with a bone marrow infiltration of less than 20%. Given the greater sensitivity of

Figure 7. Variant allele frequency of MYD88 and KMT2D mutations as determined by next-generation sequencing.

haematologica | 2017; 102(12)

2083

M. Varettoni et al.

NGS, we expected to find a significantly higher rate of mutations with this technique than with Sanger sequencing, especially in patients with IgM MGUS in whom the B-cell clone is usually under the detection limits of Sanger sequencing. Indeed, less than 10% of IgM MGUS patients were found to harbor a CXCR4 mutation by NGS, suggesting that CXCR4 mutations are later events than the MYD88 (L265P) mutation in the pathogenesis of WM/lymphoplasmacytic lymphoma. A CXCR4 mutation was found in approximately a quarter of WM patients, in agreement with results reported by Poulain et al. in a series of WM patients studied with Sanger sequencing and NGS.20 Combining Sanger technology with allele-specific PCR for the most common variant S338X, CXCR4 mutations were found in up to 43% of WM patients, suggesting that differences among studies depend mainly on the sensitivity of the method used as well as the selection of patients.18 In a study published by Xu et al., CXCR4 mutations were shown by cancer cell fraction analysis to be primarily subclonal.18 Although in a targeted sequencing approach with DNA it is often difficult to determine unequivocally whether a mutation is clonal or subclonal, because of problems with determining the tumor cell content in the biopsy, lack of knowledge about aneuploidy of chromosomes with mutations, and potential variation in the efficiency of amplifying the wild-type and mutated alleles, our data seem to indicate that CXCR4 mutations pertain to the dominant MYD88-mutated clone in the majority of cases. Of interest, CXCR4 mutations were associated with an earlier need for treatment in patients with WM. In a study of 48 asymptomatic WM patients with a long follow-up, the cumulative probability of progression to symptomatic WM or other lymphoproliferative disorders was 59% at 5 years and 68% at 10 years. The main risk factors for progression were the extent of bone marrow infiltration by lymphoplasmacytic cells, the size of monoclonal protein and hemoglobin levels.30 In our study we performed a multivariate analysis including CXCR4 mutation status and the above reported clinical risk factors for progression. CXCR4 mutation status and the size of serum monoclonal protein were independent risk factors for progression from asymptomatic to symptomatic WM requiring therapy. Based on these findings, a model including molecular and clinical variables may refine prognostication of asymptomatic WM patients and potentially lead to the design of risk-adapted follow-up strategies. In the subgroup of patients studied with NGS we analyzed the occurrence of somatic mutations in other genes

References 1. Owen RG, Treon SP, Al-Katib A, et al. Clinicopathological definition of Waldenstrom’s macroglobulinemia: consensus panel recommendations from the Second International Workshop on Waldenstrom’s Macroglobulinemia. Semin Oncol. 2003;30(2):110-115. 2. Royer RH, Koshiol J, Giambarresi TR, Vasquez LG, Pfeiffer RM, McMaster ML. Differential characteristics of Waldenström macroglobulinemia according to patterns of

2084

with potential relevance in WM based on evidence from the literature. One of the main strengths of this study is that we could validate the somatic origin of the mutations as a matched control, represented by bone marrow CD19– mononuclear cells was available in all cases. We found that subclonal mutations of KMT2D are highly recurrent in WM patients and may also be present in IgM MGUS patients. KMT2D encodes a histone methyltransferase that targets the Lys-4 position of histone H3. Somatically acquired KMT2D mutations have been reported in about 90% of follicular lymphomas, 30% of diffuse large B-cell lymphomas, 15% of splenic marginal zone lymphomas, as well as in 10-15% of mantle cell lymphomas.31-35 Germline KMT2D mutations have been reported in 6070% of patients with Kabuki syndrome, an autosomal dominant disorder characterized by craniofacial, intellectual, and cardiac abnormalities as well as by recurrent infections. In these patients, KMT2D mutations are associated with dysregulation of terminal B-cell differentiation, leading to humoral immune deficiency and autoimmune complications.36 Abnormally low serum levels of IgA, IgG, or both have been observed in more than 50% of WM patients before treatment and remain low after treatment also in patients achieving a complete remission.37 Whether there is a relationship between KMT2D mutations and hypogammaglobulinemia or autoimmune phenomena commonly observed in WM patients deserves further investigations. Using NGS we found TP53 mutations in 10% of WM patients and in 5% of IgM MGUS patients; however, so far these mutations have not been seen to have a prognostic impact, probably because of the short follow-up of patients studied with NGS. Interestingly, Poulain et al. recently reported a similar prevalence of TP53 mutations in WM (11%) and showed that TP53 alterations were associated with shorter overall survival, independently of WM International Prognostic Scoring System score.38 In conclusion, this study adds more knowledge about the clinical and prognostic implications of MYD88 and CXCR4 mutations and reveals the presence of further somatic mutations potentially relevant to the pathogenesis of WM. Longitudinal studies with sequential evaluations during the course of the disease are needed to understand the clonal evolution underlying progression from IgM MGUS to WM and clonal dynamics under treatment. Acknowledgments This study was supported by Fondazione Cariplo & Regione Lombardia, Milan (grant ID 42916996) to MGDP.

familial aggregation. Blood. 2010;115(22): 4464-4471. 3. Kristinsson SY, Björkholm M, Goldin LR, McMaster ML, Turesson I, Landgren O. Risk of lymphoproliferative disorders among first- degree relatives of lymphoplasmacytic lymphoma/Waldenstrom macroglobulinemia patients: a population-based study in Sweden. Blood. 2008;112(8):3052-3056. 4. Treon SP, Hunter ZR, Aggarwal A, et al. Characterization of familial Waldenstrom’s macroglobulinemia. Ann Oncol. 2006;17(3): 488-494. 5. 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. 6. Yang G, Zhou Y, Liu X, et al. A mutation in MYD88 (L265P) supports the survival of lymphoplasmacytic cells by activation of Bruton tyrosine kinase in Waldenstrom macroglobulinemia. Blood. 2013;122(7): 1222-1232. 7. Varettoni M, Arcaini L, Zibellini S, et al. Prevalence and clinical significance of the MYD88 (L265P) somatic mutation in Waldenstrom’s macroglobulinemia and related lymphoid neoplasms. Blood.

haematologica | 2017; 102(12)

Somatic mutations in WM or IgM MGUS

2013;121(13):2522-2528. 8. Xu L, Hunter ZR, Yang G, et al. MYD88 L265P in Waldenstrom macroglobulinemia, immunoglobulin M monoclonal gammopathy, and other B-cell lymphoproliferative disorders using conventional and quantitative allele-specific polymerase chain reaction. Blood. 2013;121(11):2051-2058. 9. Jimenez C, Sebastian E, Chillon MC, et al. MYD88 L265P is a marker highly characteristic of, but not restricted to, Waldenstrom’s macroglobulinemia. Leukemia. 2013;27(8): 1722-1728. 10. Poulain S, Roumier C, Decambron A, et al. MYD88 L265P mutation in Waldenstrom macroglobulinemia. Blood. 2013;121(22): 4504-4511. 11. Kyle RA, Therneau TM, Rajkumar SV, et al. Long-term follow-up of IgM monoclonal gammopathy of undetermined significance. Blood. 2003;102(10):3759-3764. 12. Landgren O, Staudt L. MYD88 L265P somatic mutation in IgM MGUS. N Engl J Med. 2012;367(23):2255-2256. 13. Varettoni M, Zibellini S, Arcaini L, et al. MYD88 (L265P) mutation is an independent risk factor for progression in patients with IgM monoclonal gammopathy of undetermined significance. Blood. 2013;122(13): 2284-2285. 14. Hunter ZR, Xu L, Yang G, et al. The genomic landscape of Waldenstrom macroglobulinemia is characterized by highly recurring MYD88 and WHIM-like CXCR4 mutations, and small somatic deletions associated with B-cell lymphomagenesis. Blood. 2014;123 (11):1637-1646. 15. Hernandez PA, Gorlin RJ, Lukens JN, et al. Mutations in the chemokine receptor gene CXCR4 are associated with WHIM syndrome, a combined immunodeficiency disease. Nat Genet. 2003;34(1):70-74 16. Roccaro AM, Sacco A, Jimenez C, et al. C1013G/CXCR4 acts as a driver mutation of tumor progression and modulator of drug resistance in lymphoplasmacytic lymphoma. Blood. 2014;123(26):4120-4131. 17. Cao Y, Hunter ZR, Liu X, et al. The WHIMlike CXCR4(S338X) somatic mutation activates AKT and ERK, and promotes

haematologica | 2017; 102(12)

18.

19.

20.

21.

22.

23. 24.

25.

26.

27.

resistance to ibrutinib and other agents used in the treatment of Waldenstrom's macroglobulinemia. Leukemia. 2015;29(1): 169-176. Xu L, Hunter ZR, Tsakmaklis N, et al. Clonal architecture of CXCR4 WHIM-like mutations in Waldenström macroglobulinaemia. Br J Haematol. 2016;172(5):735-744. Treon SP, Cao Y, Xu L, Yang G, Liu X, Hunter ZR. Somatic mutations in MYD88 and CXCR4 are determinants of clinical presentation and overall survival in Waldenstrom macroglobulinemia. Blood. 2014;123(18):2791-2796. 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. Schmidt J, Federmann B, Schindler N, et al. MYD88 L265P and CXCR4 mutations in lymphoplasmacytic lymphoma identify cases with high disease activity. Br J Haematol. 2015;169(6):795-803. Treon SP, Tripsas CK, Meid K, et al. Ibrutinib in previously treated Waldenström's macroglobulinemia. N Engl J Med. 2015;372(15):1430-1440. Li H, Durbin R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics. 2010;26(5):589-595. DePristo MA, Banks E, Poplin R, et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet. 2011;43(5):491498. Cibulskis K, Lawrence MS, Carter SL, et al. Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nat Biotechnol. 2013;31(3):213219. Gustine J, Meid K, Xu L, Hunter ZR, Castillo JJ, Treon SP. To select or not to select? The role of B-cell selection in determining the MYD88 mutation status in Waldenström macroglobulinaemia. Br J Haematol. 2017;176(5):822-824. Treon SP, Xu L, Hunter Z. MYD88 Mutations and response to ibrutinib in Waldenström's macroglobulinemia. N Engl J Med. 2015;373(6):584-586.

28. Ngo VN, Young RM, Schmitz R, et al. Oncogenically active MYD88 mutations in human lymphoma. Nature. 2011;470(7332): 115-119. 29. Hunter ZR, Xu L, Yang G, et al. Transcriptome sequencing reveals a profile that corresponds to genomic variants in Waldenström macroglobulinemia. Blood. 2016;128(6):827-838. 30. Kyle RA, Benson JT, Larson DR, et al. Progression in smoldering Waldenstrom macroglobulinemia: long-term results. Blood. 2012;119(19):4462-4466. 31. Beà S, Valdés-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. 32. Morin RD, Mendez-Lago M, Mungall AJ, et al. Frequent mutation of histone-modifying genes in non-Hodgkin lymphoma. Nature. 2011;476(7360):298-303. 33. 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. 34. Rossi D, Trifonov V, Fangazio M, et al. The coding genome of splenic marginal zone lymphoma: activation of NOTCH2 and other pathways regulating marginal zone development. J Exp Med. 2012;209(9):15371551. 35. Zhang J, Jima D, Moffitt AB, et al. The genomic landscape of mantle cell lymphoma is related to the epigenetically determined chromatin state of normal B cells. Blood. 2014;123(19):2988-2996. 36. Lindsley AW, Saal HM, Burrow TA, et al. Defects of B-cell terminal differentiation in patients with type-1 Kabuki syndrome. J Allergy Clin Immunol. 2016;137(1):179-187. 37. Hunter ZR, Manning RJ, Hanzis C, et al. IgA and IgG hypogammaglobulinemia in Waldenström's macroglobulinemia. Haematologica. 2010;95(3):470-475. 38. Poulain S, Roumier C, Bertrand E, et al. TP53 mutation and its prognostic significance in Waldenstrom's macroglobulinemia. Clin Cancer Res. 2017 Jul 28. [Epub ahead of print]

2085

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Non-Hodgkin Lymphoma

Ferrata Storti Foundation

Haematologica 2017 Volume 102(12):2086-2096

Results and conclusions of the European Intergroup EURO-LB02 trial in children and adolescents with lymphoblastic lymphoma

Eva Landmann,1* Birgit Burkhardt,2* Martin Zimmermann,3 Ulrike Meyer,1 Wilhelm Woessmann,1 Wolfram Klapper,4 Grazyna Wrobel,5 ┼Angelo Rosolen,6 Marta Pillon,6 Gabriele Escherich,7 Andishe Attarbaschi,8 Auke Beishuizen,9 Karin Mellgren,10 Robert Wynn,11 Richard Ratei,12 Adriana Plesa,13 Martin Schrappe,14 Alfred Reiter,1 Christophe Bergeron,15 Catherine Patte16 and Yves Bertrand 15

Department of Pediatric Hematology and Oncology, Justus-Liebig-University, Giessen, Germany; 2Department of Pediatric Hematology and Oncology, Children's University Hospital, Münster, Germany; 3Department of Pediatric Hematology and Oncology, Hannover Medical School, Germany; 4Department of Hematopathology and Lymph Node Registry, University Hospital, Kiel, Germany; 5Department of Bone Marrow Transplantation, Children's Oncology and Hematology, Wroclaw Medical University, Poland; 6Clinica di Oncoematologia Pediatrica, Università di Padova, Italy; 7Clinic for Pediatric Hematology and Oncology, University Medical Center, Hamburg, Germany; 8Department of Pediatric Hematology and Oncology, St. Anna Children's Hospital, Medical University of Vienna, Austria; 9Department of Pediatric Hematology/Oncology, Erasmus MC - Sophia Children's Hospital, Rotterdam, the Netherlands and the Dutch Childhood Oncology Group, the Hague, the Netherlands; 10Department of Pediatric Oncology and Hematology, The Queen Silvia Children’s Hospital, Göteborg, Sweden; 11Central Manchester University Hospitals, Great Britain; 12Department of Hematology, Oncology and Tumor Immunology, Helios Klinikum, Berlin-Buch, Germany; 13Department of Hematopathology and Flow Cytometry, CHU, Lyon-HCL, France; 14Department of Pediatrics, Christian-Albrechts-University, Kiel, Germany; 15Institut d'Hematologie et d'Oncologie Pediatrique, Centre Léon Bérard and HCL, Claude Bernard University, Lyon, France and 16Department of Gustave Roussy, Villejuif, France 1

*EL and BB contributed equally to this work.

Correspondence: birgit.burkhardt@ukmuenster.de

Received: December 1, 2015. Accepted: September 28, 2017. Pre-published: October 5, 2017. doi:10.3324/haematol.2015.139162 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2086 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2086

Some results were presented at the 4th International Symposium on Childhood, Adolescent and Young Adult NonHodgkin Lymphoma, New York, NY, USA, Nov. 1-3, 2012. Part of the data were presented at the 14th International Conference on Malignant Lymphoma in Lugano, Switzerland, June 17-20, 2015.

ABSTRACT

n the European Intergroup EURO-LB02 trial, children and adolescents with lymphoblastic lymphoma underwent the non-Hodgkin lymphoma Berlin-Frankfurt-Münster protocol without prophylactic cranial radiotherapy. The primary aims of this trial were to test whether replacing prednisone with dexamethasone during induction increases event-free survival in the subgroups with T-cell lymphoblastic lymphoma and whether therapy duration could be reduced from 24 to 18 months (factorial design, randomizations). These questions could not be answered due to premature closure of the trial. Here we report on the secondary aims of the trial: whether the results of the NHLBFM90 study could be reproduced and evaluation of disease features and prognostic factors. Three hundred and nineteen patients (66 with precursor B-cell lymphoblastic lymphoma, 233 with T-cell lymphoblastic lymphoma, 12 with mixed phenotype, 8 not classifiable) were enrolled. In induction, 215 patients received prednisone and 104 patients received dexamethasone. The median follow-up was 6.8 years (range, 3.0-10.3). The 5-year event-free survival was 82±2% [12 toxic deaths, 5 secondary malignancies, 43 non-response/relapse (central nervous system n=9; all received prednisone during induction)]. The event-free survival rate was 80±5% for patients with precursor B-cell lymphoblastic lymphoma, 82±3% for those with T-cell lymphoblastic haematologica | 2017; 102(12)

Dexamethasone and prednisone in LBL

lymphoma, and 100% for patients with a mixed phenotype. During induction, significantly more grade III/IV toxicities were observed in patients receiving dexamethasone, resulting in significant treatment delays. The number of toxic deaths did not differ significantly. The only variable associated with outcome was performance status at diagnosis. The 90% event-free survival rate for patients with Tcell lymphoblastic lymphoma shown in study NHL-BFM90 was not replicated, mainly due to more toxic deaths and central nervous system relapses. Dexamethasone in induction may prevent central nervous system relapse more effectively than prednisone but produces a higher burden of toxicity. (#NCT00275106).

Introduction Although lymphoblastic lymphoma (LBL) is an orphan disease, it is the second most frequent non-Hodgkin lymphoma (NHL) observed in children and adolescents. Acute lymphoblastic leukemia (ALL)-type therapeutic strategies have been demonstrated to be efficacious in LBL. Treatment protocols derived from the Berlin-Frankfurt MĂźnster (BFM) group ALL strategy result in event-free survival rates of 75% to 90%. These survival rates are, however, achieved at the expense of considerable toxicity.1-7 On the other hand, children who do not respond to this treatment or who relapse after it still have an extremely poor chance of surviving.8 Thus, many questions regarding optimal treatment of childhood LBL remain to be clarified. Although LBL and ALL are biologically similar, they probably represent different diseases, as suggested by recent research,9-12 and advances achieved in the optimization of childhood ALL treatment cannot be adopted as LBL treatment strategies. In particular, meaningful parameters allowing risk-adapted therapies, such as chromosomal translocations and minimal residual disease monitoring,13,14 are not yet available for LBL. The lack of meaningful prognostic parameters for childhood LBL is mostly explained by the nature of the disease and the limited number of patients who are treated uniformly. In contrast to ALL, peripheral blood and bone marrow are limited sources of tumor cells. More extensive surgery to obtain appropriate tumor material for biological studies, may not be feasible given the often life-threatening condition of the patients at initial presentation. Thus, even the immunological classification of the lymphoma is not exact in some cases. The EURO-LB 02 study was the second inter-group clinical trial of the newly established European Intergroup Cooperation on Childhood and Adolescent NHL (EICNHL). The patients were diagnosed and classified according to a standardized work-up, including a central pathology review. The treatment strategy of the trial was adapted from the NHL-BFM90 study, with the omission of prophylactic cranial irradiation.3,15 The primary aims of the EUROLB02 study were to test, in a randomized manner, whether replacing prednisone with dexamethasone during the induction phase increases event-free survival (EFS) in the subgroup of patients with T-cell LBL (T-LBL) and whether therapy duration could be reduced from 24 to 18 months (factorial design). We could not answer these questions because the trial had to be closed prematurely due to a substantial number of toxic deaths. Here we report on the results of the secondary aims of this study: to determine whether the outstanding results haematologica | 2017; 102(12)

for patients with T-LBL in the NHL-BFM90 study could be reproduced in a large European inter-group trial, to assess features of the disease, and to evaluate prognostic factors. Furthermore, we provide comprehensive information on toxicity and specific caveats regarding this treatment strategy, which is currently the backbone of LBL therapy in many countries. The trial is registered at http://www.clinicaltrials.gov (clinicaltrials.gov identifier: #NCT00275106).

Methods Patients Patients <22 years old with newly diagnosed LBL were eligible for entry into the study. Patients with T-LBL (local assessment) were eligible for the first randomization. For participation in the second randomization, T-LBL had to have been confirmed by the national reference laboratory. Further exclusion criteria are listed in Online Supplement 2.

Diagnostic work-up The diagnosis of LBL was established by tumor biopsy and/or cytological and immunological examinations of malignant effusions. A review by national reference pathologists or national reference cytomorphology/immunophenotyping laboratories was requested for all patients. Minimal diagnostic requirements and staging procedures are outlined in Online Supplements 3 and 4.

Treatment plan and responses Chemotherapy was based on the NHL-BFM90 protocol (Table 1).5 Patients with non-T-cell LBL received standard treatment stratified according to stage and central nervous system (CNS) status, whereas the treatments were randomized in patients with T-LBL (Table 1 and Figure 1A,B). CNS-positive patients received cranial radiotherapy after re-intensification. Online Supplements 5 and 6 contain recommendations on infection prophylaxis and details regarding the randomization process. Response criteria are outlined in Online Supplement 7.

Study design and statistics EURO-LB02 was conducted by eight national/multinational cooperative study groups in 14 European countries. The primary endpoint for both study questions was EFS, computed as the time from randomization to date of the last follow-up visit or the first event, i.e., a non-response on day 33 (defined as >5% blasts in the bone marrow and/or blasts in the cerebrospinal fluid) and/or <35% tumor regression), relapse, secondary malignancy, or death from any cause. The power and sample size calculations are detailed in Online Supplement 8. 2087

E. Landmann et al.

Results Patients’ enrollment and premature trial closure When a fifth case of toxic death occurred among 115 patients with a follow-up duration of at least 30 weeks or a toxic death, patients were no longer enrolled in the study because the stopping rule was met. After an amendment to the protocol that included auditing of the participating centers, advice on the particular toxicity of the induction and re-induction phases, and setting P0 of the Wald sequential plan for monitoring the incidence of toxic deaths to 2% (Online Supplement 9), enrollment was re-opened 9 months after the initial termination. Enrollment was ultimately closed 2 years later, when the 12th case of toxic death occurred among 268 patients with a follow-up duration of at least 30 weeks or a toxic death. Up to that time point, 351 patients had been assessed for eligibility (Figure 1A), and 319 patients were eligible for study entry and included in the subsequent analyses.

Patients’ characteristics The patients’ characteristics are presented in Table 2. The diagnoses were T-LBL in 233 (73%) patients, precursor Bcell LBL (pB-LBL) in 66 patients (21%), mixed phenotype LBL in 12 patients (4%), and other constellations in eight patients, as specified in Table 2. In 30 patients (9%), the disease stage could not be precisely determined due to the lack of pre-therapeutic bone marrow collection and/or lumbar puncture. In 28 cases, this process was consistent with the recommendation to postpone invasive measures when a critical mediastinal tumor syndrome was present and to initiate treatment immediately. 2088

Treatment Consistent with the protocol, 35 of the 40 patients with stage I/II tumors received induction, consolidation M and maintenance therapy, whereas 279 patients with stage III, IV tumors or an unknown stage received additional reintensification therapy. Five of the 40 patients with stage Table 1. Treatment protocol.

Drug

60 mg/m2/day 12 mg*

Administered on days 1-7 1

Randomization 1**

Cytoreductive pre-phase Prednisone Methotrexate (it) Induction, phase Ia Prednisone (oral or iv)

Dose

60 mg/m²/day 8-28, then tapered over 3x3 days versus Dexamethasone (oral or iv) 10 mg/m²/day 8-28, then tapered over 3x3 days Vincristine (iv) 1.5 mg/m2 8, 15, 22, 29 (max 2 mg) Daunorubicin (iv over 1 h) 30 mg/m2 8, 15, 22, 29 E. coli asparaginase (iv over 1 h) 10,000 IU/m2 12, 15, 18, 21, 24, 27, 30, 33 Methotrexate (it) 12 mg* 12, 33† Induction, phase Ib Cyclophosphamide‡ (iv over 1 h) 1000 mg/m2 36, 64 Cytarabine (iv) 75 mg/m2 38-41, 45-48, 52-55, 59-62 6-Mercaptopurine (oral) 60 mg/m2 36-63 Methotrexate (it) 12 mg* 45, 59 14-day pause Consolidation, phase M 6-Mercaptopurine (oral) 25 mg/m2 1-56 Methotrexate§ 5 g/m2 8, 22, 36, 50 Methotrexate (it) 12 mg* 8, 22, 36, 50 14-day pause For stages III and IV only: re-intensification, phase IIa Dexamethasone (oral or iv) 10 mg/m2 1-21, then tapered over 3x3 days Vincristine (iv) 1.5 mg/m2 8, 15, 22, 29 (max 2 mg) Doxorubicin (iv over 1 h) 30 mg/m2 8, 15, 22, 29 E. coli asparaginase (iv over 1 h) 10,000 IU/m2 8, 11, 15, 18 For stages III and IV only: re-intensification, phase IIb Cyclophosphamide‡ (iv over 1 h) 1000 mg/m2 36 Cytarabine (iv) 75 mg/m2 38-41, 45-48 Methotrexate (it) 12 mg* 38, 45 6-Thioguanine (oral) 60 mg/m2 36-49 14-day pause Maintenance 6-Mercaptopurine (oral) 50 mg/m² Daily until 24 months of total treatment Methotrexate (oral) 20 mg/m² Weekly until 24 months of total treatment versus 6-Mercaptopurine (oral) 50 mg/m² Daily until 18 months of total treatment Methotrexate (oral) 20 mg/m² Weekly until 18 months of total treatment

Randomization 2§§

The secondary endpoints were overall survival and toxicity. Overall survival was calculated as the time from randomization to the date of death or the last follow-up visit. EFS and overall survival were estimated using the Kaplan-Meier method. Differences were compared with the log-rank test. Cumulative incidence functions for relapse/non-response and toxic death were constructed according to the method reported by Kalbfleisch and Prentice16 and were compared using Gray test.17 The multivariate analysis of prognostic factors was conducted using a Cox regression analysis with backward elimination.18 Toxicity was assessed after each treatment phase using the National Cancer Institute Common Toxicity Criteria (NCI-CTC).19 Differences in the distribution of individual parameters between subsets of patients were analyzed using the chi-square test or Fisher exact test for categorized variables and the Mann-Whitney U test for continuous variables. A P-value <0.05 was considered statistically significant. The incidence of toxic deaths was monitored with a Wald sequential plan assuming a toxic death rate of 1% as acceptable, based on previous NHL-BFM studies. In an amendment the stopping rule was set to accept a toxic death rate of 2%. Details of the toxic death monitoring procedure are provided in Online Supplement 9. The analysis was based on follow-up data available as of March 2014. The data were analyzed using SAS version 9.3 (SAS Institute, Cary, NC, USA). Signed informed consent was obtained. The study was performed after receiving approval from the responsible ethics committees and was conducted in accordance with the Declaration of Helsinki of 1975 and its revision in 2000.

iv, intravenous; it, intrathecal. *Doses were adjusted for children <3 years of age. **Randomization of patients with T-LBL only, with dexamethasone representing the experimental arm. †Additional doses were administered to central nervous system (CNS)-positive patients and for patients with blasts in cerebrospinal fluid and <5 cells/µL cerebrospinal fluid on days 18 and 27. In December 2007, the study was amended to state that for children in whom the CNS status was unknown, two additional doses should be administered on days 18 and 27. ‡With mesna. §Ten percent of the dose was administered over 30 min and the remaining 90% was administered as a 23.5-h continuous iv infusion. Leucovorin rescue: 30 mg/m2 iv at hour 42 and 15 mg/m2 iv at hours 48 and 54. The serum methotrexate levels should be <3 mmol/L at hour 36, <1 µmol/L at hour 42, and <0.4 mmol/L at hour 48 after initiation of the methotrexate infusion. §§Randomization of patients with T-LBL only, with an 18-month total treatment duration as the experimental arm.

haematologica | 2017; 102(12)

Dexamethasone and prednisone in LBL

I/II tumors also received re-intensification in violation of the protocol due to individual decisions. Of the 319 patients, 239 were eligible for the randomization treatment with prednisone or dexamethasone in induction phase Ia (Figure 1A). One hundred and eighty-six of these 239 patients were randomized; 88 were selected to receive prednisone and 98 were selected to receive dexam-

ethasone, two of whom received prednisone due to individual decisions. Of the 53 non-randomized patients, three received dexamethasone. Of the 80 patients who were not eligible for randomization, five were given dexamethasone based on individual decisions. Thus, a total of 215 patients received prednisone and 104 patients received dexamethasone in induction phase Ia.

Figure 1. Consort flow diagram and flow chart of the EURO-LB02 trial. (A) CONSORT flow diagram. *In 11 patients, the diagnosis of T-LBL was subsequently revised (precursor B-cell LBL, n=1; biphenotypic/bilineal LBL, n=5; T-NHL not further classified, n=3; undifferentiated lymphoma, n=2). Six of these patients were randomized. (B) Flow chart of the EURO-LB02 trial. Patients with precurcor B-cell-LBL received the standard arm. *Resulting in a total therapy duration of 18 or 24 months.

haematologica | 2017; 102(12)

2089

E. Landmann et al.

Outcomes With a median follow-up period of 6.8 years (range, 3.010.3), the 5-year EFS rate for the 319 eligible patients was 82±2%. Sixty events were reported (Table 2). Twelve patients (3.8%) died from toxicity. Four patients had not responded by day 33 of induction therapy, 39 patients relapsed and five developed secondary cancers. Secondary malignancies included: acute myeloid leukemia, myelodysplastic syndrome progressing to acute myeloid leukemia, colorectal adenocarcinoma, ALL, and Epstein-Barr virusassociated lymphoma. A local lack of response or relapse was the predominant failure in patients with T-LBL, whereas systemic relapse and testicular relapse were the predominant sites of failure in patients with pB-LBL. The median time from diagnosis to relapse was 12 months (range, 1-77 months) for patients

with T-LBL and 34 months (range, 3-70 months) for patients with pB-LBL. CNS relapses occurred early (at a median of 12 months from diagnosis; range, 3-27 months). All nine patients who had a CNS relapse had received prednisone during induction phase Ia. The 5-year overall survival rate was 87±2%. The causes of death in the 41 patients who died were disease progression or relapse in 27 patients, toxicity in 12 patients and secondary malignancies in two patients.

Toxicity Eight of the 215 patients given prednisone (3.7%) and four of the 104 patients (3.8%) given dexamethasone in induction phase Ia died due to toxicity. Five toxic deaths occurred during induction phase Ia (prednisone, n=2 and dexamethasone n=3). One patient died in induction phase

Table 2. Protocol patients: characteristics and events.

Total Male gender Age, range [years] Age, median [years] Age ≥10 years Stage I Stage II Stage III Stage IV BM-positive CNS-positive CNS type 3 (blasts detected but <5 cells/mL CSF) CNS not tested Stage not evaluable • No pre-therapeutic BM • No pre-therapeutic LP • No pre-therapeutic BM or LP B-symptoms Mediastinal mass Need for intensive care EVENTS Toxic death (n) Non-response (n) Relapse (n) • Local ± new • BM • CNS • BM and CNS • BM ± local ± new • Testes • Other Second cancer

Total

Immunophenotype T-cell

Precursor B-cell

Mixed phenotype

Other*

319 (100%) 229 (72%) 0.3-18.8 8.76 139 (44%) 11 (3%) 29 (9%) 176 (55%) 73 (23%) 67 (21%) 11 (3%) 9 (3%) 28† (9%) 30 (9%) 3 22‡ 5 82 (26%) 229 (72%) 35 (11%)

233 (73%) 176 (76%) 1.0-18.8 8.74 105 (76%) 8 150 49 44 9 7 24 26 3 18 5 71 216 34

66 (21%) 40 (17%) 0.3-17.2 8.43 22 (16%) 11 18 20 14 14 1 2 3 3 3 6 4 -

12 (4%) 8 (4%) 3.8-18.6 11.52 7 (5%) 3 2 7 6 1 2 4 1

8 5 4.1-18.4 14.35 5 4 3 3 1 1 1 3 5 -

12 (3.8%) 4 (1.3%) 39 (12.2%) 15 6 7 2 4 3 2 5

8 3 31 14 5 7 1 3 1 2

3 1 7 1 1 1 3 1 3

1 1 1 -

LP: lumbar puncture; CNS: central nervous system; CSF: cerebrospinal fluid; BM: bone marrow. *Immunophenotype not determined, n=2; no lineage-specific markers, n=2; FABL1 cytomorphology but mature B-cell phenotype, n=1; T-non-Hodgkin lymphoma not further classified (lymphoblastic nature could neither be proven nor excluded due to insufficient material), n=3. †Includes one patient in whom LP was not performed but who was BM-positive and was thus stage IV. Of these 28 patients, three were treated as CNS-positive (1 due to epidural involvement and 2 due to suspicious cells in the CSF on day 3 or 33 of treatment), 19 were treated as CNS-negative, and five received two additional methotrexate doses (it) during the induction phase; for one patient, there were no data on CNS treatment. Five events (all relapses) occurred in these 28 patients (1 CNS relapse, 1 BM relapse, 3 local relapses); the CNS relapse occurred in a patient treated as CNS-negative [the first of 11 scheduled methotrexate (it) treatments was omitted]. ‡Does not include the one patient in whom LP was not performed but who was BM-positive and was thus stage IV.

2090

haematologica | 2017; 102(12)

Dexamethasone and prednisone in LBL

Ib, two patients died during consolidation phase M, three patients died during re-induction phase IIa and one patient died during the maintenance phase. Detailed causes of death are provided in Table 3. The toxic death rate per study group ranged from 0% to 9.7%. Some of the participating study groups had not previously used the BFM strategy. The toxic death rate was higher in study groups that had not previously used the BFM strategy (4 toxic deaths among 78 patients, cumulative incidence of toxic death, 5±3%) than in the groups in which the BFM strategy had already been established prior to the EURO-LB 02 study (8 toxic deaths among 241 patients; cumulative incidence of toxic death, 3±1%); this difference was not statistically significant (P=0.46). A total of 65 non-fatal severe adverse events were reported in 51 patients. The number of such events did not differ significantly between patients receiving prednisone in induction Ia (19%, 41/215 patients) and those receiving dexamethasone (23%, 24/104 patients). The incidence of nonfatal severe adverse events also did not differ significantly among the patients randomized to receive either prednisone or dexamethasone in induction Ia (data not shown). Among all patients, 29 of the severe adverse events occurred during induction Ia [12 in 215 patients receiving prednisone (6%) and 17 in 104 patients receiving dexamethasone (16%)], three during induction Ib, ten during consolidation phase M, 16 during re-intensification phase IIa, four during re-intensification phase IIb and three during maintenance. Details are provided in Online Supplement 10. Grade III and IV NCI-CTC toxicities observed in each treatment phase, with the exception of the maintenance phase, are shown in Online Supplements 11 and 12. The most frequently reported toxicities were hematologic toxicity, coagulation problems, infection, and liver toxicity. Hematologic toxicity was the most frequent toxicity observed during all phases, with the exception of the cytoreductive prephase. Toxicity due to coagulation and

thrombosis most frequently occurred in induction phase Ia and re-intensification phase IIa. Regarding the corticosteroid administered in induction phase Ia, the following grade III and IV toxicities were reported significantly more frequently in patients given dexamethasone than in patients given prednisone: hematologic toxicity, infection, stomatitis, thrombosis, arrhythmia and peripheral neurotoxicity. Details are given in Online Supplements 11 and 12. In the subset of patients randomized to receive either prednisone or dexamethasone in induction phase Ia, only grade III/IV hematologic toxicity, infection and peripheral neurotoxicity occurred significantly more frequently in patients given dexamethasone than in those given prednisone (Online Supplement 13). In the total group and in the subset of randomized patients, the higher frequency of grade III/IV toxicities observed in patients receiving dexamethasone in induction phase Ia was associated with a significantly greater delay in the initiation of induction phase Ib (a median of 4 days for patients receiving dexamethasone in induction phase Ia and a median of 1 day for patients receiving prednisone in induction phase Ia; P=0.0003 for the total group, P=0.01 in the subset of randomized patients only; Online Supplement 14).

Prognostic factors Results of univariate analyses of variables with possible impacts on treatment outcomes in the entire group are presented in Table 4 and by immunophenotypic subgroup in Online Supplements 15 and 16. The patients’ immunophenotypes had no statistically significant impact on outcomes (Figure 2). The disease stage had no statistically significant impact on the entire group or on patients with T-LBL. In the subset of patients with pB-LBL, stage III disease was associated with a significant reduction in EFS, whereas the cumulative incidences of relapse or non-response were not significantly different (Online Supplement 16). The nine events occurring among the 20 patients with stage III pB-LBL were four relapses,

Table 3. Treatment-related mortality.

Treatment phase Induction Ia

TRM (n)

Steroid received at time of death

Age at diagnosis Immuno(years) phenotype

Prednisone Prednisone

2.1 15.8

T-LBL early B-LBL

Dexamethasone Dexamethasone Dexamethasone

2.2 2.5 9.7

T-LBL T-LBL T-LBL

9.1

pB-LBL

IIa

* * Dexamethasone* Dexamethasone* Dexamethasone*

9.3 0.7 9.9 11.0 11.6

pB-LBL pB-LBL T-LBL T-LBL T-LBL

IIb Maintenance

0 1

†

7.5

T-LBL

Cause of death Septic shock Intracerebral hemorrhage following sinus venous thrombosis Sepsis (S. aureus, Acinetobacter) Necrotizing adenovirus enteritis and ARDS Acute respiratory failure with acute pulmonary edema, coma, and cardiac arrest Enterovirus infection, interstitial pneumonia, myocarditis, and pontine myelinolysis Multi-organ failure Septic shock Septicemia Pulmonary aspergillosis and ARDS Mycotic infection of the lung and pulmonary hemorrhage Varicella infection

ARDS: acute respiratory distress syndrome; pB-LBL: precursor B-cell lymphoblastic lymphoma; T-LBL: T-cell lymphoblastic lymphoma; TRM: treatment-related mortality. *Received prednisone during induction phase Ia. †Received dexamethasone during induction phase Ia.

haematologica | 2017; 102(12)

2091

E. Landmann et al.

one non-response, one toxic death, and three secondary malignancies (myeloproliferative disease: n=1, EpsteinBarr virus associated lymphoma: n=1, and mixed preT/common-ALL: n=1). Performance status 5 (i.e., a need for intensive care treatment) at diagnosis before the start of treatment was associated with poor outcomes and a higher incidence of relapse or non-response in the entire group and in patients with T-LBL. Performance status at diagnosis was documented for 232 of 233 patients with T-LBL. Thirty-four of the 35 patients with performance status 5 had T-LBL. Their outcomes were significantly worse than those of the 198 patients with T-LBL and performance status 1-4: EFS 66±8% versus 85±3%, P=0.003; cumulative incidence of non-response/relapse 29±8% versus 11±2%, P=0.006. In the univariate analyses, none of the other factors examined had an impact on outcomes, either in the entire group or in the subgroups of patients with T-LBL or pBLBL. In the multivariate analysis (Cox regression with backward elimination) of the entire group, which included the variables immunophenotype, age (<10 versus ≥10 years), gender, stage (<III versus ≥III), bone involvement, bone marrow involvement, CNS involvement, lactate dehydro-

genase levels (< 2 versus ≥2x the normal value), B-symptoms, mediastinal mass, superior vena cava syndrome, pleural effusion, pericardial effusion, tumor lysis syndrome, impaired renal function, and performance status (<5 versus 5), only a performance status of 5 remained in the models for EFS (hazard ratio: 3.0; 95% confidence interval: 1.3-6.9; P=0.01) and for cumulative incidence of non-response/relapse (hazard ratio: 2.9; 95% confidence interval: 1.4-6.9; P=0.004). This finding was also true when the analysis was restricted to only patients with T-LBL: EFS hazard ratio 3.1; 95% confidence interval: 1.3-7.2; P=0.01, incidence of non-response/relapse hazard ratio 2.8; 95% confidence interval: 1.3-6.9; P=0.006). Patients who received dexamethasone in induction phase Ia had a lower cumulative incidence of non-response or relapse than patients who received prednisone, with the difference being of borderline statistical significance (Table 4). Regarding the site of failure, the cumulative incidence of CNS relapses at 5 years was 4±1% among the 215 patients who received prednisone in the induction phase compared to 0% in the 104 patients who received dexamethasone (P=0.03), whereas the cumulative incidences of non-CNS relapse or non-response at 5 years were 10±2% and 9±3%, respectively (P=0.45).

Table 4. Univariate analyses of prognostic factors in all patients.

Stage

Stage IV

Lactate dehydrogenase

Mediastinal tumor Pleural effusion Pericardial effusion Bone involvement B-symptoms (any) Performance status *

Initial complications, any**

I II III IV unknown BM+/CNSBM-/CNS+ BM+/CNS+ ≤2 UNL >2 UNL ≤4 UNL >4 UNL no yes no yes no yes no yes no yes 1 2 3 4 5 1 to 4 5 no yes

EFS (%)

SE (%)

11 29 176 73 30 58 6 5 203 110 285 28 90 229 165 154 233 86 286 33 237 82 100 97 52 34 35 283 35 158 161

100 86 78 88 83 86 83 100 81 85 82 86 82 81 83 81 84 78 83 76 83 80 86 80 88 82 66 84 66 82 83

0 6 3 4 7 5 15 0 3 3 2 7 3 4 3 3 2 4 2 7 2 4 3 4 4 7 8 2 8 3 3

P Non-response/ (log-rank test) relapse (cumulative incidence (%)) 0.19

0.69

0.24 0.62 0.80 0.66 0.18 0.51 0.59 0.055

0.0047 0.93

0 10 14 10 17 10 17 0 13 11 12 11 10 14 13 12 11 16 12 21 11 17 9 14 10 9 29 11 29 13 12

SE (%) P (Gray)

0 6 3 4 7 4 17 0 2 3 2 6 3 2 3 3 2 4 2 7 2 4 3 4 4 5 8 2 8 3 3

0.39

0.68

0.46 0.76 0.64 0.9 0.19 0.21 0.13 0.06

0.003 0.92

continued on the next page

2092

haematologica | 2017; 102(12)

Dexamethasone and prednisone in LBL

Discussion We report data from 319 children and adolescents suffering from LBL enrolled in the EURO-LB02 trial. We derived important conclusions and useful information from this large multi-group, multinational study, although neither randomized study questions could be answered in the con-

firmatory analysis because the trial had to be closed prematurely due to an excess of toxic deaths. Firstly, this trial was instrumental in establishing a large European network of inter-group collaborators in the field of childhood and adolescent NHL, which will foster future progress in the understanding and treatment of this orphan disease.12

Figure 2. The 5-year event-free survival (EFS, from diagnosis) of protocol patients with Tcell, precursor B-cell and biphenotypic lymphoblastic lymphoma. EFS: event-free survival SE: standard error. The median time to an event was 0.9 and 2.3 years (P=0.21) in patients suffering from T-LBL and pB-LBL, respectively.

continued from the previous page

Acute tumor lysis syndrome Impaired renal function Mediastinal tumor with respiratory impairment Vena cava syndrome Immunophenotype

Immunophenotype in stages III/IV only Age (years)

Gender age>10 age>15 Corticosteroid administered in Ia Treatment delay on day 8 – the beginning of phase Ib (days)

no yes no yes no yes no yes T pB mixed T pB mixed <10 ≥10 <15 ≥15 male female male female male female Pred Dex ≤34 >34

EFS (%)

SE (%)

304 13 308 9 183 134 285 32 233 66 12 199 34 9 180 139 286 33 229 90 99 40 24 9 215 104 227 74

82 77 82 89 81 83 82 78 82 80 100 82 67 100 81 83 82 78 82 81 84 82 78 78 80 86 84 86

2 12 2 10 3 3 2 7 3 5 0 3 8 0 3 3 2 7 3 4 4 6 9 14 3 3 2 4

P Non-response/ (log-rank test) relapse (cumulative incidence (%)) 0.63 0.60 0.79

0.52 0.27

0.64

0.59 0.72 0.65 0.57 0.92 0.11 0.55

13 15 13 0 13 13 13 6 14 11 0 13 18 0 12 13 12 19 14 9 15 8 22 11 15 9 12 10

SE (%) P (Gray) 2 10 2 0 2 3 2 4 2 4 0 2 7 0 2 3 2 7 2 3 4 4 9 11 2 3 2 3

0.76 0.24 0.94 0.23 0.33

0.40

0.86 0.41 0.27 0.44 0.51 0.09 0.48

EFS: event-free survival; SE: standard error; BM+,: bone marrow positive, BM-: bone marrow negative; CNS+: central nervous system positive; CNS-: central nervous system negative; UNL: upper normal limit; T-LBL: T-lymphoblastic lymphoma; pB-LBL: precursor B-cell lymphoblastic lymphoma; Pred: prednisone; Dex: dexamethasone;*i.e., performance status at diagnosis before start of treatment, with a performance status of 5 defined as need for admission to intensive care unit. Intensive Care Unit admission was at the discretion of the responsible physician; **i.e., acute tumor lysis syndrome and/or one or more of the following conditions: impaired renal function, mediastinal tumor with respiratory impairment, vena cava syndrome, cardiac insufficiency, paraplegia, life-threatening sepsis, life-threatening bleeding, and respiratory impairment due to pleural effusions.

haematologica | 2017; 102(12)

2093

E. Landmann et al.

The reference treatment arm of the EURO-LB02 trial was the protocol of the NHL-BFM90 study, except that preventive cranial irradiation was omitted.5 The outstanding result of a 5-year EFS of 90% for patients with T-LBL reported in the BFM group NHL-BFM90 trial5 was not replicated in the inter-group EURO-LB02 study. The lower 5-year EFS of 82% for patients with T-LBL in the EURO-LB02 trial was primarily due to higher rates of toxic death (0% in NHLBFM 90, 3.4% in EURO-LB02) and CNS relapse (cumulative incidence of CNS relapse at 5 years: 0% in NHL-BFM 90 and 3±1% in EURO-LB02; P=0.05), but the cumulative incidence of non-CNS relapse/non-response was comparable (cumulative incidence of non-CNS relapse/nonresponse at 5 years: 8±3% in NHL-BFM 90 and 10±2% in EURO-LB02; P=0.27). The higher incidence of toxic deaths in the EURO-LB02 trial compared to the previous NHL-BFM studies3,5 cannot be clearly ascribed to a distinct reason. At the premature stopping of the EURO-LB02 trial, the rate of toxic deaths in the 319 enrolled patients was 3.8% (no further toxic deaths occurred during follow-up) and was higher than expected from previous BFM group studies, since no toxic deaths were observed in the NHL-BFM90 trial5 and the rate of such deaths in the NHL-BFM95 trial was 1.3%.3 Infection was the most frequent cause of death, and induction phase Ia and re-intensification phase IIa were clearly the most dangerous phases of this therapy. According to several ALL studies, compared to prednisone, dexamethasone increases the toxicity of induction therapy.20-23 In the randomized ALL AIEOP-BFM 2000 trial, toxic deaths also occurred significantly more frequently in patients who were randomized to receive dexamethasone in the induction phase than in patients randomized to receive prednisone.24 In the EURO-LB02 study, there was no significant difference in the number of toxic deaths between patients receiving dexamethasone or prednisone in induction phase Ia, either in the number of toxic deaths observed during induction phase Ia itself or in the total number of toxic deaths. However, non-fatal grade III and IV toxicity occurred significantly more frequently in patients receiving dexamethasone, resulting in a significant delay in subsequent treatment phases. Some of the participating study groups had not previously used the BFM strategy. Although the toxic death rate was higher in those study groups than in the groups in which the BFM strategy had been established prior to the EUROLB02 study, this difference was not statistically significant. Nevertheless, familiarity with treatment protocols should be considered when planning further multi-group trials. An unexpectedly high number of CNS relapses was observed in the EURO-LB02 study; all of these relapses occurred in patients who had received prednisone during induction phase Ia. The prevention of CNS relapse remains an unsolved problem. The A5971 trial by the Children’s Oncology Group (COG)25 showed comparable outcomes when CNS prophylaxis depended on the frequent delivery of intrathecal methotrexate compared to high-dose methotrexate. Thus, additional intrathecal methotrexate applications might represent a promising approach. Additional high-dose methotrexate infusions are another promising approach. In the SIOP LMT96 trial,26 ten courses of high-dose methotrexate (3 g/m2) were administered. Only one of the 79 trial participants experienced a CNS relapse. The administration of additional 2094

high-dose methotrexate infusions during the maintenance phase has also been shown to treat CNS disease effectively in patients with T-LBL.7 Based on our data, the use of dexamethasone during induction phase Ia is more efficacious than prednisone in preventing CNS relapse, but is more toxic. The higher rate of infections observed in the dexamethasone group is of particular concern. This finding is consistent with the results from several,20-24,27 but not all,28,29 trials in patients with ALL. However, some studies have reported higher EFS and overall survival, despite higher rates of severe infection, in patients treated with dexamethasone than in patients treated with prednisone.23,27 An important point to consider is the dose equivalence between dexamethasone and prednisone with regard to anti-tumor efficacy and toxicity. Data regarding these drugs’ relative anti-leukemic activity are inconsistent, and data on their toxic potential are lacking.30,31 In animal models, the cerebrospinal fluid penetration of dexamethasone is greater than that of prednisone.32 Thus, a dose of dexamethasone lower than 10 mg/m2/day, which was used in this study, might be less toxic but more efficacious in protecting against CNS toxicity, as has been shown in children suffering from ALL.29 A shorter duration of dexamethasone treatment might be another reasonable option for taking advantage of the efficacy of dexamethasone at a tolerable toxicity. The administration of dexamethasone for 2 weeks versus prednisone for 4 weeks in patients with high-risk precursorB-ALL during induction therapy was recently shown to result in better efficacy of dexamethasone in patients younger than 10 years, but not in patients aged 10 years or older. The rate of CNS relapses was lower in patients treated with dexamethasone.33 In the EURO-LB02 study, patients with stage I or II disease did not receive re-intensification therapy. Given the biological similarities between LBL and ALL, we reasonably asked whether patients with LBL would also benefit from delayed intensification therapy, like patients with ALL.34 Notably, the 5-year EFS rates for patients with stage I/II disease were 90% in both the EURO-LB02 study (5-year EFS, 90±5%, n=40) and the COG A5971 study35 (5-year EFS, 90%, n=51). However, delayed intensification therapy for patients with stage I/II disease was only administered in the COG A5971 study, resulting in a higher cumulative dose of anthracycline and cyclophosphamide. Improving the outcome for pediatric/adolescent patients with LBL remains a challenge. The toxic death rate of 3.8% in our study indicates a critical balance between treatment efficacy and risks. However, the salvage of relapsed patients has been shown to be poor.1,8,25 Thus, treatment optimization involves finding highly predictive markers that enable the early identification of patients who are not curable with current treatments. These patients could be designated to alternative treatment strategies, whereas all others could be protected from the risk of experimental treatments. We performed a comprehensive analysis of variables with possible impacts on outcomes in this large group of children who were classified and treated uniformly. Apart from localized versus advanced stage, the only variable associated with decreased EFS was a poor performance status at diagnosis. Researchers have not clearly determined whether a poor performance status reflects disease with a more aggressive biology or difficulties in treatment realization. In any case, this parameter is not very useful for risk stratification. Nevertheless, this finding might alert physihaematologica | 2017; 102(12)

Dexamethasone and prednisone in LBL

cians to pay special attention to these patients. The EURO-LB02 study contributed reliable up-to-date information on the classification of patients with LBL. A central pathology review was performed in as many as 90.4% of the cases. Additional detailed immunohistochemical analyses of this large series were performed by the European Childhood Lymphoma Panel. The results enabled the development of a step-wise and material-sparing diagnostic approach as a complementary strategy to the strategy recommended by the World Health Organization.36 Based on this reliable classification, we also showed that the immunophenotype lineage had no additional impact on outcome. A novel observation from the EURO-LB02 study was that lineage promiscuity was present in as many as 4% of all patients diagnosed with LBL. Given the aggressive course observed in children with leukemia of an ambiguous lineage,37 the outcomes of the 12 patients with mixed phenotype LBL in the EURO-LB02 study were rather favorable, with no events observed among this subgroup. The prognostic power of new biomarkers must be evaluated for the standardized classification and treatment of patients. NOTCH1 and FBXW7 mutations38,39 and loss of heterozygosity at chromosome 6q38,40 were recently shown to be promising prognostic biomarkers of pediatric T-LBL. Other candidates for prospective evaluation may be parameters of the kinetics of the response to treatment, such as

References 1. Abromowitch M, Sposto R, Perkins S, et al. Shortened intensified multi-agent chemotherapy and non-cross resistant maintenance therapy for advanced lymphoblastic lymphoma in children and adolescents: report from the Children's Oncology Group. Br J Haematol. 2008;143(2):261-267. 2. Amylon MD, Shuster J, Pullen J, et al. Intensive high-dose asparaginase consolidation improves survival for pediatric patients with T cell acute lymphoblastic leukemia and advanced stage lymphoblastic lymphoma: a Pediatric Oncology Group study. Leukemia. 1999;13(3):335-342. 3. Burkhardt B, Woessmann W, Zimmermann M, et al. Impact of cranial radiotherapy on central nervous system prophylaxis in children and adolescents with central nervous system-negative stage III or IV lymphoblastic lymphoma. J Clin Oncol. 2006;24(3):491499. 4. Pillon M, Piglione M, Garaventa A, et al. Long-term results of AIEOP LNH-92 protocol for the treatment of pediatric lymphoblastic lymphoma: a report of the Italian Association of Pediatric Hematology and Oncology. Pediatr Blood Cancer. 2009;53(6):953-959. 5. Reiter A, Schrappe M, Ludwig WD, et al. Intensive ALL-type therapy without local radiotherapy provides a 90% event-free survival for children with T-cell lymphoblastic lymphoma: a BFM group report. Blood. 2000;95(2):416-421. 6. Sandlund JT, Pui CH, Zhou Y, et al. Effective treatment of advanced-stage childhood lymphoblastic lymphoma without prophylactic cranial irradiation: results of St Jude NHL13

haematologica | 2017; 102(12)

those determined by positron emission tomography and minimal disseminated disease and its monitoring during treatment.41,42 The network of collaborators established in the EURO-LB02 study may provide a platform for this purpose. Acknowledgments An independent Data Safety and Monitoring Committee (DSMC) monitored the progress of this study and received reports on the current status of the trial every 6 months. The authors would like to thank the members of the external DSMC, P. Brice (Paris, France), D. Hasenclever (Leipzig, Germany), M. Hudson (Memphis, USA), M. Link (Palo Alto, USA), M. Parmar (London, UK), and D. Poplack (Houston, USA), for their critical review of the study progress and their advice. This work was supported by the Deutsche Krebshilfe (grants 102595 and 107813) (ARe), PHRC 2003 (national grant) (YB) and the Kinderkrebsinitiative, Buchholz, Holm-Seppensen (KKI) (WK). The authors would like to thank the European Intergroup Cooperation for Childhood Non-Hodgkin Lymphoma for its participation in this study. A complete membership list appears in the â&#x20AC;&#x153;Online Supplementâ&#x20AC;?. The authors would also like to thank the doctors, nurses and data managers at the participating hospitals who cared for these sick children and supplied data. Finally, the authors would particularly like to thank the children and their caregivers for participating in the trial.

study. Leukemia. 2009;23(6):1127-1130. 7. Uyttebroeck A, Suciu S, Laureys G, et al. Treatment of childhood T-cell lymphoblastic lymphoma according to the strategy for acute lymphoblastic leukaemia, without radiotherapy: long term results of the EORTC CLG 58881 trial. Eur J Cancer. 2008;44(6):840-846. 8. Burkhardt B, Reiter A, Landmann E, et al. Poor outcome for children and adolescents with progressive disease or relapse of lymphoblastic lymphoma: a report from the Berlin-Frankfurt-Muenster group. J Clin Oncol. 2009;27(20):3363-3369. 9. Bonn BR, Huge A, Rohde M, et al. Whole exome sequencing hints at a unique mutational profile of paediatric T-cell lymphoblastic lymphoma. Br J Haematol. 2015;168 (2):308-13. 10. Burkhardt B, Moericke A, Klapper W, et al. Pediatric precursor T lymphoblastic leukemia and lymphoblastic lymphoma: differences in the common regions with loss of heterozygosity at chromosome 6q and their prognostic impact. Leuk Lymphoma. 2008;49(3):451-461. 11. Feng H, Stachura DL, White RM, et al. Tlymphoblastic lymphoma cells express high levels of BCL2, S1P1, and ICAM1, leading to a blockade of tumor cell intravasation. Cancer Cell. 2010;18(4):353-366. 12. Minard-Colin V, Brugieres L, Reiter A, et al. Non-Hodgkin lymphoma in children and adolescents: progress through effective collaboration, current knowledge, and challenges ahead. J Clin Oncol. 2015;33(27): 2963-2974. 13. Conter V, Bartram CR, Valsecchi MG, et al. Molecular response to treatment redefines all prognostic factors in children and adolescents with B-cell precursor acute lymphoblastic leukemia: results in 3184 patients

14.

15.

16. 17. 18. 19.

20.

21.

22.

of the AIEOP-BFM ALL 2000 study. Blood. 2010;115(16):3206-3214. Schrappe M, Valsecchi MG, Bartram CR, et al. Late MRD response determines relapse risk overall and in subsets of childhood T-cell ALL: results of the AIEOPBFM-ALL 2000 study. Blood. 2011;118(8): 2077-2084. Reiter A, Schrappe M, Tiemann M, et al. Improved treatment results in childhood Bcell neoplasms with tailored intensification of therapy: a report of the Berlin-FrankfurtMunster group trial NHL-BFM 90. Blood. 1999;94(10):3294-3306. Kalbfleisch JD, Prentice RL. The Statistical Analysis of Failure Time Data. Wiley, New York, 1980. Gray RJ. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat. 1988;16(3):1141-1154. Cox DR. Regression models and life-tables. J R Stat Soc B. 1972;34(2):187-220. Common Toxicity Criteria V 2, 1999 https://ctep.cancer.gov/protocolDevelopme nt/electronic_applications/docs/ctcv20_430-992.pdf NC. De Moerloose B, Suciu S, Bertrand Y, et al. Improved outcome with pulses of vincristine and corticosteroids in continuation therapy of children with average risk acute lymphoblastic leukemia (ALL) and lymphoblastic non-Hodgkin lymphoma (NHL): report of the EORTC randomized phase 3 trial 58951. Blood. 2010;116(1):36-44. Hurwitz CA, Silverman LB, Schorin MA, et al. Substituting dexamethasone for prednisone complicates remission induction in children with acute lymphoblastic leukemia. Cancer. 2000;88(8):1964-1969. Igarashi S, Manabe A, Ohara A, et al. No advantage of dexamethasone over prednisolone for the outcome of standard- and

2095

E. Landmann et al.

23.

24.

25.

26.

27.

28.

2096

intermediate-risk childhood acute lymphoblastic leukemia in the Tokyo Children's Cancer Study Group L95-14 protocol. J Clin Oncol. 2005;23(27):6489-6498. Vrooman LM, Stevenson KE, Supko JG, et al. Postinduction dexamethasone and individualized dosing of Escherichia coli Lasparaginase each improve outcome of children and adolescents with newly diagnosed acute lymphoblastic leukemia: results from a randomized study--Dana-Farber Cancer Institute ALL Consortium Protocol 00-01. J Clin Oncol. 2013;31(9):1202-1210. Moricke A, Zimmermann M, Valsecchi MG, et al. Dexamethasone vs prednisone in induction treatment of pediatric ALL: results of the randomized trial AIEOP-BFM ALL 2000. Blood. 2016;127(17):2101-2112. Termuhlen AM, Smith LM, Perkins SL, et al. Disseminated lymphoblastic lymphoma in children and adolescents: results of the COG A5971 trial: a report from the Children's Oncology Group. Br J Haematol. 2013;162 (6):792-801. Bergeron C, Coze C, Segura C, et al. Treatment of childhood T-cell lymphoblastic lymphoma-long-term results of the SFOP LMT96 trial. Pediatr Blood Cancer. 2015;62(12):2150-2156. Schrappe M, Zimmermann M, Moricke A, et al. Dexamethasone in induction can eliminate one third of all relapses in childhood acute lymphoblastic leukemia (ALL): results of an international randomized trial in 3655 patients (Trial AIEOP-BFM ALL 2000). ASH Annual Meeting Abstracts. 2008;112(11):7. Bostrom BC, Sensel MR, Sather HN, et al. Dexamethasone versus prednisone and daily oral versus weekly intravenous mercaptopurine for patients with standard-risk acute lymphoblastic leukemia: a report from the Children's Cancer Group. Blood.

2003;101(10):3809-3817. 29. Mitchell CD, Richards SM, Kinsey SE, Lilleyman J, Vora A, Eden TO. Benefit of dexamethasone compared with prednisolone for childhood acute lymphoblastic leukaemia: results of the UK Medical Research Council ALL97 randomized trial. Br J Haematol. 2005;129(6):734-745. 30. Ito C, Evans WE, McNinch L, et al. Comparative cytotoxicity of dexamethasone and prednisolone in childhood acute lymphoblastic leukemia. J Clin Oncol. 1996;14(8):2370-2376. 31. Kaspers GJ, Veerman AJ, Popp-Snijders C, et al. Comparison of the antileukemic activity in vitro of dexamethasone and prednisolone in childhood acute lymphoblastic leukemia. Med Pediatr Oncol. 1996;27(2):114-121. 32. Balis FM, Lester CM, Chrousos GP, Heideman RL, Poplack DG. Differences in cerebrospinal fluid penetration of corticosteroids: possible relationship to the prevention of meningeal leukemia. J Clin Oncol. 1987;5(2):202-207. 33. Larsen EC, Devidas M, Chen S, et al. Dexamethasone and high-dose methotrexate improve outcome for children and young adults with high-risk B-acute lymphoblastic leukemia: a report from Children's Oncology Group study AALL0232. J Clin Oncol. 2016;34(20):23802388. 34. Tubergen DG, Gilchrist GS, O'Brien RT, et al. Improved outcome with delayed intensification for children with acute lymphoblastic leukemia and intermediate presenting features: a Childrens Cancer Group phase III trial. J Clin Oncol. 1993;11(3):527-537. 35. Termuhlen AM, Smith LM, Perkins SL, et al. Outcome of newly diagnosed children and adolescents with localized lymphoblastic lymphoma treated on Children's Oncology

36.

37.

38.

39.

40.

41.

42.

Group trial A5971: a report from the Children's Oncology Group. Pediatr Blood Cancer. 2012;59(7):1229-1233. Oschlies I, Burkhardt B, Chassagne-Clement C, et al. Diagnosis and immunophenotype of 188 pediatric lymphoblastic lymphomas treated within a randomized prospective trial: experiences and preliminary recommendations from the European childhood lymphoma pathology panel. Am J Surg Pathol. 2011;35(6):836-844. Gerr H, Zimmermann M, Schrappe M, et al. Acute leukaemias of ambiguous lineage in children: characterization, prognosis and therapy recommendations. Br J Haematol. 2010;149(1):84-92. Bonn BR, Rohde M, Zimmermann M, et al. Incidence and prognostic relevance of genetic variations in T-cell lymphoblastic lymphoma in childhood and adolescence. Blood. 2013;121(16):3153-3160. Callens C, Baleydier F, Lengline E, et al. Clinical impact of NOTCH1 and/or FBXW7 mutations, FLASH deletion, and TCR status in pediatric T-cell lymphoblastic lymphoma. J Clin Oncol. 2012;30(16):1966-1973. Burkhardt B, Bruch J, Zimmermann M, et al. Loss of heterozygosity on chromosome 6q14-q24 is associated with poor outcome in children and adolescents with T-cell lymphoblastic lymphoma. Leukemia. 2006;20(8): 1422-1429. Coustan-Smith E, Sandlund JT, Perkins SL, et al. Minimal disseminated disease in childhood T-cell lymphoblastic lymphoma: a report from the Children's Oncology Group. J Clin Oncol. 2009;27(21):3533-3539. Mussolin L, Buldini B, Lovisa F, et al. Detection and role of minimal disseminated disease in children with lymphoblastic lymphoma: the AIEOP experience. Pediatr Blood Cancer. 2015;62(11):1906-1913.

haematologica | 2017; 102(12)

ARTICLE

Non-Hodgkin Lymphoma

Dose-adjusted EPOCH chemotherapy for untreated peripheral T-cell lymphomas: a multicenter phase II trial of West-JHOG PTCL0707

Yoshinobu Maeda,1 Hisakazu Nishimori,1 Isao Yoshida,2 Yasushi Hiramatsu,3 Masatoshi Uno,4 Yasufumi Masaki,5 Kazutaka Sunami,6 Taro Masunari,7 Yuichiro Nawa,8 Hiromichi Yamane,9 Hiroshi Gomyo,10 Tsutomu Takahashi,11 Tomofumi Yano,12 Keitaro Matsuo,13 Koichi Ohshima,14 Shigeo Nakamura,15 Tadashi Yoshino16 and Mitsune Tanimoto1

Department of Hematology and Oncology, Okayama University Hospital; 2Department of Hematologic Oncology, Shikoku Cancer Center, Ehime; 3Department of Hematology and Oncology, Japanese Red Cross Society Himeji Hospital, Hyogo; 4Department of Internal Medicine, Kaneda Hospital, Okayama; 5Department of Hematology and Immunology, Kanazawa Medical University Hospital, Ishikawa; 6Department of Hematology, National Hospital Organization Okayama Medical Center; 7Department of Hematology, Chugoku Central Hospital, Hiroshima; 8Department of Hematology, Ehime Prefectural Central Hospital, Ehime; 9Department of Internal Medicine, Sumitomo Besshi Hospital, Ehime; 10 Department of Hematology, Hyogo Cancer Center, Hyogo; 11Department of Hematology and Oncology, Shimane University Hospital; 12Department of Hematology, Okayama Rosai Hospital; 13Division of Molecular and Clinical Epidemiology, Aichi Cancer Center Research Institute, Nagoya; 14Department of Pathology, School of Medicine, Kurume University, Fukuoka; 15Department of Pathology and Clinical Laboratories, Nagoya University Hospital, and 16Department of Pathology, Okayama University Graduate School of Medicine, Dentistry, and Pharmaceutical Sciences, Japan 1

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(12):2097-2103

ABSTRACT

he standard CHOP therapy for peripheral T-cell lymphoma has resulted in unsatisfactory outcomes and it is still not clear what is the optimal front-line therapy. We conducted a multicenter phase II study of dose-adjusted etoposide, doxorubicin, and cyclophosphamide with vincristine and prednisone (EPOCH) for untreated peripheral T-cell lymphoma patients. In this prospective study, 41 patients were treated with dose-adjusted-EPOCH as initial therapy: peripheral T-cell lymphoma-not otherwise specified, n=21; angioimmunoblastic T-cell lymphoma, n=17; anaplastic lymphoma kinase-positive anaplastic large cell lymphoma, n=2; and anaplastic lymphoma kinase-negative anaplastic large cell lymphoma, n=1. Median patient age was 64 years (range: 3279 years). According to the International Prognostic Index criteria, 51.2% were at high-intermediate or high risk. The overall response and complete response rates were 78.0% [95% confidence interval (CI): 62.4-89.4%] and 61.0% (95%CI: 44.5-75.8%), respectively. At the median follow up of 24.0 months, the 2-year progression-free survival and overall survival were 53.3% (95%CI: 36.4-67.5%) and 73.2% (95%CI: 56.8-84.1%), respectively. The younger patients (≤ 60 years old) had a high response rate (overall response 94.1% and complete response 70.6%) and survival rate (progression-free survival 62.5% and overall survival 82.4%). The most common grade ≥ 3 adverse events were neutropenia (74.5%), anemia (40.8%), thrombocytopenia (22.0%), and febrile neutropenia (9.0%). Dose-adjusted-EPOCH had a high response rate with a tolerable toxicity profile. Our results indicate that doseadjusted-EPOCH is a reasonable first-line approach for peripheral T-cell lymphoma patients and may improve outcomes. (UMIN trial registration number: UMIN000000829).

Introduction Peripheral T-cell lymphomas (PTCLs) are rare, heterogeneous diseases that comprise 10-15% of all adult non-Hodgkin lymphoma (NHL) cases. PTCLs have been classified into four groups according to the World Health Organization (WHO) classification syshaematologica | 2017; 102(12)

Correspondence: yosmaeda@md.okayama-u.ac.jp

Received: March 10, 2017. Accepted: September 20, 2017. Pre-published: September 29, 2017. doi:10.3324/haematol.2017.167742 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2097 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2097

Y. Maeda et al.

tem (2008), and the most common subtypes are nodal T-cell lymphomas, PTCL-not otherwise specified (PTCL-NOS), angioimmunoblastic T-cell lymphoma (AITL), anaplastic lymphoma kinase (ALK)-positive anaplastic large cell lymphoma (ALCL), and ALK-negative ALCL.1 There is no standard therapy for PTCLs; CHOP therapy (cyclophosphamide, prednisone, vincristine, and hydroxyl doxorubicin) is the most widely used, but overall survival (OS) is poor.2,3 In a systematic review and meta-analysis of 2815 patients with PTCL, treatment with CHOP or CHOPlike regimens produced a complete response (CR) in 4464% of PTCL-NOS and in 36-70% of AITL patients, although ALK-positive ALCL had a higher CR rate than other T-cell lymphomas.2 Disease progression during chemotherapy occurred in 30-40% of the patients and durable remissions after CHOP alone are not common. In the PTCL-NOS patients treated primarily with CHOP at the British Columbia Cancer Agency, the 5-year progression-free survival (PFS) and OS were only 29% and 35%, respectively.4 Patients with ALK-negative ALCL and AITL had similar 5-year OS, of 34% and 36%, respectively. Given the poor outcomes of PTCL patients, several studies are investigating the role of high-dose chemotherapy and autologous stem cell transplantation in the upfront setting; however, the benefits in terms of preventing relapse are still a subject of debate.5-8 The addition of etoposide to CHOP-based regimens improved the CR in PTCL in some studies.9,11 The German High-grade Non-Hodgkin Lymphoma Study Group (DSHNHL) assessed 320 patients from eight prospective trials and reported that patients who were young adults (< 60 years) with normal lactate dehydrogenase (LDH) had a significantly improved outcome with CHOP plus etoposide (CHOEP) versus CHOP alone.3 A dose-adjusted EPOCH (etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin) plus rituximab regimen had excellent outcomes in primary mediastinal B-cell lymphoma, germinal center B-cell diffuse large B-cell lymphoma, and Burkitt lymphoma.12-14 Recently, Dunleavy et al. reported the encouraging results of a phase II trial of dose-adjusted EPOCH in 24 patients with untreated ALK-positive (n=15) and ALK-negative (n=9) ALCL.15 The event-free survival (EFS) in ALK-positive and ALK-negative ALCL was 72.0% and 62.5%, respectively, and the OS was 78.0% and 87.5% at a median potential follow up of 14.4 years. Based on the hypothesis that dose-adjusted EPOCH may improve treatment outcome, we undertook a phase II prospective study of dose-adjusted EPOCH in patients with nodal PTCL.

Methods The study prospectively enrolled 41 consecutive patients with untreated PTCL from September 2007 to October 2011. The study population was made up of nodal PTCLs, i.e. PTCL-NOS, AITL, ALK-positive ALCL, and ALK-negative ALCL. Diagnoses were based on the WHO classification. All patients were confirmed to be human T-cell lymphotropic virus type 1 (HTLV-1) negative. The pathology was reviewed by hematopathologists (TYo, SN, and KO) to confirm the diagnoses. All patients gave written informed consent. This study was conducted in compliance with the principles of the Declaration of Helsinki and approved by the institutional review boards of Okayama University Hospital, Shikoku Cancer Center, Japanese Red Cross Society Himeji Hospital, Kaneda Hospital, Kanazawa Medical University Hospital, Okayama 2098

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

Characteristic Total patients Sex Male Female Age Median (range), years â&#x2030;¤ 60 > 60 Performance status ECOG 0 to 1 ECOG 2 B symptoms No Yes Clinical stage I/II III/IV LDH level Normal Above normal Extranodal sites 0 to 1 >2 BM involvement No Yes Soluble IL-2 receptor Under median Over median Pathology PTCL-NOS AITL ALK (+) ALCL ALK (-) ALCL IPI score Low (0 to 1) Low intermediate (2) High intermediate (3) High (4 to 5) PIT group Group 1 (0) Group 2 (1) Group 3 (2) Group 4 (3 to 4) Dose escalation No Yes

Number of patients (%) 41 (100) 21 (51.2) 20 (48.8) 64 (32-79) 17 (41.5) 24 (58.5) 35 (85.4) 6 (14.6) 20 (48.8) 21 (51.2) 8 (19.5) 33 (80.5) 13 (31.7) 28 (68.3) 37 (90.2) 4 ( 9.8) 13 (31.7) 28 (68.3) 17 (48.6) 18 (51.4) 21 (51.2) 17 (41.5) 2 ( 4.9) 1 ( 2.4) 10 (24.4) 10 (24.4) 16 (39.0) 5 (12.2) 5 (12.2) 9 (22.0) 20 (48.8) 7 (17.0) 17 (41.5) 24 (58.5)

ECOG: Eastern Cooperative Oncology Group; LDH: lactate dehydrogenase; BM: bone marrow; PTCL-NOS: peripheral T-cell lymphomas-not otherwise specified; AITL: angioimmunoblastic T-cell lymphoma; ALK: anaplastic lymphoma kinase; ALCL: anaplastic large cell lymphoma; IPI: International Prognostic Index; PIT: prognostic index for T-cell lymphoma.

haematologica | 2017; 102(12)

DA-EPOCH for untreated PTCL

Medical Center, Chugoku Central Hospital, Ehime Prefectural Central Hospital, Sumitomo Besshi Hospital, Hyogo Cancer Center, Shimane University Hospital, and Okayama Rosai Hospital (UMIN trial registration number: UMIN000000829).

Under these assumptions, 38 patients needed to be recruited for an α error of 0.05 and power of 80%. After considering patient loss as a result of dropout, we set the target number of patients to 42. The current study had enrolled 41 without drop out; therefore, it was over required sample size (n=38).

Chemotherapy and dose adjustments The dose-adjusted EPOCH starting doses were doxorubicin 10 mg/m2, etoposide 50 mg/m2, and vincristine 0.4 mg/m2 daily as a continuous infusion over days 1 to 4 (96 hours in total); cyclophosphamide 750 mg/m2 as a 2-hour infusion on day (d)5; and prednisone 60 mg/m2 on d1 to d5, as previously described.16 Patients received granulocyte colony-stimulating factor (G-CSF) on d6 until the ANC was more than 5.0x109 cells/L past the nadir. Fluconazole and sulfamethoxazole/trimethoprim were prophylactically administered according to the risk of infection in each patient. Twiceweekly complete blood counts were obtained three days apart and the doses of etoposide, doxorubicin, and cyclophosphamide were adjusted every subsequent cycle based on the neutrophil nadir. If the ANC nadir was 0.5x109 cells/L or over, doses were increased by 20% from those of the previous cycle. If the nadir ANC was less than 0.5x109 cells/L continued for one or two monitoring points, the doses were not changed. If the nadir ANC was less than 0.5x109 cells/L continued for at least three monitoring points or the platelet nadir was less than 25.0x109 cells/L, the doses were reduced by 20% from those of the previous cycle. If patients younger than 70 years of age had a dose reduction below the starting-dose level, only cyclophosphamide was reduced by 20%. If the patient was 70 years old or older, the starting doses were reduced by 20%. Vincristine was reduced by 25% or 50% for grade 2 or 3 motor neuropathy, respectively, and by 50% for grade 3 sensory neuropathy. The sites of disease were restaged after cycle 2 and every 2 cycles thereafter. The cycles were repeated every three weeks, and patients received at least 2 cycles beyond CR, for a minimum of 6 and a maximum of 8 cycles.

Primary end point The primary end point was patient response rate, and secondary end points were the safety of the treatment, OS, and PFS. Tumor responses were evaluated with computer-assisted tomography scan according to the International Workshop criteria17 and were classified as CR, unconfirmed CR (uCR), or partial response (PR).

Sample size estimation Using historical control data,2 we assumed that the lower limit of interest was 40% and the expected response rate was 60%.

Statistical analysis The PFS and OS were estimated using the Kaplan-Meier method and statistical differences among the curves were evaluated with the log rank test. The ability of the following clinical variables to predict outcome was evaluated in univariate analyses: sex, age, performance status (PS), B symptoms, clinical stage, LDH level, extranodal involvement, bone marrow involvement, soluble interleukin (IL)-2 receptor, pathology, IPI, prognostic index for prognostic index for T-cell lymphoma (PIT), and dose escalation. The significance of these variables was also evaluated in a multivariate analysis using a stepwise Cox regression model. All reported P-values are two-sided; P≤0.05 was considered statistically significant. All statistical analyses were performed using STATA software (v.12; StataCorp, College Station, TX, USA).

Results Patients’ characteristics The histological diagnoses were reviewed by a panel of expert hematopathologists and 41 patients with untreated PTCL were included in the analysis of toxicity and response. Patients' characteristics are summarized in Table 1. Median patient age was 64 years (range: 32-79 years) and more than half of the patients were older than 60; 9 patients (22.0%) were over 70. PTCL-NOS was the predominant histological finding (n=21, 51.2%), followed by AITL (n=17, 41.5%), ALK-positive ALCL (n=2, 4.9%), and ALKnegative ALCL (n=1, 2.4%). The majority of the patients were stage III and IV. According to IPI criteria, 51.2% were at high-intermediate or high risk. Twenty-seven patients (65.9%) were categorized as PIT Group 3-4.

Response to treatment Of the 41 patients assessed, 61.0% had a CR [95% confidence interval (CI): 44.5-75.9%] and 17.0% had a PR (95%CI: 7.2-32.1%) to dose-adjusted EPOCH treatment. The overall response rate (ORR) was 78.0% (95%CI: 62.4-89.4%). In patients with PTCL-NOS, the ORR and

Table 2. Adverse event of dose-adjusted-EPOCH treatment (total 255 cycles).

Adverse event Neutropenia Thrombocytopenia Anemia AST/ ALT elevation ALP elevation Creatinine elevation Febrile neutropenia Neuropathy Infection Constipation Headache Nausea Mucositis haematologica | 2017; 102(12)

Grade 2 (%)

Grade 3 (%)

Grade 4 (%)

8.2 16.5 37.3 4.3 3.9 0.8 0 3.5 0 3.1 3.1 1.6 1.2

16.5 15.3 40.4 0 0 0 9.0 0.8 3.1 0 0 0 1.6

58.0 6.7 0.4 0 0 0 0 0 0.4 0 0 0 0 2099

Y. Maeda et al.

CR were 71.4% (95%CI: 47.8-88.7%) and 47.6% (95%CI: 25.7-70.2%), respectively. In AITL patients, ORR and CR were 82.4% (95%CI: 56.6-96.2%) and 76.5% (95%CI: 50.1-93.2%), respectively. The univariate analysis showed that age 60 years or under was the only parameter related to a better response to therapy (ORR: P=0.036; CR: P=0.29). The younger patients (≤ 60 years old) had a high response rate (ORR 94.1%, 95%CI: 71.399.9% and CR 70.6%, 95%CI: 44.0-89.7%). For elderly patients (>60 years old), ORR and CR rates were 66.7% (95%CI: 44.7-84.4%) and 54.2% (95%CI: 32.8-74.4%), respectively. Histology subtype, LDH, PS, and systemic B symptoms were not significantly related to the response to therapy. Recurrence or disease progression occurred in 9 (22.0%) of the 41 patients who initially responded to chemotherapy.

Toxicity and dose adjustments In a total of 255 cycles (median: 6 cycles) of dose-adjusted EPOCH treatment, 20% of dose escalation was performed in 60 cycles (23.5%), and 24 of all 41 patients (58.5%) experienced dose escalation (Figure 1). The patients under 70 years of age received a total of 208 cycles of dose-adjusted EPOCH treatment, and 36.1% of cycles had escalation above and 10.1% had reduction below the starting-dose level. In patients aged 70 years or older, the starting doses were reduced by 20%, and 60.9% of cycles had escalation above and 6.5% had reduction below the starting-dose level. The final dose level for each patient aged under 70 years and for those aged 70 years or older was above the full dose (100%) level in 17 of 32 patients (53.1%) and 5 of 9 patients (55.6%), respectively (Figure 1). Toxicity data of grade ≥2 adverse events are listed in Table 2. The most common grade ≥3 adverse events were neutropenia (74.5%), anemia (40.8%), and thrombocytopenia (22.0%). Febrile neutropenia occurred in 9.0% of the total cycles. Grade 2 and 3 neuropathy occurred in 3.5%

and 0.8%, respectively. Grade 3 infections occurred in 8 cycles (3.1%; 4 catheter-related infections, 2 cases of sepsis, and 2 cases of pneumonia). One patient with rheumatoid arthritis who had previously been treated with methotrexate (MTX) developed severe (grade 4) progressive multifocal leukoencephalopathy (PML) after the third cycle. Gastrointestinal toxicities were generally mild. Grade 2 constipation occurred in 3.1%. Mucositis was observed in 4.3% (grade 3: 1.6%; grade 2: 1.2%; and grade 1: 1.6%, respectively). Liver or renal toxicity was also mild; no grade 3 to 4 AST/ALT, ALP or creatinine elevation was seen. General fatigue was observed in 7.1%; however, all events were grade 1. There were no cardiac complications or treatment-related deaths.

Survival and prognostic factors At the median follow up of 24.0 months, the 2-year PFS and OS were 53.3% (95%CI: 36.4-67.5%) and 73.2% (95%CI: 56.8-84.1%), respectively (Figure 2A). Two-year PFS and OS of younger patients (≤ 60 years old) were 62.5% (95%CI: 34.9-81.1%) and 82.4% (95%CI: 54.7-93.9%), respectively (Figure 2B). In patients with PTCL-NOS, the 2year PFS and OS were 47.1% (95%CI: 25.1-66.4%) and 61.9% (95%CI: 38.1-78.8%) (Figure 2C and D). In AITL patients, the 2-year PFS and OS were 55.2% (95%CI: 28.175.7%) and 88.2% (95%CI: 60.6-96.9%) (Figure 2C and D). No statistically significant trend was observed between AITL and PTCL in 2-year OS (P=0.069). No patients were treated with high-dose chemotherapy and autologous stem cell transplantation in first remission. In the univariate analyses, there was a significant association with PFS in the IPI low-risk group (low + low-intermediate) versus IPI highrisk group (high intermediate + high), and female versus male patients, while there was no significant difference in OS (Table 3). In the multivariate analysis, no factors were significant predictors of either PFS or OS. In PTCL-NOS patients, there was no significant difference in PFS and OS

Figure 1. Map of dose levels achieved, according to cycle. Patients started with full dose (100%) EPOCH (etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin) and subsequent cycles were dose adjusted every cycle based on the previous cycle. If the patient was 70 years or older, the starting doses were reduced by 20%. (Right) Final dose level for each patient aged under 70 years of age and for those aged 70 years or older.

2100

haematologica | 2017; 102(12)

DA-EPOCH for untreated PTCL Table 3. Patients’ outcome.

Characteristic Total patients Sex Male Female Median age, y (range) ≤ 60 > 60 Performance status ECOG 0 to 1 ECOG 2 B symptoms No Yes Clinical stage I/II III/IV LDH level Normal Above normal Extranodal sites 0 to 1 >2 BM involvement No Yes Soluble IL-2 receptor Under median Over median Pathology PTCL-NOS AITL ALK (+) ALCL ALK (-) ALCL IPI score Low (L: 0 to 1) Low-intermediate (LI:2) High-intermediate (HI:3) High (H: 4 to 5) L + LI HI + H PIT group Group 1 (0) Group 2 (1) Group 3 (2) Group 4 (3 to 4) Group 1-2 Group 3-4 Dose escalation No Yes

N (%)

% PFS at 2 years

%OS at 2 years

41 (100)

53.3

73.2

21 (51.2) 20 (48.8)

33.8 73.3

0.014

61.9 85.0

0.083

17 (41.5) 24 (58.5)

62.5 47.1

0.254

82.4 66.7

0.322

35 (85.4) 6 (14.6)

55.3 40.0

0.527

74.3 66.7

0.676

20 (48.8) 21 (51.2)

57.9 47.4

0.716

85.0 61.9

0.109

8 (19.5) 33 (80.5)

87.5 44.0

0.059

87.5 69.7

0.348

13 (31.7) 28 (68.3)

75.0 42.9

0.095

76.9 71.4

0.708

37 (90.2) 4 ( 9.8)

52.0 66.7

0.778

75.7 50.0

0.161

13 (31.7) 28 (68.3)

51.3 56.3

0.826

76.7 70.0

0.600

17 (48.6) 18 (51.4)

70.1 42.4

0.133

88.2 61.1

0.061

21 (51.2) 17 (41.5) 2 ( 4.9) 1 ( 2.4)

47.1 55.2 100.0 100.0

0.629

61.9 88.2 50.0 100.0

0.244

10 (24.4) 10 (24.4) 16 (39.0) 5 (12.2) 20 (48.8) 21 (51.2)

80.0 55.6 33.3 50.0 68.4 37.7

5 (12.2) 9 (22.0) 20 (48.8) 7 (17.0) 14 (34.1) 27 (65.9)

80.0 62.5 43.6 51.4 69.2 44.5

17 (41.5) 24 (58.5)

55.5 51.8

0.200 0.049

90.0 70.0 68.8 60.0 80.0 66.7

0.211

100.0 66.7 75.0 57.1 78.6 70.4

0.740

76.5 70.8

0.572

0.591 0.348

0.433 0.575

0.666

N: number; PFS: progression-free survival; OS: overall survival; y: years; ECOG: Eastern Cooperative Oncology Group; LDH: lactate dehydrogenase; BM: bone marrow; PTCL-NOS: peripheral T-cell lymphoma-not otherwise specified; AITL: angioimmunoblastic T-cell lymphoma; ALK: anaplastic lymphoma kinase; ALCL: anaplastic large cell lymphoma; IPI: International Prognostic Index; PIT: prognostic index for T-cell lymphoma.

haematologica | 2017; 102(12)

2101

Y. Maeda et al. A

Figure 2. Kaplan-Meier estimates of progression-free and overall survival of patients with peripheral T-cell lymphomas (PTCLs) receiving dose-adjusted EPOCH (etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin) (DA-EPOCH). Analysis of progression-free survival (PFS) and overall survival (OS) for all patients in this study (A) and younger patients (≤ 60 years old) (B). PFS (C) and OS (D) of angioimmunoblastic T-cell lymphoma (AITL) and PTCL-not otherwise specified (PTCL-NOS) patients.

between PIT group1-2 and PIT group 3-4 (2-year PFS=55.6% vs. 40.0%, P=0.630; 2-year OS=77.8% vs. 50.0%, P=0.201). The Prognostic Index for AITL (PIAI), comprising age over 60 years, PS ≥ 2, extranodal sites >1, B symptoms, and platelet count less than 150x109/L, was demonstrated to be the better risk model for AITL.18 In patients with AITL, there was no significant difference in PFS and OS between the PIAI low-risk group (0-1 factors, n=9) and high-risk group (2-5 factors, n=8) (2-year PFS=66.7% vs. 29.2%, P=0.119; 2-year OS=100% vs. 75.0%, P=0.121).

Discussion There is evidence for increased expression of the multidrug resistance gene (mdr-1) in relapsed and refractory lymphoma.19-21 Rodriguez-Antona et al. reported that genomic gain was shown in mdr-1 genes in PTCLs.21 The EPOCH regimen was designed on the basis of in vitro findings that mdr-1 tumor cells showed relatively less resistance with prolonged, low-concentration exposure to vincristine and doxorubicin than with brief higher-concentration exposure.22,23 The dose-adjusted EPOCH regimen has yielded excellent outcomes in B-cell lymphoma and ALCL.12-15 In the current study, the dose-adjusted EPOCH produced a high response rate in patients with nodal PTCLs. The ORR and CR rates were 78.0% and 61.0%, respectively, and the 2year OS was 73.2%. Selection bias does not appear to account for the favorable results achieved with dose-adjusted EPOCH in the current study. More than half of the patients were older than 60 years and 27 (65.9%) patients were included in PIT Group 3-4, suggesting that more 2102

patients with aggressive disease status were treated with dose-adjusted EPOCH. Hyper CVAD/MA (hyperfractionated cyclophosphamide, vincristine, adriamycin, dexamethasone/ methotrexate, cytarabine) produced a better overall response rate (85%) and CR rate (80%) than CHOP-like therapy; however, there was no significant difference in 3year OS between front-line regimens (CHOP-like 55%, hyper CVAD/MA 49%).5 For younger patients, the addition of etoposide to CHOP improved the response rates. Kim et al. reported that CHOP plus etoposide and gemcitabine had an ORR of 72%, with 61% CR in 18 PTCL patients.9 The Nordic Lymphoma Group study used biweekly CHOEP as part of an up-front high-dose chemotherapy and autologous stem cell transplantation strategy and achieved an ORR of 82%, with 51% attaining CR.24 The DSHNHL reported that, in patients aged 60 years or younger with a normal LDH, CHOEP significantly improved the 3-year EFS (75.4% vs. 51.0%, P=0.003), although the OS was not significantly affected compared with CHOP.3 In this study, younger patients (≤ 60 years old) had high response (ORR 94.1% and CR 70.6%) and survival (PFS 62.5% and OS 82.4%) rates. EPOCH therapy administered using a unique infusional approach plus a dose-adjustment strategy may shed new light on the treatment of PTCL. Dose-adjusted EPOCH had an encouraging survival curve (2-year OS=70.4%) even in PIT Group 3-4, and longer follow up may demonstrate a more robust survival benefit as first-line treatment in younger PTCL patients. Unlike in younger patients, the addition of etoposide to CHOP in older patients seemed to result in inappropriately high toxicity and did not improve the outcome, although the patient numbers were too small to be compared in the DSHNHL haematologica | 2017; 102(12)

DA-EPOCH for untreated PTCL

analysis.3 Wilson et al. adopted a dose-adjustment strategy based on the hematopoietic nadir, because pharmacokinetic analyses of etoposide and doxorubicin had showed significant interpatient variation in steady-state plasma concentrations.16,25 In this study, we also adopted a dose-adjustment strategy. The starting doses were reduced by 20% in patients aged 70 years or older and 60.9% of cycles had escalation above. In previously reported prospective studies of dose-adjusted EPOCH, dose escalation could also be conducted according to the protocols even in elderly patients (around 50%13,16); figures which are compatible with our study. For elderly patients (> 60 years old), the ORR and CR rates were 66.7% and 54.2%, respectively. Elderly patients (> 60 years old) had significantly worse response rates than younger patients (â&#x2030;¤ 60 years old), while there was no significant difference in either OS and PFS between young and elderly patients. Although dose-adjusted EPOCH was associated with a high incidence of neutropenia, this was not accompanied by a large number of febrile neutropenia episodes, and there were no treatment-related deaths. This might be due to the short duration of neutropenia and the dose-adjustment paradigm introduced to reduce the dose if severe hematologic toxicity occurred.

References 1. Vose J, Armitage J, Weisenburger D, International TCLP. International peripheral T-cell and natural killer/T-cell lymphoma study: pathology findings and clinical outcomes. J Clin Oncol. 2008;26(25):41244130. 2. Abouyabis AN, Shenoy PJ, Sinha R, Flowers CR, Lechowicz MJ. A Systematic Review and Meta-Analysis of Front-line Anthracycline-Based Chemotherapy Regimens for Peripheral T-Cell Lymphoma. ISRN Hematol. 2011;2011:623924. 3. Schmitz N, Trumper L, Ziepert M, et al. Treatment and prognosis of mature T-cell and NK-cell lymphoma: an analysis of patients with T-cell lymphoma treated in studies of the German High-Grade NonHodgkin Lymphoma Study Group. Blood. 2010;116(18):3418-3425. 4. Savage KJ, Chhanabhai M, Gascoyne RD, Connors JM. Characterization of peripheral T-cell lymphomas in a single North American institution by the WHO classification. Ann Oncol. 2004;15(10):1467-1475. 5. Abramson JS, Feldman T, Kroll-Desrosiers AR, et al. Peripheral T-cell lymphomas in a large US multicenter cohort: prognostication in the modern era including impact of frontline therapy. Ann Oncol. 2014;25(11):2211-2217. 6. Rodriguez J, Conde E, Gutierrez A, et al. Frontline autologous stem cell transplantation in high-risk peripheral T-cell lymphoma: a prospective study from The GelTamo Study Group. Eur J Haematol. 2007;79(1):32-38. 7. Mercadal S, Briones J, Xicoy B, et al. Intensive chemotherapy (high-dose CHOP/ESHAP regimen) followed by autologous stem-cell transplantation in previously untreated patients with peripheral T-cell lymphoma. Ann Oncol. 2008;19(5):958963. 8. Reimer P, Rudiger T, Geissinger E, et al. Autologous stem-cell transplantation as

haematologica | 2017; 102(12)

10.

11.

12.

13.

14.

15.

16.

This study has several limitations. It included a relatively small number of patients and the median follow-up time was short. In addition, in this study, the biology of PTCL, including molecular subtypes within PTCL diagnoses, was not evaluated. Although we cannot avoid any selection bias beyond PIT, stage, and age, etc. The 2-year PFS and OS were encouraging, and more robust data with longer-term follow up in the form of randomized controlled trials are needed to confirm the excellent response rates and survival with dose-adjusted EPOCH therapy. Recently, a number of novel agents, including histone deacetylase inhibitors, antifolates, immunomodulatory agents, nucleoside analogs, and other targeted agents, have been developed. Some of these novel agents can be combined with dose-adjusted EPOCH therapy and their relevant combination therapies should be identified. In conclusion, our results indicate that dose-adjusted EPOCH had a high response rate and may improve outcomes for PTCL patients. This regimen has an acceptable toxicity profile, even in elderly patients with poor prognoses in whom approaches such as transplantation may not be feasible. Dose-adjusted EPOCH may be considered as a first-line approach for PTCL.

first-line therapy in peripheral T-cell lymphomas: results of a prospective multicenter study. J Clin Oncol. 2009;27(1):106-113. Kim JG, Sohn SK, Chae YS, et al. CHOP plus etoposide and gemcitabine (CHOPEG) as front-line chemotherapy for patients with peripheral T cell lymphomas. Cancer Chemother Pharmacol. 2006;58(1):35-39. Karakas T, Bergmann L, Stutte HJ, et al. Peripheral T-cell lymphomas respond well to vincristine, adriamycin, cyclophosphamide, prednisone and etoposide (VACPE) and have a similar outcome as high-grade B-cell lymphomas. Leuk Lymphoma. 1996;24(1-2):121-129. Nickelsen M, Ziepert M, Zeynalova S, et al. High-dose CHOP plus etoposide (MegaCHOEP) in T-cell lymphoma: a comparative analysis of patients treated within trials of the German High-Grade NonHodgkin Lymphoma Study Group (DSHNHL). Ann Oncol. 2009;20(12):19771984. Dunleavy K, Pittaluga S, Maeda LS, et al. Dose-adjusted EPOCH-rituximab therapy in primary mediastinal B-cell lymphoma. N Engl J Med. 2013;368(15):1408-1416. Wilson WH, Dunleavy K, Pittaluga S, et al. Phase II study of dose-adjusted EPOCH and rituximab in untreated diffuse large B-cell lymphoma with analysis of germinal center and post-germinal center biomarkers. J Clin Oncol. 2008;26(16):2717-2724. Dunleavy K, Pittaluga S, Shovlin M, et al. Low-intensity therapy in adults with Burkitt's lymphoma. N Engl J Med. 2013;369(20):1915-1925. Dunleavy K, Pittaluga S, Shovlin M, et al. Phase II trial of dose-adjusted EPOCH in untreated systemic anaplastic large cell lymphoma. Haematologica. 2016; 101(1):e27-29. Wilson WH, Grossbard ML, Pittaluga S, et al. Dose-adjusted EPOCH chemotherapy for untreated large B-cell lymphomas: a pharmacodynamic approach with high efficacy. Blood. 2002;99(8):2685-2693.

17. Cheson BD, Horning SJ, Coiffier B, et al. Report of an international workshop to standardize response criteria for nonHodgkin's lymphomas. NCI Sponsored International Working Group. J Clin Oncol. 1999;17(4):1244. 18. Federico M, Rudiger T, Bellei M, et al. Clinicopathologic characteristics of angioimmunoblastic T-cell lymphoma: analysis of the international peripheral Tcell lymphoma project. J Clin Oncol. 2013;31(2):240-246. 19. Goldstein LJ, Galski H, Fojo A, et al. Expression of a multidrug resistance gene in human cancers. J Natl Cancer Inst. 1989;81(2):116-124. 20. Miller TP, Grogan TM, Dalton WS, Spier CM, Scheper RJ, Salmon SE. P-glycoprotein expression in malignant lymphoma and reversal of clinical drug resistance with chemotherapy plus high-dose verapamil. J Clin Oncol. 1991;9(1):17-24. 21. Rodriguez-Antona C, Leskela S, Zajac M, et al. Expression of CYP3A4 as a predictor of response to chemotherapy in peripheral Tcell lymphomas. Blood. 2007;110(9):33453351. 22. Wilson WH, Bryant G, Bates S, et al. EPOCH chemotherapy: toxicity and efficacy in relapsed and refractory nonHodgkin's lymphoma. J Clin Oncol. 1993;11(8):1573-1582. 23. Lai GM, Chen YN, Mickley LA, Fojo AT, Bates SE. P-glycoprotein expression and schedule dependence of adriamycin cytotoxicity in human colon carcinoma cell lines. Int J Cancer. 1991;49(5):696-703. 24. d'Amore F, Relander T, Lauritzsen GF, et al. Up-front autologous stem-cell transplantation in peripheral T-cell lymphoma: NLG-T-01. J Clin Oncol. 2012;30(25):30933099. 25. Wilson WH, Jamis-Dow C, Bryant G, et al. Phase I and pharmacokinetic study of the multidrug resistance modulator dexverapamil with EPOCH chemotherapy. J Clin Oncol. 1995;13(8):1985-1994.

2103

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Non-Hodgkin Lymphoma

Ferrata Storti Foundation

Pan-phosphatidylinositol 3-kinase inhibition with buparlisib in patients with relapsed or refractory non-Hodgkin lymphoma

Anas Younes,1 Gilles Salles,2 Giovanni Martinelli,3 Robert Gregory Bociek,4 Dolores Caballero Barrigon,5 Eva González Barca,6 Mehmet Turgut,7 John Gerecitano,1 Oliver Kong,8 Chaitali Babanrao Pisal,9 Ranjana Tavorath8 and Won Seog Kim10

Memorial Sloan Kettering Cancer Center, New York, NY, USA; Weill Cornell Medical College, Cornell University, New York, NY, USA; 2Centre Hospitalier Lyon-Sud, Pierre Bénite, Lyon, France; 3European Institute of Oncology, Milan, Italy; 4University of Nebraska Medical Center, Omaha, NE, USA; 5Hospital Universitario de Salamanca, Spain; 6Hospital Duran i Reynals-ICO, Barcelona, Spain; 7Ondokuz Mayis University, Samsun, Turkey; 8Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA; 9 Novartis Healthcare Pvt. Ltd., Hyderabad, India and 10Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea 1

Haematologica 2017 Volume 102(12):2104-2112

ABSTRACT

Correspondence: younesa@mskcc.org or wskimsmc@skku.edu

Received: April 14, 2017. Accepted: September 20, 2017. Pre-published: September 29, 2017. doi:10.3324/haematol.2017.169656 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2104 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2104

ctivation of the phosphatidylinositol 3-kinase/mechanistic target of rapamycin pathway plays a role in the pathogenesis of nonHodgkin lymphoma. This multicenter, open-label phase 2 study evaluated buparlisib (BKM120), a pan-class I phosphatidylinositol 3kinase inhibitor, in patients with relapsed or refractory non-Hodgkin lymphoma. Three separate cohorts of patients (with diffuse large B-cell lymphoma, mantle cell lymphoma, or follicular lymphoma) received buparlisib 100 mg once daily until progression, intolerance, or withdrawal of consent. The primary endpoint was overall response rate based on a 6-month best overall response by cohort; secondary endpoints included progression-free survival, duration of response, overall survival, safety, and tolerability. Overall, 72 patients (26 with diffuse large B-cell lymphoma, 22 with mantle cell lymphoma, and 24 with follicular lymphoma) were treated. The overall response rates were 11.5%, 22.7%, and 25.0% in patients with diffuse large B-cell lymphoma, mantle cell lymphoma, and follicular lymphoma, respectively; two patients (one each with diffuse large B-cell lymphoma and mantle cell lymphoma) achieved a complete response. The most frequently reported (>20%) adverse events of any grade in the population in which safety was studied were hyperglycemia, fatigue, and nausea (36.1% each), depression (29.2%), diarrhea (27.8%), and anxiety (25.0%). The most common grade 3/4 adverse events included hyperglycemia (11.1%) and neutropenia (5.6%). Buparlisib showed activity in relapsed or refractory nonHodgkin lymphoma, with disease stabilization and sustained tumor burden reduction in some patients, and acceptable toxicity. Development of mechanism-based combination regimens with buparlisib is warranted. (This study was funded by Novartis Pharmaceuticals Corporation and registered with ClinicalTrials.gov number, NCT01693614).

Introduction Non-Hodgkin lymphoma (NHL) accounts for 4% of all cancers1 and resulted in an estimated 199,700 deaths worldwide in 2012.2 NHL subtypes include diffuse large B-cell lymphoma (DLBCL; 22-36% of cases of NHL), mantle cell lymphoma (MCL; 1%-14% of cases of NHL), and follicular lymphoma (FL; 8%-32% of cases haematologica | 2017; 102(12)

Buparlisib in patients with NHL: CBKM120Z2302

of NHL).3-5 Response to treatment is heterogeneous due to the varying biology of each NHL subtype; disease course can be indolent or aggressive.6,7 The 5-year survival rate from 2006 to 2012 was 62% for patients with DLBCL, 56% for those with MCL, and 88% for those with FL.5 Combining rituximab, a monoclonal antibody targeting the CD20 antigen on B cells, with standard chemotherapy has improved NHL treatment outcomes.8 Despite high response rates to rituximab-based regimens, many patients become resistant to treatment or relapse after initially achieving a response.8 Phosphatidylinositol 3-kinase (PI3K)/mechanistic target of rapamycin (mTOR) signaling regulates many cellular activities, including proliferation, survival, angiogenesis, and glucose metabolism.9 Due to its effects on B-cell signaling, dysregulation of PI3K signaling is associated with cancer development and may play a key role in the pathogenesis of NHL.10-14 Class I PI3K, the most common forms implicated in cancer, are heterodimers with regulatory (p85) and catalytic (p110) subunits.15 There are four class I isoform catalytic subunits, p110α, p110b, p110γ, and p110d (encoded by PIK3CA, PIK3CB, PIK3CG, and PIK3CD, respectively),15 with varying levels of expression; p110α and p110b are ubiquitous, whereas p110γ and p110d are mainly expressed in hematopoietic cells.15,16 Preclinical evidence indicates that B-cell malignancies frequently overexpress PIK3CD.17,18 The PI3Kd inhibitor idelalisib showed antitumor activity in patients with relapsed or refractory indolent NHL,19,20 and was subsequently approved in 2014 for the treatment of patients with relapsed FL and chronic lymphocytic leukemia (in the latter case in combination with rituximab).21,22 Differential PI3K isoform expression levels have also been associated with different disease stages in NHL.23 For example, in MCL, PIK3CA expression increased upon progression and was implicated as an escape mechanism for PI3Kd inhibition.14,23 Other members of the PI3K pathway, including the phosphatase and tensin homolog (PTEN; a negative regulator of PI3K activity) and mTOR (a downstream effector), also play a role in NHL.14,24 Thus, treating patients with a targeted drug that can inhibit all four isoforms of PI3K may help to block escape pathways, prolong responses, and improve outcomes compared to those achieved with treatment with an inhibitor specific for a single isoform, particularly in patients with relapsed or refractory disease.25,26 Buparlisib, a potent, oral pan-PI3K inhibitor, showed antitumor activity in lymphoma cell lines, induced apoptosis in DLBCL27 and reduced myc-dependent proliferation in MCL,28 showed preclinical activity in hematologic malignancies,28,29 and had clinical efficacy in solid tumors, including breast cancer.30-33 In this open-label phase 2 study, we evaluated the efficacy and safety of buparlisib in patients with relapsed or refractory DLBCL, MCL, or FL.

Methods Patients

Adult patients (aged ≥18 years) were eligible if they had histologically confirmed DLBCL, MCL, or FL that had relapsed or was refractory to one or more prior therapies. Patients were required to have one or more measurable nodal lesions (≥2 cm according to International Working Group criteria);34,35 if no such nodal lesion was present at baseline, one or more measurable haematologica | 2017; 102(12)

extra-nodal lesions were required. Additional inclusion criteria included an Eastern Cooperative Oncology Group performance status of ≤2, adequate bone marrow and organ function, and fasting plasma glucose ≤120 mg/dL. Patients with DLBCL were either ineligible for autologous or allogeneic stem cell transplantation or had previously received a transplant. Patients previously treated with PI3K inhibitors, or those with evidence of graft-versus-host disease, active or past central nervous system disease, poorly controlled diabetes mellitus (glycated hemoglobin >8%), active cardiac disease, a history of myocardial infarction within 6 months, or documented congestive heart failure or cardiomyopathy were excluded (additional information is available in the Online Supplementary Methods). Patients were excluded if they had received anticancer therapy within 4 weeks (6 weeks for nitrosourea, monoclonal antibodies, or mitomycin-C) prior to starting the study.

Objectives The primary objective was to determine the overall response rate, defined as the proportion of patients with a best overall response of complete or partial response, according to International Working Group criteria,34,35 in three different histological subgroups of NHL (i.e., DLBCL, MCL, and FL). The primary objective was assessed locally, per investigator, independently for each cohort, and was based on a 6-month best overall response. Secondary objectives were to assess progression-free survival, duration of response, overall survival, and safety and tolerability in the three histological subgroups (additional information is available in the Online Supplementary Methods).

Study design This was a multicenter, open-label, single-arm, phase II study in patients with relapsed or refractory NHL (NCT01693614). Patients with previously treated DLBCL, MCL, or FL received buparlisib 100 mg daily continuously in 28-day cycles until disease progression, death, or intolerable toxicity. Treatment interruptions due to toxicity lasting ≥28 days required permanent study drug discontinuation. Up to two dose reductions (to 80 mg or 60 mg daily) were allowed; patients needing additional reductions discontinued the study. Patients provided written informed consent prior to entering the study. The study protocol and informed consent form were reviewed and approved by the independent ethics committee or institutional review board for each center. The study was conducted in accordance with the International Conference on Harmonization Guidelines for Good Clinical Practice, with applicable local regulations, and with the ethical principles of the Declaration of Helsinki.

Schedule of assessments The assessment schedule is described in the Online Supplementary Methods.

Statistical analyses The intent-to-treat population included all patients who received one or more dose of study treatment; the safety set included all patients who received one or more dose of study treatment and had one or more post-baseline safety assessments. The analysis for each cohort (i.e., DLBCL, MCL, and FL) was based on an exact binomial test comparing the overall response rate to a reference level of 10% in the intent-to-treat population, with a significance level of 5% (two-sided). An exact 95% Clopper-Pearson confidence interval (CI) for the overall response rate was calculated. Time-to-event endpoints (i.e., progression-free survival, duration of response, and overall survival) 2105

A. Younes et al.

were described using Kaplan–Meier curves with associated summary statistics. For progression-free survival and duration of response, patients not known to have progressed or who died by the data cut-off date were censored at the time of last tumor assessment or before any further anticancer therapy had been given. For overall survival, patients not known to have died at the data cut-off were censored at the date of their last contact (additional information is available in the Online Supplementary Methods).

Table 1. Baseline characteristics and patient disposition of the patients.

DLBCL (n = 26)

MCL (n = 22)

FL All patients (n = 24) (N = 72)

Baseline characteristics Median age, years (range) 64 (28-81) 69 (47-79) 61 (41-85) Male, n (%) 18 (69.2) 18 (81.8) 13 (54.2) Stage at last assessment prior to study entry, n (%) I/II 10 (38.5) 1 (4.5) 7 (29.2) III/IV 16 (61.5) 20 (90.9) 17 (70.8) Missing 0 1 (4.5) 0 Bulky disease 5 (19.2) 1 (4.5) 4 (16.7) Elevated LDH 11 (42.3) 6 (27.3) 9 (37.5) Prior bone marrow involvement 6 (23.1) 13 (59.1) 15 (62.5) Median number of prior 3 (1-12) 2 (1-6) 2 (1-9) regimens (range) Prior therapies, n (%) Alkylating agent 26 (100) 22 (100) 22 (91.7) Rituximab* 25 (96.2) 20 (90.9) 23 (95.8) Anthracycline 26 (100) 19 (86.4) 15 (62.5) Bendamustine 3 (11.5) 6 (27.3) 12 (50.0) Purines 1 (3.8) 2 (9.1) 1 (4.2) Radiotherapy 9 (34.6) 3 (13.6) 6 (25.0) Autologous SCT 6 (23.1) 7 (31.8) 4 (16.7) Best response at last treatment, n (%) Complete response 4 (15.4) 7 (31.8) 10 (41.7) Partial response 7 (26.9) 6 (27.3) 3 (12.5) Stable disease 2 (7.7) 3 (13.6) 5 (20.8) Progressive disease 10 (38.5) 4 (18.2) 4 (16.7) Other† 3 (11.5) 2 (9.1) 2 (8.3) WHO/ECOG performance status, n (%) 0 9 (34.6) 10 (45.5) 17 (70.8) 1 14 (53.8) 12 (54.5) 6 (25.0) 2 3 (11.5) 0 1 (4.2) Patients’ disposition‡ Treatment disposition, n (%) Treatment ongoing 2 (7.7) 5 (22.7) 2 (8.3) End of treatment 24 (92.3) 17 (77.3) 22 (91.7) Primary reason for treatment cessation, n (%) Disease progression 11 (42.3) 5 (22.7) 7 (29.2) Adverse events 6 (23.1) 9 (40.9) 4 (16.7) Protocol deviation 5 (19.2) 2 (9.1) 0 Physician’s decision 1 (3.8) 1 (4.5) 4 (16.7) Patient withdrew consent 0 0 5 (20.8) Patient/guardian decision 0 0 2 (8.3) Death 1 (3.8) 0 0

65 (28-85) 49 (68.1) 18 (25.0) 53 (73.6) 1 (1.4) 10 (13.9) 26 (36.1) 34 (47.2) 2 (1-12)

70 (97.2) 68 (94.4) 60 (83.3) 21 (29.2) 4 (5.6) 18 (25.0) 17 (23.6) 21 (29.2) 16 (22.2) 10 (13.9) 18 (25.0) 7 (9.7) 36 (50.0) 32 (44.4) 4 (5.6)

9 (12.5) 63 (87.5) 23 (31.9) 19 (26.4) 7 (9.7) 6 (8.3) 5 (6.9) 2 (2.8) 1 (1.4)

Autologous SCT: autologous stem cell transplant; DLBCL: diffuse large B-cell lymphoma; ECOG: Eastern Cooperative Oncology Group; FL: follicular lymphoma; LDH: lactate dehydrogenase; MCL: mantle cell lymphoma; WHO: World Health Organization. *In the FL cohort, one patient who did not receive rituximab and two patients who did receive rituximab also received prior investigational anti-CD20 therapy. †Other includes unconfirmed complete response (n=1, DLBCL; n=0, MCL and FL), unknown (n=2, DLBCL; n=1, MCL and FL), and not applicable (n=0, DLBLC; n=1, MCL and FL).‡The date of data cut-off was February 25, 2015.

2106

Results Patients Between February 28, 2013, and June 15, 2014, 72 patients were enrolled in the trial, including 26 with DLBCL, 22 with MCL, and 24 with FL (Table 1). Their median age was 65 years (range, 28-85) and 68.1% were male. Most patients (73.6%) had stage III/IV disease at initial diagnosis, and 36.1% of patients had elevated lactate dehydrogenase levels. All patients had received prior antineoplastic therapy, and 17 patients (23.6%) had undergone a prior autologous stem cell transplantation (6 with DLBCL, 7 with MCL, and 4 with FL). The median number of prior antineoplastic regimens was three (range, 1-12) for patients with DLBCL, two (range, 1-6) for patients with MCL, and two (range, 1-9) for patients with FL. The most frequently received prior antineoplastic therapy was an alkylating agent (n=70; 97.2%), followed by rituximab (n=68; 94.4%), anthracycline (n=60; 83.3%), and bendamustine (n=21; 29.2%). At last treatment, a best response of complete or partial response was achieved by 11 patients with DLBCL (42.3%; 15.4% complete responses and 26.9% partial responses), 13 with MCL (59.1%; 31.8% complete responses and 27.3% partial responses), and 13 with FL (54.2%; 41.7% complete responses and 12.5% partial responses): ten (38.5%), four (18.2%), and four (16.7%) patients, respectively, had refractory disease. The most common reasons for discontinuing prior therapy were completion of the prescribed regimen (n=36; 50.0%), disease progression (n=24; 33.3%), and adverse events (n=5; 6.9%); other reasons were listed as “unknown” or “other”. The date of data cut-off was February 25, 2015 for all analyses, except for the analysis of best overall response, which occurred separately in each cohort when all patients within the cohort had received 6 months of treatment.

Table 2. Best overall response.

DLBCL (n = 26) MCL (n = 22)

FL (n = 24)

Best overall response at 6 months,* n (%) [95% CI] Complete response 1 (3.8) 1 (4.5) 0 Partial response 2 (7.7) 4 (18.2) 6 (25.0) Stable disease 5 (19.2) 13 (59.1) 15 (62.5) Progressive disease 12 (46.2) 2 (9.1) 1 (4.2) Unknown† 6 (23.1) 2 (9.1) 2 (8.3) Overall response rate‡ 3 (11.5)§ 5 (22.7) 6 (25.0) [2.5, 30.2] [7.8, 45.4] [9.8, 46.7] Disease control rate|| 8 (30.8) 18 (81.8) 21 (87.5) [14.3, 51.8] [59.7, 94.8] [67.6, 97.3] CI: confidence interval; CR, complete response; DLBCL: diffuse large B-cell lymphoma; FL: follicular lymphoma; MCL: mantle cell lymphoma; *Data cut-off for best overall response had a 6-month primary analysis cut-off that corresponded to June 19, 2014, for the DLBCL cohort, February 25, 2015, for the MCL cohort, and August 7, 2014, for the FL cohort. †Reasons for unknown best overall response: no valid post-baseline assessment (n=7; 4 DLBCL, 2 FL, 1 MCL), all post-baseline assessments were not evaluable for response (n=2, 1 DLBCL, 1 MCL), stable disease too early (n=1, DLBCL). ‡Defined as complete response or partial response. §Of 26 patients with DLBCL, 16 patients had a clinical response (i.e., complete response, partial response or stable disease) to the last therapy prior to study entry; the overall response rate for these patients was 6.3% (1 patient achieved a partial response) ||Defined as stable disease or complete response or partial response.

haematologica | 2017; 102(12)

Buparlisib in patients with NHL: CBKM120Z2302

Efficacy

Overall response rate The overall response rates in the DLBCL, MCL, and FL cohorts were 11.5% (95% CI: 2.5-30.2), 22.7% (95% CI: 7.8-45.4), and 25.0% (95% CI: 9.8-46.7), respectively (Table 2). The primary endpoint was not met, since none of the cohorts reached the minimum number of responses to establish a 35% response rate required for statistical significance at the 5% level, based on the lower limits of the 95% CI (i.e., the 95% CI lower bound was <10% for all three cohorts). As shown in waterfall plots of the best overall response during buparlisib treatment, evaluable patients in all three cohorts had a reduction in investigator-assessed tumor size (Figure 1A-C). The disease control rates in the DLBCL, MCL, and FL cohorts were 30.8% (95% CI: 14.3-51.8), 81.8% (95% CI: 59.7-94.8), and 87.5% (95% CI: 67.6-97.3), respectively.

Time to response and duration of response

For the patients in the DLBCL cohort who responded, the median time to response was 2.1 months (range, 1.83.5) and the median duration of response was 2.2 months (95% CI: 1.2-not estimable). Of the three patients who responded, two relapsed. In the MCL cohort, the median time to response was 1.8 months (range, 0.9-9.4), while the median duration of response was not reached; of the five patients who responded, one patient subsequently relapsed. In the FL cohort, the median time to response and duration of response were 3.5 months (range, 1.6-5.9) and 11.0 months (95% CI: 3.9-not estimable), respectively; of the six patients who responded, two relapsed.

Progression-free survival The median progression-free survival of patients with DLBCL, MCL, and FL was 1.8 months (95% CI: 1.5-4.0), 11.3 months (95% CI: 3.8-not estimable), and 9.8 months (95% CI: 3.8-not estimable), respectively (Figure 2A). The estimated 6-month progression-free survival rate was 12.6% (95% CI: 2.3-32.0) in the DLBCL cohort, 68.6% (95% CI: 39.8-85.7) in the MCL cohort, and 60.7% (95% CI: 31.7-80.6) in the FL cohort. The median follow-up time for progression-free survival was 1.7, 4.7, and 3.7 months for the DLBCL, MCL, and FL cohorts, respectively.

Overall survival With a median follow-up time of 4.6 months, the median overall survival for the DLBCL cohort was 5.2 months (95% CI: 3.1-not estimable) (Figure 2B). For the MCL and FL cohorts the median follow-up time for overall survival was 8.2 months and 12.1 months, respectively, and the median overall survival was not reached for either cohort.

Safety The median duration of exposure to buparlisib in the DLBCL, MCL, and FL cohorts was 7.5 weeks (range, 1.776.0), 20.6 weeks (range, 1.7-54.1), and 16.3 weeks (range, 3.3-81.3), respectively. As of the data cut-off, 47 patients (65.0%; 50.0%, 68.2%, and 79.2% in the DLBCL, MCL, and FL cohorts, respectively) had received buparlisib for ≥8 weeks and 21 patients (29.2%; 7.7%, 45.5%, and 37.5%, in the DLBCL, MCL, and FL cohorts, respectively) had received the drug for ≥24 weeks. Treatment was ongoing as of February 25, 2015, the data cut-off date, in nine patients (12.5%), including two with DLBCL, five haematologica | 2017; 102(12)

Figure 1. Best overall response with respect to best percentage change from baseline in investigator-assessed tumor size during buparlisib therapy for patients with measurable lesions divided by cohort The graphs include patients with measurable disease at baseline and ≥1 subsequent tumor assessments. The data cut-off was February 25, 2015, for all cohorts. (A) Patients with DLBCL (n=20), (B) patients with MCL (n=20), (C) patients with FL (n=22). Best overall response is indicated for each patient. Note that for two patients with DLBCL, the post-baseline assessment was not evaluable or best overall response was progressive disease due to a new lesion (asterisk). The dashed line shows the percentage change that represents the criterion for response, according to International Working Group criteria.34,35 CR: complete response; DLBCL: diffuse large B-cell lymphoma; FL: follicular lymphoma; IWG: International Working Group; MCL: mantle cell lymphoma; PD: progressive disease; PR: partial response; SD, stable disease.

2107

A. Younes et al.

with MCL, and two with FL. The remaining 63 patients (87.5%) had discontinued treatment (Table 1). The most common primary reasons for treatment discontinuation were disease progression (n=23; 31.9%) and adverse events (n=19; 26.4%). Of the 22 patients (30.6%; DLBCL, n=7; MCL, n=7; FL, n=8) who required a dose reduction, the majority required only one dose reduction (54.5%: DLBCL, 4/7; MCL, 2/7; FL, 6/8), while only one patient in the FL cohort required ≥3 dose reductions (Online Supplementary Table S1). Of the

patients requiring a dose reduction, all patients required a dose reduction due to adverse events (n=22: DLBCL, n=7; MCL, n=7; FL, n=8), and one patient in the FL cohort had an additional dose reduction due to a dosing error. The median percent of days with the full dose was 100% (range, 14.3-100%), 100% (range, 6.6-100%), and 97.4% (range, 4.2-100%) for patients in the DLBCL, MCL, and FL cohorts, respectively. Dose interruptions occurred in 36 patients (50.0%): 11 with DLBCL, 10 with MCL, and 15 with FL. Among patients whose dosing was interrupted,

Figure 2. Kaplan–Meier curves for progression-free survival and overall survival. Kaplan–Meier curves for the secondary endpoints by cohort (DLBCL, MCL, and FL) are shown here together, but the curves are not meant to be compared between cohorts as the study was not designed to compare the three cohorts. (A) Progression-free survival curves; median follow-up time: DLBCL, 1.7 months; MCL, 4.7 months; FL, 3.7 months. (B) Overall survival curves; median follow-up time: DLBCL, 4.6 months; MCL, 8.2 months; FL, 12.1 months. CI, confidence interval; DLBCL, diffuse large B-cell lymphoma; FL, follicular lymphoma; MCL, mantle cell lymphoma; NE, not estimable; OS, overall survival; PFS, progression-free survival.

2108

haematologica | 2017; 102(12)

Buparlisib in patients with NHL: CBKM120Z2302

the reasons for the dose interruption included adverse events [n=32 (88.9%): DLBCL, n=11; MCL, n=9; FL, n=12], scheduling conflict [n=5 (13.9%): MCL, n=2; FL, n=3], and dosing error [n=4 (11.1%): MCL, n=1; FL, n=3]. The most common adverse events (occurring in >3 patients) that caused dose reduction or interruption were hyperglycemia (n=8), fatigue (n=6), and nausea (n=4). Hyperglycemia, fatigue, and nausea (n=26; 36.1% each), depression (n=21; 29.2%), diarrhea (n=20; 27.8%), and anxiety (n=18; 25.0%) were the most frequently reported adverse events of any grade, regardless of their relationship with the study drug, occurring in >20% of patients (Table 3). The most frequently reported grade 3/4 adverse events, regardless of study drug relationship (occurring in â&#x2030;Ľ3 patients), were hyperglycemia (n=8; 11.1%), neutrope-

nia (n=4; 5.6%), fatigue, nausea, anxiety, asthenia, confusional state, and thrombocytopenia (each n=3; 4.2%). Mood disturbances (e.g., anxiety, agitation, confusional state, depression, stress, psychiatric decompensation) of any grade were reported in 29 patients (40.3%); 24 patients (33.3%) had grade 1/2 mood disorders, and five patients (6.9%) had grade 3/4 mood disorders. Laboratory hematologic and biochemical abnormalities of any grade were reported in 63 patients (87.5%) and 71 patients (98.6%), respectively (Table 3). Grade 3/4 hematologic and biochemical abnormalities occurring in more than 10% of patients were lymphocytopenia (n=16; 22.2%), hyperglycemia (n=15; 20.8%), and neutropenia (n=9; 12.5%) (Table 3). Grade 3/4 infections were reported in four patients with DLBCL, four patients with MCL, and

Table 3. Most frequently reported adverse events and laboratory abnormalities, regardless of relationship with the study drug.

Most common adverse events, n (%) All patients (n = 72) All grades

Grade 3/4

Any-cause adverse events (occurring in >5% of all patients) Fatigue 26 (36.1) 3 (4.2) Hyperglycemia 26 (36.1) 8 (11.1) Nausea 26 (36.1) 3 (4.2) Depression 21 (29.2) 2 (2.8) Diarrhea 20 (27.8) 0 Anxiety 18 (25.0) 3 (4.2) Decreased appetite 14 (19.4) 1 (1.4) Weight decrease 13 (18.1) 0 Asthenia 11 (15.3) 3 (4.2) Constipation 11 (15.3) 0 Cough 10 (13.9) 0 Abdominal pain 8 (11.1) 1 (1.4) Pruritus 8 (11.1) 0 Rash 8 (11.1) 1 (1.4) Vomiting 8 (11.1) 0 Dyspepsia 7 (9.7) 1 (1.4) Dyspnea 7 (9.7) 2 (2.8) Tremor 7 (9.7) 1 (1.4) Headache 6 (8.3) 1 (1.4) Pyrexia 6 (8.3) 0 Urinary tract infection 6 (8.3) 2 (2.8) Agitation 5 (6.9) 0 Anemia 5 (6.9) 2 (2.8) Back pain 5 (6.9) 0 Confusional state 5 (6.9) 3 (4.2) Insomnia 5 (6.9) 0 Muscle spasms 5 (6.9) 0 Thrombocytopenia 5 (6.9) 3 (4.2) Vertigo 5 (6.9) 2 (2.8) Dizziness 4 (5.6) 0 Dry skin 4 (5.6) 0 haematologica | 2017; 102(12)

Most common adverse events, n (%) All patients (n = 72)

Hypokalemia Neutropenia Pain in extremity Pleural effusion

All grades

Grade 3/4

4 (5.6) 4 (5.6) 4 (5.6) 4 (5.6)

1 (1.4) 4 (5.6) 1 (1.4) 1 (1.4)

Hematologic abnormalities Anemia 43 (59.7) 2 (2.8) Thrombocytopenia 36 (50.0) 3 (4.2) Lymphocytopenia 31 (43.1) 16 (22.2) Leukopenia 22 (30.6) 6 (8.3) Neutropenia 19 (26.4) 9 (12.5) Lymphocytosis 17 (23.6) 3 (4.2) Biochemical abnormalities (occurring in >10% of all patients) Hyperglycemia 51 (70.8) 15 (20.8) Increased AST 31 (43.1) 1 (1.4) Hypercholesterolemia 28 (38.9) 0 Hypertriglyceridemia 28 (38.9) 0 Increased AP 27 (37.5) 0 Increased GGT 23 (31.9) 6 (8.3) Increased creatinine 19 (26.4) 0 Hyponatremia 19 (26.4) 3 (4.2) Hypokalemia 18 (25.0) 2 (2.8) Increased ALT 18 (25.0) 1 (1.4) Hypophosphatemia 16 (22.2) 3 (4.2) Hypoalbuminemia 14 (19.4) 0 Hyperkalemia 12 (16.7) 0 Uric acid 12 (16.7) 1 (1.4) Hypernatremia 10 (13.9) 0 Hyperbilirubinemia 10 (13.9) 0 Increased lipase 9 (12.5) 0 ALT: alanine aminotransferase; AP: alkaline phosphatase; AST: aspartate aminotransferase; GGT: gamma-glutamyl transferase.

2109

A. Younes et al.

one patient with FL. Of these, serious infections were reported in two patients with DLBCL (pneumonia and urinary tract infection), two patients with MCL (pneumonia and septic shock), and one patient with FL (lung infection). Treatment for patients who had infectious complications was at the discretion of the investigator. Serious adverse events, regardless of a relationship with the study medication, of any grade and grade 3/4 were reported in 30 patients (41.7%: DLBCL, n=12; MCL, n=12; FL, n=6) and 26 patients (36.1%: DLBCL, n=10; MCL, n=12; FL, n=4), respectively. The most common serious adverse events were confusional state (n=3), pleural effusion (n=3), nausea (n=2), vomiting (n=2), dysphagia (n=2), and pneumonia (n=2). Serious adverse events suspected to be related to study medication were reported in 14 patients (19.4%: DLBCL, n=3; MCL, n=9; FL, n=2), and were all grade 3/4. Overall, 18 patients died, including eight on treatment (occurring ≤30 days after study drug discontinuation): seven due to underlying DLBCL (n=6) and MCL (n=1), and one in the DLBCL cohort due to gastrointestinal hemorrhage suspected to be related to the study drug.

Exploratory analysis Biomarkers were assessed using baseline archival tumor samples from 52 patients (72.2%) and fresh tumor samples from one patient (1.4%); all but one of the archival tumor samples were derived from the primary tumor. Of the patients with evaluable PI3KCA mutation status (19 with DLBCL, 11 with MCL, and 18 with FL) or PTEN mutation status (19 with DLBCL, 11 with MCL, and 17 with FL), none had mutations in PI3KCA or PTEN (Online Supplementary Table S2). Loss of PTEN expression was evaluable in 20, 12, and 20 patients with DLBCL, MCL, and FL, respectively; none of these patients met the criteria for loss of PTEN expression.

Discussion Buparlisib elicited tumor reduction in patients with relapsed or refractory NHL; patients in all cohorts had reductions in tumor size, and the median progression-free survival was 1.8, 11.3, and 9.8 months in the DLBCL, MCL, and FL cohorts, respectively. The median overall survival, although not yet matured, was 5.2 months (95% CI: 3.1-not estimable) for the DLBCL cohort and was not reached for the MCL and FL cohorts. The PI3K/mTOR pathway is frequently activated in NHL and plays an important role in the pathogenesis of this disease.14 However, mutations in key components of the PI3K pathway, such as PIK3CA, seem infrequent in this condition.10,36 Consistently, no PIK3CA or PTEN mutations were detected in the 42 tumor samples tested for mutations in this study. Despite previous reports of a reduction or loss of PTEN expression in approximately 21% of FL,36 37% of DLBCL,11 and 15% of MCL samples,23,37 no PTEN loss was observed among the 50 tumors tested in this study, which may be explained by the stringent cut-off used for defining a loss of expression (< 10% expression by immunohistochemistry). Further investigation, including a broader evaluation of PI3K activation markers or multivariate analyses of baseline characteristics, would be needed to identify subpopulations of patients who might respond to buparlisib. 2110

Preclinical models suggest that all PI3K isoforms (PI3Kα, PI3Kb, and PI3Kd) are expressed in leukocytes,38 but that their relative levels of expression vary in NHL according to disease stage.23 Specific PI3K isoform ratios are associated with relapse and have been implicated in resistance to PI3Kd inhibitors, but not PI3Kα/PI3Kd inhibitors.23 Previous studies on PI3K inhibitors that specifically target one or two PI3K isoforms have met with some success in indolent NHL (overall response rate of 65% with duvelisib),39 in FL and in MCL (overall response rates of 45% and 40%, respectively, with idelalisib),40,41 but not in DLBCL (overall response rate of 0% with rigosertib).42 However, improved long-term outcomes (e.g., progression-free survival) with these agents have been limited in patients with relapsed or refractory disease; for example, the median progression-free survival of patients treated with idelalisib was 3.7 months among those with MCL and 7.6 to 11 months for patients with indolent NHL (the majority, 58% - 59%, with FL).20,40,41 Pan-PI3K inhibition results in similar or higher overall response rates (≤25% in DLBCL, 83% in MCL, and 20% - 50% in FL),43-45 with potentially longer progression-free survival. For example, in a study of the pan-PI3K inhibitor SAR245409, the overall response rate was 50% and the median progressionfree survival was not reached after 8 months of follow-up in patients with relapsed or refractory FL.45 In our study, patients with FL on treatment with buparlisib had an overall response rate of 25.0% and a median progression-free survival of 9.8 months. Thus, it appears that sustained inhibition of all PI3K isoforms has some effect on achieving response, but may have a more pronounced benefit on long-term outcomes for patients with FL and MCL. In general, PI3K inhibitors have a manageable safety profile in NHL;46-48 however, the European Medicines Agency has recently opened a review of idelalisib due to new safety concerns over serious adverse events (mostly infections) in ongoing clinical trials. Buparlisib was generally well tolerated. Hyperglycemia, although frequently reported, is a known on-target effect of PI3Kα inhibitors9 and, as in other buparlisib studies,30 was manageable with standard antidiabetic therapy. Mood disorders have been previously reported in studies of buparlisib49 owing, in part, to the ability of this drug to cross the blood-brain barrier.50 For this reason, in this study, patients with active psychiatric disorders at screening were excluded, and the patients enrolled were prospectively followed with self-assessment questionnaires (i.e., GAD-7 and PHQ-9). As in previous studies, some patients with NHL experienced mood disorders that were not serious and were manageable with dose reduction/interruption or treatment with appropriate concomitant medication. In addition, patients developing any of these adverse events were referred for psychiatric consultation. Loss of appetite and weight (Online Supplementary Table S3) in patients may compromise quality of life, as reported in some previous studies.51 Most treatment-related adverse events with buparlisib were grade 1/2 and could be managed with dose reduction or interruption until resolution of toxicity. The rates of grade 3/4 elevations in alanine aminotransferase and aspartate aminotransferase, pneumonia, and diarrhea were lower with buparlisib than with a selective PI3Kd inhibitor in patients with relapsed or refractory NHL.20 Although the overall response rate with buparlisib was not substantial, the observed progression-free survival in haematologica | 2017; 102(12)

Buparlisib in patients with NHL: CBKM120Z2302

this study warrants further evaluation of this agent in hematologic cancers. Combination of buparlisib with other treatments (i.e., targeted therapies, monoclonal antiCD20 antibodies, checkpoint inhibitors, or standard chemotherapies) should be considered, particularly for patients with FL and MCL. Acknowledgments We thank the patients and their caregivers for their participation in this trial and the other investigators who enrolled patients into the study: from Belgium, Marc André (n=3) and Achiel van Hoof (n=1); from France, Reda Bouabdalla (n=3) and Thierry

References 14. 1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin. 2016;66(1):7-30. 2. Torre LA, Bray F, Siegel RL, Ferlay J, LortetTieulent J, Jemal A. Global cancer statistics, 2012. CA Cancer J Clin. 2015;62(2):87-108. 3. A clinical evaluation of the International Lymphoma Study Group classification of non-Hodgkin's lymphoma. The NonHodgkin's Lymphoma Classification Project. Blood. 1997;89(11):3909-3918. 4. Anderson JR, Armitage JO, Weisenburger DD. Epidemiology of the non-Hodgkin's lymphomas: distributions of the major subtypes differ by geographic locations. NonHodgkin's Lymphoma Classification Project. Ann Oncol. 1998;9(7):717-720. 5. Howlader N, Noone AM, Krapcho M, et al. (eds.) SEER Cancer Statistics Review, 19752013, National Cancer Institute. http://seer.cancer.gov/csr/1975_2013/, based on November 2015 SEER data submission, posted to the SEER web site, April 2016. 6. Armitage JO. Treatment of non-Hodgkin's lymphoma. N Engl J Med. 1993;328(14): 1023-1030. 7. National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: non-Hodgkin's lymphoma. Version 3. 2016: http://www.nccn.org. Accessed July 2016. 8. Chao MP. Treatment challenges in the management of relapsed or refractory nonHodgkin's lymphoma - novel and emerging therapies. Cancer Manag Res. 2013;5:251269. 9. Katso R, Okkenhaug K, Ahmadi K, White S, Timms J, Waterfield MD. Cellular function of phosphoinositide 3-kinases: implications for development, homeostasis, and cancer. Annu Rev Cell Dev Biol. 2001;17:615-675. 10. Rudelius M, Pittaluga S, Nishizuka S, et al. Constitutive activation of Akt contributes to the pathogenesis and survival of mantle cell lymphoma. Blood. 2006;108(5):1668-1676. 11. Abubaker J, Bavi PP, Al-Harbi S, et al. PIK3CA mutations are mutually exclusive with PTEN loss in diffuse large B-cell lymphoma. Leukemia. 2007;21(11):2368-2370. 12. Dal Col J, Zancai P, Terrin L, et al. Distinct functional significance of Akt and mTOR constitutive activation in mantle cell lymphoma. Blood. 2008;111(10):5142-5151. 13. Baohua Y, Xiaoyan Z, Tiecheng Z, Tao Q, Daren S. Mutations of the PIK3CA gene in

haematologica | 2017; 102(12)

15. 16.

17.

18.

19.

20.

21. 22. 23.

24.

25.

26.

Lamy (n=1); from Germany, Maria-Elisabeth Goebeler (n=2) and Oliver Ottmann (n=1); from Italy, Daniele Laszlo (n=5) and Paolo Corradini (n=3); from Korea, Hyeon Seok Eom (n=1); from Turkey, Fahri Sahin (n=3); and from the USA, Alice Mims (n=2) and Jason Westin (n=1). We would also like to thank Nabanita Mukherjee for support with the statistical analysis, Joe Sulovski for trial design, study conduct, and review of early versions of the manuscript, Fabian Herbst for review of early versions of the manuscript, and Pamela Tuttle, PhD, CMPP, and Katherine Mills-Luján, PhD, of ArticulateScience, LLC for medical editorial assistance, which was funded by Novartis Pharmaceuticals Corporation.

diffuse large B cell lymphoma. Diagn Mol Pathol. 2008;17(3):159-165. Jabbour E, Ottmann OG, Deininger M, Hochhaus A. Targeting the phosphoinositide 3-kinase pathway in hematologic malignancies. Haematologica. 2014;99(1): 7-18. Fruman DA, Rommel C. PI3K and cancer: lessons, challenges and opportunities. Nat Rev Drug Discov. 2014;13(2):140-156. Beer-Hammer S, Zebedin E, von Holleben M, et al. The catalytic PI3K isoforms p110gamma and p110delta contribute to B cell development and maintenance, transformation, and proliferation. J Leukoc Biol. 2010;87(6):1083-1095. Chantry D, Vojtek A, Kashishian A, et al. p110delta, a novel phosphatidylinositol 3kinase catalytic subunit that associates with p85 and is expressed predominantly in leukocytes. J Biol Chem. 1997;272(31):19236-19241. Puri KD, Gold MR. Selective inhibitors of phosphoinositide 3-kinase delta: modulators of B-cell function with potential for treating autoimmune inflammatory diseases and B-cell malignancies. Front Immunol. 2012;3:256. Benson DM, Kahl BS, Furman RR, et al. Final results of a phase I study of idelalisib, a selective inhibitor of PI3K , in patients with relapsed or refractory indolent nonHodgkin lymphoma (iNHL). J Clin Oncol. 2013;31(suppl):8526. Gopal AK, Kahl BS, de Vos S, et al. PI3Kdelta inhibition by idelalisib in patients with relapsed indolent lymphoma. N Engl J Med. 2014;370(11):1008-1018. Zydelig idelalisib. Package insert. 2014. Zydelig idelalisib. Summary of product characteristics. 2014. Iyengar S, Clear A, Bödör C, et al. P110alpha-mediated constitutive PI3K signaling limits the efficacy of p110deltaselective inhibition in mantle cell lymphoma, particularly with multiple relapse. Blood. 2013;121(12):2274-2284. Okosun J, Wolfson RL, Wang J, et al. Recurrent mTORC1-activating RRAGC mutations in follicular lymphoma. Nat Genet. 2016;48(2):183-188. Foukas LC, Berenjeno IM, Gray A, Khwaja A, Vanhaesebroeck B. Activity of any class IA PI3K isoform can sustain cell proliferation and survival. Proc Natl Acad Sci USA. 2010;107(25):11381-11386. Tabe Y, Jin L, Konopleva M, et al. Class IA PI3K inhibition inhibits cell growth and proliferation in mantle cell lymphoma. Acta Haematol. 2014;131(1):59-69.

27. Zang C, Eucker J, Liu H, et al. Inhibition of pan-class I phosphatidyl-inositol-3-kinase by NVP-BKM120 effectively blocks proliferation and induces cell death in diffuse large B-cell lymphoma. Leuk Lymphoma. 2014;55(2):425-434. 28. Walsh KJ, Fan S, Patel A, et al. PI3K inhibitors inhibit lymphoma growth by downregulation of MYC-dependent proliferation. Blood. 2009;114(22):1697. 29. Matas-Cespedes A, Rodriguez V, Kalko SG, et al. Disruption of follicular dendritic cellsfollicular lymphoma cross-talk by the panPI3K inhibitor BKM120 (buparlisib). Clin Cancer Res. 2014;20(13):3458-3471. 30. Rodon J, Bra a I, Siu LL, et al. Phase I doseescalation and -expansion study of buparlisib (BKM120), an oral pan-class I PI3K inhibitor, in patients with advanced solid tumors. Invest New Drugs. 2014;32(4):670-681. 31. Ma CX, Luo J, Naughton M, et al. A phase I study of BKM120 and fulvestrant in postmenopausal women with estrogen receptor positive metastatic breast cancer. Cancer Res. 2015;75(9):abstract PD5-6. 32. Mayer IA, Abramson VG, Isakoff SJ, et al. Stand up to cancer phase Ib study of panphosphoinositide-3-kinase inhibitor buparlisib with letrozole in estrogen receptor-positive/human epidermal growth factor receptor 2-negative metastatic breast cancer. J Clin Oncol. 2014;32(12):12021209. 33. Saura C, Bendell J, Jerusalem G, et al. Phase Ib study of buparlisib plus trastuzumab in patients with HER2-positive advanced or metastatic breast cancer that has progressed on trastuzumab-based therapy. Clin Cancer Res. 2014;20(7):1935-1945. 34. Cheson BD, Pfistner B, Juweid ME, et al. Revised response criteria for malignant lymphoma. J Clin Oncol. 2007;25(5):579586. 35. Cheson BD. The International Harmonization Project for response criteria in lymphoma clinical trials. Hematol Oncol Clin North Am. 2007;21(5):841-854. 36. Yahiaoui OI, Nunes JA, Castanier C, et al. Constitutive AKT activation in follicular lymphoma. BMC Cancer. 2014;14:565. 37. Psyrri A, Papageorgiou S, Liakata E, et al. Phosphatidylinositol 3'-kinase catalytic subunit alpha gene amplification contributes to the pathogenesis of mantle cell lymphoma. Clin Cancer Res. 2009;15(18):5724-5732. 38. Vanhaesebroeck B, Welham MJ, Kotani K, et al. P110delta, a novel phosphoinositide 3-kinase in leukocytes. Proc Natl Acad Sci

2111

A. Younes et al. USA. 1997;94(9):4330-4335. 39. Flinn I, Oki Y, Patel M, et al. A phase 1 evaluation of duvelisib (IPI-145), a PI3K-d, inhibitor, in patients with relapsed/refractory iNHL. Blood. 2014;124:(21):802. 40. Flinn IW, Kahl BS, Leonard JP, et al. Idelalisib, a selective inhibitor of phosphatidylinositol 3-kinase-delta, as therapy for previously treated indolent nonHodgkin lymphoma. Blood. 2014;123(22): 3406-3413. 41. Kahl BS, Spurgeon SE, Furman RR, et al. A phase 1 study of the PI3Kdelta inhibitor idelalisib in patients with relapsed/refractory mantle cell lymphoma (MCL). Blood. 2014;123(22):3398-3405. 42. Roschewski M, Farooqui M, Aue G, Wilhelm F, Wiestner A. Phase I study of ON 01910.Na (rigosertib), a multikinase PI3K inhibitor in relapsed/refractory B-cell malignancies. Leukemia. 2013;27(9):19201923. 43. Dreyling M, Morschhauser F, Bron D, et al. Preliminary results of a phase II study of

2112

44.

45.

46.

47.

single agent Bay 80-6946, a novel PI3K inhibitor, in patients with relapsed/refractory, indolent or aggressive lymphoma. Blood. 2013;122(21):87. Brown JR, Davids MS, Rodon J, et al. Update on the safety and efficacy of the pan class I PI3K inhibitor SAR245408 (XL147) in chronic lymphocytic leukemia and non-Hodgkin's lymphoma patients. Blood. 2013;122:4170. Brown JR, Hamadani M, Arnason J, et al. SAR245409 monotherapy in relapsed/refractory follicular lymphoma: preliminary results from the phase II ARD12130 study. Blood. 2013;122(21):86. Coutre S, Barrientos J, Brown J, et al. Safety of idelalisib in B-cell malignancies: integrated analysis of eight clinical trials. Haematologica. 2015;100(Suppl 1):P588. O'Brien S, Faia K, White K, et al. Early clinical activity and pharmacodynamic effect of duvelisib, a PI3K-delta, gamma inhibitor, in patients with treatment-naĂŻve CLL. Haematologica. 2015;100(Suppl 1):S434.

48. Burris HA, Patel MR, Fenske TS, et al. TGR1202, a novel once daily PI3K-delta inhibitor, demonstrates clinical activity with a favorable safety profile, lacking hepatotoxicity, in patients with CLL and Bcell lymphoma. Haematologica. 2015;100(Suppl 1):S432. 49. Bendell JC, Rodon J, Burris HA, et al. Phase I, dose-escalation study of BKM120, an oral pan-class I PI3K inhibitor, in patients with advanced solid tumors. J Clin Oncol. 2012;30(3):282-290. 50. Nanni P, Nicoletti G, Palladini A, et al. Multiorgan metastasis of human HER-2(+) breast cancer in Rag2(-/-);Il2rg(-/-) mice and treatment with PI3K inhibitor. PLoS One. 2012;7(6):e39626. 51. Simons JP, Aaronson NK, Vansteenkiste JF et al. Effects of medroxyprogesterone acetate on appetite, weight, and quality of life in advanced-stage non-hormone-sensitive cancer: a placebo-controlled multicenter study. J Clin Oncol. 1996;14(4):10771084.

haematologica | 2017; 102(12)

ARTICLE

Plasma Cell Disorders

The kinesin spindle protein inhibitor filanesib enhances the activity of pomalidomide and dexamethasone in multiple myeloma

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Susana Hernández-García,1 Laura San-Segundo,1 Lorena González-Méndez,1 Luis A. Corchete,1 Irena Misiewicz-Krzeminska,1,2 Montserrat Martín-Sánchez,1 Ana-Alicia López-Iglesias,1 Esperanza Macarena Algarín,1 Pedro Mogollón,1 Andrea Díaz-Tejedor,1 Teresa Paíno,1 Brian Tunquist,3 María-Victoria Mateos,1 Norma C Gutiérrez,1 Elena Díaz-Rodriguez,1 Mercedes Garayoa1* and Enrique M Ocio1* Centro Investigación del Cáncer-IBMCC (CSIC-USAL) and Hospital Universitario-IBSAL, Salamanca, Spain; 2National Medicines Institute, Warsaw, Poland and 3Array BioPharma, Boulder, Colorado, USA

Haematologica 2017 Volume 102(12):2113-2124

*MG and EMO contributed equally to this work

ABSTRACT

inesin spindle protein inhibition is known to be an effective therapeutic approach in several malignancies. Filanesib (ARRY-520), an inhibitor of this protein, has demonstrated activity in heavily pretreated multiple myeloma patients. The aim of the work herein was to investigate the activity of filanesib in combination with pomalidomide plus dexamethasone backbone, and the mechanisms underlying the potential synergistic effect.The ability of filanesib to enhance the activity of pomalidomide plus dexamethasone was studied in several in vitro and in vivo models. Mechanisms of this synergistic combination were dissected by gene expression profiling, immunostaining, cell cycle and short interfering ribonucleic acid studies. Filanesib showed in vitro, ex vivo, and in vivo synergy with pomalidomide plus dexamethasone treatment. Importantly, the in vivo synergy observed in this combination was more evident in large, highly proliferative tumors, and was shown to be mediated by the impairment of mitosis transcriptional control, an increase in monopolar spindles, cell cycle arrest and the induction of apoptosis in cells in proliferative phases. In addition, the triple combination increased the activation of the proapoptotic protein BAX, which has previously been associated with sensitivity to filanesib, and could potentially be used as a predictive biomarker of response to this combination. Our results provide preclinical evidence for the potential benefit of the combination of filanesib with pomalidomide and dexamethasone, and supported the initiation of a recently activated trial being conducted by the Spanish Myeloma group which is investigating this combination in relapsed myeloma patients. Introduction The use of novel agents has resulted in a clear improvement in the survival of multiple myeloma (MM) patients. However, most patients eventually relapse,1 denoting the need for new drugs targeting key pathogenic mechanisms of the tumor plasma cell.2 CYCLIN D deregulation is a common oncogenic event found in 98% of MM patients,3 and considerable effort has been expended in trying to identify agents targeting the cell cycle of myeloma cells. Examples of such molecules are seliciclib (an inhibitor of CDK4/CDK6)4 and AURORA KINASE inhibitors.5 Unfortunately, thus far these agents have not proved to be sufficiently effective or were stopped due to toxicity6 in myeloma patients, who subsequently continued MM disease development. Filanesib (ARRY-520), a first-in-class7 kinesin spindle protein (KSP) inhibitor, is a novel agent targeting this same pathogenic area.8 KSP (EG5/KIF11) is a member of the mitotic kinesin family that is only expressed in dividing cells9 and is essential for establishing the mitotic bipolar spindle and ensuring centrosome separation.10 haematologica | 2017; 102(12)

Correspondence: emocio@usal.es

Received: March 13, 2017. Accepted: August 29, 2017. Pre-published: August 31, 2017. doi:10.3324/haematol.2017.168666 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2113 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2113

S. Hernández-García et al.

Inhibition of KSP activity arrests cells in metaphase by forming aberrant monopolar spindles and impairing the separation of centrosomes.11 The activity of filanesib is determined by two main factors: first, the integrity of components of the spindle checkpoint, which arrests cells when an alteration in mitosis is found, and second, the loss of anti-apoptotic signals12 during mitotic blockade, particularly the decrease in the MCL-1 protein.13 This latter protein is essential for the survival of MM cells,13,14 therefore, myeloma cells might be particularly susceptible to filanesib treatment.12 This agent has already been explored in MM in a phase II clinical trial, in which it gave a 16% response rate (≥partial response (PR)) in heavily treated patients who had received all available agents and a median of six previous lines of therapy.7 This initial activity prompted the search for potential combinations that could enhance the activity of the current backbones of therapy in relapsed MM. In this context, pomalidomide in combination with dexamethasone induces a 30% overall response rate and prolongs overall survival by up to one year in patients already exposed to immunomodulatory drugs (IMiDs) and proteasome inhibitors and refractory to the last line of therapy.15,16 However, novel partners for combination with this doublet are currently being sought, with the aim of improving these results. In the present study we evaluated the preclinical antimyeloma activity of the triple combination of pomalidomide+dexamethasone+filanesib (PDF). Preliminary data reported synergy of filanesib with pomalidomide in a xenograft mouse model.17 Herein, we demonstrate that filanesib is a good partner for combination with all IMiDs plus dexamethasone, the combination with pomalidomide being particularly potent in the dexamethasone sensitive MM.1S cell line, and very effective in a large panel of other MM cell lines. This synergistic effect is partly mediated by an increase in monopolar spindle formation and the simultaneous upregulated expression and activation of the proapoptotic protein BAX in actively proliferating myeloma cells. These findings supported the ongoing clinical trial (clinicaltrials.gov Identifier: 02384083) conducted by the Spanish MM group to evaluate the safety and efficacy of this triple combination in relapsed/refractory MM patients.

Methods (For more specific information see the Online Supplementary File) Reagents and drugs Filanesib was provided by Array BioPharma Inc. (Boulder, CO, USA). Thalidomide, lenalidomide and pomalidomide were purchased from Selleckchem (Houston, TX, USA), dexamethasone from Sigma-Aldrich (St Louis, MO, USA) and bortezomib from LC Laboratories (Woburn, MA, USA).

MM cell lines, patient samples and cultures Origin, authentication and in vitro growth conditions of MM cell lines have already been characterized.18,19 The study of the activity in the presence of interleukin (IL)-6, insulin-like growth factor (IGF)-1 or co-culture with stroma was performed as described.20,21 Bone marrow (BM) samples from MM patients were obtained following institutional approval and written informed consent. 2114

Cell viability, cell cycle and apoptosis assays Cell viability of MM cells was evaluated by the MTT assay.22 The cell cycle profile, apoptosis induction and mitochondrial membrane potential was analyzed by flow as described.21 In vitro synergism was quantitated using CalcuSyn software23 (Biosoft, Ferguson, MO, USA) via a constant ratio drug combination design. Ex vivo analysis of apoptosis in freshly isolated patient cells. Cytometry analyses of apoptosis in BM tumor plasma cells and lymphocytes were performed as described.24

Immunofluorescence study MM.1S cells treated for 24h were fixed, blocked and incubated overnight with a primary anti-α-tubulin antibody (Sigma) and counterstained with Alexa Fluor 488 conjugated anti-mouse immunoglobulin G (IgG) secondary antibody (Life Technologies, Waltham, MA, USA) and a 4',6-Diamidino-2-Phenylindole, Dihydrochloride (DAPI) solution (Sigma).

Western blot All procedures were performed as described.24 All antibodies were purchased from Cell Signaling Technology (Boston, MA, USA) except for anti-MCL-1 (Santa Cruz Biotechnology. Santa Cruz, CA, USA), anti-CYTOCHROME C (Calbiochem, Billerica, MA, USA) and Horseradish peroxidase-conjugated secondary antibodies (GE Healthcare. Little Chalfont, UK).

Gene silencing with siRNA MM.1S cells were transiently transfected with either 100 nM non-targeting control short interfering RNA (NT-siRNA) or 100 nM ON TARGET plus SMART pool siRNA targeting human BAX (BAX-siRNA; Dharmacon, Lafayette, CO, USA) using the Nucleofector II system (Lonza, Allendale, NJ, USA).

Animal models The human subcutaneous plasmacytoma model in CB17severe combined immunodeficiency (SCID) mice (The Jackson Laboratory, Bar Harbor, ME, USA) was used as described.21,25 Animal experiments were conducted with permission from the local Ethical Committee for Animal Experimentation.

Histological and immunohistochemistry (IHC) analyses This technique has been previously described25 using an antiBAX antibody (Cell Signaling, Boston, MA, USA) and the EnVision anti-mouse/rabbit peroxidase complexes (Dako, Glostrup, Denmark). Peroxidase activity was identified using the 3,3′-diaminobenzidine MAPO system (Ventana Medical Systems, Roche, Tucson, AZ, USA). Sections were counterstained with hematoxylin and analyzed by standard light microscopy. Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) was performed using the In Situ Cell Death Detection Kit (Roche, Mannheim, Germany). Sections were counterstained with DAPI and visualized by confocal laser microscopy (Leica TCS SP2).

RNA isolation, cDNA synthesis and microarray hybridization and analysis Tumors and MM.1S cells (n=3 for each treatment) were hybridized in Human Gene 2.0 ST arrays (Affymetrix) following manufacturer's instructions. The raw intensity data were preprocessed using the RMA algorithm26 implemented in the Affymetrix Expression Console, version 1.4.1.46. Differentially expressed genes were identified using the significance analysis of microarrays (SAM) algorithm,27 version 4.01, selecting genes with q value <0.05. Microarray data are available in the Gene Expression Omnibus (GEO) repository database (Accession number GSE94341). haematologica | 2017; 102(12)

Filanesib, pomalidomide and dexamethasone in MM Table 1. The most significant biological processes which were deregulated with the treatments.

In vitro

Biological Processes Common to PD, F, and PDF mitotic cell cycle process mitotic cell cycle phase transition cell cycle phase transition Biological Processes Common to PD and PDF ncRNA metabolic process RNA processing rRNA processing DNA replication DNA metabolic process Biological Processes Exclusive to PD protein targeting regulation of DNA recombination mRNA-containing ribonucleoprotein. complex export from nucleus mRNA export from nucleus DNA replication checkpoint Biological Processes Exclusive to PDF mitotic sister chromatid segregation nuclear chromosome segregation metaphase plate congression mitotic metaphase plate congression establishment of chromosome localization

In vivo

Biological Processes Common to PD and PDF mitotic cell cycle process Biological Processes exclusive to PD G1/S transition of mitotic cell cycle regulation of peptidase activity integrin-mediated signaling pathway regulation of cell protein metab. process regulation of protein metab. process Biological Processes exclusive to PDF nuclear division cell migration sterol biosynthetic process secondary alcohol metab. proc. sterol metabolic process mitotic nuclear division regulation of nervous system alcohol biosynthetic process mitotic cell cycle phase transition nervous system development brain development cholesterol metabolic process regulation of viral gen. replic. cell cycle phase transition regulation of nuclear division metaphase plate congression animal organ development

PDF

Adjusted P

Genes in GO dataset

Adjusted P

Genes in GO dataset

Adjusted P

Genes in GO dataset

1.85E-19 5.29E-15 1.35E-14

258 162 168

3.55E+00 1.19E+00 1.34E+00

3 3 3

5.18E-27 1.59E-15 4.80E-15

283 166 172

2.72E-30 8.19E-31 9.22E-21 1.98E-24 4.04E-22

211 293 114 126 290

3.45E-23 1.10E-21 2.66E-19 6.87E-19 4.25E-15

199 274 113 118 273

6.36E-03 7.51E-03 1.66E-02 1.66E-02 1.96E-02

162 26 36 36 12

9.07E-06 2.43E-04 7.02E-03 1.33E-02 1.96E-02

53 84 23 19 28

PD Adjusted P

F Genes in GO dataset

PDF Adjusted P

Adjusted P

Genes in GO dataset

2.45E+00

3.34E-02

2.11E-03 1.44E-02 3.23E-02 4.19E-02 4.34E-02

8 9 5 23 24

4.02E-05 3.67E-03 3.95E-03 4.17E-03 6.80E-03 9.05E-03 9.91E-03 1.17E-02 1.42E-02 1.65E-02 2.00E-02 2.08E-02 2.40E-02 2.48E-02 2.56E-02 2.73E-02 3.22E-02

21 27 6 8 8 14 20 7 15 39 18 7 6 15 8 5 50

Genes in GO dataset

List of the most significant ontological categories of biological processes (level 5 in the GO classification) in functional enrichment analysis (DAVID) for deregulated genes in either MM.1S cells or mouse tumors treated with PD, F and PDF. The number of deregulated genes for each treatment is represented in Venn diagrams (Figure 2C, D). Terms of the significant biological processes, Benjaminiâ&#x20AC;&#x201C;Hochberg adjusted P-value, and the number of genes present in each category are shown. Biological processes related to mitosis are shaded in blue. PD: pomalidomide+dexamethasone; F: filanesib; PDF: pomalidomide+dexamethasone+filanesib; GO: Gene Ontology; RNA: ribonucleic acid; G1/S: Gap 1/synthesis.

haematologica | 2017; 102(12)

2115

S. Hernández-García et al.

Figure 1. IMiDs, especially pomalidomide and dexamethasone, strongly synergized with filanesib. A) Cell viability of MM.1S cells treated with F plus TD, LD, PD, or the triple combination for 48h. Results are shown as the percentage of control. Data are summarized as the mean ± SD (n=3). Combination indexes (CI) for the combination of filanesib and each of the IMiDs together with dexamethasone are shown in Online Supplementary Figure S1B & S1D. Tumor growth curves of SCID mice randomized and treated when their tumors became palpable (B; small plasmocytomas; mean volume around 125 mm3; range 30 - 237; n=6 per group) or reached a substantial volume (D; large plasmacytomas; mean volume around 2500 mm3; range 1617 - 3134; n=3 per group) at the moment of treatment initiation. Experimental groups included: C (vehicle control; intraperitoneal injection (i.p.) 5 days per week), F (10 mg/kg i.p., 2 days per week), D (0.5 mg/kg i.p., 2 days per week), T (50 mg/kg i.p., 5 days per week), L (25 mg/kg i.p., 5 days per week), and P (8 mg/kg i.p., 5 days per week) in monotherapy or in double or triple combination. Data are summarized as the mean ± SEM. C & E) Analysis of survival by Kaplan-Meier estimator from mice with (C) small or (E) large tumors. F) BM cells from 9 MM patients incubated with P 500 nM, D 10 nM and F 1 nM alone and in the different combinations for 48h. Apoptosis induction was analyzed by Annexin-V staining by flow cytometry assay, both in plasma cells and lymphocytes that were identified based on the surface expression of CD38 and CD45, respectively. A statistically significant difference (P<0.05) was found between plasma cells treated with PDF and control. Error bars indicate the last “non-outlier” value at each side. F: filanesib; D: dexamethasone; T: thalidomide; L: lenalidomide; P: pomalidomide; n.s.: not significant.

2116

haematologica | 2017; 102(12)

Filanesib, pomalidomide and dexamethasone in MM

Statistical analyses Statistical analyses were performed with IBM SPSS Statistics v.21.0 (IBM Corp., Armonk, NY, USA).

Results Filanesib potentiates the anti-myeloma activity of pomalidomide in combination with dexamethasone Firstly, the ability of filanesib (F) to enhance the activity of the different IMiDs in combination with dexamethasone (D) was evaluated. MM.1S cells were treated for 48h with different doses of thalidomide + dexamethasone (TD), lenalidomide + dexamethasone (LD) or pomalidomide + dexamethasone (PD), with or without different concentrations of filanesib, and viability was analyzed by MTT assay (Figure 1A). All IMiDs + dexamethasone had a synergistic effect with filanesib, with combination indexes (CIs) in the synergistic range (most synergic CIs of 0.112, 0.103, and 0.063 for TDF, LDF and PDF, respective-

ly; Online Supplementary Figure S1). However, the most effective combination was that of F with PD (Figure 1A). This potentiation was also maintained in vivo in a subcutaneous plasmacytoma model, as filanesib enhanced the effect of TD, LD and PD in terms of delay in tumor growth (Figure 1B). In this regard, the PDF combination was particularly potent, as the addition of low doses of filanesib significantly reduced the mean tumor volume from day 26 of treatment compared with the standard of care PD. Moreover, the triple combination of PDF completely controlled tumor growth for up to 50 days. PDF was also superior to the other two tested combinations, TDF and LDF (StudentÂ´s t-test, P<0.05, from days 8 and 10 of treatment, respectively). The tumor growth control observed with PDF translated into a statistically significant improvement in the survival of treated mice, with a median survival (CI 95%) of 74 (71-76) days for PDF compared with 56 (55-61) days for mice treated with the standard backbone PD (log-rank test, P=0.004; Figure 1C). The survival analysis also favored PDF over the other IMiD com-

Figure 2. Effect of PD, F and PDF treatments on gene expression profiles. A & B) Venn diagrams of significantly deregulated genes in MM.1S cells (A) or tumors from mice (B) after treatment with PD, F and PDF vs. control. C & D) Venn diagrams of the biological processes (Gene Ontology at level 5) deregulated by the PD, F and PDF treatments vs. control in in vitro (C) and in vivo (D) studies. A list of the most significant biological processes deregulated in the in vitro and in vivo studies is shown in Table 1; only processes from level 5 Gene Ontology categories (the most specific processes) were considered, in order to avoid redundancy. P: pomalidomide; D: dexamethasone; F: filanesib; C: control.

haematologica | 2017; 102(12)

2117

S. Hernández-García et al.

binations (log-rank test, P<0.001 for PDF compared with TDF and LDF; Figure 1C). As filanesib exerts its activity on dividing cells, the PDF combination could be particularly active in highly proliferative cells. Therefore, we assessed the effect of PDF on mice bearing large plasmacytomas growing in the exponential phase. PDF was able to control these rapidly progressing tumors (Figure 1D), and there

was also a statistically significant advantage with respect to the reduction of tumor volume compared with filanesib in monotherapy from day 15, and from day 5 when compared with the PD double combination (Student’s t-test, P<0.05). As shown in Figure 1E, there was a noticeable improvement in median survival (CI 95%), from 14 (1117) and 18 (18-18) days for PD and F, respectively, to 59

Figure 3. PDF boosted the formation of monopolar spindles A) Quantitative determination of monopolar spindles in MM.1S cells treated with the vehicle (control), PD, F or PDF for 24h. Representative micrographs showing cells immunostained with anti-tubulin antibody (shown in green) to visualize microtubules and DAPI (blue) to evidence nuclei using confocal microcopy. Bar = 15 mm in lower magnification micrograph; bar = 5 mm in higher magnification insert). Data are presented as the mean ± SD of two independent experiments. Statistical significance was evaluated with Student’s t-test. B) Evaluation of monopolar spindles in paraffin sections of tumors from mice treated with vehicle (control=C), PD, F or PDF for two days and stained with hematoxylin and eosin (H&E). Thirty high-power fields (600x) were evaluated for each experimental condition (3 mice per condition). Monopolar spindles were characterized by chromosomes stained with hematoxylin (in purple) orientated in a ring. Bar = 10 μm). Data are presented as the mean ± SD. Statistical significance was evaluated with Student’s t-test. P: pomalidomide; D: dexamethasone; F: filanesib.

2118

haematologica | 2017; 102(12)

Filanesib, pomalidomide and dexamethasone in MM

Figure 4. PDF triggered cell cycle arrest and apoptosis of MM cells in pro-liferative phases. (A) Analysis of cell cycle by flow cytometry in MM.1S cells treated with the vehicle (control=C), PD, F or PDF for 48h after Draq5 staining (Draq5 binds to DNA from both living and non-living cells). B) Simultaneous analysis of cell cycle profile and apoptosis induction in MM.1S cells treated with C, PD, F or PDF for 48h by flow cytometry after Draq5/Annexin-V staining. C) Percentage of apoptotic cells in each phase of the cell cycle was also calculated. Data shown is for a representative experiment that was repeated at least twice. D) Western blot analysis of cell cycle proteins CYCLIN B1, phosphorylated HH3 and PARP in MM.1S cells treated with indicated conditions for 12h to 48h. E) Apoptosis induction in MM.1S cells treated with C, PD, F or PDF for 48h and analyzed by flow cytometry after staining with Annexin-V. F) Micrographs of tumor sections from mice treated for two days with C, PD, F or PDF and histochemically stained with TUNEL assay. Quantitative determination of TUNEL-positive cells (30 630x fields per experimental condition, 3 mice per condition). Data are expressed as the mean Âą SD (n=3); **P<0.01 (Studentâ&#x20AC;&#x2122;s t-test). P: pomalidomide; D: dexamethasone; F: filanesib.

haematologica | 2017; 102(12)

2119

S. HernĂĄndez-GarcĂa et al.

(45-73) days for animals treated with PDF (log-rank test, P<0.05 for all comparisons). It is remarkable that PDF was able to improve the survival of these end-stage mice by almost two months. With regard to toxicity, the PDF triple combination was well tolerated by mice in the initial experiment, and the effect was not worse than that of PD alone. In mice with large plasmacytomas, treatment induced a moderate decrease in body weight (not reaching 20%), which was recovered on the days without treatment, and which was partly due to the poor clinical condition of the mice (Online Supplementary Figure S2). Based on this data and the fact that the combination with pomalidomide and dexamethasone is particularly appealing in the clinical setting, we decided to expand our study of this combination in MM. First, its synergy was evaluated in some additional MM cell lines; MM144, JJN3, RPMI8226-LR5 and OPM-2; showing CIs in the synergistic range (Online Supplementary Figure S3). The potent antimyeloma effect of the PDF triple combination was maintained in the presence of IGF-1 or IL-6, two important factors in the myeloma BM microenvironment. Additionally, the PDF combination was also effective when the MM.1S cells were co-cultured with either human mesenchymal stromal cells from myeloma patients (BMSCs) or the human mesenchymal stromal line (hMSC-TERT) (Online Supplementary Figure S4). The effect of PDF was further evaluated ex vivo in BM aspirates of nine MM patients. Both filanesib alone as well as pomalidomide and dexamethasone were active in these patients. However, only the triple combination demonstrated a statistically significant difference when compared with the control (C) untreated cells (PDF vs. C, P=0.04; Figure 1F). Moreover, there was a clear therapeutic window, as the toxicity on the lymphocytes of these same patients was clearly lower. It is of note that although the efficacy of the triple combination PDF was observed in vitro in several MM cell lines as well as in patients samples, its benefit over LDF or TDF was only explored in a single cell line.

The PDF combination deregulates genes necessary for mitosis In order to investigate the mechanism of action of the triple combination, changes induced by the different treatments in the gene expression profiles of MM.1S cells in vitro or in tumors from treated mice were analyzed and compared with their respective untreated controls. In both studies, whereas filanesib alone induced a minimal genomic deregulation with respect to the control (6 and 4 genes significantly deregulated in the in vitro and in vivo experiments, respectively), treatment with PDF enhanced the effect of PD, particularly in the in vivo study (3460 vs. 3317 and 238 vs. 141 deregulated genes in the in vitro and in vivo studies, respectively; Figure 2A,B. See Online Supplementary Table S1 and Table S2 for differentially deregulated genes in the PDF combination in vitro and in vivo, respectively. A complete list of the 3460 genes deregulated in vitro is also provided as a supplement). Due to the variety of functions in which these genes were involved, a functional enrichment analysis was performed in order to identify the biological processes most significantly affected by the deregulated genes following exposure to PD, F and PDF treatments in both experiments (see Table 1 and Venn diagrams for biological processes in Figure 2C,D). Three biological processes were commonly 2120

deregulated by all treatments in the in vitro experiments (while no common process was discovered in the in vivo experiment), and all of them were related to mitosis and cell cycles (Table 1). Of interest, the top five biological processes exclusively deregulated in the PDF combination in vitro were involved in different stages of mitosis; in line with this, five out of the 17 biological processes uniquely altered in the PDF treatment in vivo were also associated with mitosis and nuclear division (see shaded processes in Table 1). Next, we selected those genes deregulated in the triple combination in vitro which were involved in mitotic and cell cycle processes (Online Supplementary Table S3). Of note, CCNB1 and CCNB2, encoding for B-type cyclins and being major regulators of the G2/M transition of cell cycle, are upregulated in the PDF treatment. The upregulated expression of PSMD3, a 26S proteasome subunit, and NEK2, involved in the anaphase promoting complex and centrosomal separation, may also be mediating the increased mitotic and cell cycle progression processes in this combination. Similarly, CDC25C and CDC25B, required for entry into mitosis, show an increased expression in the PDF combination.

The PDF combination enhances the formation of aberrant monopolar spindles One mechanism underlying the synergy of the triple combination could be the enhancement of the ability of filanesib to block mitosis and to induce monopolar spindles, the characteristic sequelae of KSP inhibition. To test this hypothesis, the formation of aberrant monopolar spindles was evaluated after a 24h treatment of MM.1S cells with vehicle, F, PD or PDF. Whereas the majority of untreated cells and those exposed to PD displayed normal mitoses with typical bipolar spindles, almost all F- and PDF-treated cells that were in mitosis showed a monopolar spindle phenotype (Figure 3A). Interestingly, and differentiating these latter two conditions, treatment with the triple combination increased the percentage of cells in mitosis, which resulted in a significant increase in the absolute number of monopolar spindles (0, 0, 8 and 19 aberrant spindles per 100 cells, for the control, PD, F and PDF, respectively (P=0.01 for the F vs. PDF comparison) (Figure 3A). These results were confirmed in vivo by immunohistochemistry, since no aberrant monopolar spindles were found in the control and PD-treated tumors, while F and PDF displayed an average of 6 and 11 monopolar spindles per field (P=0.008 for the F vs. PDF comparison; Figure 3B).

PDF causes cell cycle arrest in G2/M phases and specifically induces apoptosis in cells arrested in proliferative phases The increase in mitotic cells observed with the triple combination was intriguing, and therefore its specific effect on the cell cycle profile of treated tumor cells was studied (Figure 4A). To check whether there was a particular susceptibility to apoptosis in cells at certain phases of the cell cycle, simultaneous staining with Draq5 and Annexin-V was performed (Figure 4B). The treatment with filanesib in monotherapy arrested cells in the proliferative phases of the cell cycle, with 49% of cells in synthesis (S) and gap 2 mitosis (G2-M) phases compared with 36% in control cells (two-tailed Student's t-test from three independent experiments P=0.03). The PDF triple combihaematologica | 2017; 102(12)

Filanesib, pomalidomide and dexamethasone in MM

Figure 5. Role of the increase and activation of the proapoptotic protein BAX on the anti-myeloma efficacy of F and PDF treatments. (A) Cell viability of a panel of MM cell lines incubated with increasing concentrations of F for 48h and analyzed by MTT assay. Data are expressed as a percentage of control values. B) Western blot analysis of basal levels of six BCL-2 family members (BCL-2, BCL-XL, MCL-1, BAX, BAK, BAD) studied in the same MM cell lines. C) Expression levels for BAX were determined by densitometry analysis of bands (using ImageJ software) and relative to those of b-ACTIN. Relation between the basal levels of BAX protein and drug sensitivity to F was calculated with respect to IC50 for 48h. Statistical analysis was performed with the Mann-Whitney U test. D) Immunoblot of MM.1S cells 24h and 48h after transfection with BAX-specific siRNA. Nontransfected cells (NT) were also analyzed for comparison. 24h after siRNA transfection, cells were treated with either filanesib (F) or bortezomib (B) for an additional 24h, and apoptosis was assessed by Annexin-V+ cells and evaluated by flow cytometry. E) MM.1S cells were treated for 12-48h with 10 nM F, and expression levels of MCL-1 and BAX in the mitochondrial and cytosolic fractions were analyzed by western blot. COX IV and GAPDH expression were used as loading controls for mitochondrial and cytosolic, respectively. F & G) Immunoblot of total (F), mitochondrial and cytosolic (G) extracts from MM.1S cells treated for 12-48h with the vehicle (control=C), PD, F or PDF for 48h. The expression of COX IV and GAPDH proteins were used as mitochondrial and cytosolic loading controls, respectively. H) Analysis of mitochondrial membrane potential in MM.1S cells treated with different conditions as assessed by flow cytometry after TMRE staining. I) Immunohistochemical staining of BAX in tumors from mice treated for two days under the indicated conditions. Bar = 20 mm. In Figure E, “BAX over” indicates overexposure of the film to better visualize mild changes in the levels of this protein; “p18 BAX” indicates the proapoptotic 18 kDa BAX fragment. P: pomalidomide; D: dexamethasone.

haematologica | 2017; 102(12)

2121

S. HernĂĄndez-GarcĂa et al.

nation did not show much difference in this regard compared with filanesib alone, with 44% of cells in G2-M arrest (Figure 4A). However, treatment with filanesib specifically induced apoptosis of cells in G2-M phases with 58% of them being apoptotic compared with only 33% in non-proliferating phases (Figure 4B,C). This fact was significantly enhanced in the triple combination as most apoptotic cells (88%) were seen to be in proliferative phases (G2 or M) (Figure 4B,C). In order to confirm these findings, various cell cycle markers were biochemically analyzed. MM.1S cells treated with filanesib and the triple combination showed an accumulation of CYCLIN B1 levels at 12h and 24h as shown by western blot, this being indicative of the activation of the spindle assembly checkpoint (SAC),28 which correlated with the percentage of cells arrested in G2-M phases. In contrast to treatment with filanesib, CYCLIN B1 levels rapidly decreased after 48h in the PDF combination, possibly indicating an attempt of the cells to exit from mitosis.29,30 However, they proved unable to do so, as after 48h cells treated with PDF died in mitosis via apoptosis, as shown by an increase in phosphorylated histone H3 protein levels, a marker of mitotic cells,31 and PARP cleavage and degradation32 (Figure 4D). The increase in apoptotic induction was confirmed by Annexin-V/IP staining, as treatment of MM.1S with both PD and filanesib for 48h induced an increase of AnnexinV-positive cells (23% and 56%, respectively) compared with the control (5%), and this effect was much greater with the triple combination (70% apoptosis; Figure 4E). In the same manner, the TUNEL assay of treated tumors revealed a significant increase in apoptotic cells in the triple combination compared with filanesib or PD alone (Figure 4F).

Anti-myeloma efficacy of filanesib in monotherapy depends on the levels of both the anti-apoptotic protein MCL-1 and the proapoptotic protein BAX To gain insight into the mechanism underlying the enhanced apoptotic response in dividing cells with PDF treatment, key molecular features associated with sensitivity to filanesib alone were investigated. A dose response to filanesib at 48h in 11 cell lines showed that, although all of them were sensitive to this agent, some cell lines were more sensitive to filanesib treatment, with almost no viable cells at 2.5 nM (Figure 5A). Other cells, however, had higher IC50 values and 30-50% of cells remained viable even at higher doses of the drug (Figure 5A). Given that the sensitivity to filanesib has previously been correlated with levels of MCL-1, the basal expression of six representative BCL-2 family members was evaluated in the same panel of MM cell lines showing different patterns of expression (Figure 5B). No correlation was observed between MCL-1 expression and filanesib sensitivity. However, the sensitivity to filanesib showed a direct correlation with basal expression levels of BAX, as shown in Figure 5C (P<0.05), whereby cell lines with higher basal BAX levels were particularly sensitive to this agent. In contrast, no correlation was observed between BAX expression and sensitivity to the proteasome inhibitor bortezomib, which we used here as a control (Online Supplementary Figure S5). To confirm the role of BAX in filanesib-induced cell death, MM.1S cells were transfected with BAX-specific siRNA. BAX knockdown induced resistance to filanesib treatment, reducing the 2122

number of apoptotic cells by half (Figure 5D). A similar effect related to BAX levels was observed in OPM-2 cells (data not shown). Moreover, subcellular fractionation studies indicated that treatment with filanesib triggered the translocation of BAX from the cytoplasm to the mitochondria, where it was cleaved into the very potent proapoptotic 18 kDa fragment33 (Figure 5E). Simultaneously, the levels of the prosurvival protein MCL-1 also decreased in both subcellular compartments (Figure 5E). These findings suggest the potential importance of the modulation of both BAX and MCL-1 protein levels in filanesib-induced apoptosis.

The PDF triple combination induces augmented expression and activation of BAX protein Interestingly, and unlike filanesib alone, MCL-1 levels were not significantly affected after treatment with PDF. In contrast, two events were clearly potentiated in the triple combination compared with filanesib alone: first, PDF induced the expression of total BAX levels in vitro (Figure 5F), and second, PDF enhanced the presence of this protein in the mitochondria, where it exerts its apoptotic activity (Figure 5G). This event led to the permeabilization of the external mitochondrial membrane, resulting in the release of the apoptogenic factor CYTOCHROME C into the cytosol (Figure 5G), and an associated decrease in mitochondrial membrane potential, as observed by flow cytometry with tetramethylrhodamine, ethyl ester (TMRE; Figure 5H). Finally, in vivo experiments confirmed the PDF-induced BAX upregulation (Figure 5I).

Discussion MM remains an incurable disease in most patients34 and there is an urgent need for novel drugs to improve this situation. In the work herein, we have preclinically evaluated the combination of various IMiDs together with dexamethasone and filanesib, a novel KSP inhibitor that particularly affects dividing cells and depends on the survival protein MCL-1, which is essential for MM cell endurance.13,14 This novel therapeutic agent has already demonstrated an anti-myeloma effect with dexamethasone,7 and is currently being explored in combination with proteasome inhibitors.35 Preliminary data have demonstrated synergy of filanesib with pomalidomide in a mouse xenograft model,17 and our in vitro, ex vivo and, notably, in vivo results clearly confirm this effect and provide evidence of the benefit of the addition of dexamethasone. This strong synergistic effect was confirmed in several MM cell lines, even in the presence of stromal cells. The potency of the combination with pomalidomide was shown to be better than those including thalidomide or lenalidomide, however, one limitation of the present study is that the advantages bestowed by PDF over TDF or LDF was only demonstrated in one cell line, that of MM.1S, particularly in the in vivo model. Nevertheless, taking into account that pomalidomide and dexamethasone is currently considered a backbone for combinations we decided to elaborate on this particular combination. Since filanesib is an agent designed to interfere with the mechanisms of cell division, it is reasonable to think that it might be particularly effective in actively dividing cells. In line with this, the in vivo effect of the triple combination was particularly potent in mice bearing large tumors in the haematologica | 2017; 102(12)

Filanesib, pomalidomide and dexamethasone in MM

exponential phase of growth. Interestingly, this combination resulted in an extended survival of almost two months, on average, in these end-stage mice. Analysis of the gene expression profile of both MM cells and PDFtreated tumors showed a substantial proportion of upregulated genes to be involved in the centrosome separation (NEK2), control of spindle assembly checkpoint (CCNB1, CCNB2) and entry into mitosis (CDC25 C and CDC25 B), implying that the triple combination significantly alters mitotic processes. Recently, the overexpression of CCNB2 was associated with an acceleration in centrosome separation.36 Based on these data, the antimyeloma effect of the triplet could be due to an increased mitosis and an increased monopolar spindle formation. In order to investigate this hypothesis, the phenotype of cells in metaphase was evaluated by immunohistochemistry. The triple combination induced an increase in the frequency of cells with an aberrant monopolar spindle phenotype compared with filanesib monotherapy. Accordingly, we demonstrated that the PDF combination generated the same degree of G2-M arrest as did filanesib alone, but the combination provoked a substantially more specific apoptosis of these cells when arrested in proliferative phases. A delicate balance between pro-apoptotic and anti-apoptotic members of the BCL-2 family and their subcellular localization generally determines the fate of proliferative cells treated with anti-mitotic agents.37-40 In particular, it has been reported that apoptosis is enhanced by the activation of BAX in cancer cells using other KSP inhibitors.41,42 Accordingly, silencing BAX prompted cell survival under filanesib treatment in monotherapy. In line with this, PDF treatment induced increased apoptosis in proliferative phases that was associated with a significantly higher expression of

References 1. Kumar SK, Lee JH, Lahuerta JJ, et al. Risk of progression and survival in multiple myeloma relapsing after therapy with IMiDs and bortezomib: a multi-center international myeloma working group study. Leukemia. 2012;26(1):149-157. 2. Ocio EM, Richardson PG, Rajkumar SV, et al. New drugs and novel mecha-nisms of action in multiple myeloma in 2013: a report from the International Myeloma Working Group (IMWG). Leukemia. 2014; 28(3):525-542. 3. Bergsagel PL, Kuehl WM, Zhan F, Sawyer J, Barlogie B, Shaughnessy J,Jr. Cyclin D dysregulation: an early and unifying pathogenic event in multiple mye-loma. Blood. 2005; 106(1):296-303. 4. Niesvizky R, Lentzsch S, Badros AZ, et al. A phase I study of PD 0332991: complete CDK4/6 inhibition and tumor response in sequential combination with bortezomib and dexamethasone for relapsed and refractory multiple myeloma. ASH Annual Meeting Abstracts. 2010; 116(21):860. 5. Rosenthal A, Kumar S, Hofmeister C, et al. A phase Ib study of the combination of the aurora kinase inhibitor alisertib (MLN8237) and bortezomib in relapsed multiple myeloma. Br J Haematol. 2016; 174(2):323325.

haematologica | 2017; 102(12)

the proapoptotic BAX protein, promoting its activation. In parallel with these results, the tumors of mice treated with PDF also showed higher immunoreactivity for BAX. Moreover, we detected a release of the apoptogenic factor CYTOCHROME C into the cytosol, possibly due to permeabilization of the mitochondrial membrane induced by BAX. These experiments suggest that the expression levels and activation of BAX are crucial in determining the fate of MM cells treated with PDF. In conclusion, we report the strong synergy of the PDF triple combination in vitro and especially in vivo in preclinical models of MM. The activity of PDF relies on the induction of monopolar spindles and arrest in mitosis, characterized by primary activation of SAC; increased apoptosis in these mitosis-arrested cells is characterized by the increased expression and activation of BAX, which could be a useful predictive biomarker of response to PDF in MM. These results support the phase II clinical trial POMDEFIL, which is currently being conducted by the Spanish MM group to evaluate this combination in refractory MM patients. Funding This work was funded in part by Array BioPharma, by the Spanish ISCIII-FIS (PI 15/0067 and PI 15/2156) and FEDER, the Spanish RTICC (RD12/0036/0058), Spanish Association Against Cancer (AECC, GCB120981SAN) and the Regional Council of Castilla y LeĂłn (GRS 1029/A/14, GRS 1175/A/15 and FIC335U14). Acknowledgments The authors thank Phil Mason for his help in reviewing the English language of our manuscript.

6. Kollareddy M, Zheleva D, Dzubak P, Brahmkshatriya PS, Lepsik M, Hajduch M. Aurora kinase inhibitors: progress towards the clinic. Invest New Drugs. 2012; 30(6):2411-2432. 7. Lonial S, Shah JJ, Zonder J, Bensinger, W.I., Cohen AD, Kaufman JL, et al. Prolonged survival and improved response rates with ARRY-520 in relapsed/refractory multiple myeloma (RRMM) patients with low Îą-1 acid glycoprotein (AAG) levels: results from a phase 2 study. Blood. 2013; 122(21):285. 8. Ocio EM, Mitsiades CS, Orlowski RZ, Anderson KC. Future agents and treatment directions in multiple myeloma. Expert Rev Hematol. 2014;7(1):127-141. 9. Jackson JR, Patrick DR, Dar MM, Huang PS. Targeted anti-mitotic therapies: can we improve on tubulin agents? Nat Rev Cancer. 2007;7(2):107-117. 10. Blangy A, Lane HA, d'Herin P, Harper M, Kress M, Nigg EA. Phosphoryla-tion by p34cdc2 regulates spindle association of human Eg5, a kinesin-related motor essential for bipolar spindle formation in vivo. Cell. 1995;83(7):1159-1169. 11. Stern BM, Murray AW. Lack of tension at kinetochores activates the spindle checkpoint in budding yeast. Curr Biol. 2001; 11(18):1462-1467. 12. Tunquist BJ, Woessner RD, Walker DH. Mcl-1 stability determines mitotic cell fate

13. 14. 15.

16.

17.

18.

of human multiple myeloma tumor cells treated with the kinesin spindle protein inhibitor ARRY-520. Mol Cancer Ther. 2010;9(7):2046-2056. Peperzak V, Vikstrom I, Walker J, et al. Mcl1 is essential for the survival of plasma cells. Nat Immunol. 2013;14(3):290-297. Zhang B, Gojo I, Fenton RG. Myeloid cell factor-1 is a critical survival factor for multiple myeloma. Blood. 2002;99(6):1885-93. San Miguel J, Weisel K, Moreau P, et al. Pomalidomide plus low-dose dex-amethasone versus high-dose dexamethasone alone for patients with relapsed and refractory multiple myeloma (MM-003): a randomised, open-label, phase 3 trial. Lancet Oncol. 2013;14(11):1055-1066. Dimopoulos MA, Palumbo A, Corradini P, et al. An updated analysis of the STRATUS trial (MM-010): safety and efficacy of pomalidomide plus low-dose dexamethasone (POM + LoDEX) in patients (Pts) with relapsed/refractory multiple myeloma (RRMM). Blood. 2015;126(23):4225. Humphries MJ, Anderson D, Williams L, Rieger R, Tunquist B, Walker D. ARRY-520 combined with pomalidomide displays enhanced anti-tumor activity in preclinical models of multiple myeloma. Blood. 2013;122(21):3167. Paino T, Garcia-Gomez A, GonzalezMendez L, et al. The novel pan-PIM kinase inhibitor, PIM447, displays dual antimyelo-

2123

S. Hernández-García et al.

19.

20.

21.

22.

23.

24.

25.

2124

ma and bone-protective effects, and potently synergizes with current standards of care. Clin Cancer Res. 2017;23(1):225238. Herrero AB, San Miguel J, Gutierrez NC. Deregulation of DNA double-strand break repair in multiple myeloma: implications for genome stability. PLoS One. 2015;10(3):e0121581. Garcia-Gomez A, Quwaider D, Canavese M, et al. Preclinical activity of the oral proteasome inhibitor MLN9708 in Myeloma bone disease. Clin Cancer Res. 2014;20(6):1542-1554. Ocio EM, Maiso P, Chen X, et al. Zalypsis: a novel marine-derived com-pound with potent antimyeloma activity that reveals high sensitivity of malignant plasma cells to DNA double-strand breaks. Blood. 2009;113(16):3781-3791. Paino T, Sarasquete ME, Paiva B, et al. Phenotypic, genomic and functional characterization reveals no differences between CD138++ and CD138low sub-populations in multiple myeloma cell lines. PLoS One. 2014;9(3):e92378. Chou TC. Drug combination studies and their synergy quantification using the Chou-Talalay method. Cancer Res. 2010; 70(2):440-446. Maiso P, Carvajal-Vergara X, Ocio EM, et al. The histone deacetylase in-hibitor LBH589 is a potent antimyeloma agent that overcomes drug resistance. Cancer Res. 2006;66(11):5781-5789. Ocio EM, Vilanova D, Atadja P, et al. In vitro and in vivo rationale for the triple combination of panobinostat (LBH589) and dexamethasone with either bortezomib or

26.

27.

28. 29. 30. 31.

32.

33.

34.

lenalidomide in multiple myeloma. Haematologica. 2010;95(5):794-803. Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP. Summa-ries of Affymetrix GeneChip probe level data. Nucleic Acids Res. 2003;31(4):e15. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA. 2001;98(9):51165121. Musacchio A, Salmon ED. The spindleassembly checkpoint in space and time. Nat Rev Mol Cell Biol. 2007;8(5):379-393. Pines J. Cubism and the cell cycle: the many faces of the APC/C. Nat Rev Mol Cell Biol. 2011;12(7):427-438. Murray A. Cyclin ubiquitination: the destructive end of mitosis. Cell. 1995;81(2):149-152. Schimming TT, Grabellus F, Roner M, et al. pHH3 immunostaining improves interobserver agreement of mitotic index in thin melanomas. Am J Dermatopa-thol. 2012;34(3):266-269. Kaufmann SH, Desnoyers S, Ottaviano Y, Davidson NE, Poirier GG. Spe-cific proteolytic cleavage of poly(ADP-ribose) polymerase: an early marker of chemotherapyinduced apoptosis. Cancer Res. 1993;53(17):3976-3985. Cao X, Deng X, May W. Cleavage of Bax to p18 Bax accelerates stress-induced apoptosis, and a cathepsin-like protease may rapidly degrade p18 Bax. Blood. 2003;102(7):2605-2614. Barlogie B, Mitchell A, van Rhee F, Epstein J, Morgan GJ, Crowley J. Curing myeloma at last: defining criteria and providing the

evidence. Blood. 2014;124(20):3043-3051. 35. Shah JJ, Thomas S, Weber DM, Wang M, Orlowski R. Novel kinesin spindle protein inhibitor ARRY–520 + carfilzomib(Car) in patients with relapsed and/or refractory multiple myeloma (RRMM). Haematologica. 2013;98(S1):Abstract-S579. 36. Nam HJ, van Deursen JM. Cyclin B2 and p53 control proper timing of cen-trosome separation. Nat Cell Biol. 2014;16(6):538549. 37. Ferlini C, Cicchillitti L, Raspaglio G, et al. Paclitaxel directly binds to Bcl-2 and functionally mimics activity of Nur77. Cancer Res. 2009;69(17):6906-6914. 38. Vijapurkar U, Wang W, Herbst R. Potentiation of kinesin spindle protein inhibitor-induced cell death by modulation of mitochondrial and death receptor apoptotic pathways. Cancer Res. 2007;67(1):237-245. 39. Haschka MD, Soratroi C, Kirschnek S, et al. The NOXA-MCL1-BIM axis de-fines lifespan on extended mitotic arrest. Nat Commun. 2015;6:6891. 40. Wang P, Lindsay J, Owens TW, et al. Phosphorylation of the proapoptotic BH3only protein bid primes mitochondria for apoptosis during mitotic arrest. Cell Rep. 2014;7(3):661-671. 41. Tao W, South VJ, Zhang Y, et al. Induction of apoptosis by an inhibitor of the mitotic kinesin KSP requires both activation of the spindle assembly check-point and mitotic slippage. Cancer Cell. 2005;8(1):49-59. 42. Tao W, South VJ, Diehl RE, et al. An inhibitor of the kinesin spindle protein activates the intrinsic apoptotic pathway independently of p53 and de novo pro-tein synthesis. Mol Cell Biol. 2007;27(2):689-698.

haematologica | 2017; 102(12)

ARTICLE

Cell Therapy & Immunotherapy

Steroid treatment of acute graft-versus-host disease grade I: a randomized trial

Andrea Bacigalupo,1 Giuseppe Milone,2 Alessandra Cupri,2 Antonio Severino,3 Franca Fagioli,4 Massimo Berger,4 Stella Santarone,5 Patrizia Chiusolo,1 Simona Sica,1 Sonia Mammoliti,6 Roberto Sorasio,7 Daniela Massi,8 Maria Teresa Van Lint,9 Anna Maria Raiola,9 Francesca Gualandi,9 Carmine Selleri,10 Maria Pia Sormani,11 Alessio Signori,11 Antonio Risitano12 and Francesca Bonifazi;13 for the Gruppo Italiano Trapianto di Midollo Osseo (GITMO)

Istituto di Ematologia, Fondazione Policlinico Universitario A Gemelli, Università Cattolica, Roma; 2Cattedra di Ematologia, Universita’ di Catania; 3Divisione di Ematologia Ospedale San Camillo, Roma; 4Divisione di Oncologia Pediatrica, Ospedale Regina Margherita, Torino; 5Divisione di Ematologia, Ospedale di Pescara; 6GITMO Clinical Trial Co-ordinator; 7Divisione di Ematologia, Ospedale Santa Croce, Cuneo; 8 Divisione di Anatomia Patologica, Dipartimento di Chirurgia e Medicina Traslazionale, Università degli Studi di Firenze; 9Divisione di Ematologia, IRCCS AOU San Martino IS, Genova; 10Divisione di Ematologia, Dipartimento di Medicina e Chirurgia, Università di Salerno; 11Cattedra Di Statistica Medica, Universita’ di Genova; 12Cattedra di Ematologia, Università Federico II, Napoli and 13Ematologia “Seràgnoli”, Azienda OspedalieraUniversitaria S. Orsola–Malpighi, Bologna, Italy

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(12):2125-2133

ABSTRACT

atients with acute graft-versus-host disease (GvHD) grade I were randomized to an observation arm (n=85) or to a treatment arm (n=86) consisting of 6-methylprednisolone 1 mg/kg/day, after stratification for age and donor type. The primary end point was development of grade II-IV GvHD. The cumulative incidence of grade II-IV GvHD was 50% in the observation arm and 33% in the treatment arm (P=0.005). However, grade III-IV GvHD was comparable (13% vs. 10%, respectively; P=0.6), and this was true for sibling and alternative donor transplants. Moderate/severe chronic GvHD was also comparable (17% vs. 9%). In multivariate analysis, an early interval between transplant and randomization (<day +20) was the only negative predictor of grade III-IV GvHD. Patients in the observation arm had less infectious bacterial episodes (12 vs. 25; P=0.04), less severe infectious fungal episodes (0 vs. 3; P=0.04), and less severe adverse events (3 vs. 11; P=0.07). At five years, non-relapse mortality was 20% versus 26% (P=0.2), relapse-related mortality 25% versus 21%, and actuarial survival was 51% versus 41% (P=0.3) in the observation and treatment arms, respectively. In multivariate analysis, advanced disease phase, older age and an early onset of GvHD were significant negative predictors of survival, independent of the randomization arm. In conclusion, steroid treatment of acute grade I GvHD prevents progression to grade II but not to grade III-IV GvHD, and there is no effect on non-relapse mortality and survival. Patients treated with steroids are at a higher risk of developing infections and have more adverse events. (Trial registered as EUDTRACT 2008-000413-29).

Correspondence: apbacigalupo@yahoo.com

Received: May 11, 2017. Accepted: September 22, 2017. Pre-published: September 29, 2017. doi:10.3324/haematol.2017.171157 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2125 ©2017 Ferrata Storti Foundation

Introduction There is uncertainty as to whether grade I acute graft-versus-host disease (GvHD), that is a skin rash over less than 50% of the body surface, without liver or gut involvement, should be treated or not. In three prospective trials of firstline treatment, also patients with grade I acute GvHD (aGvHD) were enrolled;1-3 however, most Centers would probably treat only GvHD of grade II or over. One argument in favor of steroid treatment would be early intervention, thus possibly preventing progression to more severe GvHD. This is a general rule of medicine, but there is no evidence that this is also the case in patients with aGvHD.1,2 In one randomized study back in 1998, the Gruppo Italiano Trapianti di Midollo Osseo haematologica | 2017; 102(12)

2125

Author et al.

(GITMO) had shown that early intervention with highdose steroid treatment (10 mg/kg) was equally effective as a conventional dose of steroids (2 mg/kg) in first-line treatment of aGvHD.3 In that study, the proportion of patients who progressed to grade III-IV was similar in the two groups, despite a median interval between transplant and treatment of 12 days, arguing against the hypothesis that early aggressive intervention would be more effective than standard therapy and would be able to modify the natural disease course.3 Similar results were seen in a more recent prospective randomized trial, once again comparing two different doses of steroids as first-line treatment, and again showing no difference in the rate of progression to severe GvHD.4 In addition, steroids cause immune deficiency and promote infectious complications.5 On the other hand, early treatment of GvHD could be beneficial. In a retrospective study of unrelated donor transplants in two different Centers, non-relapse mortality (NRM) was lower in one Center using anti-thymocyte globulin (ATG) for GvHD prophylaxis and steroid treatment of grade I aGvHD.6 In any case, whether grade I GvHD should be treated or not has not been tested in a prospective trial, and this led GITMO to undertake this trial. The aim was how to calculate the lowest and highest success rate. We used data from the previous GITMO study:2 25% of patients with grade I GvHD treated with 6methylprednisolone (6MPred) 2 mg/kg progressed to grade II-IV GvHD. We hypothesized that patients left untreated would have twice the risk of progression to grade II-IV GvHD, and 170 patients were needed to test this hypothesis. We report the results of this trial in patients with grade I GvHD, randomized to receive steroid treatment or no treatment.

Methods Study design This is a Gruppo Italiano Trapianto di Midollo Osseo (GITMO) study (study name: RAMP08; registered as EUDTRACT N 2008000413-29). The study was conducted according to Good Clinical Practice (GCP) and the Declaration of Helsinki. The study protocol was approved by all local ethical committees. Data entry was made by electronic CRF provided by Clinical Research Technolgy (CRT), Naples, Italy. The study is an open label multicenter, phase III randomized study comparing no treatment versus treatment with 6-methylprednisolone (6MPred) 1 mg/kg per day for transplanted patients with grade I aGvHD according to Gucksberg's criteria.7 Randomization was managed centrally via the web in a 1:1 ratio. Patients were randomized using a dynamic randomization algorithm, with minimization of differences between arms A and B to no more than 2 patients overall and 3 patients per strata. Patients were stratified according to phase (early/advanced) and donor type (matched sibling/alternative donors). We applied a modified intention-to-treat analysis, and all patients with at least one day of follow up were analyzed in the arm to which they had been allocated: 173 patients were randomized, between July 2009 and August 2014, and 171 were analyzed. The study outline is shown in Figure 1. Patients randomized to the observation arm were left untreated. Patients progressing to grade II-IV GvHD were considered to have reached the primary end point of the study, independent of the interval from randomization, and were treated according to standard procedures of each Center. Patients randomized to observation and not progressing were followed up. 2126

Table 1. Clinical data of patients randomized.

N Age Sex M/F Diagnosis SAA AML ALL CML MDS Myelofibrosis Myeloma CLL NHL HD Other Disease phase:early Donor type Matched SIBS UD HAPLO CB Conditioning MA/RIC

Observation 85

Treatment 86

46 (1-69) 37/48

38 (0.4-68) 35/51

0.06 0.7

2 41 16 3 5 4 3 3 4 2 2 43 (53%)

2 34 23 1 7 5 6 2 3 1 2 38 (47%)

0.7

36 (42.4%) 36 (42.4%) 6 (7.1%) 7 (8.2%) 64/21

34 (39.5) 44 (51.2%) 7 (8.1% 1 (1.2%) 61/25

0.1

0.7

0.8

M/F: male / female; SAA: severe aplastic anemia; AML: acute myeloid leukemia; ALL: acute lymphoblastic leukemia; CML: chronic myeloid leukemia; MDS: myelodysplastic syndromes; CLL: chronic lymphocytic leukemia; NHL: non-Hodgkin lymphoma; HD: Hodgkin disease; UD: unrelated donor; SIBS: siblings; HAPLO: HLA haploidentical donors; CB: cord blood; MA: myeloablative conditioning.; RIC: reduced intensity conditioning.

Patients randomized to treatment received 6MPred 1 mg/kg/day for five days. Patients progressing to grade II-IV GvHD, had reached the primary end point of the sudy and were treated according to the policy of each Center. If GvHD did not progress, 6MPred was tapered as follows: 0.75 mg/kg/day on days 6-10, 0.5 mg/kg/day on days 11-15, 0.25 mg/kg/day on days 16-20, 0.12 mg/kg/day on days 21-30, and discontinued on day +30.

End points The primary end point was the cumulative incidence of patients progressing to grade II-IV aGvHD. Secondary end points were: proportion of patients with grade III-IV GvHD, proportion of bacterial infections, viral infection, fungal infections, number of adverse events and severe adverse events, cumulative incidence of non-relapse mortality (NRM), cumulative incidence of relapse, proportion of patients developing chronic GvHD (limited and extensive), actuarial overall survival (OS).

Inclusion and exclusion criteria Study inclusion criteria were: age 0-70 years, having received an allogeneic stem cell transplant for malignant or non-malignant diseases, developing a skin rash over 10-49% of the body surface within the previous 48 hours, having received an unmanipulated graft from any donor type, and not having received previous treatment with steroids. Signed informed consent was obtained from adults or, in the case of pediatric cases, their tutors. Conventional GvHD prophylaxis was given to all patients with cyclosporin methotrexate, with the addition of ATG for unrelated donors, and post-transplant cyclophosphamide (PT-CY) for the small number (n=15) of HAPLO grafts. A skin biopsy , was recommended but haematologica | 2017; 102(12)

Running Title

Figure 1. Study outline. Patients randomized to the observation (n=85) or treatment (n=86) arms all went forward for analysis. Two patients were not evaluable because essential data were missing (1 observation arm; 1 treatment arm). 6MPred: 6 mthylprednisolone; FU: follow up.

not mandatory; centralized histopathology was provided (D Massi, Florence, Italy). Exclusion criteria were: life-threatening infections, evidence of hematologic relapse, investigational drugs for GvHD prophylaxis, patients on steroid treatment (>0.5 mg/kg for 48 hours), grade IIIV GvHD. Progression to gut GvHD, but not liver GvHD, was confirmed by histology.

Patientsâ&#x20AC;&#x2122; characteristics Clinical characteristics of the two groups (observation/treatment) are outlined in Table 1. Patients were well balanced in terms of diagnosis (P=0.7): the most frequent diagnosis was acute myeloid leukemia (AML) (n=75), followed by acute lymphoblastic leukemia (ALL) (n=39), and myelodysplastic syndromes (MDS) (n=12). Median age for observation/treatment was 46 years (1-69) versus 38 years (0.4-68) (P=0.06). The proportion of patients over 50 years was 51% in the observation arm and 49% in the treatment arm (P=0.8). Donor type was: HLA identical siblings n=36 and n=34, unrelated cord blood (CB) n=7 and n=1, unrelated donor (UD) n=36 and n=44, and haploidentical family donors (HAPLO) 6 and 7 (P=0.1) in the observation and treatment arms, respectively. The proportion of 1 antigen mismatched unrelated donors was 7 and 9, respectively (P=0.9). Disease phase was classified as early in 43 observation arm patients and in 38 treatment arm patients (P=0.7). The conditioning regimen was myeloablative in most patients (n=64 and n=61, respectively; P=0.8).

Supportive care Antibacterial prophylaxis with quinolones was given during the neutropenic phase. All Centers used PJV prophylaxis with cotrihaematologica | 2017; 102(12)

Table 2. Patient outcome.

N Interval transplant-random (days) Interval random-GvHD II-IV (days) GvHD II-IV (n. pts) GvHD III-IV (n. pts) Chronic GvHD moderate/severe Steroid dose mg/kg <100 days Causes of death GvHD Infections Toxicity Relapse

Observation Treatment 85 86

20 (4-120) 20 (3-102) 3 (0-37) 9 (0-63) 44 29 11 9 10 15 9.5 (0-105) 24 (13.5-180)

0.5 0.03 0.01 0.6 0.3 0.01

9 7 1 20

13 9 1 16

0.5

Random: randomization; GvHD: graft-versus-host disease; N, n.: number; pts.: patients.

moxazole and monitored cytomegalovirus (CMV) reactivation by PCR or antigenemia twice weekly. Pre-emptive therapy was given in case of CMV reactivation. Epstein Barr Virus (EBV) was monitored by PCR weekly and treated pre-emptively if positive. Aspergillus antigenemia with galactomannan was also monitored weekly; diagnosis of invasive fungal disease was performed by standard criteria and treated accordingly. Specific infectious disease policies were performed according to standard procedures of each Center. 2127

Author et al.

P=0.005

Figure 2. Cumulative incidence of grade II-IV acute graft-versus-host disease (GvHD) in patients allocated to no treatment (observation) or treatment with prednisolone 1 mg/kg (treatment).

Statistical analysis Analysis of the primary end point was performed using the cumulative incidence (CI) of grade II-IV GvHD, calculated with mortality due to any cause as a competing risk. NRM was the competing risk for relapse-related death (RRD) and vice versa. Gray test was used to calculate difference between CI curves. Survival was calculated with Kaplan Meier curves, and the log rank test was used to test for difference between survival curves. Cox test was used for multivariate analysis. Ď&#x2021;2 tables, Fisher exact test, and 2-sample t-test were used as appropriate; these statistical analyses were carried out using NCSS10 for Windows. Infections and adverse events within day 100 from randomization in the two arms were assessed using Poisson or Negative Binomial (NB) regression model. Each infection type was considered as a single dependent variable and the decision on whether to use the Poisson or the NB model was based each time on a Likelihood-ratio test for overdispersion of the dependent variable considered. The treatment group indicator was considered as independent variable and the likelihood-ratio test was used to test the association with infections. The total follow up of each patient was considered as an exposure variable into the model. Stata (v.14) was used for the computation. Sample size calculation was made using data from a previous GITMO study:2 25% of patients with grade I GvHD treated with 6methylprednisolone (6MPred) 2 mg/kg, progressed to grade II or over GvHD. We hypothesized that patients left untreated would have twice the risk of progression to grade II-IV GvHD; 170 patients were needed to reject the null hypotesis with a power of 90%.

Results Primary end point and GvHD The cumulative incidence (CI) of acute GvHD grade II was 50% in the observation arm and 33% in the treatment arm (P=0.005) (Figure 2). This difference was more pronounced in sibling donor grafts (SIB) (61% vs. 32%; P=0.01), as compared to alternative donor grafts (ALT) (42% vs. 33%; P=0.1). For patients who progressed, the 2128

interval between randomization and progression was significantly shorter in the observation arm compared to the treatment arm (3 vs. 9 days; P=0.03) (Table 2). Figure 3 outlines the CI of grade III-IV GvHD in the observation versus the treatment arms (13% vs. 10%; P=0.6); it was seen in 7 vs. 4 sibling grafts, and in 4 vs. 5 alternative donor grafts, respectively (P=0.8). It was seen more frequently in patients randomized before day 20 from transplant (n=88, 17%) as compared to patients randomized later (n=83, 6%) (P=0.02) irrespective of randomization to the observation or treatment arms: 18% vs. 16% for early GvHD (<day 20), and 7% vs. 5%, for late GvHD (day 20). Moderate/severe chronic GvHD was comparable and was diagnosed in 10 patients in the observation arm versus 15 in the treatment arm (P=0.3).

Greatest severity of >grade 1 GvHD Skin GvHD, stage 2, 3 was diagnosed respectively in 22, 1 observation vs. 6, 2 treatment patients; liver GvHD stage 1,2, 3, was diagnosed in 4, 3, 1 observation vs. 7, 0, 0 treatment patients; gut GvHD stage 1, 2, 3, 4, was diagnosed respectively in 10, 3, 3, 1 observation vs. 3, 2, 3, 1 treatment patients.

Steroid dose and additional treatment The median cumulative dose of 6MPred received in the first 100 days was 9.5 mg/kg (range 0-105) and 24 mg/kg (range 13-180) in the observation and treatment arms, respectively (P=0.01) (Table 2). Of the 86 patients in the treatment arm, 57 (66%) were off steroids by day +30, whereas 29 were on steroids having progressed to grade II-IV GvHD. Of the 85 patients in the observation arm, 41 (48%) never received steroids. The use of a second immunosuppressive drug for GvHD in addition to corticosteroids was reported in 27 and 17 treatment and observation arms, respectively (P=0.08); administration of a third drug was reported in 12 and 7 patients (P=0.2) and a fourth drug in 4 and 2 patients, respectively (P=0.4). The second drug included mycophenolate mophetil (MMF) (in 6 and 7 patients, respectively), extracorporeal photopherehaematologica | 2017; 102(12)

Running Title

P=0.6

Figure 3. Cumulative incidence of grade III-IV acute graft-versus-host disease (GvHD) in patients allocated to no treatment (observation) or treatment with prednisolone 1 mg/kg (treatment).

sis (ECP) (12 and 4 patients, respectively), or infliximab, etanercept, rituximab, basiliximab, sirolimus, anti-CD26 antibody, in a few patients each. The third added drug included ECP (3 and 2 patients, respectively), MMF (5 and 1, respectively), etanercept or anti-CD26 antibody. The fourth added drug included MMF and ECP.

Infections and adverse events A summary of the adverse events in the two randomization arms in the first three months of treatment is shown in Table 3. The observation arm had less bacterial, fungal and CMV infections compared to the treatment arm; other adverse events and other severe adverse events were also fewer in the observation arm (Table 3), although these were mainly not statistically significant. Other adverse events were reported in 17 and 33 patients in the two arms, respectively (P=0.11), of which 3 and 11, respectively, were classified as severe (P=0.07). Adverse events included steroid-associated diabetes (0 vs. 9), acute renal failure (2 vs. 2), cystitis (5 vs. 5), hip necrosis (0 vs. 2), multi-organ failure (0 vs. 2), respiratory failure (0 vs. 3), and thrombosis (0 vs. 2) in the observation and treatment arms, respectively. Median blood counts were comparable on day +60 from randomization between the two arms. On day +60, chemistry results in the observation and treatment arms were also comparable.

Non-relapse mortality The 5-year CI of NRM was 20% (observation arm) versus 26% (treatment arm) (P=0.2) (Figure 4), and was comparable also after stratifying patients for age: <40 years (12% vs. 19%) and >40 years (28% vs. 34%). In univariate analysis, there was a very strong influence of the interval between transplant and randomization on NRM: median interval 20 days, CI of NRM 31% versus 18% (P=0.0006) for patients randomized before or after day +20 from transplant, respectively. For early randomization (<20 days from transplant) NRM was 24% vs. 46% (P=0.02) in observation versus treatment patients, due to an excess of haematologica | 2017; 102(12)

Table 3. Infectious episodes and adverse events in the two arms <100 days from randomization.

N FU days at 100 days FU days/patient Bacterial infections Severe bacterial inf. Lethal bacterial inf. Fungal infections Severe fungal inf. Lethal fungal inf. CMV Severe CMV inf. Lethal CMV infections Other viral infections PTLD Other AE Other severe AE Lethal AE

Observation 85

Treatment 86

8040 94.5 12 3 2 8 0 0 63 3 0 24 2 17 3 0

7934 92.2 25 8 5 8 3 2 84 3 3 15 1 33 11 4

0.045 0.16 0.29 0.94 0.047 0.11 0.48 0.94 0.046 0.32 0.53 0.11 0.077 0.041

One patient can have more than one infectious episode or adverse event. FU: follow up; AE: adverse events; inf: infections; CMV: cytomegalovirus; PTLD: post transplant lymphoproliferative disease. *P-value: Poisson or Negative Binomial (NB) regression model (see Statistical Analysis section).

infections in the treatment arm (2 vs. 8); for late randomization (>20 days) NRM was comparable in the two arms (22% and 14%; P=0.3).

Relapse-related death and survival Relapse-related death was 25% in patients in the observation arm versus 21% in patients in the treatment arm (Figure 5); patients with early disease had a significantly lower probability of RRD in univariate analysis (RR 0.3, P=0.006). RRD was unaffected by the interval between 2129

Author et al. Table 4. Multivariate analysis.

Base

Compared

Age (y)

<20

Sex Donor Phase Dx SC source

M ALT ADV AL BM

Rand Int.Tx-Rand

OBS <20

>20 >40 F SIB EARLY Other PB CB TREAT >20 dd

3.6 5.6 0.9 1.2 0.6 0.9 0.7 0.4 0.5 0.7

GvHD II-IV 95% CI 1.03-12.8 1.71-18.6

0.35-0.99

0.04 0.004 0.8 0.4 0.1 0.8 0.3 0.4 0.04 0.2

2.4 2.9 0.9 0.8 0.5 1.1 0.9 0.4 1.4 0.4

Overall survival 95%CI 1.09-5.58 1.28-6.13

0.35-0.9

0.26-0.71

P 0.02 0.009 0.9 0.4 0.02 0.6 0.8 0.3 0.1 0.001

Base: baseline value; Compared: compared value; RR: relative risk; CI: Confidence Interval; y: years; M: male; F: female; SC: stem cell; ADV: advanced; BM: bone marrow; PB: peripheral blood; CB: cord blood; P: P-value; GvHD graft-versus-host disease; Rand: randomization group; OBS: observation; TREAT: treatment; Dx: diagnosis; AL: acute leukemia; Int. TxRand: interval in days (dd) between stem cell transplantation and randomization; ALT: alternative donor; SIB: identical sibling.

P=0.2

Figure 4. Comparable cumulative incidence of nonrelapse mortality (NRM) in the two randomization groups.

transplant and randomization. Actuarial 5-year survival was 51% versus 41% in the observation and treatment arms, respectively (P=0.3) (Figure 6). Predictors of survival in univariate analysis were younger age, early disease phase, and randomization beyond day +20 from transplant. Causes of death in the two study groups were: GvHD in 9 versus 13 patients, infection in 7 versus 9 patients, toxicity in 1 patient in each group, and leukemia relapse in 20 versus 16 patients (P=0.9) (Table 2). There was no significant difference in NRM between different Centers (P=0.5).

Skin biopsies A skin biopsy to prove or disprove skin GvHD was not mandatory for eligibility in this trial. It was performed before randomization and reviewed centrally by one of the Authors (DM) in 38 patients. Of these, 36 (95%) were compatible with aGvHD (proven, probable, and possible in 9, 15 and 12 patients, respectively); these were equally distributed among the treatment and observation arms (P=0.7). 2130

Multivariate analysis Progression to grade II-IV GvHD was predicted in a Cox analysis by age over 20 years (P=0.003) or 40 years (P=0.005), and randomization to the observation arm (P=0.02) (Table 4). A short interval between transplant and randomization (< 20 days), was the only variable predicting progression to grade III-IV GvHD (RR 0.4, 95%CI: 0.12-0.98; P=0.04) and was also associated with a higher risk of death (P=0.006). Survival was also predicted by patients' age and disease phase. NRM was predicted only by age over 20 years (RR 2.8, 95%CI: 0.88-9.18; P=0.07) and age over 40 years (RR 3.0, 95%CI: 0.99-9.67; P=0.051), and by early onset of GvHD before day +20 from transplant (RR 2.38, 95%CI: 1.06-4.0; P=0.03).

Discussion Treatment of aGvHD remains a difficult issue, despite several decades of studies and the many immunosuppressive/immunomodulating agents tested.8 There are difficulhaematologica | 2017; 102(12)

Running Title

P=0.6

Figure 5. Comparable cumulative incidence of relapse-related death (RRD) in the two randomization groups.

ties not only in the treatment, but problems start with staging of involved organs and overall grading of the disease, with several possible grading options and a degree of variability according to the assessor.7-13 Despite differences in grading, and the difficulty in assessing response rates, it is recognized that mortality increases with increasing GvHD severity, and this is true both in the short and in the long term.14 In a large group of patients (n=4174), NRM at three years was 21% for grade 0-I aGvHD, 32% for grade II, 60% for grade III, and 89% for grade IV; the corresponding 3-year OS was 79%, 64%, 37% and 10%, respectively.14 This study exemplifies on one hand, the major impact of aGvHD grading on the outcome of allogeneic transplants, and on the other, the lack of effective treatment when the disease is beyond grade II. In keeping with the latter observation, a recently developed risk score for aGvHD identifies patients at high risk of mortality according to the number of involved organs and the severity of GvHD at onset.13 Mortality at six months is 22% for standard-risk and 44% for high-risk GvHD.13 A set of GvHD biomarkers have recently been described; these identify at the onset of the disease severe cases with a high risk of mortality eligible for early intervention.15 It would thus seem reasonable to try and prevent progression of aGvHD, and this may be achieved if aGvHD is treated at a very early stage (earliest being grade I, or a skin rash involving <50% of the body surface). We, therefore, asked whether steroid treatment of grade I GvHD would be beneficial, and we selected evolution to grade II or more as the primary end point of the study. Patients randomized in the observation arm would become eligible for treatment when diagnosed as grade II GvHD, also if this occurred 24 hours after randomization. This facilitated the informed consent procedure with the patients since there would be no delay in treatment once the disease had progressed to grade II. As expected, patients randomized to receive treatment at diagnosis of grade I GvHD had a significantly lower probability to progress to grade II or more GvHD compared to untreated patients (33% vs. 50%). The fact that patients in the observation arm, grafted from identical siblings, had a higher proportion of grades II-IV GvHD (61%) compared to patients in haematologica | 2017; 102(12)

the observation arm receiving alternative donor grafts (44%) can be explained by the fact that, in the latter, GvHD prophylaxis included either ATG or PT-CY, in addition to CyA and MTX (UD grafts) or CyA and mycophenolate (HAPLO grafts). The unexpected finding was that the CI of patients progressing to severe (grade III-IV) GvHD was comparable in the two groups (13% vs. 10%). Therefore, the primary end point of the study was reached, but this was due to skin GvHD progressing from stage II to stage III in the observation arm (22 observation vs. 6 treatment patients) and stage 1 gut GvHD (10 observation vs. 3 treatment patients). On the other hand, patients with stage II-IV gut GvHD were comparable in the two randomization groups (7 and 6, respectively), and liver GvHD was seen in a few patients only. When looking at adverse events, we found that patients in the treatment arm had more infections and more adverse events than observation patients, in particular, bacterial infections, severe fungal infections, and CMV reactivation. As a consequence of similar severe GvHD and more infections, NRM was 20% in the observation versus 26% in the treatment arm, and survival at five years was 51% versus 41%, respectively. In a multivariate Cox analysis, there was a trend for inferior survival (P=0.09) in the treatment arm, despite a median younger age (38 vs. 46 years). Other studies have tested early treatment of GvHD.2-4 Etanercept and topical steroids have been reported by Gatza et al.16 in grade I GvHD. Of the 34 patients enrolled in that study, 3% progressed to grade III-IV, significantly lower than another group of patients receiving topical steroids alone, 18% of whom progressed to grade III-IV GvHD.16 Although Gatza et al. suggested that etanercept was able to modify the natural course of the disease, 2-year NRM was 19%,16 comparable to the 20% NRM of our observation arm, and the 26% of our treatment arm. Another non-steroid approach was tested propectively, randomizing patients to receive or not 2.5 mg/kg ATG, on day +7 after an alternative donor transplant.17 Grade III-IV GvHD was significantly reduced in the ATG group (5%) compared to the untreated group (15%), though NRM was only marginally reduced from 35% to 29% (P=ns).17 Finally, high-dose cyclophosphamide post transplant is 2131

Author et al.

Figure 6. Comparable 5-year overall survival (OS) in the two randomization groups.

P=0.3

being widely and successfully used to prevent severe GvHD,18-20 but again this is given very early (day +3) and possibly interfers with the activation phase of T cells rather than with the effector phase. We found a strong association of early GvHD with GvHD severity and survival. Patients developing grade I GvHD within day +20 from transplant had a higher probability (RR 2.7) of developing grade III-IV GvHD, as compared to patients randomized later (17% vs. 6%; P=0.02) and a higher risk of NRM (31% vs. 18%; P=0.0006). Randomization to steroids was not beneficial in these early grade I GvHD patients, with progression to grades III-IV in 18% observation versus 16% treatment patients. In addition, more cases of infectious mortality were found in patients randomized day 20 or later to the treatment arm. Univariate and multivariate analysis predicted survival by the time of randomization; 4-year survival of patients randomized before day +20 from transplant was 33% compared to 60% for patients randomized later (P=0.001), regardless of randomization to the observation or to the treatment arms. In our data base of 2445 allogeneic transplants, the proportion of grade III-IV GvHD in patients developing GvHD within day 20, between days 21-40, or beyond day 40 is 11%, 9% and 3%, respectively (P=0.0002), and NRM is 35%, 28% and 25%, respectively (P=0.0006) (A Bacigalupo et al., 2017, unpublished data), confirming other reports on the association of early onset as a risk factor for grade III-IV GvHD.21

References 1. BolaĂąos-Meade J, Logan BR, Alousi AM, et al. Phase 3 clinical trial of steroids/mycophenolate mofetil vs steroids/placebo as therapy for acute GVHD: BMT CTN 0802. Blood. 2014; 124(22):3221-3227 2. Van Lint MT, Milone G, Leotta S, et al. Treatment of acute graft-versus-host dis-

2132

In conclusion, steroid treatment of grade 1 GvHD prevents progression to grade II GvHD, but not to grade IIIIV GvHD, and there is no beneficial effect on NRM and survival. In addition, patients receiving steroids are at a higher risk of developing infections and have more adverse events, especially if GvHD develops within day +20 from transplant. A small proportion of patients develop life-threatening GvHD, irrespective of early steroid treatment, suggesting that the severity of GvHD is determined at onset. Early identification of high-risk patients with recently described biomarkers,22 and pre-emptive treatment with non-steroidal agents should be investigated with the aim of changing the natural course of the disease. Study centers The following Centers participated in the trial: Ospedale San Martino, Genova (A Bacigalupo); Ospedale Ferrarotto, Catania (G Milone); Ospedale San Camillo, Roma (A Locasciulli); Ospedale Civile, Pescara (A Santarone); Ospedale Regina Margherita, Torino (F Fagioli); Unviersitaâ&#x20AC;&#x2122; Cattolica, Roma (S Sica, P Chiusolo); Ospedale S. Croce, Cuneo (N Mordini); Ospedale Civile, Alessandria (R Sorasio). Funding This trial was supported by the Gruppo Italiano Trapianto di Midollo Osseo (GITMO), FARITMO, Genova, and AIRC, Milano.

ease with prednisolone: significant survival advantage for day +5 responders and no advantage for nonresponders receiving anti-thymocyte globulin. Blood. 2006; 107(10):4177-4181. 3. van Lint MT, Uderzo C, Locasciulli A, et al. Early treatment of acute graft-versus-host disease with high- or low-dose 6-methylprednisolone: a multicenter randomized trial from the Italian Group for Bone Marrow Transplantation. Blood. 1998; 92(7):2288- 2293.

4. Mielcarek M, Furlong T, Storer BE, et al. Effectiveness and safety of lower dose prednisone for initial treatment of acute graft-versus-host disease: a randomized controlled trial . Haematologica. 2015; 100(6);842-848. 5. Mikulska M, Raiola AM, Bruno B, et al. Risk factors for invasive aspergillosis and related mortality in recipients of allogeneic SCT from alternative donors: an analysis of 306 patients. Bone Marrow Transplant. 2009;44(6):361-370.

haematologica | 2017; 102(12)

Running Title

6. Remberger M, Storer B, Ringdén O, Anasetti C. Association between pretransplant Thymoglobulin and reduced nonrelapse mortality rate after marrow transplantation from unrelated donors. Bone Marrow Transplant. 2002;29(5):391-397. 7. Glucksberg H, Storb R, Fefer A, et al. Clinical manifestations of graft-versus-host disease in human recipients of marrow from HL-A-matched sibling donors. Transplantation. 1974;18(4):295-304. 8. Deeg HJ. How I treat refractory acute GvHD. Blood. 2007;109(9):4119-4126. 9. Przepiorka D, Weisdorf D, Martin P, et al. 1994 Consensus Conference on Acute GVHD Grading. Bone Marrow Transplant. 1995;15(6):825-828. 10. Rowlings PA, Przepiorka D, Klein JP, et al. IBMTR Severity Index for grading acute graft-versus-host disease: retrospective comparison with Glucksberg grade. Br J Haematol. 1997;97(4):855-864. 11. Martino R, Romero P, Subirá M, et al. Comparison of the classic Glucksberg criteria and the IBMTR Severity Index for grading acute graft-versus-host disease following HLA-identical sibling stem cell transplantation. International Bone Marrow Transplant Registry. Bone Marrow Transplant. 1999;24(3):283-287. 12. MacMillan ML, Weisdorf DJ, Wagner JE, et al. Response of 443 patients to steroids as

haematologica | 2017; 102(12)

13.

14.

15.

16.

17.

18.

primary therapy for acute graft-versus-host disease: comparison of grading systems. Biol Blood Marrow Transplant. 2002; 8(7):387-394. MacMillan ML, Robin M, Harris AC, et al. A Refined Risk Score for Acute Graft-versus-Host Disease that Predicts Response to Initial Therapy, Survival, and TransplantRelated Mortality . Biol Blood and Marrow Transpl. 2015;21(4):761-767 Gratwohl A, Brand R, Apperley J, et al. Graft vs host diseaseand outcome in HLA identical sibling transplantation, for chronic myeloid leukemia. Blood. 2002; 100(12):3877- 3886 Levine JE, Braun TM, Harris AC, et al. A prognostic score for acute graft-versus-host disease based on biomarkers: a multicentre study. Lancet Haematol. 2015;2(1):e21-29 Gatza E, Braun T, Levine JE, et al. Etanercept plus topical corticosteroids as initial therapy for grade one acute graft-versus-host disease after allogeneic hematopoietic cell transplantation.Biol Blood Marrow Transplant. 2014; 20(9):1426-1434. Bacigalupo A, Lamparelli T, Milone G, et al Pre-emptive treatment of acute GVHD: a randomized multicenter trial of rabbit antithymocyte globulin, given on day+7 after alternative donor transplants. Bone Marrow Transplant. 2010;45(2):385-391. Luznik L, O'Donnell PV, Symons HJ, et al.

19.

20.

21.

22.

23.

HLA-haploidentical bone marrow ransplantation for hematologic malignancies using nonmyeloablative conditioning and high-dose, posttransplantation cyclophosphamide. Biol Blood Marrow Transplant. 2008;14(6):641-650. Solomon SR,Sizemore CA, Sanacore M, et al. Total Body Irradiatione Based Myeloablative Haploidentical Stem Cell Transplantation Is a Safe and Effective Alternative to Unrelated Donor Transplantation in Patients Without Matched Sibling Donors. Biol Blood Marrow Transpl. 2015;21(7):1299-1307 Brammer JE, Khouri I, Gaballa S, et al. Outcomes of Haploidentical Stem Cell Transplantation for Lymphoma with Melphalan-Based Conditioning. Biol Blood Marrow Transpl. 2016;22(3):493-498. Kanakry CG, Ganguly S, Zahurak M, et al. Aldehyde dehydrogenase expression drives human regulatory T cell resistance to posttransplantation cyclophosphamide. Sci Transl Med. 2013;5(211):211ra157. Saliba RM, De Lima M,Giralt S, et al. Hyperacute GVHD: risk factors, outcomes, and clinical implications. Blood. 2007; 109(7):2751-2758. Hartwell MJ, Özbek U, Holler E, et al. An early-biomarker algorithm predicts lethal graft-versus-host disease and survival. JCI Insight. 2017;2(3):e89798.

2133

ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Cell Therapy & Immunotherapy

Ferrata Storti Foundation

Mixed phenotype acute leukemia: outcomes with allogeneic stem cell transplantation. A retrospective study from the Acute Leukemia Working Party of the EBMT Reinhold Munker,1 Myriam Labopin,2 Jordi Esteve,3 Christoph Schmid,4 Mohamad Mohty2 and Arnon Nagler5,6

Haematologica 2017 Volume 102(12):2134-2140

Section of Hematology & Medical Oncology, Tulane University, New Orleans, LA, USA; Hôpital Saint Antoine, Paris, France; 3Hematology Department, IDIBAPS, Hospital Clinic, Barcelona, Spain; 4II Medizinische Klinik, Klinikum Augsburg, Germany; 5Hematology Division, Chaim Sheba Medical Center, Tel HaShomer, Israel and 6ALWP Office Hôpital Saint Antoine and Pierre and Marie Curie University, Paris, France 1 2

ABSTRACT

Correspondence: rmunker@tulane.edu

Received: June 11, 2017. Accepted: September 22, 2017. Pre-published: September 29, 2017.

ixed phenotype acute leukemias are infrequent and considered high risk. The optimal treatment approach and the role of allogeneic hematopoietic stem cell transplantation are not entirely clear. In this study, we investigated 519 patients with mixed phenotype acute leukemia in first complete remission who underwent allogeneic hematopoietic stem cell transplantation between 2000 and 2014, and who were reported to the Acute Leukemia Working Party of the European Society for Blood and Marrow Transplantation (EBMT). Median age was 38.1 years (range 18-75). Cytogenetics classified 49.3% as poor risk. At three years, relapse incidence was 31.4% (26.9-35.9), non-relapse mortality was 22.1% (18.4-26.1), the leukemia-free survival was 46.5% (41.7-51.4), and the overall survival was 56.3% (51.5-61.2). At six months, 32.5% had developed acute graft-versus-host disease, while at three years, 37.5% had developed chronic graft-versus-host disease (32.6-42.3). In a multivariate analysis, age and year of transplant had a strong impact on outcome. Myeloablative conditioning using total body irradiation correlated with a better leukemia-free survival. Our study suggests that mixed phenotype acute leukemia is potentially sensitive to graft-versus-leukemia and thus can benefit from allogeneic hematopoietic stem cell transplantation with a potential for cure.

doi:10.3324/haematol.2017.174441

Introduction Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/12/2134 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

2134

Major progress was made in the last decades in the diagnosis and treatment of acute and chronic leukemias.1 Mixed phenotype acute leukemias (MPAL), also designated as acute biphenotypic leukemias or hybrid acute leukemias are rare (1-4% of all acute leukemias) and were described many years ago.2-6 MPAL (bearing markers of myeloid and lymphoid lineage) are considered enigmatic because of their cell of origin which may be a multipotent stem cell. It is now known that MPAL, like other acute leukemias, are heterogeneous.7 In 1995, the European Group for the Immunologic Characterization of Leukemias (EGIL) established criteria for acute biphenotypic leukemias in which points are assigned to specific markers of B-lymphoid, T-lymphoid and myeloid origin.8 In 2008, the World Health Organization (WHO) revised the criteria for lineage assignment and introduced the term “mixed phenotype acute leukemia”,9 but excluded those which could be classified under other cytogenetic or clinical categories. The 2016 WHO update of the classification of acute leukemias maintained the definition of MPAL, but introduced clarifications and added the subcategory of Ph1+ MPAL.10 The optimal treatment approach to MPAL is not entirely clear. In previous case series, ranging in patient numbers between 13 and 117, allogeneic hematopoietic stem cell transplantation (alloSCT) was performed in 7-61%.6,11 However, many cases were not haematologica | 2017; 102(12)

MPAL with alloHSCT

classified according to WHO and most studies did not report transplant outcomes. In reviews by 2 experts, chemotherapy according to acute lymphoblastic leukemia (ALL), followed by alloHSCT is the preferred approach,7,12 but definitive data are still lacking. Generally, MPAL is considered to be high risk with a poor prognosis, although younger patients may have a better outcome. In the pretransplant era, or in countries with limited resources, a longer-term survival of 15-35% was described.6,11 In a smaller series from the Center for International Blood and Marrow Transplant Research (CIMBTR) a 3-year leukemia-free survival (LFS) of 56±10% and an overall survival (OS) of 67±10% was described.13 The aim of the current study is to establish the outcomes of alloHSCT in a large (n=519 patients), recent (2000-2014), adult cohort from the Acute Leukemia Working Party (ALWP) of the European Society for Blood and Marrow Transplantation (EBMT).

Methods Data source and methods This retrospective multicenter study was approved by the ALWP of the EBMT and included adults (≥18 years) diagnosed with de novo MPAL and receiving HSCT from a matched related or unrelated donor. Patients with MPAL transplanted between 2000 and 2014 were included in the study. Demographics, MPAL disease characteristics, transplantation and post-transplantation data were extracted from the EBMT database (Med-A forms). The list of centers contributing patients to this study is available in the Online Supplementary Appendix. The EBMT is a voluntary working group of more than 600 transplant centers that are required to report all consecutive HSCT procedures and their follow up once a year. Audits are routinely performed to determine the accuracy of the data. Patients have been required to provide informed consent authorizing the use of their personal information for research purposes since 1990. To verify which classification was used, the 11 centers transplanting more than 7 patients with MPAL were contacted. Ten centers responded: 8 used WHO, 1 predominantly WHO, and 1 both EGIL and WHO criteria. HLA matching was performed by standard criteria.

Definitions Cytogenetic abnormalities were classified as favorable, intermediate or high risk, as previously described.14 The conditioning regimen was defined as reduced intensity conditioning (RIC) when fludarabine was associated with low-dose total body irradiation (TBI) (6 Gy) or a dose of oral busulfan <8 mg/kg or a dose of intravenous (IV) busulfan <6.4 mg/kg or other immunosuppressive or chemotherapeutic drugs, such as melphalan or cyclophosphamide. Myeloablative conditioning (MAC) was defined as a preparative regimen that contained TBI or busulfan at higher doses.15 Neutrophil engraftment was defined as an absolute neutrophil count (ANC) over 0.5x109/L for three consecutive days. Platelet engraftment was defined as an absolute platelet count over 20x109/L for three consecutive days.

Statistical analysis The clinical outcomes studied were overall survival (OS), leukemia-free survival (LFS), relapse incidence (RI), non-relapse mortality (NRM), chronic graft-versus-host disease (cGvHD). OS was defined as the time from day 0 of allo-SCT to death or last follow up for survivors. LFS was defined as time from day 0 of alloSCT to time without evidence of relapse or disease progression haematologica | 2017; 102(12)

Figure 1. Overall survival (OS) of patients with mixed phenotype acute leukemias (MPAL) who underwent allogeneic hematopoietic stem cell transplantation (alloSCT) according to age groups: 18-35 years (yo), 36-55 yo, and ≥ 56 yo.

censored at the date of death or last follow up. Relapse was defined as any event related to re-occurrence of the disease. NRM was defined as death from any cause without previous relapse or progression. Probabilities of OS and LFS were calculated using the log rank test and Kaplan-Meier graphical representation. Further end points were engraftment, incidence and severity of acute and chronic GvHD (grading of acute GvHD was performed as previously published16). Cumulative incidence functions (CIF)17 were used to estimate RI and NRM in a competing risks setting. In order to study cGvHD, we considered death and relapse as competing events. Survival probabilities are presented as percentages and 95% confidence intervals (95%CI). Univariate analyses were performed using the log rank test for OS and LFS and the Gray test for CIF. Multivariate analyses were performed using the Cox proportional hazard model. To allow for potential confounding factors between treatments that could influence outcome, propensity score matching was also performed, using the exact matching.17 Matching on the propensity score was then used to reduce or eliminate confounding effects and estimate treatment effects by matching 3 AML and 4 ALL patients with each MPAL patient. The following factors were included in the propensity score model: patient age [18-35 years (y), 36-55 y and above 56 y], year of transplant, time from diagnosis to transplant (per quantile), conditioning (RIC, MAC TBI, MAC chemotherapy), source of SC [bone marrow (BM)/peripheral blood (PB)], patient sex, female donor to male recipient, type of donor [unrelated donor (UD)/matched sibling donor (MSD)]. The purpose of the propensity score-matching strategy was to reduce confounding effects of these variables, and strengthen causal inferences.18 Details about the patients used for matching are given in the Online Supplementary Appendix. All tests were two-sided and P<0.05 was considered statistically significant. Analyses were performed using the R statistical software v.3.2.3 (available online at: http://www.R-project.org), and propensity score analysis was performed using the ‘MatchIt’ program (last accessed May 18th, 2015; http://cran.project.org/web/packages/MatchIt/MatchIt.pdf). Patients with missing values were excluded from propensity score analyses. 2135

R. Munker et al.

Results Patients’ and disease characteristics Patients', disease and transplant characteristics are summarized in Table 1. A total of 519 patients (all in first complete remission) underwent allo-SCT in the study period. These patients were contributed by 189 centers, 30 of which transplanted 5 or more patients with MPAL during the study period. Sixty-three percent of the patients were male. Median age at transplant was 38.1 years. Complete cytogenetics were available in 203 patients. Among these, 50.7% had intermediate risk and 49.3% poor risk cytogenetics, respectively. Among the poor risk patients, 50%

were defined by positivity for bcr/abl or Ph+. Four hundred patients received MAC (260 including TBI, 140 without TBI), 119 received RIC. The majority of the patients were transplanted with stem cells harvested from PB (73.0%), while 26.4% received BM. Hematopoietic recovery from transplant was observed in 492 patients (97.0%). In 6 patients (1.2%), the engraftment was transient. Nine patients (1.8%) never engrafted. Donor lymphocyte infusions (DLI) post transplant were documented in 55 patients (25 prophylactic or pre-emptive, 29 for clinical relapse). Among the 29 patients who received DLI for relapse, 10 also underwent a second transplant. The 1- and 2-year OS of these patients was 44 and 35%, respectively.

Prognostic factors for outcome Table 1. Patients’, disease and transplant characteristics.

Characteristic

N (%)

Total patient number 519 (100%) Median age, years (range) 38.1 (18-75) 18-35 232 (44.9) 36-55 212 (40.9) ≥ 56 74 (14.3) Sex male / female 329 (63.4) / 190 (36.6) Time of transplant 2000- 2004 69 (13.3) 2005- 2010 238 (45.9) 2011- 2014 212 (40.9) Cytogenetics Available 203 (39.1) Good risk 0 (0) Intermediate risk 103 (50.7) Poor risk (including 50 bcr-abl or Ph+ and 11 11q23+ patients) 100 (49.3) Unavailable 316 (60.9) Conditioning regimen MA with TBI 260 (50.1) MA, no TBI 140 (27.0) RIC 119 (22.9) Median time from CR-transplant (days) 99 (± 36) Graft type Bone marrow 137 (26.4) Peripheral blood stem cells 379 (73.0) Both 3 (0.6) In vivo T-cell depletion ATG 153 (33.0) Alemtuzumab 56 (12.1) No in vivo T-cell depletion 254 (54.9) CMV status Patient positive 285 (63.2) Patient negative 166 (36.8) Donor positive 240 (53.7) Donor negative 207 (46.3) Follow up of survivors, median (range), months 32.1 (0.9-181.2) MA: myeloablative conditioning; TBI: total body irradiation; RIC: reduced intensity conditioning; ATG: anti-T-cell globulin. Definition of cytogenetics: Good risk: t(8;21) or inv16. Poor risk: complex or del5 or del7 or mono5 or mono7 or 11q23 or 3q26 or inv3 or t(6;9) or t(11;19) or Ph+ or t(4;11). Intermediate risk: all other.

2136

A univariate analysis of outcomes at three years is shown in Table 2. Age at transplant had a strong impact on LFS, RI and NRM. More recent transplants had less NRM and a trend for better OS. No impact was observed according to donor type, patient sex, or cytomegalovirus (CMV) status of donor or recipient. Female donors were associated with more cGvHD. The combination of a female donor for a male recipient was particularly strongly associated with cGvHD. MAC, especially with TBI, correlated with better LFS and less RI. The source of stem cells did not impact on LFS or OS, though PB had a trend for higher RI and more cGvHD, but less NRM. In a multivariate analysis (Table 3), the best outcome (NRM, LFS, and OS) was seen in younger patients and in patients transplanted more recently. Matched unrelated donors compared with matched sibling donors and female donors for male recipients had a significant increase in cGvHD, but similar RI, LFS and OS. Adverse cytogenetics (both Ph+ and other poor cytogenetics) were associated with decreased OS. MAC with TBI compared with myeloablation with chemotherapy alone was associated with better LFS and a lower RI. In vivo T-cell depletion (TCD) was associated with less cGvHD, but did not impact on survival. Similarly, the use of PB versus BM correlated with more

Figure 2. Leukemia-free survival (LFS) of patients with mixed phenotype acute leukemias (MPAL) who underwent allogeneic hematopoietic stem cell transplantation (allo-SCT) according to age groups: 18-35 years (yo), 36-55 yo, and ≥56 yo.

haematologica | 2017; 102(12)

MPAL with alloHSCT Table 2. Univariate analysis of outcomes at three years after allogeneic hematopoietic stem cell transplantation for mixed phenotype acute leukemia.

Characteristic Age (years) 18-35 36-55 ≥56 Year of transplant 2004-2004 2005-2010 2011-2014 Donor type Matched sib donor Unrelated donor Patient sex Male Female Donor sex Male Female Matching by sex No F-> M F-> M CMV status patient Negative Positive CMV status donor Negative Positive Type of conditioning MAC chemotherapy MAC TBI RIC Cytogenetics Intermediate Poor Not available Source of stem cells Bone marrow Peripheral blood

LFS (% ± SD) P

OS (% ± SD) P

RI (% ± SD) P

NRM (% ± SD) P

cGvHD (% ± SD) P

56.6 (49-63) 40.5 (33-48) 31.3 (20-43) P<0.0001

62.9 (56-70) 53.3 (46-61) 43.7 (31-56) P=0.0017

27.4 (21-34) 32.1 (25-39) 41.8 (29-54) P=0.031

16.0 (11-21) 27.4 (21- 34) 26.9 (17-38) P=0.026

39.5 (32-47) 38.1 (31-46) 30.2 (19-42) P=0.567

41.8 (30-54) 46.8 (40-54) 46.0 (37-55) P=0.361

44.7 (33-57) 55.6 (49-62) 61.8 (53-71) P=0.057

24.9 (15-36) 32.9 (27-39) 33.7 (25-43) P=0.464

33.2 (22-45) 20.2 (15-26) 20.3 (14-27) P=0.039

31.2 (20-43) 37.7 (31-45) 40.3 (32-49) P=0.639

45.6 (39-52) 47.7 (41-55) P=0.690

54.5 (48-61) 58.0 (50-66) P=0.477

32.6 (27-39) 29.7 (23-37) P=0.569

21.8 (17-27) 22.4 (17-28) P=0.918

34.6 (28-41) 40.7 (33-48) P=0.310

45.3 (39-51) 48.7 (41-57) P=0.922

55.3 (49-61) 58.0 (50-66) P=0.897

32.3 (27-38) 29.7 (23-37) P=0.805

22.4 (18-28) 21.6 (16-28) P=0.930

41.1 (35-47) 31.2 (24-39) P=0.116

45.9 (40-52) 47.1 (39-55) P=0.978

59.1 (53-65) 51.4 (43-59) P=0.131

34.8 (29-41) 25.9 (19-33) P=0.079

19.3 (15-24) 27.0 (20-34) P=0.065

33.3 (27-39) 43.0 (35-51) P=0.028

45.4 (40-51) 49.8 (39-60) P=0.492

57.5 (52-63) 51.3 (41-62) P= 0.624

33.9 (28-39) 22.6 (15-32) P=0.062

20.7 (17-25) 27.6 (19-37) P=0.229

32.5 (27-38) 53.6 (42-64) P=0.000

50.2 (42-59) 45.4 (39-52) P=0.789

59.2 (51-68) 55.1 (49-62) P=0.420

29.7 (22-38) 29.6 (24-36) P=0.293

20.0 (14-27) 24.9 (20-31) P=0.166

35.1 (27-43) 40.0 (33-47) P=0.235

48.4 (41-56) 46.9 (40-54) P=0.379

58.3 (51-66) 56.8 (50-64) P=0.306

30.1 (23-37) 29.7 (23-36) P=0.729

21.5 (16-28) 23.4 (18-30) P=0.499

40.4 (33-48) 37.2 (30-44) P=0.502

43.1 (34-53) 56.3 (17-29) 29.1 (38-58) P<0.001

55.2 (46-65) 60.0 (53-67) 49.4 (39-60) P=0.151

33.2 (24-42) 22.6 (17-29) 48.3 (38-58) P<0.001

23.7 (17-2) 21.1 (16-27) 22.6 (15-31) P=0.764

30.9 (22-41) 42.3 (35-49) 33.5 (24-43) P=0.146

54.8 (44-65) 46.1 (36-56) 43.3 (37-50) P=0.132

65.6 (56-76) 55.0 (45-65) 53.2 (47-60) P=0.159

28.4 (19-38) 34.1 (25-44) 31.5 (26-38) P=0.574

16.8 (10-25) 19.8 (12-28) 25.2 (20-31) P=0.344

37.2 (27-48) 34.4 (25-44) 38.8 (32-45) P=0.616

45.7 (37-55) 46.9 (41-53) P=0.993

53.6 (45-63) 57.4 (52-63) P=0.314

25.9 (19-34) 33.5 (28-39) P=0.066

28.3 (21-37) 19.6 (16-24) P=0.055

30.9 (23-40) 40.3 (34-46) P=0.055

LFS: leukemia-free survival; SD: Standard Deviation; OS: overall survival; RI: relapse incidence; NRM: non-relapse mortality; cGvHD: chronic graft-versus-host disease; P: P-value; sib: sibling; F: female; M: male; CMV: cytomegalovirus; MAC: myeloablative conditioning; TBI: total body irradiation; RIC: reduced intensity conditioning.

haematologica | 2017; 102(12)

2137

R. Munker et al. Table 3. Multivariate analysis of major outcomes after transplant.

HR Age per decade 0.96 Year of transplant 0.98 UD vs. MSD 0.79 Female D -> male R 0.80 Intermediate cytogenetics (ref) 1 Poor cytogenetics 1.13 Cytogenetics NA or failed 1.39 MAC chemo (ref) 1 MAC TBI vs. MAC chemo 0.50 RIC vs. MAC chemo 1.34 In vivo TCD 1.40 PB vs. BM 0.99

RI CI

0.83-1.11 0.92-1.04 0.52-1.21 0.50-1.28

0.59 0.42 0.28 0.35

1.43 0.92 1.08 1.19 1 1.76 1.40 1 0.78 0.63 0.82 0.83

0.71-1.79 0.61 0.83-2.30 0.21 0.31-0.79 0.82-2.19 0.88-2.22 0.63-1.55

0.003 0.24 0.16 0.97

NRM CI

1.19-1.71 <0.001 0.86 -0.98 0.01 0.64-1.84 0.77 0.71-1.97 0.51 0.98-3.15 0.06 0.73-2.70 0.31 0.46-1.31 0.33-1.18 0.46-1.46 0.51-1.35

0.35 0.15 0.51 0.44

1.13 0.95 0.89 0.94 1 1.39 1.40 1 0.61 1.00 1.12 0.92

LFS CI

1.01-1.27 0.91-0.99 0.64-1.24 0.67-1.33

0.03 0.02 0.49 0.72

1.19 0.93 0.93 1.11 1 1.52 1.40 1 0.73 0.86 0.95 0.89

0.97-1.99 0.07 0.94-2.09 0.10 0.43-0.86 0.68-1.47 0.78-1.61 0.66-1.27

0.005 1.00 0.54 0.60

OS CI

1.05-1.35 0.89-0.98 0.64-1.35 0.77-1.59

0.006 0.003 0.69 0.59

1.02-2.26 0.04 0.91-2.22 0.13 0.50-1.06 0.56-1.34 0.64-1.42 0.63-1.26

0.10 0.51 0.80 0.51

Chronic GvHD HR CI P 1.01 0.96 1.82 2.23 1 1.31 1.02 1 1.35 1.82 0.52 1.66

0.87-1.18 0.91-1.01 1.19-2.79 1.51-3.30

0.88 0.15 0.006 <0.001

0.86-2.01 0.21 0.62-1.67 0.95 0.85-2.14 1.02-3.25 0.33-0.84 1.08-2.55

0.20 0.043 0.01 0.02

RI: relapse incidence; NRM: non-relapse mortality; LFS: leukemia-free survival; OS: overall survival; GvHD: graft-versus-host disease; HR: Hazard Ratio; CI: Confidence Interval; P: P-value; UD: unrelated donor; MSD: matched-sibling donor; D: donor; R: recipient; ref: reference value; NA: not available; MAC: myeloablative conditioning; chemo: chemotherapy; TBI: total body irradiation; RIC: reduced intensity conditioning; TCD: T-cell depletion; PB: peripheral blood; BM: bone marrow.

cGvHD, but was not associated with RI, NRM, LFS or OS. The outcome in smaller centers (1-5 patients with MPAL in the study period) was not statistically different from larger centers (more than 5 patients with MPAL; data not shown). In a univariate analysis, the occurrence of â&#x2030;Ľgrade 2 aGvHD and cGvHD correlated with higher NRM and lower RI, but made no significant impact on LFS and OS (although a trend for worse LFS was observed with aGvHD; data not shown). An additional 51 patients with MPAL were transplanted in second complete remission (CR2). Three-year survival was inferior, mostly due to higher NRM (data not shown).

Matched-pair analysis In a matched-pair analysis, 498 patients with MPAL were matched with 1371 patients with acute lymphoblastic leukemia (ALL) and 498 patients were matched with 1412 patients with acute myelogenous leukemia (AML). Overall, the groups were well-matched although the ALL group had more high-risk cytogenetics. The comparison MPAL versus ALL showed no significant differences. When MPAL was compared with AML, MPAL had higher NRM and lower LFS (See Online Supplementary Tables 1 and 2).

Discussion Mixed phenotype acute leukemias are high-risk acute leukemias often with poor cytogenetics. As in other types of acute leukemia, the prognosis can potentially be improved by allo-SCT. We present here, extracted from the EBMT database, the largest study of patients with MPAL who underwent allo-SCT. Our results (56.3% OS at 3 years and 46.5% LFS at 3 years, with an RI of 31%) are highly encouraging and definitely improved over registry data from the Surveillance, Epidemiology and End Results Program (SEER). In the SEER study, a survival of 20-40% at three years was reported for patients aged over 20 years.6 In Table 4, a synopsis with outcomes of previous smaller studies investigating allo-SCT for MPAL is presented. In single institution studies, 9-59 patients with 2138

MPAL were transplanted. In a survey of 100 patients with MPAL treated at European hematology centers and classified according to WHO, only 20 underwent allogeneic or autologous transplantation. In this survey, no data on outcome of transplant are given. However, the overall median survival of adults with MPAL was only 11 months.23 The Center for International Blood and Marrow Transplant Research (CIBMTR) recently published a thoroughly investigated study which included 95 patients with a median age of 20 years. The CIBMTR study differs by including cord blood as source of stem cells and patients transplanted in CR2. Only 33% of patients in the CIBMTR study had PB as source of stem cells and only 11% received RIC (compared with 23% in the present study). The present study confirms and extends the CIBMTR study: allo-SCT is an effective treatment for adult patients with MPAL if a matched donor can be found. Classically, chronic myelogenous leukemia, AML, ALL, and to a lesser degree, myelodysplastic syndromes, are diseases considered to be sensitive to the graft-versusleukemia (GvL) effect of allo-SCT.24,25 In all these diseases, GvL effects are associated to a variable degree with GvHD. We propose here, supported by the CIBMTR study, to add MPAL to the list of potentially GvL-sensitive leukemias. A slightly lower rate of aGvHD and cGvHD than in the previous CIBMTR study was observed in the present study which may be due to the common use of in vivo TCD. However, treatment intensity also plays a role since MAC (especially with TBI) in our study yields better outcomes than RIC. In the study presented here, a matched pair-analysis showed transplant outcomes for MPAL are comparable with ALL and slightly worse than AML. This has to be put in perspective given that allo-SCT is a treatment offered to most patients with standard-risk AML, whereas transplant for ALL in younger patients is offered only in high-risk situations. The overall outcome in the multicenter EBMT setting (with a more homogeneous patient population: all patients in CR1, only matched related or unrelated donors) is comparable or slightly better than in the CIBMTR sethaematologica | 2017; 102(12)

MPAL with alloHSCT

Table 4. Synopsis of previous and current studies of patients with MPAL undergoing alloHSCT.

Region

Years of study

Patient #

Korea China

1995- 2008 20022011 2001- 2010 2006- 2013 1996- 2011 2000- 2014

9 24@ 35@ 18 29 95 519

Japan China CIBMTR EBMT

Median age (years) % in CR1 6 22 26 n.r. 30 20 38

n.r. n.r. n.r. n.r. 72 82 100

% of aGvHD LFS at 3 years n.r. n.r. n.r. 75 n.r. 48 33

n.r. 17 %# 56 %# 40 %# n.r. 56 % 47 %

OS at 3 years ≈ 30 % 24 %# 64 %# 48 %# 77 % 67 % 56 %

Reference 19 20 21 22 13 This study

MPAL: mixed phenotype acute leukemias; alloHSCT: allogeneic hematopoietic stem cell transplantation; CR1: first complete remission; aGvHD: graft-versus-host disease; LFS: leaukemia-free survival; OS: overall survival; CIBMTR: Center for International Blood and Marrow Transplantation Research; EBMT: European Society for Blood and Marrow Transplantation; @Patients treated according to 2 different protocols; #at five years.

ting. In the current, more recent EBMT study, the outcomes improved in recent years. Although not documented (being a registry-based study), we can assume that most patients who were bcr/abl positive were treated with tyrosine kinase inhibitors. A limiting factor of our study is that no central review was performed and that the WHO classification for MPAL was introduced only in 2008. This is compensated by the large patient numbers in a multi-center and multi-national setting. An important and unreported finding is the favorable effect of TBI as part of conditioning. In the multivariate analysis (see Table 3), MAC with TBI had a lower RI and higher LFS. Similar data were recently published for adult patients with T-cell ALL.26,27 Based on our present study, we recommend MAC with TBI for fit patients with MPAL as standard regimen. The superiority of MAC over RIC was recently also shown for AML and myelodysplastic syndromes.28 Looking into the future, two groups have performed whole exome or genome sequencing in patients with

References 1 Freireich EJ, Wiernik PH, Steensma DP. The leukemias: a half century of discovery. J Clin Oncol. 2014;32(31):3463-3469. 2 Ben-Bassat I, Gale RP. Hybrid acute leukemias. Leuk Res. 1984;8(6):929-936. 3 Weinberg OK, Arber DA. Mixed-phenotype acute leukemia: historical overview and a new definition. Leukemia. 2010;24(11):1844-1851. 4 Steensma DP. Oddballs: Acute leukemias of mixed phenotype and ambiguous origin. Hematol Oncol Clin North Am. 2011;25(6):1235-1253. 5 Yan L, Ping N, Zhu M, et al. Clinical, immunophenotypic, cytogenetic, and molecular genetic features in 117 adult patients with mixed-phenotype acute leukemia defined by WHO-2008 classification. Haematologica. 2012;97(11):17081712. 6 Shi R, Munker R. Survival of patients with mixed phenotype acute leukemias: A large population-based study. Leuk Res. 2015;39(6):606-616. 7 Wolach O, Stone RM. Mixed phenotype acute leukemia; current challenges in diagnosis and therapy. Curr Opin Hematol. 2017;24(2):139-145. 8 Bene MC, Castoldi G, Knapp W, et al. Proposals for the immunologic characteri-

haematologica | 2017; 102(12)

12 13

MPAL. The first study examined 23 adult and pediatric cases and found frequent mutations of epigenetic modifiers especially DNMT3A.29 The second study examined 115 pediatric cases of MPAL and found 35 recurrently mutated genes (including WT1, FLT3, NRAS, JAK3, and numerous other genes) and correlated these mutations with subtypes of MPAL.30 If targeted agents can be introduced into the treatment algorithm of MPAL, the prognosis of this rare, and until recently poor prognosis leukemia, may further improve. In conclusion, consolidation with alloHSCT in CR1 provides a favorable disease control to adult patients with MPAL with a moderate relapse risk. This resembles the outcome observed in patients with ALL. The observation of a possible beneficial impact of TBI as part of the conditioning regimen deserves further investigation. Acknowledgments The authors thank Emmanuelle Polge for helping with data collection.

zation of acute leukemias. European Group for the Immunologic Characterization of Leukemias (EGIL). Leukemia. 1995; 9(10):1783-1786. Borowitz MJ, Bene MC, Harris NL, et al. Acute leukaemias of ambiguous lineage. In: Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, et al. eds. World Health Organization (WHO) Classification of Tumours of Haematopoietic and Lymphoid Tissues, 4th ed. Lyon (France): International Agency for Research on Cancer. IARC Press. 2008, pp.150-155. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016;127(20):2391-2405. Deffis-Court M, Alvarado-Ibarra M, RuizArgüelles, et al. Diagnosing and treating mixed phenotype acute leukemia: a multicenter 10-year experience in México. Ann Hematol. 2014;93(4):595-601. Wolach O, Stone RM. How I treat: mixed phenotype acute leukemia. Blood. 2015; 125(16):123-131. Munker R, Brazauskas R, Wang HL, et al. Allogeneic hematopoietic cell transplantation for patients with mixed phenotype acute leukemia. Biol Blood Marrow Transplant. 2016;22(6):1024-1029. 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. Bacigalupo A, Ballen K, Rizzo D, et al. Defining the intensity of conditioning regimens: working definitions. Biol Blood Marrow Transplant. 2009;15(12):16281633. Przepiorka D, Weisdorf D, Martin P, et al. 1994 Consensus conference on acute GVHD grading. Bone Marrow Transplant. 1995;15(6):825-828. Gooley TA, Leisenring W, Crowley J, Storer BE. Estimation of failure probabilities in the presence of competing risks: new representations of old estimators. Stat Med. 1999;18(6):695-706. Ho D, Imai K, King G, Stuart E. Matching as Non-parametric Preprocessing for Reducing Model Dependence in Parametric Causal Inference. Political Analysis. 2007;15:199-236. Park JA, Ghim TT, Bae KW, et al. Stem cell transplant in the treatment of childhood biphenotypic acute leukemia. Pediatr Blood Cancer. 2009;53(3):444-452. Liu QF, Fan ZP, Wu MQ, et al. Allo-HSCT for acute leukemia of ambiguous lineage in adults: the comparison between standard conditioning and intensified conditioning regimens. Ann Hematol. 2013;92(5):679687.

2139

R. Munker et al. 21 Shimizu H, Saitoh T, Machida S, et al. Allogeneic hematopoietic stem cell transplantation for adult patients with mixed phenotype acute leukemia: results of a matched pair analysis. Eur J Haematol. 2015;95(5):455-460. 22 Tian H, Xu Y, Liu L, et al. Comparison of outcomes in mixed phenotype acute leukemia patients treated with chemotherapy and stem cell transplantation versus chemotherapy alone. Leuk Res. 2016;45:40-46. 23 Matutes E, Pickl WF, vanâ&#x20AC;&#x2122;t Veer M, et al. Mixed-phenotype acute leukemia: clinical and laboratory features and outcome in 100 patients defined according to the WHO 2008 classification. Blood. 2011;117(11):3163-3171.

2140

24 Negrin RS. Graft-versus-host disease versus graft-versus-leukemia. Hematology Am Soc Hematol Educ Program. 2015;2015:225-230. 25 Stern M, de Wreede LC, Brand R, et al. Sensitivity of hematological malignancies to graft-versus-host effects: an EBMT megafile analysis. Leukemia. 2014; 28(11):2235-2240. 26 Cahu X, Labopin M, Giebel S, et al. Impact of conditioning with TBI in adult patients with T-cell ALL who receive a myeloablative allogeneic stem cell transplantation: a report from the acute leukemia working party of EBMT. Bone Marrow Transplant. 2016;51(3):351-357. 27 Hamilton BK, Rybicki L, Abounader D, et al. Allogeneic hematopoietic cell trans-

plantation (HCT) adult T-cell acute lymphoblastic leukemia (T-ALL). Biol Blood Marrow Transplant. 2017;23(7):11171121. 28 Scott BL, Pasquini MC, Logan BR, et al. Myeloablative versus reduced-intensity hematopoietic cell transplantation for acute myeloid leukemia and myelodysplastic syndromes. J Clin Oncol. 2017; 35(11):1154-1161. 29 Eckstein OS, Wang L, Punia JN, et al Mixed-phenotype acute leukemia (MPAL) exhibits frequent mutations in DMNT3A and activated signaling genes. Exp Hematol. 2016;44(8):740-744. 30 Alexander TB, Gu Z, Choi JK, et al. Genomic landscape of mixed phenotype acute leukemia. Blood. 2016;128(22):454.

haematologica | 2017; 102(12)