Page 1


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 April 2017.


haematologica calendar of events

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

ESH-EBMT 21st Training Course on Haemopoietic Stem Cell Transplantation ESH EBMT Chairs: C Dufour, GJ Ossenkoppele, M Mohty, R Zeiser May 4-6, 2017 Saggart (Dublin), Ireland

Indolent Lymphoma Workshop SocietĂ Italiana di Ematologia (SIE) Chair: PL Zinzani May 15-16, 2017 Bologna, Italy

Annual Meeting PETHEMA Group PETHEMA Foundation May 25-27, 2017 Caceres, Spain

Korean Society of Hematology - Joint Symposium May 26-27, 2017 Seoul, South Korea 22nd Congress of the European Hematology Association European Hematology Association June 22 - 25, 2017 Madrid, Spain

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

EHA Tutorial on Biology and Management of Myeloid Malignancies October 20-22, 2017 Yerevan, Armenia

Russian Onco-Hematology Society's Conference on Malignant Lymphoma - Joint Symposium October 25-26, 2017 Moscow, Russian Federation

Turkish Society of Hematology - EHA Joint Symposium November 1 - 4, 2017 Antalya, Turkey

Argentinian Society of Hematology - EHA Joint Education Day November 17 - 18, 2017 Mar del Plata, Argentina

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

Calendar of Events updated on April 6, 2017


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

Table of Contents Volume 102, Issue 5: May 2017 Cover Figure Structure of the hemochromatosis protein (also known as the HFE protein) (image generated by www.somersault1824.com).

Editorials 805

Acute myeloid leukemia including favorable-risk group samples engraft in NSG mice: just be patient Dominique Bonnet

806

Better acute graft-versus-host disease outcomes for allogeneic transplant recipients in the modern era: a tacrolimus effect? Mahasweta Gooptu and John Koreth

Review Article 809

Pathophysiological consequences and benefits of HFE mutations: 20 years of research Ina Hollerer et al.

Articles Hematopoiesis

818

Bone marrow mesenchymal stromal cells induce nitric oxide synthase-dependent differentiation of CD11b+ cells that expedite hematopoietic recovery Cristina Trento et al.

Red Cell Biology & its Disorders

826

Loss of Forkhead box M1 promotes erythropoiesis through increased proliferation of erythroid progenitors Minyoung Youn et al.

Hemostasis

835

Obstetric antiphospholipid syndrome: early variations of angiogenic factors are associated with adverse outcomes Éva Cochery-Nouvellon et al.

Chronic Myeloid Leukemia

843

Increased peroxisome proliferator-activated receptor Îł activity reduces imatinib uptake and efficacy in chronic myeloid leukemia mononuclear cells Jueqiong Wang et al.

Acute Myeloid Leukemia

854

Long-term observation reveals high-frequency engraftment of human acute myeloid leukemia in immunodeficient mice Anna M. Paczulla et al.

865

Minimal residual disease prior to allogeneic hematopoietic cell transplantation in acute myeloid leukemia: a meta-analysis Sarah A. Buckley et al.

Chronic Lymphocytic Leukemia

874

Lymphocyte activation gene 3: a novel therapeutic target in chronic lymphocytic leukemia Mika Shapiro et al.

Haematologica 2017; vol. 102 no. 5 - May 2017 http://www.haematologica.org/


haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation Non-Hodgkin Lymphoma

883

c-Myc dysregulation is a co-transforming event for nuclear factor-ฮบB activated B cells Amandine David et al.

895

Association between quality of response and outcomes in patients with newly diagnosed mantle cell lymphoma receiving VR-CAP versus R-CHOP in the phase 3 LYM-3002 study Gregor Verhoef et al.

903

Safety and efficacy of abexinostat, a pan-histone deacetylase inhibitor, in non-Hodgkin lymphoma and chronic lymphocytic leukemia: results of a phase II study Vincent Ribrag, et al.

Plasma Cell Disorders

910

A concise revised Myeloma Comorbidity Index as a valid prognostic instrument in a large cohort of 801 multiple myeloma patients Monika Engelhardt et al.

922

Recovery of polyclonal immunoglobulins one year after autologous stem cell transplantation as a long-term predictor marker of progression and survival in multiple myeloma Verรณnica Gonzรกlez-Calle et al.

Cell Therapy & Immunotherapy

932

Plasma-derived proteomic biomarkers in human leukocyte antigen-haploidentical or human leukocyte antigen-matched bone marrow transplantation using post-transplantation cyclophosphamide Christopher G. Kanakry et al.

941

The effect of inter-unit HLA matching in double umbilical cord blood transplantation for acute leukemia Claudio Brunstein et al.

948

IL-2 promotes early Treg reconstitution after allogeneic hematopoietic cell transplantation Brian C. Betts et al.

958

Improved survival after acute graft-versus-host disease diagnosis in the modern era Hanna J. Khoury et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/102/5.toc

e163

Prostacyclin-analog therapy in sickle cell pulmonary hypertension Nargues A. Weir et al. http://www.haematologica.org/content/102/5/e163

e166

Prospective study of thrombosis and thrombospondin-1 expression in Chuvash polycythemia Adelina Sergueeva et al. http://www.haematologica.org/content/102/5/e166

e170

Suppression of RUNX1/ETO oncogenic activity by a small molecule inhibitor of tetramerization Julia Schanda et al. http://www.haematologica.org/content/102/5/e170

e175

Hyperlipidemia is a risk factor for osteonecrosis in children and young adults with acute lymphoblastic leukemia Signe Sloth Mogensen et al. http://www.haematologica.org/content/102/5/e175

e179

Infection as a cause of childhood leukemia: virus detection employing whole genome sequencing Christoph Bartenhagen et al. http://www.haematologica.org/content/102/5/e179

Haematologica 2017; vol. 102 no. 5 - May 2017 http://www.haematologica.org/


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

e184

Hyper-CVAD + epratuzumab as a salvage regimen for younger patients with relapsed/refractory CD22-positive precursor B-cell acute lymphocytic leukemia Patrice Chevallier et al. http://www.haematologica.org/content/102/5/e184

e187

BCR-ABL translocation as a favorable prognostic factor in elderly patients with acute lymphoblastic leukemia in the era of potent tyrosine kinase inhibitors Ja Min Byun et al. http://www.haematologica.org/content/102/5/e187

e191

The Bruton tyrosine kinase inhibitor ibrutinib abrogates triggering receptor on myeloid cells 1-mediated neutrophil activation Nicole Stadler et al. http://www.haematologica.org/content/102/5/e191

e195

Deltex-1 mutations predict poor survival in diffuse large B-cell lymphoma Leo Meriranta et al. http://www.haematologica.org/content/102/5/e195

e199

Genetic polymorphism at BCL2 as a predictor for rituximab, cyclophosphamide, doxorubicin, vincristine and prednisone efficacy in patients with diffuse large B-cell lymphoma Morteza Bashash et al. http://www.haematologica.org/content/102/5/e199

e203

Second-line rituximab, lenalidomide, and bendamustine in mantle cell lymphoma: a phase II clinical trial of the Fondazione Italiana Linfomi Francesco Zaja et al. http://www.haematologica.org/content/102/5/e203

Case Reports Case Reports are available online only at www.haematologica.org/content/102/5.toc

e207

Changes in allele frequencies of CSF3R and SETBP1 mutations and evidence of clonal evolution in a chronic neutrophilic leukemia patient treated with ruxolitinib Zohra Nooruddin et al. http://www.haematologica.org/content/102/5/e207

e210

Repeated fecal microbiota transplantations attenuate diarrhea and lead to sustained changes in the fecal microbiota in acute, refractory gastrointestinal graft-versus-host-disease Walter Spindelboeck et al. http://www.haematologica.org/content/102/5/e210

Comments Comments are available online only at www.haematologica.org/content/102/5.toc

e214

The TP53 Pro72Arg SNP in de novo acute myeloid leukemia Eduard Schulz and Heinz Sill http://www.haematologica.org/content/102/5/e214

Haematologica 2017; vol. 102 no. 5 - May 2017 http://www.haematologica.org/


EDITORIALS Acute myeloid leukemia including favorable-risk group samples engraft in NSG mice: just be patient Dominique Bonnet The Francis Crick Institute, Haematopoietic Stem Cell Laboratory, London, UK E-mail: dominique.bonnet@crick.ac.uk

I

doi:10.3324/haematol.2017.165159

n this issue of Haematologica, Paczulla et al. demonstrate that by extending the time to read out the engraftment of primary acute myeloid leukemia (AML) cells in immunodeficient IL2RGnull NOD/SCID (NSG) mice all types of AML samples, including favorable prognosis AML, a group that has been previously reported to be particularly difficult to engraft, could be studied.1 Xenotransplantation of human AML in immunocompromised animals has been critical for defining leukemic stem cells and remains the primary method for functional assessment of primary human AML biology as well as providing the best in vivo preclinical model. However, the use of immunodeficient mouse models to study primary AML samples has shown some limitations. Over the past decades, important advances have been made to improve patientderived xenograft (PDX) modeling of AML through the use of mice that are more immunodeficient, such as beta 2 micro-globulinnull NOD/SCID mice and more recently the IL2RG knockout NOD/SCID mice.2-4 Even with the most immunodeficient mouse model (NSG mice), only 66% of AML engraft at 10-16 weeks after transplantation.2,3 Interestingly, the capacity to engraft has been correlated with patients’ clinical outcome, with non-engrafting samples being those from the more favorable risk group.4,5 In this new study by Paczulla et al., it appears that the dominant factor in engraftment is the speed of developing a detectable disease, as all primary AML samples, including those from the favorable-risk group, might be able to engraft in NSG mice if the duration of the experiment is prolonged for up to 1 year. The authors also found that the latency of developing a detectable level of engraftment is correlated with the frequency of leukemic stem cells (LSC) in a patient. This notion is further supported by recent data from Griessinger et al., who showed ex vivo that non-engrafter AML samples have lower levels of leukemia long-term culture-initiating cells.6 In addition, others have found that a high frequency of phenotypically defined CD34+CD38CD123+ LSC is correlated with the risk of a poor clinical outcome.7 Lastly, using gene expression data, Dick’s group identified a “leukemic stem cell gene signature” and correlated the presence of this gene signature in AML samples with the aggressiveness of the disease.8 They subsequently refined this signature to 17 genes and confirmed their initial data in an extended cohort of patients.9 By investigating other components that could explain the long latency of engraftment of certain samples, the authors explored the possibility that samples from patients with a certain genetic background (favorable- or some intermediate-risk AML) might be more sensitive to microenvironmental factors for their survival and growth. Despite not detecting any significant difference in the proliferation capacity between samples from short- and long-latency “engrafter” patients, they only focused on the late time-point when a high leukemic burden was already present. Differences in the proliferative index between “non-engrafters” and haematologica | 2017; 102(5)

“engrafters” have been reported using an ex vivo co-culture system5 suggesting that the proliferative capacity of “longlatency” engrafters (originally called non-engrafters) is potentially playing a role. Indeed, the notion that some AML cells proliferate less at an early stage in the murine environment is supported by the results obtained with the new MISTRG transgenic mouse model.10 In this model, human cytokines are knocked into the endogenous mouse loci. In these mice, Ellegast et al. recently demonstrated reproducible engraftment of favorable-risk group AML samples with a shorter latency. Supporting the notion that engraftment could be impaired by cross-species differences, Majeti’s group, JJ Schuringa’s group and Bonnet’s group reported in parallel the feasibility of engrafting previously “non-engrafter” AML samples by implanting a three-dimensional humanized bone marrow stroma scaffold/ossicles.11-13 In this humanized microenvironment, non-engrafter samples were able to engraft with a shorter latency than that reported by Paczulla et al., further suggesting that long-latency engrafters might be more sensitive to microenvironmental signals than “engrafters”. Thus, the engraftment capacity or more exactly the latency to engraft primary AML samples in immunodeficient mice is likely dependent on homing factors, survival in a foreign niche, absence of specific growth factors and supporting stroma cells as well as intrinsic differences in LSC frequency. Humanizing the microenvironment via the use of new transgenic mice and/or via the use of three-dimensional scaffolds should clearly help to shorten the latency of disease development, extending the use of the PDX model to all AML samples. Shortening the latency to detect engraftment will not only reduce the cost of maintaining mice but will also provide a more useful PDX model for predicting patients’ responses to potential therapies. As new strains of mice become available and novel and more complex three-dimensional humanized scaffolds are developed, it will be necessary to examine in more detail the value of the different PDX models for predicting patients’ responses. Indeed, as indicated by assessment of the subclonal architecture of individual PDX, some reports suggest highly selective engraftment of specific subclones in PDX mice.14, 15 Paczulla et al. report that, despite the long latency to develop leukemia, the phenotypic and genetic features in mouse-derived versus pre-transplanted AML cells are conserved. Since the subclonal composition of the AML after PDX will influence responses to therapies, reproducing similar clonal architecture in PDX as in pre-transplant samples is clearly an important aspect in the development of valuable preclinical PDX models.

References 1. Paczulla AM, Dirnhofer S, Konantz M, et al. Long-term observation reveals high-frequency engraftment of human acute myeloid leukemia in immunodeficient mice. Haematologica. 2017;102(5):854-864.

805


Editorials 2. Feuring-Buske M, Gerhard B, Cashman J, Humphries RK, Eaves CJ, Hogge DE. Improved engraftment of human acute myeloid leukemia progenitor cells in beta 2-microglobulin-deficient NOD/SCID mice and in NOD/SCID mice transgenic for human growth factors. Leukemia. 2003;17(4):760-763. 3. Sanchez PV, Perry RL, Sarry JE, et al. A robust xenotransplantation model for acute myeloid leukemia. Leukemia. 2009;23(11):21092117. 4. Vargaftig J, Taussig DC, Griessinger E, et al. Frequency of leukemic initiating cells does not depend on the xenotransplantation model used. Leukemia. 2012;26(4):858-860. 5. Pearce DJ, Taussig D, Zibara K, et al. AML engraftment in the NOD/SCID assay reflects the outcome of AML: implications for our understanding of the heterogeneity of AML. Blood. 2006;107(3):1166-1173. 6. Griessinger E, Anjos-Afonso F, Vargaftig J, et al. Frequency and dynamics of leukemia-initiating cells during short-term ex vivo culture informs outcomes in acute myeloid leukemia patients. Cancer Res. 2016;76(8):2082-2086. 7. Vergez F, Green AS, Tamburini J, et al. High levels of CD34+CD38low/-CD123+ blasts are predictive of an adverse outcome in acute myeloid leukemia: a Groupe Ouest-Est des Leucemies Aigues et Maladies du Sang (GOELAMS) study. Haematologica. 2011;96(12):1792-1798. 8. Eppert K, Takenaka K, Lechman ER, et al. Stem cell gene expression

9. 10. 11.

12. 13. 14. 15.

programs influence clinical outcome in human leukemia. Nat Med. 2011;17(9):1086-1093. Ng SW, Mitchell A, Kennedy JA, et al. A 17-gene stemness score for rapid determination of risk in acute leukaemia. Nature. 2016;540 (7633):433-437. Ellegast JM, Rauch PJ, Kovtonyuk LV, et al. inv(16) and NPM1mut AMLs engraft human cytokine knock-in mice. Blood. 2016;128(17):2130-2134. Reinisch A, Thomas D, Corces MR, et al. A humanized bone marrow ossicle xenotransplantation model enables improved engraftment of healthy and leukemic human hematopoietic cells. Nat Med. 2016;22(7):812-821. Antonelli A, Noort WA, Jaques J, et al. Establishing human leukemia xenograft mouse models by implanting human bone marrow-like scaffold-based niches. Blood. 2016;128(25):2949-2959. Abarrategi A, Foster K, Hamilton A, et al. Versatile humanized niche model enables study of normal and malignant human hematopoiesis. J Clin Invest. 2017;127(2):543-548. Klco JM, Spencer DH, Miller CA, et al. Functional heterogeneity of genetically defined subclones in acute myeloid leukemia. Cancer Cell. 2014;25(3):379-392. Quek L, Otto GW, Garnett C, et al. Genetically distinct leukemic stem cells in human CD34- acute myeloid leukemia are arrested at a hemopoietic precursor-like stage. J Exp Med. 2016;213(8):15131535.

Better acute graft-versus-host disease outcomes for allogeneic transplant recipients in the modern era: a tacrolimus effect? Mahasweta Gooptu and John Koreth 1

Division of Hematologic Malignancies, Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, USA

E-mail: jkoreth@partners.org

A

doi:10.3324/haematol.2017.165266

cute graft-versus-host disease (GvHD) continues to be an important complication following allogeneic hematopoietic stem-cell transplantation (HSCT) in the modern era. With matched related and unrelated donors, the cumulative incidence of acute GvHD remains approximately 40-60%, respectively.1 Survival outcomes for patients undergoing HSCT have however improved over the last few decades because of improvements in non-relapse mortality rather than relapse incidence.2,3 It is an interesting conundrum that improvement in nonrelapse mortality and survival has occurred despite a lack of sentinel advancements in acute GvHD prophylaxis or treatment. Calcineurin inhibitors are the cornerstone of prophylaxis, while steroids remain the mainstay of treatment.4 The question arises whether improvements in nonrelapse mortality and survival are due: (i) solely to improved management of acute GvHD complications (infections and organ toxicity); (ii) to better rates of acute GvHD response to steroid-based therapy; or (iii) to a secular shift in the nature and severity of acute GvHD over time. Khoury et al. now offer some insights into these important questions in this issue of Haematologica.5 In a large registry analysis (n=2905) from the Center for International Blood and Marrow Transplant Research (CIBMTR), they evaluate the incidence and outcomes of grade II-IV acute GvHD developing within 100 days after myeloablative, HLA-matched HSCT over three successive time periods [1999-2001 (n=497), 2002-2005 (n=962), 2006-2012 (n=1446)]. These periods overlap with important advances in supportive care (e.g., azoles for fungal 806

infections, valacyclovir for cytomegalovirus).6,7 The predominant GvHD prophylaxis regimens were tacrolimusbased (n=1767; 60.7%) or cyclosporine (CsA)-based (n=1077; 37.1%). Patients in the tacrolimus and CsA groups were well-balanced with regard to baseline characteristics (except for more matched unrelated donor and peripheral blood stem-cell grafts in the tacrolimus cohort). The authors then compared the outcomes of patients in each time period stratified by GvHD prophylaxis (CsAbased versus tacrolimus-based) and grade of acute GvHD (grade II versus grades III-IV). Several interesting observations resulted. Firstly, the severity of acute GvHD appears to have decreased over time. The proportion of patients with grades III-IV severe acute GvHD in the most recent time period (2006-2012) has decreased by 20% compared to that in the earliest time period (1999-2001). This could be due to a true decrease in acute GvHD severity or a drift within acute GvHD categories, with more grade II patients being identified and reported to the CIBMTR. Simultaneously, there are fewer patients with concurrent three-organ (gut/skin/liver) involvement in recent years compared to previous years, while the proportion of patients with gut acute GvHD with or without skin involvement has increased significantly. Secondly, on multivariate analysis, in the subgroup of HSCT recipients with grades II-IV acute GvHD who received tacrolimus prophylaxis, overall survival (Figure 1, from the original article) and non-relapse mortality have improved in the modern era. The improvement appears to be due to fewer deaths from organ toxicity and infection. Interestingly, this improvement is not seen in HSCT recipihaematologica | 2017; 102(5)


Editorials

A

B

Figure 1. Overall survival. (A) Adjusted probability of overall survival following a diagnosis of grade II-IV acute GvHD among patients treated with tacrolimusbased GvHD prophylaxis. (B) Adjusted probability of overall survival following a diagnosis of grade II-IV acute GvHD among patients treated with cyclosporine-based GvHD prophylaxis. (Figure adapted from original article5)

Table 1. Multivariate analysis results - effect of time cohort on overall survival and treatment-related mortality.

Grade II acute GvHD Tacrolimus

Year of transplant

Overall survival Hazard Confidence ratio interval

P-value

Year of transplant

Treatment-related mortality Hazard Confidence P-value ratio interval

0.0494 1999-2001 2002-2005 2006-2012

1.00 1.02 0.80

0.69-1.51 0.55-1.16

0.91 0.25

0.0397 1999-2001 2002-2005 2006-2012

1.00 0.60 0.51

0.34-1.04 0.30-0.87

0.071 0.013

Grade III-IV acute GvHD Tacrolimus

Year of transplant

Overall survival Hazard Confidence ratio interval

P-value

Year of transplant

Treatment-related mortality Hazard Confidence P-value ratio interval

0.19 1999-2001 2002-2005 2006-2012

1.00 0.98 0.82

0.71-1.37 0.59-1.14

0.92 0.24

0.0785 1999-2001 2002-2005 2006-2012

1.00 0.90 0.70

0.62-1.30 0.48-1.03

0.56 0.07

Adapted from original article5. GvHD: graft-versus-host disease.

ents with acute GvHD who received CsA prophylaxis. Finally, in the tacrolimus subgroup, it is the patients with grade II acute GvHD who have had a significant reduction in hazard for mortality and treatment-related mortality over time, rather than the patients with severe grades III-IV acute GvHD patients (Table 1, adapted from the original article5). There could be a number of reasons for these findings. The authors speculate that the changes in acute GvHD severity and organ involvement could be caused by changing practices in GvHD prophylaxis over time, with increased use of tacrolimus rather than CsA. While one alternative factor underlying the reduction in severity of acute GvHD could be better high resolution HLA-matching techniques, particularly in matched unrelated donor HSCT,8 this should be applicable uniformly to both tacrolimus- and CsA-based regimens. In support of their conjecture, results from randomized phase III trials in HLA-matched HSCT have shown that tacrolimus-based haematologica | 2017; 102(5)

prophylaxis is associated with less acute GvHD (both grades II-IV as well as III-IV), albeit with similar overall survival, infections and relapse, when compared to CsAbased prophylaxis.9 A number of other subsequent trials have echoed these results.10–12 Consequently over the years, most transplant centers have adopted tacrolimusbased GvHD prophylaxis. This is reflected in the authors’ data, with tacrolimus-based prophylaxis having largely replaced CsA-based prophylaxis, being used in 80% of cases in the 2006-2012 period compared to 27% in the 1999-2001 period. The improvements in non-relapse mortality and overall survival in acute GvHD patients, on the other hand, almost certainly reflect improvements in supportive care and infection prophylaxis/treatment in transplant recipients. Thus, even when patients develop acute GvHD, they have an improved chance of survival. This was also suggested in a prior study by El-Jawahiri et al.,13 in which 807


Editorials

improved overall survival and treatment-related mortality were seen in patients with grade IV acute GvHD. This effect is not seen in CsA recipients in this study; however, this may simply reflect the limited numbers treated with CsA-based regimens in the modern era. We could also speculate that, in the current era, there is better management of tacrolimus toxicity and more stringent monitoring of tacrolimus drug levels, also perhaps accounting for better outcomes. In summary, the current study traces secular trends in the incidence of acute GvHD, demonstrating that the severity of this complication has decreased over time, with a concomitant reduction in three-organ involvement (gut/skin/liver). It is a plausible but unproven inference that these improvements are related to increased utilization of tacrolimus-based prophylaxis (versus CsA-based). Furthermore, in the tacrolimus-treated subgroup, acute GvHD patients with milder manifestations (grade II disease) have had improved non-relapse mortality and survival, with reduction in deaths from organ toxicity and infection. Overall, these results reflect the strides made in transplantation practice, where improvements in infection management, supportive care, more stringent monitoring of immunosuppressive drugs, such as tacrolimus, as well as early recognition and management of drug toxicities, can lead to improved outcomes even in the absence of radical advances in acute GvHD therapy. Major therapeutic advances are however still awaited for those with severe acute GvHD (grades III-IV), who are in the most need.

References 1. Jagasia M, Arora M, Flowers MED, et al. Risk factors for acute GVHD and survival after hematopoietic cell transplantation. Blood. 2012;119(1):296–307.

808

2. Gooley TA, Chien JW, Pergam SA, et al. Reduced mortality after allogeneic hematopoietic cell transplantation. N Engl J Med. 2010;363(22):2091–2101. 3. Hahn T, McCarthy PL, Hassebroek A, et al. Significant improvement in survival after allogeneic hematopoietic cell transplantation during a period of significantly increased use, older recipient age, and use of unrelated donors. J Clin Oncol. 2013;31(19):2437–2449. 4. Bolaños-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; quiz 3335. 5. Khoury HJ, Wang T, Hemmer MT, et al. Improved survival after acute graft-versus-host disease diagnosis in the modern era. Haematologica. 2017;102(5):958-963. 6. Slavin MA, Osborne B, Adams R, et al. Efficacy and safety of fluconazole prophylaxis for fungal infections after marrow transplantation—a prospective, randomized, double-blind study. J Infect Dis. 1995;171(6):1545–1552. 7. Ljungman P, de la Camara R, Milpied N, et al. Randomized study of valacyclovir as prophylaxis against cytomegalovirus reactivation in recipients of allogeneic bone marrow transplants. Blood. 2002;99(8):3050–3056. 8. Flomenberg N, Baxter-Lowe LA, Confer D, et al. Impact of HLA class I and class II high-resolution matching on outcomes of unrelated donor bone marrow transplantation: HLA-C mismatching is associated with a strong adverse effect on transplantation outcome. Blood. 2004;104(7):1923–1930. 9. Nash RA, Antin JH, Karanes C, et al. Phase 3 study comparing methotrexate and tacrolimus with methotrexate and cyclosporine for prophylaxis of acute graft-versus-host disease after marrow transplantation from unrelated donors. Blood. 2000;96(6):2062–2068. 10. Ziakas PD, Zervou FN, Zacharioudakis IM, Mylonakis E. Graft-versushost disease prophylaxis after transplantation: a network meta-analysis. Plos One. 2014;9(12):e114735. 11. Sabry W, Le Blanc R, Labbé A-C, et al. Graft-versus-host disease prophylaxis with tacrolimus and mycophenolate mofetil in HLA-matched nonmyeloablative transplant recipients is associated with very low incidence of GVHD and nonrelapse mortality. Biol Blood Marrow Transplant. 2009;15(8):919–929. 12. Törlén J, Ringdén O, Garming-Legert K, et al. A prospective randomized trial comparing cyclosporine/methotrexate and tacrolimus/sirolimus as graft-versus-host disease prophylaxis after allogeneic hematopoietic stem cell transplantation. Haematologica. 2016;101(11):1417–1425. 13. El-Jawahri A, Li S, Antin JH, et al. Improved treatment-related mortality and overall survival of patients with grade IV acute GVHD in the modern years. Biol Blood Marrow Transplant. 2016;22(5):910–918.

haematologica | 2017; 102(5)


REVIEW ARTICLE

Pathophysiological consequences and benefits of HFE mutations: 20 years of research Ina Hollerer,1,2,3 André Bachmann4 and Martina U. Muckenthaler1,3

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

1 Molecular Medicine Partnership Unit (MMPU), Heidelberg, Germany; 2European Molecular Biology Laboratory (EMBL), Heidelberg, Germany; 3Department of Pediatric Oncology, Hematology and Immunology, University of Heidelberg, Germany and 4Private Dental Practice, Zürich, Switzerland

ABSTRACT

Haematologica 2017 Volume 102(5):809-817

M

utations in the HFE (hemochromatosis) gene cause hereditary hemochromatosis, an iron overload disorder that is hallmarked by excessive accumulation of iron in parenchymal organs. The HFE mutation p.Cys282Tyr is pathologically most relevant and occurs in the Caucasian population with a carrier frequency of up to 1 in 8 in specific European regions. Despite this high prevalence, the mutation causes a clinically relevant phenotype only in a minority of cases. In this review, we summarize historical facts and recent research findings about hereditary hemochromatosis, and outline the pathological consequences of the associated gene defects. In addition, we discuss potential advantages of HFE mutations in asymptomatic carriers.

Introduction Iron plays a key role in various physiological pathways. All cells of the human body contain iron as an integral part of FeS-proteins. These are essential for oxidation-reduction reactions that occur, for example, in the mitochondrial respiratory chain where biochemical energy is generated. In red blood cells, iron binds oxygen in the hemoglobin molecule to enable its transport from respiratory organs throughout the human body. Sufficient amounts of serum iron are mainly ensured by macrophages, which recycle approximately 25 mg of iron a day from aging or damaged erythrocytes. Iron losses resulting from bleeding or skin desquamation are compensated for by dietary iron absorption, which amounts to approximately 1-2 mg/day.1 Duodenal iron absorption is a highly regulated process. If duodenal iron uptake exceeds the requirements for physiological iron consuming processes, such as erythropoiesis, excess iron accumulates in parenchymal organs, including the liver, heart, and pancreas. Iron overload causes oxidative damage in these organs and, ultimately, the iron-overload disease hereditary hemochromatosis (HH) may develop when genetic and environmental risk factors are present.2 Ever since it was first described in the 19th century, scientists and clinicians have been increasing their efforts to unravel the causes and consequences of HH. Modern research has been successful in identifying HH-causing gene mutations and unraveling molecular mechanisms responsible for elevated dietary iron uptake. HH is mainly caused by mutations in the HFE (hemochromatosis) gene. The most pathologically influential HFE mutation, p.Cys282Tyr (C282Y), is frequently inherited in a heterozygous state (overall carrier frequency approx. 1:16) in the Caucasian population, but accounts for a phenotype only in a minority of cases.3 This observation has led to the hypothesis that HFE mutations may confer a genetic advantage to asymptomatic hemochromatosis 'patients' and have, therefore, continued to spread.4

A history of hereditary hemochromatosis Hemochromatosis was first described in the mid-1800s by French physicians who referred to the disease as “bronze diabetes” and “pigmented cirrhosis”.5-7 A few years later, the German pathologist Friedrich Daniel von Recklinghausen linked the syndrome to iron metabolism after he had observed an excess of iron in the tissues of patients and introduced the present-day term hemochromatosis.8 In haematologica | 2017; 102(5)

Correspondence: martina.muckenthaler@med.uni-heidelberg.de

Received: November 24, 2016. Accepted: March 1, 2017. Pre-published: March 24, 2017. doi:10.3324/haematol.2016.160432 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/809 ©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.

809


I. Hollerer et al.

those days, scientists and physicians believed that the syndrome was exclusively caused by environmental factors and conditions such as diabetes, alcohol abuse, metabolic disturbances, or infections. The English gerontologist Sheldon rejected this hypothesis in 1935 and postulated that hemochromatosis was a hereditary disease.9 This assumption was experimentally verified in the 1970s in a study that associated hemochromatosis with the HLA (human leukocyte antigen) gene locus and identified it as an autosomal recessive disease that mainly affects men.10 The discovery of the intracellular iron storage protein ferritin11 and the iron-binding blood plasma protein transferrin12 in the first half of the 20th century gave researchers the first molecular hints about the biological mechanisms regulating iron metabolism. Those and subsequent mechanistic studies have provided an understanding of the molecular processes involved in hemochromatosis. Twenty years ago, the genetic factor responsible for HLAlinked HH was finally identified as a mutation in the gene encoding the non-classical MHC class I-like molecule HFE.13 Subsequently, rare non-HFE-related HH subtypes were discovered, hallmarked by mutations in newly identified iron-related proteins such as hemojuvelin (HJV; hemochromatosis type IIA),14 hepcidin, the master regulator of systemic iron homeostasis, (HAMP; hemochromatosis type IIB),15 transferrin receptor 2 (TfR2; hemochromatosis type III),16 or ferroportin (FPN or SLC40A1; type IV or ferroportin disease).16,17 Large-scale population studies have revealed that most HFE mutation carriers with a mild iron overload-phenotype lack any clinically relevant disease symptoms.19 This has led to a debate about the potential benefits of HFE gene defects and has been motivating the scientific community to identify potential advantages of these mutations.

The molecular and genetic causes of hereditary hemochromatosis The iron-regulated hormone hepcidin is of central importance to the pathogenesis of HH. It is mainly produced in hepatocytes and secreted into the blood stream. It binds to and degrades its target receptor, the iron exporter ferroportin, to inhibit iron release from duodenal enterocytes, iron-recycling macrophages, and hepatocytes. Any defect that ultimately impairs hepcidin function and inhibits the hormone’s ability to monitor and regulate serum iron levels may provoke an iron overload phenotype. In HH patients, hepcidin levels are reduced, causing excess ferroportin-mediated iron export and, as a consequence, increased dietary iron uptake and release from macrophages.20 Elevated iron levels in the blood stream subsequently lead to the saturation of the binding capacity of the iron transporter transferrin. Above a transferrin saturation of approximately 75%21 highly reactive, non-transferrin bound iron species (especially labile plasma iron, LPI) appear in the blood, which will preferentially be taken up by parenchymal cells of the liver, the pancreas and other organs,22 and ultimately provoke an HH phenotype. Mutations in either the hepcidin gene itself, in genes affecting upstream activators of hepcidin expression (HFE, TFR2, HJV), or in ferroportin (the iron exporter which acts as the hepcidin receptor) can cause different classes and subtypes of HH20 (Figure 1). Most cases of HH are due to mutations in the HFE gene. The most prevalent disease810

causing HFE mutation in the general population is the 845G polymorphism, which causes a p.Cys282Tyr amino acid substitution (C282Y) in the HFE protein. Today, approximately 0.4% of Caucasians carry a homozygous23,24 and approximately 6% a heterozygous HFE C282Y mutation.24 HFE is a non-classical major histocompatibility (MHC) protein13 located on the cell surface. The C282Y mutation disrupts the formation of a disulfide bond in the HFE protein and impairs its capability to bind β2-microglobulin.25 As a consequence, HFE is unable to reach the cell surface and aggregates intracellularly.25,26 This causes impaired signaling leading to reduced hepcidin mRNA expression, decreased plasma hepcidin levels, and excessive systemic iron accumulation in adults (aged over 40 years).20 The molecular mechanism by which HFE regulates hepcidin expression is not yet fully understood. It has been proposed that HFE plays a regulatory role in the sensing of serum TF-Fe concentrations, involving protein-protein interactions between HFE, the ubiquitous transferrin receptor TfR1 and the liver-specific transferrin receptor TfR2. High concentrations of iron-bound transferrin dissociate HFE from TfR127,28 which have been found to strengthen the interaction between HFE and Tfr2.29 Increased iron levels further stimulate the expression of BMP6 (bone morphogenetic protein 6), a member of the TGFβ superfamily, whose genetic impairment causes severe iron overload in mice30,31 and in patients.32 BMP6 binds to the GPI-anchored receptor hemojuvelin (HJV) at the cell surface,33 as well as to type 1 (ALK2 and ALK334) and type 2 (BMPR2 and ACTR2a35) serine threonine kinase receptors. As a result, SMAD1/5/8 is phosphorylated and binds to SMAD4, which translocates to the nucleus to induce hepcidin transcription.36 Stimulation of hepcidin expression by BMP6 has recently been reported to also take place in an HJV-independent manner.37 HFE has been found to prevent the ubiquitination and proteasomal degradation of the BMP6-receptor ALK3,38 as well as to engage in a tertiary complex with the BMP6-receptor Hjv and TfR2.39 These observations suggest a role of HFE in BMP/SMAD signaling and provide a first mechanistic explanation for the impairment of BMP/SMAD signaling in patients with HFE-hemochromatosis.40 HFE gene sequencing approaches have identified additional HFE mutations with different pathological impact. These include the amino acid alteration S65C, which is not considered clinically meaningful,20,41,42 or H63D, which may, in rare cases, contribute to abnormal iron parameters in H63D/C282Y compound heterozygote individuals.43-45 By contrast, individuals carrying one H63D and one healthy allele are asymptomatic as long as no additional risk factors are present.41 In addition, researchers discovered a deletion (p.Y231del) in an Huh7 hepatoma cell line derived from a Japanese HH patient46 which prevents HFE cell surface expression. The identical mutation has more recently been discovered in another Japanese patient,47 showing, for the first time, that HFE-associated HH can also occur in Asians. Furthermore, a few Sardinian individuals show deletions of the entire HFE gene.48,49 Additional HFE mutations have been detected to influence iron levels when co-inherited with heterozygous C282Y mutations, such as the p.Arg226Gly (R226G) mutation50 or the nonsense mutations HFE-Brianza and HFE-Ossola, termed after the Italian provinces they were detected in.51 Although the prevalence of such mutations is low, they show that the presence of genetic factors may contribute haematologica | 2017; 102(5)


Pathophysiological consequences and benefits of HFE mutations

Figure 1. Hereditary hemochromatosis (HH). Three different HH classes (HFE-related HH, non-HFErelated HH, and ferroportin disease) are subcategorized into different subtypes (I, II A + B, III and IV) depending on mutations in iron-related genes (HFE, HJV, HAMP, TFR2, FPN/SLC40A1) and related pathophysiology. FD: ferroportin disease.

to the clinical manifestation of hemochromatosis in C282Y heterozygotes. Non-HFE-hemochromatosis (commonly referred to as hemochromatosis type II to IV) is much rarer than HFEHH. In contrast to HFE-hemochromatosis, it also occurs in individuals of non-European descent, in both adult and juvenile onset forms.52 Mutations causing non-HFE-HH are detected in genes encoding hepatocytic membrane proteins [HJV (hemojuvelin), TFR2], which play a role in the monitoring of iron levels and signaling to hepcidin (see above), or HAMP itself. Mutations in HJV14 and HAMP15 cause juvenile non-HFE-HH forms (HH types II A and II B) which are characterized by very low circulating hepcidin levels. Patients usually develop a severe iron overload phenotype before the age of 30 years, including cardiovascular, liver, and endocrine complications.53 Adult onset nonHFE-HH forms (HH types III and IV) are either caused by mutations in TFR2 encoding transferrin receptor 2,16 or the hepcidin receptor ferroportin (FPN/SLC40A1), generally referred to as HH type IV.55 Gain-of-function mutations within ferroportin confer resistance to hepcidin binding and thus prevent ferroportin internalization and degradation (HH type IV, classical ferroportin disease with gainof-function mutation). As a consequence, uncontrolled iron export from cell types expressing ferroportin, such as duodenal enterocytes or macrophages, causes high levels haematologica | 2017; 102(5)

of iron overload.56 By contrast, loss-of-function mutations in ferroportin diminish the ability of ferroportin for iron export (HH type IV, classical ferroportin disease with lossof-function mutations), which is characterized by iron accumulation in macrophages of the spleen and the liver, and is associated with low serum iron levels.2

Symptoms, diagnosis and treatment of HH Elevated liver enzymes and/or iron parameters indicative of iron overload (serum ferritin and transferrin saturation levels) usually precede a symptomatic manifestation of hemochromatosis.20 Individuals with elevated body iron levels despite fully functional erythropoiesis are thus commonly diagnosed with hemochromatosis, regardless of whether or not they show disease symptoms.57 Genetic and biochemical screening of the general population for HFE mutations has been discussed but is currently not recommended because of its high costs and the low penetrance of the disease.20,58 Depending on the degree of iron accumulation, cytotoxic hydroxyl and lipid radicals will be produced causing various organ pathologies, including those of the liver, heart, and pancreas, and ultimately lead to organ failure in affected individuals who remain untreated.2 The symptoms of hemochromatosis range from chronic fatigue, hyperpigmentation, joint and bone symptoms to diabetes, and liver 811


I. Hollerer et al.

diseases, such as fibrosis, cirrhosis, and hepatocellular carcinoma20 (Table 1 and Figure 2). The correlation between increased ferritin levels and transferrin saturation with organ damage is poor,59,60 highlighting the need for better markers for iron-induced organ damage. Despite this, symptoms develop especially in men with homozygous p.Cys282Tyr mutations and serum ferritin concentrations greater than 1000 mg/L.61 Liver biopsies may thus be performed to test for liver damage in HH when serum ferritin levels exceed 1000 Âľg/L, while diverse arthropathies may also be observed in C282Y mutation carriers without severe iron overload.60 Patients diagnosed with clinical HH face a higher risk of developing liver cancer (when cirrhosis is present at diagnosis) compared to controls.62 Although highly debated, several studies have suggested a direct association of HFE mutations with the progression of various cancers including colon, breast, prostate, and epithelial ovarian cancer.63-66 Patients with iron overload are also more susceptible to some infections, such as those caused by Listeria monocytogenes, Vibrio vulnificus, or HIV.67,68 Consistently, studies in mice with iron overload resulting from a hepcidin deficiency showed decreased survival of animals infected by Vibrio vulnificus69 or the malaria-causing Plasmodium berghei.70 In addition, Hfe-deficiency in mice prevents a proper response to infection in that recruitment of pulmonary neutrophils upon an inflammatory stimulus is reduced.71 Until today, traditional phlebotomy (venesection) has remained the standard treatment for hemochromatosis. It was introduced in 1950 and has been applied ever since to improve the survival of HH patients.72 Phlebotomy removes large amounts of iron localized in red blood cells, which stimulates erythropoiesis and mobilizes iron stored in peripheral tissues. It is a relatively simple and inexpensive treatment but has limited effects on specific HH-associated pathologies, including musculoskeletal symptoms.73 Whether unaffected mutation carriers should undergo phlebotomy has been highly debated, especially because evidence for a beneficial effect of phlebotomy in these individuals has not been documented.20 More recent insights into the molecular mechanisms maintaining iron homeostasis led to the development of alternative therapies, including the use of iron chelators, (eg. the most recent available oral chelator deferasirox, which effectively removes excess iron in HFE-HH patients74), and minihepcidin,75 which might be applicable for patients with HH in the future.76

Prevalence and penetrance of HFE-HH The HFE mutation C282Y occurs in approximately 6% of Caucasians77 and thus represents the most common genetic variant among this population. The frequency of the C282Y allele decreases from Northern to Southern Europe suggesting that it initially occurred as a ‘master’ mutation in the Neolithic Age in the Celtic population and then spread throughout the rest of Europe later on.78 Immigration from Europe to the US and Australia led to the widespread distribution of HH among Caucasian adults living in those geographic areas.79 Despite the high prevalence of the C282Y allele, numerous population screening studies provided evidence that the mutation clinically manifests in hemochromatosis in only a subgroup of carriers (Table 2). Differences in study design, inclusion criteria, and the definition of disease penetrance make it difficult to confirm an exact number for 812

Table 1. Symptoms of hereditary hemochromatosis. Neurological ataxia (lack of voluntary co-ordination of muscle movements) depression impaired memory lethargy, chronic fatigue weakness Gastrointestinal abdominal pain cirrhosis hepatocellular carcinoma hepatomegaly (enlarged liver) Musculoskeletal arthralgia (joint pain) arthritis chondrocalcinosis Dermatological hyperpigmentation loss of body hair Endocrinal diabetes gynecomastia (non-cancerous increase in the size of male breast tissue) hypogonadism (diminished functions of gonads) impotence testicular atrophy (diminished size of testes) Cardiovascular cardiomyopathy ("heart muscle disease") heart failure adrenal insufficiency

the overall penetrance of the C282Y mutation. A metaanalysis taking into account the data from 16 independent studies reported a penetrance of C282Y homozygosity of 14%.24 It was shown that homozygous men face a much higher risk of developing an iron overload phenotype than women,61 which may be explained by the recurrent physiological blood loss in women or by the frequency disparity of certain HLA haplotypes that occurs between male and female patients.80 The penetrance of the heterozygous C282Y mutation is even lower, with only approximately 3% of mutation carriers showing disease symptoms.3 Despite the fact that HFE mutations are not always pathogenic per se, but rather represent polymorphisms that predispose to iron overload, murine disease models with Hfe-deficiency81 or engineered mutations corresponding to the human C282Y mutation show an iron overload phenotype. Like in humans, the genetic background of the mouse strain strongly affects the severity of tissue iron accumulation.82 The disease manifestation of other HFE mutations, such as H63D in combination with C282Y, is negligible41 (Table 2). In rare cases, H63D mutations can lead to phenotypic expression of HH when additional risk factors such as heavy alcohol consumption or hepatitis virus infections are present.83 The discrepancy between the frequency of HFE mutations and their phenotypic expression indicates that additional environmental or genetic factors contribute to the manifestation of HH in affected individuals. Despite significant efforts to determine these parameters, only a few risk factors have been identified, including alcohol abuse, being heavily overweight, liver disease, or viral infections.57 The consumption of more than 60 g of alcohol per haematologica | 2017; 102(5)


Pathophysiological consequences and benefits of HFE mutations

Figure 2. The positive and negative effects of HFE mutations. Hereditary hemochromatosis (HH)-associated mutations affect the health of their carriers in various ways. Besides provoking classical HH disease symptoms (left), these mutations can provide benefits for affected individuals (right). Some of the benefits associated with HH have remained speculative to date and need further experimental validation (marked with a question mark).

day leads to higher serum iron and ferritin concentrations, contributing to the progression of cirrhosis in C282Y homozygotes.84 Certain dietary habits, such as heme iron intake and meat consumption, are also associated with increased iron loading.85 In addition, recent genome-wide association studies have identified SNPs in so-called ‘modifier’ genes that are associated with iron-related phenotypes. These include polymorphisms in genes with known roles in iron metabolism, such as those encoding transferrin receptor 1 (TFRC) and ferroportin (FPN/SCL40A1),86 but also in novel genomic loci, eg. in transmembrane protease, serine 6 (TMPRSS6)87 and in the vicinity of the genes FADS2, NAT2, ABO and TEX14.86 SNPs in transferrin (TF),86,88 transferrin receptor 2 (TFR2),86 bone morphogenetic protein 2 (BMP2),89 aryl hydrocarbon receptor nuclear translocator-like protein 1 (ARNTL),20 glyceronephosphate O-acyltransferase (GNPAT),90 and cytochrome B reductase 1 (CYBRD1)91 were further found to influence iron-related parameters in C282Y homozygotes. In addition, a large cohort study reported an association between common variants in the genes coding for bone morphogenetic protein 4 (BMP4) and hemojuvelin (HJV) and increased serum ferritin levels in homozygous C282Y mutation carriers.89 Moreover, a polymorphism in the PCSK7 (proprotein convertase subtilisin/kexin type 7) gene has been linked with liver cirrhosis and advanced fibrosis in C282Y homozygotes.92 These variants are extremely rare and may thus determine disease development on an individual basis. Although carriers of heterozygous HFE mutations mostly haematologica | 2017; 102(5)

lack a hemochromatosis phenotype, enhanced iron loading and manifestation of HH may occur in persons with heterozygous HFE mutations and additional mutations in the hepcidin (HAMP)93,94 HJV,93,95 or TFR296 genes. Additional HH modifiers, such as haptoglobin (HP)97 and ceruloplasmin (CP) have been identified in mice. Hfe knock-out mice (Hfe -/-) with a heterzogous R435X nonsense mutation in the CP gene and Hfe(+/-) mice with a homozygous R435X nonsense mutation showed lower liver iron levels compared to single mutant animals, revealing a protective effect of this specific ceruloplasmin variant in mice.98 In line with this, ceruloplasmin levels were found to be lower in HH patients when compared to control subjects.99

Do HFE mutations offer any advantage to asymptomatic patients? The observation that C282Y HFE mutations are frequent but only cause a disease-related phenotype in a subgroup of carriers led to the hypothesis that this HFE gene variant may be of an environmental or genetic advantage to asymptomatic carriers and that this is the reason for which it has been inherited with such a high frequency. Increased iron uptake, a hallmark of HH, may have helped humankind, and especially women of reproductive age, to better cope with the iron-reduced cereal grain-based diet which replaced the paleo diet rich in red meat in Europe in the Neolithic Age, at the time when the first HFE mutation occurred.100 Over centuries, the mutation may have 813


I. Hollerer et al. Table 2. Prevalence and penetrance of HFE (hemachromatosis) gene mutations.

C282Y/+

C282Y/ C282Y

H63D/+

H63D/ H63D

S65C/+

S65C/ S65C

H63D/ C282Y

S65C/ C282Y

Prevalence

10%23

0.4%24

~21%23

~1%23

~3%129,42

NA

~1%23

~0.3%129,42

Penetrance

3%3

controversial: 14% (in meta analysis)24

n.s.41

n.s.41

n.s.42

NA

n.s.44

~0.6%42

NA: not available; n.s.: not significant.

spread throughout the Caucasian population because it actively promoted the health of its carriers. Recent studies have provided evidence that the HFE C282Y variant as well as the H63D and S65C variants can positively influence the immune system, the general fitness and reproductive status of mutation carriers, and might even diminish the risk of developing diseases such as amyotrophic lateral sclerosis (ALS), Alzheimer’s disease (AD), Parkinson’s disease, and atherosclerosis (Figure 2). Micro-organisms heavily depend on the availability of iron for their proliferation. Iron levels in macrophages of carriers of the C282Y allele are reduced, which makes them less susceptible to those bacteria that depend on macrophage iron, such as Mycobacterium tuberculosis,101 Leishmania amazonensis,102 and Chlamydia or Legionella species.103 In line with this, proliferation of Salmonella Typhimurium in macrophages of mice with hetero- and homozygotic Hfe-deficiencies is strongly attenuated.104 In addition, in a mouse model with an HFE H63D mutation, increased survival over wild-type mice was observed following an infection with Plasmodium falciparum, a parasite that causes cerebral malaria.105 Besides being critical for micro-organismal growth, intramacrophage iron levels influence the inflammatory state of macrophages. While reduced iron levels attenuate the inflammatory response of macrophages to Salmonella infection or LPSstimulation,106 increased iron levels trigger a phenotypic switch of macrophages towards a pro-inflammatory state.107 HFE mutations may thus interfere with the inflammatory response of macrophages by reducing iron levels in macrophages in a hepcidin-dependent manner and protect carriers from infectious diseases. This may provide an explanation of why HFE mutations were inherited with such a high frequency. In addition to influencing the host immune system, mutations in the HFE gene have also been linked to the increased fitness of affected individuals. Intravenous and oral iron supplementation can treat fatigue in non-anemic women with low iron stores,108,109 showing that elevated iron levels positively influence fitness. A recent study demonstrated that 80% of successful French athletes carry a heterozygous HFE mutation (C282Y, H63D or S65C) suggesting that the resulting enhanced iron supply during physical activity accounts for the superior physical performance of these sportsmen.110 Large-scale studies among the Sicilian population further showed that HFE C282Y heterozygous individuals, particularly women, have a significantly increased life expectancy compared to controls.111,112 A trend towards an extended life span has also been observed in Sardinian women carrying the H63D mutation.113 Treated C282Y homozygotes with serum ferritin concentrations less than 1000 mg/L further exhibit a lower mortality related to cardiovascular events and extra814

hepatic cancers when compared to the general population.114 Interestingly, people diagnosed with HFEHH and verified iron overload are also taller on average, potentially because augmented iron absorption has a beneficial effect on growth.115 Unlike the carriers of homozygous HFE C282Y mutations, who may suffer from hypogonadism, heterozygous individuals may also have a reproductive advantage. This is supported by a study that reported higher levels of the sex hormone-binding globulin in the blood of C282Y heterozygous men.116 Women of reproductive age carrying a heterozygous C282Y mutation were found to suffer less commonly from low serum ferritin concentrations compared to control subjects, which might positively affect their fertility.117 HFE C282Y mutations have been associated with the attenuation of various disease states. For example, a lower incidence rate of atherosclerosis in HFE C282Y carriers has been reported. Possible explanations include lower serum cholesterol and low-density lipoprotein cholesterol levels,118,119 or iron deficiencies of macrophages that may contribute to a diminished inflammatory response106 and a decreased tendency to form atherosclerotic plaques.120 However, other studies failed to establish a link between hemochromatosis and atherosclerosis.121-124 Screening studies and corresponding meta-analyses further associated mutations in the HFE gene with a decreased risk of developing neurodegenerative diseases. So far, experimental validations for these findings are lacking and they are still considered highly controversial. Screening among patients with AD detected an association with the C282Y allele,125 a result in line with data from a meta-analysis that revealed a correlation between AD and the C282Y but not the H63D polymorphism.126 Similarly, two different meta-analyses reported an association of the C282Y mutation and sporadic ALS127 or Parkinson’s disease,128 both excluding a role for the H63D polymorphism in disease manifestation. Taken together, several studies suggest that HFE mutation carriers might benefit from mutation-associated increased iron levels as long as the iron overload phenotype is mild. This demonstrates that we clearly need to distinguish between hemochromatosis 'patients' who suffer from the pathophysiological complications associated with iron overload and individuals who carry HHassociated mutations without showing any disease symptoms. In fact, it is possible that the beneficial effects of the mutations observed in heterozygotes might also manifest in homozygote individuals during growth and early adulthood before significant organ iron overload develops. Future clinical and biochemical studies may reveal more detailed information about the impact of HFE mutations on the health of their carriers and may provide an haematologica | 2017; 102(5)


Pathophysiological consequences and benefits of HFE mutations

answer to the long-standing question as to why HFE mutations have so far been inherited with such a high frequency. Acknowledgments The authors would like to thank Prof. Peter Nielsen, Prof.

References 1. Muckenthaler MU, Rivella S, Hentze MW, Galy B. A Red Carpet for Iron Metabolism. Cell. 2017;168(3):344-361. 2. Yun S, Vincelette ND. Update on iron metabolism and molecular perspective of common genetic and acquired disorder, hemochromatosis. Crit Rev Oncol Hematol. 2015;95(1):12-25. 3. Gallego CJ, Burt A, Sundaresan AS, et al. Penetrance of Hemochromatosis in HFE Genotypes Resulting in p.Cys282Tyr and p.[Cys282Tyr];[His63Asp] in the eMERGE Network. Am J Hum Genet. 2015;97(4):512520. 4. Barton JC. Hemochromatosis and iron overload: from bench to clinic. Am J Med Sci. 2013;346(5):403-412. 5. Trousseau A. Glycosurie; diabete sucre. Clinique Med de l'Hotel de Paris. 1865;2: 663-698. 6. Hanot V, Chauffard A. [Cirrhose hypertrophique pigmentaire dans le diabète sucré]. Rev Med. 1882;2:385-403. 7. Troisier M. [Diabète sucré]. Bull Soc Anatomique Paris. 1871;44:231-235. 8. von Recklinghausen FD. [Uber Haemochromatose. Taggeblatt der 62 Versammlung deutscher Naturforscher and Aerzte in Heidelberg]. 1889:324-325. 9. Sheldon JH. Haemochromatosis London: Oxford University Press, 1935. 10. Simon M, Pawlotsky Y, Bourel M, Fauchet R, Genetet B. [Idiopathic hemochromatosis associated with HL-A 3 tissular antigen]. Nouv Presse Med. 1975;4(19):1432. 11. Laufberger V. [Sur la cristallisation de la ferritine]. Soc Chim Biol. 1937. 12. Schade AL, Caroline L. An iron-binding component in human blood plasma. Science. 1946;104(2702):340. 13. 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. 14. 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. 15. Roetto A, Papanikolaou G, Politou M, et al. Mutant antimicrobial peptide hepcidin is associated with severe juvenile hemochromatosis. Nat Genet. 2003;33(1):21-22. 16. 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. 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. Njajou OT, Vaessen N, Joosse M, et al. A mutation in SLC11A3 is associated with autosomal dominant hemochromatosis. Nat

haematologica | 2017; 102(5)

Pierre Brissot and Prof. Graça Porto for their critical reading of the manuscript. MUM acknowledges funding from the Deutsche Forschungsgemeinschaft (SFB1118 and SFB1036) and the Dietmar Hopp Stiftung. We would like to apologize to all colleagues whose work could not be cited because of space constraints.

Genet. 2001;28(3):213-214. 19. Seckington R, Powell L. HFE-Associated Hereditary Hemochromatosis. In: Pagon RA, Adam MP, Ardinger HH, Wallace SE, Amemiya A, Bean LJH, et al., eds. GeneReviews(R). Seattle (WA), 1993. 20. Porto G, Brissot P, Swinkels DW, et al. EMQN best practice guidelines for the molecular genetic diagnosis of hereditary hemochromatosis (HH). Eur J Hum Genet. 2016;24(4):479-495. 21. Garbowski MW, Ma YM, Fucharoen S, Srichairatanakool S, Hider R, Porter JB. Clinical and methodological factors affecting non-transferrin-bound iron values using a novel fluorescent bead assay. Transl Res. 2016;177:19-30. 22. Brissot P, Ropert M, Le Lan C, Loreal O. Non-transferrin bound iron: a key role in iron overload and iron toxicity. Biochim Biophys Acta. 2012;1820(3):403-410. 23. Adams PC, Reboussin DM, Barton JC, et al. Hemochromatosis and iron-overload screening in a racially diverse population. N Engl J Med. 2005;352(17):1769-1778. 24. European Association For The Study Of The L. EASL clinical practice guidelines for HFE hemochromatosis. J Hepatol. 2010;53(1):322. 25. Feder JN, Tsuchihashi Z, Irrinki A, et al. The hemochromatosis founder mutation in HLA-H disrupts beta2-microglobulin interaction and cell surface expression. J Biol Chem. 1997;272(22):14025-14028. 26. Waheed A, Parkkila S, Zhou XY, et al. Hereditary hemochromatosis: effects of C282Y and H63D mutations on association with beta2-microglobulin, intracellular processing, and cell surface expression of the HFE protein in COS-7 cells. Proc Natl Acad Sci USA. 1997;94(23):12384-12389. 27. Giannetti AM, Bjorkman PJ. HFE and transferrin directly compete for transferrin receptor in solution and at the cell surface. J Biol Chem. 2004;279(24):25866-25875. 28. West AP Jr, Giannetti AM, Herr AB, et al. Mutational analysis of the transferrin receptor reveals overlapping HFE and transferrin binding sites. J Mol Biol. 2001;313(2):385397. 29. Goswami T, Andrews NC. Hereditary hemochromatosis protein, HFE, interaction with transferrin receptor 2 suggests a molecular mechanism for mammalian iron sensing. J Biol Chem. 2006;281(39):28494-28498. 30. Andriopoulos B Jr, Corradini E, Xia Y, et al. BMP6 is a key endogenous regulator of hepcidin expression and iron metabolism. Nat Genet. 2009;41(4):482-487. 31. Meynard D, Kautz L, Darnaud V, CanonneHergaux F, Coppin H, Roth MP. Lack of the bone morphogenetic protein BMP6 induces massive iron overload. Nat Genet. 2009;41(4):478-481. 32. Daher R, Kannengiesser C, Houamel D, et al. Heterozygous Mutations in BMP6 Propeptide Lead to Inappropriate Hepcidin Synthesis and Moderate Iron Overload in Humans. Gastroenterology. 2016;

150(3):672-683 e4. 33. Babitt JL, Huang FW, Wrighting DM, et al. Bone morphogenetic protein signaling by hemojuvelin regulates hepcidin expression. Nat Genet. 2006;38(5):531-539. 34. Steinbicker AU, Bartnikas TB, Lohmeyer LK, et al. Perturbation of hepcidin expression by BMP type I receptor deletion induces iron overload in mice. Blood. 2011;118(15):42244230. 35. Mayeur C, Leyton PA, Kolodziej SA, Yu B, Bloch KD. BMP type II receptors have redundant roles in the regulation of hepatic hepcidin gene expression and iron metabolism. Blood. 2014;124(13):2116-2123. 36. Wang RH, Li C, Xu X, et al. A role of SMAD4 in iron metabolism through the positive regulation of hepcidin expression. Cell Metab. 2005;2(6):399-409. 37. Latour C, Besson-Fournier C, Meynard D, et al. Differing impact of the deletion of hemochromatosis-associated molecules HFE and transferrin receptor-2 on the iron phenotype of mice lacking bone morphogenetic protein 6 or hemojuvelin. Hepatology. 2016;63(1):126-137. 38. Wu XG, Wang Y, Wu Q, et al. HFE interacts with the BMP type I receptor ALK3 to regulate hepcidin expression. Blood. 2014;124(8):1335-1343. 39. D'Alessio F, Hentze MW, Muckenthaler MU. The hemochromatosis proteins HFE, TfR2, and HJV form a membrane-associated protein complex for hepcidin regulation. J Hepatol. 2012;57(5):1052-1060. 40. Ryan JD, Ryan E, Fabre A, Lawless MW, Crowe J. Defective bone morphogenic protein signaling underlies hepcidin deficiency in HFE hereditary hemochromatosis. Hepatology. 2010;52(4):1266-1273. 41. Gochee PA, Powell LW, Cullen DJ, Du Sart D, Rossi E, Olynyk JK. A population-based study of the biochemical and clinical expression of the H63D hemochromatosis mutation. Gastroenterology. 2002;122(3):646651. 42. Holmstrom P, Marmur J, Eggertsen G, Gafvels M, Stal P. Mild iron overload in patients carrying the HFE S65C gene mutation: a retrospective study in patients with suspected iron overload and healthy controls. Gut. 2002;51(5):723-730. 43. Ramakrishna R, Gupta S, Sarathy K, Bowen A. Phenotypic and clinical manifestations of compound heterozygous genetic haemochromatosis (CHGH): a non-invasive approach to clinical management. Intern Med J. 2013;43(3):254-261. 44. Gurrin LC, Bertalli NA, Dalton GW, et al. HFE C282Y/H63D Compound Heterozygotes Are at Low Risk of Hemochromatosis-Related Morbidity. Hepatology. 2009;50(1):94-101. 45. Walsh A, Dixon JL, Ramm GA, et al. The clinical relevance of compound heterozygosity for the C282Y and H63D substitutions in hemochromatosis. Clin Gastroenterol Hepatol. 2006;4(11):14031410.

815


I. Hollerer et al. 46. Vecchi C, Montosi G, Pietrangelo A. Huh-7: a human "hemochromatotic" cell line. Hepatology. 2010;51(2):654-659. 47. Takano A, Niimi H, Atarashi Y, et al. A novel Y231del mutation of HFE in hereditary haemochromatosis provides in vivo evidence that the Huh-7 is a human haemochromatotic cell line. Liver Int. 2011;31(10):1593-1597. 48. Le Gac G, Gourlaouen I, Ronsin C, et al. Homozygous deletion of HFE produces a phenotype similar to the HFE p.C282Y/p.C282Y genotype. Blood. 2008;112(13):5238-5240. 49. Pelucchi S, Mariani R, Bertola F, Arosio C, Piperno A. Homozygous deletion of HFE: the Sardinian hemochromatosis? Blood. 2009;113(16):3886. 50. Cezard C, Rabbind Singh A, Le Gac G, Gourlaouen I, Ferec C, Rochette J. Phenotypic expression of a novel C282Y/R226G compound heterozygous state in HFE hemochromatosis: molecular dynamics and biochemical studies. Blood Cells Mol Dis. 2014;52(1):27-34. 51. Piperno A, Arosio C, Fossati L, et al. Two novel nonsense mutations of HFE gene in five unrelated italian patients with hemochromatosis. Gastroenterology. 2000; 119(2):441-445. 52. Wallace DF, Subramaniam VN. Non-HFE haemochromatosis. World J Gastroenterol. 2007;13(35):4690-4698. 53. Santos PC, Dinardo CL, Cancado RD, Schettert IT, Krieger JE, Pereira AC. NonHFE hemochromatosis. Rev Bras Hematol Hemoter. 2012;34(4):311-316. 54. Roetto A, Totaro A, Piperno A, et al. New mutations inactivating transferrin receptor 2 in hemochromatosis type 3. Blood. 2001;97(9):2555-2560. 55. Detivaud L, Island ML, Jouanolle AM, et al. Ferroportin Diseases: Functional Studies, a Link Between Genetic and Clinical Phenotype. Hum Mutat. 2013;34(11):15291536. 56. 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. 57. Pietrangelo A. Genetics, Genetic Testing, and Management of Hemochromatosis: 15 Years Since Hepcidin. Gastroenterology. 2015;149(5):1240-1251.e4. 58. Coppin H, Bensaid M, Fruchon S, Borot N, Blanche H, Roth MP. Longevity and carrying the C282Y mutation for haemochromatosis on the HFE gene: case control study of 492 French centenarians. BMJ. 2003;327 (7407):132-133. 59. Allen KJ, Bertalli NA, Osborne NJ, et al. HFE Cys282Tyr Homozygotes With Serum Ferritin Concentrations Below 1000 mu g/L Are at Low Risk of Hemochromatosis. Hepatology. 2010;52(3):925-933. 60. Bacon BR, Adams PC, Kowdley KV, Powell LW, Tavill AS. Diagnosis and Management of Hemochromatosis: 2011 Practice Guideline by the American Association for the Study of Liver Diseases. Hepatology. 2011;54(1):328-343. 61. Allen KJ, Gurrin LC, Constantine CC, et al. Iron-overload-related disease in HFE hereditary hemochromatosis. N Engl J Med. 2008;358(3):221-230. 62. Fracanzani AL, Conte D, Fraquelli M, et al. Increased cancer risk in a cohort of 230 patients with hereditary hemochromatosis in comparison to matched control patients with non-iron-related chronic liver disease. Hepatology. 2001;33(3):647-651.

816

63. Gannon PO, Medelci S, Le Page C, et al. Impact of hemochromatosis gene (HFE) mutations on epithelial ovarian cancer risk and prognosis. Int J Cancer. 2011;128(10):2326-2334. 64. Osborne NJ, Gurrin LC, Allen KJ, et al. HFE C282Y homozygotes are at increased risk of breast and colorectal cancer. Hepatology. 2010;51(4):1311-1318. 65. Shaheen NJ, Silverman LM, Keku T, et al. Association between hemochromatosis (HFE) gene mutation carrier status and the risk of colon cancer. J Natl Cancer Inst. 2003;95(2):154-159. 66. Syrjakoski K, Fredriksson H, Ikonen T, et al. Hemochromatosis gene mutations among Finnish male breast and prostate cancer patients. Int J Cancer. 2006;118(2):518-520. 67. Ashrafian H. Hepcidin: the missing link between hemochromatosis and infections. Infect Immun. 2003;71(12):6693-6700. 68. Khan FA, Fisher MA, Khakoo RA. Association of hemochromatosis with infectious diseases: expanding spectrum. Int J Infect Dis. 2007;11(6):482-487. 69. Arezes J, Jung G, Gabayan V, et al. Hepcidininduced hypoferremia is a critical host defense mechanism against the siderophilic bacterium Vibrio vulnificus. Cell Host Microbe. 2015;17(1):47-57. 70. Wang HZ, He YX, Yang CJ, Zhou W, Zou CG. Hepcidin Is Regulated during BloodStage Malaria and Plays a Protective Role in Malaria Infection. J Immunol. 2011; 187(12):6410-6416. 71. Benesova K, Vujic Spasic M, Schaefer SM, et al. Hfe deficiency impairs pulmonary neutrophil recruitment in response to inflammation. PloS One. 2012;7(6):e39363. 72. Kanwar P, Kowdley KV. Diagnosis and treatment of hereditary hemochromatosis: an update. Expert Rev Gastroenterol Hepatol. 2013;7(6):517-530. 73. Brissot P. Optimizing the diagnosis and the treatment of iron overload diseases. Expert Rev Gastroenterol Hepatol. 2016;10(3):359370. 74. Phatak P, Brissot P, Wurster M, et al. A phase 1/2, dose-escalation trial of deferasirox for the treatment of iron overload in HFE-related hereditary hemochromatosis. Hepatology. 2010;52(5):1671-1779. 75. Ramos E, Ruchala P, Goodnough JB, et al. Minihepcidins prevent iron overload in a hepcidin-deficient mouse model of severe hemochromatosis. Blood. 2012;120(18): 3829-3836. 76. Powell LW, Seckington RC, Deugnier Y. Haemochromatosis. Lancet. 2016;388 (10045):706-716. 77. Liver EAFTSOT. EASL clinical practice guidelines for HFE hemochromatosis. J Hepatol. 2010;53(1):3-22. 78. Distante S, Robson KJ, Graham-Campbell J, Arnaiz-Villena A, Brissot P, Worwood M. The origin and spread of the HFE-C282Y haemochromatosis mutation. Hum Genet. 2004;115(4):269-279. 79. Merryweather-Clarke AT, Pointon JJ, Jouanolle AM, Rochette J, Robson KJ. Geography of HFE C282Y and H63D mutations. Genet Test. 2000;4(2):183-198. 80. Barton JC, Wiener HW, Acton RT, Go RC. HLA haplotype A*03-B*07 in hemochromatosis probands with HFE C282Y homozygosity: frequency disparity in men and women and lack of association with severity of iron overload. Blood Cells Mol Dis. 2005;34(1):38-47. 81. Muckenthaler M, Roy CN, Custodio AO, et al. Regulatory defects in liver and intestine implicate abnormal hepcidin and Cybrd1

82.

83. 84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

98.

expression in mouse hemochromatosis. Nat Genet. 2003;34(1):102-107. Levy JE, Montross LK, Andrews NC. Genes that modify the hemochromatosis phenotype in mice. J Clin Invest. 2000;105(9):12091216. Piperno A, Sampietro M, Pietrangelo A, et al. Heterogeneity of hemochromatosis in Italy. Gastroenterology. 1998;114(5):996-1002. Fletcher LM, Dixon JL, Purdie DM, Powell LW, Crawford DHG. Excess alcohol greatly increases the prevalence of cirrhosis in hereditary hemochromatosis. Gastroenterology. 2002;122(2):281-289. Cade JE, Moreton JA, O'Hara B, et al. Diet and genetic factors associated with iron status in middle-aged women. Am J Clin Nutr. 2005;82(4):813-820. Benyamin B, Esko T, Ried JS, et al. Novel loci affecting iron homeostasis and their effects in individuals at risk for hemochromatosis. Nat Commun. 2015;6:6542. Benyamin B, Ferreira MAR, Willemsen G, et al. Common variants in TMPRSS6 are associated with iron status and erythrocyte volume. Nat Genet. 2009;41(11):1173-1175. de Tayrac M, Roth MP, Jouanolle AM, et al. Genome-wide association study identifies TF as a significant modifier gene of iron metabolism in HFE hemochromatosis. J Hepatol. 2015;62(3):664-672. Milet J, Dehais V, Bourgain C, et al. Common variants in the BMP2, BMP4, and HJV genes of the hepcidin regulation pathway modulate HFE hemochromatosis penetrance. Am J Hum Genet. 2007;81(4):799807. McLaren CE, Emond MJ, Subramaniam VN, et al. Exome sequencing in HFE C282Y homozygous men with extreme phenotypes identifies a GNPAT variant associated with severe iron overload. Hepatology. 2015;62(2):429-439. Pelucchi S, Mariani R, Calza S, et al. CYBRD1 as a modifier gene that modulates iron phenotype in HFE p.C282Y homozygous patients. Haematol-Hematol J. 2012;97(12):1818-1825. Stickel F, Buch S, Zoller H, et al. Evaluation of genome-wide loci of iron metabolism in hereditary hemochromatosis identifies PCSK7 as a host risk factor of liver cirrhosis. Hum Mol Genet. 2014;23(14):3883-3890. Barton JC, LaFreniere SA, Leiendecker-Foster C, et al. HFE, SLC40A1, HAMP, HJV, TFR2, and FTL mutations detected by denaturing high-performance liquid chromatography after iron phenotyping and HFE C282Y and H63D genotyping in 785 HEIRS Study participants. Am J Hematol. 2009;84(11):710714. Merryweather-Clarke AT, Cadet E, Bomford A, et al. Digenic inheritance of mutations in HAMP and HFE results in different types of haemochromatosis. Blood. 2003;102(11): 758a-758a. Pietrangelo A, Caleffi A, Henrion J, et al. Juvenile hemochromatosis associated with pathogenic mutations of adult hemochromatosis genes. Gastroenterology. 2005; 128(2):470-479. Biasiotto G, Belloli S, Ruggeri G, et al. Identification of new mutations of the HFE, hepcidin, and transferrin receptor 2 genes by denaturing HPLC analysis of individuals with biochemical indications of iron overload. Clin Chem. 2003;49(12):1981-1988. Tolosano E, Fagoonee S, Garuti C, et al. Haptoglobin modifies the hemochromatosis phenotype in mice. Blood. 2005;105(8): 3353-3355. Gouya L, Muzeau F, Robreau AM, et al.

haematologica | 2017; 102(5)


Pathophysiological consequences and benefits of HFE mutations Genetic study of variation in normal mouse iron homeostasis reveals ceruloplasmin as an HFE-hemochromatosis modifier gene. Gastroenterology. 2007;132(2):679-686. 99. Cairo G, Conte D, Bianchi L, Fraquelli M, Recalcati S. Reduced serum ceruloplasmin levels in hereditary haemochromatosis. Br J Haematol. 2001;114(1):226-229. 100. Naugler C. Hemochromatosis: a Neolithic adaptation to cereal grain diets. Med Hypotheses. 2008;70(3):691-692. 101. Olakanmi O, Schlesinger LS, Britigan BE. Hereditary hemochromatosis results in decreased iron acquisition and growth by Mycobacterium tuberculosis within human macrophages. J Leukoc Biol. 2007;81(1):195204. 102. Ben-Othman R, Flannery AR, Miguel DC, Ward DM, Kaplan J, Andrews NW. Leishmania-Mediated Inhibition of Iron Export Promotes Parasite Replication in Macrophages. Plos Pathog. 2014;10(1): e1003901. 103. Paradkar PN, De Domenico I, Durchfort N, Zohn I, Kaplan J, Ward DM. Iron depletion limits intracellular bacterial growth in macrophages. Blood. 2008;112(3):866-874. 104. Nairz M, Theurl I, Ludwiczek S, et al. The co-ordinated regulation of iron homeostasis in murine macrophages limits the availability of iron for intracellular Salmonella typhimurium. Cell Microbiol. 2007; 9(9):2126-2140. 105. Leitner DF, Stoute JA, Landmesser M, Neely E, Connor JR. The HFE genotype and a formulated diet controlling for iron status attenuate experimental cerebral malaria in mice. Int J Parasitol. 2015;45(12):797-808. 106. Wang L, Johnson EE, Shi HN, Walker WA, Wessling-Resnick M, Cherayil BJ. Attenuated inflammatory responses in hemochromatosis reveal a role for iron in the regulation of macrophage cytokine translation. J Immunol. 2008;181(4):27232731. 107. Vinchi F, Costa da Silva M, Ingoglia G, et al. Hemopexin therapy reverts heme-induced proinflammatory phenotypic switching of macrophages in a mouse model of sickle cell disease. Blood. 2016;127(4):473-486. 108. Krayenbuehl PA, Battegay E, Breymann C, Furrer J, Schulthess G. Intravenous iron for

haematologica | 2017; 102(5)

the treatment of fatigue in nonanemic, premenopausal women with low serum ferritin concentration. Blood. 2011;118(12):32223227. 109. Vaucher P, Druais PL, Waldvogel S, Favrat B. Effect of iron supplementation on fatigue in nonanemic menstruating women with low ferritin: a randomized controlled trial. CMAJ. 2012;184(11):1247-1254. 110. Hermine O, Dine G, Genty V, et al. Eighty percent of French sport winners in Olympic, World and Europeans competitions have mutations in the hemochromatosis HFE gene. Biochimie. 2015;119:1-5. 111. Lio D, Balistreri CR, Colonna-Romano G, et al. Association between the MHC class I gene HFE polymorphisms and longevity: a study in Sicilian population. Genes Immun. 2002;3(1):20-24. 112. Balistreri CR, Candore G, Almasio P, et al. Analysis of hemochromatosis gene mutations in the Sicilian population: implications for survival and longevity. Arch Gerontol Geriatr Suppl. 2002;8:35-42. 113. Carru C, Pes GM, Deiana L, et al. Association between the HFE mutations and longevity: a study in Sardinian population. Mech Ageing Dev. 2003;124(4):529-532. 114. Bardou-Jacquet E, Morcet J, Manet G, et al. Decreased cardiovascular and extrahepatic cancer-related mortality in treated patients with mild HFE hemochromatosis. J Hepatol. 2015;62(3):682-689. 115. Cippa PE, Krayenbuehl PA. Increased height in HFE hemochromatosis. N Engl J Med. 2013;369(8):785-786. 116. Yeap BB, Beilin J, Shi Z, et al. The C282Y polymorphism of the hereditary hemochromatosis gene is associated with increased sex hormone-binding globulin and normal testosterone levels in men. J Endocrinol Invest. 2010;33(8):544-548. 117. Bulaj ZJ, Griffen LM, Jorde LB, Edwards CQ, Kushner JP. Clinical and biochemical abnormalities in people heterozygous for hemochromatosis. N Engl J Med. 1996;335(24):1799-1805. 118. Pankow JS, Boerwinkle E, Adams PC, et al. HFE C282Y homozygotes have reduced low-density lipoprotein cholesterol: the Atherosclerosis Risk in Communities (ARIC) Study. Transl Res. 2008;152(1):3-10.

119. Adams PC, Pankow J, Barton JC, et al. Hfe C282y Homozygosity Is Associated with Lower Total and Ldl Cholesterol: The Hemochromatosis and Iron Overload Screening (Heirs) Study. Hepatology. 2008; 48(4):437a-438a. 120. Vinchi F, Muckenthaler MU, Da Silva MC, Balla G, Balla J, Jeney V. Atherogenesis and iron: from epidemiology to cellular level. Front Pharmacol. 2014;5:94. 121. Miller M, Hutchins GM. Hemochromatosis, multiorgan hemosiderosis, and coronary artery disease. JAMA. 1994;272(3):231-233. 122. Sullivan JL, Zacharski LR. Hereditary haemochromatosis and the hypothesis that iron depletion protects against ischemic heart disease. Eur J Clin Invest. 2001;31(5):375-377. 123. Munoz-Bravo C, Gutierrez-Bedmar M, Gomez-Aracena J, Garcia-Rodriguez A, Navajas JF. Iron: protector or risk factor for cardiovascular disease? Still controversial. Nutrients. 2013;5(7):2384-2404. 124. Sullivan JL. Do Hemochromatosis Mutations Protect Against Iron-Mediated Atherogenesis? Circ-Cardiovasc Gene. 2009;2(6):652-657. 125. Correia AP, Pinto JP, Dias V, Mascarenhas C, Almeida S, Porto G. CAT53 and HFE alleles in Alzheimer's disease: a putative protective role of the C282Y HFE mutation. Neurosci Lett. 2009;457(3):129-132. 126. Lin M, Zhao L, Fan J, et al. Association between HFE polymorphisms and susceptibility to Alzheimer's disease: a meta-analysis of 22 studies including 4,365 cases and 8,652 controls. Mol Biol Rep. 2012;39(3):3089-3095. 127. Li M, Wang L, Wang W, Qi XL, Tang ZY. Mutations in the HFE gene and sporadic amyotrophic lateral sclerosis risk: a metaanalysis of observational studies. Braz J Med Biol Res. 2014;47(3):215-222. 128. Xia J, Xu H, Jiang H, Xie J. The association between the C282Y and H63D polymorphisms of HFE gene and the risk of Parkinson's disease: A meta-analysis. Neurosci Lett. 2015;595:99-103. 129. Mura C, Raguenes O, Scotet V, Jacolot S, Mercier AY, Ferec C. A 6-year survey of HFE gene test for hemochromatosis diagnosis. Genet Med. 2005;7(1):68-73.

817


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Hematopoiesis

Ferrata Storti Foundation

Bone marrow mesenchymal stromal cells induce nitric oxide synthase-dependent differentiation of CD11b+ cells that expedite hematopoietic recovery

Cristina Trento,1 Ilaria Marigo,2 Alice Pievani,1,3 Antonio Galleu,1 Luigi Dolcetti,1 Chun-Yin Wang,1 Marta Serafini,3 Vincenzo Bronte4 and Francesco Dazzi1,2

Haematologica 2017 Volume 102(5):818-825

Division of Cancer Studies, King’s College London, UK; 2Department of Haematology, Imperial College, London, UK; 3M.Tettamanti Research Center, Department of Pediatrics, University of Milano-Bicocca, Italy and 4Department of Medicine, Immunology Section, Verona University Hospital, Italy 1

ABSTRACT

B Correspondence: francesco.dazzi@kcl.ac.uk

Received: September 5, 2016. Accepted: February 1, 2017. Pre-published: February 9, 2017. doi:10.3324/haematol.2016.155390 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/818 Š2017 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.

818

one marrow microenvironment is fundamental for hematopoietic homeostasis. Numerous efforts have been made to reproduce or manipulate its activity to facilitate engraftment after hematopoietic stem cell transplantation but clinical results remain unconvincing. This probably reflects the complexity of the hematopoietic niche. Recent data have demonstrated the fundamental role of stromal and myeloid cells in regulating hematopoietic stem cell self-renewal and mobilization in the bone marrow. In this study we unveil a novel interaction by which bone marrow mesenchymal stromal cells induce the rapid differentiation of CD11b+ myeloid cells from bone marrow progenitors. Such an activity requires the expression of nitric oxide synthase-2. Importantly, the administration of these mesenchymal stromal cell-educated CD11b+ cells accelerates hematopoietic reconstitution in bone marrow transplant recipients. We conclude that the liaison between mesenchymal stromal cells and myeloid cells is fundamental in hematopoietic homeostasis and suggests that it can be harnessed in clinical transplantation.

Introduction Mesenchymal stromal cells (MSC) play a crucial role in tissue homeostasis whereby they control inflammation and regulate stem cell renewal and differentiation. Their immunomodulatory properties, which target both adaptive and innate immune responses, have been extensively documented in vitro and in vivo.1-7 Although the underlying mechanisms are only partially known, it is widely accepted that the combination of soluble factors and contact-dependent interactions plays a fundamental role. Upregulation of inducible nitric oxide synthase (iNOS or NOS2), one of the key MSC transcriptional responses resulting in the secretion of a short-lived molecule with potent immunomodulatory effects, nitric oxide (NO),8,9 is not sufficient in itself to explain MSC the immunosuppressive activities. In fact, one of the main MSC effector mechanisms is the recruitment and functional modulation of myeloid cells, such as inflammatory monocytes and tissue macrophages.6,10-14 MSC also contribute to the hematopoietic stem cell (HSC) niche in which they regulate hematopoietic cell number and differentiation.15,16 These properties have been harnessed therapeutically to promote hematopoietic regeneration. However, early in vivo animal studies have not been unequivocally confirmed by clinical investigations.18,19 Although the mechanisms by which MSC regulate HSC are still unknown, it is arguable that, resembling what has been described for their immunosuppressive actions, MSC require other cells to execute their functions.20 In particular, a few studies have described that the interaction between MSC and bone marrow (BM) macrophages contributes to the retention of HSC in the BM21 and prevents their haematologica | 2017; 102(5)


MSC induce regenerative myeloid cells

exhaustion.20-24 The nature of this interaction has not, however, been elucidated. In this work, we have tested the hypothesis that MSC may skew the differentiation and expansion of BM myeloid progenitors with the ability to accelerate hematopoietic reconstitution. We have observed that MSC selectively promote the expansion and differentiation of CD11b+ cells from the BM and that this function is largely dependent on NOS2. Ex-vivo generated MSCinduced CD11b+ cells exhibit the ability to accelerate hematopoietic engraftment and reconstitution.

Methods Cell cultures and media Murine BM MSC were generated from crushed femora and tibiae of wild type (WT) C57Bl/6 or Nos2-/- mice (for further information, see the Online Supplementary Appendix). Human BM MSC were generated from human BM mononuclear cells (MNC). All samples were collected after informed consent had been obtained in accordance with Ethics Committee approval from the Province of Monza-Brianza (Italy) (approval date: 16 Oct 2014, approval file name: BM-NICHE). Further details on the methods used to generate murine and human MSC, and on the generation of MSCinduced myeloid cells, are presented in the Online Supplementary Appendix.

LSRII or BD Fortessa (BD Biosciences, NJ, USA). Data were subsequently analyzed using FlowJo software (Oregon, USA). A complete list of antibodies used is given in the Online Supplementary Appendix.

In vivo experiments For the adoptive transfer of MSC, sublethally irradiated (split dose of 800 cGy) WT CD45.1 C57Bl/6 recipients were transplanted by tail vein injection with 2x106 BM cells and 0.2x106 WT or Nos2-/- MSC 4 h after the second dose of irradiation. Mice were sacrificed after 13 days and spleens and BM analyzed by FACS. For the adoptive transfer of CD11b+ cells, sublethally irradiated (split dose of 800 cGy) WT CD45.1 C57Bl/6 recipients were transplanted with 5x104 BM cells alone or with 2x106 CD45.2+ MSCinduced CD11b+ cells 4 h after the second dose of irradiation. Peripheral blood samples were taken every 2 weeks after the transplant up to 4 months. Peripheral blood was lysed with lysis buffer (150 mM NH4Cl, 1 mM KHCO3, 0.1 mM EDTA, pH 7.3), and cells were stained for flow cytometry.

Statistical analysis Data were analyzed using GraphPad Prism Software. An unpaired two-tailed Student t-test was run with a confidence interval of 95%, and expressed as mean ± standard error of the mean (SEM) or standard deviation (SD). For analysis of three groups of data, one-way analysis of variance with Bonferroni’s multiple comparison test was used, and expressed as mean ± SEM.

Flow cytometry Unspecific binding sites were blocked with phosphate-buffered saline supplemented with 1% fetal bovine serum and either Fc blocker (CD16/32, eBioscience) or whole mouse IgG (Sigma Aldrich) before cells were incubated with the respective monoclonal antibody at 4°C for 30 min. After incubation, cells were washed twice with phosphate-buffered saline and analyzed by flow cytometry with a BD FACSCalibur, BD FACS CantoII, BD

Results Mesenchymal stromal cells induce the expansion and differentiation of CD11b+ cells from bone marrow myeloid progenitors Unfractionated BM MNC were cultured in the presence

A

B

haematologica | 2017; 102(5)

C

Figure 1. Mesenchymal stromal cells induce the differentiation of bone marrow mononuclear cells into myeloid cells. BM MNC were cultured alone or with MSC (ratio 5:1) for 4 days. (A) Proportion of CD11b+Gr1high and CD11b+Gr-1low-neg cells in the live gate. A typical result of ten independently performed experiments is shown. (B) Percentage of Gr-1high and Gr-1low-neg cells in CD11b+ cells. Mean of ten independent experiments, ± SEM. *** P<0.001 **** P<0.0001 unpaired t test. (C) Absolute number of CD11b+ cells recovered from initial seeding from BM cultured alone (white bars) or with MSC (black bars) for 4 days. Mean of ten independent experiments, ± SEM **P<0.005, unpaired t test.

819


C. Trento et al.

or absence of MSC. After 4 days, cultures were analyzed for the expression of the myeloid markers CD11b and Gr125 (Figure 1A). Whilst in the control cultures the vast majority of BM cells consisted of CD11b+Gr-1high, in those with MSC there was a marked skew towards the formation of a large proportion of CD11b+Gr-1low-neg (50.6% ± 3.9%) (Figure 1B). Analysis of absolute cell numbers revealed that the presence of MSC could not only maintain the survival of total CD11b+ cells as compared to BM MNC cultured alone (5.7x106 ± 1.8x106 compared to 1.7x106 ±1.4x106) (Figure 1C), but also drive a selective retention and/or expansion of the Gr-1low-neg population. At morphological analysis the MSC-induced CD11b+ myeloid cells consisted of a fairly homogeneous population of large cells with reniform nuclei and abundant pale vacuolated cytoplasm with granules (Figure 2A). The immunophenotype of CD11b+ sorted cells revealed a 6fold increase in F4/80+ (36.5%±10.3%), a 3-fold increase in IL4Rα+ (18.2%±7.5%), and a 2-fold increase in CD169+ (2.3%±0.6%) cells when compared to BM MNC cultured alone (Figure 2B, left panel). BM MNC cultured with MSC also expressed CD115 (48.6%±12.4%), CD206 (20.6%±2%) and CD68 (16.5%±4.9%) (Figure 2B, left panel). These macrophage markers were expressed only in the Gr-1low-neg subset (Figure 2B, right panel), whilst CD115 was detected both in the Gr-1high and the Gr-1low-neg subsets. To understand the target cells of MSC-induced myeloid differentiation, FACS-sorted HSC, common myeloid progenitors (CMP) or granulocyte/macrophage progenitors (GMP) were cultured with MSC. Megakaryocyte/erythroid progenitors (MEP) and unfractionated BM MNC were used as negative or positive control of differentiation, respectively. MSC induced the differentiation of only CMP and GMP into CD11b+Gr-1high and CD11b+Gr-1low-neg cells, with no effect on HSC or MEP (Figure 3A). The proportion of Gr-1low-neg cells from CMP cultures was higher than in the cultures with unfractionated BM (Gr-1low-neg: 60.1% ± 8.9% versus 35% ± 12.8% in unfractionated BM+MSC) (Figure 3B), and, accordingly, a 2-fold increase in the percentage of CD11b+F4/80+ cells (63.6% ± 9.8% versus 36.8% ± 13.7% in BM+MSC) and a higher percentage of CD11b+CD115+ cells (85.8% ± 1.3% versus 38.6% ± 18.9% in BM+MSC) (Figure 3C).

A

B

Mesenchymal stromal cell-induced CD11b+ Gr-1low-neg formation is Nos2-dependent Since NOS2 is one of the key effector molecules in MSC immunomodulatory properties,8,9 we tested the hypothesis that it could also be fundamental for the generation of Gr-1low-neg cells. The differentiation of CD11b+Gr-1low-neg cells was significantly impaired in cultures with Nos2-/- MSC (16.6% ± 1.2% versus 30.8% ± 2.5% in BM+MSC WT), which correlated with an increased proportion in CD11b+Gr-1high cells (82.3% ± 1.2% versus 68.2% ± 2.2% in BM+MSC WT) (Figure 4A,B). Nos2-/- BM cells did not affect the ability of MSC to induce the generation of CD11b+Gr-1low-neg cells (Figure 4A).

Mesenchymal stromal cells increase macrophage formation during hematopoietic reconstitution The ability of MSC to drive the expansion and differentiation of CD11b+ cells was then studied in vivo. Sublethally irradiated mice were transplanted with BM cells from a congenic (CD45.1+) donor either alone or with WT or Nos2-/- MSC. The proportion of the donor myeloid 820

Figure 2. Mesenchymal stromal cell-induced CD11b+ cells consist of a large proportion of M0 macrophages. (A) May-Grünwald Giemsa staining of cytospin preparations of CD11b+ cells isolated from BM MNC cultured with MSC for 4 days. (B) BM MNC cultured alone or with MSC for 4 days were evaluated for the expression of macrophage surface markers within the CD11b+ gated population (open histograms) against their matched isotype controls (filled histograms). Contour plots within the CD11b+ gated population show the expression of each surface marker versus Gr-1 expression in BM MNC cultured with MSC. Contour and histograms plots from one out of six independent experiments, and mean fluorescence intensity values presented as mean ± SD of six independent experiments. *P<0.05, **P<0.01, ***P<0.001, unpaired t test, all comparisons between ‘BM’ versus ‘BM+MSC’.

haematologica | 2017; 102(5)


MSC induce regenerative myeloid cells

A

B

C

Figure 3. Mesenchymal stromal cell-induced CD11b+ differentiation targets committed myeloid progenitors but not hematopoietic stem cells. Unfractionated BM MNC (BM+MSC) or sorted CMP (CMP+MSC), GMP (GMP+MSC), MEP (MEP+MSC) and HSC (HSC+MSC) were cultured with MSC for 4 days. (A) Proportion of CD11b+Gr1high and CD11b+Gr-1low-neg cells in the live gate. A typical result of four independently performed experiments is shown. (B) Percentage of Gr-1high and Gr-1low-neg cells within CD11b+ cells from BM, CMP or GMP cultures with MSC. Mean of four independent experiments, ± SEM. *P<0.05 unpaired t test. (C) Percentage of F4/80+ and CD115+ cells within the CD11b+ gate. Mean of four independent experiments, ± SEM *P<0.05, unpaired t test.

populations was analyzed in the BM and spleens of recipient mice 13 days after the transplant. At this time-point there was no difference in CD45.1+ engraftment (Figure 5A). However, the proportion and absolute numbers of Gr-1negF4/80+SSClow macrophages were increased in mice receiving BM and WT MSC compared to the control group (3.2%±0.3% versus 1.44%±0.18%; absolute number: 0.46x106±0.12x106 versus 0.16x106±0.06x106) (Figure 5B,C). An increase in the proportion (Figure 5D) but not in the absolute number (Online Supplementary Figure S4E) of Ly6G-Ly6C+ monocytes was found in the WT MSC-treated group (23.5%±1.9% versus 16.9%±1.5%; absolute number: 1.06x106±0.14x106 versus 0.96x106±0.25x106). Confirming in vitro results, the adoptive transfer of Nos2/MSC failed to increase the proportion and absolute number of CD11b+Gr-1negF4/80+SSClow macrophages and the proportion of Ly6G-Ly6C+ monocytes within the donor CD45.1+ engraftment (Gr-1negF4/80+SSClow macrophages: 2.1%±0.26% BM+Nos2-/- MSC; absolute number: 0.32x106±0.07x106 BM+Nos2-/- MSC. Ly6G-Ly6C+ monocytes: 18.1%±1.3% BM+Nos2-/- MSC) (Figure 5B-D). There was no difference in the proportion or absolute number of neutrophils or eosinophils in the donor engraftment in all treated groups (Online Supplementary Figure S4C,D). The adoptive transfer of MSC and BM did not affect hematopoietic generation of donor myeloid cells in the spleen of recipient mice (data not shown). haematologica | 2017; 102(5)

Ex-vivo mesenchymal stromal cell-induced CD11b+ cells accelerate hematopoietic reconstitution In order to further characterize the role of MSC-induced CD11b+ cells in vivo, we investigated their effect on hematopoietic reconstitution. For this purpose, sublethally irradiated mice received a BM transplant from a congenic donor with or without CD11b+ cells purified from the BM-MSC co-cultures. The engraftment of donor cells was monitored in the peripheral blood every 2 weeks. The group given CD11b+ cells showed a higher proportion (11.7%±1.5% versus 20.9%±2.1%) and absolute number (17.0±4.2 versus 43.3±7.5 CD45.1+ leukocytes/mL) of donor cells as compared to the control group already at 2 weeks. Such an increment was also observed in each leukocyte compartment (neutrophils: 37.4%±3.4% versus 53.3%±2.9%, and 8.4±2.4 versus 22.8±3.8 CD45.1+ neutrophils/mL; monocytes: 16.4%±2.4 versus 21.2%±2.9%, and 4.9±1.2 versus 12.1±2.9 CD45.1+ monocytes/mL, P=0.0195; B lymphocytes: 10.7%±1.5% versus 18.2%±2.4%, and 3.9±0.7 versus 9.3±2.7 CD45.1+ B lymphocytes/mL; T lymphocytes: 0.06%±0.03% versus 0.3%±0.09%, and 0.02±0.01 versus 0.1±0.04 CD45.1+ T lymphocytes/mL) (Figure 6A,B). Analysis of the absolute numbers of leukocyte populations in peripheral blood showed that CD11b+ cells not only expanded donor engraftment but they also enhanced the recovery of total hematopoiesis. At 2 weeks, the group 821


C. Trento et al.

of mice given CD11b+ cells showed higher absolute counts of neutrophils (15.2±3.4 versus 42.1±5.9 neutrophils/mL), B lymphocytes (27.8±4.8 versus 51.5±7.7 B lymphocytes/mL) and monocytes (22.6±3.7 versus 37.9±4.1 monocytes/mL) (Online Supplementary Figure S5A). Monitoring of the hematopoietic engraftment at later stages revealed that the CD11b+-driven hematopoietic reconstitution was still evident 4 weeks after the transplant (Online Supplementary Figures S5B and S6A,B). At 6 weeks the enhancement effect was selectively observed within the B and T lymphocytes compartments, whilst no

difference in the quality and quantity of engraftment was found from 8 weeks onwards (Online Supplementary Figures S5B and S6A,B).

Human mesenchymal stromal cells induce the differentiation of CD14+ cells from bone marrow mononuclear cells Finally, we asked whether the newly discovered ability of murine MSC to induce macrophage differentiation is also a property of human MSC. For this purpose, human BM MNC were cultured alone or with human MSC for 7

B

A

Figure 4. Mesenchymal stromal cell-induced myeloid cell formation is Nos2-dependent. (A, B) BM MNC isolated from WT or Nos2-/- mice were cultured with MSC isolated from WT or Nos2-/- mice for 4 days. (A) Contour plots represent CD11b+Gr-1high and CD11b+Gr-1low-neg proportions in the live gate. A typical result of three independently performed experiments is shown. (B) Percentages of Gr-1high and Gr-1low-neg within the CD11b+ gate. Mean of three independent experiments ±SEM, **P<0.01, unpaired t test.

A

B

C

D

822

Figure 5. Adoptive transfer of mesenchymal stromal cells increases macrophage formation during hematopoietic reconstitution. Sublethally irradiated (800 cGy) CD45.2 WT recipients were injected with 2x106 CD45.1 WT BM cells, either alone (BM – white bars) or in combination with 0.2x106 CD45.2 WT MSC (BM+WT MSC – black bars) or 0.2x106 CD45.2 Nos2-/- MSC (BM+Nos2-/- MSC – gray bars). 13 days after the transfer, BM and spleen were analyzed by FACS. (A) Percentage of donor engraftment in BM and spleen. (B, C) Percentage (B) and absolute number (C) of Gr1negF4/80+SSClow macrophages within donor hematopoiesis (CD45.1+ cells) in the BM. Mean of three independent experiments ±SEM, *P<0.05, **P<0.01 One-way analysis of variance (ANOVA) with the Bonferroni multiple comparison test. (D) Percentage of Ly6G-Ly6C+ monocytes within donor hematopoiesis (CD45.1+ cells) in the BM. Mean of three independent experiments ±SEM, **P<0.01 One-way ANOVA with the Bonferroni multiple comparison test.

haematologica | 2017; 102(5)


MSC induce regenerative myeloid cells

A

B

Figure 6. Adoptive transfer of mesenchymal stromal cell-induced CD11b+ cells accelerates engraftment. Sublethally irradiated (800 cGy) CD45.2 WT recipients were injected with 5x104 CD45.1 WT BM cells, either alone (BM) or in combination with 2x106 MSC-induced CD11b+ cells (BM+CD11b+). Peripheral blood samples were taken 14 days after the transplant, and the frequency of donor hematopoiesis analyzed by FACS as described in the Methods. (A) Frequency of donor hematopoiesis in MNC, T cells (CD3+), B cells (CD19+), monocytes (Gr-1negCD115+) and neutrophils (Gr-1highCD115+). Mean of four independent experiments ± SEM, **P<0.01, ***P<0.001, unpaired t test. (B) Absolute number of donor hematopoietic cells in MNC, T cells (CD3+), B cells (CD19+), monocytes (Gr-1negCD115+) and neutrophils (Gr-1highCD115+). Mean of four independent experiments ±SEM, *P<0.05, **P<0.01, unpaired t test.

days. We decided to analyze the proportions of CD14+, CD14+HLA-DRlow-neg and CD14+CD16+ cells, which best represent total myeloid population, macrophages and non-classical monocytes, respectively. The presence of human MSC induced an increment in CD14+ cells with a selective effect on non-classical monocytes (CD14+: 15.5%±2.2% versus 25.8%±3.6%, in human BM and human BM cultured with human MSC, respectively; CD14+CD16+: 2.2%±0.6% versus 7.5%±1.6%, in human BM and human BM cultured with human MSC, respectively) (Figure 7A). The analysis of the absolute numbers within the CD45+ population confirmed the statistically significant increase in non-classical monocytes (CD14+CD16+: 0.01x106±0.003x106 versus 0.08x106±0.03x106 in human BM and human BM cultured with human MSC, respectively) (Figure 7B). In contrast to what we observed in the murine system, the addition of the NOS2 inhibitor 1400W to cultures did not affect myeloid differentiation (data not shown).

A

B

Discussion Our study unveils a previously unknown property of BM MSC consisting in the ability to differentiate and expand in vitro and in vivo myeloid cells from BM progenitors. Importantly, the administration of these myeloid cells accelerates hematopoietic reconstitution in BM transplant recipients. The mechanism by which MSC promote myeloid expansion is largely dependent on NOS2. Although NOS2 haematologica | 2017; 102(5)

Figure 7. Human mesenchymal stromal cells induce the differentiation of human bone marrow mononuclear cells into myeloid cells. Human BM MNC were cultured alone or with MSC (ratio 10:1). After 7 days, the percentages (A) and absolute numbers (B) of CD14+, CD14+HLA-DRlow-neg and CD14+CD16+ in the CD45+ population were analyzed. Mean of six independent experiments, ± SEM. *P<0.05, unpaired t test.

823


C. Trento et al.

has been extensively identified as the main effector mechanism accounting for MSC immunosuppressive activity and similarly implicated in myeloid immunobiology, this is the first study that attributes NOS2 the ability to differentiate and expand myeloid cells and macrophages. In contrast to the MSC-mediated immunosuppressive activity,8 we observed that the generation and expansion of myeloid cells was not influenced by inflammatory molecules such as tumor necrosis factor-α and interferon-γ (data not shown). This suggests that, despite using similar mechanisms, MSC-induced myeloid differentiation occurs independently of inflammation. The potent ability of myeloid cells expanded ex-vivo by MSC to accelerate hematopoietic regeneration has at least two implications. The first is that the role of MSC in regulating HSC differentiation is not exclusively a direct one but can also be exerted by recruiting accessory cells. This has already been documented in the landscape of MSC immunomodulation.26 Furthermore, the regenerative properties on the hematopoietic system may apply to other tissues. The molecular and functional profile of MSC-educated myeloid cells recapitulates that of the resident macrophages which have been described as involved in tissue repair activity in other organs.27,28 Therefore, our data might lend support to the notion that stroma/fibroblasts orchestrate tissue homeostasis. The second set of implications affects the clinical approach to the use of MSC. MSC have long been studied for their ability to promote hematopoietic engraftment based on the evidence that they are a crucial component of the stem cell niche. After the initial investigation in xenogeneic models showing that MSC co-transplanted with umbilical cord blood CD34+ cells resulted in an increase in BM engraftment,17 subsequent studies indicated that such engraftment is only transient.29,30 Similar confounding factors can be found in clinical studies in pediatric31,32 and adult19,33-35 cohorts with small and heterogeneous groups of

References 7. 1. Di Nicola M, Carlo-Stella C, Magni M, et al. Human bone marrow stromal cells suppress T-lymphocyte proliferation induced by cellular or nonspecific mitogenic stimuli. Blood. 2002;99(10):3838-3843. 2. Corcione A, Benvenuto F, Ferretti E, et al. Human mesenchymal stem cells modulate B-cell functions. Blood. 2006;107(1):367-372. 3. Krampera M, Glennie S, Dyson J, et al. Bone marrow mesenchymal stem cells inhibit the response of naive and memory antigen-specific T cells to their cognate peptide. Blood. 2003;101(9):3722-3729. 4. Glennie S, Soeiro I, Dyson PJ, Lam EW, Dazzi F. Bone marrow mesenchymal stem cells induce division arrest anergy of activated T cells. Blood. 2005;105(7):2821-2827. 5. Ramasamy R, Fazekasova H, Lam EW, et al. Mesenchymal stem cells inhibit dendritic cell differentiation and function by preventing entry into the cell cycle. Transplantation. 2007;83(1):71-76. 6. Nemeth K, Leelahavanichkul A, Yuen PS, et al. Bone marrow stromal cells attenuate sepsis via prostaglandin E(2)-dependent reprogramming of host macrophages to increase

824

8.

9.

10.

11.

12.

patients. If MSC graft facilitating activity is the consequence of inducing BM progenitors to differentiate into regenerative myeloid cells, the variability in clinical outcome may result from the content of progenitors in the original donor inoculum rather than the actual direct activity of MSC on HSC differentiation. Our data highlight the complexity of the MSC ‘clinical niche effect’. Although the adoptive transfer of MSC at the time of BM transplantation significantly increased the percentage and number of donor macrophages in the BM (Figure 5B,C), we could not observe a consequent increment in the overall donor hematopoietic engraftment (Figure 5A). This discrepancy could be explained by the low number of macrophages produced by the MSC infusion. In fact, when BM cells were infused with a high number of exvivo MSC-induced myeloid cells, the hematopoietic engraftment was greatly enhanced (Figure 6). The crucial role of myeloid cells in facilitating engraftment is also supported by recent clinical evidence in which MSC have been used to condition cord blood HSC ex-vivo in order to accelerate hematopoietic recovery. Our data shed light on the interpretation of these successful clinical results.36 In that study, the MSC-conditioned cord blood unit did not engraft, indicating it mainly played a facilitating effect on the engraftment of the co-transplanted unmanipulated unit. Our study suggests that the graft facilitating effect might have been mediated by MSC-expanded monocytes, as confirmed by the phenotypic analysis of their MSC-educated cord blood HSC. Therefore, we propose the intriguing possibility that ex-vivo MSC induced myeloid cells could be harnessed in the clinical setting to expedite hematopoietic recovery and immune reconstitution in highrisk transplantation procedures.37 Funding This work was supported by Bloodwise 15029 and Stellar FP7 EU grant.

their interleukin-10 production. Nat Med. 2009;15(1):42-49. Gupta N, Su X, Popov B, et al. Intrapulmonary delivery of bone marrowderived mesenchymal stem cells improves survival and attenuates endotoxin-induced acute lung injury in mice. J Immunol. 2007;179(3):1855-1863. Ren G, Zhang L, Zhao X, et al. Mesenchymal stem cell-mediated immunosuppression occurs via concerted action of chemokines and nitric oxide. Cell Stem Cell. 2008;2(2):141-150. Sato K, Ozaki K, Oh I, et al. Nitric oxide plays a critical role in suppression of T-cell proliferation by mesenchymal stem cells. Blood. 2007;109(1):228-234. Melief SM, Geutskens SB, Fibbe WE, Roelofs H. Multipotent stromal cells skew monocytes towards an anti-inflammatory interleukin-10-producing phenotype by production of interleukin-6. Haematologica. 2013;98(6):888-895. Maggini J, Mirkin G, Bognanni I, et al. Mouse bone marrow-derived mesenchymal stromal cells turn activated macrophages into a regulatory-like profile. PLoS One. 5(2):e9252. Francois M, Romieu-Mourez R, Li M,

13.

14.

15.

16.

17.

Galipeau J. Human MSC suppression correlates with cytokine induction of indoleamine 2,3-dioxygenase and bystander M2 macrophage differentiation. Mol Ther. 2012;20(1):187-195. Choi H, Lee RH, Bazhanov N, Oh JY, Prockop DJ. Anti-inflammatory protein TSG-6 secreted by activated MSCs attenuates zymosan-induced mouse peritonitis by decreasing TLR2/NF-kappaB signaling in resident macrophages. Blood. 2011;118(2): 330-338. Ortiz LA, Dutreil M, Fattman C, et al. Interleukin 1 receptor antagonist mediates the antiinflammatory and antifibrotic effect of mesenchymal stem cells during lung injury. Proc Natl Acad Sci USA. 2007;104(26):11002-11007. Mendelson A, Frenette PS. Hematopoietic stem cell niche maintenance during homeostasis and regeneration. Nat Med. 2014;20(8):833-846. Mendez-Ferrer S, Michurina TV, Ferraro F, et al. Mesenchymal and haematopoietic stem cells form a unique bone marrow niche. Nature. 2010;466(7308):829-834. Noort WA, Kruisselbrink AB, in't Anker PS, et al. Mesenchymal stem cells promote engraftment of human umbilical cord blood-

haematologica | 2017; 102(5)


MSC induce regenerative myeloid cells

18.

19.

20.

21.

22.

23.

24.

derived CD34(+) cells in NOD/SCID mice. Exp Hematol. 2002;30(8):870-878. Koc ON, Gerson SL, Cooper BW, et al. Rapid hematopoietic recovery after coinfusion of autologous-blood stem cells and culture-expanded marrow mesenchymal stem cells in advanced breast cancer patients receiving high-dose chemotherapy. J Clin Oncol. 2000;18(2):307-316. Lazarus HM, Koc ON, Devine SM, et al. Cotransplantation of HLA-identical sibling culture-expanded mesenchymal stem cells and hematopoietic stem cells in hematologic malignancy patients. Biol Blood Marrow Transplant. 2005;11(5):389-398. Lymperi S, Ersek A, Ferraro F, Dazzi F, Horwood NJ. Inhibition of osteoclast function reduces hematopoietic stem cell numbers in vivo. Blood. 2011;117(5):15401549. Chow A, Lucas D, Hidalgo A, et al. Bone marrow CD169+ macrophages promote the retention of hematopoietic stem and progenitor cells in the mesenchymal stem cell niche. J Exp Med. 2011;208(2):261-271. Ludin A, Itkin T, Gur-Cohen S, et al. Monocytes-macrophages that express alpha-smooth muscle actin preserve primitive hematopoietic cells in the bone marrow. Nat Immunol. 2012;13(11):1072-1082. Kollet O, Dar A, Shivtiel S, et al. Osteoclasts degrade endosteal components and promote mobilization of hematopoietic progenitor cells. Nat Med. 2006;12(6):657-664. Mansour A, Abou-Ezzi G, Sitnicka E, et al. Osteoclasts promote the formation of

haematologica | 2017; 102(5)

25.

26. 27.

28.

29.

30.

31.

hematopoietic stem cell niches in the bone marrow. J Exp Med. 2012;209(3):537-549. Peranzoni E, Zilio S, Marigo I, et al. Myeloid-derived suppressor cell heterogeneity and subset definition. Curr Opin Immunol. 22(2):238-244. Marigo I, Dazzi F. The immunomodulatory properties of mesenchymal stem cells. Semin Immunopathol. 2011;33(6):593-602. de Couto G, Liu W, Tseliou E, et al. Macrophages mediate cardioprotective cellular postconditioning in acute myocardial infarction. J Clin Invest. 2015;125(8):31473162. Liu C, Wu C, Yang Q, et al. Macrophages mediate the repair of brain vascular rupture through direct physical adhesion and mechanical traction. Immunity. 2016;44(5): 1162-1176. Jaganathan BG, Tisato V, Vulliamy T, et al. Effects of MSC co-injection on the reconstitution of aplastic anemia patient following hematopoietic stem cell transplantation. Leukemia. 2010;24(10):1791-1795. Fernandez-Garcia M, Yanez RM, SanchezDominguez R, et al. Mesenchymal stromal cells enhance the engraftment of hematopoietic stem cells in an autologous mouse transplantation model. Stem Cell Res Ther. 2015;6:165. Ball LM, Bernardo ME, Roelofs H, et al. Cotransplantation of ex vivo expanded mesenchymal stem cells accelerates lymphocyte recovery and may reduce the risk of graft failure in haploidentical hematopoietic stem-cell transplantation. Blood. 2007;110

(7):2764-2767. 32. Bernardo ME, Ball LM, Cometa AM, et al. Co-infusion of ex vivo-expanded, parental MSCs prevents life-threatening acute GVHD, but does not reduce the risk of graft failure in pediatric patients undergoing allogeneic umbilical cord blood transplantation. Bone Marrow Transplant. 2011;46(2):200207. 33. Gonzalo-Daganzo R, Regidor C, MartinDonaire T, et al. Results of a pilot study on the use of third-party donor mesenchymal stromal cells in cord blood transplantation in adults. Cytotherapy. 2009;11(3):278-288. 34. Wu Y, Wang Z, Cao Y, et al. Cotransplantation of haploidentical hematopoietic and umbilical cord mesenchymal stem cells with a myeloablative regimen for refractory/relapsed hematologic malignancy. Ann Hematol. 2013;92(12): 1675-1684. 35. Wu Y, Cao Y, Li X, et al. Cotransplantation of haploidentical hematopoietic and umbilical cord mesenchymal stem cells for severe aplastic anemia: successful engraftment and mild GVHD. Stem Cell Res. 2014;12(1):132138. 36. de Lima M, McNiece I, Robinson SN, et al. Cord-blood engraftment with ex vivo mesenchymal-cell coculture. N Engl J Med. 2012;367(24):2305-2315. 37. Ballen KK, Koreth J, Chen YB, Dey BR, Spitzer TR. Selection of optimal alternative graft source: mismatched unrelated donor, umbilical cord blood, or haploidentical transplant. Blood. 2012;119(9):1972-1980.

825


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Red Cell Biology & its Disorders

Ferrata Storti Foundation

Loss of Forkhead box M1 promotes erythropoiesis through increased proliferation of erythroid progenitors

Minyoung Youn, Nan Wang, Corinne LaVasseur, Elena Bibikova, Sharon Kam, Bertil Glader, Kathleen M. Sakamoto* and Anupama Narla* Department of Pediatrics, Stanford University School of Medicine, CA, USA

Haematologica 2017 Volume 102(5):826-834

*Co-senior corresponding authors

ABSTRACT

F

Correspondence: anunarla@stanford.edu

Received: September 7, 2016. Accepted: January 24, 2017. Pre-published: February 2, 2017.

orkhead box M1 (FOXM1) belongs to the forkhead/winged-helix family of transcription factors and regulates a network of proliferation-associated genes. Its abnormal upregulation has been shown to be a key driver of cancer progression and an initiating factor in oncogenesis. FOXM1 is also highly expressed in stem/progenitor cells and inhibits their differentiation, suggesting that FOXM1 plays a role in the maintenance of multipotency. However, the exact molecular mechanisms by which FOXM1 regulates human stem/progenitor cells are still uncharacterized. To understand the role of FOXM1 in normal hematopoiesis, human cord blood CD34+ cells were transduced with FOXM1 short hairpin ribonucleic acid (shRNA) lentivirus. Knockdown of FOXM1 resulted in a 2-fold increase in erythroid cells compared to myeloid cells. Additionally, knockdown of FOXM1 increased bromodeoxyuridine (BrdU) incorporation in erythroid cells, suggesting greater proliferation of erythroid progenitors. We also observed that the defective phosphorylation of FOXM1 by checkpoint kinase 2 (CHK2) or cyclin-dependent kinases 1/2 (CDK1/2) increased the erythroid population in a manner similar to knockdown of FOXM1. Finally, we found that an inhibitor of FOXM1, forkhead domain inhibitor-6 (FDI-6), increased red blood cell numbers through increased proliferation of erythroid precursors. Overall, our data suggest a novel function of FOXM1 in normal human hematopoiesis.

doi:10.3324/haematol.2016.156257

Introduction Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/826 Š2017 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.

826

Hematopoiesis is the critical process by which normal blood cells are derived from self-renewing and pluripotent hematopoietic stem cells (HSCs).1,2 Erythropoiesis, which leads to the formation of mature red blood cells, requires that each cell division is simultaneously coupled with differentiation.3,4 Erythropoiesis can be divided into 3 stages: early erythropoiesis, terminal erythroid differentiation, and reticulocyte maturation.5,6 During early erythropoiesis, HSCs sequentially give rise to a common myeloid progenitor, megakaryocyte-erythrocyte progenitor, burst-forming unit-erythroid (BFU-E), and colony-forming uniterythroid (CFU-E) cells that differentiate into proerythroblasts.7-11 In terminal erythroid differentiation, morphologically recognizable proerythroblasts undergo sequential mitosis to become basophilic, polychromatic, and orthochromatic erythroblasts that expel their nuclei to become reticulocytes.12 At the final step of erythropoiesis, multinucleated reticulocytes mature into red blood cells accompanied by the loss of intracellular organelles, a decrease in cell volume, and extensive membrane remodeling.13-18 Erythropoiesis is tightly regulated by various regulatory growth factors and transcription factors,3,6 with erythropoietin (EPO) and GATA-1 playing essential roles.7,10 The binding of EPO to its receptor (EPOR) activates the Jak2-Stat5 signaling pathhaematologica | 2017; 102(5)


FOXM1 in erythropoiesis

way to promote the proliferation of erythroid progenitor cells and to rescue erythroid progenitors from cell death.1921 GATA-1 leads to the expression of erythroid-specific genes including EPOR, adult globin genes, heme biosynthesis enzymes, and erythroid membrane proteins.22 The expression or transcriptional activation of erythroid-specific growth and transcription factors is critical for normal erythropoiesis. FOXM1 belongs to the forkhead/winged-helix family of transcription factors and binds to a specific DNA consensus sequence through a highly conserved DNA-binding domain (DBD).23-26 It is a key transcription factor in the regulation of a network of proliferation-associated genes including the G1/S transition, S phase progression, G2/M transition, and M phase progression, and is also critical for DNA replication, mitosis, spindle assembly, and genomic stability.24,27-30 Consistent with its role in cell-cycle progression, FOXM1 expression is highly upregulated in a number of human cancers, including liver, ovarian, breast, prostate, colon, and brain tumors.31-32 Its abnormal upregulation has been shown to be a key driver of cancer progression and an initiating factor of oncogenesis.33 FOXM1 is also highly expressed in multipotent progenitor cells and inhibits differentiation of progenitors, suggesting that FOXM1 plays a role in the maintenance of multipotent progenitor cells.34-36 It is reported that Foxm1 participates in the maintenance of pluripotency of mouse P19 embryonal carcinoma cells by directly regulating Oct4 transcription.37 In addition, the overexpression of Foxm1 alone in human newborn fibroblasts restarts the expression of pluripotent genes, including Oct4, Nanog, and Sox2. Recently, it was also reported that Foxm1 is essential for the quiescence and maintenance of hematopoietic stem cells in the mouse model.38 Loss of Foxm1 induced the decrease of cyclin-dependent kinase inhibitors by suppressing the Nurr1 gene (a critical regulator of HSC quiescence) expression. However, the exact molecular mechanisms by which FOXM1 regulates human hematopoietic stem and progenitor cells are still uncharacterized. In this study, we have examined the role of FOXM1 in normal hematopoiesis using human cord blood CD34+ cells and lentivirus to target FOXM1. We found that knockdown of FOXM1 resulted in an increase in the erythroid population compared to the myeloid population, with a higher expression of CD71+ (erythroid) cells compared to CD11b+ (myeloid) cells. We also found that FOXM1 knockdown increased BrdU incorporation in CD71+ cells only, suggesting a greater proliferation of erythroid progenitors. Taken together, these studies suggest a novel function of FOXM1 in normal human hematopoiesis. Our data indicate that FOXM1 inhibitors, such as FDI-6, may be beneficial in treating patients with anemia due to decreased red blood production.

Research Protection Program and Institutional Review Board. CD34+ cells were purified using MACS cell separation (Miltenyi Biotec) and cryopreserved. Primary human bone marrow CD34+ cells were purchased from Lonza. Upon thawing, cells were cultured in x-Vivo15 medium (Lonza) containing 10% fetal bovine serum (FBS), 1x Penicillin-Streptomycin-Glutamine (PSG) (Invitrogen), FLT-3 (Miltenyi Biotec), thrombopoietin (TPO; Miltenyi Biotec), interleukin (IL)-3 (Miltenyi Biotec), IL-6 (Miltenyi Biotec), stem cell factor (SCF; Miltenyi Biotec), EPO, and transferrin (Sigma-Aldrich) by a 3 phase liquid culture system as described in the Online Supplementary Table S1. HEK 293 and K562 cell lines were cultured in Iscove's Modification of Dulbecco's Medium (Corning) containing 10% FBS and 1x PSG.

Methods

Drug treatment

Cell culture Primary human CD34+ hematopoietic stem/progenitor cells were purified from umbilical cord blood units obtained from The National Cord Blood Program at the Howard P. Milstein Cord Blood Center of New York Blood Center. These are unprocessed, non-clinical grade cord blood units and are considered “Research Units”. These materials are de-identified and are therefore considered not to involve human subjects as per the Stanford Human haematologica | 2017; 102(5)

Colony assays GFP+ or mCherry+ sorted hematopoietic cells were seeded in methylcellulose medium containing IL-3, SCF, granulocytemacrophage colony-stimulating factor (GM-CSF), and EPO (H4434, StemCell Technologies), in triplicate, with a density of 1000 cells per plate. Erythroid (BFU-E and CFU-E) and myeloid colony-forming unit granulocyte/monocyte (CFU-GM) colonies were counted 2 weeks later by an investigator blinded to the conditions. Methylcellulose medium containing only EPO (H4330; StemCell Technologies) was used for erythroid colony (CFU-E) and colonies were counted 1 week later. Methylcellulose medium containing SCF, IL-3, granulocytecolony stimulating factor (G-CSF), and GM-CSF (H4035; StemCell Technologies) was used for the myeloid colony, and colonies were counted 2 weeks later.

Flow cytometry For cell surface flow cytometry, cells were washed with phosphate-buffered saline (PBS) and then incubated with indicated antibodies for 20 minutes at room temperature. After washing, cells were analyzed by fluorescence-activated cell sorting (FACS). For intracellular flow cytometry, cells were fixed in 3.2% paraformaldehyde for 10 minutes at 37°C, and permeabilized with 100% methanol for 30 minutes at -80°C. After washing, cells were incubated with indicated antibodies and then analyzed by FACS. Data were collected on a FACS Calibur (BD Biosciences) or a DxP10 (Cytek) flow cytometer and analyzed using FlowJo Software (v.10). All antibodies are listed in the Online Supplementary Methods. Cells were plated in a 48 well plate and incubated with 10 μl of 1 mM BrdU (BD Pharmingen) for 45 minutes. Cells were fixed in 3.2% paraformaldehyde for 10 minutes at 37°C, and permeabilized with 100% methanol for 30 minutes at -80°C. After washing, cells were treated with RQ1 RNase-free DNase (Promega) for 1 hour at 37°C. After washing, cells were incubated with BrdU and CD71-PE antibodies and then analyzed by FACS. Antibody against BrdU (Bu20A: APC-conjugated: 17-5071-42) was purchased from eBioscience.

BrdU incorporation

FDI-6 (Axon2384; Axon Medchem) was dissolved in dimethyl sulfoxide (DMSO) to create a 40 mM stock and added to cells at a final concentration of 10 mM. Roscovitine (S1153; Selleck Chemicals), U0126-EtOH (S1102; Selleck Chemicals), and ADZ7762 (S1532; Selleck Chemicals) were dissolved in DMSO and added to cells at a final concentration of 10 mM, 5 mM, and 50 nM, respectively. Cells were treated with the respective drugs from day 1 to day 5 after thawing; the drug was then washed out for the remainder of the culture time. 827


M. Youn et al. A

B

C Figure 1. FOXM1 is highly expressed in the erythroid population. Human cord blood CD34+ hematopoietic progenitor cells were cultured in vitro in differentiation media. At the indicated days, cells were stained with each cell surface marker and sorted. (A) RNA was collected from the indicated populations and analyzed by RT-qPCR. (B) At the indicated days, cell were stained with each cell surface marker and FOXM1 antibody. Stained cells were analyzed by flow cytometry. (C) Data courtesy of An et al.39 (please see Reference for experimental details). Error bars represent standard deviations of 4 independent replicates. Significance was evaluated using Student's t test. *P<0.05, **P<0.01, ***P<0.001. d: day; mRNA: messenger RNA; GlyA: glycophorin A; BFU: burstforming unit; CFU: colony-forming unit; Pro: proerythroblasts; EBaso: early basophilic erythroblasts; LBaso: late basophilic erythroblasts; Poly: polychromatic erythroblasts; Ortho: orthochromatic erythroblasts.

Results FOXM1 is highly expressed in the erythroid population To understand the role of FOXM1 in hematopoietic cells, we first investigated the expression profiles of FOXM1 in individual cell populations. Trilineage differentiation of human cord blood CD34+ hematopoietic progenitor cells was induced in vitro using a 3-phase liquid culture system, as described in the Online Supplementary Table S1. We used the following cell surface markers: CD71 for early erythroid, glycophorin A (GlyA) for late erythroid, CD33 for early myeloid, CD11b for late myeloid, and CD41a for megakaryocytes. Populations were analyzed on day 6 or day 12 of culture. Sorting efficiencies for each population were confirmed by running reverse transcription-quantitative polymerase chain reaction (RT-qPCR) with each cell marker (Online Supplementary Figure S1). We found that FOXM1 had a 3-fold increased RNA level in CD71+ cells compared to CD11b+ cells (Figure 1A). To observe the effects on the protein level, we performed intracellular flow cytometry for FOXM1 with each cell surface marker. Consistent with the RNA levels, FOXM1 protein was high in the CD71+ population, suggesting that FOXM1 plays a specific role in erythropoiesis (Figure 1B and Online Supplementary Figure S2). Data from a previously published paper confirms our findings that FOXM1 has a role in erythropoiesis. The authors performed a detailed transcriptome analysis of the various stages of normal human erythropoiesis including BFU-E, CFU-E, proerythroblasts (Pro), Early and Late basophilic erythroblasts (E Baso and L Baso), polychromatic erythroblasts (Poly), and orthochromatic erythroblasts (Ortho).39 Interestingly, FOXM1 expression was elevated in erythroid cells, suggesting a functional role for 828

FOXM1 in erythroid population (Figure 1C; data courtesy of An et al.39). We chose to focus on the earlier stages of erythropoiesis which also demonstrated a higher expression of FOXM1, since it was challenging to study the later stages of erythropoiesis in our liquid culture system using shRNA knockdown.

FOXM1 downregulation increases the erythroid population To understand the function of FOXM1 in human erythropoiesis, we examined whether FOXM1 knockdown affects normal hematopoiesis. We confirmed the knockdown efficiency of two different shRNAs against FOXM1 and observed significantly decreased RNA levels in human cord blood CD34+ cells (Figure 2A). We also confirmed decreased protein levels with FOXM1 knockdown in the K562 cell line by immunoblot analysis and in the human cord blood CD34+ cells by intracellular flow cytometry (Online Supplementary Figure S3A,B). After transduction of human CD34+ cells with lentivirus expressing FOXM1 shRNA, we studied hematopoietic differentiation using FACS analysis with a range of cell surface markers. We found that knockdown of FOXM1 resulted in an increase of the erythroid population as measured by CD71 and GlyA and a decrease in the myeloid population as measured by CD11b (Figure 2B, Online Supplementary Figures S4 and S5). Additionally, we detected the effects on specific stages of erythroid differentiation by using GlyA, CD49d, and Band3 as markers.40 Consistent with our findings in the CD71+/GlyA+ population, the populations of Pro, E Baso, and L Baso stages were increased by FOXM1 knockdown (Online Supplementary Figure S6A,B). Similarly, methylcellulose colony assays demonstrated increased numbers of CFU-E colonies and decreased numbers of haematologica | 2017; 102(5)


FOXM1 in erythropoiesis

A

C

B

D

E

Figure 2. FOXM1 downregulation increases the erythroid population. Human cord blood CD34+ hematopoietic progenitor cells were transduced with lentivirus carrying shRNA against FOXM1 or luciferase (Luc) control. (A) Cells were sorted for GFP+ at 5 days after transduction. RNA was collected and analyzed by RT-qPCR. (B) Transduced cells were analyzed for CD71, GlyA, and CD11b expression by flow cytometry at the indicated days after transduction. (C) 1000 cells of GFP+ cells were plated in methylcellulose media and cultured for 2 weeks. Colonies were counted by an investigator blinded to the conditions. (D) GFP+ cells following cytospin and Wright-Giemsa staining at the indicated days post-transduction, and imaged at 63 x magnification. Arrows indicate a representative erythrocyte. (E) Cells in erythroid media or myeloid media were transduced with lentivirus and then were sorted for GFP+ at 5 days after transduction. 1000 cells of each GFP+ cells were plated in methylcellulose media for detection of CFU-E or CFU-GM, and cultured. Colonies were counted by an investigator blinded to the conditions. *P<0.05, **P<0.01, ***P<0.001. sh: short hairpin; d: day; GlyA: glycophorin A; CFU-E: colony-forming unit-erythroid; BFU-E: burst-forming unit-erythroid; CFU-GM: colony-forming unit granulocyte/monocyte.

CFU-GM colonies, but we did not note any morphological changes in FOXM1 knockdown cells (Figure 2C,D and Online Supplementary Figure S7). To determine whether the increase in the erythroid population resulting from FOXM1 knockdown was not due to a decrease in myelopoiesis, we cultured human cord blood CD34+ cells in medium supporting either erythroid or myeloid differentiation as described in the Online Supplementary Tables S2 and S3. After transduction with lentivirus expressing FOXM1 shRNA, we measured the colony forming activity of cells cultured in methylcellulose media specific for either erythroid progenitors or myeloid progenitors. We observed an increase in CFU-E colonies with erythroid specific medium in CD34+ FOXM1 knockdown cells, but no change in CFU-GM colonies with myeloid media (Figure 2E). We observed similar results using FACS analysis (Online Supplementary Figure S8). These results suggest that FOXM1 knockdown directly affects erythropoiesis. To study the effects of FOXM1 overexpression during erythropoiesis, we cloned full-length cDNA of FOXM1 into a lentiviral vector (Online Supplementary Figure S9A,B). Human CD34+ cells were transduced with lentivirus expressing FOXM1 cDNA, and FACS analysis and colony assays were performed. FOXM1 overexpression did not affect erythroid or myeloid differentiation (Online Supplementary Figure S9C,D). Taken together, these results suggest that FOXM1 knockdown directly increases the erythroid population and overexpression does not increase erythroid or myeloid colony formation. haematologica | 2017; 102(5)

FOXM1 downregulation increases the proliferation of erythroid progenitors FOXM1 is known to play a critical role in cell proliferation.37 In order to understand its potential mechanism of action in erythropoiesis, we investigated its role on the proliferation of hematopoietic progenitor cells. We again used an in vitro liquid culture system to differentiate and separate transduced CD34+ cells into erythroid and nonerythroid or myeloid populations using CD71 and CD11b. Viable cells were counted every 3 days to observe effects on proliferation. We found that FOXM1 knockdown increased the number of CD71+ cells, but not CD71-, CD11b+, or total cell populations after 11 days in culture (Figure 3A,B). This is similar to the increase in the CD71+/GlyA+ population, which we noted in earlier experiments (Figure 2B). To verify that the increased cell number is caused by increased cell proliferation, we performed BrdU incorporation assays at 13 days after transduction and found that knockdown of FOXM1 resulted in increased BrdU positive cells in the CD71+ population only, indicating more rapid proliferation of erythroid progenitors (Figure 3C left panel and Online Supplementary Figure S10). Similarly, H3P-ser10, which is a specific marker for M phase, was increased in FOXM1 knockdown CD71+ cells (Figure 3C right panel), indicating increased mitotic activity in the erythroid population with knockdown of FOXM1. Additionally, we transduced sorted CD71+ erythroid cells with FOXM1 shRNA lentivirus and observed increased BrdU positive cells (Online Supplementary Figure S11), confirming the direct action of 829


M. Youn et al. A

B

C

Figure 3. FOXM1 downregulation increases the proliferation of erythroid progenitors. Human cord blood CD34+ hematopoietic progenitor cells were transduced with lentivirus carrying shRNA against FOXM1 or Luc control. Cells were sorted for CD71+/GFP+ (for the CD71+ group), CD71–/GFP+ (for the CD71– group), CD11b+/GFP+ (for the CD11b+ group), or GFP+ (for the All group) cells at 5 days after transduction. (A and B) 4000 sorted cells were cultured for an additional 12 days or 9 days. Cells were counted every 3 days. (C) Sorted cells were cultured for an additional 8 days. At 13 days post-transduction, cells were incorporated with BrdU, and then stained with BrdU and CD71 antibodies. Or, cells were stained with H3P-ser10 and CD71 antibodies. Stained cells were analyzed by flow cytometry. *P<0.05, **P<0.01, ***P<0.001. sh: short hairpin; d: day; BrdU: bromodeoxyuridine; Luc: luciferase.

FOXM1 knockdown in erythroid proliferation. In summary, these data suggest that FOXM1 downregulation increases the proliferation of the erythroid progenitors.

FOXM1 effects on erythropoiesis are independent of the p53 pathway Previous reports established that p53 and FOXM1 negatively regulate each other’s activity.27,41 In our system, we confirmed that FOXM1 expression was decreased in CD34+ progenitor cells treated with Nutlin-3, a drug that leads to p53 stabilization by blocking MDM2 (Online Supplementary Figure S12A,B). It has also been reported that Foxm1-deficient MEFs have increased transcriptional activity of p53 with corresponding stimulation of their target gene expression.27 This suggests the possibility that p53 activity is involved in regulating FOXM1 activity during erythropoiesis. To test this hypothesis, we studied the expression of p53 target genes in FOXM1 knockdown cells. FOXM1 downregulation did not affect the expression of p21, WIG-1, BAX, and GADD45A in either the early or late stages of erythropoiesis (Online Supplementary Figure S13A). The expression of GATA1 (a canonical erythroid transcription factor) and TNF-α (whose role in erythropoiesis we and others have studied) were also not affected (Online Supplementary Figure S13B).42 Finally, we could not detect any increase in the p53 protein level and cPARP+ apoptotic cells in FOXM1 knockdown cells, although their levels were increased in RPS19 knockdown cells, which were used as a positive control (Online Supplementary Figure S13C,D).42,43 Taken together, these results suggest that the effects of FOXM1 on erythropoiesis are p53-independent. 830

FOXM1 function on erythroid proliferation requires its phosphorylation by CHK2 or CDK1/2 kinases It is known that FOXM1 activity is regulated by phosphorylation involving multiple regulatory kinases. For example, FOXM1 transcriptional activity requires binding of the CDK-Cyclin complexes and subsequent phosphorylation to regulate G2/M cell cycle regulatory genes.44-46 Additionally, FOXM1 is phosphorylated by DNA damageinduced CHK2, resulting in the stabilization of the FOXM1 protein.27 To understand the effects of the upstream signaling pathways on FOXM1 during erythroid proliferation and differentiation, we studied three representative kinases, checkpoint kinase 2 (CHK2), cyclin-dependent kinase 1/2 (CDK1/2), and polo-like kinase 1 (PLK1). We examined their expression levels at various time points during hematopoiesis using our in vitro CD34+ system. All observed kinases were expressed at the protein level throughout hematopoiesis, with a relatively high expression in the earlier stage (Figure 4A). We then made alanine-substituted FOXM1 mutants on serine or threonine residues that would be predicted to affect normal phosphorylation (Figure 4B). We confirmed that mRNA and protein of the FOXM1 mutants were normally expressed in CD34+ cells by RT-qPCR and in HEK 293 cells by immunoblot analysis (Online Supplementary Figure S14A,B). We then transduced CD34+ progenitor cells with these constructs and performed methylcellulose colony assays as described previously. We found increased BFU-E and CFU-E colonies in mt1, mt2/3, and mt1-5 but not mt4/5 (Figure 4C), suggesting that FOXM1 phosphorylation by CHK2 and CDK1/2 kinases is important in its role in regulating the proliferation of erythroid progenitors. haematologica | 2017; 102(5)


FOXM1 in erythropoiesis

A

C

B

D

Figure 4. FOXM1 function on erythroid proliferation requires its phosphorylation by CHK2 or CDK1/2 kinases. (A) Human cord blood CD34+ hematopoietic progenitor cells were cultured for 21 days. At the indicated days, cells were harvested. Protein was collected and analyzed by immunoblot analysis. GATA1 was used as an erythroid-specific marker protein. β-Actin was used as a loading control. (B) The schema of human FOXM1 phosphorylation sites by CHK2, CDK1/2, and PLK1. (C) Human cord blood CD34+ hematopoietic progenitor cells were transduced with lentivirus carrying FOXM1 mutant cDNA. Cells were sorted for mCherry+ at 5 days after transduction. 1000 cells of mCherry+ cells were plated in methylcellulose media and cultured for 2 weeks. Colonies were counted by an investigator blinded to the conditions. (D) Human cord blood CD34+ hematopoietic progenitor cells were treated with the each kinase inhibitor for 5 days. Cells were analyzed for CD71, GlyA, and CD11b expression by flow cytometry at the indicated days after washing out the drug. *P<0.05, **P<0.01, ***P<0.001. WT: wild-type; d: day; CHK2: checkpoint kinase 2; CDK1/2: cyclin-dependent kinase 1/2; PLK1: polo-like kinase 1; GlyA: glycophorin A; DMSO: dimethyl sulfoxide; mt: mutant.

Similarly, we observed increased BFU-E at 5 days, increased CFU-E with decreased BFU-E at 8 days, and increased Pro at 11 days after culture with all mutant forms, indicating an effect on promoting erythropoiesis (Online Supplementary Figure S15). Finally, we studied whether these kinase inhibitors show similar effects to FOXM1 mutants. We treated CD34+ cells with roscovitine (CDK1/2 inhibitor) or ADZ7762 (CHK1/2 inhibitor) and then performed FACS analysis on the indicated days. We observed an increase in the CD71+/GlyA+ population and a decrease in the CD11b+ population upon treatment with the kinase inhibitors (Figure 4D and Online Supplementary Figure S16). Additionally, we found similar results with U0126-EtOH, an inhibitor of MEK1/2, which is the upstream kinase of CDK1/2. These results indicate that CHK2 or CDK1/2 signaling pathways are required for FOXM1 function on erythroid differentiation.

FDI-6 increases the erythroid population through increased proliferation of its progenitors Recently, Gormally et al. identified a small molecule, FDI6, as a specific inhibitor of FOXM1.47 FDI-6 inhibits the tranhaematologica | 2017; 102(5)

scriptional activity of FOXM1 by its displacement from target gene promoters. To observe whether FDI-6 shows the same effects to FOXM1 downregulation on erythroid cells, we treated cord blood CD34+ cells with 10 mM of FDI-6 for 5 days. We confirmed that 10 mM of FDI-6 was enough to inhibit FOXM1 transcriptional activity in cord blood CD34+ cells by observing the expression of several known FOXM1 target genes, FOXM1, CDC25B, CyclinD1, CENPF, CyclinB2, and p27 (Online Supplementary Figure S17). We performed FACS analysis every 3 days and found dramatic increases of CD71+ and CD71+/GlyA+ populations with FDI6 treatment (Figure 5A and Online Supplementary Figure S18). Similar results were observed in methylcellulose colony assays containing 10 mM of FDI-6, with increased numbers of CFU-E colonies and decreased numbers of CFU-GM colonies with treatment (Figure 5B). Finally, we studied the effects of FDI-6 on cellular proliferation as described in Figure 3, and noted that FDI-6 increased the number of CD71+ cells and BrdU positive CD71+ cells (Figure 5C,D and Online Supplementary Figure S19). These results suggest that inhibition of FOXM1 by FDI6 increases erythroid cells by enhancing erythroid progenitor proliferation. 831


M. Youn et al. A

C

B

D

Figure 5. FDI-6 increases the erythroid population through increased proliferation of its progenitors. Human cord blood CD34+ hematopoietic progenitor cells were treated with FDI-6 for 5 days. (A) The drug was washed out after 5 days. Cells were analyzed for CD71 and GlyA expression by flow cytometry at the indicated days after culture. (B) 1000 drug-treated cells were plated in methylcellulose media containing 10 mM of FDI-6 and cultured for 10 days. Colonies were counted by an investigator blinded to the conditions. (C) Cells were sorted for CD71+ or CD71- cells at 5 days after drug treatment. 2250 sorted cells were cultured for an additional 9 days in differentiation media. Cells were counted every 3 days. (D) Sorted cells were cultured for an additional 4 days in differentiation media. At 9 days post-culture, cells were incorporated with BrdU, and then stained with BrdU and CD71 antibodies. Stained cells were analyzed by flow cytometry. *P<0.05, **P<0.01, ***P<0.001. d: day; CFU-E: colony-forming unit-erythroid; BFU-E: burst-forming unit-erythroid; CFU-GM: colony-forming unit granulocyte/monocyte; BrdU: bromodeoxyuridine; DMSO: dimethyl sulfoxide.

In summary, inhibition of FOXM1 activity either through decreased phosphorylation, expression (knockdown) or transcriptional activity (chemical inhibitor), increased proliferation of erythroid progenitors, thereby promoting erythropoiesis (Figure 6).

Discussion In the study herein, we found that decreased levels of the FOXM1 protein increased red blood cell production through increased proliferation of human erythroid progenitors. We found that knockdown of FOXM1 by shRNA resulted in an increased number of erythroid cells with decreased myeloid cells. FOXM1 expression was relatively low in the myeloid population, and we suspect that the reduction in the myeloid population is an indirect effect from the promotion of erythropoiesis in our in vitro system. Indeed, we observed that the myeloid population was not affected by FOXM1 downregulation in isolation (Figures 2E and 3B and Online Supplementary Figure S8). In contrast, there is a significant expression of FOXM1 throughout normal human erythropoiesis, suggesting that FOXM1 plays a critical role and may be a unique target for certain forms of anemia. Since FOXM1 is normally associated with proliferation, we would predict that knockdown would inhibit clonal expansion of stem/progenitor cells inhibiting differentiation. Interestingly, our results demonstrated that knockdown of FOXM1 increased proliferation of erythroid progenitors. Our findings suggest that FOXM1 could be a repressor of erythropoiesis. FOXM1 has been shown to function as a transcriptional repressor to inhibit mammary luminal cell differentiation.48 FOXM1 directly represses the 832

messenger RNA (mRNA) expression of GATA3, a master regulator of luminal cell differentiation, through a RBdependent and DNMT-dependent mechanism. We tested whether FOXM1 also has a role as repressor in our in vitro model of hematopoiesis. We observed mRNA levels of known FOXM1 target genes, such as proliferation-associated genes including the cell cycle regulatory genes (CyclinD1, p27, CyclinB1/B2, CyclinA1/A2, PCNA, CDC25b, PLK1, NEK2, CENPF) and the proliferation genes (TGF-Îą, JNK1, IGF-1, NEDD4) in FOXM1 knockdown cells. However, we did not observe any significant changes in their gene expression by FOXM1 knockdown in the pure CD71+ erythroid population as well as in the all mixed CD71+ and CD71â&#x20AC;&#x201C; population (data not shown). We therefore suspect that the effects of FOXM1 on the proliferation of erythroid progenitors are due to alternative mechanisms. We studied the effects of FOXM1 on erythroid population using shRNA knockdown, mutant proteins, and drug inhibitors. There was some discordance among the phenotypes, specifically with regard to the colony forming assays. While FOXM1 knockdown increases CFU-E numbers only, FOXM1 mutants exhibit increased BFU-E and CFU-E. Although we examined three representative kinases, there are still multiple regulatory pathways for FOXM1 activity such as protein stability by ubiquitin-proteasome pathway and protein expression by microRNAs (miRNAs). Therefore, it is conceivable that regulation of FOXM1 function in erythroid progenitors may not solely rely on its phosphorylation. We also studied the effects of CHK2 and CDK1/2 kinase inhibitors on erythropoiesis. These kinases are wellknown to regulate various pathways, including cell cycle pathways and we, therefore, expected that it would be difficult to observe the effects of kinase inhibition on eryhaematologica | 2017; 102(5)


FOXM1 in erythropoiesis

Figure 6. Model of increased erythroid production by FOXM1 deficiency. FOXM1 is functionally activated through the CHK2 or CDK1/2 kinase cascade. When FOXM1 cannot be activated because of certain mutations or defective expression, the proliferation of erythroid progenitors is amplified, resulting in increased erythroid cell production. We also observed that the FOXM1 inhibitor, FDI-6, shows a similar effect to FOXM1 downregulation. CHK2: checkpoint kinase 2; CDK1/2: cyclin-dependent kinase 1/2.

thropoiesis. For this reason, we initially used a range of drug concentrations and found that most of the cells were dead or arrested at the higher concentrations. However, we did observe an increase in the erythroid population and a decrease in the myeloid population without any cell growth defects at the lower concentrations, which is similar to our results with the FOXM1 mutants which cannot be phosphorylated by these kinases. Previous work in zebrafish models using morpholino antisense against foxm1 showed a blood defect in 70% of embryos, which is not consistent with our results.49 We hypothesize that FOXM1 may have different functions during embryogenesis in zebrafish or due to effects in the bone marrow microenvironment. There are heterogeneous cells in the bone marrow microenvironment, and they regulate each other through paracrine as well as autocrine mechanisms. Currently, optimal human cell model systems to study the extrinsic effects of FOXM1 in vitro are not available. In addition, Hou et al. previously reported that in vivo deletion of Foxm1 in mice leads to more proliferation and fewer quiescent cells of CD34+ cells, but did not affect the differentiation of mature blood cells or the frequency of erythroid blasts.38 One consideration is the source of CD34+ cells (i.e., cord blood for our studies and bone marrow for previous studies). We did examine the effects of FOXM1 knockdown and FOXM1 inhibitor FDI-6 in human bone marrow CD34+ cells and we could still observe the same effects (Online Supplementary Figure S20). haematologica | 2017; 102(5)

Therefore, the difference in our findings is likely due to the variability of FOXM1 expression. The aforementioned paper showed higher expression of Foxm1 in primitive hematopoietic cells than in differentiated cells including myeloid cells, B cells, erythroblasts, and T cells. However, we observed that FOXM1 is highly expressed in more differentiated hematopoietic progenitors, especially the erythroid progenitor. Our data suggest that FOXM1 plays a unique role in erythroid cells rather than in CD34+ cells. In addition, there may be differences between the mouse and human hematopoietic systems.39 In summary, our results demonstrate a novel function of FOXM1 in normal human hematopoiesis, in which FOXM1 deficiency leads to an increased proliferation of erythroid progenitors, resulting in an increased erythroid differentiation. These results suggest that FOXM1 inhibitors, such as FDI-6, may be beneficial in treating certain patients with anemia. Acknowledgments The authors would like to thank Narla Mohandas for critical discussion, reading of our manuscript, and sharing the RNA-seq data. This research was funded by NIH R01HL097561, R01DK107286, Department of Defense BM110060 (KMS), K08 DK090145-06 (AN), the Child Health Research Institute at Stanford Postdoctoral Fellowship 1111239-280-JHACT (MY), Stanford Bio-X Undergraduate Summer Research Program (SK), and USHHS Ruth L. Kirschstein Institutional National Research Service Award # T32 CA009056 (EB). 833


M. Youn et al.

References 1. Yu J, Thomson JA. Pluripotent stem cell lines. Genes Dev. 2008;22(15):1987-1997. 2. Ogawa M. Differentiation and proliferation of hematopoietic stem cells. Blood. 1993;81(11):2844-2853. 3. Tsiftsoglou AS, Vizirianakis IS, Strouboulis J. Erythropoiesis: model systems, molecular regulators, and developmental programs. IUBMB Life. 2009;61(8):800-830. 4. Dzierzak E, Philipsen S. Erythropoiesis: development and differentiation. Cold Spring Harb Perspect Med. 2013;3(4): a011601. 5. Gregory CJ, Eaves AC. Three stages of erythropoietic progenitor cell differentiation distinguished by a number of physical and biologic properties. Blood. 1978;51(3):527537. 6. Hattangadi SM, Wong P, Zhang L, Flygare J, Lodish HF. From stem cell to red cell: regulation of erythropoiesis at multiple levels by multiple proteins, RNAs, and chromatin modifications. Blood. 2011;118(24):62586268. 7. Stephenson JR, Axelrad AA, McLeod DL, Shreeve MM. Induction of colonies of hemoglobin-synthesizing cells by erythropoietin in vitro. Proc Natl Acad Sci USA. 1971;68(7):1542-1546. 8. McLeod DL, Shreeve MM, Axelrad AA. Improved plasma culture system for production of erythrocytic colonies in vitro: quantitative assay method for CFU-E. Blood. 1974;44(4):517-534. 9. Iscove NN, Sieber F, Winterhalter KH. Erythroid colony formation in cultures of mouse and human bone marrow: analysis of the requirement for erythropoietin by gel filtration and affinity chromatography on agarose-concanavalin A. J Cell Physiol. 1974;83(2):309-320. 10. Moriyama Y, Fisher JW. Effects of testosterone and erythropoietin on erythroid colony formation in human bone marrow cultures. Blood. 1975;45(5):665-670. 11. Gregory CJ, Eaves AC. Human marrow cells capable of erythropoietic differentiation in vitro: definition of three erythroid colony responses. Blood. 1977;49(6):855-864. 12. Wong P, Hattangadi SM, Cheng AW, et al. Gene induction and repression during terminal erythropoiesis are mediated by distinct epigenetic changes. Blood. 2011; 118(16):e128-138. 13. Gronowicz G, Swift H, Steck TL. Maturation of the reticulocyte in vitro. J Cell Sci. 1984;71:177-197. 14. Zhang J, Randall MS, Loyd MR, et al. Mitochondrial clearance is regulated by Atg7-dependent and -independent mechanisms during reticulocyte maturation. Blood. 2009;114(1):157-164. 15. Kundu M, Lindsten T, Yang CY, et al. Ulk1 plays a critical role in the autophagic clearance of mitochondria and ribosomes during reticulocyte maturation. Blood. 2008;112(4): 1493-1502. 16. Chasis JA, Prenant M, Leung A, Mohandas N. Membrane assembly and remodeling during reticulocyte maturation. Blood. 1989;74(3):1112-1120.

834

17. Waugh RE, Mantalaris A, Bauserman RG, Hwang WC, Wu JH. Membrane instability in late-stage erythropoiesis. Blood. 2001; 97(6):1869-1875. 18. Mel HC, Prenant M, Mohandas N. Reticulocyte motility and form: studies on maturation and classification. Blood. 1977; 49(6):1001-1009. 19. Middleton SA, Barbone FP, Johnson DL, et al. Shared and unique determinants of the erythropoietin (EPO) receptor are important for binding EPO and EPO mimetic peptide. J Biol Chem. 1999;274(20):1416314169. 20. Livnah O, Johnson DL, Stura EA, et al. An antagonist peptide-EPO receptor complex suggests that receptor dimerization is not sufficient for activation. Nat Struct Biol. 1998;5(11):993-1004. 21. Liu Y, Pop R, Sadegh C, et al. Suppression of Fas-FasL coexpression by erythropoietin mediates erythroblast expansion during the erythropoietic stress response in vivo. Blood. 2006;108(1):123-133. 22. Suzuki M, Shimizu R, Yamamoto M. Transcriptional regulation by GATA1 and GATA2 during erythropoiesis. Int J Hematol. 2011;93(2):150-155. 23. Laoukili J, Stahl M, Medema RH. FoxM1: at the crossroads of ageing and cancer. Biochim Biophys Acta. 2007;1775(1):92-102. 24. Wang IC, Chen YJ, Hughes D, et al. Forkhead box M1 regulates the transcriptional network of genes essential for mitotic progression and genes encoding the SCF (Skp2-Cks1) ubiquitin ligase. Mol Cell Biol. 2005;25(24):10875-10894. 25. Clark KL, Halay ED, Lai E, Burley SK. Cocrystal structure of the HNF-3/fork head DNA-recognition motif resembles histone H5. Nature. 1993;364(6436):412-420. 26. Kaestner KH, Knochel W, Martinez DE. Unified nomenclature for the winged helix/forkhead transcription factors. Genes Dev. 2000;14(2):142-146. 27. Tan Y, Raychaudhuri P, Costa RH. Chk2 mediates stabilization of the FoxM1 transcription factor to stimulate expression of DNA repair genes. Mol Cell Biol. 2007;27(3):1007-1016. 28. Costa RH. FoxM1 dances with mitosis. Nat Cell Biol. 2005;7(2):108-110. 29. Wonsey DR, Follettie MT. Loss of the forkhead transcription factor FoxM1 causes centrosome amplification and mitotic catastrophe. Cancer Res. 2005;65(12):5181-5189. 30. Laoukili J, Kooistra MR, Bras A, et al. FoxM1 is required for execution of the mitotic programme and chromosome stability. Nat Cell Biol. 2005;7(2):126-136. 31. Pilarsky C, Wenzig M, Specht T, Saeger HD, Grutzmann R. Identification and validation of commonly overexpressed genes in solid tumors by comparison of microarray data. Neoplasia. 2004;6(6):744-750. 32. Koo CY, Muir KW, Lam EW. FOXM1: From cancer initiation to progression and treatment. Biochim Biophys Acta. 2012; 1819(1):28-37. 33. Radhakrishnan SK, Gartel AL. FOXM1: the Achilles' heel of cancer? Nat Rev Cancer. 2008;8(3):c1; author reply c2. 34. Karsten SL, Kudo LC, Jackson R, et al. Global

35.

36.

37.

38.

39.

40.

41. 42.

43.

44.

45.

46.

47.

48. 49.

analysis of gene expression in neural progenitors reveals specific cell-cycle, signaling, and metabolic networks. Dev Biol. 2003;261(1): 165-182 Ahn JI, Lee KH, Shin DM, et al. Temporal expression changes during differentiation of neural stem cells derived from mouse embryonic stem cell. J Cell Biochem. 2004; 93(3):563-578. Lee D, Prowse DM, Brissette JL. Association between mouse nude gene expression and the initiation of epithelial terminal differentiation. Dev Biol. 1999; 208(2):362-374. Xie Z, Tan G, Ding M, et al. Foxm1 transcription factor is required for maintenance of pluripotency of P19 embryonal carcinoma cells. Nucleic Acids Res. 2010; 38(22):80278038. Hou Y, Li W, Sheng Y, et al. The transcription factor Foxm1 is essential for the quiescence and maintenance of hematopoietic stem cells. Nat Immunol. 2015;16(8):810818. An X, Schulz VP, Li J, et al. Global transcriptome analyses of human and murine terminal erythroid differentiation. Blood. 2014;123(22):3466-3477. Hu J, Liu J, Xue F, et al. Isolation and functional characterization of human erythroblasts at distinct stages: implications for understanding of normal and disordered erythropoiesis in vivo. Blood. 2013;121(16): 3246-3253. Pandit B, Halasi M, Gartel AL. p53 negatively regulates expression of FoxM1. Cell Cycle. 2009;8(20):3425-3427. Bibikova E, Youn MY, Danilova N, et al. TNF-mediated inflammation represses GATA1 and activates p38 MAP kinase in RPS19-deficient hematopoietic progenitors. Blood. 2014;124(25):3791-3798. Moniz H, Gastou M, Leblanc T, et al. Primary hematopoietic cells from DBA patients with mutations in RPL11 and RPS19 genes exhibit distinct erythroid phenotype in vitro. Cell Death Dis. 2012;3:e356. Chen YJ, Dominguez-Brauer C, Wang Z, et al. A conserved phosphorylation site within the forkhead domain of FoxM1B is required for its activation by cyclin-CDK1. J Biol Chem. 2009;284(44):30695-30707. Major ML, Lepe R, Costa RH. Forkhead box M1B transcriptional activity requires binding of Cdk-cyclin complexes for phosphorylation-dependent recruitment of p300/CBP coactivators. Mol Cell Biol. 2004;24(7):26492661. Fu Z, Malureanu L, Huang J, et al. Plk1dependent phosphorylation of FoxM1 regulates a transcriptional programme required for mitotic progression. Nat Cell Biol. 2008;10(9):1076-1082. Gormally MV, Dexheimer TS, Marsico G, et al. Suppression of the FOXM1 transcriptional programme via novel small molecule inhibition. Nat Commun. 2014;5:5165. Carr JR, Kiefer MM, Park HJ, et al. FoxM1 regulates mammary luminal cell fate. Cell Rep. 2012;1(6):715-729. Eckfeldt CE, Mendenhall EM, Flynn CM, et al. Functional analysis of human hematopoietic stem cell gene expression using zebrafish. PLoS Biol. 2005;3(8):e254.

haematologica | 2017; 102(5)


ARTICLE

Hemostasis

Obstetric antiphospholipid syndrome: early variations of angiogenic factors are associated with adverse outcomes

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Éva Cochery-Nouvellon,1,2 Érick Mercier,1,2,3 Sylvie Bouvier,1,2,3 Jean-Pierre Balducchi,4 Isabelle Quéré,2,5 Antonia Perez-Martin,2,6 Eve Mousty,7 Vincent Letouzey7 and Jean-Christophe Gris1,2,3

Department of Hematology, University Hospital, Nîmes; 2Research team UPRES EA2992 "Caractéristiques féminines des dysfonctions des interfaces vasculaires CaFeDIVa", University of Montpellier; 3Laboratory of Hematology, Faculty of Pharmacy and Biological Sciences, University of Montpellier; 4Department of Internal Medicine, University Hospital, Nîmes; 5Department of Vascular Medicine and Internal Medicine, University Hospital, Montpellier; 6Department of Vascular Medicine, University Hospital, Nîmes and 7Department of Gynecology and Obstetrics, University Hospital, Nîmes, France 1

Haematologica 2017 Volume 102(5):835-842

ABSTRACT

T

he prognostic value of angiogenic factors in newly pregnant women with obstetric antiphospholipid syndrome (oAPS) has not been documented. We observed 513 oAPS who experienced three consecutive spontaneous abortions before the 10th week of gestation or one fetal loss at or beyond the 10th week. We assessed the plasma concentrations of the proangiogenic factor placenta growth factor (PIGF) and of the antiangiogenic factor soluble fms-like tyrosine kinase-1 on the eve and on the 4th day of the low-molecular weight heparin-low-dose aspirin treatment. Placenta growth factor and fms-like tyrosine kinase-1 plasma concentrations showed marked increases. Treatment-associated variations of PIGF and of soluble fms-like tyrosine kinase-1 were antagonist risk factors for placenta-mediated complications (PMC) and for severe PMC, for fetal death, stillbirth and neonatal death. The ratio between PIGF increase and soluble fms-like tyrosine kinase-1 was a summary variable whose best cut-off values (1.944.10-2) had high negative predictive values for PMC (0.918) and may be used to help rule out the development of PMC in evolutive pregnancies after 19 completed weeks. The early variations of PIGF and soluble fms-like tyrosine kinase-1 concentrations in newly pregnant oAPS may help to detect patients at low risk of PMC. (clinicaltrials.gov identifier: 02855047)

Introduction Antiphospholipid antibody syndrome (APS) is defined by precise clinical symptoms associated with repeated positive results in laboratory tests for IgG or IgM antiphospholid antibodies (aPLAbs) such as lupus anticoagulant (LA), anticardiolipin antibody (aCL) and anti-β2 glycoprotein I antibody (aβ2GP1), present at moderate or high titers for diagnosis. Purely obstetric APS (oAPS) is an APS clinical presentation characterized by precisely defined morbidities occurring during pregnancy in women with no history of thrombosis.1-5 The use of low-dose aspirin (LDA) in association with heparin is generally recommended for the treatment of oAPS.6 Among heparins, low-molecular weight heparins (LMWH) are favored for safety reasons and practical considerations. In oAPS defined on pregnancy loss criteria, the LDA-LMWH association, however, shows some limitations.7 Available data support the hypothesis that alterations in placental trophoblast production of angiogenic proteins may help to explain the pathogenesis of preeclampsia,8 placental abruption,9 intrauterine growth restriction, and stillbirths.10 The overall theory is that excessive release of antiangiogenic factors, such as soluble vascular endothelial growth factor (VEGF) receptor (sFlt1), by haematologica | 2017; 102(5)

Correspondence: jean.christophe.gris@chu-nimes.fr

Received: August 23, 2016. Accepted: January 20, 2017. Pre-published: January 25, 2017.

doi:10.3324/haematol.2016.155184 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/835 ©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.

835


E. Cochery-Nouvellon et al.

hypoxic trophoblasts, antagonize proangiogenic factors such as placenta growth factor (PlGF). This causes aberrant placental angiogenesis and vasculogenesis.11 In pregnant women receiving LMWH therapy, heparin increases circulating sFlt12 and PlGF13 immunoreactivity. Results of the PROMISSE study recently showed that circulating angiogenic factors, measured as early as 12-15 weeks in heterogeneous pregnant women with systemic lupus erythematous and/or aPLAbs, have a high negative predictive value in ruling out the development of severe adverse outcomes.14 The Nimes Obstetricians and Hematologists – AntiphosPholipid Syndrome (NOH-APS) observational cohort study7 allowed us to evaluate the prognostic value on pregnancy outcomes of early sFlt1 and PlGF variations associated with the start of the LDA-LMWH treatment in oAPS women.

Methods Patients This study focuses on a subgroup of patients included in the NOH-APS cohort: those with an oAPS diagnosis who were treated for a new pregnancy during the 18 months individual observational period after diagnosis. The time window for recruitment was January 1st 1995 to January 1st 2005. Clinical follow up started on July 1st 1995 and ended on September 1st 2007. None of the observed treated pregnant oAPS women was lost to follow up. Patient recruitment and definition of the patient groups included in the NOH-APS study have been described in detail elsewhere7,15 (Online Supplementary Figure S1). Briefly, all women fulfilled one of the following inclusion criteria: 1) three unexplained consecutive spontaneous abortions before the 10th week of gestation that could not be accounted for by maternal anatomic or hormonal abnormalities, or paternal or maternal chromosomal causes (the recurrent embryo loss subgroup); or 2) one unexplained death of a morphologically normal fetus (fetal loss) at or after the 10th week of gestation, with the normal morphology of the fetus confirmed by ultrasound scan or direct examination of the fetus (fetal loss subgroup). The exclusion criteria were: a history of thrombotic events (at least one clinical episode of venous, arterial, or small-vessel thrombosis in any tissue or organ other than the placenta confirmed by objective validated criteria, ie. unambiguous findings in appropriate imaging or histological studies), or any treatment given during previous pregnancies that might have modified the natural course of the condition, such as antithrombotic agents, or immunosuppressive or immunomodulatory drugs. We also excluded women whose pregnancy losses could be explained by infectious, metabolic, anatomic or hormonal factors. Women seropositive for HIV, or hepatitis B or C were also excluded. Patients were classified as primary aborters (no previous successful pregnancy) or secondary aborters. Complete thrombophilia screening tests were systematically performed, leading to the definition of subgroups.15 Among the 4801 screened patients for thrombophilia15 (Online Supplementary Figure S1), the present study focuses on the APS subgroup, conventionally defined as women being persistently positive for LA, and/or aCL, and/or aβ2GP1 (initially: n=517) assayed as described.7,15 This subgroup was restricted to the women who initiated a new pregnancy during the 18 months observational period after oAPS diagnosis (n=513)7 (Online Supplementary Figure S1). Pregnant oAPS women were regularly followed by obstetricians involved in the NOHA network, as needed, and were sys836

tematically evaluated once a month by internists and hematologists in our outpatient department of hematology, with no loss to follow up. Any initial missing values for clinical characteristics could be obtained during the subsequent consultations and with the help of the network obstetricians and general practitioners. The study was approved by the University Hospital of Nîmes Institutional Review Board and ethics committee and by the local Comité de Protection des Personnes Soumises à la Recherche Biomédicale. This clinical investigation was conducted in accordance with the Declaration of Helsinki of 1975 as revised in 1996. All the women gave their informed consent to participate. Financial support was provided by Nîmes University hospital via an internal funding scheme; the funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Antithrombotics during new observed pregnancies Low-molecular weight heparin [enoxaparin, 40 mg per day (4000 U/day)] was added to LDA (100 mg/day) from the day of positive pregnancy test result until delivery; these two treatments were administered concomitantly. Compliance to LMWH treatment was monitored by self-declaration of the patients and their partners, and systematic examination of the subcutaneous injection sites at each medical examination. Platelet counts were checked on the day before the first LMWH injection, twice a week during the first three weeks, and then once a month.

Outcomes The primary end point was a composite outcome that included any of the following events occurring after 19 completed weeks of the observed pregnancy: preeclampsia, abruptio placenta, or small-gestational age newborn (< 10th percentile), summarized as the so-called placenta-mediated complications (PMCs). The secondary outcome analysis included a composite outcome that included any of the severe PMCs: severe preeclampsia, severe small-gestational-age newborn less than 5th percentile, abruptio placenta leading to emergency delivery, pregnancy loss categorized as embryonic loss (before 10 week’s gestation, WG), fetal death (before 20 WG), stillbirths (from 20 WG to delivery), and neonatal death defined before reaching 28 days of age. The diagnosis of preeclampsia was the association of systolic blood pressure 140 mmHg or over or diastolic blood pressure 90 mmHg or over in a woman who was normotensive before 20 weeks’ gestation and a significant proteinuria defined as the presence of 0.3 g or more of protein in a 24-hour urine specimen.16 The diagnosis of severe preeclampsia was made according to American College of Obstetricians and Gynecologist criteria. Briefly, preeclampsia was considered severe if one or more of the following criteria was present: systolic blood pressure 160 mmHg or over or diastolic blood pressure 110 mmHg or over on two occasions at least six hours apart while the patient is on bed rest, proteinuria 5 g or over in a 24-hour urine specimen or 3+ or greater on two random urine samples collected at least four hours apart, oliguria less than 500 mL in 24 hours, eclamptic seizures, persistent headache or visual disturbances, abruptio placenta, pulmonary edema or cyanosis, epigastric or right upper-quadrant pain, impaired liver function (twice the normal range), thrombocytopenia (<100,000 cellsx10-9L), severe fetal growth restriction (< 5th percentile).17 Abruptio placenta was defined according to classical clinical prenatal signs and symptoms: vaginal bleeding accompanied by nonreassuring fetal status or uterine hypertonicity, or sonographic visualization of abruption, and evidence of retroplacental clots during examination of the delivered placenta. Cases were confirmed by histopathological diagnosis. Birthweights were assessed by birthweight percentile charts haematologica | 2017; 102(5)


Angiogenic factors in obstetric APS

customized for maternal age, pre-pregnancy body mass index, parity, gestational age at delivery, and sex.18 Small-gestational-age newborn was defined as birthweight under the 10th percentile, and severe when under the 5th percentile. Chronic hypertension was defined as hypertension (blood pressure ≥140 mmHg systolic or 90 mmHg diastolic or more) that was present before pregnancy or that was diagnosed before the 20th week of gestation.16 The diagnosis of superimposed preeclampsia in chronic hypertensive women needed one of the following findings: 1) in women with hypertension and no proteinuria early in pregnancy (<20 weeks’ gestation), new-onset proteinuria as defined above; 2) in women with hypertension and proteinuria before 20 weeks’ gestation; 3) sudden increase in proteinuria; 4) in women whose hypertension has previously been well controlled sudden increase in blood pressure; 5) thrombocytopenia (platelet count <100,000 cellsx10-9 L); or 6) increase in alanine aminotransferase or aspartate aminotransferase to abnormal levels.16 HELLP syndrome was defined by the presence of all 3 of the following criteria: hemolysis [characteristic peripheral blood smear and serum lactate dehydrogenase (LDH) ≥ 600 U/L or serum total bilirubin ≥ 20 mM/L-1], elevated liver enzymes [serum aspartate aminotransferase (AST) ≥ 70 U/L], and low platelet counts (< 100,000 cellsx10-9 L).19

Samples We used blood samples collected for platelet monitoring under LMWH treatment. EDTA-anticoagulated blood samples were obtained by clean venipuncture: the first one the day before LMWH starting, the second one the 4th day of LMWH treatment, four hours after subcutaneous injection; both at 11.00 am (+ 30 minutes). After taking a platelet count, whole blood samples were centrifuged twice at 4000 g for 20 minutes, aliquots of plateletpoor plasma were then stored at -80°C until tested.

Assays Stored plasma samples were subsequently used for sFlt1 and PlGF plasma level measurements, performed in a blind fashion for PMC outcome, and in duplicate using enzyme-linked immunosorbent assay kits (R&D Systems Europe, Lille, France). The calculated interassay coefficients of variation for sFlt-1 and PlGF were 4.0% and 7.5%, respectively; the calculated intraassay coefficients of variation were 2.5% and 4.5%, respectively. In order to minimize interassay variations, plasmas obtained from a given patient before and during LMWH treatment were assayed at the same time using the same standard curve (paired testing). We had no missing blood sample; 2.3% of the samples had to be assayed a second time due to a first accidental technical failure. Complete biological results were obtained for all study subjects.

Statistical analysis Quantitative data are presented as the median, interquartile range (IQR) and minimum-maximum values. Qualitative data are presented as values and percentages. Mann-Whitney test, KruskalWallis test, χ² tests and Fisher’s exact tests were used, as appropriate, for comparisons between baseline characteristics. All analyses were based on pregnancy outcomes that occurred during the first pregnancy after oAPS diagnosis, treated as described above. Variations in the plasma concentrations of the tested angiogenic factors, defined as the difference “D” between plasma concentrations on day 4 of LMWH-LDA treatment and before treatment, were calculated for each patient and used for analysis (DPlGF; DsFlt1). The DPlGF/DsFlt1 ratio was analyzed as a continuous variable and after categorization into quartiles. A priori selected factors of the various outcomes (primary outhaematologica | 2017; 102(5)

come: PMCs; secondary outcomes: severe PMCs, pregnancy loss/neonatal death: abortion, fetal death, stillbirth, neonatal death) among the clinical predictors (age, body mass index, thrombotic familial antecedents, pregnancy loss familial antecedents, ethnicity, smoking history, pre-existing diabetes mellitus, preexisting hypertension, pre-existing embryonic/fetal pregnancy loss, primary/secondary pregnancy loss and initial inflammatory disease), the metabolic markers at inclusion (hypercholesterolemia defined as a fasting cholesterol concentration above 5.2 mM/L-1 and hypertriglyceridemia as a fasting triglyceride concentration above 1.7 mM/L-1), the biological predictors at inclusion (positive LA, positive aCL-G, positive aCL-M, positive aβ2GP1-G, positive aβ2GP1-M, triple positivity and the F5 rs6025 or F2 rs1799963 polymorphisms) and the variations of angiogenic factors associated with the LMWH-LDA treatment (DPlGF; DsFlt1; or the DPlGF / DsFlt1 ratio) were evaluated first by univariate then by multivariate logistic regression analysis. For multivariate models, and because not all confounders were known for our study model, a stepwise backward elimination was performed after selecting all the variables identified by the univariate models as potential predictors at P<0.20, with adjustment being finally performed for all variables with P<0.20 in the multivariate models. The final model only included main effects with P<0.05. The goodness-of-fit of the model was assessed by the Hosmer-Lemeshow test. Discrimination was assessed using computing receiver operating characteristic (ROC) curves and calculating the area under the curve (AUC). The best cut-off value was identified as the point maximizing the computed Youden index (sensivity Se + specificity Sp -1). All tests were two-sided and assessed at the 5% significance level. The study design was based on our recruitment capacities assessed over a 10-year period, thus no sample size calculation was performed. However, hypothesizing a minimum clinically relevant relative risk of PMC of 10% and a 0.05 2-sided alpha level the study power to detect a 50% increase in the relative risk with the 513 observed patients is 0.935. Statistical analyses were performed using StatView®-windows software v.5.0 (SAS Institute Inc., Cary, NC, USA) and XLSTAT® software v.2015.4.01.20116 (Addinsoft SARL, Paris, France).

Results Patients' characteristics and new pregnancy outcomes Patients' characteristics are available in the Online Supplementary Table S1 and new pregnancy outcomes are shown in Table 1. Most of the patients were non-overweight Caucasian European women, under 35 years of age, with no positive family history of venous or arterial thrombotic disease or significant pregnancy loss (Online Supplementary Table S1). Primary oAPS and fetal death were the main clinical presentation and the main criterion, respectively. Pre-existing hypertension or/and hyperlipidemia were present in a minority of the patients and only one out of 10 women were current tobacco smokers. ACL-M or LA were the most frequent aPlAb features. Seventy-five percent of the patients were positive for more than one marker and approximately 20% of the women were positive only for aCL-Ab (8.5% had only aCL-M antibodies). None had only aβ2GP1-Ab. Over 25% of the patients exhibited triple positivity, which was due in part to the high frequency of aCL-M-positive patients. 837


E. Cochery-Nouvellon et al.

In the APS group, 3.3% of women were positive for the F5 rs6025 or F2 rs1799963 polymorphisms. Despite the systematic prescription of the LMWH-LDA treatment during new pregnancies, only 75% of the women had an ongoing viable pregnancy after 19 completed weeks of gestation, with an 18.1% rate of abortions before 10 WG and a 7.2% rate of fetal deaths after 9 completed WG (Table 1). Fetal karyotype investigations on early loss before ten weeks through genomic hybridization on miscarriage tissue was only available in 78.5% of the cases and evidenced a significant rate of chromosomal abnormalities, mainly aneuploidies, in 59% of the investigated cases. The same test performed on fetal death was available for all cases and showed a 5.4% chromosomal abnormality rate, all in fetal loss cases during weeks 10 to 12. An additional 5.7% of the women experienced stillbirth (normal karyotypes). Neonatal death before 28 days was reported in 3.5% of the women (normal karyotypes). PMCs were diagnosed in 17.7% of the women, the main feature being preeclampsia (11.1%; severe preeclampsia: 7.8%), followed by small-gestational-age newborn (14.2%; severe small-gestational-age newborn: 5.1%) and abruptio placenta (2.1%).

Safety outcomes outcomes for women taking antithrombotic treatment No treatment had to be stopped or replaced for safety reasons. There were no cases of heparin-induced thrombocytopenia.

Angiogenic factors The LMWH-LDA treatment was globally associated with a marked increase of PlGF plasma concentrations and a strong increase of sFlt1 concentrations, leading to a collapse of the PlGF/sFlt1 ratio values: in the whole oAPS group as well as in the two subgroups of women who later developed or did not develop any PMC (Table 2). The

treatment-associated variations in PlGF and in sFlt1 concentrations were correlated (Spearman’s rank-order correlation coefficient: 0.424; P<0.0001) (Figrue 1). Focusing on angiogenic factors on the 4th day of treatment, PlGF concentrations were lower, sFlt1 concentrations higher and the PlGF/sFlt1 values lower in women who finally developed PMCs than in those who did not (Table 2). Treatment-associated variations of PlGF and of sFlt1 concentrations were respectively lower and higher in women who developed PMCs than in those who did not (Table 2 and Figure 2). Consequently, the ratio of the treatmentassociated variations of PlGF and of sFlt1 plasma concenTable 1. Pregnancy outcomes of the antiphospholipid syndromes (APS) women who initiated a new pregnancy during the 18 months individual observational period after obstetric APS diagnosis.

New pregnancy outcomes

N (%)

Abortion < 10 WG Fetal death ≥ 10 WG and < 20 WG Ongoing pregnancies at 20 WG Stillbirths ≥ 20 WG to delivery Any pregnancy loss Perinatal death > 22 WG and < 8 days Neonatal death < 28 days PE Severe PE Abruptio placenta SGA < p10 SGA newborn < p5 PMCs

93 (18.1%) 37 (7.2%) 83 (74.7%) 29 (5.7%, 7.6%) 159 (31%) 43 (8.4%, 11.2%) 18 (3.5%, 4.7%) 57 (11.1%, 14.9%) 40 (7.8%, 10.4%) 11 (2.1%, 2.9%) 73 (14.2%, 19.1%) 26 (5.1%, 6.8%) 91 (17.7%, 23.8%)

When two percentages are given, the first is calculated on the whole number of pregnancies, the second is restricted to ongoing pregnancies at 20 weeks of gestation (WG). PE: preeclampsia; SGA: small gestational age; p10: 10th percentile; p5: 5th percentile; PMC: placenta-mediate complications; PE and/or abruptio placenta and/or fetal growth restriction.

Table 2. Levels of angiogenic factor during early pregnancy in obstetric antiphospholipid syndromes (APS) women (oAPS) receiving the low-molecular weight heparins-low-dose aspirin (LMWH-LDA) treatment.

Angiogenic factors PlGF, ng.L-1 Before treatment During treatment P D sFlt1, ng.L-1 Before treatment During treatment P D PlGF:sFlT1 ratio Before treatment During treatment P DPlGF/DsFlt1

oAPS Whole group (n=513)

PMC-negative (n=422)

PMC-positive (n=91)

P

9.3 [1.68] (5.6-12.6) 31.8 [5.97] (16.8-49.5) < 0.0001 22.5 [6.21] (5.9-39.4)

9.3 [1.71] (5.6-12.6) 31.9 [5.57] (16.8-49.5) < 0.0001 22.6 [6.18] (5.9-39.4)

9.3 [1.64] (5.9-12.3) 29.8 [7.1] (16.8-44.8) < 0.0001 20.5 [7.1] (9.1-36.3)

0.88 0.001

32 [20] (1-81) 1283 [728] (72-2992) < 0.0001 1244 [718] (30-2928)

31 [20] (1-81) 1247 [697] (72-2992) < 0.0001 1216 [712] (30-2928)

34 [21] (2-62) 1462 [782] (293-2560) < 0.0001 1429 [756] (265-2511)

0.296 [0.234] (0.069-11.3) 0.025 [0.012] (0.012-2.26) < 0.0001 0.018 [0.009] (0.007-0.631)

0.302 [0.227] (0.069-11.3) 0.026 [0.013] (0.012-2.26) < 0.0001 0.018 [0.01] (0.008-0.631)

0.265 [0.274] (0.099-5.42) 0.021 [0.009] (0.012-0.089) < 0.0001 0.015 [0.007] (0.007-0.063)

0.0016 0.39 0.0006 0.0007 0.851 < 0.0001 <0.0001

Before treatment then on the 4th day of treatment (During treatment); in the whole group of patients and according to the subsequent development of placenta-mediated complications (PMC; PMC negative vs. PMC positive). Results are given as median, (interquartile range) and (minimum-maximum values). D: levels during treatment minus levels before treatment.

838

haematologica | 2017; 102(5)


Angiogenic factors in obstetric APS

trations, namely the (DPlGF/DsFlt1) variable, was lower in women who developed PMCs (Table 2 and Figure 2). We systematically looked for any associations between each of the five APS markers, the aPLAbs, and the angiogenic factors (basal concentrations, treatment-associated variations and ratio of the treatment-associated variations). Only basal sFlt1 concentrations and ab2Gp1-IgG titers were correlated (Spearmansâ&#x20AC;&#x2122; rank correlation coefficient r: 0.117; P=0.010) and higher values of sFlt1 were evidenced in ab2GP1-IgG positive women (P=0.0105). Assuming that the fixed dose of 40 mg enoxaparin may not fit all women, we looked for any correlations between basal concentrations of angiogenic factors and their releases, and BMI values: no significant association could be evidenced (data not shown). No significant associations between BMI values and PMCs, severe PMCs, pregnancy loss before ten weeks, fetal death, stillbirth and neonatal death in patients receiving enoxaparin were observed (data not shown).

Risk factors for PMCs Risk factors for PMCs are shown in Table 3. PMCs are pathologies arising after the 19th completed WG; therefore, risk factors for PMCs were investigated in the subgroup of 383 women with an ongoing pregnancy during the 20th gestational week. The treatment-associated variations of PlGF concentrations (DPlGF) were overall highly significant protecting factors against the development of PMCs whereas variations of sFlt1 (DsFlt1) were highly significant risk factors for PMCs. Both DPlGF and DsFlt1 remained independent indicators after adjustment for prior fetal death and triple positivity. The analysis of the summary variable (DPlGF/DsFlt1) ratio showed values belonging to the 2nd, 3rd and 4th quartiles to protect against the occurrence of PMCs even after adjustment for prior fetal death

and for a triple positivity for aPLAbs, the risk values decreasing when moving to the highest quartile. The AUC related to the ROC computed from the (DPlGF/DsFlt1)*102 ratio values (Figure 2) was 0.745 (0.686-0.803; P<0.0001). Looking for any cut-off values, 1.944 maximized Youden index and was mainly associated with an interesting negative predictive value NPV for PMCs (0.918; negative likelihood ratio LR-: 0.286), thus ruling out PMCs, whereas its positive predictive value PPV was poor (0.363; positive likelihood ratio LR+: 1.830).

Risk factors for secondary outcomes The study of severe PMCs also pointed out increasing DPlGF values to protect against severe PMCs and increasing DsFlt1 values to expose to severe PMCs (Online Supplementary Table S2). Indeed, the three highest quartiles of the (DPlGF/DsFlt1) ratio were also associated with a decreasing clinical risk, which was the lowest for values belonging to the upper quartile. These associations persisted after adjustment for prior fetal death and age. The AUC related to the ROC computed from the (DPlGF/DsFlt1)*102 ratio values was 0.737 (0.666-0.809; P<0.0001). The 1.589 value maximized Youden index and was mainly associated with an interesting NPV (0.927; LR-: 0.431) but with a poor PPV value (0.328; LR+: 2.679). Angiogenic factors best predicted severe preeclampsia [area under the ROC analyzing the (DPlGF / DsFlt1) *102 ratio in predicting severe preeclampsia: 0.769 (0.6880.850)]. The 1.967 value maximized Youden index; its NPV was 0.982 (LR-: 0.118) with a poor PPV value (0.171; LR+: 1.763). The study of risk factors for difficult pregnancies within the whole cohort, ie. which were not concluded by a viable neonate at 28 days of age, led to final results which depended on the type of pregnancy failure. Angiogenic factors could not predict the occurrence of

Figure 1. Correlation between the lowmolecular weight heparins-low-dose aspirin (LMWH-LDA) treatment-associated early variations of the proangiogenic factor plancenta growth factor (PIGF) plasma concentrations (DPlGF) and of sFlt1 plasma concentrations (DsFlt1). Spearman coefficient of rank correlation 0.446, P<0.0001.

haematologica | 2017; 102(5)

839


E. Cochery-Nouvellon et al.

abortions/embryonic losses before 10 WG, which was the most frequent event; in this setting, an abnormal karyotype was a huge clinical predictor. An abnormal karyotype also strongly predicted fetal death. Deaths occurring after nine completed weeks, ie. fetal losses, stillbirths and neonatal deaths, were highly associated with the angiogenic factor variations during the onset of the LMWHLDA treatment, increasing DPlGF values and increasing DsFlt1 values being negative and positive risk factors, respectively, with a significant progressive risk decrease from the second to the fourth quartiles of the (DPlGF / DsFlt1) *102 ratio values. Multivariate analysis also confirmed prior fetal death to predict for recurrence; familial thromboembolism and familial atherothrombosis to independently increase the risk of stillbirth and maternal tobacco smoking or triple positivity for aPLAbs to enhance the risk of neonatal death. We thus analyzed the discrim-

ination power of the (DPlGF / DsFlt1)*102 ratio values on the occurrence of non-embryonic unhappy pregnancies in the 383 women with an ongoing pregnancy at 10 WG. The AUC was 0.725 (0.662-0.789; P<0.0001). The 1.545 value maximized Youden index; its NPV was 0.885 (LR-: 0.522) with a poor PPV value (0.381; LR+: 2.458).

Discussion With this observational study we describe for the first time the early systemic blood variations of two angiogenic factors: PlGF, an agonist of placenta development, and sFlt1, a PlGF antagonist, in newly pregnant oAPS women clinically defined on the basis of previous pregnancy loss and receiving the LMWH-LDA treatment. Our observations highlighted significant PlGF and sFlt1 increases, which in turn have antagonistic effects on the risk of PMCs (PlGF: protecting, sFlt1: precipitating) of severe PMCs, and of fetal death / stillbirth / neonatal death. As a consequence, the categorization of the PlGF:sFlt1 ratio values into quartiles allowed a decrease of risks to be described as values jumped from the lower to the higher quartiles. Interestingly, a negative predictive value for the risk of PMCs and of severe PMCs were obtained, which may help to rule out the ulterior development of these still abnormally prevalent syndromes in conventionally-treated oAPS women.7

Table 3. Risk factors of placenta-mediated complications (PMCs) in antiphospholipid syndromes (APS) women.

Models

P

Variables

OR (95%CI)

Fetal death aCL-M Triple positivity DPlGF.10-1 DsFlt1.10-3 (DPlGF/DsFlt1).102 Q1 (0.75-1.39) Q2 (1.40-1.77) Q3 (1.78-2.32) Q4 (2.33-63.1)

1.772 (1.072-2.89) 1.818 (1.033-3.232) 1.637 (0.995-2.692) 0.321 (0.188-0.548) 2.852 (1.784-4.556)

0.025 0.041 0.052 < 0.0001 < 0.0001

1 0.348 (0.189-0.640) 0.163 (0.080-0.332) 0.083 (0.035-0.197)

0.0007 < 0.0001 < 0.0001

DsFlt1.10-3 DPlGF.10-1 Fetal death

11.19 (5.688-22.0) 0.071 (0.034-0.150) 2.088 (1.166-3.739)

< 0.0001 < 0.0001 0.0133

(DPlGF/DsFlt1).102 Q1 (0.75-1.39) Q2 (1.40-1.77) Q3 (1.78-2.32) Q4 (2.33-63.1) Fetal death Triple positivity

1 0.309 (0.163-0.586) 0.145 (0.070-0.302) 0.065 (0.026-0.159) 1.903 (1.097-3.3.3) 1.893 (1.083-3.308)

0.0003 < 0.0001 < 0.0001 0.0220 0.0250

Univariate

Multivariate 1*

Multivariate 2**

Figure 2. Variations of the proangiogenic factor placenta growth factor (PIGF) plasma concentrations (DPlGF), of sFlt1 plasma concentrations (DsFlt1) and of the DPlGF. Δ sFlt1 ratio (ΔPlGF/ΔsFlt1) associated with the beginning of the low molecular weight heparin–low dose aspirin treatment in obstetrical APS women who later developed (PMC-Pos) or did not develop (PMC-Neg) placenta-mediated complications.

840

PMCs occurred in 91 of the 383 women with an ongoing pregnancy after the 19th completed gestational week. Univariate analysis: results are restricted to putative predictors with P-value <0.20. Multivariate analysis: final models include only main effects with P<0.05, adjusted for variables with P<0.20. aCL-M: positive for anticardiolipin IgM. D: levels during treatment minus levels before treatment.Q1: first quartile; Q2: second quartile; Q3: third quartile; Q4: fourth quartile. *Multivariate model 1 does not include the [(DPlGF/DsFlt1). 102] variable; adjusted for triple positivity. **Multivariate model 2 does not include the [DPlGF. 10-1] and the [DsFlt1 . 10-3] variables.

haematologica | 2017; 102(5)


Angiogenic factors in obstetric APS

Administration of LMWH to preeclamptic women to enhance sFlt1 renal elimination has been described.20 The proposed mechanism is a direct binding leading to mask sFlt1 positive charges.20 The SFlt1 molecule is too big (â&#x2030;&#x2C6;100 kDa) to be filtered into urine in the absence of renal damage. The range of renal manifestations associated with APS has been broadened.21 Renal function and the sFlt1 renal loss may thus be a strong modulating factor impacting on the negative prognostic value of circulating sFlt1. On the other hand, PlGF is a much smaller protein (â&#x2030;&#x2C6;30 kDa) which is readily filtered; however, decreased urinary PlGF at mid gestation is strongly associated with subsequent early development of preeclampsia.22 Future studies must investigate the putative impact of LMWHassociated sFlt1 and PlGF renal clearances on APS pregnancies. Unfractionated heparin, 70 IU/kg body weight, intravenously administered over one minute has been described to increase plasma sFlt1 and PlGF in non-pregnant women with previous preeclampsia and uncomplicated pregnancy.23 This resulted in a large vascular store, partly glycocalix-bound, of releasable sFlt1 and a smaller store of releasable PlGF.23 VEGF/PlGF local regulation partly depends on the availability of their decoy receptor sFlt1, determined by sFlt1 local storage versus its systemic release. Heparin displaces natural sFlt1 stores from vascular smooth muscle cells and vessel walls in vitro, a phenomenon which is regulated by heparanase.24 The same phenomenon is observed with human term placental villi explants.24 As strong trophoblastic involvement is unlikely in our patients with an early new pregnancy, the treatment-associated PlGF and sFlt1 variations we could observe are presumably due to the heparin-mediated mobilization of vascular stores. A recent study investigated the dysregulation of angiogenic factors in women with systemic lupus and/or antiphospholipid antibodies in relation to the development of pregnancy complications.14 In this study, aPlAb positive patients were a minority (32%; 20% with only aPlAbs), treatments were non-uniformly applied, the first blood sample was taken at 6-11 weeks without any basal evaluation, the primary outcome was the development of adverse pregnancy outcomes (a composite of preeclampsia at any time, fetal death > 12 weeks, neonatal death prior to hospital discharge due to prematurity, preterm delivery < 36 weeks due to any placenta-mediated complications and small for gestational age < 5th percentile at birth). Despite these significant heterogeneities, a high NPV ruling out the development of severe adverse outcomes could be seen for PlGF and sFlt1 among the 12th-15th week measures, which is in line with the observations of our study. Furthermore, our findings are consistent with available studies demonstrating a deleterious association between an angiogenic imbalance and pregnancy complications in which a dysfunctional placental organ is thought to play a key pathophysiological role. The similar risk profiles for preeclampsia, abruptio placenta and fetal growth restriction provide compelling evidence to suggest that these conditions may share common pathophysiological mechanisms.25 Pro- and anti-angiogenic factors are dysregulated in patients with APS.26-28 They are also dysregulated in pregnant women with PMCs.29 In vitro studies showed that the modulation of 1st-trimester trophoblast angiogenic factor secretion by aPLAbs is insensitive to heparin.30 Thus, the early variations of angiogenic factors may play a direct haematologica | 2017; 102(5)

Figure 3. Receiver operating characteristic curve analyzing the discrimination power of the [(DPlGF / DsFlt1) * 102] ratio in predicting placenta-mediated complications.

role in the outcome of APS pregnancies. Given this, heparin injections may serve as a truly dynamic sensitization test for assessing active mediators poorly accessible to plasma exploration in usual conditions. Results showed non-uniform associations between pregnancy loss/death subtypes. Overall, prognostic values were only seen in non-embryonic losses, ie. the most severe phenotypes prone to the most devastating psychological consequences.31 The high rate of chromosomal abnormalities in miscarriage tissues before ten weeks may at least partly explain the disappointing value of angiogenic factors in this clinical setting. Isolated recurrent early, embryonic loss defines a subgroup of oAPS patients which differ from other oAPS, with a better overall prognosis.7 Studies suggest that the decidualized endometrium can select good quality embryos and rejects the incompetent embryos: the nature of the cellular and molecular dialogue between the maternal endometrium and the implanting embryo is central to early pregnancy failures. It has been proposed that patients experiencing recurrent implantation failure possess a selection mechanism that inappropriately rejects good quality embryos, recurrent embryo loss being caused by failure of natural embryo quality control.32,33 No central role for angiogenic factors is currently considered in this setting. The strength of this study includes a homogeneous and well characterized group of women receiving a single type of treatment during pregnancy, an accurate phenotyping of the various PMCs, and measurements performed in a single laboratory by personnel blinded to the outcomes. Limitations include the single-center design, the overall low number of PMCs, the retrospective study of frozen samples, the absence of any control group allowing analysis of the aPlAb-dependence and the LMWH-LDA treatment-dependence of the variations in angiogenic factors. In summary, PlGF and sFlt1 plasma concentrations are 841


E. Cochery-Nouvellon et al.

up-regulated in newly-pregnant obstetric APS beginning the traditional LMWH-LDA treatment. Variations of PlGF and of sFlt1 are risk factors for PMCs and for severe PMCs; and for fetal death, stillbirth and neonatal death. The ratio between PlGF increase and sFlt1 increase may help to rule out early on the development of PMCs or of severe PMCs in LMWH-LDA treated pregnancies: a prospective validation cohort is now mandatory to confirm the robustness of these data. Confirmation would lead to the non-selection of conventionally-treated patients with an overall good prognosis in the very first trials testing new therapeutic developments. Acknowledgments We thank all the study participants, the patients who agreed to

References 1. Wilson WA, Gharavi AE, Koike T, et al. International consensus statement on preliminary classification criteria for definite antiphospholipid syndrome: report of an international workshop. Arthritis Rheum. 1999;42(7):1309-1311. 2. Miyakis S, Lockshin MD, Atsumi T, et al. International consensus statement on an update of the classification criteria for definite antiphospholipid syndrome (APS). J Thromb Haemost. 2006;4(2):295-306. 3. Ruiz-Irastorza G, Crowther M, Khamashta M. Antiphospholipid syndrome. Lancet. 2010;376(9751):1498-509. 4. Giannakopoulos B, Passam F, Ioannou Y, Krilis SA. How we diagnose the antiphospholipid syndrome. Blood. 2009; 113(5):985-994. 5. de Jesus GR, Agmon-Levin N, Andrade CA, et al. 14th International Congress on Antiphospholipid Antibodies Task Force report on obstetric antiphospholipid syndrome. Autoimmun Rev. 2014;13(8):795813. 6. Giannakopoulos B, Krilis SA. How I treat the antiphospholipid syndrome. Blood. 2009;114(10):2020-2030. 7. Bouvier S, Cochery-Nouvellon E, LavigneLissalde G, et al. Comparative incidence of pregnancy outcomes in treated obstetric antiphospholipid syndrome: the NOH-APS observational study. Blood. 2014; 123(3):404-413. 8. Levine RJ, Maynard SE, Qian C, et al. Circulating angiogenic factors and the risk of preeclampsia. N Engl J Med. 2004; 350(7):672-683. 9. Signore C, Mills JL, Qian C, et al. Circulating angiogenic factors and placental abruption. Obstet Gynecol. 2006; 108(2):338-344. 10. Smith GC, Crossley JA, Aitken DA, et al. Circulating angiogenic factors in early pregnancy and the risk of preeclampsia, intrauterine growth restriction, spontaneous preterm birth, and stillbirth. Obstet Gynecol. 2007;109(6):1316-1324. 11. Zygmunt M, Herr F, Münstedt K, Lang U, Liang OD. Angiogenesis and vasculogenesis in pregnancy. Eur J Obstet Gynecol Reprod Biol. 2003;110(Suppl 1):S10-S18.

842

join us in this long-term adventure. We thank the NOHA network of gynecologists, obstetricians and general practitioners who actively contributed to this study. We also thank M.L. Tailland, D. Dupaigne, C. Ferrer, S. Ripart, A. Cornille, F. Masia, L. Boileau, F. Grojean, R. de Tayrac, P. Marès, E. Arnaud and A. Sotto for their assistance with patient management. We thank the research staff of the “Direction de la Recherche Clinique et de l’Innovation” of Nîmes University Hospital: S. Clément, C. Meyzonnier, N. Best, A. Megzari, R. Jacquet, S. Granier, B. Lafont, C. Masseguin, H. Obert, H. Léal, O. Albert, C. Suehs, P. Rataboul and M.P. Francheschi. We thank the staff of the "Département de Biostatistique, Epidémiologie, Santé Publique et Information Médicale BESPIM" of Nîmes University Hospital: P. Landais, J.P. Daurès and P. Fabbro-Peray. We thank M. Lomma for expert English editing.

12. Rosenberg VA, Buhimschi IA, Lockwood CJ, et al. Heparin elevates circulating soluble fms-like tyrosine kinase-1 immunoreactivity in pregnant women receiving anticoagulation therapy. Circulation. 2011; 124(23):2543-2553. 13. Yinon Y, Ben Meir E, Margolis L, et al. Low molecular weight heparin therapy during pregnancy is associated with elevated circulatory levels of placental growth factor. Placenta. 2015;36(2):121-124. 14. Kim MY, Buyon JP, Guerra MM, et al. Angiogenic factor imbalance early in pregnancy predicts adverse outcomes in patients with lupus and antiphospholipid antibodies: results of the PROMISSE study. Am J Obstet Gynecol. 2016;214(1):108.e1108.e14. 15. Gris JC, Bouvier S, Molinari N, et al. Comparative incidence of a first thrombotic event in purely obstetric antiphospholipid syndrome with pregnancy loss: the NOH-APS observational study. Blood. 2012;119(11):2624-2632. 16. Report of the National High Blood Pressure Education Program. Working group report on high blood pressure in pregnancy. Am J Obstet Gynecol . 2000;183(1):S1-S22. 17. ACOG Practice Bulletin No. 33. American College of Obstetricians and Gynecologists. Diagnosis and management of preeclampsia and eclampsia. Obstet Gynecol. 2002;77(1):67-75. 18. [Estimation de la croissance neonatale et postnatale-Audipog.] Available at: www. audipog. net/module_ligne/php. Last accessed 13 March 2017. 19. Sibai BM, Ramadan MK, Usta I, Salama M, Mercer BM, Friedman SA. Maternal morbidity and mortality in 442 pregnancies with hemolysis, elevated liver enzymes, and low platelets (HELLP syndrome). Am J Obstet Gynecol. 1993;169(4):1000-1006. 20. Hagmann H, Bossung V, Belaidi AA, et al. Low-molecular weight heparin increases circulating sFlt-1 levels and enhances urinary elimination. PLoS One. 2014; 9(1):e85258. 21. Sciascia S, Cuadrado MJ, Khamashta M, Roccatello D. Renal involvement in antiphospholipid syndrome. Nat Rev Nephrol. 2014;10(5):279-289. 22. Levine RJ, Thadhani R, Qian C, et al. Urinary placental growth factor and risk of

preeclampsia. JAMA. 2005;293(1):77-85. 23. Weissgerber TL, Rajakumar A, Myerski AC, et al. Vascular pool of releasable soluble VEGF receptor-1 (sFLT1) in women with previous preeclampsia and uncomplicated pregnancy. J Clin Endocrinol Metab. 2014;99(3):978-987. 24. Sela S, Natanson-Yaron S, Zcharia E, Vlodavsky I, Yagel S, Keshet E. Local retention versus systemic release of soluble VEGF receptor-1 are mediated by heparinbinding and regulated by heparanase. Circ Res. 2011;108(9):1063-1070. 25. Ananth CV, Vintzileos AM. Ischemic placental disease: epidemiology and risk factors. Eur J Obstet Gynecol Reprod Biol. 2011;159(1):77-82. 26. Cuadrado MJ, Buendía P, Velasco F, et al. Vascular endothelial growth factor expression in monocytes from patients with primary antiphospholipid syndrome. J Thromb Haemost. 2006;4(11):2461-2469. 27. Williams FM, Parmar K, Hughes GR, Hunt BJ. Systemic endothelial cell markers in primary antiphospholipid syndrome. Thromb Haemost. 2000;84(5):742-746. 28. Smadja D, Gaussem P, Roncal C, Fischer AM, Emmerich J, Darnige L. Arterial and venous thrombosis is associated with different angiogenic cytokine patterns in patients with antiphospholipid syndrome. Lupus. 2010;19(7):837-843. 29. Hladunewich M, Karumanchi SA, Lafayette R. Pathophysiology of the clinical manifestations of preeclampsia. Clin J Am Soc Nephrol. 2007;2(3):543-549. 30. Carroll TY, Mulla MJ, Han CS, et al. Modulation of trophoblast angiogenic factor secretion by antiphospholipid antibodies is not reversed by heparin. Am J Reprod Immunol. 2011;66(4):286-296. 31. Daugirdait V, van den Akker O, Purewal S. Posttraumatic stress and posttraumatic stress disorder after termination of pregnancy and reproductive loss: a systematic review. J Pregnancy. 2015;2015:646345. 32. Quenby S, Vince G, Farquharson R, Aplin J. Recurrent miscarriage: a defect in nature's quality control? Hum Reprod. 2002; 17(8):1959-1963. 33. Koot YE, Teklenburg G, Salker MS, Brosens JJ, Macklon NS. Molecular aspects of implantation failure. Biochim Biophys Acta. 2012;1822(12):1943-1950.

haematologica | 2017; 102(5)


ARTICLE

Chronic Myeloid Leukemia

Increased peroxisome proliferator-activated receptor γ activity reduces imatinib uptake and efficacy in chronic myeloid leukemia mononuclear cells

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Jueqiong Wang,1,2 Liu Lu,1,2 Chung H. Kok,1,2 Verity A. Saunders,1 Jarrad M. Goyne,1 Phuong Dang,1 Tamara M. Leclercq,1,2 Timothy P. Hughes1,2,3,4 and Deborah L. White1,2,4

Cancer Theme, South Australian Health and Medical Research Institute (SAHMRI), Adelaide; 2School of Medicine, University of Adelaide; 3Department of Haematology, SA Pathology, Adelaide and 4Australasian Leukaemia and Lymphoma Group, Melbourne, Australia 1

Haematologica 2017 Volume 102(5):843-853

ABSTRACT

I

matinib is actively transported by organic cation transporter-1 (OCT1) influx transporter, and low OCT-1 activity in diagnostic chronic myeloid leukemia blood mononuclear cells is significantly associated with poor molecular response to imatinib. Herein we report that, in diagnostic chronic myeloid leukemia mononuclear cells and BCR-ABL1+ cell lines, peroxisome proliferator-activated receptor γ agonists (GW1929, rosiglitazone, pioglitazone) significantly decrease OCT-1 activity; conversely, peroxisome proliferator-activated receptor γ antagonists (GW9662, T0070907) increase OCT-1 activity. Importantly, these effects can lead to corresponding changes in sensitivity to BCR-ABL kinase inhibition. Results were confirmed in peroxisome proliferator-activated receptor γ-transduced K562 cells. Furthermore, we identified a strong negative correlation between OCT-1 activity and peroxisome proliferator-activated receptor γ transcriptional activity in diagnostic chronic myeloid leukemia patients (n=84; P<0.0001), suggesting that peroxisome proliferator-activated receptor γ activation has a negative impact on the intracellular uptake of imatinib and consequent BCR-ABL kinase inhibition. The inter-patient variability of peroxisome proliferator-activated receptor γ activation likely accounts for the heterogeneity observed in patient OCT-1 activity at diagnosis. Recently, the peroxisome proliferator-activated receptor γ agonist pioglitazone was reported to act synergistically with imatinib, targeting the residual chronic myeloid leukemia stem cell pool. Our findings suggest that peroxisome proliferator-activated receptor γ ligands have differential effects on circulating mononuclear cells compared to stem cells. Since the effect of peroxisome proliferatoractivated receptor γ activation on imatinib uptake in mononuclear cells may counteract the clinical benefit of this activation in stem cells, caution should be applied when combining these therapies, especially in patients with high peroxisome proliferator-activated receptor γ transcriptional activity.

Introduction The first-generation Abl kinase inhibitor imatinib has proven effective in chronic phase chronic myeloid leukemia (CP-CML) patients with minimal toxic side effects. While responses to imatinib are generally excellent, 20-30% of patients will demonstrate suboptimal response / tyrosine kinase inhibitor (TKI) resistance, and 5-10% will still progress to the generally fatal blast crisis stage, despite TKI therapy.1-3 Hence, inter-patient variability in response to TKIs is evident despite the haematologica | 2017; 102(5)

Correspondence: deborah.white@sahmri.com

Received: July 24, 2016. Accepted: January 31, 2017. Pre-published: February 2, 2017. doi:10.3324/haematol.2016.153270 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/843 ©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.

843


J. Wang et al.

universal presence of the driving oncoprotein BCR-ABL. On this basis, there is growing scientific and clinical interest to define factors that underpin this response variability. Peroxisome proliferator-activated receptors (PPARs) are a family of transcription factors that regulate several metabolic pathways in a tissue-selective manner.4 Of the three PPAR subtypes, PPARγ has been studied most extensively in diverse biological pathways and disease conditions, including adipocyte differentiation/metabolism, insulin sensitivity, and inflammation. PPARγ has two isoforms, PPARγ1 and PPARγ2. While PPARγ2 is mostly found in adipose tissue,5 PPARγ1 is ubiquitously expressed in many tissues and cell types, including immunocytes (i.e., activated macrophages, lymphocytes and dendritic cells).6 PPARγ and its agonists have been implicated in hematological malignancies playing antitumor roles, such as inhibiting cell proliferation, inducing cell differentiation, and inducing apoptosis.7,8 Prost et al. recently demonstrated that the PPARγ agonist, pioglitazone, could target the residual CML stem cell pool by suppressing signal transducer and activator of transcription 5 (STAT5) and its downstream targets HIF2α and CITED2.9 This was supported by the work of Glodkowska-Mrowka et al., suggesting the clinical potential of the combination of pioglitazone and second- or third-generation TKIs in CML.10 The importance of the PPAR complex has also been demonstrated by several groups, indicating PPARα/PPARγ activation can increase organic cation uptake by inducing human SLC22A1 (encoding OCT-1) or murine Slc22a1 messenger ribonucleic acid (mRNA) expression.11,12 The functional activity of OCT-1 (OCT-1 activity, OA) in mononuclear cells (MNC) of de novo CP-CML patients is a powerful predictor of molecular response, overall, eventfree and progression-free survival.13-17 Patients with low OA demonstrate significantly inferior responses to standard imatinib therapy than those with high OA, due to low intracellular imatinib concentrations and corresponding reduced BCR-ABL kinase inhibition.14,15 Although the negative impact of low OA may be partially overcome by escalating the imatinib dose,14,16 this regimen is not tolerated by all patients and may lead to higher rates of adverse events.18,19 In a previous study, we demonstrated that the use of diclofenac, a competitive PPARγ antagonist, significantly increased OA in CML cells.20 Herein we assess the correlation between PPARγ activation and OA using primary MNC from de novo CP-CML patients and BCRABL1+ cell lines. Paradoxically we demonstrate that, in these cells, PPARγ agonists have an opposing effect on intracellular imatinib uptake and OA. In addition, a previous study from our laboratory has demonstrated that OA in patient MNC varies with cell lineage in the peripheral blood.21 Given the critical role of PPARγ in cell differentiation, the present study also explores the correlation between OA and the expression of the myeloid surface markers in CP-CML patient MNC at diagnosis.

Methods Cell lines BCR-ABL1+ KU812 and K562 cell lines were obtained from the American Type Culture Collection (ATCC, USA). BCR-ABL1transduced HL60 cells (HL60-BCRABL) were generated as described previously.21 844

Primary samples from CP-CML patients or healthy donors MNC and plasma samples were collected from de novo CP-CML patients enrolled in the TIDEL II study22 prior to the commencement of imatinib therapy. Normal MNC were obtained from healthy volunteers. All samples were collected with informed consent in accordance with the Declaration of Helsinki. Use of clinical trial patients samples were approved by the institutional review boards of the SA Pathology and the Royal Adelaide Hospital Research Ethics Committee.

Drugs Imatinib mesylate (STI571) and 14C-labelled imatinib were kindly provided by Novartis Pharmaceuticals (Switzerland). The potent OCT-1 inhibitor prazosin and PPARγ ligands GW1929, rosiglitazone, pioglitazone, GW9662 and T0070907 were all purchased from Sigma-Aldrich.

Lentivirus production and cell transduction The lentiviral plasmids expressing FLAG-tagged wild-type (WT) PPARγ1 and dominant negative (DN) PPARγ1-L466A/E469A,23 together with empty vector (EV), were constructed from a previously characterized vector, pLenti4/TO-IRES EGFP.24 K562 cells were transduced as previously described,25 and GFP+ cells were isolated for subsequent experiments.

Imatinib intracellular uptake and retention (IUR) assay and OCT-1 activity (OA) The IUR assay was performed and OA was determined as previously described.13 Cells were pre-incubated with 40 mM PPARγ ligands for one hour, and cell viability prior to the IUR assay was confirmed as greater than 98% by trypan blue exclusion assay. The assays were performed in the presence and absence of 100 mM prazosin, which is a potent inhibitor of OCT-1. OCT-1 activity was determined by calculating the difference between the IUR in the absence of prazosin and the IUR in the presence of prazosin.

Western blotting analyses and determination of IC50imatinib values Western blotting analyses for phosphorylated CRKL (p-CRKL) were performed to IC50imatinib as previously described.26,27 Cells were pre-incubated with 40 mM PPARγ ligands for one hour prior to exposure to imatinib. Anti-CRKL, anti-FLAG M2, anti-PPARγ and anti-GAPDH antibodies were employed in western blotting analyses.

Cell viability Analyses

KU812 cells were incubated with 10 mM PPARγ ligands for 24 hours prior to an additional 72-hour treatment with PPARγ ligands and varying concentrations of imatinib (range: 0-5 mM). Cell viability was assessed by Annexin V/7-AAD staining and fluorescence-activated cell sorting (FACS) analysis. The half maximal effective concentration (ED50) that induces cell apoptosis was estimated using non-linear regression as implemented in the GraphPad Prism software program (version 7.0a, GraphPad Software, USA).

Examination of PPARG and SLC22A1 mRNA expression in BCR-ABL1+ CML cell lines and MNC of de-novo CP-CML patients The expression level of PPARG and SLC22A1 (encoding OCT-1) mRNA in KU812 cells were examined by real-time quantitative polymerase chain reaction (RQ-PCR). PPARG and SLC22A1 mRNA expression levels in MNC of CP-CML patients were evaluated using the Illumina HumanHT-12v4 platform. haematologica | 2017; 102(5)


PPARγ activation reduces active imatinib uptake in CML MNC

A

B

C

D

Figure 1. Treatment with PPARγ ligands significantly alters intracellular uptake and retention (IUR) of imatinib and OCT-1 activity (OA) in BCR-ABL1+ CML cell lines. Cells were pre-incubated with 40 mM PPARγ ligands for 1 hour prior to the IUR assay. The IUR of imatinib in BCR-ABL1+ cell lines were significantly decreased with (A) PPARγ agonists GW1929, rosiglitazone (Rosi) or pioglitazone (Pio), and increased with (B) antagonists GW9662 or T0070907. The OCT-1 activity (OA) was determined by calculating the difference of IUR with or without the potent OCT-1 inhibitor, prazosin. OA in both cell lines were significantly decreased with (C) PPARγ agonists and increased with (D) antagonists. Results (mean ± SEM) are expressed as ng of imatinib per 200,000 cells, for at least 3 biological replicates. *P<0.05; **P<0.01; compared with DMSO control. DMSO: dimethyl sulfoxide; OCT-1: organic cation transporter-1; PPARγ: peroxisome proliferator-activated receptor γ.

PPARγ transcriptional activity in MNC of de-novo CP-CML patients

Results

Nuclear extracts from CP-CML patient MNC were prepared using the Nuclear Extract Kit (Active Motif, USA). PPARγ transcriptional activity was then measured using the PPARγ Transcription Factor Assay Kit (Active Motif). Linear regression analysis was used to determine whether the PPARγ transcriptional activity level could predict OA.

Treatment with PPARγ ligands significantly alters OCT-1 activity in BCR-ABL1+ CML cell lines

Enzyme immunoassays for 15-deoxy-D12,14-PGJ2 (15d-PGJ2) The 15d-PGJ2 levels in plasma samples from CP-CML patients were analyzed using a 15d-PGJ2 ELISA kit (Enzo Life Sciences, USA).

Immunophenotyping Cryopreserved MNC were stained with antibodies specifically targeting myeloid lineage markers (CD14-PE, CD15-FITC and CD16-PerCP-Cy5.5 antibodies, all from BD Biosciences). Neutrophils were identified as CD15+/CD14-,28 with additional marker CD16 to indicate the different stages of neutrophil maturation.29

Statistical Analyses All statistical analyses were performed using GraphPad Prism. Differences were considered to be statistically significant when the P-value was less than 0.05. For details of the methods see the Online Supplementary Material. haematologica | 2017; 102(5)

Treatment with the PPARγ agonist GW1929, rosiglitazone (Rosi), or pioglitazone (Pio) significantly decreased the IUR of imatinib in KU812 and BCR-ABL1-transduced HL60 cells (HL60-BCRABL, Figure 1A). An opposite effect on IUR was observed in both cell lines following treatment with PPARγ antagonists (Figure 1B). The addition of prazosin, a potent inhibitor of OCT-1, allowed us to further evaluate the activity of the OCT-1 protein in the transport of imatinib. Treatment with the PPARγ agonist GW1929 significantly decreased OA in KU812 (from 10.8 to 7.5 ng/200,000 cells, P=0.0280) and BCR-ABL1-transduced HL60 cells (HL60-BCRABL, from 11.9 to 8.9 ng/200,000 cells, P=0.0228, Figure 1C). Similar results were observed when cells were treated with the PPARγ agonist Rosi (KU812: from 10.8 to 5.5 ng/200,000 cells, P=0.0010; HL60-BCRABL: from 11.9 to 8.9 ng/200,000 cells, P=0.0391) and Pio (KU812: from 10.8 to 6.5 ng/200,000 cells, P=0.0057; HL60-BCRABL: from 11.9 to 4.6 ng/200,000 cells, P=0.0001, Figure 1C). The opposite effect on OA was also observed in both cell lines following treatment with PPARγ antagonists (Figure 1D). The presence of GW9662 significantly increased the OA (KU812: from 10.8 to 15.4 ng/200,000 cells, P=0.0011; HL60-BCRABL: from 11.9 to 15.1 845


J. Wang et al.

A

F

B

G

C

H

D

I

E

J

Figure 2. Treatment with PPARγ ligands significantly alters OCT-1 activity in MNC from de novo CP-CML patients. OA assays were performed on thawed MNC isolated from the peripheral blood of newly diagnosed patients with CP-CML. Cells were treated with 40 µM PPARγ ligands for 1 hour prior to IUR assay. PPARγ agonists (A) GW1929, (B) rosiglitazone (Rosi) or (C) pioglitazone (Pio) significantly decreased OA in CP-CML MNC with high OA. PPARγ antagonists (D) GW9662 or (E) T0070907 significantly increased OA in CP-CML MNC with low OA. (F-J) Treatment with PPARγ ligands had no significant effect on OA in normal MNC isolated from healthy donors. The MNC samples treated with Pio (C, H) were different from those treated with GW1929 or Rosi (A, B, F, G), as indicated by different symbols. Dotted line indicates the cutoff value of OA (4 ng/200,000 cells) to define “high OA” and “low OA”. ns, P>0.05. DMSO: dimethyl sulfoxide; OCT-1: organic cation transporter-1; PPARγ: peroxisome proliferator-activated receptor γ.

846

haematologica | 2017; 102(5)


PPARγ activation reduces active imatinib uptake in CML MNC

A

B

Figure 3. Treatment with PPARγ ligands significant alters IC50imatinib and ED50imatinib. (A) The in vitro reduction in the level of p-Crkl by imatinib was detected using the IC50imatinib assay. KU812 cells were incubated with 40 mM PPARγ ligands for 1 hour prior to the treatment with increasing concentrations of imatinib for 2 hours. IC50imatinib was significantly increased with PPARγ agonists GW1929, rosiglitazone (Rosi) or pioglitazone (Pio), and decreased with antagonists GW9662 or T0070907. (B) Cell viability was determined using Annexin V-PE/7-AAD staining. KU812 cells were incubated with 10 mM PPARγ ligands for 24 hours prior to an additional 72-hour treatment with PPARγ ligands and varying concentrations of imatinib, ranging from 0 mM to 5 mM. PPARγ antagonists co-administered with imatinib resulted in a significantly lower ED50imatinib. Data are mean ± SEM for at least 3 biological replicates. *P<0.05; **P<0.01; compared with DMSO control. DMSO: dimethyl sulfoxide; PPARγ: peroxisome proliferator-activated receptor γ; ED: the half maximal effective concentration.

ng/200,000 cells, P=0.0330). Similarly, there was a significant increase in OA when cells were treated with T0070907 (KU812: from 10.8 to 16.8 ng/200,000 cells, P=0.0025; HL60-BCRABL: from 11.9 to 16.9 ng/200,000 cells, P=0.0040).

Treatment with PPARγ ligands significantly alters OCT-1 activity in MNC from de novo CP-CML patients Our previous studies demonstrated that CP-CML patients with low MNC OA (less than 4.0 ng/200,000 cells, lowest OA quartile) at diagnosis have the poorest response to imatinib treatment and the highest rate of transformation to accelerated phase or blast crisis.15 Herein we examined the effect of PPARγ ligands on OA in cryopreserved MNC isolated from CP-CML patients at diagnosis. Patient baseline MNC OA values were divided into two groups (“high OA” and “low OA”) using the cutoff as 4.0 ng/200,000 cells. In patients with high OA, treatment with PPARγ agonists resulted in consistently reduced OA. The average OA in these samples was significantly reduced by GW1929 (from 7.9 to 5.2 ng/200,000 cells, P=0.0075, Figure 2A) or Rosi (from 7.9 to 4.7 ng/200,000 cells, P<0.0001, Figure 2B). In another 5 MNC samples with high OA, treatment with PPARγ agonist Pio resulted in a similar decrease in average OA (from 9.5 to 3.9 ng/200,000 cells, P<0.0001, Figure 2C). As a result of this decrease, OA values in 60% high OA cases (6 out of 10) were moved into the low OA group in the presence of PPARγ agonists. Notably, treatment with PPARγ antagonists increased the OA in patients with low OA, (n=6). The average OA in these patients was increased from 2.7 to 5.1 ng/200,000 cells by GW9662 (P=0.0129, Figure 2D) and to 4.7 ng/200,000 cells by T0070907 (P=0.0155, Figure 2E). In addition, this increase in OA afforded by the PPARγ antaghaematologica | 2017; 102(5)

onists resulted in 5 out of 6 low OA samples (83.33%) moving into high OA groups. In contrast to the results in CP-CML patient samples, no significant change was observed in OA in peripheral blood MNC isolated from healthy donors after incubation with any PPARγ ligand (P>0.05, Figures 2F-2J).

Treatment with PPARγ ligands significantly alters IC50imatinib and cell viability when co-administered with imatinib The IC50imatinib was examined in KU812 cells to assess whether the observed effects of PPARγ ligands on OA translate into corresponding changes in BCR-ABL tyrosine kinase inhibition. Consistent with the results of the OA assay, a significant increase in IC50imatinib was observed in KU812 cells when treated with the PPARγ agonists GW1929 (from 4.7 to 7.2 mM, P=0.0078) or Rosi (from 4.7 to 8.5 mM, P=0.0013). In the presence of Pio, about 70% increment (from 4.7 to 6.4 mM) in IC50imatinib was observed compared to the control, although this increment was not statistically significant (P=0.1151). In contrast, treatment with PPARγ antagonists significantly reduced the IC50imatinib in KU812 cells (GW9662: from 4.2 to 2.5, P=0.0078, or T0070907: from 4.2 to 1.4 mM, P=0.0055, Figure 3A). Annexin V-PE and 7-AAD staining was performed in KU812 cells to investigate the effects of the PPARγ ligands on cell viability when co-administered with imatinib. In the presence of varying concentrations of imatinib, cotreatment with 10 mM PPARγ antagonists significantly reduced the half-maximal effective concentration (ED50) that induces cell apoptosis (GW9662: P=0.0236, T0070907: P=0.0011) compared with vehicle control (Figure 3B). There was no significant effect on cell viability when treating KU812 cells with PPARγ agonists. 847


J. Wang et al. A

B

C

D

E

Figure 4. Over-expression of PPARγ significantly reduced OCT-1 activity and increased IC50imatinib . K562 cells were transduced with (A) lentiviral pLenti4/TO-IRES EGFP plasmids expressing FLAG-tagged wild-type (WT) PPARγ1 and dominant negative (DN) PPARγ1-L466A/E469A, together with empty vector (EV). The overexpression of PPARγ was confirmed by (B) western blotting with anti-FLAG M2 antibodies and (C) RT-PCR analyses. Compared with the EV control, WT PPARγ overexpression in K562 cells significantly decreased (D) OA and increased (E) IC50imatinib. There was no difference in OA or IC50imatinib between the EV control and cells transduced with DN PPARγ. Data are mean ± SEM for at least 3 biological replicates. *P<0.05; **P<0.01; compared with EV control. OCT-1: organic cation transporter-1; PPARγ: peroxisome proliferator-activated receptor γ.

Lentiviral over-expression of PPARγ significantly decreases OCT-1 activity and increases IC50imatinib

Neither PPARG gene expression nor PPARγ protein is associated with OCT-1 activity

WT and DN PPARγ (Figure 4A for construct schematics) transduced K562 cells were FACS-sorted isolated based on green fluorescent protein (GFP) intensity. Over expression of FLAG-tagged PPARγ was confirmed by western blotting with anti-FLAG M2 antibodies (Figure 4B) and RQ-PCR (Figure 4C). Compared to WT PPARγ, DN PPARγ has impaired ligands binding affinity and significantly reduced transcriptional activity.23 As shown in Figure 4D, compared with the empty vector (EV) control (mean OA=24.0), the OA was significantly decreased in WT PPARγ transduced K562 cells (mean OA=16.0, n=4, P=0.0286). There was no significant difference in OA between the EV control and DN PPARγ transduced K562 cells (mean OA=23.6, n=4, P>0.5). When examining the IC50imatinib in transduced K562 cells, a significant increase was observed in cells transduced with WT PPARγ (mean 13.3 mM) compared with EV control cells (mean 6.8 mM, n=3, P=0.0074, Figure 4E). No significant change in IC50imatinib was observed in cells transduced with PPARγ-DN (mean 5.7 mM) compared to EV control cells (n=3, P>0.5).

The effect of PPARγ ligands on OA strongly suggests the involvement of PPARγ in OA regulation. No significant change in PPARG gene expression was observed in KU812 cells after 3-hour treatment with PPARγ ligands (P>0.5, Online Supplementary Figure S1A). The PPARG mRNA level in diagnostic MNC of CP-CML patients was measured using the Illumina HumanHT-12v4 platform and compared between high and low OA groups to determine any association between PPARG gene expression and OA in primary cells. As shown in the Online Supplementary Figure S1B, across 120 CP-CML patient MNC samples tested, the average PPARG mRNA level in low OA patients was not different from that in the high OA group (mean 4.52 vs. 4.51, P=0.6673). The expression of total PPARγ protein also remained unchanged in KU812 cells treated with PPARγ ligands (P>0.5, Online Supplementary Figure S1C). In whole cell lysates prepared from CP-CML patient MNC samples, no significant difference was observed in PPARγ total protein levels between patients in low OA (n=6) and those in high OA groups (n=7, P=0.7732, Online Supplementary Figure S1D).

848

haematologica | 2017; 102(5)


PPARγ activation reduces active imatinib uptake in CML MNC

B

A

C

Figure 5. PPARγ transcriptional activity negatively correlates with OCT-1 activity in MNCs of de novo CP-CML patients. (A) Levels of PPARγ transcriptional activity were compared between low and low high OA groups. The error bars represent 95% confidence interval (CI) of the mean value. (B) Correlation between PPARγ transcriptional activity and OA in 84 CP-CML patient MNC samples by Pearson product-moment correlation. (C) The percentage of patients that had low OA in the high (n=13) and low (n=71) PPARγ transcriptional activity patient group was determined based on a linear regression model using the PPARγ activity level of 0.2 as cutoff. OR denotes odds ratio and P value are derived from Fisher's exact test. OA: OCT-1 activity; OCT-1: organic cation transporter-1; PPARγ: peroxisome proliferator-activated receptor γ.

Notably, there was no significant difference in the mRNA expression level of SLC22A1 (encoding OCT-1) in KU812 cells treated with PPARγ ligands compared with vehicle control (P>0.5, Online Supplementary Figure S1E). In addition, when assessing MNC of de-novo CP-CML patients, the SLC22A1 mRNA expression levels between the two OA groups were comparable (P=0.3006, Online Supplementary Figure S1F).

PPARγ transcriptional activity negatively correlates with OCT-1 activity in MNC of de novo CP-CML patients PPARγ plays an important role in activating the transcription of its downstream target genes that mediate multiple signaling pathways.30 However, the level of PPARγ transcriptional activity has not previously been investigated in CP-CML, in particular its link with OA. To further evaluate this relationship, CML patients were grouped into low and high OA groups as previously defined, and the nuclear PPARγ transcriptional activity was compared between the two groups. The result confirmed that nuclear PPARγ transcriptional activity was significantly higher in the low OA group (average 0.1742, n=33) compared with the high OA group (average 0.0889, n=51, P<0.0001, Figure 5A). Additionally, a significant negative correlation was observed between the transcriptional activity of PPARγ and the OA in individual samples (n=84, r=-0.5677, P<0.0001, Figure 5B). Linear regression analysis revealed a significant relationship between PPARγ transcriptional activity level and OA (P<0.0001), with the model described as OA=8.0-21.3×(PPARγ activity level). Using this fitted model, we identified a PPARγ transcriptional activity level of 0.2 or greater (rounded from 0.19 to haematologica | 2017; 102(5)

be more conservative) to be associated with a low OA. Hence, samples with high PPARγ activity levels (>0.2) were predicted to be low OA, whereas low PPARγ activity levels (≤0.2) were predicted as high OA. As such, samples with high PPARγ activity levels (n=11/13, 85%) were significantly enriched for low OA, compared to the samples of the group with low PPARγ transcriptional activity levels (n=21/71, 30%) (OR=13.1; 95% CI: 2.7-64.3; P=0.0003; Figure 5C).

No significant difference was observed in plasma 15d-PGJ2 between CP-CML patients with low and high OCT-1 activity or PPARγ activity One of the major regulatory mechanisms of PPARγ transcriptional activity is the direct binding of PPARγ ligands, such as 15d-PGJ2,31 that result in conformational changes of PPARγ and subsequent changes in its transcriptional activity.32 To investigate the possibility that 15d-PGJ2 plays a role in activating PPARγ in CP-CML, the plasma levels of 15d-PGJ2 were examined in 150 CP-CML patient samples prior to imatinib treatment. No significant difference was observed in plasma 15d-PGJ2 levels between the patients in the low or high OA groups (P=0.2446, Online Supplementary Figure S2A). In 59 samples with matched PPARγ transcriptional activity results, there was no significant correlation between plasma 15d-PGJ2 levels and PPARγ transcriptional activity (P=0.4112, Online Supplementary Figure 2B).

Cell composition of CP-CML patient MNC varies significantly between patients with low and high OA Our previous study reported that MNC OA varies great849


J. Wang et al. A

B

C

D

Figure 6. Cell composition of CP-CML patient MNC was significantly correlated with OA and PPARγ transcriptional activity. Using multi-parameter flow cytometry, the expression of CD14, CD15 and CD16 in cryopreserved MNC collected from de novo CP-CML patients and their relationship to MNC OCT-1 activity or PPARγ transcriptional activity were examined. (A) A significant higher percentage of CD15+CD16brightCD14- cells was observed in patients with low OA. (B) CD15+CD16-CD14cells were significantly enriched in high OA patients. (C) There was a significant negative correlation between the percentage of CD15+CD16brightCD14- cells and MNC OA in CML diagnosis patients. The Pearson product-moment was used to assess the correlation. (D) The mean percentage of CD15+CD16brightCD14- cells in samples with high PPARγ transcriptional activity levels was significantly higher than samples with low PPARγ transcriptional activity levels. The error bars represent 95% confidence interval (CI) of the mean value. *P<0.05; **P<0.01. OA: OCT-1 activity; OCT-1: organic cation transporter-1; PPARγ: peroxisome proliferator-activated receptor γ.

ly between cell lineages in CML and is significantly associated with the OA in isolated neutrophils.21 It is possible that the specific cell composition within individual patient samples may underlie their specific OA. Given that the MNC compartment in CML patients at diagnosis is predominantly comprised of immature and mature neutrophils,21 here the expression of the granulocytic surface markers CD15 and CD16 in the MNC population was examined and correlated with OA. As shown in Figure 6A,B, compared with high OA patients, patients with low OA had a higher percentage of CD15+CD16brightCD14- cells (44.47% vs. 20.97%, P=0.0048) and a lower percentage of CD15+CD16-CD14- neutrophils (12.87% vs. 23.52%, P=0.0113) in the MNC samples. In keeping with the above findings there was a significant negative correlation between the percentage of CD15+CD16brightCD14- cells and OA (r=-0.5273, P=0.0169, n=20, Figure 6C). To determine the role of PPARγ in the MNC composition, the link between the percentages of granulocytic markers and PPARγ transcriptional activity was then examined in 13 samples with matched PPARγ transcriptional activity results. The mean percentage of CD15+CD16brightCD14- cells measured in samples with 850

high PPARγ transcriptional activity levels (38.85%, n=4) was significantly higher than samples with low PPARγ transcriptional activity levels (17.77%, n=9, P=0.0013, Figure 6D).

Discussion The functional OA in primary CML mononuclear cells at diagnosis is a strong and reliable predictor of both shortand long-term imatinib responses and clinical outcomes in independent clinical trials.14-17 Modulation of the OCT-1 transporter to increase the uptake of imatinib into CML cells could potentially improve the efficacy of imatinib therapy for patients with low OA. Many studies have been published investigating the SLC22A1 (encoding OCT-1) genetic variants and its link with imatinib uptake. However, as reviewed by Watkins et al.,33 this is controversial and the mechanism regulating imatinib uptake via OCT-1 remains unclear. Our previous data demonstrated that the treatment with PPARγ ligand diclofenac significantly increased imatinib uptake mediated by OCT-1.20 In the study herein, the haematologica | 2017; 102(5)


PPARγ activation reduces active imatinib uptake in CML MNC

negative link between OA and PPARγ activation has further been elucidated by treating BCR-ABL1+ cell lines and primary MNC of CP-CML patients with various synthetic PPARγ ligands. Over-expression of WT PPARγ in K562 cells resulted in significantly decreased OA, confirming these effects as specific to PPARγ. In addition, by investigating nuclear transcriptional activity of PPARγ in CPCML patient MNC samples, we provide evidence that activation of PPARγ negatively impacts OA and therefore reduces imatinib uptake and retention. We have previously demonstrated that there is a significant correlation between the in vitro BCR-ABL kinase activity inhibitory concentration 50% for imatinib (IC50imatinib) and OA.14 These findings are substantiated herein, by demonstrating that alterations of OA by PPARγ ligands translated to corresponding changes in sensitivity to BCR-ABL kinase inhibition, as demonstrated in IC50imatinib . However, when used in combination with imatinib, significant changes in cell viability were only observed following PPARγ antagonist treatments. The change in OA and IC50imatinib did not extend to an increase in ED50imatinib in PPARγ agonists treated cells. Of note, the baseline IC50imatinib value in KU812 cells is at a relatively high level (4.2 mM). Therefore, it is tempting to speculate that the cell viability has already reached its peak and cannot be further improved by PPARγ agonist treatments. In addition, activation of PPARγ has been recently reported to decrease STAT5 transcription in CML stem cells.9 It is possible that the impaired intracellular imatinib uptake by PPARγ agonists may be counterbalanced by their inhibitory effect on STAT5. Different from the synergistic effect of pioglitazone and imatinib in CML stem cells,9 we observed an opposing effect of PPARγ and imatinib, probably due to the different target populations (MNC vs. CD34+ cells) with varying SLC22A1 mRNA expression and imatinib uptake.34 As OA in CD34+ cells has been proven to be significantly low or even below the level of detection,34 it is unlikely that OA will be decreased significantly, or measurably, within the confines of this assay, by the use of a PPARγ agonist. In addition, OA in CD34+ is not predictable for the achievement of major molecular response (MMR).35 Given that diagnostic peripheral blood MNC samples is the cell population in which the predictive value of OA was established, this cell population was investigated in the current study. While the proposed PPARγ ligand-TKI combination therapy can effectively target leukemic stem cells, the contradictory effect of PPARγ on intracellular imatinib uptake and retention observed in circulating MNC, suggests that it may not be an ideal option for de novo CP-CML patients on imatinib, as a rapid initial decline in BCR-ABL1 transcripts is critical for improved event-free survival.36 In addition, the inter-patient variation in PPARγ transcriptional activity we demonstrate herein may make this combinational therapy only applicable to those patients with low PPARγ transcriptional activity. However, patients on second- or third-generation TKIs may benefit from such therapy,10 as these TKIs are not transported via OCT-1.37-39 The role of PPARα in imatinib transport has been reported by Wang et al. using KCL22 cells and primitive CML CD34+ cells, whereby PPARα agonists upregulated imatinib uptake by increasing PPARA and SLC22A1 mRNA expression levels.11 Despite the high homology at the protein level, different or even contrary biologic functions of haematologica | 2017; 102(5)

PPARα and PPARγ have been implicated in monocytes/macrophages40-42 or cardiomyocytes43 by several groups. In the current study, we did not observe significant changes in PPARG or SLC22A1 mRNA expression in PPARγ ligand treated cells. Therefore, it is likely that PPARγ interacts with the OCT-1 transporter through a different mechanism from PPARα. Instead of altering the expression of SLC22A1, we speculated that the regulation of OA observed here by PPARγ is through a PPARγ direct target gene network. The downstream target genes of PPARγ, together with the overlapping and competing actions of PPAR transcription factors in CML cells, are the focus of ongoing studies. PPARγ is known to interact with several endogenous ligands.31 Given the important role of PPARγ in inflammation regulation44 and the potential link between CML and the cyclooxygenase/prostaglandin pathway,45 we examined the plasma levels of 15d-PGJ2, the dehydration end product of prostaglandin D2. The result suggested that ligand binding by 15d-PGJ2 is not critical for the inter-patient variability in PPARγ activation, which is in agreement with the previous report that in vivo 15d-PGJ2 is insufficient to activate PPARγ.46,47 Similar to the previous findings with diclofenac,20 our results demonstrate that PPARγ regulates OA in a BCRABL-dependent manner, as the effects on OA by PPARγ ligands were only observed in CP-CML patient MNC but not in healthy donor MNC. It has been reported that the constitutively active tyrosine kinase BCR-ABL promotes neutrophil differentiation by downregulating c-Jun expression, while BCR-ABL inhibition by imatinib promotes monocytic differentiation in KCL22/α cells.48 Our previous work has also indicated that BCR-ABL may have an indirect effect on OA by promoting granulocytic differentiation.21 In addition, PPARγ is required for terminal maturation in the granulocytic lineage in vitro, but to a lesser extent for the early stages of hematopoietic cell development.49 Herein we examined differentiation stages of neutrophils and demonstrated that a subset of mature neutrophils (CD15+CD16brightCD14-) was enriched in low OA patients. Furthermore, a higher percentage of these cells were observed in patients with high PPARγ transcriptional activity, which suggests a correlation between PPARγ activation, neutrophils maturity and OA. Based on our findings, we speculate that the variation in patient MNC OA may reflect the heterogeneous leukemia cells composition influenced by BCR-ABL and PPARγ. More research will be necessary to determine the role of PPARγ in granulocytic lineage differentiation procedures, especially in the presence of BCR-ABL. In conclusion, the findings presented in the study herein demonstrate that treatment with PPARγ ligands significantly alters OA via a BCR-ABL-dependent mechanism. PPARγ transcriptional activity, rather than mRNA or protein expression level, has a significant correlation with OA. Furthermore, the significant enrichment of mature neutrophils in patients with low OA and high PPARγ transcriptional activity indicates the involvement of PPARγ in the granulocytic differentiation driven by BCR-ABL. These findings suggest that, while PPARγ ligand has been shown to efficiently affect CML stem cells, inter-patient variability in PPARγ plays a critical role in the heterogeneity in patients’ MNC OA at diagnosis. Personalized combination therapeutic strategy may be needed when targeting different leukemia cell populations. 851


J. Wang et al.

Acknowledgments The authors would like to thank Associate Professor Claudine Bonder (The Centre for Cancer Biology, Australia) for kindly providing the lentiviral plasmids and Professor Andrew Zannettino

References 1. Hughes TP, Hochhaus A, Branford S, et al. Long-term prognostic significance of early molecular response to imatinib in newly diagnosed chronic myeloid leukemia: An analysis from the International Randomized Study of Interferon and STI571 (IRIS). Blood. 2010;116(19):37583765. 2. Jabbour E, Saglio G, Hughes TP, Kantarjian H. Suboptimal responses in chronic myeloid leukemia: implications and management strategies. Cancer. 2012;118(5):1181-1191. 3. Hughes T, White D. Which TKI? An embarrassment of riches for chronic myeloid leukemia patients. Hematology Am Soc Hematol Educ Program. 2013;2013:168-175. 4. Braissant O, Foufelle F, Scotto C, Dauca M, Wahli W. Differential expression of peroxisome proliferator-activated receptors (PPARs): tissue distribution of PPAR-alpha,beta, and-gamma in the adult rat. Endocrinology. 1996;137(1):354-366. 5. Mukherjee R, Jow L, Croston GE, Paterniti JR, Jr. Identification, characterization, and tissue distribution of human peroxisome proliferator-activated receptor (PPAR) isoforms PPARgamma2 versus PPARgamma1 and activation with retinoid X receptor agonists and antagonists. J Biol Chem. 1997;272(12):8071-8076. 6. Beamer BA, Negri C, Yen CJ, et al. Chromosomal localization and partial genomic structure of the human peroxisome proliferator activated receptorgamma (hPPAR gamma) gene. Biochem Biophys Res Commun. 1997;233(3):756759. 7. Garcia-Bates TM, Lehmann GM, SimpsonHaidaris PJ, Bernstein SH, Sime PJ, Phipps RP. Role of peroxisome proliferator-activated receptor gamma and its ligands in the treatment of hematological malignancies. PPAR Res. 2008;834612. 8. Bertz J, Zang C, Liu H, et al. Compound 48, a novel dual PPAR alpha/gamma ligand, inhibits the growth of human CML cell lines and enhances the anticancer-effects of imatinib. Leuk Res. 2009;33(5):686-692. 9. Prost S, Relouzat F, Spentchian M, et al. Erosion of the chronic myeloid leukaemia stem cell pool by PPAR agonists. Nature. 2015;525(7569):380-383. 10. Glodkowska-Mrowka E, Manda-Handzlik A, Stelmaszczyk-Emmel A, et al. PPAR ligands increase antileukemic activity of second-and third-generation tyrosine kinase inhibitors in chronic myeloid leukemia cells. Blood Cancer J. 2016;6(1):e377. 11. Wang L, Giannoudis A, Austin G, Clark RE. Peroxisome proliferator-activated receptor activation increases imatinib uptake and killing of chronic myeloid leukemia cells. Exp Hematol. 2012;40(10):811-819. 12. Nie W, Sweetser S, Rinella M, Green RM. Transcriptional regulation of murine

852

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

(The University of Adelaide, Australia) for kindly providing the packaging cells and lentiviral packaging plasmids. This study was supported by the National Health and Medical Research Council of Australia (NHMRC) project grants (No. 1026932).

Slc22a1 (Oct1) by peroxisome proliferator agonist receptor-alpha and -gamma. Am J Physiol Gastrointest Liver Physiol. 2005;288(2):G207-212. White DL, Saunders VA, Dang P, et al. OCT-1-mediated influx is a key determinant of the intracellular uptake of imatinib but not nilotinib (AMN107): Reduced OCT-1 activity is the cause of low in vitro sensitivity to imatinib. Blood. 2006;108(2):697-704. White DL, Saunders VA, Dang P, et al. Most CML patients who have a suboptimal response to imatinib have low OCT-1 activity: Higher doses of imatinib may overcome the negative impact of low OCT1 activity. Blood. 2007;110(12):4064-4072. White DL, Dang P, Engler J, et al. Functional activity of the OCT-1 protein is predictive of long-term outcome in patients with chronic-phase chronic myeloid leukemia treated with imatinib. J Clin Oncol. 2010;28(16):2761-2767. White DL, Radich J, Soverini S, et al. Chronic phase chronic myeloid leukemia patients with low OCT-1 activity randomized to high-dose imatinib achieve better responses and have lower failure rates than those randomized to standard-dose imatinib. Haematologica. 2012;97(6):907-914. White DL, Saunders VA, Dang P, Engler J, Hughes TP. OCT-1 activity measurement provides a superior imatinib response predictor than screening for single-nucleotide polymorphisms of OCT-1. Leukemia. 2010;24(11):1962-1965. Cortes JE, Baccarani M, Guilhot F, et al. Phase III, randomized, open-label study of daily imatinib mesylate 400 mg versus 800 mg in patients with newly diagnosed, previously untreated chronic myeloid leukemia in chronic phase using molecular end points: tyrosine kinase inhibitor optimization and selectivity study. J Clin Oncol. 2010;28(3):424-430. Hughes TP, Branford S, White DL, et al. Impact of early dose intensity on cytogenetic and molecular responses in chronicphase CMLpatients receiving 600 mg/day of imatinib as initial therapy. Blood. 2008;112(10):3965-3973. Wang J, Hughes TP, Kok CH, et al. Contrasting effects of diclofenac and ibuprofen on active imatinib uptake into leukaemic cells. Br J Cancer. 2012;106(11):1772-1778. Engler JR, Zannettino ACW, Bailey CG, Rasko JEJ, Hughes TP, White DL. OCT-1 function varies with cell lineage but is not influenced by BCR-ABL. Haematologica. 2011;96(2):213-220. Yeung DT, Osborn MP, White DL, et al. TIDEL-II: first-line use of imatinib in CML with early switch to nilotinib for failure to achieve time-dependent molecular targets. Blood. 2015;125(6):915-923. Parham KA, Zebol JR, Tooley KL, et al. Sphingosine 1-phosphate is a ligand for peroxisome proliferator-activated receptorgamma that regulates neoangiogenesis.

FASEB J. 2015;29(9):3638-3653. 24. Barrett JM, Parham KA, Pippal JB, et al. Over-expression of sphingosine kinase-1 enhances a progenitor phenotype in human endothelial cells. Microcirculation. 2011;18 (7):583-597. 25. Isenmann S, Arthur A, Zannettino AC, et al. TWIST family of basic helix-loop-helix transcription factors mediate human mesenchymal stem cell growth and commitment. Stem Cells. 2009;27(10):2457-2468. 26. White D, Saunders V, Lyons AB, et al. In vitro sensitivity to imatinib-induced inhibition of ABL kinase activity is predictive of molecular response in patients with de novo CML. Blood. 2005;106(7):2520-2526. 27. White D, Saunders V, Grigg A, et al. Measurement of in vivo BCR-ABL kinase inhibition to monitor imatinib-induced target blockade and predict response in chronic myeloid leukemia. J Clin Oncol. 2007;25(28):4445-4451. 28. Dumitru CA, Moses K, Trellakis S, Lang S, Brandau S. Neutrophils and granulocytic myeloid-derived suppressor cells: immunophenotyping, cell biology and clinical relevance in human oncology. Cancer Immunol Immunother. 2012;61(8):11551167. 29. Elghetany MT. Surface antigen changes during normal neutrophilic development: a critical review. Blood Cells Mol Dis. 2002;28(2):260-274. 30. Gelman L, Feige JN, Tudor C, Engelborghs Y, Wahli W, Desvergne B. Integrating nuclear receptor mobility in models of gene regulation. Nucl Recept Signal. 2006; 4:e010. 31. Sauer S. Ligands for the nuclear peroxisome proliferator-activated receptor gamma. Trends Pharmacol Sci. 2015;36(10):688-704. 32. Auwerx J, Baulieu E, Beato M, et al. A unified nomenclature system for the nuclear receptor superfamily. Cell. 1999;97(2):161163. 33. Watkins DB, Hughes TP, White DL. OCT1 and imatinib transport in CML: is it clinically relevant? Leukemia. 2015;29(10):19601969. 34. Engler JR, Frede A, Saunders VA, Zannettino AC, Hughes TP, White DL. Chronic myeloid leukemia CD34+ cells have reduced uptake of imatinib due to low OCT-1 activity. Leukemia. 2010;24(4):765770. 35. Engler JR, Frede A, Saunders V, Zannettino A, White DL, Hughes TP. The poor response to imatinib observed in CML patients with low OCT-1 activity is not attributable to lower uptake of imatinib into their CD34+ cells. Blood. 2010;116(15):2776-2778. 36. Stein AM, Martinelli G, Hughes TP, et al. Rapid initial decline in BCR-ABL1 is associated with superior responses to second-line nilotinib in patients with chronic-phase chronic myeloid leukemia. BMC Cancer. 2013;13(1):1. 37. Eadie L, Hughes TP, White DL. Nilotinib does not significantly reduce imatinib OCT-

haematologica | 2017; 102(5)


PPARÎł activation reduces active imatinib uptake in CML MNC

38.

39.

40.

41.

1 activity in either cell lines or primary CML cells. Leukemia. 2010;24(4):855-857. Hiwase DK, Saunders V, Hewett D, et al. Dasatinib cellular uptake and efflux in chronic myeloid leukemia cells: therapeutic implications. Clin Cancer Res. 2008;14 (12):3881-3888. Lu L, Saunders VA, Leclercq TM, Hughes TP, White DL. Ponatinib is not transported by ABCB1, ABCG2 or OCT-1 in CML cells. Leukemia. 2015;29(8):1792-1794. Chang-Yeop H, Soo-Young P, PAK YK. Role of endocytosis in the transactivation of nuclear factor- B by oxidized low-density lipoprotein. Biochem J. 2000;350(3):829837. Fischer B, von Knethen A, Brune B. Dualism of oxidized lipoproteins in provoking and attenuating the oxidative burst in macrophages: role of peroxisome proliferator-activated receptor-gamma. J

haematologica | 2017; 102(5)

Immunol. 2002;168(6):2828-2834. 42. Lee H, Shi W, Tontonoz P, et al. Role for peroxisome proliferator-activated receptor alpha in oxidized phospholipid-induced synthesis of monocyte chemotactic protein-1 and interleukin-8 by endothelial cells. Circ Res. 2000;87(6):516-521. 43. Son N-H, Yu S, Tuinei J, et al. PPAR induced cardiolipotoxicity in mice is ameliorated by PPAR deficiency despite increases in fatty acid oxidation. J Clin Invest. 2010;120(10):3443. 44. Daynes RA, Jones DC. Emerging roles of PPARs in inflammation and immunity. Nat Rev Immunol. 2002;2(10):748-759. 45. Giles FJ, Kantarjian HM, Bekele BN, et al. Bone marrow cyclooxygenase-2 levels are elevated in chronic-phase chronic myeloid leukaemia and are associated with reduced survival. Br J Haematol. 2002;119(1):38-45. 46. Bell-Parikh LC, Ide T, Lawson JA,

McNamara P, Reilly M, FitzGerald GA. Biosynthesis of 15-deoxy-delta12,14-PGJ2 and the ligation of PPARgamma. J Clin Invest. 2003;112(6):945-955. 47. Zhang GS, Liu DS, Dai CW, Li RJ. Antitumor effects of celecoxib on K562 leukemia cells are mediated by cell-cycle arrest, caspase-3 activation, and downregulation of Cox-2 expression and are synergistic with hydroxyurea or imatinib. Am J Hematol. 2006;81(4):242-255. 48. Kobayashi S, Kimura F, Ikeda T, et al. BCRABL promotes neutrophil differentiation in the chronic phase of chronic myeloid leukemia by downregulating c-Jun expression. Leukemia. 2009;23(9):1622-1627. 49. Labrecque J, Allan D, Chambon P, Iscove NN, Lohnes D, Hoang T. Impaired granulocytic differentiation in vitro in hematopoietic cells lacking retinoic acid receptors alpha1 and gamma. Blood. 1998;92(2):607-615.

853


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Acute Myeloid Leukemia

Ferrata Storti Foundation

Long-term observation reveals high-frequency engraftment of human acute myeloid leukemia in immunodeficient mice

Anna M. Paczulla,1 Stephan Dirnhofer,2 Martina Konantz,1 Michael Medinger,3 Helmut R. Salih,4,5 Kathrin Rothfelder,4,5 Dimitrios A. Tsakiris,6 Jakob R. Passweg,3 Pontus Lundberg6 and Claudia Lengerke1,3,5

Haematologica 2017 Volume 102(5):854-864

University of Basel and University Hospital Basel, Department of Biomedicine, Switzerland; 2University of Basel and University Hospital Basel, Department of Pathology, Switzerland; 3University of Basel and University Hospital Basel, Clinic for Hematology, Switzerland; 4Clinical Collaboration Unit Translational Immunology, German Cancer Consortium (DKTK) and German Cancer Research Center (DKFZ), Partner site Tübingen, Department for Internal Medicine II, Tübingen, Germany; 5Department of Hematology and Oncology, Eberhard-Karls-University, Tübingen, Germany and 6 University of Basel and University Hospital Basel, Diagnostic Hematology, Switzerland 1

ABSTRACT

R Correspondence: claudia.lengerke@unibas.ch

Received: July 28, 2016. Accepted: February 6, 2017. Pre-published: February 9, 2017. doi:10.3324/haematol.2016.153528 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/854 ©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.

854

epopulation of immunodeficient mice remains the primary method for functional assessment of human acute myeloid leukemia. Published data report engraftment in ~40-66% of cases, mostly of intermediate- or poor-risk subtypes. Here we report that extending follow-up beyond the standard analysis endpoints of 10 to 16 weeks after transplantation permitted leukemic engraftment from nearly every case of xenotransplanted acute myeloid leukemia (18/19, ~95%). Xenogeneic leukemic cells showed conserved immune pheno-types and genetic signatures when compared to corresponding pre-transplant cells and, furthermore, were able to induce leukemia in re-transplantation assays. Importantly, bone marrow biopsies taken at standardized time points failed to detect leukemic cells in 11/18 of cases that later showed robust engraftment (61%, termed “long-latency engrafters”), indicating that leukemic cells can persist over months at undetectable levels without losing disease-initiating properties. Cells from favorable-risk leukemia subtypes required longer to become detectable in NOD/SCID/IL2Rγnull mice (27.5±9.4 weeks) than did cells from intermediate-risk (21.9±9.4 weeks, P<0.01) or adverse-risk (17±7.6 weeks; P<0.0001) subtypes, explaining why the engraftment of the first was missed with previous protocols. Mechanistically, leukemic cells engrafting after a prolonged latency showed inferior homing to the bone marrow. Finally, we applied our model to favorable-risk acute myeloid leukemia with inv(16); here, we showed that CD34+ (but not CD34–) blasts induced robust, long-latency engraftment and expressed enhanced levels of stem cell genes. In conclusion, we provide a model that allows in vivo mouse studies with a wide range of molecular subtypes of acute myeloid leukemia subtypes which were previously considered not able to engraft, thus enabling novel insights into leukemogenesis.

Introduction The proliferation and survival of acute myeloid leukemia (AML) cells depend largely on environmental cues that are yet to be deciphered, making in vivo models mandatory for functional studies on AML.1 In contrast to genetically modified mice, human AML xenografts better depict the disease heterogeneity observed in patients. haematologica | 2017; 102(5)


Long-latency engraftment of human AML in NSG mice

A variety of different strains of immunosuppressed mice are available for xenograft studies.2,3 Overall, a higher degree of immune suppression appears to facilitate human cell engraftment. As such, robust engraftment was reported from ~40% of human AML samples transplanted via intrafemoral injection into non-obese diabetes/severe combined immunodeficiency (NOD/SCID) mice that were given pre-transplant irradiation conditioning.4 The more severely immunosuppressed NOD/SCID/IL2Rγnull (NSG) mice, which lack T, B and functional natural killer cells,3,5 enabled engraftment of 66% of transplanted AML samples, but mice received 10-fold higher cell numbers (107 cells/mouse) and a particularly low threshold of >0.1% human among murine bone marrow (BM) cells was set to define engraftment.6 With both protocols, engraftment was preferentially observed from FLT3mutated high and intermediate molecular-risk AML (defined according to European LeukemiaNet criteria)7,8 suggesting that especially favorable-risk AML cannot be studied using these models.9 Insufficient cross-reactivity between murine and human proteins may lead to inadequate cytokine activity in human cells growing in the murine environment, thereby specifically hampering engraftment of certain subtypes of AML;10,11 consistently, human cytokine knock-in mice were recently shown to improve engraftment.12 Time to AML development is not assessable in patients, since diagnosis is typically made at the time of overt disease. However, a longer time to relapse after apparent remission is observed in patients with favorable-risk as compared to adverse-risk AML, suggesting differences in in vivo leukemia kinetics.13,14 We, therefore, hypothesized that subsets of AML (e.g. favorable-risk types) may require longer time to produce detectable engraftment and induce leukemia in mice. We transplanted a mixed cohort of 19 human AML of various genetic backgrounds, including four favorable-risk AML and two cases of acute promyelocytic leukemia (APL), and extended the posttransplant observation period to 1 year, instead of the 10 to 16 weeks used in previous studies.4,6 Indeed, only 7/19 transplanted AML (~37%, termed “standard engrafters”) showed detectable engraftment by week 16 after transplantation, while 11/19 (~58%, termed “long-latency engrafters”) repopulated mice later. Consistent with our hypothesis, all favorable-risk AML were long-latency engrafters. Importantly, longitudinal assessment of murine BM at 8 to 16 weeks after transplantation showed no evidence of leukemic cells in long-latency engrafters, indicating that they would have been missed with standard protocols. Next, we used this model to investigate the mechanisms underlying the observed differences in engraftment latency and the leukemia-initiating cell (LIC) compartment of favorable risk AML with inv(16).

Methods Primary acute myeloid leukemia cells Peripheral blood (PB) samples from patients with AML (Table 1 and Online Supplementary Table S1) were collected following approval by the Ethics Review Board of the University Hospitals of Basel and Tuebingen, enriched for mononuclear cells using a Ficoll (Biocoll, Merck Millipore, Darmstadt, Germany) gradient and viably frozen in 10% dimethyl sulfoxide solution (AppliChem, Darmstadt, Germany). haematologica | 2017; 102(5)

Mice and xenotransplantation assays NOD.Cg-Prkdcscid IL2rgtmWjl/Sz (also termed NOD/SCID/IL2Rγnull, NSG) mice purchased from Jackson Laboratory (Bar Harbor, ME, USA) were maintained under pathogen-free conditions according to German and Swiss federal and state regulations. Freshly thawed AML cells were used for primary transplants. Details on sample preparation are provided in the Online Supplementary Material. Gender-matched, 7- to 10-week old animals with or without prior sublethal irradiation were injected intrafemorally15 or via the tail vein with AML cells resuspended in 25 or 200 mL phosphatebuffered saline, respectively (Table 2). Engraftment, (defined as ≥1% leukemic cells in murine PB or BM),1,4 was assessed in PB and BM at signs of distress (e.g. decreased food and water consumption, rapid breathing, altered movement)16 or routinely every 4 to 5 weeks in one mouse per group for each AML case. Mice were euthanized at sickness (weight loss, ruffled coat, weakness, reduced motility, other severe pathology) or detection of engraftment.6,17 Kaplan-Meier survival analysis and final assessment were performed on all animals. For secondary transplants, BM cells freshly isolated from mice showing >40% BM infiltration and belonging to one experimental group were pooled, subjected to MACS purification for human CD33 to eliminate contamination by murine cells and then used for transplantation in equal numbers as in the corresponding primary transplants. Limiting dilution and homing assays were performed according to standard protocols (see Online Supplementary Material).

Flow cytometry and histopathology For flow cytometry analyses, fluorescent antibodies against human CD33, CD34, CD133, CD117, CD45, (BD Biosciences), CD14, CD13 (eBiosciences, San Diego, CA, USA), CD3 and CD19 (Biolegend, San Diego, CA, USA) were used. The analyses were performed on either a FACS CantoII or LSR II Fortessa (both BD Biosciences). SytoxBlue or 7-aminoactinomycin D was used to discriminate living and dead cells. Histological studies (with hematoxylin & eosin) and immunohistochemical analyses (CD33, CD34; Ventana, Tucson, Arizona, USA; CD117, DAKO, Glostrup, Denmark) were performed on formalin-fixed paraffin-embedded organs as described elsewhere.18 Samples were evaluated for Ki67 protein expression using the monoclonal anti-Ki67 antibody (clone Mib 1, DAKO IR626, ready-to-use). The percentage of stained leukemic cells was counted and recorded in 5% increments.

Colony-forming unit assays For colony-forming assays, 2.5x104 CD34+ or respectively CD34– cells, sorted by FACS, were plated in methylcellulose (Methocult H4434, StemCell Technologies, Vancouver, Canada) in triplicate, incubated at 37°C in 5% CO2 and scored for colonyforming units at day 14.

Next-generation sequencing and microarray gene expression analyses For next-generation sequencing, the AML community panel from Thermo Fisher containing 19 genes frequently mutated in AML (see ampliseq.com), the Ion PGM platform and the Ion Reporter AML pipeline (version 5.0) were employed. Agilent SurePrint G3 Human Gene Expression 8x60K v2 Microarrays analyses were performed on RNA extracts. For a detailed description of the next-generation sequencing and microarrays see the Online Supplementary Data.

Statistical analysis Data are presented as the mean ± standard error of the mean. P values are derived from the application of the Mann-Whitney U test or two-tailed Fisher exact test. 855


A.M. Paczulla et al.

Results Extended follow-up time improved the detection rate of human acute myeloid leukemia engraftment in NSG mice Nineteen AML cases of various genetic backgrounds, including four favorable-risk AML [three with inv(16) and one with t(8;21)] and two APL, were investigated regarding engraftment in NSG mice (Table 1 and Online Supplementary Table S1 for details of the patients and AML characteristics). Xenotransplants were performed following standard procedures via the tail vein (18/19 AML cases, 7x105 to 1x106 cells per mouse) or intrafemoral injection (4/19 AML cases, 4x105 cells per mouse) (Table 1). In previous studies, mice transplanted with human AML cells were assessed at defined time-points (10, 12 or 16 weeks) and considered engrafted if >0.1-1% of human among total BM cells were detected.4,6 Using this method, a considerable proportion of AML, mostly of intermediateand favorable-risk subtypes, were scored non-engraftable (35-60%, depending on the transplantation procedure and detection threshold). We hypothesized that these AML would also engraft, if longer observation times were applied. We prolonged follow-up to 1 year after transplantation, during which we closely monitored animals by inspection (every other day) and by BM puncture and multi-parameter flow cytometry analysis (performed at signs of distress or routinely every 4 to 5 weeks in one

mouse per group for every AML case). This longitudinal assessment in the BM revealed that engraftment detection did indeed progress over time (Figure 1A and Online Supplementary Table S2). Compared to analyses at 8-10 or 12-14 weeks after transplantation, the standard analysis at 16 weeks already improved engraftment, but late analysis at 1 year revealed better engraftment than at any of these early time-points (Figure 1A, P<0.0001, P<0.0001 and P<0.001, respectively). In fact, only 7/19 transplanted AML cases (37%) were standard engrafters, showing ≥1% of human leukemic cells in the BM within the first 16 weeks after transplantation, while the majority (11/19, 58%) of cases were long-latency engrafters and repopulated mice at detectable levels only later. Importantly, 11/11 and respectively 10/11 of long-latency engrafters were screened negative for human leukemic cells in routine BM biopsies performed at 8-10, 12-14 and 16 weeks after transplantation (Figure 1A and Online Supplementary Table S2), indicating that their engraftment would have been missed with protocols stopping experimental follow-up at these timepoints. Consistently, the majority (63%) of transplanted mice were alive until 16 weeks after transplantation, but developed robust leukemic engraftment and were removed from the experiments progressively within the next months (Figure 1B). Of note, in the three AML cases transplanted using both techniques (patients #1, #2, and #9; Table 1), intrafemoral transplantation was more efficient and allowed comparable engraftment starting from 2-fold reduced cell numbers (Online Supplementary Figure S1).

Table 1. Molecular risk groups of the transplanted cases of acute myeloid leukemia, transplantation procedure and outcome details.

Molecular risk group†

Transplanted cells per mouse

Route

Irradiation

Engrafted/ transplanted

% Human engraftment in BM

Time to engraftment

1

intermediate-I

2

adverse

3 4

APL intermediate-I

5 6 7 8

intermediate-I intermediate-I adverse favorable [inv(16)(p13.1q22)] favorable [inv(16)(p13.1q22)] intermediate-I adverse APL intermediate-I intermediate-I intermediate-I favorable [inv(16)(p13.1q22)] intermediate-II favorable [t(8;21)] intermediate-I

4.0x105 7.0x105 4.0x105 8.0x105 7.0x105 1.0x106 CD34+ 1.0x106 CD341.0x106 1.0x106 1.0x106 1.0x106 CD34+ 1.0x106 CD344.0x105 1.0 x 106 4.0x105 1.0x106 1.0x106 7.5x105 1.0x106 6.0x105 1.0x106 CD34+ 1.0x106 CD341.0x106 1.0x106

i.f. i.v. i.f. i.v. i.v. i.v. i.v. i.v. i.v. i.v. i.v. i.v. i.f. i.v. i.f. i.v. i.v. i.v. i.v. i.v. i.v. i.v. i.v. i.v.

yes yes yes yes yes yes yes yes yes yes no no yes yes yes yes yes no yes no no no yes yes

3/3 5/5 5/5 5/5 5/5 5/5 0/5 5/5 5/5 5/5 5/5 0/5 4/4 4/4 3/3 3/5 5/5 4/4 3/3 3/5 5/5 0/5 4/4 3/3

49.23 + 11.81 14.15 + 7.57 78.40 + 6.59 70.63 + 14.91 3.84 + 1.04 74.3 + 21.81 0 83.45 + 3.97 4.86 + 4.63 68.51 + 11.36 68.6 + 1.06 0 87.45 + 9.67 95.34 + 4.65 40.38 + 22.61 62.42 + 21.31 2.01 + 0.7 8.34 + 6.6 3.84 ± 1.04 22.32 + 16.38 37.94 + 18.61 0 11.40 + 8.21 1.6 + 0.35

7 weeks 7 weeks 9 weeks 9 weeks 10 weeks 13 weeks 15 weeks 15 weeks 16 weeks 17 weeks 20 weeks 21 weeks 22 weeks 25 weeks 26 weeks 27 weeks 28 weeks 28 weeks 30 weeks 38 weeks 39 weeks

1.0x106

i.v.

yes

0/6

0

-

Long-latency engrafters (58%)

Standard engrafters (37%)

Pat #

9 10 11# 12 13 14 15 16 17# 18

NE 19

according to Mrozek et al.8; #relapse sample; APL: acute promyelocytic leukemia; i.v.: intravenously; i.f. intrafemorally. Note that for patients #4, #8 and #16 both CD34+ and CD34- blasts were transplanted, but engraftment resulted only from the CD34+ cells. Patients’ samples #1, #2 and #9 were transplanted both intravenously and intrafemorally. Standard and long-latency engrafters are idicated at the side of the table. NE: Non-engrafter (5%). †

856

haematologica | 2017; 102(5)


Long-latency engraftment of human AML in NSG mice

Conserved immunophenotypic and molecular features in xenogeneic versus corresponding pre-transplant acute myeloid leukemia cells Mice engrafted with human AML cells were analyzed by multicolor flow cytometry, next-generation sequencing and whole-body histopathology. Various degrees of leukemic infiltration were detected in murine PB, BM and organs (Figure 2A,B). Engraftment of healthy human CD3+ and CD19+ cells was not observed at any time-point (presumably because CD33 or CD34 pre-selection was applied to cells from AML samples showing <95% blasts). Xenogeneic leukemic cells derived from murine BM mostly showed conserved immunophenotypes when compared to corresponding pre-transplant samples (Figure 2C and Online Supplementary Figure S2), with the exception of a down-regulation of CD117 (cKIT), which was observed in 15/18 analyzed liver and 17/18 PB but not in corresponding spleen or BM-derived samples (Online Supplementary Figure S2B). Of note, 10/18 (56%) of AML samples that showed robust engraftment in BM and/or organs did not show leukemic cells in the PB (Figure 2A). These data suggest that there are differences in leukemic cells residing in different sites in the mouse, and show that screening of PB alone cannot be used to monitor engraftment reliably. Outgrowth of human leukemic cells in a xenograft animal model â&#x20AC;&#x201C; especially if observed after periods of long latency â&#x20AC;&#x201C; might indicate selection of specific clones or de novo acquisition of permissive mutations. To investigate this aspect, we analyzed the mutational status of pretransplant and mouse-derived (post-transplant) leukemic cells using next-generation sequencing. Post-transplant samples were generated from pooled BM of all mice engrafted with one AML case. Paired samples of pre-transplant and corresponding xenogeneic cells were available

A

for analysis from 12 of 18 engrafted AML cases. As shown in Figure 2D, high conservation of genetic signatures was observed in both standard and long-latency engrafters (see also Online Supplementary Table S3 for a detailed annotation of individual mutations). In one mouse-derived sample, a de novo-occurring ASLX1 mutation was detected at a low allelic burden (4%), while in another a low allelic burden NRAS mutation was lost, suggesting that genetic evolution can occur, but may not be a frequent event (Figure 2D, Online Supplementary Table S3). Furthermore, reduced frequencies of all five detected mutations were observed in post- versus pre-transplant samples of another case, suggesting possible co-engraftment with non-leukemic human cells (patient #13, Figure 2D). Finally, xenogeneic leukemic cells were shown to induce leukemia efficiently upon retransplantation in secondary recipients (Figure 3). Interestingly, faster repopulation was observed with xenogeneic secondary AML cells compared with the primary AML cells from the same donor for both standard and long-latency engrafting AML (Figure 3). Taken together, leukemic engraftment in mice was confirmed by phenotype, histopathology, molecularly by next-generation sequencing and functionally in retransplantation assays. Mouse xenografts appear to mimic human disease faithfully, even though engraftment occurs after a long latency.

Favorable-risk acute myeloid leukemia required a longer time to become detectable in mice Favorable-risk AML cells were previously scored nonengraftable, because they were not detectable in transplanted NSG mice analyzed 10 to 16 weeks after transplantation. Our study confirms these results: at these time-points, leukemic cells could not be detected in BM

B

Figure 1. Extended post-transplant follow-up enhances the detection efficiency of human acute myeloid leukemia cell engraftment in NSG mice. Longitudinal engraftment analysis by routine bone marrow punctures performed every 4 weeks revealed dependency of engraftment efficiency on post-transplant follow-up time. Lower percentages of engrafted mice were detected at standard post-transplantation analysis time-points used in previous protocols as shown by (A) quantification of engrafted samples at 8-10 weeks, 12-14 weeks and 16 weeks versus 1 year and by (B) Kaplan-Meier survival analysis of 109 mice xenotransplanted with human AML cells, indicating that with extended follow-up time, ~95% of all transplanted mice develop leukemia. A Fisher exact test was used to calculate statistical significance between engrafters and non-engrafters at each analyzed time to our endpoint of 1 year of observation in (A).

haematologica | 2017; 102(5)

857


A.M. Paczulla et al. A

B

C

D

Figure 2. Conserved immunophenotypic and genetic features in patient-derived and corresponding xenogeneic acute myeloid leukemia cells. (A) Whole-body flow cytometry and (B) histopathological analysis of peripheral blood (PB), bone marrow (BM), spleen and liver of transplanted mice revealed heterogeneous infiltration with human leukemic blasts. (A) Semi-quantitative analysis summarizing high (>20%, +++), medium (5-20%, ++) and low (<5%, +) degrees of human leukemic cells among murine cells. Note that cells could be undetectable in PB despite robust BM/organ infiltration. (B) A representative histopathological analysis of BM and spleen from one engrafted sample [patient #9, AML with inv(16)] using hematoxylin & eosin and antibody staining against human CD33, CD34 and CD117 indicating the leukemic origin. (C) A representative multicolor flow cytometric analysis of immunophenotypic profiles of pre- (patient PB-derived) versus post-transplant (mouse BM-derived) AML cells showing no phenotypic changes upon engraftment [patient #8, AML with inv(16)]. (D) Highly conserved genetic patterns in pre- (patientderived) and post-transplant (mouse BM-derived) samples. Allele frequencies for each mutation as detected by next-generation sequencing are shown. Note the loss of one out of two NRAS mutations (*p.Gln61Hi, # p.Gln61Lys; patient #8) and the gain of a low level ASXL1 mutation (patient #15).

858

haematologica | 2017; 102(5)


Long-latency engraftment of human AML in NSG mice

punctures performed on mice transplanted with favorablerisk AML cells (Online Supplementary Table S2), even though, when followed-up for longer periods, these mice eventually showed robust engraftment (Figure 1, Table 1 and Online Supplementary Table S2). To analyze the relationship between molecular risk classification and engraftment capacity in NSG mice, we divided our cohort according to the European LeukemiaNet molecular risk score into favorable-, intermediate (I or II)- or adverse-risk AML. The time to detectable engraftment was indeed longer in mice transplanted with favorable-risk versus adverse- or intermediate-risk AML cells (favorable versus adverse P<0.0001, favorable versus intermediate P=0.004, Figure 4A, see also Table 1 for detailed numbers of engrafted among transplanted mice). Consistently, the longest survival was observed in mice transplanted with favorable-risk AML (27.5±9.4 weeks), followed by those transplanted with intermediate-risk AML (21.9±9.4 weeks, P=0.008) and adverse-risk cells (17±7.6 weeks, P<0.0001) (Figure 4B). The two analyzed APL samples showed heterogeneous behavior, with one sample belonging to the standard engraftment group and the other to the long-latency group (Online Supplementary S3A). Consistent with previous data reporting a high frequency of FLT3-ITD mutations among engrafting AML, FLT3-ITD mutations were observed in 4/7 (57%) of standard versus only 3/11 (27%) long-latency engrafters (Online Supplementary Table S1). However, when all mice were analyzed together, FLT3-ITD mutated AML did not show significantly faster engraftment or shorter survival when compared to samples without a mutated FLT3 (Online Supplementary Figure 3B).

Investigation of factors regulating engraftment latency Leukemia has been proposed to initiate from rare subpopulations of blasts termed leukemia-initiating cells (LIC).1,4,19 The observed differences in leukemia latency could derive from: (i) different numbers of LIC (resulting

from different LIC frequency and/or homing capacity to BM niches); (ii) differences in survival and/or proliferation capacity of human AML cells in the murine microenvironment; or (iii) a combination of these factors. Limiting dilution assays did in fact reveal a strong association between engraftment latency (Figure 5A, left), mouse survival (Figure 5A, right) and, respectively, the number of transplanted cells, and lower LIC frequency in long-latency versus standard engrafters (1:197.599 versus 1:102.237). Interestingly, reduced homing rates to the BM20 were noted in long-latency versus standardly engrafting AML (Figure 5B) and the lowest percentage of homing cells was actually observed in the one AML sample that did not engraft (patient #19, data not shown). To investigate in vivo proliferation, we performed Ki67 staining on the BM of mice infiltrated with human AML cells; for reasons of consistency, only mice showing >20% leukemic BM infiltration were included in the analysis. Interestingly, similar high Ki67 positivity was detected in standardly versus long-latency engrafting AML cells, indicating comparable in vivo proliferation at this advanced disease stage (Figure 5C,D). Supporting the notion that AML cells of both groups can rapidly expand in the BM microenvironment of NSG mice, the time from the last negative BM puncture to detectable engraftment was comparably short in standard versus long-latency engrafters (3.36±1.03 weeks versus 3.17±1.17 weeks, P=0.72) (Online Supplementary Table S2).

Favorable-risk acute myeloid leukemia with inv(16) engraft from CD34-expressing leukemia-initiating cells Next, we used this model to investigate leukemia initiation from understudied favorable-risk AML with inv(16). All three inv(16) samples available to us were transplanted and showed engraftment, although only after a long latency (patients #8, #9 and #16; see Table 1 and Online Supplementary Table S1). Engraftment was confirmed as described above by multicolor flow cytometry and

Figure 3. Xenogeneic human cells robustly induce leukemia in secondary transplantation assays. Fresh xenogeneic cells derived from murine BM purified by MACS recognizing leukemic antigens were retransplanted at equal numbers as in the corresponding primary transplantation assay. Note that robust engraftment occurred from all retransplanted cases, including the long-latency engrafting AML. Of note, engraftment occurred faster (left) and mice survived for a shorter time (right) after secondary transplantation (blue line) than after the primary transplantations (red line). This is also summarized in the table below. A log-rank (Mantel-Cox) test was applied to calculate statistical significance.

haematologica | 2017; 102(5)

859


A.M. Paczulla et al.

histopathology analyses detecting human leukemic cells of conserved phenotype in murine PB, BM and organs (Figure 2A,B and Online Supplementary Figure 2A). As shown by next-generation sequencing analysis, two NRAS mutations were found in the patient who engrafted first (patient #8), of which one was lost upon engraftment, while no further AML-specific mutations were detected with our panel in patient #9 in either pre- or post-transplant cells (Figure 2D and Online Supplementary Table S3). AML cells from patient #18, which required the longest time to engraft, showed well-conserved mutations in BRAF, KRAS (at low allele frequency) and ASXL1 (Figure 2D and Online Supplementary Table S3). To investigate in more detail the relevance of CD34 expression as a marker for LIC in this genetic subtype, we sorted CD34+ and CD34– AML blasts with inv(16) and analyzed them functionally in colony-forming unit and xenotransplantation assays, as well as molecularly by gene expression arrays (Figure 6A). Interestingly, colony-formation capacity was higher in CD34+ blasts than in CD34– ones (Figure 6B). Even with prolonged follow-up, in vivo leukemia initiation was only observed from transplanted CD34+ but not CD34– leukemic inv(16) cells (Figure 6C), indicating that LIC are comprised in the CD34+ subpopulation. When observation times were longer, CD34+ inv(16) AML blasts induced engraftment even in the absence of pre-transplant irradiation conditioning. In mice, CD34+ blasts recapitulated the original AML phenotype generating both CD34+ and CD34– subpopulations (Online Supplementary Figure S2). Supporting these functional results, microarray gene expression analyses performed on CD34+ and CD34– inv(16) AML blasts revealed enhanced stem cell gene expression in CD34+ cells, as shown by gene set enrichment analysis, comparing our results with common datasets (Figure 6D and Online Supplementary Figure S4).4 Collectively these data indicate that, albeit with longer latency, favorable-risk AML with inv(16) can robustly engraft NSG mice from CD34+ LIC (Figure 6 and Table 1).

A

Discussion Human healthy and malignant hematopoietic cells engraft immunosuppressed mice, thereby providing a valuable tool for studies on human hematopoiesis and leukemogenesis. Higher degrees of murine immune suppression facilitate repopulation with human cells,3,5 yet a substantial proportion of AML, particularly of favorablerisk and APL subtypes, remained non-engraftable in antecedent studies.21-24 The current study indicates that some AML require more time to become detectable in mice, and thus may have been falsely scored non-engraftable in previous reports. By performing longitudinal BM analyses of transplanted mice at 10, 12 and 16 weeks, and afterwards prolonging follow-up time to 1 year, we demonstrate that AML that score negative for human leukemic cells at conventional time-points eventually do engraft and induce symptomatic leukemia. Besides providing a tool for in vivo investigation of additional AML subtypes, these data demonstrate that leukemic cells can survive long periods of time at undetectable levels in mice before they eventually spread out and induce leukemia. This evolution in NSG mice mimics well the clinical course of AML in patients. As hypothesized, the molecular risk group of the transplanted AML significantly influenced engraftment and leukemia kinetics in mice, with animals transplanted with favorable-risk AML showing the slowest engraftment and longest survival. FLT3-mutated AML, in previous studies overrepresented within engrafted AML,7 did not show significantly shorter time to engraftment nor was the overall survival of transplanted mice shorter compared to outcomes of FLT3-non-mutated samples.6,7 Higher sample numbers or exclusion of specific genetic backgrounds (e.g. aggressive AML without FLT3 mutation but with EVI1 overexpression, patient #2) might be required to detect significant differences.9 These findings are consistent with

B

Figure 4. Time to in vivo engraftment and mouse survival correlate with molecular risk group stratification of the transplanted acute myeloid leukemia. (A) Favorable-risk AML showed longer time to engraftment when compared to intermediate- or adverse-risk AML. (B) Kaplan-Meier survival analysis of 99 mice transplanted with human AML indicates that mouse survival depends on the AML molecular risk group. Note that animals transplanted with favorable-risk AML cases showed the longest survival. (black: favorable, n=4; blue: intermediate (I+II), n=8; red: adverse risk AML, n=3). See Table 1 for transplanted/engrafted mouse numbers per individual AML case. A log-rank (Mantel-Cox) test was applied to calculate statistical significance.

860

haematologica | 2017; 102(5)


Long-latency engraftment of human AML in NSG mice

A

B

C

D

Figure 5. Investigation of factors regulating leukemia latency. (A) Limiting dilution assays showed that lowering the number of transplanted cells under certain thresholds resulted in engraftment failure (left) and better survival (right) in both standard (upper) and long-latency (lower) engrafters. (B) The frequency of leukemic cells homing to the BM was higher in standard engrafters than in long-latency engrafters.The data shown are from three standard (patients #2, #5, #6) and three long-latency engrafting AML cases (patients #11, #17, #18). (C, D) Comparable Ki67 positivity in standard vs. long-latency engrafting AML cells infiltrating murine BM (C). Representative pictures for both CD33 and Ki67 staining of two standard engrafters (patients #2, #4) and two long-latency engrafters (patients #13, #16) (D). For reasons of consistency only BM of mice showing >20% infiltration with human cells were included in the analysis.

haematologica | 2017; 102(5)

861


A.M. Paczulla et al. A

B

C

D

Figure 6. Leukemia initiating cells are within the CD34+ compartment of acute myeloid leukemia with inv(16). (A) Experimental setup to analyze CD34+ versus CD34- blasts in AML with inv(16). (B, C) Functional analyses of sorted CD34+ versus CD34- AML with inv(16) blasts revealed that CD34+ subsets are enriched for cells with in vitro colony-forming capacity (B) and contain in vivo engrafting LIC. Kaplan-Meier survival analysis indicating survival differences in mice transplanted with non-leukemogenic CD34- blasts versus leukemia-inducing CD34+ blasts (C); (red: CD34+ blasts (engraftable); gray: CD34- blasts (non-engraftable) of the same patient in each panel; patients #8 and #16; five transplanted mice per condition, 1x106 transplanted cells/mouse). For colony-forming unit (CFU) counts, three biological replicates performed on samples from patients #8 and #16 are shown. (D) Enrichment plot of a common LIC-gene signature4 in gene expression profiles of CD34+ versus CD34- blasts from two inv(16) AML cases (patients #8 and #16) indicating a common gene signature in both comparisons (left) and a heatmap of the CD34+ versus CD34- gene signatures in AML with inv(16) (right). NES denotes normalized enrichment score.

previously published data that correlate the capacity of AML cells to engraft NOD-SCID mice with patientsâ&#x20AC;&#x2122; outcome,22 and the side-by-side analysis of NOD-SCID and NSG mice indicating that AML considered nonengraftable show equally impaired repopulation.25 There are several possible explanations for the engraftment deficiency observed with some AML subtypes. Delayed engraftment, as assessed during 1 year in our study, might reflect clonal adaptation or slower growth rate of some AML subtypes in the mouse environment. In our study, both standard and long-latency engrafters showed conserved phenotypic and genetic features in mouse-derived cells in comparison with the correspon862

ding pre-transplant cells, findings which stand in contrast to those of previous studies in which engraftment of individual subclones was detected.26,27 We hypothesize that the more homogeneous results consistently observed in the xenogeneic samples analyzed in our study reflect the fact that they were induced by higher numbers of LIC, which enhances the chance that several subclones are depicted in the eventually analyzed sample. In contrast to the previous studies, we used pre-transplant irradiation conditioning26 and comparatively younger mice,27 which might augment LIC homing to BM niches and thus enable long-term engraftment and disease initiation from higher numbers of cells. In support of this hypothesis, previous haematologica | 2017; 102(5)


Long-latency engraftment of human AML in NSG mice

studies also showed engraftment of different subclones from one AML case in individual mice, suggesting that the ability to repopulate mice is not in fact restricted to certain genetic features. Further investigations (e.g. using exome sequencing comparison of patient-derived versus corresponding xenogeneic AML cells) are required to investigate whether atypical mutations (not captured by our next-generation sequencing platform) may however be acquired to facilitate growth in the murine environment. This latter could actually also explain the faster outgrowth of leukemic cells observed in retransplant assays. AML with specific genetic backgrounds (e.g. favorable risk) may particularly rely on cytokine and/or niche support for optimal in vivo survival and growth. Their growth in NSG mice and capacity to engraft by outcompeting endogenous hematopoietic stem cells may thus be hampered by limited cross-reactivity of murine factors with human receptors. Immunosuppressed mice engineered to express human cytokines (“humanized mice”), or Kit mutations impairing murine host hematopoietic stem cells, respectively, were found to improve engraftment of healthy human hematopoietic and recently also leukemic cells.28-30 Further supporting the notion that engraftment is impaired by cross-species differences, implanted humanized BM ossicles – which provide a humanized niche environment within mice – were shown to facilitate repopulation with APL and myelofibrosis cells.31 In our study, similar (high) Ki67 positivity was detected in the BM of mice infiltrated with different AML at times of high leukemic burden. Furthermore, the time period from the last detected negative BM sample to engraftment and leukemia induction remained short in all transplanted AML irrespectively of engraftment latency or molecular risk subtype (data not shown). Thus, at times of advanced disease, the murine environment appears to support the proliferation of AML cells of various genetic backgrounds (including favorable-risk) comparably. However, AML cells are known to secrete cytokines and by themselves contribute to a pro-leukemogenic environment; if they are expanded to detectable numbers, inherent milieu disadvantages (e.g. lack of cross-reactive cytokines in NSG mice) might thereby be overruled. The situation might be different at the time of disease initiation, when only a few human AML cells are present in the mouse. Thus, the data presented here do not exclude the possibility that some AML engraft with long latency because they proliferate less in the murine BM at early (pre-detection) time-points, due to oncogene-specific differences in interactions with the healthy BM environment. Further studies are needed to investigate whether, specifically at stages of minimal disease burden, NSG mice provide suboptimal conditions for the outgrowth of certain genetic backgrounds, and to test whether further genetic modification of NSG mice via knock-in of human cytokines could accelerate disease induction thereby reducing more elevated costs raised by long-term mouse maintenance. Previous studies in mouse xenograft models have shown that leukemia derives from rare LIC.1,4,19,25,32 Consistently, lowering transplanted cell numbers below a critical level (as part of limiting dilution assays) resulted in engraftment failure, which could not be compensated by extending follow-up. Reduced LIC frequency, as also measured by us in long-latency versus standardly engrafting AML, might thus be another explanation for the negahaematologica | 2017; 102(5)

tive or delayed engraftment observed with certain AML subtypes (particularly intermediate- and favorable-risk). Supporting this notion, Griessinger and colleagues recently reported a surrogate ex vivo short-term co-culture system showing lower frequencies of long-term culture-initiating cells in good- versus intermediate- and poor-risk AML.33 Furthermore, indirect support of a lower LIC frequency in favorable-risk AML was provided by mRNA analyses, in which favorable clinical outcome was associated with reduced expression of stem cell genes.4,32,34 Alternatively, the reduced expression of stem cell genes might derive from qualitative alterations of LIC in these subtypes.4 LIC evaluation ultimately requires in vivo studies. LIC frequency and phenotype are thus largely understudied in AML previously considered non-engraftable, particularly in favorable-risk AML. In our study, we observed lower LIC frequencies in long-latency versus standard engrafters, suggesting that this factor does indeed contribute to the observed difference in leukemia induction latency. However, follow-up studies involving larger numbers of patients are necessary to confirm this conclusion. Furthermore, long-latency engrafting AML contained fewer numbers of leukemic cells able to home to the BM, when compared to standard engrafting AML. The low homing capacity and corresponding delay in leukemia induction observed with some AML may result from reduced expression of adhesion molecules, which in healthy hematopoietic stem and progenitor cells have been shown to essentially regulate BM homing.35,36,37 Homing is a core property of healthy hematopoietic stem cells.38 Although understudied in the setting of AML, homing appears as a pre-requisite for BM niche occupation and long-term leukemia induction, and LIC should be included within the subset of homing AML cells. Future studies are needed to evaluate homing as a surrogate assay to measure LIC content and perhaps even as a short-term functional assay that can predict disease aggressiveness. Several attempts have been made to identify the phenotype of LIC. While they were reported to reside in the CD34+ or CD34+CD38– blast subpopulations, CD34+CD38+ and CD34– LIC have also been described (e.g. recently via expression of GPR56)32 and specifically linked to certain genetic backgrounds (e.g. AML with NPM1 mutation).1,4,19 We next used our model to investigate in vivo leukemogenesis and LIC from AML with inv(16). All three transplanted samples of this subtype robustly engrafted NSG mice, albeit with longer latency. Engraftment was achieved even in the absence of prior irradiation, indicating that conditioning and knock-in of human cytokines are not mandatory for leukemia induction, although they likely accelerate this process.12 However, in spite of extended follow-up, in vivo leukemogenesis was only observed from CD34+ (but not CD34–) inv(16) blasts. Furthermore, gene expression analyses demonstrated enhanced expression of stem cell genes in CD34+ cells, reinforcing CD34 surface expression as a LIC marker in this AML subtype. In conclusion, we present a model that enables in vivo studies of AML subtypes previously considered nonengraftable, using widely accessible and well-characterized NSG mice. We show that in vivo xenotransplantation of human AML cells in NSG mice faithfully recapitulates human disease since xenogeneic leukemic cells: (i) retain the phenotypic, genetic and functional leukemia-initiating 863


A.M. Paczulla et al.

properties of the corresponding pre-transplant AML samples; (ii) follow disease kinetics and mortality induction in mice according to molecular risk groups established in humans; and (iii) importantly, can persist in animals over several months at undetectable levels without losing disease-initiating properties, thus mimicking the clinical course of AML in humans. Mechanistically, the longer time to detectable engraftment observed with some AML (e.g. favorable molecular risk) may be due to lower numbers of transplanted LIC or, alternatively, to qualitative alterations in LIC resulting in lower homing to the BM and/or reduced growth rates in the mouse microenviron-

References 1. Lapidot T, Sirard C, Vormoor J, et al. A cell initiating human acute myeloid leukaemia after transplantation into SCID mice. Nature. 1994;367(6464):645-648. 2. Meyer LH, Debatin KM. Diversity of human leukemia xenograft mouse models: implications for disease biology. Cancer Res. 2011;71(23):7141-7144. 3. Theocharides AP, Rongvaux A, Fritsch K, Flavell RA, Manz MG. Humanized hematolymphoid system mice. Haematologica. 2016;101(1):5-19. 4. Eppert K, Takenaka K, Lechman ER, et al. Stem cell gene expression programs influence clinical outcome in human leukemia. Nat Med. 2011;17(9):1086-1093. 5. McDermott SP, Eppert K, Lechman ER, Doedens M, Dick JE. Comparison of human cord blood engraftment between immunocompromised mouse strains. Blood. 2010;116(2):193-200. 6. Sanchez PV, Perry RL, Sarry JE, et al. A robust xenotransplantation model for acute myeloid leukemia. Leukemia. 2009;23(11): 2109-2117. 7. Rombouts WJ, Blokland I, Löwenberg B, Ploemacher RE. Biological characteristics and prognosis of adult acute myeloid leukemia with internal tandem duplications in the Flt3 gene. Leukemia. 2000;14(4):675683. 8. Mrózek K, Marcucci G, Nicolet D, et al. Prognostic significance of the European LeukemiaNet standardized system for reporting cytogenetic and molecular alterations in adults with acute myeloid leukemia. J Clin Oncol. 2012;30(36):45154523. 9. Rombouts WJ, Martens AC, Ploemacher RE. Identification of variables determining the engraftment potential of human acute myeloid leukemia in the immunodeficient NOD/SCID human chimera model. Leukemia. 2000;14(5):889-897. 10. Rongvaux A, Takizawa H, Strowig T, et al. Human hematolymphoid system mice: current use and future potential for medicine. Annu Rev Immunol. 2013;31:635-674. 11. Manz MG. Human-hematolymphoid-system mice: opportunities and challenges. Immunity. 2007;26(5):537-541. 12. Ellegast JM, Rauch PJ, Kovtonyuk LV, et al. inv(16) and NPM1mut AMLs engraft human cytokine knock-in mice. Blood. 2016;128 (17):2130-2134. 13. Estey EH. Treatment of relapsed and refractory acute myelogenous leukemia. Leukemia. 2000;14(3):476-479. 14. Breems DA, Van Putten WL, Huijgens PC, et

864

15.

16.

17.

18.

19.

20.

21.

22.

23.

24. 25.

26.

ment. When applied to favorable-risk AML with inv(16), our model indicates that in this subtype leukemia initiation occurs from CD34+ blasts with enhanced expression of stem cell genes. Acknowledgments This study was supported by grants from the DFG (Deutsche Forschungsgemeinschaft, LE 2483/7-1) and SNF (Schweizerischer Nationalfonds, NMS1820) to CL. We thank Joëlle S. Müller for help with the isolation of mononuclear cells from patients’ samples and the animal and FACS core facilities of the Department of Biomedicine at the University Hospital Basel.

al. Prognostic index for adult patients with acute myeloid leukemia in first relapse. J Clin Oncol. 2005;23(9):1969-1978. Mazurier F, Doedens M, Gan OI, Dick JE. Rapid myeloerythroid repopulation after intrafemoral transplantation of NOD-SCID mice reveals a new class of human stem cells. Nat Med. 2003;9(7):959-963. Jacobsen KR, Kalliokoski O, Teilmann AC, Hau J, Abelson KS. Postsurgical food and water consumption, fecal corticosterone metabolites, and behavior assessment as noninvasive measures of pain in vasectomized BALB/c mice. J Am Assoc Lab Anim Sci. 2012;51(1):69-75. Konantz M, André MC, Ebinger M, et al. EVI-1 modulates leukemogenic potential and apoptosis sensitivity in human acute lymphoblastic leukemia. Leukemia. 2013;27(1):56-65. Kunder S, Calzada-Wack J, Hölzlwimmer G, et al. A comprehensive antibody panel for immunohistochemical analysis of formalinfixed, paraffin-embedded hematopoietic neoplasms of mice: analysis of mouse specific and human antibodies cross-reactive with murine tissue. Toxicol Pathol. 2007;35(3):366-375. Taussig DC, Vargaftig J, Miraki-Moud F, et al. Leukemia-initiating cells from some acute myeloid leukemia patients with mutated nucleophosmin reside in the CD34(-) fraction. Blood. 2010;115(10): 1976-1984. Wang C, Sashida G, Saraya A, et al. Depletion of Sf3b1 impairs proliferative capacity of hematopoietic stem cells but is not sufficient to induce myelodysplasia. Blood. 2014;123(21):3336-3343. Bonnet D, Dick JE. Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nat Med. 1997;3(7):730-737. Pearce DJ, Taussig D, Zibara K, et al. AML engraftment in the NOD/SCID assay reflects the outcome of AML: implications for our understanding of the heterogeneity of AML. Blood. 2006;107(3):1166-1173. Ailles LE, Gerhard B, Kawagoe H, Hogge DE. Growth characteristics of acute myelogenous leukemia progenitors that initiate malignant hematopoiesis in nonobese diabetic/severe combined immunodeficient mice. Blood. 1999;94(5):1761-1772. Grimwade D, Enver T. Acute promyelocytic leukemia: where does it stem from? Leukemia. 2004;18(3):375-384. Vargaftig J, Taussig DC, Griessinger E, et al. Frequency of leukemic initiating cells does not depend on the xenotransplantation model used. Leukemia. 2012;26(4):858-860. Klco JM, Spencer DH, Miller CA, et al.

27.

28.

29.

30.

31.

32.

33.

34.

35. 36. 37.

38.

Functional heterogeneity of genetically defined subclones in acute myeloid leukemia. Cancer Cell. 2014;25(3):379392. Quek L, Otto GW, Garnett C, et al. Genetically distinct leukemic stem cells in human CD34- acute myeloid leukemia are arrested at a hemopoietic precursorlike stage. J Exp Med. 2016;213(8):15131535. Rongvaux A, Willinger T, Martinek J, et al. Development and function of human innate immune cells in a humanized mouse model. Nat Biotechnol. 2014;32(4):364-372. Cosgun KN, Rahmig S, Mende N, et al. Kit regulates HSC engraftment across the human-mouse species barrier. Cell Stem Cell. 2014;15(2):227-238. Feuring-Buske M, Gerhard B, Cashman J, Humphries RK, Eaves CJ, Hogge DE. Improved engraftment of human acute myeloid leukemia progenitor cells in beta 2-microglobulin-deficient NOD/SCID mice and in NOD/SCID mice transgenic for human growth factors. Leukemia. 2003;17(4):760-763. Reinisch A, Thomas D, Corces MR, et al. A humanized bone marrow ossicle xenotransplantation model enables improved engraftment of healthy and leukemic human hematopoietic cells. Nat Med. 2016;22(7):812-821. Pabst C, Bergeron A, Lavallée VP, et al. GPR56 identifies primary human acute myeloid leukemia cells with high repopulating potential in vivo. Blood. 2016;127(16):2018-2027. Griessinger E, Anjos-Afonso F, Vargaftig J, et al. Frequency and dynamics of leukemiainitiating cells during short-term ex vivo culture informs outcomes in acute myeloid leukemia patients. Cancer Res. 2016;76(8): 2082-2086. Verhaak RG, Wouters BJ, Erpelinck CA, et al. Prediction of molecular subtypes in acute myeloid leukemia based on gene expression profiling. Haematologica. 2009;94 (1):131-134. Cancelas JA, Williams DA. Stem cell mobilization by beta2-agonists. Nat Med. 2006;12(3):278-279. Lapidot T, Dar A, Kollet O. How do stem cells find their way home? Blood. 2005;106(6):1901-1910. Jin L, Hope KJ, Zhai Q, Smadja-Joffe F, Dick JE. Targeting of CD44 eradicates human acute myeloid leukemic stem cells. Nat Med. 2006;12(10):1167-1174. Suárez-Álvarez B, López-Vázquez A, López-Larrea C. Mobilization and homing of hematopoietic stem cells. Adv Exp Med Biol. 2012;741:152-170.

haematologica | 2017; 102(5)


ARTICLE

Acute Myeloid Leukemia

Minimal residual disease prior to allogeneic hematopoietic cell transplantation in acute myeloid leukemia: a meta-analysis

Sarah A. Buckley,1 Brent L. Wood,2 Megan Othus,3 Christopher S. Hourigan,4 Celalettin Ustun,5 Michael A. Linden,6 Todd E. DeFor,7 Michele Malagola,8 Chloe Anthias,9,10 Veronika Valkova,11 Christopher G. Kanakry,12,13 Bernd Gruhn,14 Francesco Buccisano,15 Beth Devine16-18 and Roland B. Walter19-21

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(5):865-873

Hematology/Oncology Fellowship Program, University of Washington, Seattle, WA, USA; Division of Hematopathology, Department of Laboratory Medicine, University of Washington, Seattle, WA, USA; 3Public Health Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, WA, USA; 4Hematology Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD, USA; 5Division of Hematology, Oncology and Transplantation, Department of Medicine, University of Minnesota, Minneapolis, MN, USA; 6Division of Hematopathology, Department of Laboratory Medicine and Pathology, University of Minnesota, Minneapolis, MN, USA; 7Biostatistics and Bioinformatics Core, Masonic Cancer Center, University of Minnesota, Minneapolis, MN, USA; 8Unit of Blood Diseases and Stem Cell Transplantation, University of Brescia, A.O. Spedali Civili, Italy; 9Anthony Nolan Research Institute, London, UK; 10Royal Marsden Hospital, London, UK; 11Institute of Haematology and Blood Transfusion, Prague, Czech Republic; 12Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University School of Medicine, Baltimore, MD, USA; 13Experimental Transplantation and Immunology Branch, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA; 14Department of Pediatrics, Jena University Hospital, Germany; 15Department of Hematology, Fondazione Policlinico Tor Vergata, Rome, Italy; 16 Pharmaceutical Outcomes Research and Policy Program, University of Washington, Seattle, WA, USA; 17Department of Health Services, University of Washington, Seattle, WA, USA; 18Department of Biomedical Informatics, University of Washington, Seattle, WA, USA; 19 Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, WA, USA; 20 Department of Medicine, Division of Hematology, University of Washington, Seattle, WA, USA and 21Department of Epidemiology, University of Washington, Seattle, WA, USA 1 2

ABSTRACT

Correspondence:

M

inimal residual disease prior to allogeneic hematopoietic cell transplantation has been associated with increased risk of relapse and death in patients with acute myeloid leukemia, but detection methodologies and results vary widely. We performed a systematic review and meta-analysis evaluating the prognostic role of minimal residual disease detected by polymerase chain reaction or multiparametric flow cytometry before transplant. We identified 19 articles published between January 2005 and June 2016 and extracted hazard ratios for leukemia-free survival, overall survival, and cumulative incidences of relapse and non-relapse mortality. Pre-transplant minimal residual disease was associated with worse leukemia-free survival (hazard ratio=2.76 [1.90-4.00]), overall survival (hazard ratio=2.36 [1.73-3.22]), and cumulative incidence of relapse (hazard ratio=3.65 [2.53-5.27]), but not nonrelapse mortality (hazard ratio=1.12 [0.81-1.55]). These associations held regardless of detection method, conditioning intensity, and patient age. Adverse cytogenetics was not an independent risk factor for death or relapse. There was more heterogeneity among studies using flow cytometry-based than WT1 polymerase chain reaction-based detection (I2=75.1% vs. <0.1% for leukemia-free survival, 67.8% vs. <0.1% for overall survival, and 22.1% vs. <0.1% for cumulative incidence of relapse). These results demonstrate a strong relationship between pre-transplant minimal residual disease and post-transplant relapse and survival. Outcome heterogeneity among studies using flow-based methods may underscore site-specific methodological differences or differences in test performance and interpretation.

haematologica | 2017; 102(5)

buckleys@uw.edu

Received: November 2, 2016. Accepted: January 20, 2017. Pre-published: January 25, 2017. doi:10.3324/haematol.2016.159343 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/865 Š2017 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.

865


S.A. Buckley et al.

Introduction Morphologic complete remission (CR), defined by the presence of <5% bone marrow blasts and recovery of peripheral blood counts, is the long-standing standard for response assessment in acute myeloid leukemia (AML).1-5 Based on estimates of normal marrow cellularity,6 however, this cutoff allows for the presence of up to 1010 leukemic blasts or more. It is therefore not surprising that relapse remains the major cause of treatment failure among patients who have achieved a morphologic CR.4,5 Significant effort has gone into developing tools to identify minimal (or, perhaps more appropriately, measurable) residual disease (MRD), including multi-parametric flow cytometry (MFC) to enumerate myeloid cell populations with immunophenotypic abnormalities, polymerase chain reaction (PCR) to quantify leukemia-associated mutations or RNA transcript levels, and cytogenetic / fluorescence in situ hybridization to detect chromosome level changes specific to the malignant clone. Among these modalities, MFC- and PCR-based approaches have the highest sensitivity and are increasingly employed in the clinic.7-12 A large number of studies has demonstrated worse outcomes for patients who have MRD compared to similarly treated patients in whom no MRD can be detected. This relationship has been observed during/after induction and post-remission chemotherapy courses as well as before and after hematopoietic cell transplantation (HCT).7-12 The magnitude of the association between MRD status and

risk of relapse varies widely between studies, however, as do the details of the detection methods. In addition to differences in the specifics of the MRD techniques across institutions, there are also differences in cut-points chosen to define MRD positivity, the patient material that is used to perform the MRD assay on (i.e., peripheral blood or bone marrow), and the timing as well as frequency with which MRD assessments are obtained. In this meta-analysis, we focused on MRD assessed immediately before allogeneic HCT in patients with AML, other than acute promyelocytic leukemia (APL). Besides ascertaining the relationship between pre-HCT MRD and post-transplant outcomes, we also investigated whether, and to what degree, the prognostic role of MRD is influenced by the method of MRD detection.

Methods We searched PubMed/MEDLINE and EMBASE (Online Supplementary Table S1) for English language articles published between January 2005 and June 2016 that reported on the association between pre-HCT MRD (by PCR and/or MFC) and postHCT survival in patients with non-APL AML in morphologic CR. Two authors (S.A.B. and R.B.W.) independently reviewed the search results. We excluded studies with <15 patients or <6 months of follow up. If needed, authors of included studies were contacted for additional information. Our search yielded 344 reports, which were screened according to 2009 PRISMA

Figure 1. PRISMA flow diagram for study selection. HCT: hematopoietic cell transplantation; MRD: minimal residual disease.

866

haematologica | 2017; 102(5)


Pre-HCT MRD and Post-HCT Outcomes in AML

Guidelines (Figure 1). For studies of interest, we collected data on the number of patients, median/range age, median follow-up time, percentage of patients with adverse-risk cytogenetics (using the classification criteria reported by study authors), percentage of patients receiving myeloablative (MA) vs. reduced intensity conditioning (RIC), interval between MRD detection and HCT, and details of the MRD detection method. We assessed risk of bias using an instrument based on the Quality in Prognostic Studies (QUIPS),13 modified to reflect our judgment about potential biases (Online Supplementary Table S2). Finally, we obtained data for leukemia-free survival (LFS), overall survival (OS), and cumulative incidence of relapse (CIR) and non-relapse mortality (NRM) from the date of HCT. We used a hierarchical approach14 to compare outcomes of MRDpos and MRDneg subjects: (i) when available, we used observed hazard ratios (HRs) and confidence intervals (CIs); (ii) when Kaplan-Meier curves were provided, we used Enguage Digitizer version 4.1 to calculate HRs and CIs based on an established algorithm,15 and (iii) for articles providing survival data at single time points, we estimated HRs based on exponential decay. We performed a random-effects meta-analysis, with inter-study heterogeneity described using the I2 statistic16 (STATA version 14; StataCorp, College Station, TX, USA). Cut-points between MRD positivity and negativity were based on criteria specified by the individual publications. In one,17 no cut-point was specified for Wilms Tumor 1 (WT1) transcript level. As other studies used cut-points in the range of 50-70 copies/104 reference gene copies,1820 and as no events were observed at WT1 levels <65, a cutoff of 70 was used. In another study18 that used a WT1 cutoff of 50, there were no relapses in the MRDneg group (n=25) by 6.6 years. As no HR could be calculated, this study was not incorporated into pooled CIR results. In two studies in which HRs were extracted from survival curves,21,22 curves were portrayed for subgroups within MRDpos and MRDneg patients; here, a weighted average of the HR between groups by number of patients per group was used to obtain a final HR. In one study19 reporting results by MFC and by WT1 PCR, we used MFC results for overall analysis, as these data were more complete. Subgroup analyses involved stratification by MRD detection method, age, and conditioning intensity. We calculated the ratio of the percentage of patients with adverse cytogenetics in the MRDpos and MRDneg groups. If HRs for survival outcomes were higher in studies where this ratio was greater, it would indicate that adverse cytogenetics might be an independent negative prognostic factor.23 We used meta-regression to test this hypothesis.

Results Included studies Our search yielded 19 unique publications with a total of 1,431 patients (Table 1).17-19,21,22,24-37 Details of transplant and conditioning regimens are shown in the Online Supplementary Table S3. The sole method of MRD detection was MFC in 9 studies22,24,26-29,33,36,37 and WT1 PCR in 5,17,18,30-32 while one study reported results separately for MFC- and WT1 PCR-based detection.19 Four studies used combination methods;21,25,34,35 all of these included MFC, and 3 also included PCR-based detection. Among studies using MFC-based detection, the cut-point between MRD positivity and negativity was fairly uniform: 11 of 14 used the limit of detection for the assay (around 0.1%), while 3 specified a cutoff of 0.1%,26,33,36 which corresponded roughly to the limit of detection in these cases. In other words, heterogeneity in cut-points was primarily determined by differences in performance characteristics and haematologica | 2017; 102(5)

interpretation of the assay rather than the cut-points selected. Among studies that only used PCR-based methods, all assessed quantitative PCR for WT1, while one study31 utilized a panel of other genes in addition to WT1. Two studies, both using combination approaches for MRD detection, targeted PCR at AML-specific mutations (e.g., Fms related tyrosine kinase 3 internal tandem duplifusions genes (e.g., cation [FLT3/ITD])21 or RUNX1/RUNX1T1)25 present at diagnosis. Among studies quantifying WT1 transcript levels, most normalized against expression of ABL1; MRDpos cutoff levels varied between 50-70 copies of WT1 per 104 copies of ABL1.17,19,32,34 Five studies were considered as having a high risk of bias: the MRD measurement technique was implicated in all cases, and study confounding was felt possible in 2 of these cases (Figure 2). For 11 studies, we were able to obtain HRs for all reported outcomes from the manuscript or personal communication; for the other 8 studies, HRs were extrapolated from Kaplan-Meier curves or survival point estimates (n=4).18,26,34,36 MRD was measured within 60 days of HCT in all studies in which this information was reported, and within 30 days in all but one study.35

Association between pre-HCT MRD status and post-HCT outcomes Overall, MRD positivity was associated with worse LFS (HR=2.76 [1.90-4.00], I2=70.0%), OS (HR=2.36 [1.73-3.22], I2=59.7%), and CIR (HR=3.65 [2.53-5.27], I2=37.9%) but not NRM (HR=1.12 [0.81-1.55], I2<0.1%). After removing studies with a high risk of bias in any domain, MRD remained strongly associated with worse LFS (HR=3.24 [2.17-4.83], I2=64.5%), OS (HR=2.64 [1.87-3.72], I2=57.8%), and CIR (HR=4.06 [2.70-6.12], I2=48.0%) while, again, there was no statistically significant association with NRM (HR=1.18 [0.80-1.75] I2=0.9%).

Effect of MRD detection method on post-HCT outcomes In subgroup analyses, being MRDpos was associated with an increased risk of relapse and mortality regardless of the detection method (Table 2). For CIR, the HR for WT1 PCRbased methods was statistically significantly larger than for MFC-based methods. Figure 3 shows a forest plot for the 17 studies reporting on the primary outcome of LFS, while similar plots for OS, CIR, and NRM can be found in the Online Supplementary Figures S1-S3. Results for studies using MFC-based methods were more heterogeneous than those using WT1 PCR or combination methods for LFS (I2=75.1% vs. <0.1% and 57.2%), OS (I2=67.8% vs. <0.1% and 12.5%), and CIR (I2=22.1% vs. <0.1% and 6.7%). After excluding studies with a high risk of bias in any domain, WT1 PCR-based studies and combination methods continued to have low heterogeneity for LFS, OS, and CIR (all I2<0.1%), whereas MFC-based studies showed persistent and considerable heterogeneity for LFS (I2=81.5%), OS (I2=73.8%), and CIR (I2=46.4%). While all MFC-based studies analyzed bone marrow tissue, WT1 PCR-based studies were mixed between the use of marrow and peripheral blood for analysis. Restriction to studies that reported data from peripheral blood18,31,32 yielded essentially identical results. Outcomes for MFC-based studies were similar regardless of whether residual disease was detected via gating for the original leukemia-associated immunophenotype or based on detecting a phenotype 867


S.A. Buckley et al.

different from normal, although results for the latter were more heterogeneous (I2 89.7% vs. 32.5% for LFS, 88.3% vs. 0.0% for OS, and 70.8% vs. 0.0% for CIR). There were no significant differences in outcomes between MFCbased studies by number of fluorochromes (<6 vs. ≥6) used.

Effect of patient age on post-HCT outcomes On subgroup analysis of age 0-20,18,25,26,30,36 21-40,28,31,34 and >40,17,19,21,22,24,27,29,32,33,35 we found no difference in the effect of MRD between groups. The same was true after exclusion of studies with a high risk of bias. Among studies reporting on older patients, there was sufficient data to further stratify into ages 40-60 and >60 for the LFS endpoint; the HR for this outcome was similar in these subgroups (HR=2.67 [1.46-4.86], I2=81.1%; HR=3.02 [0.9010.08], I2=52.3%, respectively). When we restricted our

analysis to studies using primarily MA conditioning, 2 were primarily pediatric (median age 0-20),25,30 4 involved young adults (median age 20-40),27,28,31,34 and 5 involved older adults (median age >40).19,21,24,32,35 The association between MRD and LFS was similar in all age groups, though between-study heterogeneity was high (age 0-20: HR=3.45 [0.39-30.86], I2=89.5%; age 20-40: HR=2.35 [1.10-5.02], I2=72%; age >40: HR=3.56 [1.79-7.05], I2=77.5%).

Effect of conditioning intensity on post-HCT outcomes Next, we considered whether differences in conditioning regimen intensity might explain between-study heterogeneity, particularly in light of conflicting results from Ustun et al.27 showing in a large cohort (n=203) that MA conditioning could compensate for the increased hazard for relapse and mortality associated with being MRDpos,

Table 1. Characteristics of included studies.

MRD method

MRD source

Cutoff for MRDpos

MRDneg (n)

MRDpos (n)

Age, median (range)

% MA

Bleyzac et al.26 Ustun et al.27

MFC (LAIP) MFC (DFN, 4-color)

BM BM

0.1% Limit of detection (0.1%)

18 178

14 25

9 (0-19) 47 (0-74)

Zheng et al.25

MFC (LAIP, 4-color) or PCR (fusion genes, multiple) MFC (DFN, 10-color)

BM

40

32

235

76

PCR (WT1, multi-gene)

PB

MFC: 0.01% PCR: limit of detection Limit of detection (0.1%) Different for each gene

38

10

BM

Tian et al.28

MFC (LAIP, 6-color) PCR (WT1) MFC (LAIP, 4-color)

MRDneg 16 (3-28) MRDpos 19 (6-36) MRDneg 47 (19-71) MRDpos 51 (18-72) MRDneg 34 (12-59) MRDpos 34 (16-53) 44 (18-64)

NR 39% MRDneg 60% MRDpos 100%

Walter et al.29

MFC (DFN, 10-color)

BM

PCR (WT1)

PB or BM

MFC (LAIP, 3-color)

Study

Araki et al.24 Goswami et al.31 Rossi et al.19

BM

MFC: 0.1% 22 (MFC) WT1: 64 / 104 copies ABL 19 (PCR) Limit of detection 21 65

21

17

23

53

35

0.1%

18

11

MFC: limit of detection PCR: limit of detection MFC: limit of detection WT1: 60/104 copies ABL Limit of detection

76

25

110

20

26 (3-54)

100%

40

19

BM PB

0.1% 50 / 104 copies ABL1

27 29

9 13

PCR (WT1)

BM

70 / 104 copies ABL1

5

13

90% MRDneg 84% MRDpos 100% 79% MRDneg 85% MRDpos 0%

PCR (WT1)

PB

25

11

MFC (LAIP, 3-color)

BM

0.5 (normalized to WT1 level in control cells) Limit of detection

MRDneg 43 (20-65) MRDpos 50 (28-65) (Pediatric) MRDneg 43 (20-63) MRDpos 51 (36-63) MRDneg 61 (39-66) MRDpos 61 (36-68) 10 (3-22)

12

5

(Adult)

100%

BM

Bastos-Oriero et al.33 MFC (LAIP, 4-color)

BM

Kanakry et al.21 MFC (LAIP), PCR (FLT3, NPM1), and/or cytogenetics / FISH Wang et al.34 MFC (LAIP) and PCR (WT1) Grubovikj et al.35 MFC (DFN) or cytogenetics / FISH Leung et al.36 MFC (LAIP, 4-color) Valkova et al.32 PCR (WT1)

BM

Candoni et al.17 Jacobsohn et al.18

Anthias et al.22

Laane et al.37

89% MRDneg 100% MRDpos 100%

MRDneg 31 (15-55) MRDpos 32 (16-58) MRDneg 62 (20-75) MRDpos 63 (33-74) MRDneg 4 (1-21) MRDpos 13 (2-18) MRDneg 44 (18-70) MRDpos 52 (21-70) MRDneg 41 (19-62) MRDpos 50 (19-63) 51 (20-66)

BM

Limit of detection (0.1%) -3 5×10 normalized to β2M expression Limit of detection (0.4%)

Woehlecke et al.30

8 (MFC) 10 (PCR) 32

100%

BM

NR 0% 100% 40% MRDneg 60% MRDpos 100% MRDneg 72% MRDpos 100%

100%

PB: peripheral blood; BM: bone marrow; NR: not reported; MFC: multiparametric flow cytometry; LAIP: leukemia-associated immunophenotype; MA: myeloablative; DFN: different from normal; FISH: fluorescence in situ hybridization; MRD: minimal residual disease; PCR: polymerase chain reaction; pos: positive; neg: negative.

868

haematologica | 2017; 102(5)


Pre-HCT MRD and Post-HCT Outcomes in AML

while Walter et al.29 showed no such effect in 241 patients. Among studies reporting LFS as an outcome, 14 reported on the fraction of MRDpos and MRDneg patients who underwent MA versus RIC HCT. To test whether higher intensity transplant might reduce the negative impact of being MRDpos, we specifically analyzed studies in which >75% of MRDpos and MRDneg patients received MA HCT (n=12 for LFS endpoint), and compared the results with studies where 0% of patients received MA HCT (n=3 for LFS endpoint). Results from Ustun et al.27 were reported separately for MA and RIC patients within their publication, and for the purposes of this analysis, we treated these sets of results as two separate studies. As shown in Table 2 and as a forest plot in the Online Supplementary Figure S4, there was no indication that MA conditioning was able to attenuate the negative effects associated with MRD positivity on LFS, OS, or CIR. In contrast, the HRs for MA studies were numerically higher than for the few RIC studies, although the large confidence intervals exclude a definitive conclusion as to whether conditioning intensity affects the association between MRD status and post-HCT outcomes. The exclusion of high-risk studies did not fundamentally change these results and conclusions. Not surprisingly, all studies using RIC conditioning involved older adults (the >40 age group as stratified above).

ria,21,24,29,35 while one incorporated mutational profiling.34 The ratio of the proportion of adverse-risk cytogenetics among MRDpos to MRDneg patients ranged from roughly equal to 7.5 times higher in the MRDpos group. We used meta-regression to measure how HRs for LFS changed with variations in this risk ratio and found that differences in adverse-risk cytogenetics between MRDpos and MRDneg groups did not account for a significant proportion of between-study variance (R2): R2 -9.15% (P=0.82) for all studies, and 14.83% (P=0.92) after excluding high-risk studies (Figure 4). Results were similar when the study with the highest ratio of 7.5 was excluded from this analysis (P=0.62). Similarly, adverse-risk cytogenetics was not an independent prognostic factor for OS (P=0.11), CIR (P=0.85), or NRM (P=0.99).

Effect of cytogenetic risk on post-HCT outcomes

Discussion

Most studies reporting cytogenetics in MRDpos and MRDneg patients used the Southwest Oncology Group17,27,28,32,33 or 2010 Medical Research Council crite-

The findings from this meta-analysis support our main conclusion that the presence of MRD before allogeneic

Testing for publication bias Funnel plot analyses for each survival outcome are shown in Figure 5 as a graph of log-HR versus the variance in the log-HR. These plots did not suggest a publication bias, although they indicated that the publication of studies considered to have a high risk of bias could bias overall study results towards the null for LFS, OS, and CIR.

2016

2013

Figure 2. Risk of bias assessment illustrating review authorsâ&#x20AC;&#x2122; judgments about each risk of bias item for each included study.

haematologica | 2017; 102(5)

869


S.A. Buckley et al.

HCT identifies patients at a higher risk of relapse and shorter survival relative to patients in whom no evidence of MRD is found. Although we were unable to incorporate results from one study with an incalculable HR for CIR (based on the lack of relapses among MRDneg patients), the findings from that report similarly supported our conclusion. The association between MRD and postHCT relapse and mortality is robust, and is seen within all patient ages and regardless of which detection method is used. It is similarly found in those undergoing MA conditioning as well as RIC transplants without discernible difference in strength of association between these cohorts, suggesting that higher conditioning regimen intensity may not be able to overcome the adverse impact of MRD. To the extent that we were able to control for differences in cytogenetic risk with meta-regression, the negative impact of being MRDpos superseded any potential adverse effects of having poor-risk cytogenetics. In comparison, our analysis indicates that pre-HCT MRD is not associated with a significantly increased risk of NRM, in line with the notion that the association between pre-HCT MRD and OS is entirely accounted for by disease relapse without significant contribution from HCT toxicity. Although our meta-analysis demonstrates a significant association between pre-HCT MRD status and post-HCT outcomes with both WT1 PCR- and MFC-based assays, we found a greater degree of heterogeneity in survival

estimates in studies with MFC-based detection methods. This heterogeneity could not be accounted for by differences in patient age, conditioning regimen intensity, or cytogenetic risk. In addition, the cut-points between MRD positivity and negativity were primarily determined by the limits of detection of each particular assay, indicating that chosen cut-points are unlikely to account for heterogeneity. We were, however, able to show that at least some of this heterogeneity may be accounted for by study-specific differences in approach to MFC, with studies detecting residual disease based on initial leukemiaassociated immunophenotypes having more uniform results than those using a “different from normal” approach. Other possible causes of heterogeneity might include site-specific differences in MFC methodology, including differences in antigens and fluorochromes used, methods for cell lysis, number of events collected, or specifics of the aspirate used for analysis, with an increasing risk of hemodilution with each pull. If such speculation is correct, efforts towards standardization/harmonization of MRD methods; as pioneered in acute lymphoblastic leukemia38 and currently underway for AML; might ultimately lead to less heterogeneous data with MFC-based MRD assays. In contrast, despite some heterogeneity in PCR targets and cut-points, PCR methodology may be relatively more standardized, accounting for more uniform results. As an illustration, the risk estimates

Table 2. Pooled HRs [95% CI] and inter-study heterogeneity for all studies (above) and excluding a high risk of bias (below).

Subset Method MFC PCR Combination Median age 0-20 21-40 >40 Conditioning >75% MA 0% MA

Subset Method MFC PCR Combination Median age 0-20 21-40 >40 Conditioning >75% MA 0% MA

All Studies LFS

OS

CIR

NRM

1.98 [1.26-3.10]  5.25 [3.08-8.95] ☐ 1.86 [1.25-2.77] ☐

2.41 [1.36-4.29]  5.80 [3.57-9.42] ☐ 1.79 [1.06-3.01] 

2.81 [1.94-4.08] ☐ 9.53 [4.48-20.29] ☐ 3.73 [1.94-7.18] ☐

1.11 [0.63-1.95] ☐ 1.51 [0.57-4.00] ☐ 1.15 [0.57-2.33] ☐

3.12 [1.29-7.57]  2.60 [1.36-4.99]  2.25 [1.47-3.47] 

3.33 [0.95-11.6]  3.02 [1.27-7.16]  2.69 [1.64-4.42] 

3.57 [0.67-18.91]  5.13 [2.37-9.64] ☐ 3.33 [2.18-5.11] 

1.13 [0.52-2.4] ☐ -1.23 [0.77-1.97] ☐

2.64 [1.77-3.93]  2.05 [0.78-5.39] 

2.86 [1.80-4.55]  2.09 [1.33-3.29] ☐

4.21 [2.70-6.58]  3.23 [1.88-5.53] ☐

1.39 [0.94-2.07] ☐ 0.58 [0.22-1.52] ☐

OS

Excluding Studies with High Risk of Bias LFS

CIR

NRM 1.11 [0.63-1.95] ☐ 1.28 [0.41-4.03] 

2.19 [1.29-3.72]  4.60 [2.60-8.14] ☐ 2.57 [1.52-4.33] ☐

2.77 [1.39-5.50]  5.14 [3.04-8.72] ☐ 2.81 [1.70-4.66] ☐

2.90 [1.81-4.64]  9.53 [4.48-20.29] ☐ 4.53 [2.30-8.92] ☐

4.41 [1.65-11.8]  3.29 [1.39-7.79]  2.46 [1.56-3.86] 

5.89 [1.90-18.2]  4.13 [1.19-14.3]  3.06 [1.85-5.05] 

-5.66 [2.80-11.4] ☐ 3.33 [2.18-5.11] 

1.16 [0.18-7.58]  -1.23 [0.77-1.97] ☐

3.39 [2.20-5.22]  2.05 [0.78-5.39] 

4.09 [2.53-6.62]  2.09 [1.33-3.29] ☐

4.72 [2.97-7.50]  3.23 [1.88-5.53] ☐

1.42 [0.90-2.25] ☐ 0.58 [0.22-1.52] ☐

Only fields pooled from ≥2 studies are reported; otherwise, fields are left blank. Colored boxes indicate degree of heterogeneity as defined by the I2 statistic: 0-24.9% = low (☐), 25-75% = moderate (), 75.1-100% = high ().39 Cells are filled only if two or more studies contribute to the analysis. HR: hazard ratio; CI: confidence interval; OS: overall survival; LFS: leukemia-free survival; CIR: cumulative incidence of relapse; NRM: non-relapse mortality; MFC: multi-parametric flow cytometry; PCR: polymerase chain reaction; MA: myeloablative.

870

haematologica | 2017; 102(5)


Pre-HCT MRD and Post-HCT Outcomes in AML

P = 0.000

P = 0.660

P = 0.072

Figure 3. Forest plot showing hazard ratio (effect size, ES) for leukemia-free survival with pooling of results for each minimal residual disease detection method. Columns indicate study size (N) and whether each study carries a high risk of bias (Bias Risk). Within groups, studies are listed by year of publication. CI: confidence interval; MFC: multi-parametric flow cytometry; PCR: polymerase chain reaction; MRD: minimal residual disease.

B

A P = 0.82

P = 0.92

Figure 4. Figure 4. Meta-regression analysis showing the effect of the ratio of the percentage of MRDpos patients with adverse cytogenetics to the percentage of neg MRD patients with adverse cytogenetics on log-hazard for leukemia-free survival. A flat line indicates no relationship, and this is shown for all studies (A) and after excluding studies with a high risk of bias (B). MRD: minimal residual disease; HR: hazard ratio.

haematologica | 2017; 102(5)

871


S.A. Buckley et al.

A

B

C

D

Figure 5. Funnel plot analysis for survival outcomes. Shown are (A) leukemia-free survival (LFS), (B) overall survival (OS), (C) cumulative incidence of relapse (CIR), (D) non-relapse mortality (NRM). HR: hazard ratio.

for being MRDpos by PCR-quantified WT1 transcript levels are very similar across several studies, indicating that this method yields highly reproducible results for pre-HCT risk stratification. Even in the smallest studies,17,19 in which there was no statistically significant relationship between MRD and LFS, observed HRs were consistent with the other, larger studies. One might wonder whether using more than one method to detect MRD might lead to more sensitive detection and stronger associations with relapse and survival, indicated by higher HRs. We found that studies using combination methods of MRD detection did not show stronger associations with survival outcomes over studies using either MFC- or WT1 PCR-based methods. That said, all four of these â&#x20AC;&#x2DC;combinationsâ&#x20AC;&#x2122; involved MFCbased detection, and the heterogeneity within the combination group may simply underscore the heterogeneity in MFC-based studies as a whole. Alternatively, MFC and WT1 PCR are both potentially highly sensitive tests, and using multiple modalities may not add much additional sensitivity in detection, or increases in assay sensitivity beyond current limits may not lead to appreciably stronger associations with survival outcomes. Although our studies highlight the importance of preHCT MRD, we were unable to account for inter-study differences in the selection of patients for HCT, which may impact post-HCT results. It is conceivable that different strategies in allocating patients to different post872

remission treatment strategies could affect our study results. Given the nature of our analysis, we were only able to test the effects of select covariates and only in an aggregate fashion. Similarly, we were not able to control for the considerable heterogeneity in transplant conditioning regimens, donor sources, graft characteristics, and immunosuppression, all of which could potentially influence relapse and death. Due to absent individual patient data, we were not able to assess whether higher levels of MRD were associated with higher risk of relapse. Regardless of these limitations, our results demonstrate a strong relationship between pre-HCT MRD status and post-HCT relapse and survival, but not NRM. Further studies are needed to determine how preHCT MRD status should guide therapeutic decisions, either through treatment intensification for MRD pos patients, or possibly de-intensification for patients who are found to be MRDneg by a reliable method. Acknowledgements SAB is supported by a fellowship training grant from the National Heart, Lung, and Blood Institute/National Institutes of Health (NHLBI/NIH: T32-HL007093). R.B.W. is a Leukemia & Lymphoma Society Scholar in Clinical Research. This work was supported in part by the Intramural Research Programs of the National Heart, Lung, and Blood Institute and the National Cancer Institute of the National Institutes of Health. haematologica | 2017; 102(5)


Pre-HCT MRD and Post-HCT Outcomes in AML

References 1. Bisel HF. Criteria for the evaluation of response to treatment in acute leukemia. Blood. 1956;11:676-677. 2. Cheson BD, Bennett JM, Kopecky KJ, et al. RevisedRevised recommendations of the International Working Group for diagnosis, standardization of response criteria, treatment outcomes, and reporting standards for therapeutic trials in acute myeloid leukemia. J Clin Oncol. 2003;21(24):4642-4649. 3. 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. 4. Ferrara F, Schiffer CA. Acute myeloid leukaemia in adults. Lancet. 2013;381 (9865):484-495. 5. Döhner H, Weisdorf DJ, Bloomfield CD. Acute myeloid leukemia. N Engl J Med. 2015;373(12):1136-1152. 6. Harrison WJ. The total cellularity of the bone marrow in man. J Clin Pathol. 1962; 15:254-259. 7. Coustan-Smith E, Campana D. Should evaluation for minimal residual disease be routine in acute myeloid leukemia? Curr Opin Hematol. 2013;20(2):86-92. 8. Hourigan CS, Karp JE. Minimal residual disease in acute myeloid leukaemia. Nat Rev Clin Oncol. 2013;10(8):460-471. 9. Grimwade D, Freeman SD. Defining minimal residual disease in acute myeloid leukemia: which platforms are ready for "prime time"? Blood. 2014;124(23):33453355. 10. Paietta E. Minimal residual disease in acute myeloid leukemia: coming of age. Hematology Am Soc Hematol Educ Program. 2012;2012:35-42. 11. Hokland P, Ommen HB, Mulé MP, Hourigan CS. Advancing the minimal residual disease concept in acute myeloid leukemia. Semin Hematol. 2015;52(3):184-192. 12. Ommen HB. Monitoring minimal residual disease in acute myeloid leukaemia: a review of the current evolving strategies. Ther Adv Hematol. 2016;7(1):3-16. 13. Hayden JA, van der Windt DA, Cartwright JL, Cote P, Bombardier C. Assessing bias in studies of prognostic factors. Ann Intern Med. 2013;158(4):280-286. 14. Broglio KR, Quintana M, Foster M, et al. Association of pathologic complete response to neoadjuvant therapy in HER2-positive breast cancer with long-term outcomes: a meta-analysis. JAMA Oncol. 2016;2(6):751760. 15. Tierney JF, Stewart LA, Ghersi D, Burdett S, Sydes MR. Practical methods for incorporating summary time-to-event data into metaanalysis. Trials. 2007;8:16. 16. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21(11):1539-1558. 17. Candoni A, Toffoletti E, Gallina R, et al. Monitoring of minimal residual disease by quantitative WT1 gene expression following

haematologica | 2017; 102(5)

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

reduced intensity conditioning allogeneic stem cell transplantation in acute myeloid leukemia. Clin Transplant. 2011;25(2):308316. Jacobsohn DA, Tse WT, Chaleff S, et al. High WT1 gene expression before haematopoietic stem cell transplant in children with acute myeloid leukaemia predicts poor event-free survival. Br J Haematol. 2009;146(6):669-674. Rossi G, Carella AM, Minervini MM, et al. Optimal time-points for minimal residual disease monitoring change on the basis of the method used in patients with acute myeloid leukemia who underwent allogeneic stem cell transplantation: a comparison between multiparameter flow cytometry and Wilms' tumor 1 expression. Leuk Res. 2015;39(2):138-143. Olszewski M, Chou PM, Huang W, Tallman S, Kletzel M. Correlation of minimal residual disease by assessing Wilms tumor gene expression and engraftment by variable number of tandem repeats in children with leukemia posthematopoietic stem cell transplantation. Pediatr Dev Pathol. 2006;9(3): 203-209. Kanakry CG, Tsai HL, Bolanos-Meade J, et al. Single-agent GVHD prophylaxis with posttransplantation cyclophosphamide after myeloablative, HLA-matched BMT for AML, ALL, and MDS. Blood. 2014;124 (25):3817-3827. Anthias C, Dignan FL, Morilla R, et al. Pretransplant MRD predicts outcome following reduced-intensity and myeloablative allogeneic hemopoietic SCT in AML. Bone Marrow Transplant. 2014;49(5):679-683. Buccisano F, Maurillo L, Spagnoli A, et al. Cytogenetic and molecular diagnostic characterization combined to postconsolidation minimal residual disease assessment by flow cytometry improves risk stratification in adult acute myeloid leukemia. Blood. 2010;116(13):2295-2303. Araki D, Wood BL, Othus M, et al. Allogeneic hematopoietic cell transplantation for acute myeloid leukemia: time to move toward a minimal residual diseasebased definition of complete remission? J Clin Oncol. 2016;34(4):329-336. Zheng C, Zhu X, Tang B, et al. The impact of pre-transplant minimal residual disease on outcome of intensified myeloablative cord blood transplant for acute myeloid leukemia in first or second complete remission. Leuk Lymphoma. 2016;57(6):1398-1405. Bleyzac N, Cuzzubbo D, Renard C, et al. Improved outcome of children transplanted for high-risk leukemia by using a new strategy of cyclosporine-based GVHD prophylaxis. Bone Marrow Transplant. 2016; 51(5):698-704. Ustun C, Courville EL, DeFor T, et al. Myeloablative, but not reduced-intensity, conditioning overcomes the negative dffect of flow-cytometric evidence of leukemia in acute myeloid leukemia. Biol Blood Marrow Transplant. 2016;22(4):669-675. Tian H, Chen GH, Xu Y, et al. Impact of pretransplant disease burden on the outcome of allogeneic hematopoietic stem cell trans-

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

plant in refractory and relapsed acute myeloid leukemia: a single-center study. Leuk Lymphoma. 2015;56(5):1353-1361. Walter RB, Gyurkocza B, Storer BE, et al. Comparison of minimal residual disease as outcome predictor for AML patients in first complete remission undergoing myeloablative or nonmyeloablative allogeneic hematopoietic cell transplantation. Leukemia. 2015;29(1):137-144. Woehlecke C, Wittig S, Arndt C, Gruhn B. Prognostic impact of WT1 expression prior to hematopoietic stem cell transplantation in children with malignant hematological diseases. J Cancer Res Clin Oncol. 2015;141(3): 523-529. Goswami M, McGowan KS, Lu K, et al. A multigene array for measurable residual disease detection in AML patients undergoing SCT. Bone Marrow Transplant. 2015 ;50(5):642-651. Valkova V, Polak J, Markova M, et al. Minimal residual disease detectable by quantitative assessment of WT1 gene before allogeneic stem cell transplantation in patients in first remission of acute myeloid leukemia has an impact on their future prognosis. Clin Transplant. 2013;27 (1):E21-29. Bastos-Oreiro M, Perez-Corral A, MartinezLaperche C, et al. Prognostic impact of minimal residual disease analysis by flow cytometry in patients with acute myeloid leukemia before and after allogeneic hemopoietic stem cell transplantation. Eur J Haematol. 2014;93(3):239-246. Wang Y, Liu DH, Liu KY, et al. Impact of pretransplantation risk factors on post transplantation outcome of patients with acute myeloid leukemia in remission after haploidentical hematopoietic stem cell transplantation. Biol Blood Marrow Transplantat. 2013;19(2):283-290. Grubovikj RM, Alavi A, Koppel A, Territo M, Schiller GJ. Minimal residual disease as a predictive factor for relapse after allogeneic hematopoietic stem cell transplant in adult patients with acute myeloid leukemia in first and second complete remission. Cancers (Basel). 2012;4(2):601-617. Leung W, Pui CH, Coustan-Smith E, et al. Detectable minimal residual disease before hematopoietic cell transplantation is prognostic but does not preclude cure for children with very-high-risk leukemia. Blood. 2012;120(2):468-472. Laane E, Derolf AR, Bjorklund E, et al. The effect of allogeneic stem cell transplantation on outcome in younger acute myeloid leukemia patients with minimal residual disease detected by flow cytometry at the end of post-remission chemotherapy. Haematologica. 2006;91(6):833-836. van Dongen JJ, van der Velden VH, Bruggemann M, Orfao A. Minimal residual disease diagnostics in acute lymphoblastic leukemia: need for sensitive, fast, and standardized technologies. Blood. 2015;125(26): 3996-4009. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in metaanalyses. BMJ. 2003;327(7414):557-560.

873


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Chronic Lymphocytic Leukemia

Ferrata Storti Foundation

Lymphocyte activation gene 3: a novel therapeutic target in chronic lymphocytic leukemia Mika Shapiro,1* Yair Herishanu,1,2* Ben-Zion-Katz,1,2 Nili Dezorella,1,2 Clare Sun,3 Sigi Kay,1 Aaron Polliack,4 Irit Avivi,1,2 Adrian Wiestner3 and Chava Perry1

Department of Hematology, Tel Aviv Sourasky Medical Center, Israel; 2Sackler Faculty of Medicine, Tel Aviv University, Israel; 3Hematology Branch, National Heart, Lung, and Blood Institute, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA and 4Hadassah University Hospital, Jerusalem, Israel 1

Haematologica 2017 Volume 102(5):874-882

*MS and YH contributed equally to this work

ABSTRACT

A

Correspondence: yairh@tlvmc.gov.il

Received: May 8, 2016. Accepted: January 25, 2017. Pre-published: February 2, 2017. doi:10.3324/haematol.2016.148965 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/874

novel therapeutic approach in cancer, attempting to stimulate host anti-tumor immunity, involves blocking of immune checkpoints. Lymphocyte activation gene 3 (LAG3) is an immune checkpoint receptor expressed on activated/exhausted T cells. When engaged by the major histocompatibility complex (MHC) class II molecules, LAG3 negatively regulates T-cell function, thereby contributing to tumor escape. Intriguingly, a soluble LAG3 variant activates both immune and malignant MHC class II-presenting cells. In the study herein, we examined the role of LAG3 in the pathogenesis of chronic lymphocytic leukemia, an MHC class II-presenting malignancy, and show that chronic lymphocytic leukemia cells express and secrete LAG3. High levels of surface and soluble LAG3 were associated with the unmutated immunoglobulin variable heavy chain leukemic subtype and a shorter median time from diagnosis to first treatment. Utilizing a mechanism mediated through MHC class II engagement, recombinant soluble LAG3-Ig fusion protein, LAG3-Fc, activated chronic lymphocytic leukemia cells, induced anti-apoptotic pathways and protected the cells from spontaneous apoptosis, effects mediated by SYK, BTK and MAPK signaling. Moreover, LAG3 blocking antibody enhanced in vitro T-cell activation. Our data suggest that soluble LAG3 promotes leukemic cell activation and anti-apoptotic effects through its engagement with MHC class II. Furthermore, MHC class II-presenting chronic lymphocytic leukemia cells may affect LAG3-presenting T cells and impose immune exhaustion on their microenvironment; hence, blocking LAG3-MHC class II interactions is a potential therapeutic target in chronic lymphocytic leukemia.

Introduction Š2017 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.

874

Chronic lymphocytic leukemia (CLL) is a lymphoproliferative disorder (LPD) characterized by the progressive accumulation of small CD5+ mature-looking B cells in the peripheral blood, bone marrow (BM) and secondary lymphoid organs.1 Despite recent advances in understanding the pathophysiology of CLL, it is still mostly regarded as an incurable disorder, despite the long-term remissions observed in some of the patients treated with the fludarabine-cyclophosfamide-rituximab (FCR) regimen, or patients who underwent allogeneic stem cell transplantation.2,3 There are two main subgroups of CLL based on the presence or absence of somatic mutations in the immunoglobulin heavy chain variable domain (IGHV).1 The presence of a mutated IGHV (M-IGHV) identifies a leukemic subtype that has a stable or slowly progressive course, while the expression of an unmutated IGHV (UM-IGHV) gene is associated with a more aggressive disease and an inferior rate of survival.4-6 The inability of the immune system to eradicate malignancy is one of the fundahaematologica | 2017; 102(5)


LAG3: a novel therapeutic target in CLL

mental hallmarks of cancer. Due to chronic antigen stimulation induced by cancer cells, effector T cells may gradually lose their effector activities, a process termed “exhaustion".7 In this respect, the expression of immune checkpoint receptors is regarded as a hallmark of "exhaustion". Cytotoxic Tlymphocyte-associated antigen 4 (CTLA4) and programmed cell death protein 1 (PD1) are particularly important immune checkpoint receptors.8-10 The CD4 homolog lymphocyte activation gene 3 (LAG3;CD223) is an immune checkpoint receptor. Among others, LAG3 is expressed on exhausted T cells as well as on tumor-infiltrating lymphocytes (TILs).11,12 LAG3 binds to MHC Class II (MHCII) molecules on antigen presenting cells (APC), but with higher affinity than CD4, an interaction that negatively regulates CD3-T-cell receptor (TCR) complex signaling, thus affecting T-cell proliferation, function and homeostasis.11 In humans, a 52kDa soluble LAG3 protein variant (LAG3V3, sLAG3) is formed by an alternatively spliced RNA13,14 (Online Supplementary Figure S1). sLAG3 has also been shown to bind MHCII, yet this variant was reported to activate APCs and enhance tumor-specific cytotoxic T cells.15 However, in melanoma cells that express MHCII, the interaction with sLAG3 activates MAPK/ERK and PI3K/AKT pathways, thus contributing to the resistance of the malignant cells to apoptosis.15 Interestingly, LAG3 expression was recently suggested as a prognostic marker in patients with CLL, as gene expression profiling of CLL cells detected increased LAG3 expression levels that were in correlation with UM-IGHV and with reduced treatment-free survival.16 We hypothesized that LAG3-MHCII interaction may play an important role in the pathogenesis of CLL and contribute to leukemic cells resistance to apoptosis and their ability to evade anti-cancer immunity. For that reason, we analyzed the expression of LAG3 and its soluble variant, sLAG3, in patients with CLL, and explored the effects of LAG3-MHCII interaction on CLL cells activation, survival and downstream signaling pathways that mediate these effects.

Stimulation: CLL cells were incubated with FcR blocking reagent, before being stimulated by either a recombinant soluble human LAG3-Ig fusion protein (LAG3-Fc) (1mg/ml) or control Ig-Fc (1mg/ml). Analysis of cell viability and apoptosis; After incubation with LAG3Fc or Ig-Fc, cells were either harvested for Western blot assays or stained with the Annexin V/propidium Iodide MEBCYTO® Apoptosis Kit (MBL, Nagoya, Japan), according to the manufacturer’s instructions. For inhibition assays, CLL cells were pre-incubated with wortmannin (50nM), PD98059 (100mM), idelalisib (10mM), R406 (100mM) or ibrutinib (0.5mM) for 1 hour prior to stimulation, then cultured for 48 hours and analyzed by flow cytometry. LAG3-Fc binding: CLL cells were incubated for 15 min with either LAG3-Fc or Ig-Fc and stained for CD19 and anti-human IgG (Fc γ-specific) for flow cytometry analysis. For Inhibition assays: LAG3-Fc (1mg/ml) was incubated with antiLAG3 (aimed at the MHCII molecules binding site (clone 17B4), 10µg/ml) for 30 min before being added to the cultured CLL cells for 1 hour of incubation. Subsequently cells were washed and incubated for 72 hours.

Methods

Quantitative PCR

Patients and samples After obtaining informed consent in accordance with the Declaration of Helsinki and approval from the institutional ethics committee, peripheral blood samples were collected from CLL patients17 and healthy controls. Lymph nodes and spleen samples were also collected from CLL patients. Handling protocol is available in the Online Supplementary Material and Methods.

Reverse transcription was performed using oligo(dT) priming and Verso cDNA kit (Thermo Fisher Scientific, ABgene, Epsom, UK) according to the manufacturer’s instructions.

IGHV gene analysis Analysis of IGHV gene status was performed as described in Wiestner et al.18 and detailed in the Online Supplementary Material and Methods.

Cell stimulation, apoptosis and LAG3-Fc binding assay

Activation of T cells and blocking antibody treatment Cells were incubated for 48 hours in the presence of i) antiLAG3 (17B4) (20mg/mL), ii) anti-PD-1 (J116) with anti-PD-L1 (M1H1) (10 mg/mL each), iii) the combined three antibodies, or iv) IgG1 isotype control, then activated by CD3/CD28 Dynabeads for 6 hours, followed by flow cytometry analysis.

Western blotting and Flow Cytometry These are detailed in the Online Supplementary Material and Methods.

Real-time polymerase chain reaction (PCR) was performed using LightCycler® 480 SYBR Green I Master and analyzed on a LightCycler 480 II (Roche, Basel, Switzerland). Primers are presented in the Online Supplementary Material and Methods.

Enzyme-linked immunosorbent assay (ELISA)

Reagents and antibodies

sLAG3 serum concentrations were determined using RayBio Human LAG3 Elisa kit (RayBiotech, GA, USA) following the manufacturer's instructions, using SpectraMax M2 ELISA reader (Molecular Devices, CA, USA).

These are detailed in the Online Supplementary Material and Methods.

Statistical analysis

Enrichment of CLL cells Peripheral blood mononuclear cells were magnetically labeled either using CD19 microbeads for positive selection or by B-CLL Cell Isolation kit for negative selection and then separated on a magnetic cell separation column (MACS), all from Miltenyi Biotec Inc., Auburn, CA, USA.

We used unpaired and paired t-tests or one-way ANOVA to assess differences in the means of two groups or three groups, respectively. A P-value <0.05 was considered significant.

Results LAG3 expression in CLL cells and disease course

RNA extraction and cDNA synthesis RNA was extracted using RNeasy kit (Qiagen, CA, USA). haematologica | 2017; 102(5)

Based on previously reported gene expression profiles that have shown overexpression of LAG3 in UM-IGHV 875


M. Shapiro et al. A

B

C

D

E

F

G

H

I

Figure 1. Full-length LAG3 expression in CLL patients. (A-C) Full-length LAG3 mRNA levels in CD19+ selected normal B cells and CLL cells quantitated by qPCR, and normalized to GAPDH. (A) LAG3 mRNA expression in normal B cells (n=7) vs. CLL cells (n=28). (B) LAG3 mRNA expression in M-IGHV cells (n=15) compared to UMIGHV CLL cells (n=13). (C) Kaplan-Meier analysis of time from diagnosis to first treatment in CLL patients with “low” and “high” LAG3 mRNA levels (n=28), using the median value as cutoff level. (D-E) Detection of LAG3 protein levels by Western blot assay in CD19+ purified CLL cells, M-IGHV cells compared to UM-IGHV CLL cells. (D) Representative blot analyzing CLL cells from 8 patients, 4 M-IGHV and 4 UM-IGHV CLL (patients’ characteristics are presented in the Online Supplementary Table S1, according to the designated numbers). Actin was used to verify equal loading. (E) Statistic analysis of CLL cells from 16 patients with CLL, 8 M-IGHV compared to 8 UM-IGHV CLL cells. (F-I) LAG3 expression in peripheral blood samples, evaluated by flow cytometry. (F) Representative dot plots showing surface and intracellular expression of LAG3 in CLL cells. (G) Summary of surface LAG3 mean fluorescence intensity (MFI) normalized to isotype control, in normal B cells (n=8) compared to CLL cells (n=22), isolated from peripheral blood of healthy controls and patients with CLL, respectively. (H) Surface LAG3 MFI normalized to isotype control (MFIR) as detected in peripheral blood of M-IGHV (n=11) and UM-IGHV (n=11) CLL cells. (I) Summary of surface LAG3 MFIR in CLL, CD4+ and CD8+ cells isolated from peripheral blood of patients with CLL (n=22). CLL: chronic lymphocytic leukemia; LAG3: lymphocyte activation gene 3; ns: not significant.

CLL,16 we first evaluated the expression of full-length LAG3 messenger RNA (mRNA) in CLL cells from patients with M-IGHV and UM-IGHV CLL as well as in B cells from normal controls. Patient characteristics are presented in the Online Supplementary Table S1. Peripheral blood CLL and normal B cells were purified using positive selection to obtain B cell purity (>96%) and LAG3 expression was analyzed by RT-PCR. Full-length LAG3 mRNA expression levels were increased in CLL cells compared to normal B cells (P=0.0028; Figure 1A). When evaluated among patients with CLL, LAG3 mRNA levels were significantly increased in UM-IGHV CLL cells compared to cells with the MIGHV gene (P=0.026; Figure 1B). Moreover, patients with higher levels of full-length LAG3 mRNA (defined as being above the median LAG3 mRNA level) had a shorter median time from diagnosis to first treatment (Figure 1C). At the protein level, LAG3 was detected by Western blot in CD19+ purified CLL cells in all analyzed patients. However, no differences were detected in LAG3 levels between MIGHV and UM-IGHV CLL cells (Figure 1D,E). Using flow 876

cytometry, we evaluated LAG3 cellular localization in CLL cells. LAG3 was detected at very low levels on the surface of CLL cells, and only a small fraction of the cells expressed substantial levels of surface LAG3 (Figure 1F). Most CLL cells, however, expressed high levels of intracellular LAG3 (6.4±5.4% expressed surface LAG3 while 60.9±24.8% expressed intracellular LAG3, Figure 1F). The intensity of surface LAG3 expression was further evaluated in peripheral blood lymphocytes; mean fluorescence intensity (MFI) of surface LAG3 was increased in CLL cells compared to normal B cells (P<0.001; Figure 1G). Surface LAG3 MFI was also increased in UM-IGHV compared to M-IGHV CLL cells (9.2±7.1 vs. 3.9±1.9, respectively; P=0.026; Figure 1H). In patients with CLL, surface LAG3 MFI was elevated in CLL cells compared to CD4+ and CD8+ lymphocytes (6.55±5.8 vs. 2.6±2.0 (P=0.005), vs. 2.3±1.7 (P=0.002), respectively, Figure 1I). No statistically significant differences were detected in the intensity of intracellular LAG3 expression between CLL and normal B cells (Online Supplementary Figure S2A). haematologica | 2017; 102(5)


LAG3: a novel therapeutic target in CLL

A

C

B

D

E

Figure 2. Soluble (s)LAG3 is associated with UM-IGHV status and progressive disease. (A-B) LAG3V3 mRNA levels encoding sLAG3 were quantitated in CD19+ selected normal B and CLL cells by qPCR, and normalized to GAPDH. (A) LAG3V3 mRNA levels in normal B cells (n=7), M-IGHV (n=12) and UM-IGHV (n=11) CLL cells. (B) Kaplan-Meier analysis of time from diagnosis to first treatment in patients with CLL (n=23), expressing “low” and “high” LAG3V3 mRNA levels, using the median value as cutoff level. (C-D) Serum sLAG3 levels in CLL patients and healthy individuals, determined by ELISA. (C) Serum sLAG3 levels in healthy controls (n=8) and patients with M-IGHV (n=16) and UM-IGHV (n=17) CLL. (D) Comparison between serum sLAG3 levels in CLL patients with either stable (n=17) or progressive (n=18) disease. (E) Measurement of sLAG3 levels in the medium of cultured CLL cells; negatively selected CLL cells were cultured and medium sLAG3 levels in the culture medium were determined by ELISA after 24, 48 and 72 hours (n=5). mRNA (RQ): messenger RNA (relative quantification).

Increased expression of sLAG3 is associated with both UM-IGHV status and shorter time to treatment The levels of LAG3V3, the soluble, shorter LAG3 isoform, encoded by alternatively spliced RNA, were determined in patients with CLL and in normal controls. In this analysis, IGHV mutational status data were available for 32 patients. Thirteen out of 17 patients with UM-IGHV, but only 3 out of 15 with M-IGHV, had progressive disease (Online Supplementary Table S1). Increased levels of LAG3V3 mRNA were evident in UM-IGHV CLL cells compared to both M-IGHV CLL cells (P=0.039) and normal B cells (P=0.03; Figure 2A). Elevated levels of LAG3V3 mRNA (defined as levels higher than the median value) were significantly associated with a shorter time to first treatment (Figure 2B). sLAG3 protein levels were determined in the serum of patients with CLL and healthy controls, and were found to be higher in patients with UM-IGHV CLL compared to patients with the M-IGHV gene and healthy controls (Figure 2C). The median serum sLAG3 levels were 2.5ng/ml (0.11-6.67), 0.2ng/ml (0.03-13.0) and 0.15 ng/ml (0.09-1.79) in patients with UM-IGHV, M-IGHV CLL and healthy controls, respectively, (Figure 2C). High sLAG3 levels were also detected in patients whose disease progressed compared to patients with stable CLL (median levhaematologica | 2017; 102(5)

els of 2.9 ng/ml (0.15-13.0) and 0.06 ng/ml (0.03-1.48), respectively, P<0.001; Figure 2D). Next, we explored whether CLL cells secrete sLAG3. sLAG3 levels progressively increased in the culture medium of negatively selected CLL cells, and the highest levels were detected at the 72 hour time point (Figure 2E). Overall, our data suggest that CLL cells express and secrete sLAG3.

LAG3 binds MHC class II molecules on CLL cells As CLL cells express MHCII molecules on their cell surface,19 we further determined the specific binding of LAG3 to CLL cells. As shown in Figure 3A,B, LAG3-Fc (a fusion protein that consists of an extracellular portion of LAG3, fused to the Fc fraction of human IgG1, that binds to MHCII) was found to bind a large proportion of CD19+ CLL cells, as opposed to Ig-Fc control. MFI value, representing LAG3-Fc binding to CLL cells, was 226 as compared to 51 in cells incubated with the Ig-Fc control. The addition of anti-LAG3 antibody, directed to the extra loop of the Ig-like domain 1 of LAG3 that binds MHCII molecules,20,21 completely abolished soluble LAG3 ligation to CLL cells. Therefore, our results suggest that sLAG3 binds to CLL cells through interaction with MHCII molecules. 877


M. Shapiro et al.

sLAG3 activates CLL cells and exerts an anti-apoptotic effect We further studied the biological effects of sLAG3 on CLL cells. For this purpose, peripheral blood CLL cells were incubated with LAG3-Fc, and its effect on CLL cell activation was studied by evaluating cell surface CD86 expression. Expression of the costimulatory B7 molecules, CD80 and CD86, is low in CLL cells, but it is upregulated upon cell activation.22 Activation of B cells via MHCII engagement was reported to induce B7 costimulatory molecules.23 As LAG3 interacts with CLL cells via MHCII, we used CD86 expression as a marker of LAG3-induced CLL cell activation.

A

After 24 hours incubation with LAG3-Fc, the expression of CD86+ CLL cells increased significantly compared to control (Figure 3C). CD86 upregulation in response to sLAG3 activation was completely blocked by pre-incubation with anti-LAG3 antibody (Figure 3C). Incubation with LAG3-Fc also induced a mild, though statistically significant, increase in the expression of another marker of CLL cell activation, CD69 (Online Supplementary Figure S2B). We next investigated the effect of LAG3-Fc on the PI3K/AKT and MAPK/ERK pathways, which have been reported to be activated following MHCII engagement.15 Stimulation of CLL cells with LAG3-Fc induced AKT and

B

C

D

Figure 3. Soluble (s)LAG3 binds and activates MHC class II molecules on CLL cells. (A-B) Detection of sLAG binding to CLL cells: peripheral blood CLL cells were incubated for 15 min with either LAG3-Fc, LAG3-Fc after pre-incubation with anti-LAG3 blocking antibody, or Ig-Fc that served as control. LAG3 binding to CLL cells was detected by flow cytometry, using a fluorophore-conjugated secondary antibody (anti-human Fc). (A) Representative dot plot analysis showing that LAG3-Fc binding to CLL cells (middle box) was completely abolished by anti-LAG3 blocking antibody (right box). (B) The mean fluorescence intensity (MFI) of LAG3-binding CLL cells after incubation with LAG3-Fc (middle bar) decreases to control levels after pre-incubation with anti-LAG3 antibodies (right bar); cumulative results of 11 experiments. (C) Measurement of CD86 surface expression on CLL cells in response to sLAG3 activation: CLL cells were incubated with either LAG3-Fc or Ig-Fc that served as control for 24 hours, and surface CD86 expression was analyzed on CD5+/CD23+ gated CLL cells by flow cytometry. C-Left: representative dot plots showing an increase in surface CD86 expression on CLL cells in the presence of LAG3-Fc. C-middle: surface CD86 MFI levels on LAG3-Fc-activated CLL cells are presented as fold change increase over control (Ig-Fc) levels, n=11. C-Right: comparison of CD86 MFI expression on CLL cells incubated with either Ig-Fc (control, left bar), LAG3-Fc (middle bar) or LAG3-Fc pre-incubated with anti-LAG3 blocking antibody [right bar, (n=7)]. (D) Changes in the mean fluorescence levels (normalized to baseline) of pERK and pAKT in CLL cells, measured by flow cytometry at 0, 5, 15, 45, 60 and 120 min after activation by LAG3-Fc (n=5). CLL: chronic lymphocytic leukemia; LAG3: lymphocyte activation gene 3.

878

haematologica | 2017; 102(5)


LAG3: a novel therapeutic target in CLL

ERK1/2 phosphorylation, an effect that peaked 15 min after activation (Figure 3D). To explore possible effects of soluble LAG3 on CLL cell survival, CLL cells were incubated with LAG3-Fc and their viability was evaluated after 24, 48, 72 and 96 hours. The percentage of live cells increased significantly after incuba-

A

tion with LAG3-Fc compared to unstimulated CLL cells. Maximal effect was detected after 48 and 72 hours incubation (Figure 4A-C and Online Supplementary Figure S2C). The effect of LAG3-Fc on CLL cell survival was abolished by PD98059 (MEK1/2 inhibitor), ibrutinib (Brutonâ&#x20AC;&#x2122;s tyrosine kinase inhibitor) and R406 (SYK inhibitor and the

B

D

C

E

F

G

H

Figure 4. Soluble (s)LAG3 protects CLL cells from spontaneous apoptosis. (A-F) Peripheral blood CLL cells were incubated for 48 or 72 hours with either Ig-Fc (control), LAG3-Fc, or with LAG3-Fc pre-incubated with anti-LAG3 blocking antibody aimed at the MHCII molecules binding site. Cell viability was determined by flow cytometry, using an Annexin V/PI apoptosis detection kit. The levels of anti-apoptotic proteins and cleaved PARP were determined by Western blot analysis and quantified. (A) Representative dot plots showing the percentage of apoptotic cells in the presence of Ig-Fc (left), LAG3-Fc (middle) and LAG3-Fc with anti-LAG3 blocking antibody (right). (B-C) Percentage of live CLL cells in the presence of either Ig-Fc (control) or LAG3-Fc, as seen in 10 independent experiments after 48 (B) and 72 hours (C) incubation. (D) The percentage of live CLL cells in the presence of Ig-Fc control [marked as (-)], or LAG3-Fc [marked as (+)], with or without 1 hour pre-incubation with PD98059, wortmannin (left graph), ibrutinib, R406 or idelalisib (right graph). After being cultured for 48 hours, cell viability was determined by flow cytometry, using an Annexin V/PI apoptosis detection kit (n=7). (E) The percentage of live CLL cells cultured with either Ig-Fc (left bar), LAG3-Fc (middle bar), or LAG3-Fc pre-incubated with anti-LAG3 (right bar), for 1 hour, washed and incubated for an additional 72 hours before being analyzed by flow cytometry (n=6). (F) Representative Western blot analysis showing the levels of cleaved PARP, MCL-1, Bcl-XL and Bcl-2 in CLL cells after 72 hours incubation with Ig-Fc as a control [marked (-)] or LAG3-Fc [marked (+)]. Actin was used to verify equal loading. (G) Cumulative results from 8 independent experiments, performed as described in Figure 4F. Shown are quantified levels of cleaved PARP, MCL-1, Bcl-XL and Bcl-2 in LAG3-Fc activated CLL cells, normalized to control (incubation with Ig-Fc). (H) The percentage of apoptotic CLL cells increased following LAG3 blockade. The levels of apoptotic CLL cells were determined after 72 hours incubation with anti-LAG3 blocking antibody and normalized to control levels in 15 independent experiments. LAG3: lymphocyte activation gene 3; CLL: chronic lymphocytic leukemia; ns: not significant.

haematologica | 2017; 102(5)

879


M. Shapiro et al. A

B P=0.026

C

P=0.001

P=0.006

P=0.01

Activation Anti-PD1+PD-L1 Anti-LAG3

P=0.047

Activation Anti-PD1+PD-L1 Anti-LAG3

Figure 5. Increased surface LAG3 expression on CD8+ tumor infiltrating lymphocytes and blocking LAG3 enhance in vitro T-cell activation. (A) The expression of surface LAG3 on CD4+ and CD8+ cells (analyzed by flow cytometry), in 10 paired samples of peripheral blood (PB) and secondary lymphoid tissues [sec. LN; lymph node (n=3) or spleen cells (n=7)] from patients with CLL. (B) A representative dot plot analysis, showing co-expression of surface LAG3 and PD1 on lymph node-derived CD8+ lymphocytes isolated from a patient with CLL (n=5). (C) CD69 expression on activated T cells, with or without LAG3 and PD1 blockade. Cells from 7 patients with CLL were incubated for 48 hours in the presence or absence of the indicated blocking antibodies or with IgG1 isotype control. T cells were then activated by CD3/CD28 beads for 6 hours and CD69 expression was analyzed in CD4+ (left) and CD8+ cells (right) by flow cytometry. LAG3: lymphocyte activation gene 3; MFI: mean fluorescence intensity; ns: not significant.

active metabolite of fostamatinib) (Figure 4D) as well as by the anti-LAG3 blocking antibody (Figure 4A and 4E). However, LAG3-Fc anti-apoptotic effect was affected neither by pre-incubating with wortmannin (phosphatidylinositol 3-kinase (PI3K) inhibitor) nor by idelalisib (a specific PI3Kδ inhibitor), Figure 4D. Incubation with LAG3-Fc was also associated with a prominent decrease in cleaved PARP and a robust increase in Mcl-1 levels. (Figure 4F,G and Online Supplementary Figure S2D,E). The levels of Bcl-XL and Bcl-2 increased slightly and inconsistently after 48 and 72 hour incubation with LAG3-Fc (Figure 4F,G and Online Supplementary Figure S2D,E). Interestingly, incubating CLL cells with anti-LAG3 antibody resulted in increased levels of apoptotic cells compared to control (Figure 4H), suggesting that blocking sLAG3-MHCII interaction prevented autocrine effects of sLAG3, excreted by the cultured CLL cells.

T cells in the CLL microenvironment express both LAG3 and PD1 We also studied the expression of LAG3 on tumor infiltrating T lymphocytes in secondary lymphoid tissues (lymph nodes and spleens) obtained from patients with CLL, and compared it to LAG3 expression on concurrently collected circulating peripheral blood T cells. There was no statistically significant difference between LAG3 880

expression on CD4+ T cells in peripheral blood and secondary lymphoid organs. However, we found that the percentage of CD8+ T cells expressing LAG3 was significantly higher in secondary lymphoid tissues compared to paired peripheral blood CD8+ lymphocytes isolated from the same patient (5.7±5.4% vs. 1.2±2.2% of CD8+ T cells in secondary lymphoid tissues and peripheral blood, respectively, P=0.026; Figure 5A). CD8+ cells, obtained from secondary lymphoid organs of patients with CLL, were analyzed further, and PD1 expression on these cells was evaluated. We found that LAG3 expression was confined to PD1 expressing CD8+ lymphocytes (Figure 5B). Next, we evaluated the possible combined effects of LAG3 and PD1 blockade on T cell activation in patients with CLL. In order to do so, we determined the expression of CD69 (as a marker of T cell activation) on T cells from peripheral blood of CLL patients, that were activated in vitro (using anti-CD3/CD28 beads), after pre-incubation with either anti-LAG3 antibody, anti-PD1 combined with anti-PD-L1 antibodies (to fully block the PD-1 pathway), or both (Figure 5C). We found that T-cell activation was increased in the presence of anti-LAG3 antibody but was unaffected by PD-1 pathway blockade. Combining antiLAG3 with anti-PD1/anti-PD-L1 antibodies abolished the positive effect induced by anti-LAG3 antibodies on both CD4+ and CD8+ T-cell activation (Figure 5C). haematologica | 2017; 102(5)


LAG3: a novel therapeutic target in CLL

Discussion In the study herein, we examined the role of the immune checkpoint receptor LAG3 and the interactions with its ligand, MHCII, in the pathogenesis of CLL. We showed that CLL cells express LAG3 and excrete its soluble isoform, LAG3V3. sLAG3 activated CLL cells and prevented them from undergoing spontaneous apoptosis, both effects mediated by its binding to MHCII molecules present on their surface. LAG3 mRNA was detected in CLL cells at higher levels than in normal B cells. Full-length LAG3 mRNA levels were also significantly higher in patients with the prognostically unfavorable UM-IGHV compared to those with the M-IGHV gene. The latter observation is similar to gene expression profile results reported earlier by Kostaskova et al.16 LAG3 was detected intracellularly in CLL cells, while only a small proportion of cells presented surface LAG3. However, in cells expressing surface LAG3, the levels were significantly higher in UM-IGHV cells, perhaps implying a role for LAG3 in the unfavorable prognosis of patients with UM-IGHV CLL. mRNA LAG3V3 and serum levels of sLAG3, the short, soluble LAG3 isoform, were elevated in the UM-IGHV subgroup of patients compared to patients with the M-IGHV gene. Increased levels of full-length LAG3 mRNA, LAG3V3 mRNA and serum sLAG3 were all associated with a more aggressive clinical course and a shorter median time to first treatment. Thus, we can conclude that higher levels of LAG3 are associated with poor prognostic features and an aggressive course of disease in patients with CLL. Previous studies have reported that increased levels of sLAG3 were associated with a favorable outcome in patients with breast cancer.24 In these cases, sLAG3 binds MHCII molecules on APCs, increases the capacity of MHCII positive immune cells to induce Tcell response and enhances tumor-specific cytotoxic T cells.15 However, in malignant melanoma cells that express MHCII, sLAG3 binding appears to upregulate anti-apoptotic pathways.15 Similarly, we found that sLAG3 binds to MHCII on CLL cells, and induces CLL cell activation and stimulation of the PI3K/AKT and MAPK/ERK pathways as well as promoting anti-apoptotic effects. Incubating CLL cells with sLAG3 resulted in an increase in the number of live cells, an effect abrogated through the inhibition of BTK, SYK and LAG3-MHCII interaction, but not through the inhibition of the PI3K pathway. Our findings are compatible with previous reports showing that ligation of MHCII generates downstream signals which is mediated through SYK, AKT and ERK.15,25 The activation of CLL cells via sLAG3 also resulted in a decreased degradation of PARP

References 1. Caligaris-Cappio F, Hamblin TJ. B-cell chronic lymphocytic leukemia: a bird of a different feather. J Clin Oncol. 1999; 17(1):399-408. 2. Moreno C, Villamor N, Colomer D, et al. Allogeneic stem-cell transplantation may overcome the adverse prognosis of unmutated VH gene in patients with chronic lymphocytic leukemia. J Clin Oncol. 2005 20;23(15):3433-3438.

haematologica | 2017; 102(5)

and an increased expression of anti-apoptotic proteins, which was substantial for Mcl-1 and more subtle for BclXL and Bcl-2. Constitutive expression of anti-apoptotic proteins and resistance to apoptosis are major hallmarks of CLL. Our data suggest a role for LAG3 in the pathogenesis of CLL, not only as an immune modulator but also in the regulation of anti-apoptotic pathways in CLL cells. We show that in CLL patients, LAG3 is expressed both by tumor cells as well as in the tumor microenvironment; we found that LAG3 expression on CD8+ T cells was increased in secondary lymphoid tissues obtained from CLL patients, compared to peripheral blood lymphocytes. This is in agreement with previous studies that reported increased expression of LAG3 on CD8+ T cells infiltrating some solid tumors as well as in a murine model of CLL.12,26,27 We also show that LAG3 expression was detected almost exclusively on PD1 presenting CD8+ lymphocytes. Co-expression of LAG3 together with PD1 on TILs identifies a highly exhausted T-cell population, and the synergy between these inhibitory receptors appears to impose tumor-induced immune tolerance in solid tumors.8,11,28-30 Blocking LAG3 enhanced both CD4+ and CD8+ T-cell activation, while blocking the PD-1/PD-L1 pathway did not affect T-cell activation. This is perhaps in agreement with recently published data showing only a modest effect for anti-PD1 pembrolizumab in patients with CLL.31 When expressed on immune cells present in the microenvironment, LAG3 may induce immune tolerance and exhaustion of LAG3-expressing cells through its interaction with the MHCII-presenting CLL cells. Hence, it is feasible that LAG3 could be targeted in an attempt to enhance anti-tumor immunogenicity. In the study herein, we demonstrated that CLL cells not only express, but also secrete sLAG3. Additionally, the mere addition of anti-LAG3 antibodies to CLL cells increased spontaneous apoptosis. This may be indicative of the existence of a vicious cycle in which LAG3 (either secreted by CLL or T cells, or presented on immune cells) and its interaction with MHCII on CLL cell surfaces promotes CLL cell activation and enhances their survival. Our data suggests that targeting LAG3-MHCII engagement could be considered as a potentially novel form of antiCLL immunotherapy. Funding The study was supported by The Varda and Boaz Dotan Research Center in Hemato-Oncology affiliated with the CBRC at Tel-Aviv University, Israel. CS and AW are supported by the intramural program of the National Heart, Lung and Blood Institute, NIH.

3. Gladstone DE, Fuchs E. Hematopoietic stem cell transplantation for chronic lymphocytic leukemia. Curr Opin Oncol. 2012; 24(2):176-181. 4. Lanham S, Hamblin T, Oscier D, Ibbotson R, Stevenson F, Packham G. Differential signaling via surface IgM is associated with VH gene mutational status and CD38 expression in chronic lymphocytic leukemia. Blood. 2003 ;101(3):1087-1093. 5. Stevenson FK, Krysov S, Davies AJ, Steele AJ, Packham G. B-cell receptor signaling in

chronic lymphocytic leukemia. Blood. 2011;118(16):4313-4320. 6. 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. 7. Valujskikh A, Li XC. Memory T cells and their exhaustive differentiation in allograft tolerance and rejection. Curr Opin organ Transplant. 2012;17(1):15-19.

881


M. Shapiro et al. 8. Matsuzaki J, Gnjatic S, Mhawech-Fauceglia P, et al. Tumor-infiltrating NY-ESO-1-specific CD8+ T cells are negatively regulated by LAG-3 and PD-1 in human ovarian cancer. Proc Natl Acad Sci USA. 2010; 107(17):7875-7880. 9. Fourcade J, Sun Z, Benallaoua M, et al. Upregulation of Tim-3 and PD-1 expression is associated with tumor antigen-specific CD8+ T cell dysfunction in melanoma patients. J Exp Med. 2010;207(10):21752186. 10. Norde WJ, Hobo W, van der Voort R, Dolstra H. Coinhibitory molecules in hematologic malignancies: targets for therapeutic intervention. Blood. 2012; 120(4): 728-736. 11. Woo SR, Turnis ME, Goldberg MV, et al. Immune inhibitory molecules LAG-3 and PD-1 synergistically regulate T-cell function to promote tumoral immune escape. Cancer Res. 2012;72(4):917-927. 12. Grosso JF, Kelleher CC, Harris TJ, et al. LAG-3 regulates CD8+ T cell accumulation and effector function in murine self- and tumor-tolerance systems. J Clin Invest. 2007;117(11):3383-3392. 13. Triebel F. LAG-3: a regulator of T-cell and DC responses and its use in therapeutic vaccination. Trends Immunol. 2003; 24(12):619-622. 14. Romano E, Michielin O, Voelter V, et al. MART-1 peptide vaccination plus IMP321 (LAG-3Ig fusion protein) in patients receiving autologous PBMCs after lymphodepletion: results of a Phase I trial. J Transl Med. 2014;12:97. 15. Hemon P, Jean-Louis F, Ramgolam K, et al. MHC class II engagement by its ligand LAG-3 (CD223) contributes to melanoma resistance to apoptosis. J Immunol. 2011; 186(9):5173-5183. 16. Kotaskova J, Tichy B, Trbusek M, et al. High expression of lymphocyte-activation

882

17.

18.

19.

20.

21.

22.

23.

gene 3 (LAG3) in chronic lymphocytic leukemia cells is associated with unmutated immunoglobulin variable heavy chain region (IGHV) gene and reduced treatmentfree survival. J Mol Diagn. 2010;12(3):328334. Hallek M, Cheson BD, Catovsky D, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood. 2008; 111(12):5446-5456. Wiestner A, Rosenwald A, Barry TS, et al. ZAP-70 expression identifies a chronic lymphocytic leukemia subtype with unmutated immunoglobulin genes, inferior clinical outcome, and distinct gene expression profile. Blood. 2003;101(12):4944-4951. Guy K, Meehan RR, Dewar AE, Larhammar D. Expression of MHC class II antigens in human B-cell leukaemia, and increased levels of class II antigens and DRspecific mRNA after stimulation with 12O-tetradecanoyl phorbol-13-acetate. Immunology. 1986;57(2):181-188. Baixeras E, Huard B, Miossec C, , et al. Characterization of the lymphocyte activation gene 3-encoded protein. A new ligand for human leukocyte antigen class II antigens. J Exp Med. 1992 ;176(2):327-337. Huard B, Mastrangeli R, Prigent P, et al. Characterization of the major histocompatibility complex class II binding site on LAG3 protein. Proc Natl Acad Sci USA. 1997; 94(11):5744-5749. Ranheim EA, Kipps TJ. Activated T cells induce expression of B7/BB1 on normal or leukemic B cells through a CD40-dependent signal. J Exp Med. 1993; 177(4):925-935. Scholl PR, Geha RS. MHC class-II signaling in B-cell activation. Immunol Today. 1994; 15(9):418-422.

24. Triebel F, Hacene K, Pichon MF. A soluble lymphocyte activation gene-3 (sLAG-3) protein as a prognostic factor in human breast cancer expressing estrogen or progesterone receptors. Cancer Lett. 2006; 235(1):147-153. 25. Nashar TO, Hirst TR, Williams NA. Modulation of B-cell activation by the B subunit of Escherichia coli enterotoxin: receptor interaction up-regulates MHC class II, B7, CD40, CD25 and ICAM-1. Immunology. 1997;91(4):572-578. 26. Gassner FJ, Zaborsky N, Catakovic K, et al. Chronic lymphocytic leukaemia induces an exhausted T cell phenotype in the TCL1 transgenic mouse model. Br J Haematol. 2015;170(4):515-522. 27. Demeure CE, Wolfers J, Martin-Garcia N, Gaulard P, Triebel F. T Lymphocytes infiltrating various tumour types express the MHC class II ligand lymphocyte activation gene-3 (LAG-3): role of LAG-3/MHC class II interactions in cell-cell contacts. Eur J Cancer. 2001;37(13):1709-1718. 28. Turnis ME, Korman AJ, Drake CG, Vignali DA. Combinatorial Immunotherapy: PD-1 may not be LAG-ing behind any more. Oncoimmunology. 2012 ;1(7):1172-1174. 29. Grosso JF, Goldberg MV, Getnet D, et al. Functionally distinct LAG-3 and PD-1 subsets on activated and chronically stimulated CD8 T cells. J Immunol. 2009;182(11): 6659-6669. 30. Baitsch L, Legat A, Barba L, et al. Extended co-expression of inhibitory receptors by human CD8 T-cells depending on differentiation, antigen-specificity and anatomical localization. PLoS One. 2012;7(2):e30852. 31. Wei Ding, Jennifer Le-Rademacher, Timothy G. Call, et al. PD-1 blockade with pembrolizumab in relapsed CLL including Richterâ&#x20AC;&#x2122;s transformation: an updated report from a phase 2 trial (MC1485). Blood. 2016; 128(22):4392.

haematologica | 2017; 102(5)


ARTICLE

Non-Hodgkin Lymphoma

c-Myc dysregulation is a co-transforming event for nuclear factor-κB activated B cells Amandine David,1,2* Nicolas Arnaud,1,2* Magali Fradet,1,2 Hélène Lascaux,1,2 Catherine Ouk-Martin,1,2,3 Nathalie Gachard,1,2 Ursula Zimber-Strobl,4 Jean Feuillard,1,2 and Nathalie Faumont1,2

1 CNRS-UMR 7276, University of Limoges, France; 2Hematology Laboratory of Dupuytren Hospital University Center (CHU) of Limoges, France; 3Platform of Cytometry and Imagery (CIM), University of Limoges, France and 4Research Unit Gene Vectors, Helmholtz Center Munich, German Research Center for Environmental Health GmbH, Germany

*AD and NA contributed equally to this work.

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(5):883-894

ABSTRACT

W

hile c-Myc dysregulation is constantly associated with highly proliferating B-cell tumors, nuclear factor (NF)-κB addiction is found in indolent lymphomas as well as diffuse large B-cell lymphomas, either with an activated B-cell like phenotype or associated with the Epstein-Barr virus. We raised the question of the effect of c-Myc in B cells with NF-κB activated by three different inducers: Epstein-Barr viruslatency III program, TLR9 and CD40. Induction of c-Myc overexpression increased proliferation of Epstein-Barr virus-latency III immortalized B cells, an effect that was dependent on NF-κB. Results from transcriptomic signatures and functional studies showed that c-Myc overexpression increased Epstein-Barr virus-latency III-driven proliferation depending on NF-κB. In vitro, induction of c-Myc increased proliferation of B cells with TLR9-dependant activation of MyD88, with decreased apoptosis. In the transgenic lc-Myc mouse model with c-Myc overexpression in B cells, in vivo activation of MyD88 by TLR9 induced splenomegaly related to an increased synthesis phase (S-phase) entry of B cells. Transgenic mice with both continuous CD40 signaling in B cells and the lc-Myc transgene developed very aggressive lymphomas with characteristics of activated diffuse large B-cell lymphomas. The main characteristic gene expression profile signatures of these tumors were those of proliferation and energetic metabolism. These results suggest that c-Myc is an NF-κB co-transforming event in aggressive lymphomas with an activated phenotype, activated B-cell like diffuse large B-cell lymphomas. This would explain why NF-κB is associated with both indolent and aggressive lymphomas, and opens new perspectives on the possibility of combinatory therapies targeting both the c-Myc proliferating program and NF-κB activation pathways in diffuse large B-cell lymphomas. Introduction NF-κB associated aggressive B-cell lymphomas can be subdivided into at least two main categories: diffuse large B-cell lymphomas (DLBCLs) of immunocompetent patients with an activated B-cell-like (ABC) gene expression signature, and Epstein-Barr virus (EBV) associated DLBCLs, either in elderly or immunocompromised patients. ABC-DLBCLs exhibit an NF-κB addiction and have a worse prognosis when compared to DLBCLs with a germinal center B-cell-like (GCB) signature, the other main molecular DLBCL subtype.1 In ABC-DLBCLs, activation of NFκB is due either to autoactivation of the CD40 signalosome2 or to NF-κB activating mutations, among them mutations of TNFAIP3 (A20) or MYD88 (the most frequent, concerning 39% of ABC-DLBCLs).3 Prognosis of EBV-associated DLBCLs of the elderly who have an ABC-DLBCL profile is poor.4,5 In these cases, NF-κB actihaematologica | 2017; 102(5)

Correspondence: nathalie.faumont@unilim.fr

Received: September 8, 2016. Accepted: February 21, 2017. Pre-published: February 23, 2017. doi:10.3324/haematol.2016.156281 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/883 ©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.

883


A. David et al.

vation is due to the expression of the latent membrane protein 1 (LMP1), the main oncoprotein of EBV. NF-κB inhibition in these tumors induces apoptosis.5,6 Immunodeficient patients are also prone to aggressive Bcell lymphoproliferative disorders, which are often associated with EBV and inhibition of NF-κB leading to apoptosis of tumor cells.7–10 The activated B-cell phenotype of all these aggressive Bcell lymphomas is largely due to NF-κB activation. These tumors also exhibit a high proliferative index,11 likely due to the dysregulation of c-Myc activity.12 Genetic alterations of MYC, although more frequent in GCB-DLBCLs, are also found in ABC-DLBCL, and c-Myc overexpression is a negative predictor of survival in both ABC and GCBDLBCLs.13 In EBV-positive DLBCLs, Epstein-Barr virus nuclear antigen 2 (EBNA2) expression is a poor prognosis factor.14 By subverting the Notch pathway through targeting of the RBP-Jκ nuclear factor, EBNA2 is the EBV protein responsible for the EBV-latency III program (also called proliferating program)15 and directly upregulates LMP1 gene expression, which in turns activates NF-κB.16 EBNA2 itself directly contributes to protection against apoptosis.15,17 EBNA2 is also responsible for c-Myc deregulation.18 In vitro, EBV-driven B-cell proliferation is directly related to the activity of two master transcription factors: c-Myc and NF-κB.19 c-Myc is the master transcription factor for cell proliferation and is involved in numerous hematological and solid cancers.20 Several transgenic mice models, including Em-Myc, lc-Myc or 3’RR-c-Myc models21–23 as well as Burkitt’s lymphomas (BLs) demonstrate that c-Myc overexpression in B cells leads to the emergence of aggressive B-cell lymphomas, but with a non-activated phenotype. Based on these features, we raised the question concerning the effect of c-Myc overexpression in B cells with an NF-κB activated B-cell-like phenotype. We showed that cMyc constantly promoted B-cell proliferation of NF-κB activated B cells in different models depending either on EBV, MyD88 or CD40, in vitro and in vivo. Co-regulation of c-Myc and CD40 in a mouse model led to very aggressive B-cell lymphomas with an activated phenotype.

Methods See the Online Supplementary Materials and Methods for a description of the techniques.

Cells EREB2.5 cells are a non-classical LCL with an estradiol inducible EBV-latency III proliferation program.24 The P493.6 cell line is an EREB2.5 derivative transfected with a Tet-Off inducible c-Myc expressing vector.25

EMSAs, Plasmid Constructs, Western Blotting and RT-PCR Methods for nuclear extracts, electrophoretic mobility shift assays (EMSAs), Western blots and reverse transcription-polymerase chain reactions (RT-PCRs) are described elsewhere.26 Complementary DNA for IκBαS32,36A (super-repressor form of IκBα) has already been published.27

Cell Labeling, proliferation and immunohistochemistry Red blood cell lysis buffer came from eBioscience, San Diego, CA, USA. To assess proliferation, carboxyfluorescein diacetate 884

succinimidyl ester (CFSE) and 5-ethynyl-2’-deoxyuridine (EdU) were both obtained from Life Technologies, while 5-bromo-2'deoxyuridine (BrdU) came from Sigma-Aldrich, Saint-Louis, MO, USA. Ki-67 labeling was also used to follow proliferation using imaging flow cytometry with the ImageStream 100 apparatus (Amnis®; Merck, Darmstadt, Germany).

Gene Expression Profiling Amplification of ribonucleic acids (RNAs) and hybridization onto microarrays were performed on an Affymetrix GeneAtlas® System with: Affymetrix® Human Genome U219 Array Strip, and Affymetrix® Mouse Gene 2.1 ST Array Strip as previously described.26

Mouse Models

Information on lc-Myc mice and mice with the CD19-Cre conditional LMP1.CD40 fusion transgene have already been published.22,28 All procedures were conducted under an approved protocol according to European guidelines for animal experimentation (French national authorization number: 87-022 and French ethics committee registration number “CREEAL”: 09-07-2012).

Results c-Myc increases NF-κB dependant EBV-latency III B-cell proliferation c-Myc and NF-κB are the two master transcriptional factors of EBV-latency III proliferating B cells.19 To understand how c-Myc interferes with the EBV-latency III proliferation program, we used the EBV-infected P493.6 B-cell line, a cell line that is double conditional for c-Myc and EBNA2, allowing for growth under the c-Myc or EBV-latency III program.25 Indeed, P493.6 cells are thought to be an in vitro model of normal B cells that can be forced to adopt four different proliferation statuses: quiescent state (c-Myc/EBNA2-), EBV-latency III proliferating program (c-Myc/EBNA2+), BL-like c-Myc proliferating program (cMyc+/EBNA2-), and both programs (c-Myc+/EBNA2+) (see the Online Supplementary Figure S1 and its legend for a detailed description of this model). We first analyzed transcriptional changes induced by EBV-latency III when associated with c-Myc overexpression. Supervised analysis led to the selection of 1,648 probes with at least a two-fold signal variation in one of the four P493.6 cell conditions when compared to the median of each probe. Genes could be grouped into five clusters by hierarchical clustering. Their main relevant biological functions are shown in Figure 1 (see also Online Supplementary Table S1). Cluster 1 corresponds to genes that were strongly repressed when both EBV-latency III and c-Myc programs were induced (c-Myc+/EBNA2+). Most genes were associated with immune response and the induction of apoptosis. Cluster 2 genes (such as CD80, CFLAR/c-Flip, TRAF1, EBI3, and TNFAIP3/A20) were induced by the EBV-latency III program alone (c-Myc/EBNA2+), these genes are known to be NF-κB targets. These genes were repressed by c-Myc (compare cMyc+/EBNA2- and c-Myc-/EBNA2+ conditions) but were still expressed in the presence of both proliferating programs. Genes belonging to cluster 3 were likely to be targets of signaling pathways repressed by c-Myc in the presence or absence of EBV-latency III (c-Myc+/EBNA2+ and c-Myc+/EBNA2-). These signaling pathways included interferon, JAK/STAT, NF-κB, Jun, Erk, and Akt pathways. haematologica | 2017; 102(5)


c-Myc overexpression in NF-κB activated B-cells

The last two clusters (clusters 4 and 5) were genes induced either by EBV-latency III (c-Myc+/EBNA2+) or c-Myc (cMyc+/EBNA2-) programs alone for which expression was over-induced when both programs were switched on (cMyc+/EBNA2+). The functions of these genes were nucleic acid metabolic processes, energy metabolism, and proliferation. We subsequently studied the functional consequences of co-activation of both EBV-latency III and c-Myc programs. Induction of c-Myc by tetracycline retrieval increased proliferation of EBV-latency III P493.6 cells in a dose-dependent manner, reaching 45% with 1 mM estradiol (c-Myc+/EBNA2+) compared to 29% and 37% with

EBV-latency III or c-Myc alone, respectively (Figure 2A). Using an imaging flow cytometer, the strongest labeling of the Ki-67 proliferation marker was observed when both proliferation programs were induced (Figure 2B-D). Morphological evaluation of apoptotic cells by quantification of nuclear fragmentation29 revealed that, as expected, induction of the EBV-latency III program increased protection against apoptosis and that induction of c-Myc alone was not deleterious in this cell type, as reported by Schuhmacher et al.30 (Figure 2E). These results suggest that proliferation was increased when c-Myc overexpression was associated with the EBV-latency III program (cMyc+/EBNA2+), maintaining protection from apoptosis.

Figure 1. Transcriptomic changes induced by c-Myc overexpression on EBV-latency III proliferating P493.6 cells. Gene expression profiles of P493.6 cells induced (+) or not (-) for EBNA2 and/or c-Myc for 48 hours. Each cell condition was repeated once. Selection of the 1,648 probes was based on at least a two-fold variation in one of the four cell conditions when compared to the median. After ascending hierarchical clustering on median centered values, five main clusters were defined as the five ultimate ascending clusters with a global positive correlation value. For each cluster and each cell condition, centered means of gene expression values with standard errors (±SE) are indicated on the left. Most representative functions with most significant genes are mentioned on the right. Log2 value color codes are shown beside the heat map. EBNA2: Epstein-Barr virus nuclear antigen 2; EBV: Epstein-Barr virus; NF-κB: nuclear factor-κB.

haematologica | 2017; 102(5)

885


A. David et al.

Since P493-6 cells could have been selected to resist the potential deleterious effect of c-Myc overexpression, we also transfected two classical LCLs, named PRI and JEF, with a doxycycline-inducible PRT-1 vector overexpressing c-Myc.31 Results suggest that mild induction of c-Myc increased proliferation of both classical LCLs while, as a control, proliferation of these cells was stable over luciferase induction levels (data not shown). Altogether, the results presented suggest that both EBVlatency III and c-Myc programs brought complementary advantages for cell transformation, both acting concomitantly on cell proliferation and cell metabolism, with EBVlatency III acting more specifically on protection against apoptosis and c-Myc on repression of interferon response genes. Albeit in a less intense manner than in EBV-latency III proliferating cells, NF-κB was clearly activated in c-Myc+/EBNA2+ P493.6 cells when compared to c-Myc+/EBNA2- cells (see the Online Supplementary Figure

A

B

C

D

E

886

S2 for details on NF-κB activation in the P493.6 cell line). We thus tested the effect of NF-κB inhibition on these cells. NF-κB was first repressed by overexpression of the super-repressor IκBαS32,36A.27 Inhibition of NF-κB decreased cell proliferation regardless of whether c-Myc was induced or not in EBV-latency III proliferating P493.6 cells (Figure 3A). The percentage of sub-G1 cells also increased after NF-κB inhibition (data not shown). We also treated cells with PHA-408 (Figure 3B,C), an inhibitor of the IκB kinase 2 (IKK2) subunit of the IKK complex.32 PHA-408 significantly decreased NF-κB DNA binding activity when the EBV-latency III program was switched on (Online Supplementary Figure S3; Figure 3B: lanes 1, 5, 9 in the presence of c-Myc, and lanes 3, 7, 11 in the absence of c-Myc). In the presence of PHA-408, P493.6 proliferation driven by both EBV-latency III plus c-Myc programs was similar to that of cells driven by c-Myc alone. As control, the addition of PHA-408 marginally affected proliferation of P493.6 cells when c-Myc alone was turned on

Figure 2. Proliferation and apoptosis of P493.6 cells after induction of c-Myc programs in EBV-latency III B cells. (A) Percentage of BrdU positive cells was assessed by flow cytometry on P493.6 cells treated with two doses of estradiol (0.1 µM: EBNA2-/+ and 1 µM: EBNA2+) or not (EBNA2-) and/or with tetracycline (100 µg/ml: c-Myc-; 3 µg/ml: c-Myc-/+; and not: c-Myc+) for 48 hours. Data are shown as mean ± standard deviation (n=3 independent experiments). (B to E) P493.6 cells were treated or not with 1 µM estradiol and/or 100 µg/ml tetracycline for 48 hours to induce either cell quiescence (EBNA2-, c-Myc-), EBV-latency III alone (EBV+, c-Myc-), c-Myc alone (EBNA2-, cMyc+), and both c-Myc and EBV-latency III (EBNA2+, c-Myc+). (B,C and D) Percentage of cells with strong staining of the Ki-67 proliferation marker in the nucleus, assessed by imaging flow cytometry. (B) Examples of cells under the c-Myc program with low or high Ki-67 expression. Selection of cells was based on colocalization of Ki-67 and the DRAQ5® nuclear dye. (C) Representative monoparametric histograms of Ki-67 intensity gated on cells with colocalization of Ki-67 and DRAQ5®. Each cell condition is indicated at the top left of each histogram. Threshold for strong Ki-67 labeling is indicated by the gray box. Percentage of cells with high Ki67 labeling is indicated on the top right of each histogram. (D) Total percentage of high nuclear stained Ki-67 cells in the four culture conditions of P493.6 cells. Data are shown as mean ± standard deviation (n=5 independent experiments). (E) Apoptosis of the uncolocalized Ki-67 and DRAQ5® population using imaging flow cytometry. Left panel: example of living and apoptotic cells under the c-Myc condition. Right panel: total percentage of apoptotic cells in the four culture conditions. Quantification was based on morphometric parameters.29 Data are shown as mean ± standard deviation (n=3 independent experiments). Statistical significance was determined by Student’s t-test (*** P<0.001; **P<0.01; *P<0.05). BrdU: 5bromo-2'-deoxyuridine; EBNA2: EpsteinBarr virus nuclear antigen 2; EBV: EpsteinBarr virus; DRAQ5: deep red anthraquinone 5.

haematologica | 2017; 102(5)


c-Myc overexpression in NF-κB activated B-cells

(Figure 3C). This suggests that when both c-Myc and the EBV-latency III program were induced, increased proliferation was mainly due to both c-Myc and NF-κB.

C-Myc increases TLR9 induced B-cell proliferation both in vitro and in vivo Since activating mutations of MyD88 are frequently found in aggressive B-cell lymphomas with activation of NF-κB,33 we decided to assay the effect of both c-Myc and TLR9 activation in B cells. In vitro, resting P493.6 cells (cMyc-/EBNA2-) were stimulated with or without either short control oligodeoxynucleotides (ODN-ctl) or with unmethylated CpG motifs (ODN-CpG). TLR9 activation moderately but significantly increased the S-phase entry of resting P493.6 cells (Figure 4A). c-Myc induction

(c-Myc+/EBNA2-) led to a considerable acceleration of cell proliferation; an effect that was further enhanced on TLR9 activated cells (Figure 4A). As expected, NF-κB targets, such as TRAF1, were upregulated upon ODN-CpG treatment (Figure 4B). When NF-κB was inhibited after the addition of PHA-408, as shown by inhibition of TRAF1 expression, a significant decrease in TLR9 activated B-cell proliferation was seen. This shows that enhanced proliferation of TLR9 activated P493.6 cells with c-Myc overexpression was dependent on NF-κB. This proliferative effect of TLR9 stimulation was confirmed in c-Myc overexpressing BL2 and BL41 BL cell lines (data not shown). We used the lc-Myc mice published by Kovalchuk et al.22 carrying an insertion of a translocated MYC gene cloned from the human BL60 cell line in order to investi-

A

B

C

haematologica | 2017; 102(5)

Figure 3. Activation of NF-κB in P493.6 cells after induction of c-Myc on EBV-latency III. P493.6 cells were treated or not with estradiol 1 mM (EBNA2- or EBNA2+) and/or with tetracycline at 100 ng/mL (c-Myc+ or c-Myc-) for 48 hours. (A) Percentage of BrdU positive cells after transient transfection with the pCDNA.3 empty vector (Ctl) or expressing the super-repressor IκBαS32,36A for 48 hours (upper panel). and Representative analysis of IκBαS32,36A, α-tubulin expression by Western blot (lower panel). (B and C) Effect of the IKK2 inhibitor PHA-408 (5 mM and 10 mM) for 48 hours on NF-κB. Representative DNA binding by EMSA (panel B) and BrdU incorporation (panel C). In the histograms, data are shown as the mean ± standard deviation (at least 3 independent experiments). Statistical significance was determined by Student’s t-test (****P<0.001; **P<0.01; *P<0.05). ns: not significant; BrdU: 5-bromo-2'-deoxyuridine; EBNA2: Epstein-Barr virus nuclear antigen 2; EBV: Epstein-Barr virus; Ctl: control; NFκB: nuclear factor-κB.

887


A. David et al.

gate the cooperation of c-Myc overexpression and TLR9 signaling in vivo. As reported, all lc-Myc mice developed high grade B-cell lymphomas with similarities to BL, and median survival was 14 weeks (Online Supplementary Figure S4A). At pre-tumor stage (5-6-week old) mice had increased spleen weight, white blood cell count, and percentage of activated CD86 positive B cells, but without affecting the percentage of total B cells in the spleen (Online Supplementary Figure S4B,C). TLR9 stimulation with ODN-CpG led to strong NF-κB activation in splenocytes from both lc-Myc and wild-type (wt) mice (Online Supplementary Figure S5A). In vivo repeated ODN-CpG intraperitoneal injections induced a significant increase in spleen size and weight in both wt and lc-Myc mice

(Figure 5A,B). This splenomegaly was much higher for half of the ODN-CpG injected lc-Myc mice when compared to the wt mice which were also stimulated: 361.3±49.7 mg vs. 216.9±14.5 mg (P = 0.0043). In vivo, ODN-CpG injection induced proliferation of B cells in spleens from both wt and lc-Myc mice. This increased proliferation was higher in ODN-CpG injected lc-Myc mice (Figure 5C). Similar results were obtained from lymph node B cells of the same mice (data not shown). To eliminate an indirect proliferating effect on ODNCpG on B cells via stimulation of the microenvironment, ex vivo ODN-CpG stimulations of total splenocytes were done. TLR9 stimulation induced a significant increase in proliferation and viability of lc-Myc splenocytes when

A

B

Figure 4. Effect of TLR9 stimulation and NF-κB implication on c-Myc overexpressing P493.6 cells. P493.6 cells were treated (quiescent state) or not (c-Myc condition) with 100 ng/mL tetracycline and stimulated with 0.5 mM oligodeoxynucleotides control (ODN-ctl) or containing CpG (ODN-CpG) for 48 hours. (A) Representative bivariate BrdU propidium iodide cytograms. Percentage of S-phase cells is indicated on each cytogram. (B) Upper panel: percentage of BrdU positive cells after ODN stimulation or not (ns) and 10 mM PHA-408 treatment. Data are shown as the mean ± standard deviation (at least 4 independent experiments). Statistical significance was determined by Student’s t-test (****P<0.0001; ***P<0.001). Lower panel: representative expression of c-Myc, TRAF1 and α-Tubulin as loading control by Western blot detection. BrdU: 5-bromo-2'-deoxyuridine.

888

haematologica | 2017; 102(5)


c-Myc overexpression in NF-κB activated B-cells

compared to wt cells (Online Supplementary Figure S5B,C). B-cell labeling after EdU incorporation showed that proliferating splenocytes belonged to the B-cell fraction with a higher proliferation when B cells were from lc-Myc spleens (Figure 5D). Cell growth assay and Ki-67 labeling on sorted B-cell subsets showed that this increased proliferation was mainly due to follicular (CD23-high, CD21low) and marginal zone (CD23-low, CD21-high) B cells (Figure 5E,F). These data show that in vivo MYC/IG translocation with c-Myc overexpression provided an additional signal driving a stronger proliferation and survival of TLR9 activated B cells.

In vivo dysregulation of both CD40 signaling and c-Myc leads to aggressive diffuse large B-cell lymphoma with an activated phenotype To assess the role of c-Myc in an in vivo model of NF-κB dependent B-cell lymphoma, we crossbred lc-Myc with

A

B

C

D

E

F

haematologica | 2017; 102(5)

L.CD40 transgenic mice (L.CD40/lc-Myc transgenic mice). L.CD40 mice are characterized by surface membrane expression of a chimeric LMP1/CD40 protein in CD19 positive B cells, resulting in a constitutively active CD40 signal in the B-cell lineage with constitutive activation of the NF-κB pathway as well as the mitogen-activated protein kinases Jnk and extracellular signal–regulated kinase.28 In these mice, deregulated CD40 signaling induces the development of indolent splenic B-cell lymphomas with an increase in the marginal zone B-cell compartment in more than 60% of mice older than 1 year. Spleen size and weight were much higher in L.CD40/lc-Myc transgenic mice when compared to control mice (wt, L.CD40, and lc-Myc mice) (Figure 6A,B). L.CD40/lc-Myc mice prematurely developed enlarged lymph nodes at 10±2 weeks (vs. 12±3 weeks for lc-Myc mice). In contrast, lymph nodes were not enlarged in L.CD40 and wt mice (Figure 6B). If percentages of CD80

Figure 5. Effect of TLR9 stimulation on B cells from 4-5-week old lc-Myc (pretumor stage) and wild-type (wt) mice. In vivo stimulation with control (ODN-ctl) or CpG containing (ODN-CpG) oligodeoxynucleotides: 10-fold 50 mg ODN intraperitoneal injection every two days. Representative spleen size (A) and scatterplot from spleen weight (B). B220 positive splenocyte proliferation after in vivo BrdU incorporation for about 24 hours (C): upper panel, overlay of representative BrdU labeling histograms gated on B220 positive cells from ODN-CpG injected lc-Myc and wt mice; and lower panel, mean of in vivo BrdU percentage on B220 positive splenocytes. Data are shown as mean ± standard deviation (n=6 mice in each experiment group). (D) Total splenocytes were ex vivo stimulated or not (ns) with control (ODN-ctl) or CpG containing motif (ODN-CpG) oligodeoxynucleotides. Proliferation at day 4 was assessed by EdU incorporation and labeling of B220 positive splenocytes. Upper panel: representative biparametric EdU side scatter cytograms gated on B220 positive cells. Percentages of EdU positive cells are indicated in each graph. Lower panel: mean percentages of EdU+ and B220+-cells. Data are shown as mean ± standard deviation (n=4 mice in each experiment group). (E) Cell growth of marginal zone (MZ), follicular (FO) and double negative (DN) spleen B cells sorted from wt and λc-Myc mice. Cells were counted by Trypan blue staining exclusion after 4 days stimulation. Red line indicates no cell growth fold increase in ODNCpG condition versus ODN-ctl condition (value = 1). Data are shown as mean ± standard deviation (n=3 mice in each experiment group). (F) BrdU labeling of sorted FO B cells from wt (left panels) and λc-Myc mice (right panels) treated (CpG, lower panels) or not (ctl, upper panels) with CpG oligonuclotides for 4 days. BrdU incorporation was performed 4 hours before cells were cytocentrifuged. No BrdU incorporation was detectable in ctl conditions (upper panels). Mean and standard deviation of percentages of BrdU positive CpG treated cells are indicated (n=3 mice in each experiment group). Statistical significance was determined by Student’s t-test (****P<0.0001; ***P<0.001; **P<0.01; *P<0.05). EdU: 5-ethynyl-2’-deoxyuridine; BrdU: 5-bromo-2'-deoxyuridine.

889


A. David et al. A

C

B

D

E

F

Figure 6. Phenotypic analysis of L.CD40/lc-Myc mice compared to wild-type (wt), L.CD40, and lc-Myc control mice. (A) Representative spleen size. (B) Scatter-plot of spleen (sp) and lymph node (ln) weights. For each group, number (n) and mean age of mice (w, week) are indicated. Data are shown as mean ± standard deviation (n=3-11 mice in each experiment group). Statistical significance was determined by Student’s t-test (**P<0.01). (C, D, and E) B220, CD80, and CD86 labeling from total splenocytes and lymph node cells. Histograms of activated B-cell percentages on B220+ cells i.e., percentage of CD80+ and/or CD86+ cells (C), mean fluorescent intensity (MFI) of CD80 (D) and CD86 (E) on B220 positive cells. MFI was normalized from wt mice (MFI=1 in red). Data are shown as mean ± standard deviation (n=3-11 mice in each experiment group). Statistical significance was determined by Student’s t-test (*P<0.05; **P<0.01; ***P<0.001; ****P<0.0001). (F) H&E and Ki-67 staining on lymph node and spleen sections.

890

haematologica | 2017; 102(5)


c-Myc overexpression in NF-κB activated B-cells

or CD86 activated B cells were increased in a comparable manner in both lc-Myc and L.CD40/lc-Myc mice (Figure 6C), CD80 and CD86 mean fluorescent intensities on B cells were significantly increased in L.CD40/lc-Myc mice compared to lc-Myc mice, both in spleen and lymph nodes (Figure 6D,E). The viability of splenocytes after

three days culture in complete medium without stimulation indicated that protection against cell death was increased in L.CD40/lc-Myc tumor cells when compared to other conditions (Online Supplementary Figure S6). In both lc-Myc and L.CD40/lc-Myc mice, lymphoid tissue architecture was lost with a massive tumor infiltration,

Figure 7. Gene expression profile of L.CD40/lc-Myc tumors compared to lc-Myc mice tumors. Supervised analysis led to the selection of 2,437 significant probes, which were initially partitioned into 30 clusters by K-means clustering so that the number of probes was above 15. The closest groups were merged 2 by 2 according to their proximity by hierarchical clustering and principal component analysis of the mean vectors. This was repeated until maximization of the absolute value of χ2.19 The final resulting number of clusters was 13 (Cluster 1 to 13). For each cluster and each sample condition, fold change value compared to wild-type (wt) samples were calculated from the mean of gene expression values and are indicated on the heat map. Most representative functions with most significant genes are mentioned on the right. NF-κB target genes are indicated in red. Log2 value color codes are shown beside the heat map. NF-κB: nuclear factor-κB; BCR: B-cell receptor.

haematologica | 2017; 102(5)

891


A. David et al.

and the Ki-67 proliferation index was high (>90%) (Figure 6F). Tumor lymph nodes from lc-Myc histologically resembled BL with medium sized monomorphic tumor cells and the presence of numerous tingible body macrophages with apoptotic bodies giving a starry sky effect. In L.CD40/lc-Myc mice, lymphoma cells were highly atypical, being large and irregular, with abundant cytoplasm, large nuclei that possessed one to three or four prominent nucleoli usually located in the periphery of the nucleus, evoking the aspect of immunoblastic lymphomas according to the Bethesda classification of lymphoid neoplasms in mice34 (Figure 6F), for which the human counterpart would be DLBCLs with an activated phenotype. Taking wt mice as a reference, gene expression profiles of tumors from lymph node samples suggest that deregulated genes could be separated into thirteen clusters. These clusters were grouped into six categories: upregulated genes in lc-Myc mice alone, in both lc-Myc and L.CD40/lc-Myc mice, in L.CD40 mice alone, in L.CD40/lc-Myc when compared to lc-Myc mice, downregulated genes in both lc-Myc and L.CD40/lc-Myc mice, and in L.CD40/lc-Myc mice alone (Figure 7, see also Online Supplementary Table S2). In agreement with the morphology, a specific macrophage signature was found in upregulated genes in lc-Myc mice alone. Both lc-Myc and L.CD40/lc-Myc tumors had numerous upregulated genes involved in cell cycle regulation, such as E2f7/8, Cemp, Ccnb2 and Ccne1/2. Metabolism processes were also among the highly upregulated functions in both mice models. Genes whose expression was increased in L.CD40 mice alone include those of the CD8 cytotoxic T cells and natural killer cells response, as well as those of the inflammatory and interferon response. This is in agreement with the original publication reporting an activated T-cell phenotype in these mice.28 In the meantime, expression of the immunosuppressive CD274/PDL.1 was increased in L.CD40 mice. CD274/PDL.1 was also increased in L.CD40/lc-Myc when compared to lc-Myc mice. Among over-expressed genes in L.CD40/lc-Myc compared to lc-Myc mice, the main represented biological functions were inflammation, transformation, cell activation, angiogenesis, transcription regulators, and NF-κB regulators such as Ltb, Tlr3, CD40, RelB, and Traf1. CD40 expression was increased in both L.CD40 and L.CD40/lcMyc mice. This could be due to expression of the L.CD40 transgene or to endogenous CD40 expression, since one of the three probe sets spanning CD40 partially matched the C-terminal CD40 moiety of the L.CD40 protein while the two others did not. NF-κB target genes were induced in L.CD40/lc-Myc mice, including, for example, CD80, Fas, Tnfaip3, Bcl2, Ebi3 and Ccnd2. Repression of interferon response was significantly attenuated in these double L.CD40/lc-Myc mice when compared to lc-Myc mice, but was still present if compared to wt or L.CD40 mice. Genes coding for T-cell signature and T-cell activation markers were decreased in both lc-Myc and L.CD40/lcMyc mice. Thus, as regards P493.6 cells in the cMyc+/EBNA2+ condition, the main transcriptomic features of L.CD40/lc-Myc mice evoked increased proliferation, metabolism with apoptosis protection and attenuated interferon response as well as expression of the CD274/PDL1 molecule that inhibits the T-cell antitumor response. Moreover, expression of some B-cell markers such as CD79a, CD19 and CD22 was specifically decreased in L.CD40/lc-Myc. Even if lower CD19 expres892

sion could be the result of using CD19-Cre mice, its strongest decrease in L.CD40/lc-Myc mice together with the decrease of the B-cell receptor (BCR) associated CD22 molecule, of two BCR signaling molecules (CD79a and BANK1) as well as of a molecule normally expressed in germinal center B-cells (BCL11a), evokes features found in B cells with partial engagement in plasma cell differentiation. These results suggest that c-Myc overexpression confers an aggressive DLBCL phenotype with a high proliferative index and an activated phenotype to L.CD40 activated B cells.

Discussion EBV-positive DLBCLs of the elderly are characterized by prominent NF-κB and JAK/STAT activation,4,35 able to upregulate c-Myc expression in ABC-DLBCLs.36 B-cell lymphomas from HIV-infected patients very often have an ABC-DLBCL phenotype and are associated with MYC/IGH translocation.37 In LCLs, blocking c-Myc directly arrests cell growth and survival.19 In our EBV-infected human cell lines, the addition of c-Myc clearly favored cell growth with increased proliferation and metabolism, meanwhile, genes involved in immune surveillance and/or the inflammatory or interferon response were repressed. Enforced MyD88-L252P expression by retroviral infection of Em-Myc fetal liver cells dramatically shortened the time of tumor onset after transplantation in irradiated mice.38 In agreement, our results show that activation of MyD88 by TLR9 stimulation increased cell growth both in vitro in human cell lines and in vivo in the lc-Myc mouse model. c-Myc may play an important role in aggressive lymphomas with an activated phenotype related to NF-κB. It is a paradox that, as a primary event, NF-κB activation is associated with both indolent and aggressive B-cell nonHodgkin lymphomas (NHL). For example, mucosa-associated lymphoid tissue (MALT) indolent lymphomas frequently exhibit recurrent inactivating mutations of A20/TNFAIP3 that results in continuous activation of the classical NF-κB pathway, and these mutations are also found in ABC-DLBCLs.39,40 Waldenström macroglobulinemias are characterized by recurrent activating MyD88 mutations in 90% of cases, mainly MyD88-L265P, first described in ABC-DLBCLs,33,41,42 which again leads to the activation of NF-κB. This could suggest that NF-κB addiction in DLBCLs is associated with other genetic events that promote cell proliferation. To our knowledge, only three mouse models for indolent lymphomas of the spleen have been published, one mimicking TRAF3 inactivation,43 the second with constitutive expression of Bcl10,44 and the last one with continuous CD40 signaling.28 The three models are characterized by an increased activation of RelB (i.e., the NF-κB alternative pathway), and B-cell lymphomas preferentially located in the spleen, with expansion of the marginal zone. B-TRAF3–/– deficient mice developed lymphomas that have been reported to be either low or high grade and to diffuse into other lymphoid organs as well as the bone marrow.43 Bcl10 transgenic mice were characterized by an expansion of marginal zone B cell due to lymphocyte accumulation, and some mice older than 8 months developed lymphoma resembling the human marginal zone lymphomas of the spleen.44 The evolution of L.CD40 haematologica | 2017; 102(5)


c-Myc overexpression in NF-κB activated B-cells

transgenic mice was characterized by a progressive increase in splenomegaly and considerable expansion of the spleen B-cell area with acquisition of monoclonality, but the proliferative index remained constantly low, with no evidence for transformation into high grade lymphoma.28 Regarding the aggressive transformation of B cells in mice, Calado et al. reported that a mouse model with constitutive activation of IKK2 in B cells only developed plasma cell hyperplasia with a serum immunoglobulin peak.45 Additional ablation of the Blimp1/Prdm1 gene blocked plasma cell differentiation and led to the development of aggressive B-cell lymphomas resembling ABCDLBCLs. Zhang et al. showed that in vivo enforced expression of NIK in germinal center B cells led to plasma cell hyperplasia, again without lymphoma. But the combination of NIK and Bcl6 enforced expression led to the development of aggressive lymphomas resembling ABC-DLBCLs.46 These results on animal models suggest that in vivo continuous NF-κB activation would favor B cells, but other events, such as Blimp1 inactivation or Bcl6 dysregulation, seem to be needed for transformation into an aggressive DLBCL. Both BLIMP1 and BCL6 are transcriptional repressors of the MYC gene.47,48 Consequently, BLIMP1 disruption can cause increased c-Myc expression.20 In DLBCLs tumor B cells become refractory to c-Myc repression by BCL6.49 Furthermore, BCL6 expression represses that of BLIMP1.48 This suggests that BCL6 may indirectly activate c‐Myc transcription through repression of BLIMP1 expression.20 Thus, c-Myc could be the good endpoint target of various secondary events to then cooperate with NF-κB to promote aggressive B-cell lymphomagenesis. Contrasting with mice models with constitutive NF-κB activation in B cells; all mice models with c-Myc overexpression in B cells develop aggressive B-cell tumors. For example, lc-Myc mice had an early aggressive tumor onset primarily located in lymph nodes with secondary dissemination to the spleen.22 Morphology of these tumors evokes that of human BL, with a very high proliferative index. Our results show that double L.CD40/lc-Myc mice have very aggressive tumors with systemic dissemination to all lymphoid compartments of the organism. Immunoblastic morphology of these tumors evokes that of ABC-DLBCLs in humans, being completely different from that of both L.CD40 and lc-Myc mice. These features can be viewed either as the c-Myc-driven transformation of a CD40dependent indolent lymphoma, or as a shift from a c-Mycdriven aggressive B-cell lymphoma resembling human BL to an activated phenotype due to continuous CD40 activation. Functional studies clearly indicate cooperation between L.CD40 and lc-Myc transgenes in B-cell transformation. This cooperation was also found at the gene expression level, with upregulated genes involved in transformation, cell activation, and angiogenesis, such as Tnfaip3, Rab20/25, CD40, Arntl2, Ctgf, Bcl2, Rras, Ccnd2, and Jun. Most of these genes are known to be NF-κB targets.

References 1. Lenz G, Staudt LM. Aggressive lymphomas. N Engl J Med. 2010;362(15):1417–1429. 2. Pham LV, Tamayo AT, Yoshimura LC, et al.

haematologica | 2017; 102(5)

lc-Myc and L.CD40/lc-Myc tumors were characterized by a marked decrease in the T-cell signature. This could be either due to massive infiltration of the tumor or to lack of immune control. The growth of c-Myc-driven tumors is rather likely to be independent of their immune microenvironment.50 By contrast, escaping immune surveillance is clearly a key step for tumor emergence and/or tumor progression in DLBCLs.50 As illustrated herein by our transcriptome results, Hömig-Hölzel et al. have previously shown that L.CD40 mice exhibited activating T-cell expansion.28 However, CD274/PDL1 expression was found to be increased in L.CD40 tumors. Even if attenuated, such CD274/PDL1 expression was also increased in L.CD40/lc-Myc tumors when compared to those of lcMyc. Thus, combining the lack of T-cell signatures, including T-cell activation markers, with CD274/PDL1 expression by the tumor (a L.CD40 effect) and decreased interferon response (a c-Myc effect) in L.CD40/c-Myc mice could in part explain the aggressiveness of tumor growth, which is under a lesser degree of immune control. Therefore, c-Myc increased the tumor potency of NF-κB activated B cells in terms of proliferation, protection against apoptosis and tumor aggressiveness. In conclusion, we demonstrated that adding c-Myc to B cells with NF-κB activated by either LMP1, TLR9 or CD40, constantly increased the proliferation potential of cells and/or the aggressiveness of the tumor both in vitro and in animal models. Furthermore, we showed that adding c-Myc to CD40 activation in mice transformed indolent into aggressive lymphomas with features of NF-κB-activated cells. These results highlight the role of these two master transcriptional systems in B-cell lymphomagenesis, explaining why NF-κB is associated with both indolent and aggressive lymphomas and opening new perspectives on the possibility of combinatory therapies targeting both the c-Myc proliferating program and NF-κB activation pathways in DLBCLs. Acknowledgments We are especially grateful to Pr GW Bornkamm, Research Unit Gene Vectors, Helmholtz Center Munich, German Research Center for Environmental Health GmbH, Germany, for helping us to initiate the project and for continuous support and helpful discussions. We thank Dr A Marfak, Statistical Unit, IFCS, Rabat, Morocco, for transcriptome analysis. We thank Dr J Cook Moreau, UMR CNRS 7276, Limoges, France, for English editing. Funding The group of J Feuillard is supported by grants from the Ligue National contre le Cancer (Equipe Labellisée Ligue), the Institut National contre le Cancer (INCa), the Comité Orientation Recherche Cancer (CORC), the Limousin Region and the Haute Vienne and Corrèze comitees of the Ligue Nationale contre le Cancer and by the Lyons Club of Corrèze. U Zimber Strobk was supported by the DFG grant: GZ: ZI 1382/4-1.

A CD40 Signalosome anchored in lipid rafts leads to constitutive activation of NFkappaB and autonomous cell growth in B cell lymphomas. Immunity. 2002;16(1):3750. 3. Roschewski M, Staudt LM, Wilson WH.

Diffuse large B-cell lymphoma-treatment approaches in the molecular era. Nat Rev Clin Oncol. 2014;11(1):12-23. 4. Montes-Moreno S, Odqvist L, Diaz-Perez JA, et al. EBV-positive diffuse large B-cell lymphoma of the elderly is an aggressive

893


A. David et al.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16. 17.

18.

894

post-germinal center B-cell neoplasm characterized by prominent nuclear factor-kB activation. Mod Pathol Off J U S Can Acad Pathol Inc. 2012;25(7):968-982. Castillo JJ, Beltran BE, Miranda RN, Young KH, Chavez JC, Sotomayor EM. EBV-positive diffuse large B-cell lymphoma of the elderly: 2016 update on diagnosis, risk-stratification, and management. Am J Hematol. 2016;91(5):529-537. Bavi P, Uddin S, Bu R, et al. The biological and clinical impact of inhibition of NF- B-initiated apoptosis in diffuse large B cell lymphoma (DLBCL). J Pathol. 2011; 224(3):355366. Liebowitz D. Epstein-Barr virus and a cellular signaling pathway in lymphomas from immunosuppressed patients. N Engl J Med. 1998;338(20):1413-1421. Feuillard J, Schuhmacher M, Kohanna S, et al. Inducible loss of NF-kappaB activity is associated with apoptosis and Bcl-2 downregulation in Epstein-Barr virus-transformed B lymphocytes. Blood. 2000; 95(6):20682075. Zou P, Kawada J, Pesnicak L, Cohen JI. Bortezomib induces apoptosis of EpsteinBarr virus (EBV)-transformed B cells and prolongs survival of mice inoculated with EBVtransformed B cells. J Virol. 2007; 81(18):10029-10036. Chao C, Silverberg MJ, Martínez-Maza O, et al. Epstein-Barr virus infection and expression of B-cell oncogenic markers in HIVrelated diffuse large B-cell Lymphoma. Clin Cancer Res Off J Am Assoc Cancer Res. 2012;18(17):4702-4712. Liu Y-H, Xu F-P, Zhuang H-G, et al. Clinicopathologic significance of immunophenotypic profiles related to germinal center and activation B-cell differentiation in diffuse large B-cell lymphoma from Chinese patients. Hum Pathol. 2008; 39(6):875-884. Ott G, Rosenwald A, Campo E. Understanding MYC-driven aggressive Bcell lymphomas: pathogenesis and classification. Hematol Educ Program Am Soc Hematol Am Soc Hematol Educ Program. 2013;122(24):3884-3891. Valera A, López-Guillermo A, CardesaSalzmann T, et al. MYC protein expression and genetic alterations have prognostic impact in patients with diffuse large B-cell lymphoma treated with immunochemotherapy. Haematologica. 2013;98(10):1554-1562. Stuhlmann-Laeisz C, Borchert A, Quintanilla-Martinez L, et al. In Europe expression of EBNA2 is associated with poor survival in EBV-positive diffuse large Bcell lymphoma of the elderly. Leuk Lymphoma. 2016;57(1):39-44. Kohlhof H, Hampel F, Hoffmann R, et al. Notch1, Notch2, and Epstein-Barr virusencoded nuclear antigen 2 signaling differentially affects proliferation and survival of Epstein-Barr virus-infected B cells. Blood. 2009;113(22):5506-5515. Young LS, Rickinson AB. Epstein-Barr virus: 40 years on. Nat Rev Cancer. 2004; 4(10):757-768. Lee JM, Lee K-H, Weidner M, Osborne BA, Hayward SD. Epstein-Barr virus EBNA2 blocks Nur77- mediated apoptosis. Proc Natl Acad Sci USA. 2002;99(18):11878-11883. Kaiser C, Laux G, Eick D, Jochner N, Bornkamm GW, Kempkes B. The protooncogene c-myc is a direct target gene of Epstein-Barr virus nuclear antigen 2. J Virol. 1999;73(5):4481-4484.

19. Faumont N, Durand-Panteix S, Schlee M, et al. c-Myc and Rel/NF-kappaB are the two master transcriptional systems activated in the latency III program of Epstein-Barr virusimmortalized B cells. J Virol. 2009; 83(10):5014-5027. 20. Wierstra I, Alves J. The c-myc promoter: still MysterY and challenge. Adv Cancer Res. 2008;99:113-333. 21. Langdon WY, Harris AW, Cory S, Adams JM. The c-myc oncogene perturbs B lymphocyte development in E-mu-myc transgenic mice. Cell. 1986 Oct 10;47(1):11–18. 22. Kovalchuk AL, Qi CF, Torrey TA, et al. Burkitt lymphoma in the mouse. J Exp Med. 2000;192(8):1183–1190. 23. Truffinet V, Pinaud E, Cogné N, et al. The 3’ IgH locus control region is sufficient to deregulate a c-myc transgene and promote mature B cell malignancies with a predominant Burkitt-like phenotype. J Immunol. 2007;179(9):6033–6042. 24. Kempkes B, Zimber-Strobl U, Eissner G, et al. Epstein-Barr virus nuclear antigen 2 (EBNA2)-oestrogen receptor fusion proteins complement the EBNA2-deficient EpsteinBarr virus strain P3HR1 in transformation of primary B cells but suppress growth of human B cell lymphoma lines. J Gen Virol. 1996;77 ( Pt 2 ):227-237. 25. Pajic A, Spitkovsky D, Christoph B, et al. Cell cycle activation by c-myc in a burkitt lymphoma model cell line. Int J Cancer. 2000;87(6):787-793. 26. Chanut A, Duguet F, Marfak A, et al. RelA and RelB cross-talk and function in EpsteinBarr virus transformed B cells. Leukemia. 2014;28(4):871-879. 27. Traenckner EB, Pahl HL, Henkel T, Schmidt KN, Wilk S, Baeuerle PA. Phosphorylation of human I kappa B-alpha on serines 32 and 36 controls I kappa B-alpha proteolysis and NFkappa B activation in response to diverse stimuli. EMBO J. 1995;14(12):2876-2883. 28. Hömig-Hölzel C, Hojer C, Rastelli J, et al. Constitutive CD40 signaling in B cells selectively activates the noncanonical NF-kappaB pathway and promotes lymphomagenesis. J Exp Med. 2008; 205(6):1317-1329. 29. Le Clorennec C, Ouk T-S, Youlyouz-Marfak I, et al. Molecular basis of cytotoxicity of Epstein-Barr virus (EBV) latent membrane protein 1 (LMP1) in EBV latency III B cells: LMP1 induces type II ligand-independent autoactivation of CD95/Fas with caspase 8mediated apoptosis. J Virol. 2008;82(13): 6721-6733. 30. Schuhmacher M, Kohlhuber F, Hölzel M, et al. The transcriptional program of a human B cell line in response to Myc. Nucleic Acids Res. 2001;29(2):397-406. 31. Bornkamm GW, Berens C, Kuklik-Roos C, et al. Stringent doxycycline-dependent control of gene activities using an episomal onevector system. Nucleic Acids Res. 2005;33(16):e137. 32. Mbalaviele G, Sommers CD, Bonar SL, et al. A novel, highly selective, tight binding IkappaB kinase-2 (IKK-2) inhibitor: a tool to correlate IKK-2 activity to the fate and functions of the components of the nuclear factor-kappaB pathway in arthritis-relevant cells and animal models. J Pharmacol Exp Ther. 2009;329(1):1-25. 33. Ngo VN, Young RM, Schmitz R, et al. Oncogenically active MYD88 mutations in human lymphoma. Nature. 2011; 470(7332):115-119. 34. Morse HC, Anver MR, Fredrickson TN, et al. Bethesda proposals for classification of lymphoid neoplasms in mice. Blood. 2002;

100(1):246-258. 35. Kato H, Karube K, Yamamoto K, et al. Gene expression profiling of Epstein-Barr viruspositive diffuse large B-cell lymphoma of the elderly reveals alterations of characteristic oncogenetic pathways. Cancer Sci. 2014;105(5):537-544. 36. Ding BB, Yu JJ, Yu RY-L, et al. Constitutively activated STAT3 promotes cell proliferation and survival in the activated B-cell subtype of diffuse large B-cell lymphomas. Blood. 2008;111(3):1515-1523. 37. Morton LM, Kim CJ, Weiss LM, et al. Molecular characteristics of diffuse large Bcell lymphoma in human immunodeficiency virus-infected and -uninfected patients in the pre-highly active antiretroviral therapy and pre-rituximab era. Leuk Lymphoma. 2014;55(3):551-557. 38. Knittel G, Liedgens P, Korovkina D, et al. Bcell-specific conditional expression of Myd88p.L252P leads to the development of diffuse large B-cell lymphoma in mice. Blood. 2016;127(22):2732-2741. 39. Kato M, Sanada M, Kato I, et al. Frequent inactivation of A20 in B-cell lymphomas. Nature. 2009;459(7247):712-716. 40. Compagno M, Lim WK, Grunn A, et al. Mutations of multiple genes cause deregulation of NF-kappaB in diffuse large B-cell lymphoma. Nature. 2009;459(7247):717721. 41. Treon SP, Xu L, Yang G, et al. MYD88 L265P somatic mutation in Waldenström’s macroglobulinemia. N Engl J Med. 2012; 367(9):826-833. 42. Gachard N, Parrens M, Soubeyran I, et al. IGHV gene features and MYD88 L265P mutation separate the three marginal zone lymphoma entities and Waldenström macroglobulinemia/lymphoplasmacytic lymphomas. Leukemia. 2013;27(1):183-189. 43. Moore CR, Liu Y, Shao C, Covey LR, Morse HC, Xie P. Specific deletion of TRAF3 in B lymphocytes leads to B-lymphoma development in mice. Leukemia. 2012;26(5):11221127. 44. Li Z, Wang H, Xue L, et al. Emu-BCL10 mice exhibit constitutive activation of both canonical and noncanonical NF-kappaB pathways generating marginal zone (MZ) Bcell expansion as a precursor to splenic MZ lymphoma. Blood. 2009;114(19):4158-4168. 45. Calado DP, Zhang B, Srinivasan L, et al. Constitutive canonical NF- B activation cooperates with disruption of BLIMP1 in the pathogenesis of activated B cell-like diffuse large cell lymphoma. Cancer Cell. 2010;18(6):580-589. 46. Zhang B, Calado DP, Wang Z, et al. An oncogenic role for alternative NF-κB signaling in DLBCL revealed upon deregulated BCL6 expression. Cell Rep. 2015;11(5):715-726. 47. Lin Y, Wong K, Calame K. Repression of cmyc transcription by Blimp-1, an inducer of terminal B cell differentiation. Science. 1997;276(5312):596-599. 48. Shaffer AL, Yu X, He Y, Boldrick J, Chan EP, Staudt LM. BCL-6 represses genes that function in lymphocyte differentiation, inflammation, and cell cycle control. Immunity. 2000;13(2):199-212. 49. Ci W, Polo JM, Cerchietti L, et al. The BCL6 transcriptional program features repression of multiple oncogenes in primary B cells and is deregulated in DLBCL. Blood. 2009; 113(22):5536-5548. 50. Fowler NH, Cheah CY, Gascoyne RD, et al. Role of the tumor microenvironment in mature B-cell lymphoid malignancies. Haematologica. 2016;101(5):531-540.

haematologica | 2017; 102(5)


ARTICLE

Non-Hodgkin Lymphoma

Association between quality of response and outcomes in patients with newly diagnosed mantle cell lymphoma receiving VR-CAP versus R-CHOP in the phase 3 LYM-3002 study

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Gregor Verhoef,1 Tadeusz Robak,2 Huiqiang Huang,3 Halyna Pylypenko,4 Noppadol Siritanaratkul,5 Juliana Pereira,6 Johannes Drach,7 Jiri Mayer,8 Rumiko Okamoto,9* Lixia Pei,10 Brendan Rooney,11 Andrew Cakana,11 Helgi van de Velde12 and Franco Cavalli13 *Current affiliation: Chibanishi General Hospital, Chiba, Japan

1 University Hospital Leuven, Belgium; 2Medical University of Lodz, Copernicus Memorial Hospital, Poland; 3Sun Yat-sen University Cancer Center, Guangzhou, Guangdong, China; 4 Cherkassy Regional Oncology Center, Ukraine; 5Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand; 6Hospital das Clinicas da Faculdade de Medicina da USP, SĂŁo Paolo, Brazil; 7University of Vienna, Vienna General Hospital, Austria; 8 Masaryk University Hospital Brno, Czech Republic; 9Tokyo Metropolitan Cancer and Infectious Diseases Center, Komagome Hospital, Japan; 10Janssen Research & Development, LLC, Raritan, NJ, USA; 11Janssen Research & Development, High Wycombe, Buckinghamshire, UK; 12Millennium Pharmaceuticals, Inc., Boston, MA, USA and 13Oncology Institute of Southern Switzerland, Ospedale San Giovanni, Bellinzona, Ticino, Switzerland

Haematologica 2017 Volume 102(5):895-902

ABSTRACT

I

n the phase 3 LYM-3002 study comparing intravenous VR-CAP with R-CHOP in patients with newly-diagnosed, measurable stage II-IV mantle cell lymphoma, not considered or ineligible for transplant, the median progression-free survival was significantly improved with VRCAP (24.7 versus 14.4 months with R-CHOP; P<0.001). This post-hoc analysis evaluated the association between the improved outcomes and quality of responses achieved with VR-CAP versus R-CHOP in LYM3002. Patients were randomized to six to eight 21-day cycles of VR-CAP or R-CHOP. Outcomes included progression-free survival, duration of response (both assessed by an independent review committee), and time to next anti-lymphoma treatment, evaluated by response (complete response/unconfirmed complete response and partial response), MIPI risk status, and maximum reduction of lymph-node measurements expressed as the sum of the product of the diameters. Within each response category, the median progression-free survival was longer for patients given VR-CAP than for those given R-CHOP (complete response/unconfirmed complete response: 40.9 versus 19.8 months; partial response: 17.1 versus 11.7 months, respectively); similarly, the median time to next anti-lymphoma treatment was longer among the patients given VR-CAP than among those treated with R-CHOP (complete response/unconfirmed complete response: not evaluable versus 26.6 months; partial response: 35.3 versus 24.3 months). Within the complete/unconfirmed complete and partial response categories, improvements in progression-free survival, duration of response and time to next anti-lymphoma treatment were more pronounced in patients with lowand intermediate-risk MIPI treated with VR-CAP than with R-CHOP. In each response category, more VR-CAP than R-CHOP patients had a sum of the product of the diameters nadir of 0 during serial radiological assessments. Results of this post-hoc analysis suggest a greater duration and quality of response in patients treated with VR-CAP in comparison with those treated with R-CHOP, with the improvements being more evident in patients with low- and intermediate-risk MIPI. LYM-3002 ClinicalTrials.gov: NCT00722137. haematologica | 2017; 102(5)

Correspondence: gregor.verhoef@uzleuven.be

Received: August 25, 2016. Accepted: February 7, 2017. Pre-published: April 14, 2017. doi:10.3324/haematol.2016.152496 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/895 Š2017 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.

895


G. Verhoef et al.

Introduction

Methods

Mantle cell lymphoma (MCL) is an aggressive and generally incurable form of non-Hodgkin lymphoma that is responsible for approximately 4.3–5% of new cases of nonHodgkin lymphoma.1-3 Prognosis is poor for patients with MCL, with 5-year relative survival rates of about 30%.2 Rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) represents a standard of care for the treatment of newly-diagnosed patients with MCL who are considered ineligible for intensive therapy and stem cell transplantation.4 Complete response/unconfirmed complete response (CR/CRu) rates of up to 48% have been demonstrated with R-CHOP in this population of patients.5 However, progression-free survival (PFS) in the study achieving this result was limited (median 16.6 months).6 VR-CAP (bortezomib, rituximab, cyclophosphamide, doxorubicin, and prednisone), which substitutes bortezomib for vincristine in R-CHOP, has been approved by regulatory agencies in the USA, European Union and Japan for the treatment of MCL.7-9 This recommendation is based on findings from the randomized, phase 3 LYM-3002 study (NCT00722137), which evaluated the efficacy and safety of VR-CAP compared with R-CHOP in patients with newly diagnosed, measurable stage II, III or IV MCL who were ineligible or not considered for stem cell transplantation. Results from the primary analysis of LYM-3002 demonstrated a 59% improvement in median independent radiology review committee-assessed PFS with VR-CAP compared with R-CHOP [24.7 versus 14.4 months, respectively; hazard ratio (HR) 0.63; 95% confidence interval (CI): 0.50–0.79; P<0.001], and a 96% improvement in investigator-assessed PFS (median 30.7 versus 16.1 months, respectively; HR 0.51; 95% CI: 0.41–0.65; P<0.001).10 Significant and clinically important improvements in secondary efficacy endpoints were also demonstrated with VR-CAP compared with R-CHOP, including higher CR/CRu rate (53% versus 42%; P=0.007), longer median time to next anti-lymphoma therapy (TTNT; 44.5 versus 24.8 months; P<0.001; HR 0.50), longer median duration of CR/CRu (42.1 versus 18.0 months), and longer median duration of overall response (DOR; 36.5 versus 15.1 months). Overall response rates were high and were similar with the two regimens (92% versus 89% for VR-CAP versus R-CHOP, respectively). Patients receiving VR-CAP experienced higher rates of toxic effects than those receiving R-CHOP, including increased frequencies of grade ≥3 thrombocytopenia (57% versus 6%), neutropenia (85% versus 67%), leukopenia (44% versus 29%), lymphocytopenia (28% versus 9%), and infections/infestations (21% versus 14%). However, there were no significant effects on the number of completed cycles, median dose intensity for drugs common to both regimens, or rates of discontinuations or deaths related to adverse events.10 These results suggest that the difference between the two regimens in terms of response quality (i.e. DOR, PFS, and TTNT) was more pronounced than the difference in response rates. We hypothesized that the improved outcomes seen with VR-CAP were more closely related to response quality (duration, depth) than simply CR/CRu rates. We therefore conducted a post-hoc analysis of the LYM-3002 study10 to evaluate the association between response, response quality and outcomes for VR-CAP compared with R-CHOP.

The details of the design of the LYM-3002 study have been described previously;10 a brief overview is given in the Online Supplementary Material. The trial protocol was approved by local

896

A

B

C

Figure 1. Time-to-event outcomes for VR-CAP and R-CHOP by response category. (A) PFS assessed by the independent radiology review committee, (B) TTNT, and (C) DOR.

haematologica | 2017; 102(5)


LYM-3002 outcomes by response

Table 1. Patients’ demographics and baseline characteristics.

Median age, years (range) Male, n (%) Race, n (%) White Asian Black/African American Other Disease stage at diagnosis,* n (%) II / III / IV ECOG performance status, n (%)† 0/1/2 IPI score (risk category), n (%) 0-1 (low) 2 (low–intermediate) 3 (high–intermediate) 4-5 (high) MIPI risk status, n (%)‡ Low / intermediate / high MIPIb risk status, n (%)‡ Low / intermediate / high Ki-67 status, n (%)¶ Positive / negative Elevated LDH, n (%) Bone marrow involvement, n (%)

VR-CAP (n=243)

R-CHOP (n=244)

65 (26–88) 178 (73)

66 (34–82) 182 (75)

151 (62) 88 (36) 3 (1) 1 (<1)

172 (71) 68 (28) 0 4 (2)

12 (5) / 49 (20) / 182 (75)

16 (7) / 42 (17) / 186 (76)

111 (46) / 101 (42) / 31 (13)

85 (35) / 127 (52) / 31 (13)

38 (16) 75 (31) 84 (35) 46 (19)

38 (16) 71 (29) 88 (36) 47 (19)

76 (31) / 96 (40) / 71 (29)

70 (29) / 93 (38) / 80 (33)

23 (14) / 74 (45) / 66 (40)

23 (14) / 72 (44) / 69 (42)

84 (52) / 79 (48) 88 (36) 165 (68)

82 (50) / 82 (50) 86 (35) 171 (70)

Also published in part by Robak et al. 201510 *American Joint Committee on Cancer NHL disease staging system; †Data missing for one patient in the R-CHOP arm (n=243 VRCAP; n=243 R-CHOP); ‡Assessed in Ki-67-evaluable patients (n=163 VR-CAP; n=164 R-CHOP); ¶Based on a cut-off of 10% Ki-67 expression on an ordinal scale. ECOG: Eastern Cooperative Oncology Group; IPI: International Prognostic Index; LDH: lactate dehydrogenase; MIPI: Mantle cell lymphoma International Prognostic Index; MIPIb: MIPI with biologic component. R-CHOP: rituximab plus cyclophosphamide, doxorubicin, vincristine and prednisone;VR-CAP: bortezomib plus rituximab, cyclophosphamide, doxorubicin and prednisone.

ethics committees/institutional review boards. All patients provided written informed consent prior to the commencement of the study.

Procedures Computed tomography (CT) scans were performed every 6 weeks during treatment, and every 6–8 weeks during follow-up, until disease progression, study discontinuation, initiation of alternative therapy, or death. By central radiology acquisition guidelines, slide thicknesses for helical and conventional CT scans were 5 mm and 7–10 mm for the chest, abdomen, and pelvis, and 3 mm and 5 mm for the neck, respectively. All CT results were assessed by a blinded independent radiology review committee and by investigators, using modified International Workshop to Standardize Response Criteria (IWRC) for Non-Hodgkin Lymphoma.11 For each patient, up to ten measurable sites of disease at baseline (clearly measurable in 2 perpendicular dimensions, >1.5 cm in the long axis, >1.0 cm in the short axis) were tracked. At each follow-up, the sum of the product of the diameters (SPD) was calculated (i.e., the sum of the long axis and short axis of all measureable sites). Lesions that became smaller than 5 mm x 5 mm were recorded as ‘too small to measure’, and given a default value of 0 mm x 0 mm (SPD nadir of 0). All other sites of disease were considered assessable (including objective evidence of disease identified by radiological imaging, physical examination, or other procedures as necessary) but were not measurable. Adverse events were graded using the National Cancer Institute’s Common Terminology Criteria for Adverse Events (NCI-CTCAE) version 3.0. haematologica | 2017; 102(5)

Outcomes Endpoints assessed in this post-hoc analysis were PFS, DOR, and TTNT. Outcomes were stratified by response category [CR/CRu and partial response (PR); defined in the Online Supplementary Material and MCL International Prognostic Index (MIPI) risk status.12,13 Overall response rate was defined as CR plus CRu plus PR. Depth of response was evaluated by measuring the maximum reduction in measurable lymph nodes from baseline, as assessed by CT, expressed as the SPD, and provided by the independent radiology review committee. Of note, achievement of an SPD nadir of 0 was not a prerequisite for achievement of CR or CRu, thus patients could meet the criteria for CR/CRu yet not achieve an SPD of 0. Equally, a patient with an SPD of 0 who did not otherwise meet the criteria for CR/CRu, or those with CR/CRu who did not have bone marrow/lactate dehydrogenase confirmation, would be classified as having a PR. An additional analysis was conducted investigating CR and CRu as individual categories. For this post-hoc analysis, response, DOR, and PFS were based on assessment by the independent review committee. At the time of analysis, overall survival data were immature (the median overall survival in the VR-CAP arm was not estimable) and thus are not included.

Statistical analysis All efficacy analyses were performed on the intention-to-treat population, except for response endpoints (analyzed in the response-evaluable population). Time-to-event distributions were 897


G. Verhoef et al. Table 2. Patients achieving a sum of the product of the diameters nadir of 0 (lesions absent or ‘too small to measure’ by computed tomograpy scan).

Patients, n/N (%)* CR/CRu CR CRu PR

VR-CAP

R-CHOP

SPD nadir 0

SPD nadir 0+

SPD nadir 0

SPD nadir 0+

87/121 (72) 83/105 (79) 4/16 (25) 42/87 (48)

34/121 (28) 22/105 (21) 12/16 (75) 45/87 (52)

54/92 (59) 52/76 (68) 2/16 (12) 30/108 (28)

38/92 (41) 24/76 (32) 14/16 (88) 78/108 (72)

*Number of lesions measured per patient: mean: 4.8, standard deviation: 2.86, median: 4, range: 1–10. Response evaluable population: VR-CAP: n=229; R-CHOP: n=229. SPD nadir calculated based on up to ten measureable lesions identified at baseline. SPD 0: defined as SPD nadir of 0 for all measureable lesions. SPD 0+: defined as SPD nadir >0. CR, complete response; CRu, unconfirmed complete response; PR, partial response. R-CHOP, rituximab plus cyclophosphamide, doxorubicin, vincristine and prednisone; SPD, sum of the product of the diameters; VR-CAP, bortezomib plus rituximab, cyclophosphamide, doxorubicin and prednisone.

estimated using Kaplan-Meier methodology, with log-rank tests and Cox models (alpha=0.05, 2-sided) used for inter-arm comparisons of time-to-event endpoints. A stratified Cochran-MantelHaenszel chi-square test with International Prognostic Index and disease stage as stratification factors was used to assess betweengroup differences.

A

Results Patients Between May 2008 and December 2011, 487 patients (243 treated with VR-CAP; 244 treated with R-CHOP) from 128 centers in 28 countries were randomized into LYM-3002, with 457 patients (229 VR-CAP; 228 R-CHOP) evaluable for response. In the overall population, the demographic and baseline characteristics of the patients were generally well balanced between the two arms (Table 1).

B

Outcomes stratified by response category When time-to-event outcomes were stratified according to response category (CR/CRu and PR), PFS was longer with VR-CAP than with R-CHOP in patients achieving CR/CRu (median 40.9 versus 19.8 months, respectively) and in those achieving PR (median 17.1 versus 11.7 months, respectively) (Figure 1A). Similarly, TTNT was longer with VR-CAP versus R-CHOP in patients achieving CR/CRu (median not evaluable versus 26.6 months, respectively) and PR (median 35.3 versus 24.3 months, respectively) (Figure 1B). DOR was also prolonged with VR-CAP compared with R-CHOP in patients achieving CR/CRu (42.1 versus 18.5 months, respectively) and PR (20.2 versus 9.6 months, respectively) (Figure 1C). Across all time-to-event outcomes, outcomes appeared similar between patients receiving VR-CAP who achieved PR and those seen in patients receiving R-CHOP who achieved CR/CRu. Median times to first evidence of ≥PR (VR-CAP: 42 days, R-CHOP: 47 days) in patients who achieved a response, and median times to CR/CRu (VR-CAP: 82 days, R-CHOP: 84 days) in patients who achieved CR/CRu appeared similar in the VR-CAP and R-CHOP arms. Within the CR/CRu and PR response categories, prolongation of time-to-event outcomes – PFS (Figure 2), TTNT (Online Supplementary Figure S1) and DOR (Online Supplementary Figure S2) – with VR-CAP compared with RCHOP were more pronounced in low- and intermediaterisk MIPI patients. In contrast, the treatment effect was less apparent in patients with high-risk MIPI scores. An analysis of response in which CR, CRu and PR were 898

C

Figure 2. Progression-free survival according to MIPI risk status. (A) High-risk patients, (B) intermediate-risk patients, (C) low-risk patients.

haematologica | 2017; 102(5)


LYM-3002 outcomes by response

considered separately was also performed. This analysis indicated that results in patients achieving CRu were closer to those reported in patients achieving PR than to those in patients achieving CR (Figure 3A). Excluding patients who achieved CRu from the combined CR/CRu category resulted in a longer median DOR (CR: 48.6 versus 23.1 months for VR-CAP and R-CHOP patients, respectively). When considering CRu and PR patients together, their median DOR was shorter than that of PR patients alone (CRu/PR: 18.8 versus 9.5 months for VR-CAP and R-CHOP patients, respectively) (Figure 3B).

Depth of response: change in nadir of the sum of the product of the diameters To further elucidate factors within each response category that could contribute to the ‘quality of response’ potentially driving differences in long-term outcomes between the VR-CAP and R-CHOP treatment groups, we investigated the maximum reduction, measured by CT scan, of the size of the measurable lymph node lesions chosen at baseline. Based on the default value for lesions recorded as ‘too small to measure’ being set as 0 mm x 0 mm, patients in whom all measurable lesions became ‘too small to measure’ were also referred to as having a SPD nadir of 0. Within each response category (CR, CRu and PR), the reduction in lymph node measurements was more pronounced in patients who received VR-CAP compared with that in patients who received R-CHOP (Table 2, Figure 4). The percentage of patients with CR in whom all measurable baseline lesions became ‘too small to measure’ was 79% for VR-CAP and 68% for R-CHOP, with similar observations made for the CR/CRu category (72% and 59%, respectively). Among the patients who achieved a PR, all measurable baseline lesions became ‘too small to measure’ in 48% of those treated with VR-CAP and 28% of those treated with R-CHOP. Patients with a CR in whom all measurable lesions became ‘too small to measure’ had a longer PFS than those in whom at least one measurable lesion could still be determined (e.g. lymph nodes had regressed to normal size but were still measurable, while all other criteria for CR were met) (Figure 5A). Similar patterns were observed for TTNT and DOR (data not shown). Patients with a PR in whom all measurable lesions became ‘too small to measure’ had a longer PFS than those who had residual measurable lesions (Figure 5B). This observation was equally present in both treatment groups. Similar patterns were observed for TTNT and DOR (data not shown). We also analyzed the achievement of an SPD nadir of 0 according to MIPI risk status (Online Supplementary Table S1). In patients with low- and intermediate-risk MIPI receiving VR-CAP, 76% and 77% of patients with CR/CRu, respectively, had an SPD nadir of 0; in those receiving R-CHOP, this was 65% and 68%, respectively. In patients with high-risk MIPI, this effect was less evident, with only 47% of CR/CRu patients treated with VRCAP achieving an SPD nadir of 0 versus 41% of CR/CRu patients who were treated with R-CHOP. For patients who achieved PR, there were very few individuals with an SPD nadir of 0 in each MIPI group. For comparison, response rates across all patients with high-risk MIPI were similar between the VR-CAP and R-CHOP groups (overall response rate: 81% versus 83%; CR: 24% versus 24%; haematologica | 2017; 102(5)

CR+CRu: 34% versus 32%; PR: 47% versus 51%) but the depth of response appeared greater with VR-CAP (patients with SPD=0, CR: 64% versus 53% with RCHOP; CR/CRu 47% versus 38%).

Discussion In the LYM3002 study, despite similar overall response rates, VR-CAP resulted in improved outcomes (PFS, TTNT, DOR) compared with R-CHOP in newly-diagnosed patients with MCL.10 The findings of our post-hoc analysis suggest that this may be driven by a longer DOR and more profound disease elimination in those patients who received VR-CAP than in those who received RCHOP, both overall and when stratified by response category. Within each response category, PFS, TTNT, and DOR were longer in patients receiving VR-CAP than in

A

B

Figure 3. Duration of response according to type of response. (A) CR, CRu or PR, (B) CR versus CRu/PR.

899


G. Verhoef et al.

those receiving R-CHOP, and a greater proportion of patients became lesion-negative (SPD nadir of 0 for all measurable lesions), indicating that responses observed with VR-CAP were both longer and deeper than those observed with R-CHOP. However, these results should be interpreted with caution, because of the small sample sizes (16 patients per arm), the observation that CRu patients had a higher percentage of SPD nadirs of 0 and a longer DOR could be due to random variations. Of particular interest was the finding that results across all time-toevent outcomes in patients receiving VR-CAP who achieved PR were comparable to those seen in patients receiving R-CHOP who achieved CR/CRu. Our analysis of response depth (based on the diameters of the lymph nodes chosen to be followed as measurable lesions by the independent radiology review committee and for which serial radiological measurements were provided) indicates that VR-CAP produced a deeper response than R-CHOP. In patients with PR, VR-CAP resulted in a higher rate of patients with an SPD of 0 compared with RCHOP. Achievement of an SPD nadir of 0 suggests a better

clearance of residual disease from the lymph nodes – a benefit that is likely to contribute to the observed longer remission duration – and was associated with improved PFS (Figure 5). Thus, although outcomes for lesion-negative PR patients were similar regardless of treatment (as were outcomes for lesion-positive PR patients) the longer PFS with VR-CAP versus R-CHOP in patients with PR is likely being driven by the higher percentage of lesion-negative patients. Although it appears counter-intuitive to have lesion-negative patients categorized as only partial responders, we highlight that the PR grouping here included patients who did not otherwise meet the criteria for CR/CRu. However, within the CR category, although being lesion-negative was again prognostic for improved outcomes versus lesion-positive CR, and although outcomes were similar in lesion-positive CR patients regardless of treatment, outcomes in lesion-negative CR patients were further enhanced in the VR-CAP group versus the RCHOP group, a finding which suggests there may be other factors at play following treatment with VR-CAP, which could not be quantified here (as discussed further below).

A

B

C

Figure 4. Percentage change in sum of the product of the diameters nadir from baseline by response category. The default value for lesions ‘too small to measure’ was 0 mm x 0 mm. (A) Patients with CR, (B) patients with CR/CRu, (C) patients with PR.

900

haematologica | 2017; 102(5)


LYM-3002 outcomes by response

Recalling that achievement of an SPD nadir of 0 was not a prerequisite for CR, these data introduce the concept that not all CR, by standard criteria/CT scanning, are equal. This suggests the need to use additional methodologies beyond radiology to further characterize the depth of complete remissions; however these methodologies were not used in this study. One such example is positron emission tomography CT scanning, which has improved the accuracy of disease staging in lymphoma over CT alone and is now considered the standard for response assessment in most lymphomas.14,15 When the LYM-3002 study was initiated, positron emission tomography CT was not a widely available technique and could not easily be applied to global studies, but may have provided improved methods of characterizing depth of response. Another example is determination of minimal residual disease. In an analysis by Pott et al. of two studies involving transplant-eligible and -ineligible patients with previously untreated MCL, molecular remission, defined as minimal residual disease negativity by quantitative polymerase chain reaction (sensitivity: 10â&#x20AC;&#x201C;5) in both peripheral blood and bone marrow, was found to be an independent predictor of clinical outcome after combined immunochemotherapy.16 In the analysis by Pott et al., most patients received R-CHOP induction; other patients received rituximab, fludarabine and cyclophosphamide or alternating R-CHOP/R-DHAP (rituximab, high-dose cytarabine and cisplatin) induction. In contrast, Howard et al. found that the attainment of a molecular remission to R-CHOP was not predictive of PFS in previously untreated patients with MCL,6 noting that the criteria for molecular remission were slightly looser (disappearance of polymerase chain reaction-detectable disease in the peripheral blood or bone marrow). These data, in conjunction with our analysis, indicate the need for further refinement of characterization of response and the importance of incorporating new techniques into studies of this nature. Since the initiation of the LYM-3002 study, another notable update to the response criteria has been the removal of the CRu category and its replacement in the IWRC11 and in the Lugano classification criteria with updated definitions of CR and PR.14 In acknowledgment of this update, we conducted an additional analysis investigating the DOR in which achievement of CR and CRu were considered separately. Long-term outcomes differed substantially when considering achievement of combined CR/CRu versus achievement of CR alone (Figure 3); thus, our work supports the elimination of CRu as a separate response category. Based on our analysis, grouping these patients under the PR category appeared more appropriate and may have more clinical relevance. This observation is of particular relevance for early clinical phase trials, in which response is often the primary endpoint, designed to characterize treatment effect of novel therapeutic approaches. In the overall LYM-3002 analysis, the improvement in PFS with VR-CAP versus R-CHOP was seen across each MIPI risk grouping, although in high-risk patients this was not statistically significant.10 In the present analysis, the beneficial effect of VR-CAP in prolonging outcomes within each response category was predominantly seen in lowand intermediate-risk MIPI patients, and was less striking in the high-risk group. This may be driven by the greater proportion of CR/CRu patients receiving VR-CAP across the MIPI groupings who became lesion-negative by SPD haematologica | 2017; 102(5)

assessment. It would be interesting to further investigate the biological drivers behind this observation â&#x20AC;&#x201C; what are the differential characteristics of MIPI high-risk disease that result in lower rates of lesion-negative CR and that obviate the improved outcomes seen with VR-CAP versus R-CHOP in these patients in other MIPI categories? For example, is it possible that patients with high-risk MIPI status have more rapidly proliferative disease, and thus the deeper response achieved with VR-CAP versus RCHOP does not translate into substantially improved outcomes due to rapid disease return? Such findings support the suggestion of Dreyling that future studies should address the goal of personalized medicine.17

A

B

Figure 5. Progression-free survival assessed by the independent radiology review committee. PFS according to reduction in baseline lesions expressed as SPD in patients with (A) CR and (B) PR.

901


G. Verhoef et al.

Since commencement of our study, maintenance treatment with rituximab has become the standard of care for the treatment of newly-diagnosed, elderly patients with MCL,4,17 and conceivably the use of VR-CAP with rituximab maintenance therapy could further prolong PFS. In addition, newer regimens showing enhanced efficacy relative to RCHOP have since been introduced;18,19 a further study comparing VR-CAP with these combinations may be warranted, and in reference to the availability of newer and more sensitive methods of detecting disease elimination, should include analysis of depth of response, relationship of response categories and long-term outcomes. Tailoring therapies to different ‘types’ or stages of disease (e.g. patients with high-risk MIPI) could perhaps also be considered. In conclusion, this post-hoc analysis of the association between the improved outcomes and quality of responses achieved with VR-CAP versus R-CHOP in the LYM-3002 study suggests a longer duration and better quality of response in VR-CAP versus R-CHOP patients, which was more evident in patients with low- and intermediate-risk MIPI. In addition, the results suggest that the extent of disease elimination, rather than the category of achieved response, may be a better predictor of treatment outcomes

References 1. Howlader, N, Noone, A., Krapcho and others. SEER Cancer Statistics Review, 19752012, National Cancer Institute, Bethesda, MD. Based on November 2014 SEER data submission, posted to the SEER website, April 2015. Available at: http://seer.cancer.gov/csr/1975_2012/ (Accessed July 2015). 2. Smith A, Crouch S, Lax S, et al. Lymphoma incidence, survival and prevalence 20042014: sub-type analyses from the UK's Haematological Malignancy Research Network. Br J Cancer. 2015;112(9):15751584. 3. Zhou Y, Wang H, Fang W, et al. Incidence trends of mantle cell lymphoma in the United States between 1992 and 2004. Cancer. 2008;113(4):791-798. 4. Dreyling M, Thieblemont C, Gallamini A, et al. ESMO Consensus Conferences: guidelines on malignant lymphoma. part 2: marginal zone lymphoma, mantle cell lymphoma, peripheral T-cell lymphoma. Ann Oncol. 2013;24(4):857-877. 5. McKay P, Leach M, Jackson R, Cook G, Rule S. Guidelines for the investigation and management of mantle cell lymphoma. Br J Haematol. 2012;159(4):405-426. 6. Howard OM, Gribben JG, Neuberg DS, et al. Rituximab and CHOP induction therapy for newly diagnosed mantle-cell lymphoma: molecular complete responses are not predictive of progression-free survival. J Clin Oncol. 2002;20(5):1288-1294.

902

in patients with newly diagnosed MCL. In an era of an increasing number of regimens being developed in MCL, additional approaches should therefore be incorporated in clinical trial designs to further define the quality of responses. Acknowledgments The authors would like to acknowledge Olga Samoilova for her contribution to collecting data for the LYM-3002 study. The authors would also like to acknowledge Catherine Crookes of FireKite, an Ashfield company, part of UDG Healthcare plc, for writing assistance during the development of this manuscript, which was funded by Millennium Pharmaceuticals, Inc. and Janssen Global Services, LLC. Prior presentation Some of the data published here were presented in poster form at the 13th International Conference on Malignant Lymphoma, Lugano, Switzerland, June 17–20, 2015: Verhoef G, Robak T, Huang H, et al. Association between quality of response and outcomes in patients with newly diagnosed mantle cell lymphoma receiving VR-CAP versus R-CHOP in the phase 3 LYM-3002 study.

7. European Medicines Agency. European Medicines Agency. VELCADE® (bortezomib). Summary of Product Characteristics (updated February 2014). Available at: http:// www.ema.europa.eu/docs/ e n _ G B / d o c u m e n t _ l i b r a r y / E PA R _ _Product_Information/human/000539/WC5 00048471.pdf (Accessed July 2015). 2014. 8. Millennium Pharmaceuticals Inc. VELCADE® (bortezomib) for injection: for subcutaneous or intravenous use. Full prescribing information. October 2014, Revision 17. Available at: http://www.velcade.com/files/ PDFs/VELCADE_PRESCRIBING_INFORMATION.pdf (Accessed July 2015). 9. Japanese Pharmaceuticals and Medical Devices Agency. List of approved products FY 2015. https://www.pmda.go.jp/ files/000208993.pdf. 10. Robak T, Huang H, Jin J, et al. Bortezomibbased therapy for newly diagnosed mantlecell lymphoma. N Engl J Med. 2015; 372(10):944-953. 11. Cheson BD, Pfistner B, Juweid ME, et al. Revised response criteria for malignant lymphoma. J Clin Oncol. 2007;25(5):579-586. 12. Hoster E, Dreyling M, Klapper W, et al. A new prognostic index (MIPI) for patients with advanced-stage mantle cell lymphoma. Blood. 2008;111(2):558-565. 13. Hoster E, Klapper W, Hermine O, et al. Confirmation of the mantle-cell lymphoma International Prognostic Index in randomized trials of the European Mantle-Cell Lymphoma Network. J Clin Oncol. 2014;32(13):1338-1346. 14. Cheson BD, Fisher RI, Barrington SF, et al.

15.

16.

17.

18.

19.

Recommendations for initial evaluation, staging, and response assessment of Hodgkin and non-Hodgkin lymphoma: the Lugano classification. J Clin Oncol. 2014; 32(27):3059-3068. Juweid ME, Wiseman GA, Vose JM, et al. Response assessment of aggressive nonHodgkin's lymphoma by integrated International Workshop Criteria and fluorine-18-fluorodeoxyglucose positron emission tomography. J Clin Oncol. 2005;23(21):4652-4661. Pott C, Hoster E, Delfau-Larue MH, et al. Molecular remission is an independent predictor of clinical outcome in patients with mantle cell lymphoma after combined immunochemotherapy: a European MCL intergroup study. Blood. 2010; 115(16):32153223. Dreyling M. Haematological cancer: bortezomib in MCL--new standard of care or just another option? Nat Rev Clin Oncol. 2015;12(7):376-378. Frosch Z, Luskin MR, Landsburg DJ, et al. RCHOP or R-HyperCVAD with or without autologous stem cell transplantation for older patients with mantle cell lymphoma. Clin Lymphoma Myeloma Leuk. 2015;15 (2):92-97. Rummel MJ, Niederle N, Maschmeyer G, et al. Bendamustine plus rituximab versus CHOP plus rituximab as first-line treatment for patients with indolent and mantle-cell lymphomas: an open-label, multicentre, randomised, phase 3 non-inferiority trial. Lancet. 2013;381(9873):1203-1210.

haematologica | 2017; 102(5)


ARTICLE

Non-Hodgkin Lymphoma

Safety and efficacy of abexinostat, a pan-histone deacetylase inhibitor, in non-Hodgkin lymphoma and chronic lymphocytic leukemia: results of a phase II study

Vincent Ribrag,1 Won Seog Kim,2 Reda Bouabdallah,3 Soon Thye Lim,4 Bertrand Coiffier,5 Arpad Illes,6 Bernard Lemieux,7 Martin J. S. Dyer,8 Fritz Offner,9 Zakia Felloussi,10 Ioana Kloos,10 Ying Luan,11 Remus Vezan,11 Thorsten Graef,11 and Franck Morschhauser12

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(5):903-909

1 Institut de Cancérologie Gustave Roussy, Villejuif, France; 2Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea; 3CLCC Institut Paoli Calmettes, Marseille, France; 4National Cancer Center Singapore, Duke-National University of Singapore Medical School, Singapore; 5Hospices Civils de Lyon, France; 6 Department of Hematology, Faculty of Medicine, University of Debrecen Medical and Health Science Center, Hungary; 7CHUM, Montreal, QC, Canada; 8Ernest and Helen Scott Haematological Research Institute, University of Leicester, UK; 9Universitair Ziekenhuis Gent, Gent, Belgium; 10Servier, Paris, France; 11Pharmacyclics LLC, an AbbVie Company, Sunnyvale, CA, USA and 12Department of Hematology, Université de Lille, EA GRIIOT, France

ABSTRACT

H

istone deacetylase inhibitors are members of a class of epigenetic drugs that have proven activity in T-cell malignancies, but little is known about their efficacy in B-cell lymphomas. Abexinostat is an orally available hydroxamate-containing histone deacetylase inhibitor that differs from approved inhibitors; its unique pharmacokinetic profile and oral dosing schedule, twice daily four hours apart, allows for continuous exposure at concentrations required to efficiently kill tumor cells. In this phase II study, patients with relapsed/refractory non-Hodgkin lymphoma or chronic lymphocytic leukemia received oral abexinostat at 80 mg BID for 14 days of a 21-day cycle and continued until progressive disease or unacceptable toxicity. A total of 100 patients with B-cell malignancies and T-cell lymphomas were enrolled between October 2011 and July 2014. All patients received at least one dose of study drug. Primary reasons for discontinuation included progressive disease (56%) and adverse events (25%). Grade 3 or over adverse events and any serious adverse events were reported in 88% and 73% of patients, respectively. The most frequently reported grade 3 or over treatment-emergent related adverse events were thrombocytopenia (80%), neutropenia (27%), and anemia (12%). Among the 87 patients evaluable for efficacy, overall response rate was 28% (complete response 5%), with highest responses observed in patients with follicular lymphoma (overall response rate 56%), T-cell lymphoma (overall response rate 40%), and diffuse large Bcell lymphoma (overall response rate 31%). Further investigation of the safety and efficacy of abexinostat in follicular lymphoma, T-cell lymphoma, and diffuse large B-cell lymphoma implementing a less doseintense week-on-week-off schedule is warranted. (Trial registered at: EudraCT-2009-013691-47) Introduction Histone deacetylases (HDAC) and histone acetyl transferases control acetylation of histone and non-histone proteins, thereby regulating gene transcription, protein function, and protein stability. The aberrant expression of HDAC proteins and haematologica | 2017; 102(5)

Correspondence: vincent.ribrag@gustaveroussy.fr

Received: August 22, 2016. Accepted: January 13, 2017. Pre-published: January 25, 2017. doi:10.3324/haematol.2016.154377 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/903 ©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.

903


V. Ribrag et al. Table 1. Patients' baseline characteristics.

Characteristic

FL (n=18)

Age, years 60 Median (range) (32–83) Sex, n (%) Female 9 (50) Median BSA, m2 1.9 (range) (1.3–2.3) Baseline ECOG performance status, n (%) 0 10 (56) 1 8 (44) 2 0 (0) Ann Arbor stage, n (%) I 1 (6) II 8 (44) III 2 (11) IV 7 (39) Prior treatments Median 4.5 (range) (1–11) Radiotherapy, n (%) 4 (22) Median time from diagnosis, months 59.5 (range) (16–156)

DLBCL (n=17)

MCL (n=16)

T-CL* (n=18)

MZL/ Other (n=15)

CLL/ LL (n=16)

All (n=100)

63 (38–85)

69 (59–82)

61 (47–78)

69 (52–78)

71 (46–80)

67 (32–85)

10 (59) 1.7 (1.4–2.4)

6 (38) 1.7 (1.4–2.1)

9 (50) 1.8 (1.5–2.1)

4 (27) 1.8 (1.5–2.1)

7 (44) 1.9 (1.4–2.1)

45 (45) 1.8 (1.3–2.4)

5 (29) 11 (65) 1 (6)

7 (44) 9 (56) 0 (0)

9 (50) 9 (50) 0 (0)

6 (40) 9 (60) 0 (0)

9 (56) 7 (44) 0 (0)

46 (46) 53 (53) 1 (1)

1 (6) 5 (29) 5 (29) 6 (35)

2 (13) 1 (6) 1 (6) 12 (75)

0 (0) 1 (6) 4 (22) 13 (72)

0 (0)† 1 (7)† 4 (27)† 5 (33)†

NA NA NA NA

4 (4) 16 (16) 16 (16) 43 (43)

3 (1–10) 8 (47) 18 (5–121)

3 (1–8) 5 (31) 38 (13–108)

2 (1–5) 3 (17) 23 (3–99)

4 (1–7) 2 (13) 90 (18–252)

2.5 (1–6) 0 (0) 69 (19–113)

3 (1–11) 22 (22) 47 (3–252)

BSA: body surface area; CLL: chronic lymphocytic leukemia; DLBCL: diffuse large B-cell lymphoma; ECOG: Eastern Cooperative Oncology Group; FL: follicular lymphoma; LL: lymphoblastic lymphoma; MCL: mantle cell lymphoma; MZL: marginal zone lymphoma; T-CL: T-cell lymphoma. *Of 18 patients with T-CL treated, 8 presented with angioimmunoblastic T-cell lymphoma, 6 with peripheral T-cell lymphoma not otherwise characterized, and 4 with cutaneous T-cell lymphoma-mycosis fungoides/Sezary syndrome. †Data not available for 5 patients.

mutation in the genes encoding HDACs have been linked to the development of cancers.1-3 The use of HDAC inhibitors causes hyperacetylation of histone and non-histone proteins, leading to the transcriptional activation of tumor suppressor genes, as well as genes involved in cell cycle control, cell division, and apoptosis, resulting in antitumor activity.4 More recently, HDAC inhibitors have also been shown to exert immunostimulatory effects on cancer cells. Exposure to HDAC inhibitors was found to up-regulate natural killer cell-activating ligands, major histocompatibility class I and II molecules, proteins involved in antigen presentation, and costimulatory molecules; in colon cancer cells, treatment with an HDAC inhibitor induced immunogenic cell death.5 In addition, HDAC inhibitors can block tumor angiogenesis and indirectly damage DNA by reducing RAD51 protein and inhibiting DNA repair.4 Currently, 4 HDAC inhibitors, vorinostat, romidepsin, belinostat, and panobinostat are approved for the treatment of cutaneous T-cell lymphoma, peripheral T-cell lymphoma, or multiple myeloma based on their potent and specific activity.6-10 These agents, along with other HDAC inhibitors under development, are showing promising results in B-cell malignancies such as non-Hodgkin lymphoma (NHL) and chronic lymphocytic leukemia (CLL),11,12 and solid tumors such as breast cancer.13 The tolerability profiles for different HDAC inhibitors vary. The most commonly reported adverse events include gastrointestinal effects (primarily diarrhea, nausea, and vomiting), constitutional effects (mainly fatigue), hematologic effects (thrombocytopenia and neutropenia), and cardiac toxicities (QTc prolongation); many of these toxicities were found to be dose-limiting.14 904

Abexinostat, a novel, orally available hydroxamate-containing pan-HDAC inhibitor, differs from approved HDAC inhibitors due to its unique pharmacokinetic profile. Abexinostat was rapidly absorbed following oral administration, and a maximal plasma concentration was achieved within 0.5-1 hour; the terminal elimination halflife was calculated to be approximately four hours.15 Pharmacokinetic modeling was used to determine that the oral dosing schedule for abexinostat, twice daily four hours apart, allows for continuous exposure at concentrations required for efficient tumor cell killing while lowering the peak concentrations associated with once-daily oral administration.16 Abexinostat may, therefore, offer a more active and potentially less toxic treatment option for cancer with a wider therapeutic index than other HDAC inhibitors under development. In a phase I/II study of 55 patients with relapsed or refractory lymphomas, abexinostat was administered on a 1-week-on / 1-week-off schedule, and the recommended phase II dose was identified as 45 mg/m2 BID (equivalent to 80 mg BID). At this dose, abexinostat was well-tolerated with significant clinical activity and highly durable responses, especially in patients with multiple relapse in follicular lymphoma (FL).17 With this dosing schedule, the incidence of grade 3 or over thrombocytopenia was 20%. Another concurrent phase I study investigated other dosing schedules.15 During phase I of our study, the safety and efficacy of abexinostat using 3 different dosing schedules in 35 patients with relapsed or refractory NHL or CLL were evaluated.15 The recommended phase II dose in this study was also 45 mg/m2 BID (equivalent to 80 mg BID) administered on a more dose-intensive 2-weeks-on / haematologica | 2017; 102(5)


Safety and efficacy of abexinostat in NHL and CLL

A

B

Figure 1. Thrombocytopenia with abexinostat treatment. (A) Platelet counts and (B) incidence of grade 3 or over thromoboctyopenia.

1-week-off schedule. The use of abexinostat was associated with a manageable toxicity profile and durable responses, and showed itself to be particularly active in patients with relapsed or refractory FL. Here, we report the safety and efficacy results of phase II of the same study.

Methods This phase II study enrolled patients with relapsed/refractory NHL or CLL aged 18 years or over with histologically confirmed NHL [including FL, T-cell lymphoma (T-CL), diffuse large B-cell lymphoma (DLBCL), mantle cell lymphoma (MCL), marginal zone lymphoma (MZL), or other subtypes] or immunophenotypically confirmed CLL. Patients with NHL other than T-CL had received 2 or more prior lines of standard therapy; patients with TCL and CLL had received 1 or more prior treatment. Other eligibility criteria included estimated life expectancy more than 12 weeks, Eastern Cooperative Oncology Group performance status 2 or under, adequate hematologic function (absolute neutrophil count >1×109/L, hemoglobin ≥8 g/dL, platelet count >100×109/L or 75×109/L in case of bone marrow involvement), adequate renal function [creatinine <1.5×upper limit of normal (ULN) or creatinine clearance ≥50 mL/min], adequate hepatic function (aspartate amino transferase and alanine aminotransferase <1.5×ULN, total bilirubin <1.5×ULN), and adequate cardiac functions with potassium levels within normal ranges with or without potassium correction. Exclusion criteria included allogeneic bone marrow transplant, major surgery within four weeks before the first abexinostat dose, corticosteroids within ten days or valproic acid within five days before the first abexinostat dose, prior treatment with HDAC inhibitor other than valproic acid, concurrent therapeutic anticoagulation by antivitamin K, immunotherapy, chemotherapy (nitrosoureas within 6 weeks), and radiotherapy within four weeks before the first abexinostat dose. haematologica | 2017; 102(5)

Table 2. Treatment-related adverse events by grade occurring in 5% or more of patients.

Adverse event, % Any related adverse event Hematologic toxicities Thrombocytopenia Neutropenia Anemia Non-hematologic toxicities Diarrhea Nausea Asthenia Decreased appetite Vomiting Fatigue Epistaxis Muscle spasms Weight loss

Any grade

Grades 3/4

98

82

87 34 20

80 27 12

42 33 27 23 17 11 7 6 5

3 2 5 0 0 0 0 0 0

Written informed consent was obtained from all patients before enrollment. The Institutional Review Board, Research Ethics Board, and Independent Ethics Committee at each site reviewed and approved the protocol prior to initiation of the study. Patients were given oral abexinostat at 80 mg BID four hours apart for 14 days of a 3-week cycle, and treatment was continued until progressive disease or unacceptable toxicity. The 80 mg BID dose, which corresponds to the recommended phase II dose of 45 mg/m2 BID, had been identified in phase I of the study.15 For patients who could not tolerate abexinostat, treatment was discontinued until resolution of the adverse event; treatment was resumed at a dose of 80 mg BID for five days per week for two weeks of a 3-week cycle in subsequent cycles. A further dose 905


Change in target lesion, %

V. Ribrag et al.

*Patient had a 1212% change in target lesion, which falls outside the scale of this figure.

reduction to 60 mg BID for five days per week for two weeks of a 3-week cycle was also permitted, if necessary. The primary end point was overall response rate (ORR). Secondary end points included duration of response, progressionfree survival (PFS), and overall survival. Patients were monitored for response at the end of every cycle. For patients with NHL, response was evaluated based on the recommendations of the International Working Group Revised Response Criteria for Malignant Lymphomas.18 Patients with CLL were evaluated for response according to the guidelines of the International Workshop on Chronic Lymphocytic Leukemia.19 Adverse events were recorded at each visit using the National Cancer Institute CTCAE v.3.0 criteria. This exploratory study enrolled approximately 16 patients per tumor type. For ORR, 95% exact (Clopper-Pearson) confidence intervals are provided. The Kaplan-Meier method was used for time-to-event analyses.

Results Patientsâ&#x20AC;&#x2122; baseline characteristics A total of 100 patients with NHL or CLL were enrolled in this phase II study between October 2011 and July 2014. Represented histologies included FL (n=18), DLBCL (n=17), MCL (n=16), T-CL (n=18), MZL and other subtypes (n=15), and CLL/LL (n=16). Patients' baseline characteristics are presented in Table 1. The median age of the population was 67 years (range 32-85 years). The median number of prior therapies was 3 (range 1-11).

Treatment All patients received at least one dose of study drug. The median duration of treatment was 2.8 months (range 0.735.4 months). Patients with FL continued abexinostat treatment longer than other NHL subtypes with a median 906

Figure 2. Change in lymph node size. Waterfall plot of maximum change from baseline of the sum of products of the greatest diameters during the treatment phase. CR: complete response; PR: partial response; SD: stable disease; PD: progressive disease.

treatment duration of 5.6 months. The most common primary reasons for withdrawal from the study were progressive disease (56%) and adverse events (25%).

Safety Treatment-emergent adverse events related to study drug were reported in 98% of patients, with 82% experiencing grade 3 or over events. The most frequently reported grade 3 or over treatment-related AEs were thrombocytopenia (80%), neutropenia (27%), and anemia (12%). The incidence of any grade diarrhea was 42%; however, only 3% of patients reported grade 3 or over diarrhea (Table 2). Serious AEs were reported in 73% of patients with 70% reporting grade 3 or over serious AEs. The most commonly reported serious AEs included thrombocytopenia (54%), neutropenia (11%), anemia (7%), and pneumonia (6%). There were 5 deaths on study, although the investigators of the study did not consider any of them to be related to treatment. Because this trial was conducted at a fixed abexinostat dose, we analyzed whether patients with lower body surface area (BSA), who would have higher drug exposure, experienced greater toxicities. As expected, grade 3 or over AEs were more common in patients with a lower BSA than the median (1.8 m2) when compared with those with BSA at or above the median (98% vs. 78%). Serious AEs were also more common in patients with BSA lower than the median compared with those with BSA at or above the median BSA (90% vs. 56%). Adverse events led to dose reductions and treatment discontinuations in 21% and 29% of patients, respectively. Thrombocytopenia was the most common toxicity that caused dose reductions, occurring in 17% of patients. Other adverse events leading to dose reductions included asthenia (2 patients) and neutropenia, decreased appetite, haematologica | 2017; 102(5)


Safety and efficacy of abexinostat in NHL and CLL

Figure 3. Progression-free survival (PFS) in follicular lymphoma (FL), diffuse large B-cell lymphoma (DLBCL), and T-cell lymphoma (T-CL). mo: months.

Table 3. Overall response and median duration of response among responders by tumor type.

Tumor type

Number evaluable

ORR, % (CR, %)

ORR 95% CI

Median DOR, months (range)

Median DOR 95% CI

FL DLBCL MCL T-CL MZL/Other CLL/LL Overall

16 16 13 15 13 14 87

56 (6) 31 (6) 15 (8) 40 (7) 15 (0) 0 (0) 28 (5)

30–80 11-59 2–45 16–68 2–45 0–23 19–38

16.0 (0.0+–27.7+) 1.9 (0.7–13.6) 3.6 (2.8–4.3) 11.5 (1.4–11.8+) 8.9 (8.8–9.0) NA 8.8 (0.0+–27.7+)

3.3−NA 0.7−13.6 2.8−4.3 1.4−NA 8.8−9.0 16.0−NA 3.3−13.6

CI: confidence intervals; CLL: chronic lymphocytic leukemia; CR: complete response; DLBCL: diffuse large B-cell lymphoma; DOR: duration of response; FL: follicular lymphoma; LL: lymphoblastic lymphoma; MCL: mantle cell lymphoma; MZL: marginal zone lymphoma; NA: not available; ORR: overall response rate; T-CL: T-cell lymphoma.

hypernatremia, and decreased platelet aggregation (each in 1 patient). Treatment-emergent adverse events that led to treatment discontinuation in at least 2 patients included thrombocytopenia (12 patients), neutropenia (3 patients), anemia (2 patients), and lung infection (2 patients). Gastrointestinal toxicities resulted in no dose reductions and only 3 discontinuations resulting from diarrhea, vomiting, and rectal hemorrhage in one patient each (Online Supplementary Table S1). Across all tumor types, 4% of patients experienced cardiac treatment-emergent adverse events, with all events being reported as grade 1 or 2. One patient experienced first-degree atrioventricular block and bradycardia. Cardiac failure, defect in intraventricular conduction, and nodal arrhythmia were reported in one patient each; none of these cardiac events were considered serious or related to treatment. Clinically significant abnormalities in electrocardiogram recordings were reported for 8 patients. T-wave inversion, atrial fibrillation, and isolated ventricular premature beat were each reported in 2 patients; localized ST depression and relevant QTc prolongations were each reported in one patient. In comparison to previous experience with abexinostat administered on a 1-week-on / 1-week-off schedule,17 weekly monitoring for thrombocytopenia showed that haematologica | 2017; 102(5)

platelet counts fell sharply during the 14 days on abexinostat and recovered before the start of the next cycle (Figure 1).

Efficacy At a median follow up of 18 months, 87 patients were evaluable for efficacy. The 13 patients who did not have baseline or at least one post-baseline tumor evaluation were excluded from the efficacy analysis. The ORR across all tumor types was 28% [95% confidence interval (CI): 19%-38%], with 5% of patients showing complete responses (CR) and a median duration of response of 8.8 months (range 0.0+ to 27.7+ months) (Table 3 and Figure 2). The highest responses were observed in FL, T-CL, and DLBCL with ORRs of 56% (95%CI: 30%-80%), 40% (95%CI: 16%-68%), and 31% (95%CI: 11%-59%), respectively. Median durations of response of 16.0 months [95%CI: 3.3-not available (NA)], 11.5 months (95%CI: 1.4-NA), and 1.9 months (95%CI: 0.7-13.6) were observed in each of these tumor types, respectively. Among patients with T-CL, responses varied by subtypes with higher responses observed in patients with angioimmunoblastic T-cell lymphoma (AITL) with an ORR of 71% (1 CR and 4 PR in 7 evaluable patients) than in patients with peripheral T-cell lymphoma not otherwise characterized (ORR 907


V. Ribrag et al.

20%; 1 PR) and cutaneous T-cell lymphoma-mycosis fungoides/Sezary syndrome (ORR 0%). Median PFS was 10.2 months (95%CI: 2.8-NA) for FL, 5.5 months (95%CI: 2.8NA) for T-CL, and 2.8 months (95%CI: 1.4-3.7) for DLBCL (Figure 3). The PFS for all tumor histologies is shown in Online Supplementary Figure S1. Median overall survival was not reached for FL (95%CI: 18.4-NA), and was 17.8 months (95%CI: 11.2-NA) for T-CL and 10.2 months (95%CI: 4.4-NA) for DLBCL (Figure 4).

Discussion In this study, we report that the pan-HDAC inhibitor abexinostat showed an ORR across all tumor types of 28%. Promising response rates were observed in patients with FL, T-CL, and DLBCL, with ORRs of 56%, 40%, and 31% in each of these NHL subtypes, respectively. Considering the difficulties involved in cross-trial comparisons, the results for abexinostat in FL are consistent with those of Evens et al., who reported an ORR of 64% with abexinostat in multiply relapsed disease.17 A median PFS of 10.2 months was observed for patients with FL in this study; a median PFS of 20.5 months was reported by Evens et al. Interestingly, oral vorinostat, which has been approved for cutaneous T-cell lymphoma, yielded an ORR of 47% and a median PFS of 15.6 months when tested in patients with relapsed or refractory FL.20 Among patients with DLBCL, the ORR of 31% obtained with abexinostat is similar to the ORR of 28% recently reported for oral panobinostat in relapsed DLBCL.21 In other B-cell malignancies, abexinostat was associated with a modest response rate (15%) among patients with MCL and MZL. No responses to abexinostat were observed among patients with relapsed/refractory CLL, a result that is consistent with reports from other trials of single-agent HDAC inhibitors in this tumor type.22-25 Response to abexinostat in T-CL varied by subtype, with the highest ORR (71%, including 14% CR) in AITL comparing favorably with the ORRs of 30% and 45% reported for romidepsin26 and belinostat.6

Abexinostat showed a manageable toxicity profile in various NHL subtypes in this study. The most frequently occurring non-hematologic events were primarily gastrointestinal, and included diarrhea (42%) and nausea (33%), with most of these events being grade 1 or 2 in severity. Although cardiac toxicities, especially QTc prolongations, are considered a dose-limiting class effect of HDAC inhibitors,14 the use of abexinostat was associated with a low incidence of cardiac events (3%), which resulted in only one patient discontinuing treatment. These toxicities and their rates of occurrence were comparable with those reported in the phase I study15 and substantially lower than the phase I/II study by Evens et al., which reported incidences of nausea and diarrhea of 63% and 50%, respectively, although most of these events were low grade.17 Hematologic adverse events in this study, however, were more frequent and generally more severe than those reported by Evens et al. The incidence of grade 3 or over hematologic adverse events in this study compared to the Evens study were thrombocytopenia (80% vs. 20%) and neutropenia (27% vs. 13%).17 The difference in the toxicity profiles of abexinostat between the 2 studies may largely be attributed to the differences in dosing schedules used. Abexinostat, when administered on a 2-weeks-on / 1-week-off schedule, is associated with higher rates of grade 3 or over hematologic events. In contrast, the 1-week-on / 1-week-off schedule was associated with lower rates of high-grade hematologic toxicities. Platelet counts fell sharply during the 14 days of abexinostat and recovered during the week patients were off treatment (Figure 1). In contrast, fluctuations in platelet counts were less sharp and less prominent with the 1-week-on / 1-week-off schedule. Given that hematologic toxicities, especially thrombocytopenia, were doselimiting in our study, the 1-week-on / 1-week-off dosing schedule will be explored in future studies. In conclusion, this phase II study showed that abexinostat is clinically active in patients with relapsed/refractory NHL, especially in patients with FL, T-CL, and DLBCL, who exhibited high response rates and durable tumor control. Abexinostat showed good tolerability during pro-

Figure 4. Overall survival (OS) in follicular lymphoma (FL), diffuse large B-cell lymphoma (DLBCL), and T-cell lymphoma (T-CL). mo: months.

908

haematologica | 2017; 102(5)


Safety and efficacy of abexinostat in NHL and CLL

longed drug administration; however, using a less doseintense schedule may mitigate the hematologic toxicity. Abexinostat is currently being tested in a variety of clinical trial settings, and further examination in FL, T-CL, and DLBCL is warranted. Acknowledgments The authors would like to thank the patients who participated

References 1. Barneda-Zahonero B, Parra M. Histone deacetylases and cancer. Mol Oncol. 2012;6(6):579-589. 2. Ropero S, Esteller M. The role of histone deacetylases (HDACs) in human cancer. Mol Oncol. 2007;1(1):19-25. 3. Marchion D, Munster P. Development of histone deacetylase inhibitors for cancer treatment. Expert Rev Anticancer Ther. 2007;7(4):583-598. 4. Mottamal M, Zheng S, Huang TL, Wang G. Histone deacetylase inhibitors in clinical studies as templates for new anticancer agents. Molecules. 2015;20(3):3898-3941. 5. West AC, Smyth MJ, Johnstone RW. The anticancer effects of HDAC inhibitors require the immune system. Oncoimmunology. 2014;3(1):e27414. 6. O'Connor OA, Horwitz S, Masszi T, et al. Belinostat in patients with relapsed or refractory peripheral T-cell lymphoma: results of the pivotal phase II BELIEF (CLN19) study. J Clin Oncol. 2015;33(23):24922499. 7. Olsen EA, Kim YH, Kuzel TM, et al. Phase IIb multicenter trial of vorinostat in patients with persistent, progressive, or treatment refractory cutaneous T-cell lymphoma. J Clin Oncol. 2007;25(21):3109-3115. 8. Piekarz RL, Frye R, Turner M, et al. Phase II multi-institutional trial of the histone deacetylase inhibitor romidepsin as monotherapy for patients with cutaneous T-cell lymphoma. J Clin Oncol. 2009; 27(32):5410-5417. 9. San-Miguel JF, Hungria VT, Yoon SS, et al. Panobinostat plus bortezomib and dexamethasone versus placebo plus bortezomib and dexamethasone in patients with relapsed or relapsed and refractory multiple myeloma: a multicentre, randomised, double-blind phase 3 trial. Lancet Oncol. 2014;15(11):1195-1206. 10. Whittaker SJ, Demierre MF, Kim EJ, et al.

haematologica | 2017; 102(5)

11.

12.

13.

14. 15.

16.

17.

18.

in the study and their supportive families, and the investigators and clinical research staff from the study centers. Funding The study was originally sponsored by Servier and subsequently by Pharmacyclics LLC, an AbbVie company. Medical writing assistance was provided by Supriya Srinivasan, PhD, and was funded by Pharmacyclics LLC, an AbbVie company.

Final results from a multicenter, international, pivotal study of romidepsin in refractory cutaneous T-cell lymphoma. J Clin Oncol. 2010;28(29):4485-4491. Veliz M, Pinilla-Ibarz J. Treatment of relapsed or refractory chronic lymphocytic leukemia. Cancer Control. 2012;19(1):3753. Watanabe T. Investigational histone deacetylase inhibitors for non-Hodgkin lymphomas. Expert Opin Investig Drugs. 2010;19(9):1113-1127. Yardley DA, Ismail-Khan RR, Melichar B, et al. Randomized phase II, double-blind, placebo-controlled study of exemestane with or without entinostat in postmenopausal women with locally recurrent or metastatic estrogen receptor-positive breast cancer progressing on treatment with a nonsteroidal aromatase inhibitor. J Clin Oncol. 2013;31(17):2128-2135. West AC, Johnstone RW. New and emerging HDAC inhibitors for cancer treatment. J Clin Invest. 2014;124(1):30-39. Morschhauser F, Terriou L, Coiffier B, et al. Phase 1 study of the oral histone deacetylase inhibitor abexinostat in patients with Hodgkin lymphoma, non-Hodgkin lymphoma, or chronic lymphocytic leukaemia. Invest New Drugs. 2015;33(2):423-431. Fouliard S, Robert R, Jacquet-Bescond A, et al. Pharmacokinetic/pharmacodynamic modelling-based optimisation of administration schedule for the histone deacetylase inhibitor abexinostat (S78454/PCI-24781) in phase I. Eur J Cancer. 2013;49(13):27912797. Evens AM, Balasubramanian S, Vose JM, et al. A phase I/II multicenter, open-label study of the oral histone deacetylase inhibitor abexinostat in relapsed/refractory lymphoma. Clin Cancer Res. 2016; 22(5):1059-1066. Cheson BD, Pfistner B, Juweid ME, et al. Revised response criteria for malignant lymphoma. J Clin Oncol. 2007;25(5):579586.

19. Hallek M, Cheson BD, Catovsky D, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood. 2008; 111(12):5446-5456. 20. Kirschbaum M, Frankel P, Popplewell L, et al. Phase II study of vorinostat for treatment of relapsed or refractory indolent non-Hodgkin's lymphoma and mantle cell lymphoma. J Clin Oncol. 2011;29(9):11981203. 21. Assouline SE, Nielsen TH, Yu S, et al. Phase 2 study of panobinostat with or without rituximab in relapsed diffuse large B-cell lymphoma. Blood. 2016;128(2):185-194. 22. Byrd JC, Marcucci G, Parthun MR, et al. A phase 1 and pharmacodynamic study of depsipeptide (FK228) in chronic lymphocytic leukemia and acute myeloid leukemia. Blood. 2005;105(3):959-967 23. Garcia-Manero G, Yang H, Bueso-Ramos C, et al. Phase 1 study of the histone deacetylase inhibitor vorinostat (suberoylanilide hydroxamic acid [SAHA]) in patients with advanced leukemias and myelodysplastic syndrome. Blood. 2008; 111(3):1060-1066. 24. Blum KA, Advani A, Fernandez L, et al. Phase II study of the histone deacetylase inhibitor MGCD0103 in patients with previously treated chronic lymphocytic leukemia. Br J Haematol. 2009;147:507514. 25. Gimsing P, Hansen M, Knudsen LM, et al. A phase I clinical trial of the histone deacetylase inhibitor belinostat in patients with advanced hematological neoplasia. Eur J Haematol. 2008;81(3):170-176. 26. Coiffier B, Pro B, Prince HM, et al. Romidepsin for the treatment of relapsed/refractory peripheral T-cell lymphoma: pivotal study update demonstrates durable responses. J Hematol Oncol. 2014; 7:11.

909


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Plasma Cell Disorders

Ferrata Storti Foundation

A concise revised Myeloma Comorbidity Index as a valid prognostic instrument in a large cohort of 801 multiple myeloma patients

Monika Engelhardt,1* Anne-Saskia Domm,1* Sandra Maria Dold,1 Gabriele Ihorst,2 Heike Reinhardt,1 Alexander Zober,1 Stefanie Hieke,2,3 Corine Baayen,3,4 Stefan Jürgen Müller,1 Hermann Einsele,5 Pieter Sonneveld,6 Ola Landgren,7 Martin Schumacher3 and Ralph Wäsch1

Haematologica 2017 Volume 102(5):910-921

Department of Medicine I, Hematology, Oncology & Stem Cell Transplantation, Medical Center - University of Freiburg, Faculty of Medicine, Germany; 2Clinical Trials Unit, Medical Center - University of Freiburg, Faculty of Medicine, Germany; 3Center for Medical Biometry and Statistics, University of Freiburg, Faculty of Medicine, Germany; 4Université de Nantes, UFR des Sciences Pharmaceutiques, Nantes Cedex, France; 5Department of Internal Medicine II, University Hospital, Würzburg, Germany; 6Department of Hematology, University Rotterdam, The Netherlands and 7Myeloma Service, Memorial Sloan-Kettering Cancer Center, New York, NY, USA 1

* ME and A-SD contributed equally

ABSTRACT

Correspondence: monika.engelhardt@uniklinik-freiburg.de

Received: December 19, 2016. Accepted: January 25, 2017. Pre-published: February 2, 2017. doi:10.3324/haematol.2016.162693 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/910 ©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.

910

W

ith growing numbers of elderly multiple myeloma patients, reliable tools to assess their vulnerability are required. The objective of the analysis herein was to develop and validate an easy to use myeloma risk score (revised Myeloma Comorbidity Index) that allows for risk prediction of overall survival and progression-free survival differences in a large patient cohort. We conducted a comprehensive comorbidity, frailty and disability evaluation in 801 consecutive myeloma patients, including comorbidity risks obtained at diagnosis. The cohort was examined within a training and validation set. Multivariate analysis determined renal, lung and Karnofsky Performance Status impairment, frailty and age as significant risks for overall survival. These were combined in a weighted revised Myeloma Comorbidity Index, allowing for the identification of fit (revised Myeloma Comorbidity Index ≤3 [n=247, 30.8%]), intermediate-fit (revised Myeloma Comorbidity Index 4-6 [n=446, 55.7%]) and frail patients (revised Myeloma Comorbidity Index >6 [n=108, 13.5%]): these subgroups, confirmed via validation analysis, showed median overall survival rates of 10.1, 4.4 and 1.2 years, respectively. The revised Myeloma Comorbidity Index was compared to other commonly used comorbidity indices (Charlson Comorbidity Index, Hematopoietic Cell Transplantation-Specific Comorbidity Index, KaplanFeinstein Index): if each were divided in risk groups based on 25% and 75% quartiles, highest hazard ratios, best prediction and Brier scores were achieved with the revised Myeloma Comorbidity Index. The advantages of the revised Myeloma Comorbidity Index include its accurate assessment of patients' physical conditions and simple clinical applicability. We propose the revised Myeloma Comorbidity Index to be tested with the "reference" International Myeloma Working Group frailty score in multicenter analyses and future clinical trials. The study was registered at the German Clinical Trials Register (DRKS-00003868).

haematologica | 2017; 102(5)


Comprehensive appraisal of a weighted myeloma comorbidity index

Introduction Over the past decade, overall survival (OS) has improved significantly in patients with multiple myeloma (MM). This is driven by better biological insights in the disease, implementation of more sensitive tests and technologies leading to earlier diagnosis, access to better combination therapies and increased access to supportive care measures.1-3 However, MM typically affects elderly patients, who face the challenge that treatment endurance is poorer and prognosis more unfavorable.4,5 Moreover, the simultaneous presence of additional diseases may complicate antimyeloma treatment.1,3 In general, comorbidities have been shown to influence cancer patients’ general health status, limit their physical condition and OS.6-11 Therefore, with a growing number of elderly patients, reliable tools to assess patients' vulnerability as expressed in chronic conditions and limitations in daily activity are required to guide therapeutic decisions.4,12-15 Historically, treatment decisions in symptomatic MM patients have been largely age-based. Today, disease biology and fitness, including patients' Karnofsky Performance Status (KPS), are considered when assessing therapeutic options.3,14 However, the KPS is often overestimated and does not reflect the entire functional status.10 Therefore, advances in more precise ways of defining fitness are warranted. Moreover, since elderly MM patients are often excluded from clinical trials due to strict inclusion criteria,16 these trial results are not necessarily transferable to elderly patients. In this context, the International Myeloma Working Group (IMWG), European Myeloma Network (EMN) and others (e.g., IFM, HOVON, DSMM, GMMG) recommended that age, physical condition and comorbidities are included in therapy decisions.1,8,10,12,14 Since cytogenetic aberrations are additional prognostic factors in MM,17-20 it may also be important to include cytogenetics in MM-specific risk scores. Risk scores for MM have indeed included disease-related risks (the International Staging System (ISS), lactate dehydrogenase (LDH), cytogenetics), combined comorbidities with cytogenetics or multiple comorbidity screening tests (IMWG frailty score).12,17-19 Our prior test8 and independent validation analyses9,10 defined impaired renal function, lung function or KPS as relevant risks via thoroughly assessed univariate and multivariate analyses. These variables were combined in an additive Initial Myeloma Comorbidity Index (I-MCI),8-10 which enabled the clear definition of risk groups with substantially different progression-free survival (PFS) and OS. Furthermore, it was found to add valuable information to the ISS.10 In order to refine and weight our I-MCI, we tested and validated a 'revised MCI' (R-MCI) based on a large cohort of 801 MM patients. Additionally, we compared the RMCI to other internationally used Comorbidity Indices (CIs), namely Charlson CI (CCI), Hematopoietic Cell Transplantation-specific CI (HCT-CI) and KaplanFeinstein Index (KFI). Frailty scores are already used clinically for various cancers, but a comprehensive comorbidity, frailty and disability evaluation is time-consuming and less applicable outside centers with oncogeriatric teams,21 which was the reason why we aimed to establish a concise, time-saving R-MCI. Since the comparison of the Initial-/R-MCI (I-/R-MCI) with the IMWG frailty score was already meticulously performed by us,22 it was not the focus of this analysis, haematologica | 2017; 102(5)

rather, the aim was to define a concise, weighted, both tested and validated MM-specific risk score in a large cohort which could be subsequently used for the measurement of frailty in multicenter analyses and future clinical trials.

Methods Patient population and study design This prospective assessment was based on the analysis of 801 consecutive MM patients at the time of initial diagnosis and first presentation at our center between 1997 and 2012. The study was registered at the German Clinical Trials Register (DRKS00003868). The primary objective was to optimize the I-MCI8-10 within a weighted R-MCI in a large myeloma cohort. Secondary objectives included the impact of the R-MCI as compared to the IMCI, CCI, HCT-CI and KFI (Online Supplementary Table S1), and their value for PFS and OS. The analysis was carried out according to the guidelines of the Declaration of Helsinki principles and good clinical practice. All patients gave their written informed consent for institutional-initiated research studies and analyses of clinical outcome studies conforming to the institutional review board guidelines.

Assessment The I-MCI consists of an additive scoring system, namely renal, lung and/or KPS impairment.8-10 In order to weight this in an even larger cohort, 13 comorbidities were assessed in 801 patients: these were graded and rated according to Common Terminology Criteria for Adverse Events (CTCAE) 4.03, which included: renal, lung and KPS impairment, cardiac, liver or gastrointestinal disease, disability, frailty, infection, thromboembolic events, peripheral neuropathy, pain and secondary malignancies (Table 1). In addition, age, cytogenetics via fluorescence in situ hybridization (FISH after CD138 selection), renal function (via estimated glomerular filtration rate [eGFRMDRD]) and lung disease, including lung function tests, were determined as described.8-11,22-25 We performed interphase FISH on CD138+ plasma cells, which were analyzed using DNA probes specific for the following chromosomal aberrations: t(11;14)(q13;q32), t(4;14)(p16;q32), t(14;16)(q32;q23) and t(14;20)(q32;q12 (Abbott Laboratories, IL, USA), XL 5q31/5p12, XCE 9, 11, 15, gain(1q21), del(1p32), del(13q14), del(17p13) and c-myc rearrangements (MetaSystems, Altlussheim, Germany). The score of Wuilleme et al.26 was used to assess ploidy by using gains of at least two of the chromosomes. For each probe a minimum of 100 nuclei were scored. European Myeloma Network (EMN) cutoff values were applied for the detection of aberrations.27 Unfavorable cytogenetics were defined as del(17p13), del(13q14), t(4;14), t(14;16), t(14;20), hypodiploidy, c-myc and chromosome 1 aberrations.1,8,10,22,24,27-31 The KPS was defined as normal (100%), mildly (90%), moderately (80%) or more substantially impaired (≤70%). Frailty and disability were assessed in order to get a more precise determination of patients’ physical condition. The Fried definition for frailty was used, which takes into account the added presence of weakness, poor endurance, low physical activity, slow gait speed and shrinking, with ≤2 factors defining frailty as moderate and with ≥3 factors determining frailty as severe.32-34 The assessment was performed by a staff member trained in oncogeriatrics (A-SD, SMD, AZ, SJM), and was performed identically throughout the study period. Patient characteristics included age, myeloma type, stage, β2-microglobulin (β2-MG), creatinine, bone marrow (BM) infiltration, cytogenetics and treatment (Table 2). 911


M. Engelhardt et al.

Statistical analysis Data were analyzed using SAS 9.2 (SAS Institute Inc., NC, USA). OS was calculated from the date of initial diagnosis until the date of death from any cause, while PFS was calculated from the date of initial diagnosis until the date of progression, relapse or death from any cause. When no event of interest occurred, observations were censored at the time the patient was last seen alive/without documented event, or at the latest on June 1st, 2015. OS and PFS rates were estimated using the Kaplan-Meier method, and compared using the log-rank test. In order to weight the MCI in a large cohort,8-10 the data set was randomly split into 2/3 and 1/3, namely a training (n=552) and validation set (n=249). The training set was built by randomly drawing 552 samples. The training set was used to develop the R-MCI, and the validation set to validate our results. Multivariate Cox proportional hazards regression models with backward variable selection were applied to the training set to evaluate the prognostic significance of the comorbidity factors. Variable selection was based on complete case analysis. For all other variables, the 552 patients without any missing data with 294 events (deaths) were used. The results of the final model with prognostic factors contributing to the R-MCI were presented as estimated hazard ratios (HRs) with two-sided 95% confidence interval (CI), corresponding log hazard

ratios and P-values (Table 3). Score weights were determined based on log hazard ratios, i.e., the regression coefficients of the prognostic factors, as these reflect the level of association with the outcome of OS on an additive scale. We assigned a score weight of 0 if the log hazard ratio was below 0.3, a score weight of 1 if the log hazard ratio was between 0.31 and 0.7, a score weight of 2 if the log hazard ratio was between 0.71 and 1.07, and a score weight of 3 if the log hazard ratio was 1.08 or higher, leading to a maximum of 9 points (Table 3). This rule very closely approximates the weights as described.35 In order to additionally evaluate whether the co-variable cytogenetics can increase the predictive performance of this score (Table 3), a multivariate Cox model including a preliminary score as a co-variable was compared to a multivariable Cox model including both the preliminary score and cytogenetics: the models were based on 353 patients without any missing data in all co-variables, with or without the inclusion of cytogenetics (Table 3).17–20 Prediction errors based on the Brier score36 were used to compare the R-MCI, with and without cytogenetics, determining that the R-MCI could be improved with the inclusion of cytogenetics (Online Supplementary Figure S1). Of note, our final 9-point weighted R-MCI can be used as a risk tool both with or without cytogenetics (e.g., if cytogenetics were unavailable). Albeit there was no missing data for the prognostic factors

Table 1. Definition and grading of 13 comorbidities and physical functions in myeloma patients.

I-MCI

Variables

Mild Moderate

Definition and grading

Severe

References

1. Renal function: eGFR / serum creatinine 2. Lung function: dyspnea or FEV1/FVCa, FEV1, TLC, respiratory insufficiency 3. Karnofsky Performance Status 4. Cardiac function: arrhythmias, myocardial infarction/CAD, heart failure 5. Hepatic function: chronic hepatitis, cirrhosis, fibrosis, hyperbilirubinemia 6. GI disease: nausea, vomiting, diarrhea, ulcer 7. Disability: help in personal care and household tasks 8. Frailty: weakness, poor endurance, low physical activity, slow gait speed 9. Infection 10. Thromboembolic event

CTCAE grade 1

CTCAE grade 2

CTCAE grade 3-4

Kleber8–10

dyspnea upon intense activity, mildly altered lung function

dyspnea at rest/few steps taken/the need for oxygen/non-invasive ventilation or FEV1<50% ≤70%

Kleber8–10

90%

dyspnea upon moderate activity, moderately altered lung function or respiratory insufficiency 80%

CTCAE grade 1

CTCAE grade 2

CTCAE grade 3-4

CTCAE, 4.0

CTCAE grade 1

CTCAE grade 2

CTCAE grade 3-4

CTCAE, 4.0

CTCAE grade 1

CTCAE grade 2

CTCAE grade 3

CTCAE, 4.0

occasional

frequent

≥1x/day

Palumbo12

1 factor

2 factors

≥3 factors

11. PNP 12. Pain 13. Secondary malignancy

local intervention venous thrombosis

oral intervention thrombosis, medical intervention indicated CTCAE grade 1 CTCAE grade 2 CTCAE grade 1 CTCAE grade 2 1. chronological criteria: before, synchronous or after MM 2. local criteria: local vs. disseminated cancer 3. etiological criteria: hematological, solid or skin tumors

i.v. intervention life-threatening, urgent intervention indicated CTCAE grade 3-4 CTCAE grade 3-4

Kleber8–10

Rodriguez-Mañas,32 Xue33 CTCAE, 4.0 CTCAE, 4.0 Kristinsson47 CTCAE, 4.0 CTCAE, 4.0 Hasskarl28 Engelhardt24 Kleber8–10

FEV1/FVC: Tiffeneau-Pinelli Index: ratio of the forced expiratory volume in 1 second and the forced vital capacity. CAD: Coronary Artery Disease; CTCAE: Common Terminology Criteria for Adverse Events; eGFR: estimated glomerular filtration rate; FEV1: forced expiratory volume in one second; FVC: forced vital capacity; GI: gastrointestinal; PNP: peripheral neuropathy; TLC: total lung capacity; i.v.: intravenous; I-MCI: Initial Myeloma Comorbidity. a

912

haematologica | 2017; 102(5)


Comprehensive appraisal of a weighted myeloma comorbidity index

in our dataset, the provision of weighing unfavorable, favorable vs. "missing/unavailable cytogenetics" within this R-MCI, illustrates its usefulness for primary or secondary institutions and non-academic centers. The R-MCI was also compared to the I-MCI, CCI, HCT-CI and KFI, evaluating the prognostic role on OS with Cox regression models (Table 4) in terms of HRs. The predictive ability of different scores was assessed using prediction error curves and Brier scores (Online Supplementary Figure S2):36 the smaller this prediction error is, the better the curves' prediction rate turns out, with the 'reference' constituting a model without co-variables. Additionally, the R-MCI was assessed on OS in patients with different antimyeloma treatments and ages. Since the comparison of the MCI with the IMWG frailty score had already been performed,22 it was not the focus of this analysis, which was rather to define a weighted, tested and validated MM-specific risk score in a large myeloma cohort.

Results Patient characteristics The analysis included 801 consecutive MM patients. The median follow up was 6.1 years. The median age was 63 years: 28% of patients were 66-75 years and 13% older than 75 years, which is very typical for tertiary centers.711,22,24,37,38 Gender distribution and myeloma subtypes corresponded to the data as described.7-11,24 Other characteristics were likewise representative of large MM centers, e.g., typical paraprotein frequencies and mostly advanced Durie-Salmon and ISS II/III disease stages. The median β2MG level was 4.5mg/dL, renal function showed a median creatinine level of 0.93mg/dL and BM plasma cell infiltration of 30%. Patients underwent treatment according to international guidelines, labels and practices as described.1,10,11,22,24 Autologous stem cell transplantation

Table 2. Patient characteristics.

Entire cohort (n=801) n (%) Median (range) Patient-specific data Male : female Age (years) MM-specific data Type of myeloma IgG / IgA IgM / IgD Light-chain MM only Biclonal (HC) Non-secretory

450 (56.2) : 351 (43.8)

Training set (n=552 / 68.9%) n (%) Median (range) 316 (57) : 236 (43)

63 (21-93)*

455 (56.8) / 152 (19.0) 6 (0.8) / 2 (0.2) 162 (20.2) 6 (0.8) 18 (2.3)

κ/l 502 (62.7) / 276 (34.5) Biclonal (LC) 5 (0.6) Non-secretory 18 (2.3) Durie-Salmon I 204 (25.5) II 117 (14.6) III 480 (59.9) A/B 665 (83.1) / 136 (16.9) ISS 759 (94.8)a I 225 (28.2) II 206 (25.8) III 328 (41.0) Laboratory parameters β2-microglobulin (mg/dL) 755 (94.3)e Creatinine (mg/dL) BM infiltration rate (%) 695 (86.8)d Cytogenetics Favorable 316 (39.5) Unfavorableb 212 (26.5) Missing 273 (34.1) Therapy SCT with novel agents 300 (37.5) SCT w/o novel agents 83 (10.4) Standard with novel agents 173 (21.6) Standard w/o novel agentsc 170 (21.2) w/o CTx# 75 (9.4)

4.5 (1.1-65.5) 0.93 (0.4-17.9) 30 (0-100)

Validation set (n=249 / 31.1%) n (%) Median (range) 134 (53.8) : 115 (46.2)

62 (21-93)

63 (32-89)

309 (56.0) / 105 (19.0) 3 (0.5) / 2 (0.4) 117 (21.0) 4 (0.7) 12 (2.2)

146 (58.6) / 47 (18.9) 3 (1.2) / 0 (0) 45 (18.1) 2 (0.8) 6 (2.4)

355 (64.1) / 181 (32.9) 4 (0.7) 12 (2.2)

147 (59.0) / 95 (38.2) 1 (0.4) 6 (2.4)

139 (25.3) 88 (15.8) 325 (58.9) 456 (82.6) / 96 (17.4) 517 (93.5) 146 (26.5) 139 (25.3) 232 (42.0)

65 (26.1) 29 (11.7) 155 (62.2) 209 (83.9) / 40 (16.1) 242 (97.2) 79 (31.7) 67 (26.9) 96 (38.6)

516 (93.5) 487 (88.2)

4.72 (1.1-65.5) 0.92 (0.4-17.9) 30 (0-100)

240 (96.4) 208 (83.5)

214 (38.8) 140 (25.4) 198 (35.9)

102 (41.0) 72 (28.9) 75 (30.1)

194 (35.1) 62 (11.2) 118 (21.4) 127 (23.1) 51 (9.3)

106 (42.6) 21 (8.4) 55 (22.1) 43 (17.3) 24 (9.6)

4.20 (1.4-52.6) 0.94 (0.5-10.5) 30 (0-90)

*13% of patients were >75 years, 7% were 76-79 years and 6.3% ≥80 years. aNot evaluated in n=42 patients because of missing data. bUnfavorable cytogenetics defined as del(17p13), del(13q14), t(4;14), t(14;16), t(14;20), hypodiploidy, c-myc and chromosome 1 aberrations. cNovel agents: e.g., thalidomide, lenalidomide, bortezomib. dNot evaluated in n=106 patients because of missing data. eNot evaluated in n=46 patients because of missing data. #Radiotherapy and steroids alone. n: number; Ig: immunoglobulin; HC: heavy-chain; MM: multiple myeloma; LC: light-chain; ISS: International Staging System; BM: bone marrow; SCT: stem cell transplantation; w/o CTx: without chemotherapy; κ / l: kappa / lambda.

haematologica | 2017; 102(5)

913


M. Engelhardt et al. Table 3. Multivariate Cox proportional hazards model of the training set analysis (n=552) based on backward selection for overall survival (OS), and the value of inclusion of cytogenetics (n=353).

Multivariate Cox proportional hazards model of the training set analysis (n=552) Definition n=552 (%) HR (2.5-97.5%) 1. Renal disease (eGFRMDRD)a

2. Lung disease 3. KPS

4. Age (years)

5. Frailty

± Cytogenetics

≥90 60-89 <60 No/mild Moderate/severe 100% 80-90% ≤70% <60 60-69 ≥70 No/mild Moderate Severe Favorable Unfavorable Unavailable

184 (33) 193 (35) 175 (32) 470 (85) 82 (15) 35 (6) 207 (38) 310 (56) 226 (41) 185 (33) 141 (26) 323 (59) 140 (25) 89 (16)

1 (-) 1.25 (0.92-1.68) 1.96 (1.43-2.68) 1 (-) 1.65 (1.24-2.18) 1 (-) 2.17 (1.04-4.52) 2.96 (1.43-6.12) 1 (-) 1.43 (1.06-1.92) 2.08 (1.50-2.89) 1 (-) 1.54 (1.17-2.04) 2.02 (1.45-2.82)

P-value

<0.0001 0.0005 0.0036

<0.0001

<0.0001

log(HR)

Score weight

0 0.22 0.67 0 0.50 0 0.77 1.08 0 0.36 0.73 0 0.43 0.70

0 0 1 0 1 0 2 3 0 1 2 0 1 1 0 1 0 9

HR (2.5-97.5%)

P-value

Maximum points

Univariate and bivariate Cox model with and without inclusion of cytogenetics (n=353) log(HR) Univariate Cox model with preliminary score Multivariable Cox model with inclusion of cytogenetics

Preliminary score

0.10

Preliminary score

0.10

Cytogenetics unfavorable

0.44

1.11 (1.08-1.14) 1.11 (1.08-1.13) 1.56 (1.13-2.15)

<0.0001 <0.0001 0.006

a eGFR calculated as MDRD 186 × (serum creatinine level [mg/dl]) -1.154 × (age [y]) -0.203 × (0.742 if female, 1.21 if black person), log hazard ratios: parameter estimates. We assigned a score weight of 0, if the log hazard ratio was below 0.3, a score weight of 1, if the log hazard ratio was between 0.31 and 0.7, a score weight of 2, if the log hazard ratio was between 0.71 and 1.07, and a score weight of 3, if the log hazard ratio was 1.08 or higher, leading to a maximum of 9 points. This rule very closely approximates the weights as previously described.35 n: number; HR: hazard ratio; KPS: Karnofsky Performance Status; eGFRMDRD: estimated glomerular filtration rate by MDRD (Modification of Diet in Renal Disease).

(ASCT) was recommended for medically fit, symptomatic patients up to the age of 70 years.22,25 Induction usually consisted of bortezomib-based regimens, such as VCD (bortezomib, cyclophosphamide and dexamethasone) or CTD (cyclophosphamide, thalidomide and dexamethasone). Mobilization and conditioning were performed as described.1,10,11,23,24 Patients ineligible for ASCT received melphalan, prednisone and bortezomib (MPV), melphalan, prednisone and thalidomide (MPT) or melphalan and prednisone (MP).1 Novel agent-based therapies included immunomodulatory drugs and proteasome inhibitor treatment according to the approved indications and in line with treatment at other international centers (Table 2). In order to revise the MCI, the data set was randomly split into 2/3 and 1/3 of patients, using a training (n=552) and validation set (n=249). The training set was used to develop the R-MCI and the validation set to validate our results. Both groups were comparable with respect to relevant patient-specific and MM-specific data, laboratory parameters and therapy. The data of the entire patient cohort, and of both the training and validation sets are displayed in Table 2. Approximately one-half (43%) of the patients received standard treatment without stem cell transplantation (SCT), the other percentage of patients includes those 914

who underwent SCT (Table 2). Patient characteristics according to treatment are displayed in the Online Supplementary Table S2. Treatment was not modified according to the comorbidity scores in line with prior studies.7-10,12,35,39,40

Frequency of specific comorbidities Frequent comorbidities (>30%) of all grades were KPS impairment (94%), renal impairment (68%), frailty (62%), cardiac impairment (45%), disability (43%) and lung impairment (32%). More severely graded comorbidities were again KPS impairment, frailty, disability, renal impairment, lung impairment, cardiac impairment and infections. Other comorbid conditions, such as liver and gastrointestinal impairment and thrombosis occurred to a lesser extent and severity (Figure 1).

Multivariate analysis for OS, weighting and risk stratification via MCI The multivariate Cox proportional hazards model based on backward selection revealed five highly significant risks as relevant for OS (Table 3). Score weights for comorbidities (Table 3) were determined based on regression coefficients of the prognostic factors, i.e., log hazard ratios. In a separate haematologica | 2017; 102(5)


Comprehensive appraisal of a weighted myeloma comorbidity index Table 4. Univariate Cox proportional hazards model for overall survival (OS) - comparison of the R-MCI with other comorbidity scores: I-MCI, CCI, HCT-CI and KFI.

Score Training analysis (n=552) R-MCI

I-MCI

CCI

HCT-CI

KFI

Validation analysis (n=249) R-MCI

I-MCI

CCI

HCT-CI

KFI

low- vs. moderate- vs. high-riska n (%) fit (n=176) intermediate-fit (n=302) frail (n=74) fit (n=187) intermediate-fit (n=225) frail (n=140) fit (n=198) intermediate-fit (n=181) frail (n=173) fit (n=155) intermediate-fit (n=220) frail (n=177) fit (n=241) intermediate-fit (n=118) frail (n=193) fit (n=71) intermediate-fit (n=144) frail (n=34) fit (n=84) intermediate-fit (n=104) frail (n=61) fit (n=120) intermediate-fit (n=70) frail (n=59) fit (n=78) intermediate-fit (n=93) frail (n=78) fit (n=125) intermediate-fit (n=48) frail (n=76)

HR (2.5-97.5%)

P-value

2.87 (2.14-3.85) 9.57 (6.52-14.03)

<0.0001

2.47 (1.82-3.37) 4.45 (3.20-6.17)

<0.0001

2.05 (1.51-2.79) 3.56 (2.65-4.80)

<0.0001

2.29 (1.67-3.15) 2.84 (2.05-3.92)

<0.0001

2.16 (1.57-2.98) 3.53 (2.67-4.67)

<0.0001

5.26 (2.41-11.49) 28.35 (12.23-65.69)

<0.0001

4.30 (2.11-8.73) 9.47 (4.61-19.44)

<0.0001

2.37 (1.43-3.93) 6.40 (3.94-10.41)

<0.0001

1.80 (1.01-3.21) 5.06 (2.95-8.67)

<0.0001

1.73 (0.98-3.04) 4.80 (3.00-7.67)

<0.0001

Scoring groups (fit, intermediate-fit, frail) based on the 25% and 75% quartiles of the scores evaluated from the training set. R-MCI: revised Myeloma Comorbidity Index; I-MCI: initial Myeloma Comorbidity Index; CCI: Charlson Comorbidity Index; HCT-CI: Hematopoietic cell transplantation-specific Comorbidity Index; KFI: Kaplan-Feinstein Index; HR: hazard ratio; n: number.

a

Figure 1. Frequency of entire (blue columns) and moderatesevere (yellow columns) organ impairment. Frequency of relevant comorbidities and impairment of general condition in all MM patients. Proportion of patients with any degree of impairment/grade of severity (blue bars) vs. proportion of patients with moderate to severe impairment/grade of severity (red bars). For cardiac function, hepatic function, GI disease, infection, thrombosis and renal function the CTCAE grading system was used, with CTCAE 1-4 (blue bars) vs. CTCAE ≥2 (red bars). KPS and lung function was graded as described.8–10,22,23,46 Frailty was graded according to the definition of Rodriguez-Mañas L et al.32 and Fried LP et al.48 Disability was graded as described.32,33 Table 1 shows the definition and grading of all assessed comorbidities. GI: gastrointestinal; KPS: Karnofsky Performance Status; pts: patients.

haematologica | 2017; 102(5)

915


M. Engelhardt et al.

A

B

C

D

E

F

G

H

Figure 2. OS (A-D) and PFS (E-H) using the R-MCI (left) and I-MCI (right) in training and validation analyses. Overall survival (OS: A-D) and progression-free survival (PFS: E-H) curves using the revised MCI (R-MCI) and initial MCI (I-MCI) in the training and validation sets. Fit patients, with the use of the R-MCI, were defined with 0-3, intermediate with 4-6 and frail patients with 7-9 R-MCI points; with the use of the I-MCI with 0, 1 and 2-3 points, respectively. The survival curves of the R-MCI were stratified based on 25% and 75% quantiles leading to a stratification of fit patients with â&#x2030;¤3 R-MCI points, intermediate-fit patients with 4-6 R-MCI points and frail patients with >6 R-MCI points. The numbers of patients at risk in the respective groups are given below each Kaplan-Meier plot. PFS with the use of the R-MCI generated better group distinction both in the training set (E) and validation set (G) than with the use of the I-MCI (F and H). In line, OS was better distinguishable in 3 risk groups via the R-MCI in the training set (A) and validation set (C) vs. with the use of the I-MCI (B and D). R-MCI: revised Myeloma Comorbidity Index; I-MCI: initial Myeloma Comorbidity Index.

916

haematologica | 2017; 102(5)


Comprehensive appraisal of a weighted myeloma comorbidity index

model, cytogenetics proved to supplement both additional and independent information to these risks (Table 3): prediction errors based on the Brier score36 compared the RMCI, with and without cytogenetics, and showed the smallest prediction errors with the R-MCI, with the inclusion of cytogenetics therein (Online Supplementary Figure S1). The weight for cytogenetics was based on the regres-

sion coefficient from the multivariable Cox model, including both the preliminary score and cytogenetics (Online Supplementary Table S3). A prognostic model was generated by combining and weighting the risks as displayed in the Online Supplementary Table S3, showing only marginal changes compared to the results of the final model used for determining score weights (Table 3). The R-MCI allowed for

A

B

C

D

E

F

Figure 3. OS with the use of the R-MCI with different treatment intensity and age groups. OS with the use of the R-MCI in different treatment (A-D) and age cohorts (E,F), demonstrated excellent risk allocations of fit, intermediate and unfit patients with vs. without stem cell transplantation (A,B), with vs. without novel agent treatment (C,D) and <65 vs. >65 year old patients (E,F). Tx: transplantation; NA: novel agents; w: with; w/o: without.

haematologica | 2017; 102(5)

917


M. Engelhardt et al.

Figure 4. Scoring factors and maximum points of the R-MCI as compared to the IMWG frailty index, CCI, HCT-CI, KFI and R-ISS. Single factors of respective internationally used comorbidity scores, namely the IMWG frailty index, HCT-CI, KFI and R-ISS and R-MCI are displayed with their respective comorbidities therein. The number of risk factors and maximum score that can be achieved is also depicted. Moreover, respective overlapping risk factors with the R-MCI are shown. Impaired general condition is differently assessed with various scores, e.g., with the R-MCI via Karnofsky Performance Status (KPS) and frailty; with the IMWG frailty index via ADL and IADL and with the KFI via assessment of any locomotor impairment. IMWG: International Myeloma Working Group; CCI: Charlson Comorbidity Index; (I)ADL: (Instrumental) Activity of Daily Living; AIDS: Acquired Immune Deficiency Syndrome. HCT-CI: Hematopoietic cell transplantation-specific Comorbidity Index; R-MCI: revised Myeloma Comorbidity Index; KFI: Kaplan-Feinstein Index; ISS: International Staging System; R-ISS: revised ISS; LDH: lactate dehydrogenase.

the definition of largely different risk groups: in both training and validation analyses, it distinguished patients with clearly different OS and PFS (Figure 2A-H): patients with a R-MCI of ≤3 were defined as fit (247 patients [30.8%]), those with a R-MCI of 4-6 as intermediate-fit (446 patients [55.7%]) and those with a R-MCI >6 as frail (108 patients [13.5%]). The median OS in these defined groups was 10.1, 4.4 and 1.2 years, respectively, in the training set (Figure 2A), and in the validation set (Figure 2C) not reached (n.r.), 5.9 and 0.8 years, respectively. The median PFS in these groups was 4.1, 1.9 and 0.9 years, respectively, in the training set (Figure 2E), and 5.0, 2.3 and 0.8 years in the validation set, respectively (Figure 2G). From Figure 2, it can also be perceived that in the entire cohort, 13.5% were frail, if assessed via the R-MCI, whereas if assessed with the IMCI, 25.1% of the cohort were determined frail. Intermediate-fit patients were fairly similar with 56% assessed intermediate-fit via the R-MCI vs. 41.1% via the IMCI. Because the percentages of frail patients are distinctive, this highlights the relevance of performing a comprehensive comorbidity, frailty and disability evaluation also in 'intermediate-old' patients, and that different types of evaluations (e.g., scores: Table 4), result in diverse percentages of frail patients, namely of 13.5% with the R-MCI, 25.1% with the I-MCI, 29% with the CCI, 32% with the HCT-CI and 33.6% with the KFI, the latter four most likely overestimating frailty in MM, which illustrates the value of a prospectively validated MCI, also in intermediate-old MM patients. 918

MCI analysis in different treatment and age groups To also determine whether the R-MCI remained a significant risk tool in different treatment and age groups, we assessed the risk group allocation of fit (R-MCI ≤3), intermediate-fit (R-MCI 4-6) and frail patients (R-MCI >6), with vs. without SCT (Figure 3A,B), with and without novel agents (NAs) (Figure 3C,D) and in younger (<65 years) vs. older patients (≥65 years; Figure 3E,F). In all subgroups, the value of the R-MCI was confirmed (P<0.0001). Of note, the R-MCI was a highly valuable risk tool for younger and older MM patients, demonstrating again that age alone is not sufficient to define fitness and therapy allocation (Figure 3E,F). Although subgroups of patients were limited, and these were non-randomized treatment comparisons, it was of interest that fit patients fared favorably, even without SCT (Figure 3B [n=72]), without NAs (Figure 3D [n=108]) or ≥65 years (Figure 3F [n=25]; all alive), whereas intermediate-fit and frail patients in these subgroups performed less favorably as compared to those with SCT (Figure 3A), with NAs (Figure 3C) or younger patients (Figure 3E). Figures 3E and 3F illustrate that patients <65 years were fit, intermediatefit and frail in 48%, 50% and 2%, respectively, whereas in the >65 year old patient group, 7%, 63% and 29% were fit, intermediate-fit and frail, respectively. This showed that in the younger age group, there were also frail patients, albeit rarely, and that in the older cohort, fit patients were present, with 2% and 7%, respectively. Moreover, this demonstrates that frail patients increase with age, whereas intermediate-fit patients remained a very prominent group with haematologica | 2017; 102(5)


Comprehensive appraisal of a weighted myeloma comorbidity index

50% and 63% in both cohorts, respectively. Treatment was not modified according to the comorbidity scores, therefore, although very infrequently, frail patients (n=6; Figure 3A) as assessed via the R-MCI received transplants; even more impressively, 72 patients, notwithstanding being scored as fit via the R-MCI (Figure 3B), did not receive transplantation, suggesting more underutilization or undertreatment than overtreatment in a very rare subset. An even more detailed comparison of treatment subgroups of patients receiving more vs. less intensive treatment is depicted in the Online Supplementary Table S2, which shows comparable disease characteristics, but, as expected, more intensive treatment in younger patients.

equal in patients with favorable and unfavorable cytogenetics, it was, as expected, more advanced in those with unavailable cytogenetics, suggesting that cytogenetic analyses were less consistently performed in older patients. Gender and MM subtype distributions appeared comparable, whereas median β2-MG levels were higher in patients with unfavorable and unavailable cytogenetics, and BM infiltration was increased in patients with unfavorable cytogenetics. In terms of therapy allocation, the differences between unfavorable and favorable cytogenetic groups were less obvious, whereas those with unavailable cytogenetics had received less intensive treatment.

Comparison of the R-MCI with other comorbidity scores

Discussion

All patients were divided into three groups of fit, intermediate-fit and frail patients based on the 25% and 75% quartiles of the respective scores: the R-MCI, I-MCI, CCI, HCTCI and KFI (Table 4). Differences and overlapping similarities are depicted in Figure 4. The univariate Cox proportional hazards model revealed that all scores allocated patients into 3 risk groups of fit, intermediate-fit and frail patients, with increasing HR. All scores reached statistical significance in terms of OS. Direct comparison of these comorbidity scores showed that the R-MCI generated the highest HRs in the training set, increasing to 2.87 and 9.57 for intermediate-fit and frail patients, respectively. Consistent with these results, the validation analysis confirmed the highest HRs of 5.26 for intermediate-fit and 28.35 for frail patients with the use of the R-MCI, which achieved much higher HRs than those generated through the use of the I-MCI, CCI, HCT-CI and KFI. The superiority of both the R- and I-MCI with the highest HRs was confirmed in the validation analysis (Table 4) and via prediction error curves and Brier scores: comparison of the reference, the R-MCI, I-MCI, KFI, CCI and HCT-CI demonstrated that the R-MCI generated the most favorable prediction curves (Online Supplementary Figure S2). For the exact comparison, both the I- and R-MCI were also assessed side by side, both for OS and PFS (Figure 2AH). Survival curves of the I-MCI groups with 0, 1 and â&#x2030;Ľ2 risk factors, generated clear group allocations of fit, intermediate-fit and frail patients in both training and validation analyses, for OS (Figure 2B and 2D) and PFS (Figure 2F and 2H), albeit subgroup allocation via R-MCI was improved (Figure 2A and 2C, 2E and 2G).

Differences in organ function and comorbidity with favorable vs. unfavorable cytogenetics To determine whether unfavorable cytogenetics were also associated with organ dysfunction, increased comorbidity scores and unfavorable MM-associated characteristics, patients were separated into those with favorable, unfavorable and unavailable cytogenetics, the latter either due to patient preference, age, their location in external centers or other reasons.11,19 Organ impairment differed in patients with unfavorable or missing vs. favorable cytogenetics: the former exhibiting higher frequencies of KPS impairment, renal impairment, cardiac impairment and hepatic impairment. Moreover, these patients had more comorbidities, such as pain, infections and limited physical conditions with disability and frailty. In line with this, the R-MCI and KFI increased in patients with unfavorable or unavailable cytogenetics (Online Supplementary Table S4A). The Online Supplementary Table S4B summarizes the respective MM patient characteristics: while the median age was haematologica | 2017; 102(5)

The results of the present study confirm the prognostic value of the R-MCI on a large independent patient cohort.22 In order to weight our prior MCI,8-10 13 risk factors were thoroughly reassessed, including organ function and MMspecific risks (Table 1). The R-MCI was also compared to non-MM-specific comorbidity scores, namely CCI, HCTCI and KFI. Since we have already compared the MCI and IMWG frailty score,22 it was not the focus of this analysis, which was instead to define a weighted validated MM-specific risk score in a large myeloma cohort. Frequent comorbidities in MM proved to be frailty, disability, heart impairment, renal impairment, lung impairment and KPS impairment in agreement with prior studies.3,12,38 Of interest, multivariate risks (renal impairment, lung impairment, KPS impairment, age, and frailty) matched with frequent comorbidities. Other previously suspected risk factors, such as liver or gastrointestinal dysfunction,4,5,15 did not reach significance and proved not to be MM-specific risks. These less relevant comorbidities were generated from retrospective studies which were based on multicenter data entries and had assessed therapy-induced adverse events rather than baseline comorbidities. Therefore, they may bear the restriction of validity and information loss. Furthermore, the frequency of comorbidities in some retrospective studies was notably low,5,12 which was possibly related to incomplete multicenter data entries and solely clinical trial patient inclusion.12 Aside from organ impairment, cytogenetic aberrations corroborate with impaired OS in MM patients. Our analysis confirmed that cytogenetics provide independent additional information,17-20,41 and that patients with unfavorable cytogenetics had higher disease stages, adverse laboratory values and reduced organ and physical function. Although cytogenetics proved to be a relevant risk factor, our analysis also demonstrated that other factors, such as physical and organ conditions, are equally important. Moreover, the development of the R-MCI showed that the multivariate risks (renal function, lung function, KPS, age, and frailty) defined patients as fit, intermediate-fit and frail, which could be improved with the inclusion of cytogenetics, but was also readily usable if this information was unavailable (Table 3). Weighting of our R-MCI verified that this 9-point score was able to define 3 patient groups with clearly different median PFS and OS. Comparison of the R-MCI with others (CCI, HCT-CI, KFI) showed that they all divide patients into risk groups with substantially different OS, however, Brier scores determined the smallest prediction errors with the R-MCI. One reason for the comparability of the R-MCI with other risk scores is that they all include risk 919


M. Engelhardt et al.

factors that have some relevance in MM, namely renal function, lung function and in some physical condition (Figure 4). Compared to our non-weighted I-MCI,8-10 the R-MCI led to an improvement in group distinction, which highlights the relevance to further improve a MM-specific risk score as performed here. Although the HCT-CI and CCI have been tested and proven their usefulness,12,35,42 studies in MM have suggested that both may not have as much impact in MM as the I-/R-MCI.1,8-10,22,37 This led to some modification of prior comorbidity scores, albeit in diseases other than MM,40,43-45 and verified that adapted scores (such as the International Prognostic Index (IPI) for diffuse large B-cell lymphoma, Mantle Cell Lymphoma International Prognostic Index (MIPI) for mantel cell lymphoma and Follicular Lymphoma International Prognostic Index (FLIPI) for follicular lymphoma) are valuable. Moreover, the CCI with 19 different factors, HCT-CI with 17 and KFI with 12 diverse factors are much more complex and time-consuming to assess, whereas the R-MCI involves 5 comorbid conditions only, uses information which is routinely collected and requires minimal interpretation (Figure 4 and Online Supplementary Table S1). The strengths of our R-MCI lie in its inclusion of few comorbid conditions, that it is readily obtainable from the collection of the medical history and was obtained from the analysis of a large sample size. Furthermore, it is valuable both in younger and older patients. Other notable assets were that the R-MCI was tested and validated in the same cohort, randomly splitting the data into a training and validation set. Additional advantages of the R-MCI are that: 1) it allows for a more accurate assessment of physical conditions than best clinical judgment, age or KPS alone, 2) it precisely divides patients into fit, intermediate-fit and frail patients with definite PFS and OS risk groups, 3) current and relevant biological risks, namely cytogenetics can, but don't have to be included, and 4) it is very concise. Moreover, compared to other international comorbidity scores (the retrospectively assessed IMWG frailty score, CCI, HCT-CI and KFI),8-10,22 the R-MCI was particularly effective in identifying patients at risk.22,25 A limitation of the present study was that the R-MCI was generated from a large, independent data set, but was acquired in a single center. Nevertheless, we have compared our patients to those of other tertiary centers,22,46 demonstrating that our patients were representative, which is currently being assessed and affirmed in subsequent analyses and in prospective multicenter studies. Moreover, the cohort represented intermediate-old patients, which is typical for comprehensive cancer centers (CCCs).7-11,19,22-24,46 Because of the accumulation of challenging cases in these CCCs, patients therein show typical comorbidities despite being "younger".11 Since age is associated with increased comorbidity, our results will be even more pertinent for older patients.8,10,11,22,25 Another criticism might be that different antimyeloma treatments were applied, nevertheless, in sub-

References 1. Engelhardt M, Terpos E, Kleber M, et al. European Myeloma Network recommendations on the evaluation and treatment of newly diagnosed patients with multiple

920

groups and prior analyses of our group,8,10,22 we demonstrated that the R-MCI was equally relevant to distinguish highly significant risk groups (Figure 2 and Figure 3). Nevertheless, the heterogeneity of therapies may in part describe some differences in terms of PFS and OS, although it is noteworthy that our Kaplan-Meier curves of frail patients with transplants (Figure 3A), with novel agent treatment (Figure 3C) and in patients <65 years (Figure 3E) distinctly demonstrated that, despite intensive treatment and younger age, comorbidities and frailty induced a poor outcome. In conclusion, although the retrospectively assessed IMWG frailty score is considered the "reference", we demonstrate the validity of this straightforward R-MCI as a valid prognostic instrument in a large cohort treated according to current standards. Based on existing recommendations, the R-MCI can be applied in routine clinical care, multicentre analyses and future clinical trials. It may further be used in research to compare risk profiles of MM cohorts, to adjust for imbalanced risk profiles and to provide a basis to establish new clinical or biologic prognostic factors. Moreover, the R-MCI might be considered as an integral part in the development of individualized risk-adapted treatment strategies to further improve outcome in MM. This includes the correct use of resources, higher inclusion rates of older patients in clinical studies and the avoidance of an undersupply of older but fit patients. In the future, the R-MCI could help to support treatment decisions, aid in improving tolerability and avoiding toxicity. Since any prospective comorbidity, frailty and disability evaluation in MM can be time-consuming, we have implemented the RMCI within a web-based technology application which allows one to perform the MCI expeditiously (www.myelomacomorbidityindex.org).34 With the retrospective "reference IMWG frailty score”,12 ours is an active website that allows one to calculate a concise MM-specific comorbidity index which can be tested side by side22,25,46 by IMWG and EMN experts. Acknowledgements The authors thank distinguished IMWG, EMN, DSMM and GMMG experts for their advice and recommendations as well as the insightful und inspiring comments of the 3 anonymous reviewers that have helped us to further improve this paper. We are specifically obliged to Jochen Knaus (center for Medical Biometry and Statistics [IMBI], University of Freiburg, Faculty of Medicine, Germany) for performing the structured programming of our RMCI website and Dr. Milena Pantic for FISH cytogenetics. We are also indebted to Dr. Marie Follo for proof reading the manuscript and thank all MM patients who participated in this study. The paper is dedicated to Prof. Dr. Justus Duyster and Prof. Dr. Roland Mertelsmann, University of Freiburg, for their exceptional support. Funding This work is supported by the Deutsche Krebshilfe (grants 1095969 and 111424 [to ME and RW]).

myeloma. Haematologica. 2014;99(2): 232–242. 2. Lohr JG, Stojanov P, Carter SL, et al. Widespread genetic heterogeneity in multiple myeloma: implications for targeted therapy. Cancer Cell. 2014;25(1):91–101. 3. Ludwig H, Sonneveld P, Davies F, et al.

European perspective on multiple myeloma treatment strategies in 2014. Oncologist. 2014;19(8):829–844. 4. Palumbo A, Bringhen S, Ludwig H, et al. Personalized therapy in multiple myeloma according to patient age and vulnerability: a report of the European Myeloma Network

haematologica | 2017; 102(5)


Comprehensive appraisal of a weighted myeloma comorbidity index (EMN). Blood. 2011;118(17):4519–4529. 5. Bringhen S, Mateos MV, Zweegman S, et al. Age and organ damage correlate with poor survival in myeloma patients: meta-analysis of 1435 individual patient data from 4 randomized trials. Haematologica. 2013;98(6): 980–987. 6. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373–383. 7. Kleber M, Ihorst G, Deschler B, et al. Detection of renal impairment as one specific comorbidity factor in multiple myeloma: multicenter study in 198 consecutive patients. Eur J Haematol. 2009;83(6):519–527. 8. Kleber M, Ihorst G, Terhorst M, et al. Comorbidity as a prognostic variable in multiple myeloma: comparative evaluation of common comorbidity scores and use of a novel MM-comorbidity score. Blood Cancer J. 2011;1(9):e35. 9. Kleber M, Ihorst G, Udi J, Koch B, Wäsch R, Engelhardt M. Prognostic risk factor evaluation in patients with relapsed or refractory multiple myeloma receiving lenalidomide treatment: analysis of renal function by eGFR and of additional comorbidities by comorbidity appraisal. Clin Lymphoma Myeloma Leuk. 2012;12(1):38–48. 10. Kleber M, Ihorst G, Gross B, et al. Validation of the Freiburg Comorbidity Index in 466 multiple myeloma patients and combination with the international staging system are highly predictive for outcome. Clin Lymphoma Myeloma Leuk. 2013;13(5): 541– 551. 11. Hieke S, Kleber M, König C, Engelhardt M, Schumacher M. Conditional survival: a useful concept to provide information on how prognosis evolves over time. Clin Cancer Res. 2015;21(7):1530–1536. 12. Palumbo A, Bringhen S, Mateos M-V, et al. Geriatric assessment predicts survival and toxicities in elderly myeloma patients: an International Myeloma Working Group report. Blood. 2015;125(13):2068–2074. 13. Dubruille S, Libert Y, Roos M, et al. Identification of clinical parameters predictive of one-year survival using two geriatric tools in clinically fit older patients with hematological malignancies: Major impact of cognition. J Geriatr Oncol. 2015;6(5):362–369. 14. Ruiz M, Reske T, Cefalu C, Estrada J. Management of elderly and frail elderly cancer patients: the importance of comprehensive geriatrics assessment and the need for guidelines. Am J Med Sci. 2013;346(1): 66–69. 15. Larocca A, Palumbo A. How I treat fragile myeloma patients. Blood. 2015;126(19): 2179–2185. 16. Murthy VH, Krumholz HM, Gross CP. Participation in cancer clinical trials: race-, sex-, and age-based disparities. JAMA. 2004;291(22):2720–2726. 17. Moreau P, Cavo M, Sonneveld P, et al. Combination of international scoring system 3, high lactate dehydrogenase, and t(4;14) and/or del(17p) identifies patients with multiple myeloma (MM) treated with front-line autologous stem-cell transplantation at high risk of early MM progressionrelated death. J Clin Oncol. 2014;32(20): 2173–2180. 18. Neben K, Jauch A, Bertsch U, et al. Combining information regarding chromosomal aberrations t(4;14) and del(17p13)

haematologica | 2017; 102(5)

19.

20. 21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32. 33.

with the International Staging System classification allows stratification of myeloma patients undergoing autologous stem cell transplantation. Haematologica. 2010;95(7):1150–1157. Palumbo A, Avet-Loiseau H, Oliva S, et al. Revised international staging system for multiple myeloma: a report from International Myeloma Working Group. J Clin Oncol. 2015;33(26):2863–2869. Corre J, Munshi N, Avet-Loiseau H. Genetics of multiple myeloma: another heterogeneity level? Blood. 2015;125(12):1870–1876. Bron D, Soubeyran P, Fulop T, SWG “Aging and Hematology” of the EHA. Innovative approach to older patients with malignant hemopathies. Haematologica. 2016;101(8): 893–895. Engelhardt M, Dold SM, Ihorst G, et al. Geriatric assessment in multiple myeloma patients: validation of the International Myeloma Working Group (IMWG) score and comparison with other common comorbidity scores. Haematologica. 2016;101(9): 1110– 1119. Domm A-S, Hieke S, Ihorst G, et al. Importance and determinants of comorbidities, functional limitations and multiple myeloma (MM)-specific risk factors: further development of an improved and weighted MM-risk score (Freiburg Comorbidity Index [FCI]). Blood. 2014;124(21):733–733. Engelhardt M, Ihorst G, Landgren O, et al. Large registry analysis to accurately define second malignancy rates and risks in a wellcharacterized cohort of 744 consecutive multiple myeloma patients followed-up for 25 years. Haematologica. 2015;100(10): 1340– 1349. Engelhardt M, Ihorst G, Caers J, Günther A, Wäsch R. Autotransplants in older multiple myeloma patients: hype or hope in the era of novel agents? Haematologica. 2016;101(11):1276–1278. Wuilleme S, Robillard N, Lodé L, et al. Ploidy, as detected by fluorescence in situ hybridization, defines different subgroups in multiple myeloma. Leukemia. 2005;19(2): 275–278. Ross FM, Avet-Loiseau H, Ameye G, et al. Report from the European Myeloma Network on interphase FISH in multiple myeloma and related disorders. Haematologica. 2012;97(8):1272–1277. Hasskarl J, Ihorst G, De Pasquale D, et al. Association of multiple myeloma with different neoplasms: systematic analysis in consecutive patients with myeloma. Leuk Lymphoma. 2011;52(2):247–259. Bergsagel PL, Mateos M-V, Gutierrez NC, Rajkumar SV, San Miguel JF. Improving overall survival and overcoming adverse prognosis in the treatment of cytogenetically highrisk multiple myeloma. Blood. 2013;121 (6):884–892. Avet-Loiseau H, Attal M, Moreau P, et al. Genetic abnormalities and survival in multiple myeloma: the experience of the Intergroupe Francophone du Myélome. Blood. 2007;109(8):3489–3495. Sekiguchi N, Ootsubo K, Wagatsuma M, et al. Impact of C-Myc gene-related aberrations in newly diagnosed myeloma with bortezomib/dexamethasone therapy. Int J Hematol. 2014;99(3):288–295. Rodriguez-Mañas L, Fried LP. Frailty in the clinical scenario. Lancet. 2015;385(9968):e7-9. Xue Q-L, Walston JD, Fried LP, Beamer BA.

34.

35.

36.

37.

38.

39.

40.

41.

42. 43.

44.

45.

46.

47.

48.

Prediction of risk of falling, physical disability, and frailty by rate of decline in grip strength: the women’s health and aging study. Arch Intern Med. 2011;171(12): 1119–1121. Engelhardt M, Dold SM, Ihorst G, Knaus J, Schumacher M. R-MCI webpage [Internet]. 2015. Available from: http://www.myelomacomorbidityindex.org/de_main.html Sorror ML, Maris MB, Storb R, et al. Hematopoietic cell transplantation (HCT)specific comorbidity index: a new tool for risk assessment before allogeneic HCT. Blood. 2005;106(8):2912–2919. Gerds TA, Schumacher M. Consistent estimation of the expected Brier score in general survival models with right-censored event times. Biom J. 2006;48(6):1029–1040. Terpos E, Kleber M, Engelhardt M, et al. European Myeloma Network guidelines for the management of multiple myeloma-related complications. Haematologica. 2015;100(10):1254–1266. Offidani M, Corvatta L, Polloni C, et al. Assessment of vulnerability measures and their effect on survival in a real-life population of multiple myeloma patients registered at Marche Region Multiple Myeloma Registry. Clin Lymphoma Myeloma Leuk. 2012;12(6):423–432. Kos FT, Yazici O, Civelek B, et al. Evaluation of the effect of comorbidity on survival in pancreatic cancer by using “Charlson Comorbidity Index” and “Cumulative Illness Rating Scale.” Wien Klin Wochenschr. 2014;126(1–2):36–41. Saussele S, Krauss M-P, Hehlmann R, et al. Impact of comorbidities on overall survival in patients with chronic myeloid leukemia: results of the randomized CML study IV. Blood. 2015;126(1):42–49. Avet-Loiseau H, Durie BGM, Cavo M, et al. Combining fluorescent in situ hybridization data with ISS staging improves risk assessment in myeloma: an International Myeloma Working Group collaborative project. Leukemia. 2013;27(3):711–717. Sorror ML. How I assess comorbidities before hematopoietic cell transplantation. Blood. 2013;121(15):2854–2863. Kallogjeri D, Gaynor SM, Piccirillo ML, Jean RA, Spitznagel EL, Piccirillo JF. Comparison of comorbidity collection methods. J Am Coll Surg. 2014;219(2):245–255. Etienne A, Esterni B, Charbonnier A, et al. Comorbidity is an independent predictor of complete remission in elderly patients receiving induction chemotherapy for acute myeloid leukemia. Cancer. 2007;109(7): 1376–1383. DeFor TE, Majhail NS, Weisdorf DJ, et al. A modified comorbidity index for hematopoietic cell transplantation. Bone Marrow Transplant. 2010;45(5):933–938. Dold SM, Zober A, Pantic M, et al. Prospective comorbidity and functional geriatric assessment in multiple myeloma patients: results from a multicenter German Study Group MM (DSMM) trial. Haematologica. 2016;101(7):2016. Kristinsson SY, Pfeiffer RM, Björkholm M, Schulman S, Landgren O. Thrombosis is associated with inferior survival in multiple myeloma. Haematologica. 2012;97(10):1603– 1607. Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci. 2001;56(3): M146-156.

921


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Plasma Cell Disorders

Ferrata Storti Foundation

Haematologica 2017 Volume 102(5):922-931

Recovery of polyclonal immunoglobulins one year after autologous stem cell transplantation as a long-term predictor marker of progression and survival in multiple myeloma Verónica González-Calle,1 Seila Cerdá,2 Jorge Labrador,3 Eduardo Sobejano,1 Beatriz González-Mena,4 Carmen Aguilera,5 Enrique María Ocio,1 María Belén Vidriales,1 Noemí Puig,1 Norma Carmen Gutiérrez,1 Ramón García-Sanz,1 José María Alonso,6 Rosa López,7 Carlos Aguilar,8 Alfonso García de Coca,9 Roberto Hernández,10 José Mariano Hernández,11 Fernando Escalante2 and María-Victoria Mateos1

Complejo Asistencial Universitario de Salamanca/Instituto de Investigación Biomédica de Salamanca (CAUSA/IBSAL); 2Complejo Asistencial de León; 3Complejo Asistencial de Burgos; 4Hospital Nuestra Señora de Sonsoles de Ávila; 5Hospital Del Bierzo Ponferrada; 6 Hospital Río Carrión de Palencia; 7Hospital Virgen del Puerto de Plasencia; 8Complejo Asistencial de Soria; 9Hospital Clínico de Valladolid; 10Complejo Asistencial de Zamora and 11Complejo Asistencial de Segovia, Spain

1

ABSTRACT

I

Correspondence: mvmateos@usal.es

Received: October 14, 2016. Accepted: January 11, 2017. Pre-published: January 25, 2017. doi:10.3324/haematol.2016.158345 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/922 ©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.

922

mmunoparesis or suppression of polyclonal immunoglobulins is a very common condition in newly diagnosed myeloma patients. However, the recovery of polyclonal immunoglobulins in the setting of immune reconstitution after autologous stem cell transplantation and its effect on outcome has not yet been explored. We conducted this study in a cohort of 295 patients who had undergone autologous transplantation. In order to explore the potential role of immunoglubulin recovery as a dynamic predictor of progression or survival after transplantation, conditional probabilities of progression-free survival and overall survival were estimated according to immunoglobulin recovery at different time points using a landmark approach. One year after transplant, when B-cell reconstitution is expected to be completed, among 169 patients alive and progression free, 88 patients (52%) showed immunoglobulin recovery and 81 (48%) did not. Interestingly, the group with immunoglobulin recovery had a significantly longer median progression-free survival than the group with persistent immunoparesis (median 60.4 vs. 27.9 months, respectively; Hazard Ratio: 0.45, 95%Confidence Interval: 0.31-0.66; P<0.001), and improved overall survival (11.3 vs. 7.3 years; Hazard Ratio: 0.45, 95%Confidence Interval: 0.27-0.74; P=0.002). Furthermore, the percentage of normal plasma cells detected by flow cytometry in the bone marrow assessed at day 100 after transplantation was associated with the immunoglobulin recovery at that time and may predict immunoglobulin recovery in the subsequent months: nine months and one year. In conclusion, the recovery of polyclonal immunoglobulins one year after autologous transplantation in myeloma patients is an independent longterm predictor marker for progression and survival.

Introduction High-dose therapy followed by autologous stem cell transplantation (ASCT) remains the standard of care for young, newly diagnosed multiple myeloma (MM) patients. It produces high rates of complete remission, and prolonged progressionfree survival (PFS) and overall survival (OS).1-3 However, MM is still an incurable disease, with a high rate of relapse or progression after ASCT.4 In recent years, several prognostic factors have been identified that predict outcomes after ASCT, such haematologica | 2017; 102(5)


Immunoglobulin recovery after ASCT in MM

as treatment response, persistence (or not) of minimal residual disease (MRD) or cytogenetic abnormalities. In addition, researchers have shown an increased interest in immune reconstitution after this procedure, based on the premise that an early, strong and sustained recovery of the immune system could help eliminate residual myeloma plasma cells (PCs), thereby improving the final outcome. In fact, several studies have reported positive impacts of early lymphocyte recovery, the presence of oligoclonal bands or an early reconstitution of natural killer (NK) cells on survival.5-8 Most MM patients (85%-90%) exhibit immunoparesis at the time of diagnosis.9 This condition is defined as a reduction in the levels of polyclonal or uninvolved immunoglobulins (Igs).10 Several mechanisms are thought to be involved in immunoparesis, such as impaired B-cell differentiation related to humoral and cellular immune dysfunction, and the reduction of the quantity of B lymphocytes due to cytokines produced by myeloma cells (TGF-β).11 Moreover, it seems that this B-cell suppression is reversible and inversely correlated with disease stage.12 In fact, the presence of immunoparesis in smoldering myeloma is considered to be a prognostic marker of progression to symptomatic myeloma,13 and is also associated with adverse outcome in newly diagnosed symptomatic myeloma patients.9 In the ASCT setting, after high doses of melphalan and the infusion of stem cells, an immune reconstitution is expected, including the reappearance of functional B lymphocytes and, thereby, the recovery of polyclonal Igs.14 On this basis, we hypothesized that persistence of immunoparesis after ASCT may predict worse progression or survival in patients with MM, similarly to the other, aforementioned markers of immune dysfunction. The primary goal of this study was to determine whether the recovery of polyclonal Igs after ASCT is of prognostic value in a cohort of patients with MM. We show that the immunoglobulin (Ig) recovery one year after ASCT is an independent prognostic factor associated with longer PFS and OS in MM patients undergoing this procedure.

Methods We retrospectively evaluated patients from our region diagnosed with symptomatic MM, according to the 2003 criteria, who consecutively underwent ASCT at either of two referral centers, the University Hospitals of Salamanca and of León, Spain, between 1993 and 2014. Clinical and biological data were collected by searching the medical records and databases of each hospital. Serum immunoglobulin levels (IgG, IgA and IgM) were measured by nephelometry. Immunoparesis was defined as a more than 25% decrease in one or both polyclonal Igs relative to the lowest limit of normality of each laboratory. The recovery of the Igs was established as a normalization of polyclonal Igs levels (presence of polyclonal IgG, IgA, IgM serum level above the minimum level of the normal range cited by each laboratory). Igs were collected at different time points (± 7 days): diagnosis, before ASCT, and 100 days, 6 months, 9 months, 1 year, 18 months after ASCT, and annually thereafter, until relapse, progression or death. In the group of patients who received tandem ASCT, Igs were evaluated after the second ASCT. However, patients who underwent tandem auto/allo-stem cell transplantation were excluded because this procedure may interfere with the pure autologous-immune haematologica | 2017; 102(5)

reconstitution; those patients without follow up at 100 days were also excluded. Response to treatment was evaluated according to 2006 response criteria for MM.15 In addition, phenotypically aberrant bone marrow plasma cells (aPCs) and normal bone marrow plasma cells (nPCs) were assessed by multiparameter flow cytometry (MFC), as previously described.16 Flow MRD-negative assessed 100 days after ASCT was defined as the absence of aPCs.17 Fluorescence in situ hybridization (FISH) analysis was performed in selected CD138 plasma cells in the bone marrow (BM) samples at diagnosis, as previously described.18,19

Statistical analyses

The χ2, Student t-test and Mann-Whitney U tests were used to establish statistically significant differences between comparison groups. P<0.05 was considered statistically significant. PFS in the whole patient cohort was defined as the time from date of transplantation to relapse, progression or death, regardless of cause. OS was considered the time from transplantation to death. Patients without a recorded progression or death date were censored for PFS or OS at their last follow up. These probabilities were estimated by the Kaplan-Meier method. Ig recovery was evaluated at different time points, mentioned above, until progression or death. To explore whether Ig recovery has a prognostic role for each of those moments, and to obtain a dynamic prediction, we calculated conditional survival (CS) probabilities using the landmark approach;20,21 we estimated PFS or OS according to Ig recovery, given that the patient was already alive and progression free at those landmark time points, which correspond to Ig-evaluation time points. Thus, only patients who were still alive and without progression at such landmark times were included in the respective analyses. Survival curves were plotted by the Kaplan-Meier method and calculated from the landmark time point, with differences assessed by the log-rank test. To explore the effects of potential risk factors for progression or survival, Cox proportional hazards regression model was used. Hazard Ratios (HR) were also estimated by conditional versions of the Cox regression model for the different landmark time points. In addition, analysis took into consideration the whole follow up after ASCT by treating Ig recovery as a time-varying covariate; this new time-varying covariate was then incorporated into the final multivariate model. All statistical analyses were performed using IBM SPSS Statistics for Windows, v.20.0 (IBM Corp., Armonk, NY, USA). The study was approved by the Institutional Review Board of one of the participating centers, in accordance with the Declaration of Helsinki.

Results Patients’ characteristics A total of 342 MM patients underwent ASCT between 1993 and 2014. A total of 295 patients met the inclusion criteria and were included in this study (Figure 1). Their baseline characteristics are summarized in Table 1. There were 171 (58%) men and 124 (42%) women. Median age at diagnosis was 57 years (range 29-71 years). Conventional chemotherapy was administered as an induction regimen in 163 patients (55%); 137 (46%) received VBMCP/VBAD (vincristine, BCNU, melphalan, cyclophosphamide, prednisone/vincristine, BCNU, doxorubicin, dexamethasone) and 24 (8%) received VAD (vincristine, adriamycin and dexamethasone). The remaining 132 patients received immunomodulatory drugs (IMIDs) 923


V. González-Calle et al.

or proteasome inhibitor-based therapies: 46 patients (16%) had received VD (bortezomib and dexamethasone) and 37 patients (13%) VTD (bortezomib, thalidomide and dexamethasone). According to the International Staging System (ISS),21

fifty-nine (20%) patients were categorized as having stage III. FISH studies were carried out at diagnosis in 206 patients, 45 of whom (22%) were classified as having high-risk cytogenetic abnormalities: 25 (12%) had t(4;14), 15 (7%) had del17p, and 5 (2%) had t(14;16).

Table 1. Baseline characteristics of 295 transplant-eligible myeloma patients and treatments received before and after autologous stem cell transplant (1993-2014).

Characteristics Male / female, n. (%) Age at diagnosis, median, years (range) Heavy chain type, n. (%) IgG IgA BJ Non-secretory MM Ig D Light chain type kappa, n. (%) lambda, n. (%) Serum M-protein, median mg/dL (range) % BM PC by morphology, mean (SD) Hemoglobin, mean g/dL (SD) Creatinine, mean mg/dL (SD) Calcium, mean mg/dL (SD) β2 microglobulin, mean mg/dL (SD) Immunoparesis at diagnosis, n. (%) NA ISS stage, n. (%) I II III NA High-risk cytogenetic, n. (%) t(4;14) del 17 p t(14;16) NA Induction treatment, n. (%) Conventional chemotherapy VBCMP/VBAD VAD Others Novel agents VD VTD VCD Others (TD, VRD, RD, VDL-PACE) Maintenance therapy Interferon-α Other combinations (bortezomib, thalidomide, lenalidomide) NA

Myeloma patients (n=295) 171 (58)/124 (42) 57 (29-71) 173 (59) 60 (20) 45 (15) 14 (5) 3 (1) 179 (61) 113 (39) 3.4 (0-12.4) 36 (25) 10.9 (2.2) 1.4 (1.3) 9.7 (1.7) 4.4 (3.5) 208 (84) 48 115 (46) 85(34) 49 (20) 46 45 (22) 25 (12) 15 (7) 5 (2) 89 163 (55) 137 (84) 24 (15) 2 (1) 132 (45) 46 (36) 37 (28) 10 (6) 39 (30) 141 (57) 113 (80) 28 (20) 49

BJ: Bence Jones myeloma; MM: multiple myeloma; n: number; BMPC: bone marrow plasma cells; ISS: International Staging System; SD: standard deviation; NA: not available; VBCMP/VBAD: vincristine, BCNU, melphalan, cyclophosphamide, prednisone/vincristine, BCNU, doxorubicin, dexamethasone; VAD: vincristine, adriamycin and dexamethasone; VD: bortezomib, dexamethasone; VTD: bortezomib, thalidomide, dexamethasone; VCD: bortezomib, cyclophosphamide, dexamethasone; TD: thalidomide and dexamethasone; VRD: bortezomib, lenalidomide, dexamethasone; RD: lenalidomide, dexamethasone; VDL-PACE: bortezomib, dexamethasone, lenalidomide, cisplatin, adriamycin, cyclophosphamide, etoposide.

924

haematologica | 2017; 102(5)


Immunoglobulin recovery after ASCT in MM

Figure 1. Kinetics of the polyclonal immunoglobulin (Ig) recovery after autologous stem cell transplantation (ASCT) in 295 myeloma patients (1993-2014). A total of 295 patients were included in the study, after excluding 47 patients who underwent tandem auto-alloSCT or those who were lost to follow up before day +100. Evaluable patients with available Igs at each time during the study are represented on the right. The gray box on the right shows the number of patients who had recovered polyclonal Igs by the different times since ASCT (center boxes). The proportion of patients with Ig recovery increases over time, indicating progressive Ig recovery after ASCT (represented by the descending black arrow) among the evaluable patients. Some patients did not have available Igs at various times (data not shown): 20 patients at 100 days (d); 71 at six months (mo); 71 at nine mo; 62 at one year (yr); 55 at two years; 60 at three years; and 57 patients at five years. The cumulative numbers of patients who had progressed or died at each time point, and who were therefore excluded from the Ig evaluation, are shown on the left.

ASCT features and treatment response The median time from diagnosis to ASCT was eight months (range 3-186 months); 200 mg/m2 melphalan was the standard conditioning regimen used for the majority of patients. The median infused CD34+ stem cell dose was more than 2 x106/kg. Only one case of engraftment failure was recorded during this period (Online Supplementary Table S1). ASCT improved the overall response rate from 90% before ASCT to 94% after the procedure, as well as the quality of response: 106 (36%) patients showed an improved response. As a result, the complete response (CR) rate, including stringent CR (sCR), improved from 27% before ASCT to 48% after the procedure.

Kinetics of polyclonal immunoglobulin recovery and association with depth of response after transplantation Most patients (208 patients, 84%) had immunoparesis at diagnosis, and this was associated with aggressive disease characteristics: renal impairment (P=0.004), IgA subtype (P=0.04), 40% or more BMPCs (P<0.001), and advanced ISS stage (P=0.004). Figure 1 provides an illustrative explanation of the prohaematologica | 2017; 102(5)

portion of patients who had Ig recovery at each time point after ASCT during the study; 100 days after ASCT, 58 of 263 (22%) evaluable patients had recovered polyclonal Igs, while the remaining 205 (78%) had immunoparesis. There were no significant differences between the two groups (with immunoparesis or Ig recovery) after 100 days with respect to sex, age, induction, double ASCT, cytogenetic or early neutrophil engraftment (Table 2). However, in the group with polyclonal Ig recovery there was a trend towards more complete responses [10 patients (17%) in sCR and 26 (45%) in CR] and fewer partial responses (PR) (21%) achieved by 100 days than in the group with immunoparesis persistence 100 days after transplantation [20 (10%), 73 (35%), 65 (32%) patients in sCR, CR and PR, respectively] (Table 2). None of the patients who recovered Igs had progressed in their disease or showed no response by 100 days. Moreover, there was more flow MRD-negative after 100 days among patients who had recovered Igs than in the group with immunoparesis: 20 of 58 (34%) versus 48 of 205 (23%), respectively (P=0.08). One year after ASCT, 169 patients were evaluable and 81 of them had immunoparesis (48%) while the remaining 88 (52%) had experienced Ig recovery during the first 925


V. Gonzรกlez-Calle et al. A

B

Figure 2. Box plots illustrating the association between total percentage of normal plasma cells (nPCs) in the bone marrow (BM) flow assessment at 100 days and subsequent immunoglobulin (Ig) recovery nine months (A) and one year (B) after transplantation. (A) Box plots showing the distribution of nPCs in the BM assessment 100 days by Ig recovery after nine months. Note that the group who recovered polyclonal Igs after nine months (left) had shown a significantly higher median percentage of nPCs in the previous BM assessment at 100 days: 0.11% versus 0.06%. (B) Box plots showing the distribution of nPCs after 100 days in the Ig recovery groups after one year. Patients who had recovered polyclonal Igs by one year after transplantation also had shown higher median percentages of nPCs at 100 days: 0.10% versus 0.08% nPCs, respectively. Therefore, the quantity of nPCs in the BM assessment after 100 days can predict subsequent Ig recovery after transplantation.

year since ASCT: 34 of these 88 patients (39%) had recovered Igs at 100 days, 16 (18%) at six months, 15 (17%) at nine months and 23 (26%) one year after ASCT. Therefore, there was a progressive Ig recovery after ASCT in those patients who were alive and without progression at one year. No significant differences in any baseline characteristics were found between these groups (Table 2). In order to determine whether the recovery of nPCs is correlated with serum Ig recovery, we compared the percentage of nPCs in the bone marrow assessed by MFC after 100 days, performed in 212 patients, with the Ig recovery at various times after ASCT. As expected, the median percentage of nPCs in the plasma cell bone marrow compartment was higher in the group of 46 patients who had recovered Igs than in the group with immunoparesis after 100 days: 85.4% versus 68.2% nPCs, respectively (P=0.004). In addition, patients who recovered the Igs later, by nine months or one year after ASCT, had shown a higher median percentage of 100-day nPCs with respect to whole bone marrow cellularity, than those who had persistent immunoparesis at those times: 0.11% versus 0.06% (P=0.003) and 0.10% versus 0.08% (P=0.013), respectively (Figure 2). Moreover, all patients who lacked nPCs in the BM after 100 days still exhibited immunoparesis six and nine months and one year later, except for 2 patients (19%) who finally recovered Igs after one year. Therefore, the percentage of nPCs after 100 days may predict subsequent Ig recovery at various time points after ASCT and absence of nPCs may predict the persistence of immunoparesis in subsequent months.

Impact on survival of immunoglobulin recovery one year after transplantation Median follow up for surviving patients was 59.7 months (range 7.3-301.1 months); 221 out of 295 patients (70%) progressed, relapsed or died after ASCT, with a median PFS of 30.2 months [95%Confidence Interval (CI): 926

25.9-34.5 months] from ASCT and a median OS for the whole cohort of patients of 7.4 years (95%CI: 6.2-8.5 years) from ASCT. Conditional PFS and OS were estimated at each landmark time point according to Ig recovery. Although there were no statistically significant differences between the groups with respect to Ig recovery at 100 days, six months or nine months, the median PFS tended to be slightly higher in the recovery than in the immunoparesis group: 36 versus 28 months, 41 versus 32 months and 50 versus 32 months, respectively, for each landmark time point (P=0.3). However, statistically significant differences in PFS and OS were found from the 1-year landmark time point (Online Supplementary Table S2). Altogether, a total of 169 patients with available Ig data were alive and progression free one year after ASCT. Median follow up for patients with Ig recovery was 78.8 months and 85.1 months for patients who had not recovered Igs (P=0.2). Interestingly, median PFS was significantly longer for the 88 patients with Ig recovery than for the 81 patients without Ig recovery according to the 1-year landmark analysis: 60.4 versus 27.9 months, respectively (HR: 0.45, 95%CI: 0.31-0.66; P<0.001) (Figure 3). We also explored whether the timing of Ig recovery during the first year among these 169 patients who were alive and progression free had an impact on PFS. Another landmark analysis of one year was performed for PFS according to the period of time when the Ig recovery had occurred within the previous 12 months, identifying three groups with different PFS: i) group 1 included those who had recovered the Igs within the first six months after ASCT; ii) group 2 included those who had recovered the Igs 6-12 months after ASCT; and iii) group 3 included those patients with no Ig recovery one year after ASCT. The shorter the Ig recovery time the longer was the PFS (69.3 vs. 52.9 vs. 27.9 months for groups 1, 2 and 3, respectively; P<0.001) (Figure 4). haematologica | 2017; 102(5)


Immunoglobulin recovery after ASCT in MM

Table 2. Distribution of baseline and disease characteristics by immunoglobulin recovery status 100 days and one year after autologous stem cell transplantation (n=263 and n=169).

Characteristics

Age at diagnosis, â&#x2030;Ľ65 year, n. (%) <65 year, n. (%) BM PC by morphology, mean % (SD) Serum M-protein, mean mg/dL (SD) High-risk cytogenetics, n. (%) NA ISS stage, n. (%): I II III NA Treatment induction, n. (%) Conventional chemotherapy Novel agents ASCT tandem, n. (%) Response after 100 days, n. (%) PD NR PR VGPR CR sCR Flow negative-MRD, n. (%) Maintenance therapy with interferon-Îą NA

Immunoparesis: 100 days n=205

Ig recovery: Ig 100 days n=58

P

Immunoparesis: 1 year n=81

Ig recovery: 1 year n=88

P

42 (20) 160 (80) 35 (25.6) 3.4 (2.6) 31 (20) 31

12 (21) 46 (79) 39 (24.6) 3.2 (2.6) 10 (24) 17

0.9

19 (23) 62 (77) 34 (21.4) 3.4 15 (23) 17

14 (16) 71 (84) 35 (27.6) 3.1 10 (14) 19

0.3

79 (45) 64 (36) 33 (19) 29

23 (46) 16 (32) 11 (22) 8

35 (46) 27 (36) 14 (18) 5

44 (53) 26 (31) 13(16) 5

110 (54) 95 (46) 18 (9)

30 (52) 28 (48) 2 (3)

46 (53) 40 (47) 7 (9)

43 (49) 45 (51) 4 (5)

0.5

5 (2) 8 (4) 65 (32) 34 (17) 73 (35) 20 (10) 48 (23)

0 0 12 (21) 10 (17) 26 (45) 10 (17) 20 (34)

-

-

-

1 (1) 24 (27) 17 (19) 31 (35) 15 (17) 24 (27) 40 (45) 6

0.1

0.08

5 (6) 23 (28) 14 (17) 31 (38) 8 (10) 19 (23) 45 (55) 2

0.3 0.6 0.5

0.8

0.8 0.2 0.1

0.7 0.5 0.2

0.7

0.3

0.5 0.3

Ig: immunoglobulin; BMPC: bone marrow plasma cells; NA: not available; ISS: International Staging System; PD: progressive disease; NR: stable disease; PR: partial response; VGPR: very good partial response; CR: complete response; sCR: stringent complete response; MRD: minimal residual disease.

Furthermore, median OS was significantly longer for the group with Ig recovery than for the group with persistent immunoparesis from the 1-year landmark point (11.3 vs. 7.3 years, P<0.001; HR: 0.45, 95%CI: 0.27-0.74, P=0.002) (Figure 5). Conditional versions of the Cox model for 100-days and 1-year landmark time points were made taking into consideration only patients who were alive and progression free at those moments after ASCT. Altogether, 4 multivariate models were performed, for both PFS and OS at each landmark point (100 days and 1 year). Covariates significantly associated with PFS and OS were identified by univariate analysis. P=0.05 was considered statistically significant. Multivariate analysis was then performed including only significant factors obtained in the univariate analysis. Finally, an additional model treating Ig recovery as a timevarying covariate was performed. The conditional version of the Cox model for 100 days, including 283 patients alive and progression free at this landmark time point is shown in Online Supplementary Table S3. A total of 138 patients who simultaneously had all the covariates were evaluated. Neutrophil engraftment before ten days was selected as an independent predictor for PFS, high-risk cytogenetic abnormalities were shown to be an independent predictor for both PFS and OS, and haematologica | 2017; 102(5)

presence of renal impairment at diagnosis for OS. Altogether, 231 patients were alive and progression-free at one year after ASCT. In all, 134 patients were included in the multivariate analysis for PFS (Table 3) showing ISS stage III, neutrophil engraftment before ten days and Ig recovery at one year were independent factors for predicting PFS at this 1-year landmark time point. Patients with Ig recovery at one year had a 2-fold lower risk of progression or death from one year after ASCT than those who had not recovered Igs (HR 0.5, 95%CI: 0.3-0.8; P=0.001). In addition, 127 patients were included in the multivariate analysis for OS and Ig recovery was also selected as an independent predictor for OS at this landmark-point, (HR: 0.35, 95%CI: 0.2-0.7; P=0.004). We incorporated Ig recovery as a time-varying covariate into the multivariate model and findings support the results already found by the landmark analysis: Ig recovery after ASCT was an independent predictor for PFS (Online Supplementary Table S4).

Discussion This retrospective study shows that polyclonal immunoglobulin recovery occurs gradually after trans927


V. González-Calle et al.

Progression-free survival probability

1.0

0.8

0.6 Figure 3. Kaplan-Meier curves for conditional progression-free survival (PFS) from the landmark time point of one year after autologous stem cell transplantation (ASCT) according to immunoglobulin (Ig) recovery. Estimated probability of PFS conditional on being alive and progression-free one year after ASCT, according to Ig recovery at this landmark time point (represented with a vertical line intersecting 12 months). There were 169 patients at risk, indicated below the figure, corresponding to those alive, progression-free and not censored at this landmark time point; 88 of 169 had Ig recovery and a median PFS significantly longer than those 81 patients who had not recovered Ig at this landmark time point. mo.: months; OS: overall survival; HR: hazard ratio.

0.4

0.2

0.0

Table 3. Univariate and multivariate analysis of covariates affecting progression-free survival and overall survival by conditional version of the Cox regression model for the one year landmark point.

Covariates

ISS III I or II NA Cytogenetic risk High Standard NA Induction Conventional Novel agents NA PMN engraftment ≤10day >10day NA Response 100 days CR Not CR NA Ig recovery at 1 year Yes No NA

N (%)

PFS since 1 year after ASCT Median Univ. Multivariate (months) P HR (95%CI)

P

Median (years)

OS since 1 year after ASCT Univ. Multivariate P HR (95%CI)

P

35 (18) 163 (82) 33

25.6 41.7

0.006

2.1 (1.2-3.6) Ref

0.01

4.9 9.8

0.005

1.1 (0.3-3.6) Ref

NS

28 (17) 138 (83) 65

31.6 43.3

NS

6.8 11.3

0.02

2.0 (0.9-4.5) Ref

NS

131 (56) 100 (44) −

40.9 48.4

NS

8.3 NR

0.02

1.8 (0.8-3.7) Ref

NS

28 (14) 174 (86) 29

67.8 36.6

0.006

0.4 (0.2-0.7) Ref

0.004

11.7 8.4

NS

113 (49) 118 (51) −

42.4 32.6

0.06

10.9 8.3

NS

88 (52) 81 (48) 62

60.4 27.9

<0.0001

0.5 (0.3-0.8) Ref

0.001

11.3 7.9

0.001

0.35 (0.2-0.7) Ref

0.004

ASCT: autologous stem cell transplantation; PFS: progression-free survival; mo: months; HR: Hazard Ratio; CI: Confidence Interval; univ: univariate analysis; HR: Hazard Ratio; OS: overall survival; ISS: International Staging System; PMN: neutrophils; CR: complete response; Ig: immunoglobulin; Ref: reference category; NA: data not available; NS: not significant; NR: not reached.

928

haematologica | 2017; 102(5)


Immunoglobulin recovery after ASCT in MM

Progression-free survival probability

1.0

0.8

0.6

0.4

0.2

0.0

plantation and that recovery one year after ASCT is an independent prognostic factor predicting longer PFS and OS, when the B-cell reconstitution is expected to be completed. In addition, the presence of nPCs in the BM 100 days after ASCT is associated with early recovery of Igs by 100 days and with subsequent Ig recovery after nine months and one year. Therefore, we propose that Ig levels should be measured during follow up of patients undergoing ASCT. To the best of our knowledge, this is the first study to evaluate the presence of immunoparesis after ASCT, as well as the kinetics of polyclonal Ig recovery and its effect on outcome after ASCT. The serum Ig findings are consistent with the biological background described by Hernรกndez et al.22 and Rueff et al.8 B-cell reconstitution is a delayed and progressive process beginning one month after ASCT, reaching a normal range at six months, and ending after one year when maximum B-lymphocyte levels are detected in BM. In this context, several observations in the study are worthy of discussion. 1) Considering the 1-year period required for complete B-cell reconstitution after ASCT, we observed that 88 patients (52%) had recovered Igs by this time. One-third of those patients had already recovered their polyclonal Igs by 100 days, half of them had recovered by six months and the other half did so between six months and one year. This timing of polyclonal Ig recovery has a prognostic value in terms of PFS, reflecting the potential benefit of early immune recovery. Patients with polyclonal immunoglobulin recovery within the first six months following ASCT had significantly longer PFS than those who recovered during the next six months. In addition, engraftment of neutrophils within ten days was significantly associated with longer PFS. Several studies have also shown that rapid immune reconstitution after ASCT, both early lymphocyte recovery5,6 and higher levels of NK haematologica | 2017; 102(5)

Figure 4. Kaplan-Meier curves for conditional progression-free survival (PFS) from the landmark time point of one year according to timing of immunoglobulin (Ig) recovery within the first year after autologous stem cell transplantation (ASCT). Again, the probability of PFS was estimated by restricting to patients who were alive and progression-free one year after ASCT, according to the Ig recovery period. Group 1 recovered Ig before six months since ASCT, group 2 between 6-12 months after ASCT, and Group 3 had not recovered Ig at one year. Median PFS was longer for the first group, who had an earlier Ig recovery. mo.: months; OS: overall survival; HR: Hazard Ratio.

cells after one month8 have a significantly positive impact on outcomes, probably due to the immune effect on residual myeloma PCs. 2) As far as nPCs are concerned, we show that a higher proportion of nPCs after 100 days was significantly associated with early polyclonal Ig recovery, although it may also predict Ig recovery in subsequent months; by contrast, the absence of nPCs after 100 days was associated with persistence of immunoparesis one year after ASCT. With respect to abnormal PCs, although immunoparesis mechanisms are not completely understood, our results also suggest that B cells are suppressed by the plasma cell clone: Ig recovery was more common in patients without aPCs or flow MRD-negative after 100 days. The findings of Tovar et al.7 provide additional evidence of a humoral response after ASCT, revealing that emergence of oligoclonal bands could be the consequence of the strong immune reconstitution that is associated with better PFS, and suggesting that there is clonal competition between myeloma PCs and polyclonal B lymphocytes. Furthermore, a sustained oligoclonal response, lasting for more than one year after ASCT, also had a positive influence on the outcomes. Finally, the most interesting finding in our study is probably that polyclonal Ig recovery one year after ASCT was associated with significantly longer PFS and OS than in those with persistent immunoparesis: median PFS of 60 versus 28 months and OS of 11 versus 7 years, respectively. However, this significant association was not evident earlier (after 100 days). One possible explanation is that the prognostic significance of the polyclonal Ig recovery could be established only in those patients who lived long enough to have experienced complete and uneventful Bcell reconstitution one year after ASCT. Therefore, if the polyclonal Igs have recovered by this time, our results would lead us to expect a positive outcome. By contrast, persistence of immunoparesis at this time was independ929


V. González-Calle et al.

1.0

Overall survival

0.8

0.6

0.4

0.2

0.0

ently associated with shorter PFS and worse OS. As a result, polyclonal Ig recovery after one year may be considered an independent long-term marker for predicting PFS and OS. Our risk-reassessment approach involves a non-invasive strategy that could be easily implemented in clinical practice. In addition, Ig quantification by standard nephelometry is a quick and highly reproducible method, at relatively low cost,23 and is widely available, compared with serum Ig heavy/light chain ratio (HLC) assays. Some recent studies have reported that HLC is a predictor of PFS in MM patients at diagnosis24 and after ASCT.25 However, further studies are required because only one of these was conducted after ASCT, and the association with treatment response or the kinetics of HLC recovery has not yet been established. Despite there being no definitive recommendations regarding consolidation and maintenance treatment for MM patients after ASCT,26 strategies that enhance the immune reconstitution might be beneficial. In fact, interferon maintenance significantly improved OS in those patients in our series who tolerated the treatment. A recent immunotherapy study showed that patients with persistent positive MRD after treatment showed upregulation of PD-L1/PD-1, suggesting that this group of patients may benefit from PD1-blockade with anti-PD1 drugs.27 In accordance with this, patients with persistent immunoparesis and absence of nPCs are a suitable cohort in which to investigate immunotherapy strategies in clinical trials that aim to enhance their immune system and subsequently achieve immune-mediated eradication of myeloma cells. However, further prospective studies are required to analyze in greater detail the impact of polyclonal Ig recovery and the immune system background after transplantation in the era of new drugs. The presence of high-risk cytogenetic abnormalities stood out in our study as one of the most important inde-

930

Figure 5. Kaplan-Meier curves for conditional overall survival (OS) from the landmark time point of one year after autologous stem cell transplantation (ASCT) according to immunoglobulin (Ig) recovery. Estimated probability of survival given that a patient has already survived or was not censored at one year after ASCT according to the Ig recovery. Median OS for the group with Ig recovery at this landmark time point was significantly longer than the median OS for the 81 patients without Ig recovery. yrs.: years; CI: Confidence Interval; HR: hazard ratio.

pendent prognostic factors for progression and survival in myeloma patients, as noted in other series.17,19,28 Interestingly, Ig recovery after one year may also help identify patients with better subsequent long-term outcomes among those high-risk patients who live for more than one year after transplantation and who have not progressed. In conclusion, this study, carried out in a representative series of MM patients, showed that the recovery of polyclonal Igs one year after ASCT, when B-cell reconstitution is expected to have concluded, had occurred in half of the patients and was an independent long-term marker of progression and survival. This recovery of Igs was a gradual process following ASCT that could be predicted on the basis of the percentage of underlying nPCs detected by flow cytometry in the bone marrow assessment after 100 days. If these results were confirmed in other studies, they might facilitate the selection of candidate patients requiring consolidation/maintenance therapy after ASCT, and even the establishment of immunotherapy strategies to enhance their immune system and improve their outcomes in the setting of clinical trials. Acknowledgments The authors would like to thank all the investigators of the Cooperative Group for the study of Monoclonal Gammopathies of Castilla y León; the patients for their participation in this study; Llorenç Badiella for assistance with statistical analysis and Phill Mason for assistance with English language. Funding Eduardo Sobejano was supported by the Fundación AMIR (Convocatoria Becas de Investigación Pregrado Academia AMIR 2014). Veronica Gonzalez-Calle was supported by the Fundación Española de Hematología y Hemoterapia and Janssen (Beca Estancias de Investigación en el Extranjero Convocatoria 2015-2016).

haematologica | 2017; 102(5)


Immunoglobulin recovery after ASCT in MM

References 1. Attal M, Harousseau JL, Stoppa AM, et al. A prospective, randomized trial of autologous bone marrow transplantation and chemotherapy in multiple myeloma. Intergroupe Francais du Myelome. N Engl J Med. 1996;335(2):91-97. 2. Child JA, Morgan GJ, Davies FE, et al. Highdose chemotherapy with hematopoietic stem-cell rescue for multiple myeloma. N Engl J Med. 2003;348(19):1875-1883. 3. Palumbo A, Cavallo F, Gay F, et al. Autologous transplantation and maintenance therapy in multiple myeloma. N Engl J Med. 2014;371(10):895-905. 4. Fernandez de Larrea C, Jimenez R, Rosinol L, et al. Pattern of relapse and progression after autologous SCT as upfront treatment for multiple myeloma. Bone Marrow Transplant. 2014;49(2):223-227. 5. Porrata LF, Gertz MA, Inwards DJ, et al. Early lymphocyte recovery predicts superior survival after autologous hematopoietic stem cell transplantation in multiple myeloma or non-Hodgkin lymphoma. Blood. 2001;98(3):579-585. 6. Kim H, Sohn HJ, Kim S, et al. Early lymphocyte recovery predicts longer survival after autologous peripheral blood stem cell transplantation in multiple myeloma. Bone Marrow Transplant. 2006;37(11):1037-1042. 7. Tovar N, de Larrea CF, Arostegui JI, et al. Natural history and prognostic impact of oligoclonal humoral response in patients with multiple myeloma after autologous stem cell transplantation: long-term results from a single institution. Haematologica. 2013;98(7):1142-1146. 8. Rueff J, Medinger M, Heim D, Passweg J, Stern M. Lymphocyte subset recovery and outcome after autologous hematopoietic stem cell transplantation for plasma cell myeloma. Biol Blood Marrow Transplant. 2014;20(6):896-899. 9. Kastritis E, Zagouri F, Symeonidis A, et al. Preserved levels of uninvolved immunoglobulins are independently associated with favorable outcome in patients with symptomatic multiple myeloma. Leukemia. 2014;28(10):2075-2079.

haematologica | 2017; 102(5)

10. Kyrtsonis MC, Mouzaki A, Maniatis A. Mechanisms of polyclonal hypogammaglobulinaemia in multiple myeloma (MM). Med Oncol. 1999;16(2):73-77. 11. Quach H, Ritchie D, Stewart AK, et al. Mechanism of action of immunomodulatory drugs (IMiDS) in multiple myeloma. Leukemia. 2010;24(1):22-32. 12. Rawstron AC, Davies FE, Owen RG, et al. Blymphocyte suppression in multiple myeloma is a reversible phenomenon specific to normal B-cell progenitors and plasma cell precursors. Br J Haematol. 1998; 100(1):176183. 13. Perez-Persona E, Vidriales MB, Mateo G, et al. New criteria to identify risk of progression in monoclonal gammopathy of uncertain significance and smoldering multiple myeloma based on multiparameter flow cytometry analysis of bone marrow plasma cells. Blood. 2007;110(7):2586-2592. 14. Guillaume T, Rubinstein DB, Symann M. Immune reconstitution and immunotherapy after autologous hematopoietic stem cell transplantation. Blood. 1998; 92(5):14711490. 15. Durie BG, Harousseau JL, Miguel JS, et al. International uniform response criteria for multiple myeloma. Leukemia. 2006; 20(9):1467-1473. 16. Paiva B, Vidriales MB, Cervero J, et al. Multiparameter flow cytometric remission is the most relevant prognostic factor for multiple myeloma patients who undergo autologous stem cell transplantation. Blood. 2008;112(10):4017-4023. 17. Kaufman GP, Gertz MA, Dispenzieri A, et al. Impact of cytogenetic classification on outcomes following early high-dose therapy in multiple myeloma. Leukemia. 2016; 30(3):633-639. 18. Ross FM, Avet-Loiseau H, Ameye G, et al. Report from the European Myeloma Network on interphase FISH in multiple myeloma and related disorders. Haematologica. 2012;97(8):1272-1277. 19. Gutierrez NC, Castellanos MV, Martin ML, et al. Prognostic and biological implications of genetic abnormalities in multiple myeloma undergoing autologous stem cell transplantation: t(4;14) is the most relevant adverse prognostic factor, whereas RB dele-

20.

21.

22.

23.

24.

25.

26.

27.

28.

tion as a unique abnormality is not associated with adverse prognosis. Leukemia. 2007;21(1):143-150. Hieke S, Kleber M, Konig C, Engelhardt M, Schumacher M. Conditional Survival: A Useful Concept to Provide Information on How Prognosis Evolves over Time. Clin Cancer Res. 2015;21(7):1530-1536. Delgado J, Pereira A, Villamor N, LopezGuillermo A, Rozman C. Survival analysis in hematologic malignancies: recommendations for clinicians. Haematologica. 2014; 99(9):1410-1420. Hernandez MD, del Canizo MC, Gonzalez M, et al. [Immune reconstitution after autologous progenitor hemopoietic cell transplantation. A study comparing autologous bone marrow and autologous peripheral blood transplantation]. Med Clin (Barc). 1998;110(20):768-773. Sherrod AM, Hari P, Mosse CA, Walker RC, Cornell RF. Minimal residual disease testing after stem cell transplantation for multiple myeloma. Bone Marrow Transplant. 2016;51(1):2-12. Bradwell A, Harding S, Fourrier N, et al. Prognostic utility of intact immunoglobulin Ig'kappa/Ig'lambda ratios in multiple myeloma patients. Leukemia. 2013; 27(1):202-207. Tovar N, Fernandez de Larrea C, Elena M, et al. Prognostic impact of serum immunoglobulin heavy/light chain ratio in patients with multiple myeloma in complete remission after autologous stem cell transplantation. Biol Blood Marrow Transplant. 2012;18(7):1076-1079. Mohty M, Richardson PG, McCarthy PL, Attal M. Consolidation and maintenance therapy for multiple myeloma after autologous transplantation: where do we stand? Bone Marrow Transplant. 2015;50(8):10241029. Paiva B, Azpilikueta A, Puig N, et al. PDL1/PD-1 presence in the tumor microenvironment and activity of PD-1 blockade in multiple myeloma. Leukemia. 2015; 29(10):2110-2113. Gertz MA, Lacy MQ, Dispenzieri A, et al. Clinical implications of t(11;14)(q13;q32), t(4;14)(p16.3;q32), and -17p13 in myeloma patients treated with high-dose therapy. Blood. 2005;106(8):2837-2840.

931


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Cell Therapy & Immunotherapy

Ferrata Storti Foundation

Haematologica 2017 Volume 102(5):932-940

Plasma-derived proteomic biomarkers in human leukocyte antigen-haploidentical or human leukocyte antigen-matched bone marrow transplantation using post-transplantation cyclophosphamide

Christopher G. Kanakry,1 Giorgos Bakoyannis,2 Susan M. Perkins,2 Shannon R. McCurdy,1 Ante Vulic,1 Edus H. Warren,3 Etienne Daguindau,4,5 Taylor Olmsted,4,5 Christen Mumaw,4,5 Andrea M.H. Towlerton,3 Kenneth R. Cooke,1 Paul V. O'Donnell,3 Heather J. Symons,1 Sophie Paczesny4,5 and Leo Luznik1

Sidney Kimmel Comprehensive Cancer Center, The Johns Hopkins University School of Medicine, Baltimore, MD; 2Department of Biostatistics, Indiana University School of Medicine, Indianapolis, IN; 3Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, WA; 4Department of Pediatrics, Indiana University School of Medicine, Indianapolis, IN; 5Department of Immunology, Indiana University School of Medicine, Indianapolis, IN, USA 1

ABSTRACT

R Correspondence: luznile@jhmi.edu or sophpacz@iu.edu Received: July 8, 2016. Accepted: January 20, 2017. Pre-published: January 25, 2017. doi:10.3324/haematol.2016.152322 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/932 ©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.

932

ecent studies have suggested that plasma-derived proteins may be potential biomarkers relevant for graft-versus-host disease and/or non-relapse mortality occurring after allogeneic blood or marrow transplantation. However, none of these putative biomarkers have been assessed in patients treated either with human leukocyte antigen-haploidentical blood or marrow transplantation or with post-transplantation cyclophosphamide, which has been repeatedly associated with low rates of severe acute graft-versus-host disease, chronic graft-versus-host disease, and non-relapse mortality. We explored whether seven of these plasmaderived proteins, as measured by enzyme-linked immunosorbent assays, were predictive of clinical outcomes in post-transplantation cyclophosphamide-treated patients using plasma samples collected at serial predetermined timepoints from patients treated on prospective clinical studies of human leukocyte antigen-haploidentical (n=58; clinicaltrials.gov Identifier: 00796562) or human leukocyte antigen-matched-related or -unrelated (n=100; clinicaltrials.gov Identifiers: 00134017 and 00809276) T-cell-replete bone marrow transplantation. Day 30 levels of interleukin2 receptor α, tumor necrosis factor receptor 1, serum STimulation-2 (IL1RL1 gene product), and regenerating islet-derived 3-α all had high areas under the curve of 0.74-0.97 for predicting non-relapse mortality occurrence by 3 months post-transplant in both the human leukocyte antigen-matched and human leukocyte antigen-haploidentical cohorts. In both cohorts, all four of these proteins were also predictive of subsequent non-relapse mortality occurring by 6, 9, or 12 months post-transplant and were significantly associated with non-relapse mortality in univariable analyses. Furthermore, day 30 elevations of interleukin-2 receptor α were associated with grade II-IV and III-IV acute graft-versus-host disease occurring after day 30 in both cohorts. These data confirm that plasma-derived proteins previously assessed in other transplantation platforms appear to retain prognostic and predictive utility in patients treated with post-transplantation cyclophosphamide.

haematologica | 2017; 102(5)


Biomarkers for post-transplant cyclophosphamide

Introduction High-dose, post-transplantation cyclophosphamide (PTCy) provides effective graft-versus-host disease (GvHD) prophylaxis after allogeneic blood or marrow transplantation (alloBMT).1 This approach has facilitated the safe performance of T-cell-replete human leukocyte antigen (HLA)-haploidentical alloBMT1-3 and can function as single-agent GvHD prophylaxis after myeloablative conditioning and HLA-matched bone marrow allografting.1,4-6 Despite these clinical successes with low rates of severe acute GvHD, chronic GvHD, and nonrelapse mortality (NRM), biomarkers prognostic for GvHD or predictive for NRM occurring despite the use of PTCy have not been explored. Such biomarkers could potentially help guide treatment decisions and direct more intensive clinical surveillance of patients at high-risk for poor outcomes. Furthermore, they may provide biologic insight into immunologic pathways that could be targeted to prevent adverse clinical events. A number of candidate plasma-derived biomarkers have been examined in other alloBMT platforms,7-23 and several have repeatedly been found to be associated with clinical outcomes. In the study herein, we focused on seven particularly promising proteins. These proteins are all biologically plausible molecules either related directly to the inflammatory response thought to mediate GvHD (interleukin [IL]-2 receptor α [IL-2Rα],24 IL-6,25 tumor necrosis factor receptor 1 [TNFR-1],26 serum STimulation-2 (IL1RL1 gene product) [ST2],27,28 and chemokine [C-X-C motif] ligand 9 [CXCL9]29), or are released from tissue directly damaged by GvHD (lower gastrointestinal tract [regenerating islet-derived 3-α, REG3α]30-32 and skin [elafin]33). When measured at the start of clinical acute GvHD, elevated plasma levels of IL-2Rα, IL-6, TNFR-1, ST2, REG3α, and elafin have been associated with the presence and predicted severity of acute GvHD, response to immunosuppressive therapy, and/or risk for NRM.7-9,12-17,21 Plasma levels of CXCL9, IL-2Rα, and elafin measured at the time of onset of chronic GvHD have been associated with chronic GvHD diagnosis; CXCL9 had the highest predictive accuracy and was also associated with chronic GvHD severity.18 A few studies have examined the utility of biomarker measurements at pre-determined timepoints. When measured at day 7 after myeloablative alloBMT, elevated plasma levels of TNFR-1 were associated with an increased incidence of NRM and an increased incidence and severity of acute GvHD.10 Elevated plasma levels of ST2 at days 14 or 21 after myeloablative alloBMT or day 14 after reduced-intensity conditioning alloBMT were associated with an increased risk of NRM.17 In patients receiving double umbilical cord blood transplantation (dUCBT), elevated day 28 plasma levels of ST2 were associated with an increased incidence of NRM and grade II-IV and III-IV acute GvHD occurring after that timepoint; elevated day 28 plasma levels of TNFR-1 and REG3α, while not associated with acute GvHD, were associated with an increased risk of NRM.20 Finally, in patients treated with myeloablative HLA-matched-related alloBMT using either tacrolimus/sirolimus or tacrolimus/methotrexate for GvHD prophylaxis, elevated day 28 plasma levels of ST2, REG3α, and IL-6 were all associated with an increased risk of NRM, but were not associated with GvHD risk.23 With the increasing use of PTCy worldwide and the haematologica | 2017; 102(5)

clinical observations (e.g., low incidence of chronic GvHD) suggesting that PTCy may modulate GvHD risk in a dissimilar way to other T-cell-replete alloBMT approaches,1 we explored whether these seven promising candidate biomarkers had utility in prognosticating clinical outcomes for PTCy-treated patients. We assessed plasma levels using blood previously collected at serial pre-determined timepoints from 158 patients treated on one of three prospective clinical studies of PTCy as GvHD prophylaxis after myeloablative alloBMT using either HLAmatched or HLA-haploidentical donors.4,6,34 We hypothesized that plasma elevations in these proteins might be associated with negative clinical outcomes, particularly NRM, for PTCy-treated patients.

Methods Study design This study was designed to assess whether plasma-derived proteins, measured at specific post-transplant timepoints, are predictive of NRM occurrence or prognostic for the development of acute or chronic GvHD after alloBMT using PTCy as GvHD prophylaxis. The sample was based on the number of available plasma specimens previously collected at pre-determined timepoints from patients treated on one of three prospective clinical studies (clinicaltrials.gov Identifiers: 00134017, 00809276, and 00796562) (Online Supplementary Figure S1). Due to differences in the donor sources and adjunct GvHD prophylaxis, patients receiving HLAmatched versus HLA-haploidentical alloBMT were analyzed in separate cohorts for all analyses.

Patients and samples All three studies exclusively employed myeloablative conditioning and T-cell-replete bone marrow allografts. Two of the clinical trials, (clinicaltrials.gov Identifiers: 00134017 and 00809276), were both prospective studies of HLA-matched-related or -unrelated donor alloBMT using PTCy as single-agent GvHD prophylaxis for adult patients with advanced hematologic malignancies.4,6 One of these, (clinicaltrials.gov Identifier: 00134017), was a single-institutional study using busulfan/cyclophosphamide conditioning (n=122);4 although this study spanned 2004-2009, plasma was only cryopreserved for a group of 35 patients transplanted in 2007 and 2008. The other, (clinicaltrials.gov Identifier: 00809276), was a multi-institutional study using busulfan/fludarabine conditioning (n=92);6 as part of that protocol, plasma samples from all 80 patients treated at two of the three participating centers (Johns Hopkins Hospital and Fred Hutchinson Cancer Research Center) were collected. The third trial, (clinicaltrials.gov Identifier: 00796562), was a single-institutional prospective study of busulfan/cyclophosphamide conditioning, HLA-haploidentical donor alloBMT, and GvHD prophylaxis using PTCy, mycophenolate mofetil (MMF), and tacrolimus for adult or pediatric patients with advanced hematologic malignancies (n=95).34 Plasma samples were collected for 92 patients. As this latter trial enrolled patients with chemorefractory hematologic malignancies and the study did not mandate specimen collection once a patient relapsed, patients on this trial that relapsed less than 6 months post-transplant (n=23) were not included in this analysis. Blood was collected only at pre-determined timepoints. Patients who experienced graft failure (n=10), suffered NRM prior to day 30 (n=4), or survived to day 30 and had sustained engraftment but did not have a day 30 plasma sample collected (n=12) were excluded (Online Supplementary Figure S1). Both studies of HLA-matched alloBMT (clinicaltrials.gov Identifiers: 00134017 and 933


C.G. Kanakry et al. Table 1. Patient, donor, and transplantation characteristics.

Characteristics Donor, n (%) Related Unrelated Conditioning Bu/Cy Bu/Flu GvHD prophylaxis Age, y, median (range) Sex, n (%) female Diagnosis, n (%) AML/MDS/CMML ALL NHL/CLL Hodgkin lymphoma CML Multiple myeloma Remission status at alloBMT, n (%) CR without MRD CR with MRD Not in CR HCT-CI score, median (range) Donor age, y, median (range) Donor sex, n (%) female Female donor into male recipient, n (%) CMV serostatus D-/RD+/RD-/R+ D+/R+ Dunk/R+ Allograft cell counts, # (range)/kg recipient weight Total nucleated cells, x 108 CD34+ cells, x 106 CD3+ cells, x 107

HLA-Matched (n=100)

HLA-Haploidentical (n=58)

All Patients (n=158)

42 (42%) 58 (58%)

58 (100%) N/A

100 (63%) 58 (37%)

30 (30%) 70 (70%) PTCy alone 51 (22-65) 57 (57%)

58 (100%) 0 (0%) PTCy, MMF, tacro 42 (2-64) 27 (47%)

88 (56%) 70 (44%) 49 (2-65) 84 (53%)

73 (73%) 16 (16%) 3 (3%) 0 (0%) 6 (6%) 2 (2%)

33 (57%) 8 (14%) 9 (16%) 4 (7%) 4 (7%) 0 (0%)

106 (67%) 24 (15%) 12 (8%) 4 (3%) 10 (6%) 2 (1%)

52 (52%) 23 (23%) 25 (25%) 2 (0-8) 41 (17-67) 56 (56%) 23 (23%)

31 (53%) 9 (16%) 18 (31%) 1 (0-5) 36 (11-69) 34 (59%) 18 (31%)

83 (53%) 32 (20%) 43 (27%) 2 (0-8) 40 (11-69) 90 (57%) 41 (26%)

31 (31%) 16 (16%) 30 (30%) 22 (22%) 1 (1%)

28 (48%) 10 (18%) 9 (16%) 11 (19%) 0 (0%)

59 (37%) 26 (16%) 39 (25%) 33 (21%) 1 (1%)

4.09 (1.3-8.82) 3.72 (1.0-9.87) 3.8 (0.6-7.75)

4.84 (2.63-11.2) 4.16 (1.68-9.73) 4.67 (1.09-11.07)

4.31 (1.3-11.2) 3.82 (1.0-9.87) 4.11 (0.6-11.07)

HLA: human leukocyte antigen; Bu: busulfan; Cy: cyclophosphamide; Flu: fludarabine; GvHD: graft-versus-host disease; tacro: tacrolimus; PTCy: post-transplantation cyclophosphamide; MMF: mycophenolate mofetil; AML: acute myeloid leukemia; MDS: myelodysplastic syndrome; CMML: chronic myelomonocytic leukemia; ALL: acute lymphoblastic leukemia; NHL: non-Hodgkin lymphoma; CLL: chronic lymphocytic leukemia; CML: chronic myelogenous leukemia; alloBMT: allogeneic bone marrow transplantation; CR: morphologic complete remission; MRD: minimal residual disease; HCT-CI, hematopoietic cell transplantation-comorbidity index; CMV: cytomegalovirus; D: donor; R: recipient; unk: unknown.

00809276) were combined together as a single cohort for analysis as these patients were identically treated except for the type of myeloablative conditioning (cyclophosphamide versus fludarabine in addition to busulfan). Ultimately, 100 patients receiving HLAmatched alloBMT and 58 patients receiving HLA-haploidentical alloBMT were included in this study (Table 1; Online Supplementary Figure S1). Plasma samples from 33 healthy donors of bone marrow for patients treated on the two latter mentioned trials were also included in these analyses. All patients and healthy donors provided written informed consent on institutional review board-approved protocols prior to specimen collection.

Proteomic analysis Enzyme-linked immunosorbent assays (ELISAs) for IL-2RÎą, IL-6, TNFR-1, ST2, elafin, REG3Îą, and CXCL9 were performed in batches on cryopreserved plasma.12-14,16-18,20 All of these biomarkers were measured using sequential ELISA as previously reported.35 These methods are described in detail in the Online Supplementary Methods. 934

Statistical analysis Statistical comparisons between donors and recipients regarding biomarker levels at various timepoints were based on the nonparametric Mann-Whitney test. In order to account for multiple hypothesis testing due to the multiple timepoints, we applied the Bonferroni type I error adjustment. All other outcomes were analyzed under the competing risks framework.36 Definitions of endpoints and competing risks are defined in the Online Supplementary Methods. To explicitly quantify the predictive accuracy of each biomarker at 30 days for the different outcomes of the study, and to account for both the time-dependency of the outcomes and the competing risks, time-dependent receiver operating characteristic (ROC) curves for competing endpoints were computed. 37 Predictive accuracy was estimated based on the area under the ROC curve at 3, 6, 9, and 12 months. To assess the association of each biomarker with outcomes, the linear effect of the biomarker (treated as a continuous variable), either alone or adjusted for the potential confounders of recipient age and recipient cytomegalovirus serostatus, on the cumulative incidence was estihaematologica | 2017; 102(5)


Biomarkers for post-transplant cyclophosphamide

Table 2. Univariable associations of day 30 biomarker levels with acute GvHD and non-relapse mortality.

Biomarker IL-2Rα (per 500 units) HLA-matched HLA-haploidentical TNFR-1 (per 1000 units) HLA-matched HLA-haploidentical ST2 (per 20 units) HLA-matched HLA-haploidentical REG3α (per 20 units) HLA-matched HLA-haploidentical Elafin (per 1000 units) HLA-matched HLA-haploidentical IL-6 (per 10 units) HLA-matched HLA-haploidentical CXCL9 (per 5 units) HLA-matched HLA-haploidentical

Grade II-IV Acute GvHD SHR (95% CI)

P

Grade III-IV Acute GvHD SHR (95% CI)

Non-Relapse Mortality P

SHR (95% CI)

P

1.27 (1.06-1.53) 1.66 (1.04-2.64)

0.01 0.03

1.47 (1.18-1.83) 1.61 (1.08-2.41)

0.001 0.021

1.55 (1.17-2.05) 1.88 (1.28-2.76)

0.002 0.001

1.04 (0.96-1.13) 1.00 (0.91-1.10)

0.33 0.99

1.09 (1.00-1.19) 1.01 (0.88-1.16)

0.05 0.93

1.22 (1.07-1.38) 1.28 (1.20-1.36)

0.002 <0.001

1.04 (0.95-1.15) 1.10 (0.92-1.33)

0.39 0.30

1.06 (0.88-1.27) 1.10 (0.89-1.37)

0.56 0.36

1.12 (1.02-1.23) 1.41 (1.26-1.56)

0.013 <0.001

1.00 (0.98-1.02) 0.99 (0.98-1.00)

0.60 0.11

1.01 (0.99-1.03) 0.99 (0.97-1.01)

0.24 0.17

1.05 (1.04-1.07) 1.01 (1.01-1.01)

<0.001 <0.001

1.02 (0.98-1.07) 1.06 (0.99-1.13)

0.32 0.09

1.00 (0.94-1.06) 1.03 (0.98-1.08)

0.98 0.31

1.00 (0.93-1.06) 1.11 (1.06-1.17)

0.88 <0.001

1.00 (0.99-1.01) 0.97 (0.83-1.13)

0.98 0.68

1.00 (0.99-1.02) 0.12 (0.02-0.80)

0.69 0.03

1.00 (0.98-1.02) 1.03 (1.02-1.04)

0.91 <0.001

1.02 (0.98-1.06) 1.50 (0.92-2.44)

0.24 0.10

1.03 (0.99-1.07) 2.70 (1.80-4.04)

0.17 <0.001

1.03 (0.98-1.08) 1.23 (0.73-2.09)

0.25 0.44

For grade II-IV acute GvHD, the numbers of events and competing risks were 31 and 23 for the HLA-matched cohort and 6 and 19 for the HLA-haploidentical cohort, respectively. For grade III-IV acute GvHD, the numbers of events and competing risks were 16 and 39 for the HLA-matched cohort and 2 and 23 for the HLA-haploidentical cohort, respectively. For non-relapse mortality, the numbers of events and competing risks were 14 and 37 for the HLA-matched cohort and 10 and 12 for the HLA-haploidentical cohort, respectively. Biomarkers for these analyses were assessed as continuous variables. The SHR listed for each biomarker/outcome is the risk per given number of biomarker units shown. GvHD: graft-versus-host disease; IL-2Rα: interleukin-2 receptor α; TNFR-1: tumor necrosis factor receptor-1; ST2: serum STimulation-2, IL1RL1 gene product; REG3α: regenerating isletderived 3-α; IL-6: interleukin-6; CXCL9: chemokine [C-X-C motif] ligand 9; HLA: human leukocyte antigen; SHR: subdistribution hazard ratio; CI: confidence interval.

mated under the semiparametric proportional subdistribution hazards model (Fine-Gray model).38 Additionally, day 30 biomarker values were dichotomized (higher than or equal to median value vs. lower than median value), and cumulative incidence functions, estimated using the nonparametric Aalen-Johansen estimator,39 were plotted. Statistical comparisons between the cumulative incidence curves for the outcome by the biomarker level categories were undertaken using Gray’s nonparametric test.40 All analyses were performed separately for the HLA-matched and HLA-haploidentical cohorts.

Results Plasma-derived putative proteomic biomarkers are elevated post-transplant Plasma levels of all seven tested proteins were significantly elevated in patients post-transplant compared with plasma levels of their healthy bone marrow donors (Figure 1). Levels of IL-2Rα, TNFR-1, ST2, and IL-6 were highest at 30 days post-transplant, declined at subsequent posttransplant measurements, and had similar temporal kinetics between the HLA-matched and HLA-haploidentical cohorts (Figure 1). Levels of the other three proteins were relatively stable during the first post-transplant year, but differed between the two cohorts: REG3α and CXCL9 levels were higher in the HLA-matched cohort, while haematologica | 2017; 102(5)

elafin levels were higher in the HLA-haploidentical cohort (Figure 1). The differences in REG3α levels between the two cohorts may be related to the much higher incidence of gastrointestinal acute GvHD in the HLA-matched cohort (39% versus 12%) (Online Supplementary Table S1). The difference in elafin levels between cohorts was not directly attributable to varying GvHD rates, as the incidence of cutaneous acute GvHD was higher in the HLAmatched cohort (Online Supplementary Table S1).

Day 30 IL-2Rα, TNFR-1, ST2, and REG3α are consistently predictive of the occurrence of non-relapse mortality Using the first post-transplant timepoint (day 30) samples, levels of each of the seven proteins were tested for associations with subsequent occurrence of NRM. First, time-dependent ROC curves, which relate a biomarker’s sensitivity and specificity, were generated to assess the overall accuracy (area under the ROC curve) of each biomarker for predicting NRM. Separate ROC curves were generated for each cohort (HLA-matched and HLA-haploidentical) (Figure 2). In both cohorts, IL-2Rα, TNFR-1, ST2, and REG3α all had high area under the curve (AUC) values of 0.74-0.97, consistent with high degrees of predictive accuracy for NRM occurrence by 3 months post-transplant (Figure 2). In the HLA-haploidentical cohort, elafin and IL-6 also had high AUC values of 0.72-0.83 (Figure 2). 935


C.G. Kanakry et al.

Figure 1. Dynamics of biomarker levels over time. Plasma biomarker levels are shown for healthy bone marrow donors (n=33) and for patients at pre-transplant and serial pre-defined (1, 2, 6, and 12 months) post-transplant timepoints. The number of samples for the HLA-haploidentical cohort is as follows: pre-transplant n=53, 1 month n=58, 2 months n=54, 6 months n=41, 12 months n=18. The number of samples for the HLA-matched cohort is as follows: pre-transplant n=93, 1 month n=100, 2 months n=90, 6 months n=33, 12 months n=36. Samples for the HLA-matched and HLA-haploidentical cohorts were drawn at the same timepoints, but are shown slightly staggered for the sake of clarity. Asterisks show statistically significant differences for each cohort compared with donor levels. IL-2Rα: interleukin2 receptor α; TNFR-1: tumor necrosis factor receptor 1; ST2: serum STimulation-2, IL1RL1 gene product; CXCL9: chemokine [C-X-C motif] ligand 9; REG3α: regenerating islet-derived 3-α; IL-6: interleukin-6; BMT: allogeneic bone marrow transplantation; HLA: human leukocyte antigen; CI: confidence interval.

In both cohorts, the AUC values remained high for each of these biomarkers when assessing NRM occurring by 6, 9, or 12 months post-transplant, suggesting that early elevations in these biomarkers may be predictive of both shorter and longer term NRM risk. In univariable analyses for association with NRM, high day 30 levels of IL-2Rα, TNFR-1, ST2, and REG3α were significantly associated with greater cumulative incidence of NRM within each cohort (Table 2). High elafin and IL6 levels were also associated with NRM in the HLA-haploidentical cohort only (Table 2), while CXCL9 was not associated with NRM in either cohort. As prior analyses had shown significant associations of recipient age and recipient cytomegalovirus serostatus with NRM,5 biomarker analyses were subsequently adjusted for these two factors and showed similar results as the unadjusted univariable analyses (Online Supplementary Table S2). To provide results comparable with one of the most analogous alloBMT biomarker studies,20 the association of each protein with NRM was also assessed for each cohort by dichotomizing each biomarker at the median and nonparametrically estimating the cumulative incidence functions of each outcome (Online Supplementary Tables S3 and S4). Of note, in all univariable analyses, there were no significant associations between day 30 levels for any of the seven proteins and subsequent relapse of malignant disease.

Day 30 IL-2Rα levels are consistently prognostic for the occurrence of acute GvHD The timing of sample collection was not ideal for assess936

ing the potential association of these biomarkers with acute GvHD as 31 patients (19.6%) had onset of grade IIIV acute GvHD prior to day 30. Despite this limitation, and in order to better understand the associations with subsequent NRM, we did assess whether these biomarkers were prognostic for acute GvHD occurring after day 30. Importantly, the NRM for patients with grade II-IV acute GvHD onset prior to day 30 was not higher than the NRM of patients who developed grade II-IV acute GvHD after day 30 (time-dependent proportional cause-specific hazard ratio (HR), grade II-IV acute GvHD developing after day 30 compared with before day 30, HR 1.14 (95% confidence interval (CI) 0.41-3.16); P=0.81). Furthermore, there was no evidence that the risk of developing grade IIIIV acute GvHD was different in those who developed grade II-IV acute GvHD after day 30 compared with before day 30 based on Fisher’s exact test (P=0.68) or the Cox proportional cause-specific hazards model (HR 1.3 (95% CI 0.5-3.35), P=0.59). In testing associations with grade II-IV acute GvHD using the day 30 protein levels, ROC curves showed high AUC values of >0.7 in both cohorts for IL-2Rα, TNFR-1, and CXCL9, consistent with high predictive accuracy for grade II-IV acute GvHD occurring after day 30 (Online Supplementary Figures S2 and S3). In the HLA-haploidentical cohort, ST2 and REG3α also had high AUC values of approximately 0.9 for grade II-IV acute GvHD, although the number of events (n=6) was low (Online Supplementary Figure S3). In testing associations with grade III-IV acute GvHD development after day 30, day 30 IL-2Rα, CXCL9, and REG3α levels all had high AUC values (>0.73 for the haematologica | 2017; 102(5)


Biomarkers for post-transplant cyclophosphamide

A

B

Figure 2. Time-dependent ROC curves for NRM separated by donor type. Using day 30 biomarker levels, ROC curves for predicting subsequent NRM occurrence by 3, 6, 9, or 12 months post-transplant are shown separately for the (A) HLA-matched and (B) HLA-haploidentical cohorts. The analysis was performed using a competing risks framework with relapse as a competing risk for NRM. IL-2Rα: interleukin-2 receptor α; TNFR-1: tumor necrosis factor receptor 1; ST2: serum STimulation2, IL1RL1 gene product; CXCL9: chemokine [C-X-C motif] ligand 9; REG3α: regenerating islet-derived 3-α; IL-6: interleukin-6; AUC: area under the curve.

haematologica | 2017; 102(5)

937


C.G. Kanakry et al.

HLA-matched cohort and >0.9 for the HLA-haploidentical cohort) (Online Supplementary Figures S4 and S5). Within the HLA-haploidentical cohort, day 30 TNFR-1, ST2, and elafin levels also had high AUC values (>0.8) for grade IIIIV acute GvHD, while IL-6 had a low AUC value (0.18), although again the event number was quite low (n=2). In univariable analyses, high day 30 IL-2Rα levels were consistently associated with a greater cumulative incidence of both grade II-IV and grade III-IV acute GvHD (Table 2 and Online Supplementary Table S2; Online Supplementary Figures S2-S5). Six-month plasma protein levels were used to evaluate for an association with chronic GvHD development. However, only 71 of the 125 patients (56.8%) that were chronic GvHD-free and still at risk for chronic GvHD at 6 months had 6-month samples collected. Univariable analyses with chronic GvHD development did not show any statistically significant associations, although CXCL9 trended towards significance in the HLA-haploidentical cohort (HLA-haploidentical cohort, HR 1.49 (95% CI 0.932.40), P=0.096; HLA-matched cohort, HR 1.12 (95% CI 0.95-1.32), P=0.17).

Discussion In the study herein, we explored whether putative proteomic biomarkers, previously assessed using other transplantation platforms, were applicable to PTCy-treated patients. Among its novel features, our study is the first to examine the relevance of promising biomarkers in PTCytreated patients or patients receiving HLA-haploidentical alloBMT. All patients were uniformly treated with myeloablative conditioning, bone marrow allografts, and PTCy for GvHD prophylaxis, and we found effects that appear to be conserved across different donor sources. Furthermore, we have used novel statistical methods to address our scientific aims. These include the Fine-Gray model to directly estimate the effect of an independent variable on the cumulative incidence function of the outcome of interest in the presence of competing risks and also time-dependent ROC curve methodology for competing risks to directly quantify the predictive accuracy of each biomarker. We found that, despite low rates of NRM and severe GvHD in PTCy-treated patients, elevated levels of several of these plasma proteins remained prognostic for adverse outcomes. Consistently in both cohorts, elevations in IL-2Rα, TNFR-1, ST2, and REG3α at 30 days post-transplant each were significantly associated with a greater cumulative incidence of NRM. Furthermore, in both cohorts, elevations in IL-2Rα at 30 days post-transplant were associated with subsequent grade II-IV and III-IV acute GvHD occurrence. Prior studies of putative proteomic biomarkers in other transplantation platforms, using plasma measurements at pre-determined timepoints during the first post-transplant month, have shown associations between levels of TNFR1, ST2, and REG3α and subsequent occurrence of NRM and/or acute GvHD.10,17,20,23 Most consistent with our results, a recent large study of patients treated with HLAmatched-related alloBMT showed that elevated day 28 ST2 and REG3α levels were associated with greater risk for NRM, but did not have a significant relationship with acute or chronic GvHD.23 Also compatible with our 938

results, a study of patients undergoing dUCBT found that elevations in TNFR-1, ST2, and REG3α at day 28 posttransplant were associated with NRM at 180 days.20 In contrast to their results, but consistent with the HLAmatched-related alloBMT study, we did not find a statistically significant association between ST2 levels and acute GvHD. Furthermore, in the dUCBT study, elevations above the median in IL-2Rα were not associated with GvHD or NRM, whereas our study showed strong associations of IL-2Rα with grade II-IV acute GvHD, grade III-IV acute GvHD, and NRM. The strong relationship we found between day 30 IL-2Rα levels and acute GvHD was consistent with another study that demonstrated a peak in IL2Rα levels at 14 days post-transplant in patients who would subsequently develop acute GvHD.7 Moreover, our results are consistent with other studies which have shown associations between IL-2Rα levels measured at the start of acute GvHD and subsequent NRM.12,16 In contrast with prior studies that showed that day 14 IL-6 levels were prognostic of subsequent grade III-IV acute GvHD occurrence or that day 28 IL-6 levels were prognostic of NRM,21,23 we did not find a consistent relationship between IL-6 and outcomes between our two cohorts. Our study has several limitations. Importantly, we do not have an independent verification cohort as we utilized all plasma samples available from three prospective studies of PTCy. Furthermore, cross-validation was not performed to evaluate the out-of-sample predictive ability of our model. This is because the goal of the present study was neither to propose a specific model for the alloBMT-related outcomes nor to provide a specific classification/prediction rule. Our goal was to evaluate the predictive potential of a set of biomarkers, and therefore we quantified this using time-dependent ROC curves for competing risks. Even so, we found consistent results between the two cohorts in terms of four predictors of NRM and one predictor of acute GvHD, suggesting that our results may have some external validity. Differences in plasma protein levels and associations for three of the tested biomarkers (CXCL9, REG3α, and elafin) between the HLA-matched and HLA-haploidentical cohorts could reflect differences in the biology of donor HLA disparity, but more likely reflect the differences in the immunosuppressants used (the former used PTCy alone, whereas the latter also incorporated MMF and tacrolimus), as demonstrated in a recent publication regarding REG3α,23 in addition to the differing incidences of acute GvHD (lower grade II-IV and III-IV acute GvHD rates for the HLA-haploidentical cohort) probably due to this adjunct immunosuppression therapy. Unfortunately, our limited number of events for NRM and GvHD outcomes made it statistically infeasible to perform multivariable analyses to attempt to thoroughly dissect the impact of other potential confounders. Another limitation of this study is that we did not adjust for type I error inflation due to multiple statistical testing in the univariable models. However, the main purpose of this work was to estimate the predictive accuracy for a set of biomarkers that had already been reported to be associated with NRM and GvHD in previous reports using other transplantation platforms. The evaluation of this predictive accuracy was not based on hypothesis testing but on the calculation of ROC AUCs. Furthermore, hypothesis tests that were conducted were specified a priori and were confirmatory in nature rather than exploratory. haematologica | 2017; 102(5)


Biomarkers for post-transplant cyclophosphamide

Lastly, our study is limited by the sample collection performed. Our samples were collected at uniform pre-determined timepoints, but samples from patient-specific timepoints (e.g., at GvHD diagnosis or longitudinally in response to GvHD treatment) were not collected. Furthermore, the first post-transplant plasma samples were taken at day 30; 20% of patients had experienced the onset of grade II-IV acute GvHD prior to that timepoint and had to be excluded from the acute GvHD analyses. Thus, our results for acute GvHD are only relevant for patients surviving to day 30 without yet developing acute GvHD. The apparent lack of statistical association of some of these proteins (e.g., ST2, REG3α, or elafin for acute GvHD) with clinical outcomes in our patients could in part be explained by these temporal and contextual differences in sampling; had samples been collected earlier post-transplant or at the time of diagnosis of acute GvHD, these proteins may have had a more apparent prognostic import. While the HLA-haploidentical cohort also suffered from exclusion of patients who relapsed within 6 months posttransplant, those patients would not have remained at risk for NRM or GvHD in any case due to the occurrence of a competing risk. Thus, the main findings of this manuscript would likely remain unchanged even if samples had been universally available for those patients. Moreover, many of the results in the HLA-haploidentical cohort were confirmed in the HLA-matched cohort, in which there was no bias in sample availability. Even though PTCy is broadly and highly effective in preventing chronic GvHD in patients undergoing alloBMT,1,2,4-6 our study has not revealed any factor that can prognosticate which of our patients will still develop chronic GvHD despite PTCy treatment. It is possible that the lack of any significant findings was due in part to the low incidence of chronic GvHD in our patients and thus low statistical power to detect true differences that may exist; however, the number of chronic GvHD events (n=24) was identical to the number of NRM events (n=24) and more than the number of grade III-IV acute GvHD events (n=18). Further insight may require a better understanding of the underlying biology of chronic GvHD, which remains poorly defined.41 Immunologic parameters could potentially be of interest given the recently demonstrated role for regulatory T cells in GvHD prevention by PTCy,42,43 but such studies would require a large number of patients for adequate statistical power given the low incidence of chronic GvHD development in PTCy-treated patients. Although not prognostic for chronic GvHD, elevations in IL-2Rα, TNFR-1, ST2, and REG3α at day 30 post-transplant were consistently predictive of NRM occurrence in our patients. While IL-2Rα was also prognostic for acute GvHD occurrence in both cohorts, the majority of our patients did not die from GvHD-related causes (Online

References 1. Kanakry CG, Fuchs EJ, Luznik L. Modern approaches to HLA-haploidentical blood or marrow transplantation. Nat Rev Clin Oncol. 2016;13(2):132.

haematologica | 2017; 102(5)

Supplementary Table S5). This apparent disconnect between predictors of GvHD and NRM was also seen in a recent study of HLA-matched alloBMT in which day 28 plasma biomarker assessments were studied.23 Thus, whether elevations in these proteins are specific for GvHD is unclear. It is possible that GvHD and many cases of NRM are both direct sequelae of a dysregulated immune system, which is what many of these proteins are directly measuring. Further dissection of what elevations in each of these biomarkers mean, particularly in cases where they are not correlated with each other, would require complex prospective assessment of a host of clinical and immunologic factors and would require a large number of patients in addition to pre-clinical supporting studies. Beyond the immediate risk for NRM, it is disconcerting that day 30 elevations of these proteins are associated with an increased risk for NRM in our patients, which remains present for at least the first post-transplant year. Given this apparent longer term risk, a fundamental outstanding question is whether this risk is modifiable with more intensive monitoring, immunosuppressive treatment, or other novel interventions. Conversely, perhaps low levels of these plasma proteins could justify less intensive monitoring in such patients (i.e., risk-adapted therapy). If the risk is modifiable, what is the earliest posttransplant timepoint at which this risk could be determined such that an intervention could be expeditiously implemented? Prospective studies are required to further clarify the role of biomarkers in facilitating personalized, risk-adapted strategies for monitoring patients after alloBMT. This may be particularly relevant as older and less fit patients are being transplanted.44,45 Moreover, the optimal cutoffs of “high” versus “low” would need to be better defined and standardized before broader clinical application of these results. Despite the need for further research, our work supports the fact that subsequent studies, such as the BMT Clinical Trials Network study 1202,46 hold high promise for resulting in validated plasmaderived biomarkers that could have prognostic and predictive utility broadly applicable to patients undergoing alloBMT, irrespective of the particular transplantation platform employed. Funding This work was supported by awards from the NIH (R01 HL110907 to LL and R01 CA16884 to SP), the Leukemia & Lymphoma Society Scholar Award (1293-15, to SP), the Lilly Physician Scientist Initiative Award (to SP), and a grant from Otsuka Pharmaceutical (to LL). Acknowledgments We thank all patients and donors who participated in these clinical studies (clinicaltrials.gov Identifiers: 00134017, 00809276, and 00796562) and who contributed specimens.

2. Luznik L, O'Donnell PV, Symons HJ, et al. HLA-haploidentical bone marrow transplantation for hematologic malignancies using nonmyeloablative conditioning and high-dose, posttransplantation cyclophosphamide. Biol Blood Marrow Transplant. 2008;14(6):641-650.

3. McCurdy SR, Kanakry JA, Showel MM, et al. Risk-stratified outcomes of nonmyeloablative HLA-haploidentical BMT with highdose posttransplantation cyclophosphamide. Blood. 2015;125(19):3024-3031. 4. Luznik L, Bolanos-Meade J, Zahurak M, et al. High-dose cyclophosphamide as single-

939


C.G. Kanakry et al.

5.

6.

7.

8.

9.

10.

11.

12. 13.

14.

15.

16.

17.

940

agent, short-course prophylaxis of graft-versus-host disease. Blood. 2010; 115(16):32243230. Kanakry CG, Tsai HL, Bolanos-Meade J, et al. Single-agent GVHD prophylaxis with posttransplantation cyclophosphamide after myeloablative, HLA-matched BMT for AML, ALL, and MDS. Blood. 2014; 124(25):3817-3827. Kanakry CG, O'Donnell PV, Furlong T, et al. Multi-institutional study of post-transplantation cyclophosphamide as single-agent graft-versus-host disease prophylaxis after allogeneic bone marrow transplantation using myeloablative busulfan and fludarabine conditioning. J Clin Oncol. 2014; 32(31):3497-3505. Miyamoto T, Akashi K, Hayashi S, et al. Serum concentration of the soluble interleukin-2 receptor for monitoring acute graftversus-host disease. Bone Marrow Transplant. 1996;17(2):185-190. Foley R, Couban S, Walker I, et al. Monitoring soluble interleukin-2 receptor levels in related and unrelated donor allogenic bone marrow transplantation. Bone Marrow Transplant. 1998;21(8):769-773. Grimm J, Zeller W, Zander AR. Soluble interleukin-2 receptor serum levels after allogeneic bone marrow transplantations as a marker for GVHD. Bone Marrow Transplant. 1998;21(1):29-32. Kitko CL, Paczesny S, Yanik G, et al. Plasma elevations of tumor necrosis factor-receptor1 at day 7 postallogeneic transplant correlate with graft-versus-host disease severity and overall survival in pediatric patients. Biol Blood Marrow Transplant. 2008;14(7):759765. Choi SW, Kitko CL, Braun T, et al. Change in plasma tumor necrosis factor receptor 1 levels in the first week after myeloablative allogeneic transplantation correlates with severity and incidence of GVHD and survival. Blood. 2008;112(4):1539-1542. Paczesny S, Krijanovski OI, Braun TM, et al. A biomarker panel for acute graft-versushost disease. Blood. 2009;113(2):273-278. Paczesny S, Braun TM, Levine JE, et al. Elafin is a biomarker of graft-versus-host disease of the skin. Sci Transl Med. 2010; 2(13):13ra12. Ferrara JL, Harris AC, Greenson JK, et al. Regenerating islet-derived 3-alpha is a biomarker of gastrointestinal graft-versus-host disease. Blood. 2011;118(25):6702-6708. Harris AC, Ferrara JL, Braun TM, et al. Plasma biomarkers of lower gastrointestinal and liver acute GVHD. Blood. 2012; 119(12):2960-2963. Levine JE, Logan BR, Wu J, et al. Acute graftversus-host disease biomarkers measured during therapy can predict treatment outcomes: a Blood and Marrow Transplant Clinical Trials Network study. Blood. 2012; 119(16):3854-3860. Vander Lugt MT, Braun TM, Hanash S, et al. ST2 as a marker for risk of therapy-resistant graft-versus-host disease and death. N Engl J

Med. 2013;369(6):529-539. 18. Kitko CL, Levine JE, Storer BE, et al. Plasma CXCL9 elevations correlate with chronic GVHD diagnosis. Blood. 2014;123(5):786793. 19. Holtan SG, Verneris MR, Schultz KR, et al. Circulating angiogenic factors associated with response and survival in patients with acute graft-versus-host disease: results from Blood and Marrow Transplant Clinical Trials Network 0302 and 0802. Biol Blood Marrow Transplant. 2015; 21(6):1029-1036. 20. Ponce DM, Hilden P, Mumaw C, et al. High day 28 ST2 levels predict for acute graft-versus-host disease and transplant-related mortality after cord blood transplantation. Blood. 2015;125(1):199-205. 21. McDonald GB, Tabellini L, Storer BE, Lawler RL, Martin PJ, Hansen JA. Plasma biomarkers of acute GVHD and nonrelapse mortality: predictive value of measurements before GVHD onset and treatment. Blood. 2015;126(1):113-120. 22. Holler E, Kolb HJ, Moller A, et al. Increased serum levels of tumor necrosis factor alpha precede major complications of bone marrow transplantation. Blood. 1990;75(4): 1011-1016. 23. Abu Zaid M, Wu J, Wu C, et al. Plasma biomarkers of risk for death in a multicenter phase 3 trial with uniform transplant characteristics post-allogeneic HCT. Blood. 2016;129(2):162-170. 24. Rubin LA, Kurman CC, Fritz ME, et al. Soluble interleukin 2 receptors are released from activated human lymphoid cells in vitro. J Immunol. 1985;135(5):3172-3177. 25. Coghill JM, Sarantopoulos S, Moran TP, Murphy WJ, Blazar BR, Serody JS. Effector CD4+ T cells, the cytokines they generate, and GVHD: something old and something new. Blood. 2011;117(12):3268-3276. 26. Holler E, Kolb HJ, Hintermeier-Knabe R, et al. Role of tumor necrosis factor alpha in acute graft-versus-host disease and complications following allogeneic bone marrow transplantation. Transplant Proc. 1993;25(1 Pt 2):1234-1236. 27. Matta BM, Lott JM, Mathews LR, et al. IL-33 is an unconventional Alarmin that stimulates IL-2 secretion by dendritic cells to selectively expand IL-33R/ST2+ regulatory T cells. J Immunol. 2014;193(8):4010-4020. 28. Schiering C, Krausgruber T, Chomka A, et al. The alarmin IL-33 promotes regulatory Tcell function in the intestine. Nature. 2014;513(7519):564-568. 29. Groom JR, Luster AD. CXCR3 ligands: redundant, collaborative and antagonistic functions. Immunol Cell Biol. 2011; 89(2):207-215. 30. Cash HL, Whitham CV, Behrendt CL, Hooper LV. Symbiotic bacteria direct expression of an intestinal bactericidal lectin. Science. 2006;313(5790):1126-1130. 31. Zheng Y, Valdez PA, Danilenko DM, et al. Interleukin-22 mediates early host defense against attaching and effacing bacterial pathogens. Nat Med. 2008;14(3):282-289.

32. Ogawa H, Fukushima K, Naito H, et al. Increased expression of HIP/PAP and regenerating gene III in human inflammatory bowel disease and a murine bacterial reconstitution model. Inflamm Bowel Dis. 2003; 9(3):162-170. 33. Alkemade JA, Molhuizen HO, Ponec M, et al. SKALP/elafin is an inducible proteinase inhibitor in human epidermal keratinocytes. J Cell Sci. 1994;107( Pt 8):2335-2342. 34. Symons HJ, Chen A, Gamper C, et al. Haploidentical BMT using fully myeloablative conditioning, T cell replete bone marrow grafts, and post-transplant cyclophosphamide (PT/Cy) has limited toxicity and promising efficacy in largest reported experience with high risk hematologic malignancies. Biol Blood Marrow Transplant. 2015;21(2):S29-S29. 35. Fiema B, Harris AC, Gomez A, et al. High throughput sequential ELISA for validation of biomarkers of acute graft-versus-host disease. J Vis Exp. 2012;68. pii 4247. 36. Bakoyannis G, Touloumi G. Practical methods for competing risks data: a review. Stat Methods Med Res. 2012;21(3):257-272. 37. Saha P, Heagerty PJ. Time-dependent predictive accuracy in the presence of competing risks. Biometrics. 2010;66(4):999-1011. 38. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. Stat Med. 1999;94(446):496-509. 39. Aalen OO, Johansen S. An empirical transition matrix for non-homogeneous Markov chains based on censored observations. Scand J Stat. 1978;5:141-150. 40. Gray RJ. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat. 1988;16:1141-1154. 41. Blazar BR, Murphy WJ, Abedi M. Advances in graft-versus-host disease biology and therapy. Nat Rev Immunol. 2012;12(6):443-458. 42. 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. 43. Ganguly S, Ross DB, Panoskaltsis-Mortari A, et al. Donor CD4+ Foxp3+ regulatory T cells are necessary for posttransplantation cyclophosphamide-mediated protection against GVHD in mice. Blood. 2014; 124(13):2131-2141. 44. Kasamon YL, Bolanos-Meade J, Prince GT, et al. Outcomes of nonmyeloablative HLAhaploidentical blood or marrow transplantation with high-dose post-transplantation cyclophosphamide in older adults. J Clin Oncol. 2015; 33(28):3152-3161. 45. McClune BL, Ahn KW, Wang HL, et al. Allotransplantation for patients age >/=40 years with non-Hodgkin lymphoma: encouraging progression-free survival. Biol Blood Marrow Transplant. 2014;20(7):960-968. 46. US National Library of Medicine. ClinicalTrials.gov [online]; 2013. Available from: https: // clinicaltrials.gov/ct2/ show/NCT01879072. Last Accessed: 21st December 2016.

haematologica | 2017; 102(5)


ARTICLE

Cell Therapy & Immunotherapy

The effect of inter-unit HLA matching in double umbilical cord blood transplantation for acute leukemia

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Claudio Brunstein,1 Mei-Jie Zhang,2,3 Juliet Barker,4 Andrew St. Martin,2 Asad Bashey,5 Marcos de Lima,6 Jason Dehn,7 Peiman Hematti,8 Miguel-Angel Perales,4 Vanderson Rocha,9 Mary Territo,10 Daniel Weisdorf1 and Mary Eapen2

1 University of Minnesota Medical Center, Minneapolis, MN, USA; 2Center for International Blood and Marrow Transplant Research, Department of Medicine, Medical College of Wisconsin, Milwaukee, WI, USA; 3Division of Biostatistics, Institute for Health and Society, Medical College of Wisconsin, Milwaukee, WI, USA; 4Adult Bone Marrow Transplant Services, Department of Medicine, Memorial Sloan-Kettering Cancer Center, and Department of Medicine, Weill Cornell Medical College, New York, NY, USA; 5Blood and Marrow Transplant Program at Northside Hospital, Atlanta, GA, USA; 6Department of Medicine, Seidman Cancer Center, University Hospitals Case Medical Center, Cleveland, OH, USA; 7National Marrow Donor Program/Be the Match, Minneapolis, MN, USA; 8Division of Hematology/Oncology/Bone Marrow Transplantation, Department of Medicine, University of Wisconsin Hospital and Clinics, Madison, WI, USA; 9Churchill Hospital, Oxford, UK and 10UCLA Center for Health Sciences, Los Angeles, CA, USA

Haematologica 2017 Volume 102(5):941-947

ABSTRACT

T

he effects of inter-unit HLA-match on early outcomes with regards to double cord blood transplantation have not been established. Therefore, we studied the effect of inter-unit HLA-mismatching on the outcomes of 449 patients with acute leukemia after double cord blood transplantation. Patients were divided into two groups: one group that included transplantations with inter-unit mismatch at 2 or less HLA-loci (n=381) and the other group with inter-unit mismatch at 3 or 4 HLA-loci (n=68). HLA-match considered low resolution matching at HLA-A and B loci and allele-level at HLA-DRB1, the accepted standard for selecting units for double cord blood transplants. Patients', disease, and transplant characteristics were similar in the two groups. We observed no effect of the degree of inter-unit HLA-mismatch on neutrophil (Hazard Ratio 1.27, P=0.11) or platelet (Hazard Ratio 0.1.13, P=0.42) recovery, acute graft-versus-host disease (Hazard Ratio 1.17, P=0.36), treatment-related mortality (Hazard Ratio 0.92, P=0.75), relapse (Hazard Ratio 1.18, P=0.49), treatment failure (Hazard Ratio 0.99, P=0.98), or overall survival (Hazard Ratio 0.98, P=0.91). There were no differences in the proportion of transplants with engraftment of both units by three months (5% after transplantation of units with inter-unit mismatch at â&#x2030;¤2 HLA-loci and 4% after transplantation of units with inter-unit mismatch at 3 or 4 HLA-loci). Our observations support the elimination of inter-unit HLA-mismatch criterion when selecting cord blood units in favor of optimizing selection based on individual unit characteristics.

Introduction Double umbilical cord blood (UCB) has allowed adults and larger adolescents to proceed to allogeneic transplantation with this donor type when a single adequately dosed unit is not available. At its inception at the University of Minnesota, the UCB units were required to be at least 4/6 HLA-matched to the patient and to each other, not necessarily at the same loci.1,2 The inter-unit HLA-match requirement was based on what was known about the effect of HLA-mismatching on the outcomes of single UCB transplantation in order to minimize the risk of cross-rejection haematologica | 2017; 102(5)

Correspondence: meapen@mcw.edu

Received: October 18, 2016. Accepted: January 20, 2017. Pre-published: January 25, 2017. doi:10.3324/haematol.2016.158584 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/941 Š2017 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.

941


C. Brunstein et al.

between the two donor units and the potential for increased risk of graft failure. However, as double UCB transplantation became more widely used, the degree of inter-unit HLA-mismatching allowed has been relaxed to adjust for institutional practice.3-5 The rationale for relaxing inter-unit HLA-matching was to allow for optimization of

Table 1. Patients', disease and transplantation characteristics.

Variables Number Age ≤ 20 years 21 – 40 years > 40 years Sex, male Performance score 90 – 100 < 90 Not reported HCT-CI score 0 1–2 ≥3 Cytomegalovirus serostatus Positive Negative Not reported Disease Acute myeloid leukemia Acute lymphoblastic leukemia Disease status First complete remission Second complete remission Third complete remission Cytogenetic risk Favorable/intermediate Poor Not reported Conditioning regimen TBI 200 + cyclophosphamide + fludarabine TBI 200 + treosulfan TBI 400 + cyclophosphamide + fludarabine + thiotepa Melphalan (< 150 mg/m2) + other agents TBI≥1000 + cyclophosphamide + fludarabine Graft-versus-host disease prophylaxis Tacrolimus + mycophenolate Cyclosporine + mycophenolate Donor-recipient HLA-match 4/6 + 4/6 4/6 + 5/6 4/6 + 6/6 5/6 + 5/6 5/6 + 6/6 6/6 + 6/6 Transplant period 2008 – 2010 2011 – 2014 Median follow up (range) months

cell dose and donor-recipient HLA-matching of each unit, instead of compromising one or both criteria in order to find a pair of UCB units that match each other. Such an approach would potentially improve the engraftment potential of each unit despite the fact that a single UCB unit predominates long term in most patients. The

≤ 2 inter-unit HLA-mismatch

≥3 inter-unit HLA-mismatch

381

68

72 (19%) 105 (27%) 204 (54%) 205 (54%)

12 (17%) 13 (19%) 43 (64%) 37 (54%)

274 (72%) 101 (27%) 6 ( 2%)

47 (69%) 19 (28%) 2 ( 3%)

136 (36%) 121 (32%) 124 (33%)

20 (30%) 20 (30%) 28 (40%)

248 (65%) 129 (34%) 4 ( 1%)

49 (72%) 17 (25%) 2 ( 3%)

247 (65%) 134 (35%)

49 (72%) 19 (28%)

125 (33%) 32 (8%)

224 (59%) 32 (47%) 5 (7%)

259 (68%) 102 (27%) 20 (5%)

42 (62%) 19 (28%) 7 (10%)

142 (37%) 14 (4%) 27 (7%) 12 (3%) 186 (49%)

21 (31%) 5 (7%) 4 (6%) 4 (6%) 34 (50%)

87 (23%) 294 (77%)

21 (31%) 47 (69%)

116 (30%) 87 (23%) 2 (1%) 120 (31%) 26 (7%) 30 (8%)

48 (71%) 20 (29%) __ __ __ __

175 (46%) 276 (54%) 36 (3-77)

32(47%) 36 (53%) 36 (4-74)

P

0.27

0.93 0.70

0.17

0.19

0.25

0.07 31 (46%) 0.25

0.43

0.15

N/A

0.86

N/A

HCT-CI: hematopoietic cell transplant co-morbidity index; TBI: total body irradiation; HLA: human leukocyte antigen; N/A: not applicable.

942

haematologica | 2017; 102(5)


Inter-unit HLA-match and double UCB transplant

engraftment of each potential unit had to be optimized as we cannot predict the predominant unit at time of selection. Moreover, when a limited number of UCB units are available, as in ethnically and racially diverse populations, it may be very difficult to meet a strict inter-unit HLAmatching criterion. As institutions tend to follow uniform practices when selecting UCB units,1-6 opportunities to study the effect of inter-unit HLA-mismatch on hematopoietic recovery and survival at individual transplant centers are limited. Reports from a single institution on the characteristics of the dominant unit in the setting of double UCB transplantation and myeloablative conditioning regimens support the fact that cell dose is the only characteristic independently associated with engraftment.7,8 Unit-unit HLA match did not affect sustained engraftment, but recipients of units closely matched to each other were more likely to demonstrate initial engraftment of both units.7 Both reports were from a single institution and included modest sample sizes of 129 and 84 double UCB transplants. Thus, we designed a study using data reported to the Center for International Blood and Marrow Transplant Research (CIBMTR) to determine if the degree of interunit HLA-mismatch defined as 2 or less HLA-loci versus 3

or 4 HLA-loci between units would affect early outcomes after double UCB transplantation. We hypothesized that any effect of inter-unit HLA-mismatching would be evident within three months after transplantation as one UCB unit typically predominates beyond this period.

Methods Patients The CIBMTR is a voluntary group of over 350 transplant centers that contribute data prospectively on consecutive transplants performed at each individual center. All patients are followed longitudinally until death or lost to follow up. Seventy-eight centers contributed patients, and transplants were performed between 2008 and 2014 in the United States. Eligible patients were aged 1 to 70 years with acute myeloid or lymphoblastic leukemia (n=449), and were in a first or subsequent complete remission. All received two UCB units, myeloablative or reduced intensity conditioning regimen, and cyclosporine or tacrolimus with mycophenolate for graft-versus-host disease (GvHD) prophylaxis. Exclusion criteria were transplants for relapse or primary induction failure (n=226) and anti-thymocyte globulin (ATG)-containing regimens (n=39). The Institutional Review Board of the National Marrow Donor Program approved this study.

End points Table 2. Effect of inter-unit UCB HLA-match on early outcomes.

Outcomes

Hazard Ratio (95% Confidence Interval)

P

Neutrophil recovery Inter-unit mismatch ≤2 HLA-loci 1.00 Inter-unit mismatch ≥3 HLA-loci 0.83 (0.63 – 1.08) 0.17 Infused TNC (sum unit 1 + unit 2) ≥4 vs. <4 x 107/kg 1.39 (1.09 – 1.79) 0.008 Cytomegalovirus serostatus Positive vs. negative 0.75 (0.61 – 0.96) 0.006 Transplant-conditioning regimen Reduced intensity vs. 1.49 (1.22 – 1.82) <0.001 myeloablative regimen Platelet recovery Inter-unit mismatch ≤2 HLA-loci 1.00 Inter-unit mismatch ≥3 HLA-loci 1.13 (0.84 – 1.53) 0.42 Transplant-conditioning regimen Reduced intensity vs. 1.82 (1.45 – 2.27) <0.001 myeloablative regimen Grade II-IV acute GvHD Inter-unit mismatch ≤2 HLA-loci 1.00 Inter-unit mismatch ≥3 HLA-loci 1.17 (0.83 – 1.66) 0.36 Transplant-conditioning regimen Reduced intensity vs. 0.56 (0.43 – 0.73) <0.001 myeloablative regimen Overall mortality Inter-unit mismatch ≤2 HLA-loci 1.00 Inter-unit mismatch ≥3 HLA-loci 0.83 (0.46 – 1.46) 0.52 Infused TNC (sum unit 1 + unit 2) ≥4 vs. <4 x 107/kg 0.95 (0.57 – 1.57) 0.84 Age 21 – 40 years vs. ≤20 years 0.89 (0.37 – 2.14) 0.80 >40 years vs. ≤20 years 2.11 (1.02 – 4.38) 0.04 >40 years vs. 21 – 40 years 2.36 (1.26 – 4.42) 0.007 UCB: umbilical cord blood; GvHD: graft-versus-host disease;TNC: total nucleated cell. haematologica | 2017; 102(5)

The primary end point was overall survival at three months and one year. Death from any cause was considered an event and surviving patients were censored at last follow up. Neutrophil recovery was defined as achieving an absolute neutrophil count of 0.5 x109/L or more for three consecutive days; and platelet recovery as platelets 20x109/L or more, without transfusion support for seven days. Graft failure was defined as 5% or less donor chimerism or absence of neutrophil recovery. Incidences of grades 2 to 4 acute GvHD were based on reports from each transplant center using standard criteria. Relapse was defined as leukemia recurrence (morphological, cytogenetic or molecular), and non-relapse mortality was defined as death in remission.

Statistical analysis

Differences between groups were compared using the χ2 test. The probability of overall survival was calculated using the Kaplan-Meier estimator.9 The probability of neutrophil and platelet recovery, and acute and chronic GvHD were calculated using the cumulative incidence estimator to accommodate competing risks.10 Cox regression models were built to study the effect of inter-unit HLA mismatch and other factors associated with hematopoietic recovery, acute GvHD, day-100 mortality and 1-year relapse, non-relapse mortality and overall mortality.11 Variables tested include: inter-unit HLA mismatch, age, sex, performance score, hematopoietic cell transplant co-morbidity (HCT-CI) score, cytomegalovirus (CMV) serostatus, disease, disease status, and transplant conditioning regimen intensity and transplant period. All variables tested met the assumptions for proportionality, and there were no first order interactions between inter-unit HLA mismatch and other variables held in the final multivariate model. All variables that achieved P≤0.05 were held in the final multivariate model, with the exception of the variable for inter-unit mismatch that was held in all steps of model building and the final model regardless of level of significance. Transplant center effect on survival was tested using the frailty approach.12 All P-values are two-sided. All analyses were carried out using SAS v.9.3 (Cary, NC, USA). 943


C. Brunstein et al.

Results Patients’, disease and transplant characteristics The characteristics of 449 patients with acute leukemia are summarized in Table 1. Donor-recipient and unit-unit HLA-match considered antigen level matching at HLA-A and -B loci and allele-level at HLA-DRB1, the accepted standard for selecting units for double UCB transplants. Clinical practice tolerated multiple HLA-mismatching between units and no more than 2 HLA-loci mismatches between each, and according to this the two recipient groups were created: 1) units were either matched (n=49) or mismatched to each other at 1 (n=113) or 2 (n=219) HLA-loci; and 2) units were mismatched to each other at 3 (n=60) or 4 HLA-loci (n=8). When inter-unit HLA mismatch was 3 or more, most units were mismatched to the recipient at 2 HLA-loci and double mismatch at the same HLA-locus. The median infused total nucleated cell (TNC) dose was 4.26 (range 2.31-5.76) x107/kg for transplants with 2 or less inter-unit mismatch and 4.93 (2.99-6.00) x107/kg for transplants with 3 or more inter-unit mismatch. There were no significant differences in patient age, sex, CMV serostatus, performance score and HCT-CI between the two groups. Although there were no differences between the groups by leukemia type or cytogenetic risk, more transplants with inter-unit mismatch at 2 or less

HLA-loci were in first complete remission. Patients were equally likely to receive a myeloablative or reduced intensity conditioning regimen, and all received a calcineurin inhibitor with mycophenolate mofetil for GvHD prophylaxis. The median follow up of surviving patients was 36 months in both groups.

Early outcomes Results of multivariate analysis for the effect of interunit HLA mismatch on outcomes are shown in Table 2. Inter-unit HLA mismatch was not associated with hematopoietic recovery, acute grade II-IV GvHD or overall survival at three months. Independent of inter-unit HLA mismatch, neutrophil recovery was more likely with infused TNC 4 or more x107/kg (sum unit1 + unit2). We investigated whether the TNC of a single unit would influence its engraftment in the setting of 2 or less HLAloci compared to 3 or more HLA-loci inter-unit mismatched transplants and did not see such an effect (P=0.20; paired t-test). Independent of inter-unit HLAmatch and total infused TNC, recovery was more likely with reduced intensity transplant conditioning regimen and less likely for CMV seropositive patients. The day-28 incidence of neutrophil recovery after 2 or less HLA-loci inter-unit mismatch transplantation was 67% [95% Confidence Interval (CI): 63-72] and after 3 or more HLAloci inter-unit mismatch transplantation, 76% (95%CI: 66-

A

B

C

D

Figure 1. Neutrophil and platelet recovery, grade II-IV acute graft-versus-host disease (GvHD) and overall survival. (A) Day 28 neutrophil recovery: the incidence of neutrophil recovery after ≤2 HLA-loci (A) and ≥3 HLA-loci (B) inter-unit mismatched transplants. (B) Day 100 platelet recovery: the adjusted incidence of platelet recovery after ≤2 HLA-loci (A) and ≥3 HLA-loci (B) inter-unit mismatched transplants. (C) Day 100 Grade II-IV acute GvHD: the incidence of acute GvHD after ≤2 HLA-loci (A) and ≥3 HLA-loci (B) inter-unit mismatched transplants. (D) Day 100 survival: the adjusted probability of survival after ≤2 HLA-loci (A) and ≥3 HLA-loci (B) inter-unit mismatched transplants.

944

haematologica | 2017; 102(5)


Inter-unit HLA-match and double UCB transplant

Table 3. Effect of inter-unit UCB HLA-match 1-year after transplantation.

Outcomes

Hazard Ratio (95% Confidence Interval)

Non-relapse mortality Inter-unit mismatch ≤2 HLA-loci 1.00 Inter-unit mismatch ≥3 HLA-loci 0.92 (0.57 – 1.50) Other factors associated with non-relapse mortality Age 21 – 40 years vs. ≤20 years 1.50 (0.85 – 2.63) >40 years vs. ≤20 years 3.06 (1.69 – 5.54) >40 years vs. 21 – 40 years 2.03 (1.26 – 3.29) Performance score <90 vs. 90-100 1.78 (1.22 – 2.60) Transplant-conditioning regimen Reduced intensity vs. 0.46 (0.29 – 0.73) myeloablative regimen Relapse Inter-unit mismatch ≤2 HLA-loci 1.00 Inter-unit mismatch ≥3 HLA-loci 1.18 (0.74 – 1.88) Other factors associated with relapse Transplant-conditioning regimen Reduced intensity vs. 3.14 (2.14 – 4.63) myeloablative regimen Overall mortality Inter-unit mismatch ≤2 HLA-loci 1.00 Inter-unit mismatch ≥3 HLA-loci 0.98 (0.69 – 1.39) Other factors associated with mortality Age 21 – 40 years vs. ≤20 years 0.96 (0.62 – 1.48) >40 years vs. ≤20 years 1.82 (1.26 – 2.64) >40 years vs. 21 – 40 years 1.90 (1.38 – 2.62) Performance score <90 vs. 90-100 1.68 (1.28 – 2.22)

Causes of death

P

0.75

0.15 <0.001 0.004

Number Recurrent leukemia Graft-versus-host disease Infection Interstitial pneumonitis Other causes

≤ 2 inter-unit HLA-mismatch

≥ 3 inter-unit HLA-mismatch

205 92 (45%) 22 (11%) 27 (13%) 10 ( 5%) 54 (26%)

37 17 (46%) 5 (17%) 6 (16%) 1 ( 3%) 8 (22%)

0.003 <0.001

0.49

<0.001

0.91

0.85 0.001 <0.001 <0.001

87) (P=0.26) (Figure 1A). The corresponding day-42 incidences were 85% (95%CI: 82-89) and 90% (95%CI: 8196) (P=0.28). Independent of inter-unit HLA mismatch, day-28 neutrophil recovery was less likely in patients who were CMV seropositive (64%, 95%CI: 59-70) compared to 74% (95%CI: 67-81) in CMV seronegative patients (P=0.03). The incidence of CMV reactivation within 100 days post transplant was higher in CMV seropositive (52%, 95%CI: 46-58) compared to seronegative patients (7%, 95%CI: 4-12) (P<0.001). There were no differences in the proportion of patients who died between the two groups: there were 85 deaths (55%; 85 of 156) in the CMV seropositive group and 7 deaths (63%; 7 of 11) in the CMV seronegative group (P=0.50). Platelet recovery was more likely with reduced intensity transplant conditioning. The day-100 incidence of platelet recovery, adjusted for transplant conditioning regimen intensity, was 72% (95%CI: 67-77) and 72% (95%CI: 6082) after 2 or less HLA-loci and 3 or more HLA-loci interunit mismatch transplantation, respectively (P=0.95) (Figure 1B). Inter-unit HLA mismatch was not associated with graft failure (primary or secondary). Seventy of 381 (18%) patients transplanted with 2 or less HLA-loci inter-unit mismatch and 8 of 68 (13%) patients transplanted with 3 or more HLA-loci inter-unit mismatch developed graft failhaematologica | 2017; 102(5)

Table 4. Causes of death.

ure (P=0.19). We also explored whether engraftment of both UCB units (dual engraftment) varied with inter-unit HLA match and did not find such an effect; 6% of recipients of 2 or less inter-unit HLA-mismatched and 4% of recipients of 3 or more inter-unit HLA-mismatch experienced dual engraftment (P=0.65). Dual engraftment was explored on the first reported chimerism assay performed 30 days +10 days after transplantation. The only factor associated with acute grade 2-4 GvHD was transplant conditioning regimen intensity; risks were lower with reduced intensity conditioning regimens. The day-100 incidence of acute grade 2-4 GvHD, after adjusting for transplant conditioning regimen intensity, was 49% (95%CI: 44-54) and 57% (95%CI: 46-69), after 2 or less HLA-loci and 3 or more HLA-loci inter-unit mismatched transplants, respectively (P=0.27) (Figure 1C). The only risk factor for early mortality was age; risks were higher for patients aged 40 years and older independently of inter-unit HLA-mismatch and TNC. TNC was not associated with early survival. The day-100 probability of survival, adjusted for age was 83% (95%CI: 7987) and 78% (95%CI: 67-87) after 2 or less HLA-loci and 3 or more HLA-loci inter-unit mismatched transplants, respectively (P=0.60) (Figure 1D). Bacterial, viral and fungal infections were common within the first 100 days in both groups; infection rates were 59% and 60% after 2 or less HLA-loci and 3 or more HLA-loci inter-unit mismatched transplants, respectively (P=0.82). We also explored transplant outcomes considering interunit HLA-match 0-1 versus 2 or more. Compared to transplants with inter-unit HLA-match 0-1, the risks of neutrophil recovery (HR 0.97, 95%CI: 0.77-1.22; P=0.82), platelet recovery (HR 0.90, 95%CI: 0.72-1.14; P=0.40), grade II-IV acute GvHD (HR 1.14, 95%CI: 0.87-1.50; P=0.33) and overall mortality (HR 1.17, 95%CI: 0.83-1.20; P=0.35) were not significantly different for transplants with inter-unit HLA-match 2 or more.

One-year overall survival, non-relapse mortality, relapse and chronic GvHD There were no differences in risks for overall mortality, non-relapse mortality or relapse by inter-unit HLA-mismatch beyond the early post-transplant period (Table 3). We tested for an effect of transplant center; none was found (P=0.37). Risks for overall mortality and non-relapse mortality were associated with poor performance scores of 80 or lower, and in patients older than 40 years of age. Transplant-conditioning regimen intensity was associated with non-relapse mortality and relapse. Non-relapse mor945


C. Brunstein et al.

tality risks were lower and relapse risks higher with reduced intensity compared to myeloablative conditioning regimens. The causes of death are shown in Table 4. Leukemia recurrence was the predominant cause of death in both groups. There was no difference in the 1-year cumulative incidences of chronic GvHD; 28% (95%CI: 24-33) and 35% (95%CI: 24-47) after transplants with inter-unit HLA-mismatch 2 or less and 3 or more, respectively (P=0.40). For most transplants with inter-unit HLA-mismatch 3 or more, each unit was mismatched to the patient at 2 HLAloci. Therefore, to ensure the observed results were independent of unit-patient HLA-mismatch, we performed a subset analysis that explored possible differences in survival by inter-unit HLA-mismatch for transplantations mismatched at 2 HLA-loci. There were 203 transplants mismatched at 2 HLA-loci with inter-unit mismatch at 2 HLA-loci and 68 transplants mismatched at 2 HLA-loci with inter-unit mismatch at 3 or 4 HLA-loci. Consistent with the main analysis, we did not observe any differences in survival by inter-unit HLA mismatch (HR 1.09, 95%CI: 0.75-1.59; P=0.63), adjusted for patient age. Similarly, there were no differences between the groups with regards to neutrophil recovery (HR 1.26, 95%CI: 0.91.75; P=0.16), platelet recovery (HR 1.23, 95%CI: 0.881.69; P=0.22), acute GvHD (1.13, 95%CI: 0.78-1.64; P=0.52), non-relapse mortality (HR 0.99, 95%CI: 0.571.72; P=0.97) and relapse (HR 1.16, 95%CI: 0.67-2.00; P=0.61).

Discussion We studied the effect of the inter-unit HLA match on outcomes of double UCB transplantation for acute leukemia and did not find an association between interunit HLA-mismatch and outcomes. Specifically, there were no differences in hematopoietic recovery, acute GvHD or survival, demonstrating that inter-unit HLAmatch is not relevant when selecting UCB units for double UCB transplantation for acute leukemia. The only unit characteristic associated with neutrophil recovery was cell dose. Transplantations of UCB units with a combined infused TNC 4x107/kg or more was associated with faster neutrophil recovery, but infused TNC was not associated with survival or non-relapse mortality. There were no differences in engraftment of the dominant and non-dominant units based on inter-unit HLA match or unit TNC within the first month after transplantation. Our observations are not in keeping with a single report of early engraftment of the non-dominant UCB unit when the cell dose of the dominant unit was low (CD34+ <1.20x105/kg).8 The current analyses used TNC dose and the single institution report8 used CD34+ dose. The heterogeneity of CD34 measurements across laboratories prevents us from studying the effects of CD34 dose in the setting of registry studies. As CD34 is a subset of TNC, a higher infused TNC implies higher infused CD34. The cryopreserved TNC dose of 2.5x107/kg or more, and at least 4/6 HLA-matching to the patient, considering HLA-A and -B at the antigen level and -DRB1 at the allele level, are the cornerstones of initial UCB unit selection for double UCB transplantation. The additional step of matching units to each other has added complexity and has, at times, limited options with respect to selecting the 946

best available UCB unit. Patients with common haplotypes will have several UCB units that meet the above criteria to compose a double UCB graft that includes units that are at least 4/6 HLA-matched to the patient and to each other. In contrast, for racial minorities, identifying multiple UCB units cryopreserved nucleated cell dose 2.5x107/kg or more and at least 4/6 HLA-matching to the patient can be challenging, and the added burden of interunit matching limited to no more than mismatching at 2 HLA-loci may result in selecting individual units that are less desirable, or at times prohibitive, to the extent that transplantation is denied.13 Our results support a focus on the selection of each UCB unit with at least the minimum desired dose of 2.5x107/kg, and thereafter the best HLAmatch to the patient. In the current analysis, 2-year overall survival for adults with acute leukemia in remission are 47% (95%CI: 42-52) and 45% (95%CI: 33-58) after double UCB transplants with inter-unit HLA-mismatch 2 or less and 3 or more, respectively. The corresponding non-relapse mortality rates were 29% (95%CI: 25-34) and 32% (95%CI: 21-44), and relapse rates 29% (95%CI: 24-34) and 31% (95%CI: 20-43) at two years post transplant. Patients older than 40 years of age and those with performance scores of 80 or lower were at higher risk for overall and non-relapse mortality. Age is not a modifiable factor, but early referral may result in transplantations with better performance scores. The decision to offer an ablative or reduced intensity regimen is based on several factors, including age, fitness and organ function. Consistent with other reports, a potential survival advantage with reduced intensity conditioning regimens was negated by higher relapse.14,15 These results support the view that UCB transplants are desirable for patients without a fully HLA-matched related or unrelated donor. Although neutrophil recovery was less likely in CMV seropositive patients, and these patients were more likely to experience CMV reactivation, this was not associated with higher mortality when compared to CMV seronegative patients. Our study has limitations that we have addressed by performing carefully controlled analyses. First, consistent with current clinical practice, only 15% of transplantations chose UCB units with inter-unit mismatch at 3 or 4 HLA-loci, implying that it is more likely to achieve interunit HLA matching than not. Among centers that contributed more than 5 patients (n=23 centers), at 6 centers 20%-30% of transplants used UCB units with inter-unit mismatch 3 or more; at the remaining 17 centers, fewer than 5% of transplants used UCB units with inter-unit mismatch 3 or more. Although this is a modest study population, the observed Hazard Ratios are close to 1.00, supporting our recommendation that inter-unit mismatch may be ignored when selecting UCB units for double UCB transplantation for acute leukemia. Second, interunit mismatch is confounded by HLA-mismatch. Therefore, a subset analysis limited to 4/6 double UCB transplants was carried out which confirmed the results of the main analysis. Third, our groupings for inter-unit mismatch considered 2 or less HLA-loci mismatch versus 3 or more HLA-loci, based on clinical practice. However, we also looked for possible differences, such as 0-1 HLAlocus versus 2 or more HLA-loci mismatch, and found none. Data, mostly from the single UCB transplantation setting, support the use of donor specific anti-HLA antibodhaematologica | 2017; 102(5)


Inter-unit HLA-match and double UCB transplant

ies,16 high resolution HLA-matching,17 matching at HLAC,18 CD34+ cell dose,7,8,19,20 and cell viability8 to refine UCB unit selection. However, these data are not always reproducible in the double UCB focused studies.21,22 Reviewing HLA typing reported to the CIBMTR for the current analyses, the majority of transplantations did not consider allele-level HLA-match or matching at the HLA-C locus for unit selection. It is plausible that matching at the HLAC locus for units that are matched to the patient at HLAA, -B and DRB1, or in the presence of a single locus mismatch at A, B or DRB1 may minimize mortality risks.18 Others have reported that selecting units for double UCB transplants based on high resolution HLA-matching is feasible for most patients without compromising cell dose.23 Only when UCB unit selection considers matching at the HLA-C locus and high-resolution HLA-match criteria can we design studies to explore the role of better HLA-match for double UCB transplants. In the setting of single UCB transplantation for leukemia, processing and banking practices at the publicly funded US Cord Blood Banks had no

References 1. Barker JN, Weisdorf DJ, DeFor TE, et al. Transplantation of 2 partially HLAmatched umbilical cord blood units to enhance engraftment in adults with hematologic malignancy. Blood. 2005; 105(3):1343-1347. 2. Brunstein CG, Barker JN, Weisdorf DJ, et al. Umbilical cord blood transplantation after nonmyeloablative conditioning: impact on transplantation outcomes in 110 adults with hematologic disease. Blood. 2007; 110(8):3064-3070. 3. Ponce DM, Zheng J, Gonzales AM, et al. Reduced late mortality risk contributes to similar survival after double-unit cord blood transplantation compared with related and unrelated donor hematopoietic stem cell transplantation. Biol Blood Marrow Transplant. 2011;17(9):1316-1326. 4. Wagner JE Jr, Eapen M, Carter S, et al. Oneunit versus two-unit cord-blood transplantation for hematologic cancers. N Engl J Med. 2014;371(18):1685-1694. 5. Kanda J, Rizzieri DA, Gasparetto C, et al. Adult dual umbilical cord blood transplantation using myeloablative total body irradiation (1350 cGy) and fludarabine conditioning. Biol Blood Marrow Transplant. 2011;17(6):867-874. 6. Ostronoff F, Milano F, Gooley T, et al. Double umbilical cord blood transplantation in patients with hematologic malignancies using a reduced-intensity preparative regimen without antithymocyte globulin. Bone Marrow Transplant. 2013; 48(6):782-786. 7. Avery S, Shi W, Lubin M, et al. Influence of infused cell dose and HLA match on engraftment after double-unit cord blood allografts. Blood. 2011;117(12):3277-3285; quiz 3478.

haematologica | 2017; 102(5)

effect on early survival,24 although a single center report concluded that units provided by the non-Netcord Foundation for the Accreditation of Cellular Therapy accredited Cord Blood Banks were associated with low recovery of viable CD34+ cells.8 Data reported to the CIBMTR suggest approximately 80% of UCB transplants in the US for patients older than 18 years use two cord blood units. Therefore, eliminating inter-unit HLA-mismatch restriction will allow for a larger number of units to be considered when selecting units for double UCB transplants. Funding The CIBMTR is supported by Public Health Service Grant U24-CA076518 from the National Cancer Institute, the National Heart, Lung and Blood Institute and the National Institute of Allergy and Infectious Diseases, a contract HHSH250201200016C with Health Resources and Services Administration (HRSA/DHHS) and grants N00014-15-1-0848 and N00014-16-1-2020 from the Office of Naval Research.

8. Purtill D, Smith K, Devlin S, et al. Dominant unit CD34+ cell dose predicts engraftment after double-unit cord blood transplantation and is influenced by bank practice. Blood. 2014;124(19):2905-2912. 9. Kaplan EL, Meier P. Nonparametric Estimation from Incomplete Observations. J Am Stat Assoc. 1958;53(282):457-481. 10. Lin DY. Non-parametric inference for cumulative incidence functions in competing risks studies. Stat Med. 1997;16(8):901910. 11. Cox DR. Regression Models and LifeTables. J R Stat Soc Series B Stat Methodol. 1972;34(2):187-220. 12. Andersen PK, Klein JP, Zhang MJ. Testing for centre effects in multi-centre survival studies: a Monte Carlo comparison of fixed and random effects tests. Stat Med. 1999; 18(12):1489-1500. 13. Gragert L, Eapen M, Williams E, et al. HLA match likelihoods for hematopoietic stemcell grafts in the U.S. registry. N Engl J Med. 2014;371(4):339-348. 14. Luger SM, Ringden O, Zhang MJ, et al. Similar outcomes using myeloablative vs reduced-intensity allogeneic transplant preparative regimens for AML or MDS. Bone Marrow Transplant. 2012;47(2):203211. 15. Marks DI, Wang T, Perez WS, et al. The outcome of full-intensity and reducedintensity conditioning matched sibling or unrelated donor transplantation in adults with Philadelphia chromosome-negative acute lymphoblastic leukemia in first and second complete remission. Blood. 2010; 116(3):366-374. 16. Cutler C, Kim HT, Sun L, et al. Donor-specific anti-HLA antibodies predict outcome in double umbilical cord blood transplantation. Blood. 2011;118(25):6691-6697. 17. Eapen M, Klein JP, Ruggeri A, et al. Impact of allele-level HLA matching on outcomes

18.

19.

20.

21.

22.

23.

24.

after myeloablative single unit umbilical cord blood transplantation for hematologic malignancy. Blood. 2014;123(1):133-140. Eapen M, Klein JP, Sanz GF, et al. Effect of donor-recipient HLA matching at HLA A, B, C, and DRB1 on outcomes after umbilical-cord blood transplantation for leukaemia and myelodysplastic syndrome: a retrospective analysis. Lancet Oncol. 2011;12(13):1214-1221. Sanz J, Sanz MA, Saavedra S, et al. Cord blood transplantation from unrelated donors in adults with high-risk acute myeloid leukemia. Biol Blood Marrow Transplant. 2010;16(1):86-94. Wagner JE, Barker JN, DeFor TE, et al. Transplantation of unrelated donor umbilical cord blood in 102 patients with malignant and nonmalignant diseases: influence of CD34 cell dose and HLA disparity on treatment-related mortality and survival. Blood. 2002;100(5):1611-1618. Brunstein CG, Petersdorf EW, DeFor TE, et al. Impact of Allele-Level HLA Mismatch on Outcomes in Recipients of Double Umbilical Cord Blood Transplantation. Biol Blood Marrow Transplant. 2016;22(3):487492. Brunstein CG, Noreen H, DeFor TE, Maurer D, Miller JS, Wagner JE. Anti-HLA antibodies in double umbilical cord blood transplantation. Biol Blood Marrow Transplant. 2011;17(11):1704-1708. Dahi PB, Ponce DM, Devlin S, et al. Donorrecipient allele-level HLA matching of unrelated cord blood units reveals high degrees of mismatch and alters graft selection. Bone Marrow Transplant. 2014;49(9):1184-1186. Ballen KK, Logan BR, Laughlin MJ, et al. Effect of cord blood processing on transplantation outcomes after single myeloablative umbilical cord blood transplantation. Biol Blood Marrow Transplant. 2015; 21(4):688-695.

947


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Cell Therapy & Immunotherapy

Ferrata Storti Foundation

Haematologica 2017 Volume 102(5):948-957

IL-2 promotes early Treg reconstitution after allogeneic hematopoietic cell transplantation Brian C. Betts,1,2 Joseph Pidala,1,2 Jongphil Kim,3 Asmita Mishra,1 Taiga Nishihori,1 Lia Perez,1 Jose Leonel Ochoa-Bayona,1 Farhad Khimani,1 Kelly Walton,1 Ryan Bookout,1 Michael Nieder,1 Divis K. Khaira,1 Marco Davila,1,2 Melissa Alsina,1 Teresa Field,1 Ernesto Ayala,1 Frederick L. Locke,1 Marcie Riches,1 Mohamed Kharfan-Dabaja,1 Hugo Fernandez1 and Claudio Anasetti1,2 1 3

Department of Blood and Marrow Transplantation; 2Department of Immunology and Department of Biostatistics and Bioinformatics, Moffitt Cancer Center, Tampa, FL, USA

ABSTRACT

G

Correspondence: brian.betts@moffitt.org

Received: July 19, 2016. Accepted: January 18, 2017. Pre-published: January 19, 2017. doi:10.3324/haematol.2016.153072 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/948

raft-versus-host disease (GvHD) remains a major cause of transplant-related mortality. Interleukin-2 (IL-2) plus sirolimus (SIR) synergistically reduces acute GvHD in rodents and promotes regulatory T cells. This phase II trial tested the hypothesis that IL-2 would facilitate STAT5 phosphorylation in donor T cells, expand regulatory T cells, and ameliorate GvHD. Between 16th April 2014 and 19th December 2015, 20 patients received IL-2 (200,000 IU/m2 thrice weekly, days 0 to +90) with SIR (5-14 ng/mL) and tacrolimus (TAC) (3-7 ng/mL) after HLA-matched related or unrelated allogeneic hematopoietic cell transplantation (HCT). The study was designed to capture an increase in regulatory T cells from 16.0% to more than 23.2% at day +30. IL-2/SIR/TAC significantly increased regulatory T cells at day +30 compared to our published data with SIR/TAC (23.8% vs. 16.0%, P=0.0016; 0.052 k/uL vs. 0.037 k/uL, P=0.0163), achieving the primary study end point. However, adding IL-2 to SIR/TAC led to a fall in regulatory T cells by day +90 and did not reduce acute or chronic GvHD. Patients who discontinued IL-2 before day +100 showed a suggested trend toward less grade II-IV acute GvHD (16.7% vs. 50%, P=0.1475). We surmise that the reported accumulation of IL-2 receptors in circulation over time may neutralize IL-2, lead to progressive loss of regulatory T cells, and offset its clinical efficacy. The amount of phospho-STAT3+ CD4+ T cells correlated with donor T-cell activation and acute GvHD incidence despite early Tcell STAT5 phosphorylation by IL-2. Optimizing IL-2 dosing and overcoming cytokine sequestration by soluble IL-2 receptor may sustain lasting regulatory T cells after transplantation. However, an approach to target STAT3 is needed to enhance GvHD prevention. (clinicaltrials.gov identifier: 01927120).

Š2017 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.

948

Introduction Allogeneic hematopoietic cell transplantation (HCT) can cure high-risk hematologic malignancies and blood disorders. Graft-versus-host disease (GvHD) remains a major cause of transplant-related mortality. Sirolimus (SIR) is an immune suppressant that inhibits mTOR (mammalian target of rapamycin) and protein S6 phosphorylation.1 SIR permits interleukin-2 (IL-2) receptor activation of JAK1/JAK3 and STAT5, favoring the differentiation of regulatory T cells (Treg).1 Tregs express high affinity IL-2 receptors2 and respond to minimal levels of cytokine.3 Tregs are implicated in ameliorating GvHD,1 while IL-6 activates JAK2/STAT3 in T cells and fuels GvHD.4 A deficiency of Tregs relative to effector T cells is also associated with increased risk of chronic GvHD.5 A randomized phase II trial using matched related haematologica | 2017; 102(5)


IL-2 facilitates Treg reconstitution

and unrelated donors at the Moffitt Cancer Center (MCC) demonstrated that SIR/tacrolimus (SIR/TAC) was superior to methotrexate (MTX)/TAC in preventing grade II-IV acute GvHD, while facilitating Treg recovery.6 Later, a multicenter, phase III Blood and Marrow Transplant Clinical Trials Network trial reported equivalent rates of acute GvHD between the two regimens among recipients of matched related allografts, though SIR/TAC was associated with significantly less mucositis.7 Adding IL-2 to SIR/TAC is a rational strategy to improve Treg recovery after allogenic HCT given the beneficial immune reconstitution effects of the regimen.6 Interleukin-2 was initially shown to induce FOXP3 expression among peripheral blood mononuclear cells (PBMCs) in patients receiving donor lymphocyte infusions.8 IL-2 (1,000,000 IU/m2/day for 8 weeks) induces significant clinical responses as therapy for steroid-refractory chronic GvHD, in part by expanding Tregs in vivo.9,10 As GvHD prophylaxis, IL-2 (1-200,000 IU/m2 3x week for 612 weeks) combined with MTX/TAC facilitates Treg recovery after 30 days of therapy.11 This pediatric clinical trial reported no cases of grade II-IV acute GvHD and few viral infections occurred after allogeneic HCT, supporting the view that IL-2 reduced alloreactivity while sparing Teffector function.11 We hypothesized that replacing broadly suppressive MTX for SIR would increase Treg reconstitution in response to IL-2. It is known that TAC also impairs Treg recovery and function.12,13 Pre-clinical evidence in mice has shown IL-2 and SIR synergistically reduce GvHD,14 but further validation of this TAC-free approach is lacking in robust non-human primate models. We cautiously retained TAC in the IL-2/SIR/TAC regimen, given that others have shown that SIR without a calcineurin inhibitor is not sufficient to prevent acute GvHD.15 Moreover, our prior randomized trial proved that significant Treg expansion could be achieved with SIR even when combined with TAC.6 We had previously demonstrated that the ratio of pSTAT5 to pSTAT3 in pooled CD4+ T cells provides a functional immune rheostat.16 As such, high pSTAT5 would favor Tregs whereas high pSTAT3 would promote alloreactive conventional T cells. Here we report our results from a prospective, phase II trial of IL-2/SIR/TAC, with the intent to maximize donor T-cell STAT5 phosphorylation and Treg reconstitution.

Methods Study design and end points Interleukin-2 was combined with SIR/TAC in this phase II, University of South Florida Institutional Review Board approved trial (clinicaltrials.gov identifier: 01927120). The primary end point was %Tregs at day +30. The study was designed to capture a 1.45-fold increase in Tregs from 16.0%, as published with SIR/TAC,6 to more than 23.2% with IL-2/SIR/TAC (upper interquartile range of Treg reconstitution after a year of SIR/TAC6). With 20 evaluable patients, a one-sided t-test or appropriate nonparametric test would achieve 86% power at a significance level of 0.05. A 10% patient replacement for non-evaluable patients was allowed. Thrombotic microangiopathy (TMA)17 and venoocclusive disease/sinusoidal obstruction syndrome (VOD/SOS)18 were diagnosed according to standard criteria. Acute GvHD was assessed weekly until day +100.19 Chronic GvHD was evaluated according to NIH consensus criteria.20 haematologica | 2017; 102(5)

Patients From 16th April 2014 and 19th December 2015, 25 patients were assessed for eligibility. Five patients failed screening and were not enrolled. Eligibility criteria are detailed in the Online Supplementary Methods.

Treatment protocol Patients received an allogeneic HCT from HLA-matched (HLAA, -B, -C, and -DRB1)-related or -unrelated donors using PBSC.6 Conditioning included intravenous busulfan (Bu) [area-under-thecurve targeting 3500 (reduced-intensity) or 5300 (myeloablative) mM/L*min/day, respectively, of Bu 3500/Flu or Bu 5300/Flu] or melphalan (140 mg/m2) with fludarabine (Flu/Mel, reduced-intensity).

Graft-versus-host disease prophylaxis Sirolimus (5-14 ng/mL) and TAC (3-7 ng/ml) were monitored per ARCHITECT. IL-2 (aldesleukin, provided by Prometheus Labs, San Diego, CA, USA) was administered at 200,000 IU/m2 subcutaneously thrice weekly days 0 to +90 (+7). IL-2 was held for fever (>100.4°F), hypotension, creatinine more than 2 mg/dL, or weight gain over 10% of baseline. IL-2 was discontinued for grade II-IV acute GvHD, with the exception of isolated upper gastrointestinal (GI) stage 1 disease.

Regulatory T-cell reconstitution and suppressive function CD4+ Tregs were identified by surface expression of CD25 and lack of CD1272,21 as gated in our prior SIR/TAC trial (Online Supplementary Figure S1A).6 Activated CD4+, conventional T cells (Tconv) were characterized by co-expression of CD25 and CD127 (Online Supplementary Figure S1A).22-24 Day +30 Treg suppressive potency was verified (n=3). Tregs and Tconv were purified by a cell sorter (BD Biosciences). Tregs were added to 5x103 Tconv, and stimulated with CD3/CD28 beads (Invitrogen, Carlsbad, CA, USA). Total T-cell proliferation was measured on day +3 [CellTiter 96 Aqueous One Solution Cell Proliferation Assay (MTS), Promega, Madison, WI, USA].4

CD4+ T-cell STAT3, STAT5, and S6 protein phosphorylation T-cell phospho-proteins were measured as described in the Online Supplementary Methods.4,16,25

Statistical analysis For comparisons of independent data sets, the Mann-Whitney test was used. The paired t-test was used for paired comparisons. ANOVA was used for group comparisons. The Gray method was used to evaluate the difference in acute GvHD incidence rates.26 Survival, cumulative incidence of relapse, and chronic GvHD were analyzed. SAS 9.3 (SAS Institute, Cary, NC, USA) and Prism software v.5.04 (GraphPad Software, San Diego, CA, USA) were used to conduct analyses. P<0.05 was considered significant. Given the sole interest in capturing a significant increase (not a decrease) in Tregs caused by IL-2 at day +30 after allogeneic HCT, a one-tailed test was used to assess the primary end point. A two-tailed test was used for all others.

Results Patientsâ&#x20AC;&#x2122; characteristics and therapy adherence A total of 20 patients received IL-2/SIR/TAC. Patients' characteristics are described in Table 1. The IL-2/SIR/TAC and SIR/TAC patients were similar according to age, sex, diagnosis, and conditioning regimens.6,27,28 The use of matched-unrelated donors was more prevalent among IL949


B.C. Betts et al. 2/SIR/TAC versus SIR/TAC recipients (P=0.02).6 Among the 20 treated patients, 10 fully completed IL-2 therapy between days 0 to +90 (+7 days). Six patients prematurely discontinued IL-2 injections due to: patient preference (n=3), VOD/SOS (n=2), and angioedema (n=1). A total of 8 patients developed grade II-IV acute GvHD on IL2/SIR/TAC. IL-2 injections were stopped for 4 of these patients as per protocol. Those with isolated stage 1 upper GI disease were allowed to continue IL-2. The mean exposure of IL-2 was 69 days (range 6-90 days) for protocoladherent patients, including those who stopped IL-2 for acute GvHD. For those who discontinued cytokine therapy for reasons other than GvHD, the mean exposure of IL2 was 34 days (range 10-62 days). All patients remained on SIR/TAC after stopping IL-2 therapy, with only one patient discontinuing SIR at day +252 due to relapsed malignancy.

IL-2/SIR/TAC toxicity Grade 3-5 unexpected or serious adverse events (AE) [according to Common Terminology Criteria for Adverse Events (CTCAE) v.4.03)] were captured up to day +130 or 30 days after the last dose of IL-2 (Table 2). A total of 3 patients experienced VOD/SOS and/or portal vein thrombosis, and the toxicities were attributed toward Table 1. Baseline characteristics of study sample (n=20).

IL-2/SIR/TAC Recipient age (median, range) Sex Male Female Diagnosis AML MDS CML ALL Disease Risk Index27 Low Intermediate High Very high HCT-CI28 0 1 2 3 4 Donor Matched sibling Matched unrelated Recipient:Donor sex Male:Male Male:Female Female:Male Female:Female Conditioning Flu/Mel Bu3500/Flu Bu5300/Flu

46 (22-70) 12 8 13 2 1 4 1 12 6 1 3 2 6 4 5 3 17 7 5 5 3 4 1 15

IL-2: interleukein-2; SIR: sirolimus; TAC: tacrolimus; AML: acute myeloid leukemia; MDS: myelodysplastic syndrome; CML: chronic myelogenous leukemia; ALL: acute lymphoblastic leukemia; HCT: hematopoietic cell transplantation; Flu: fludarabine; Mel: melphalan; Bu: busulfan.

950

IL-2/SIR/TAC as being probable (Table 2). One subject developed grade 2 arthralgia on day +84, coinciding with the last dose of planned injections. The 30% intolerance rate of low-dose IL-2 on this trial was unanticipated, as it is typically well tolerated in other post-transplant applications.11,29 No grade 4 or 5 AEs were attributed to IL2/SIR/TAC. No patients developed capillary leak syndrome or TMA while receiving IL-2/SIR/TAC. At a median follow up of 470 days (121-882 days), one patient died from a bowel obstruction related to a pre-HCT surgical site, 2 patients died from relapsed acute myeloid leukemia and 17 patients survive.

IL-2/SIR/TAC promotes an early wave of Treg expansion Among a total of 18 evaluable patient samples, IL2/SIR/TAC significantly increased Treg reconstitution at day +30 compared with published data from SIR/TAC alone and met the primary study end point6 (Figure 1A and B). IL-2/SIR/TAC Tregs obtained on day +30 were suppressive in vitro, albeit at a Treg to Tconv ratio of 1:1 (Figure 1C). The frequency of Tregs on IL-2/SIR/TAC decreased significantly by day +90, compared to day +30 values or SIR/TAC controls (Figure 1A and B). Marked contraction of Treg numbers was observed by day +180, and persisted one year after allogeneic HCT (Figure 1A and B). The sharp decline in Tregs was also observed among the subset of patients who successfully completed all planned doses of IL-2 (Figure 1D and E). Among CD4+ T cells, Tregs exhibit preferential STAT5 activation in response to IL-2 compared to Tconv.1,10,16 Therefore, we measured the amount of resting and IL-2-stimulated pSTAT5+ CD4+ T cells to determine whether the loss of Tregs was due to impaired IL-2 receptor signal transduction. The proportion of IL-2-stimulated pSTAT5+ CD4+ T cells was significantly increased at all tested time points compared to unstimulated controls, indicating IL2/SIR/TAC T cells maintained IL-2 responsiveness in vitro (Figure 1F). Conversely, the ratio of Treg to pSTAT5+ CD4+T cells decreased significantly from day +30 to day +90 even among the subset of patients actively receiving IL-2 (Figure 1G) and suggests reduced cytokine efficiency over time. While reduced Tregs among those no longer receiving IL-2 was not surprising, the severe decline in Tregs among IL-2-treated patients was unexpected.

Table 2. IL-2/SIR/TAC adverse events.

Toxicity VOD/SOS (n=2) Portal vein thrombosis (n=1) Sepsis (n=1) Atrial fibrillation (n=1) GI bleed (n=1) Acute kidney injury (n=3)

Kidney stone (n=1) Intracranial hemorrhage (n=1)

Grade •3 •3 •3 •3 •3 •3 •3 •3 •3 •3 •3

Attribution to IL-2/SIR/TAC* • Possibly • Possibly • Possibly • Unlikely (infection) • Unlikely (infection) • Unrelated (adenovirus) • Unrelated (vancomycin) • Unrelated (furosemide) • Unrelated (adenovirus) • Unrelated (metabolic) • Unrelated (trauma)

IL-2: interleukein-2; SIR: sirolimus; TAC: tacrolimus; VOD: veno-occlusive disease; SOS: sinusoidal obstruction syndrome; GI: gastrointestinal.*Attributed cause of adverse event is reported in brackets.

haematologica | 2017; 102(5)


IL-2 facilitates Treg reconstitution

B

A

D

H

C

E

G

F

I

J

Figure 1. Interleukin-2/sirolimus/tacrolimus (IL-2/SIR/TAC) promotes an early wave of Treg expansion, as well as conventional T cells (Tconv) contraction after allogeneic hematopoietic stem cell transplantation (HCT). (A and B) Kinetics of CD4+, CD25+, CD127– Tregs (median % and absolute #) from day +30 to +365 among IL-2/SIR/TAC (circle) and versus published data from SIR/TAC (triangle) alone (Mann-Whitney test). IL-2 was given from day 0 to day +90, thrice weekly. For IL2/SIR/TAC: n=18, 16, 12, and 6, at days +30, +90, +180, and +365, respectively. For SIR/TAC: n=36, 32, 21, and 17, at days +30, +90, +180, and +365, respectively. (C) Treg suppression of allo Tconv was verified among 3 independent IL-2/SIR/TAC patients at day +30. A representative experiment is shown. (D and E) Mean CD4+ Tregs (% and absolute #, ±SE) for those actively receiving IL-2 or off cytokine at days +30 and +90 (Mann-Whitney test). (F) Mean %phospho-STAT5+ CD4+ T cells [without (closed circle) or with (open circle) pulse of IL-2 for 15 minutes] at pre-transplant, day +30, and day +90 after allogeneic HCT (ANOVA). (G) Mean ratio of Tregs to IL-2-stimulated pSTAT5+ CD4+ T cells (±SE) for those actively receiving IL-2 or off cytokine at days +30 and +90 (Mann-Whitney test). (H and I) Kinetics of CD4+, CD25+, CD127+ activated Tconv (median % and absolute #) from day +30 to +365 among IL-2/SIR/TAC (circle) and versus published data from SIR/TAC (triangle) alone (Mann-Whitney test). (J) Median values for total CD4+ Tconv from day +30 to +365 among IL-2/SIR/TAC (circle) and versus SIR/TAC (triangle) (Mann-Whitney test). *P<0.05, **P=0.001-0.01, ***P=0.0001-0.001, ****P<0.0001. NS: not significant.

haematologica | 2017; 102(5)

951


B.C. Betts et al.

IL-2/SIR/TAC induces contraction of activated Tconv Interleukin-2 with SIR is known to reduce Tconv in murine allogeneic transplantation.14 IL-2/SIR/TAC led to a late but significant loss of activated CD4+ Tconv by day +180 (Figure 1H and I) identified by expression of the high affinity IL-2 receptor, CD25, and the IL-7 receptor, CD127, recognized markers of activation.22-24 Similar to the observed decline in Tregs, the loss of activated Tconv occurred among all patients treated with IL-2/SIR/TAC irrespective of cytokine treatment adherence (Online Supplementary Figure S2A and B). Distinct from observations made in rodent transplantation, IL-2/SIR/TAC was not entirely lymphodepleting and the total CD4+ T-cell count was preserved (Figure 1J).

The addition of prolonged IL-2 therapy with SIR/TAC does not further reduce acute or chronic GvHD The cumulative incidence of grade II-IV by day +100 was 40% [95% confidence interval (CI): 15.8-63.4] with IL-2/SIR/TAC (Figure 2A) and comparable to our published experience with SIR/TAC alone at 43% (95%CI: 27%-59%).6 The distribution and organ stage of acute GvHD for the 8 IL-2/SIR/TAC patients are detailed in Table 3. No participants died of acute GvHD and none developed steroid refractory disease. Of the 6 patients who discontinued IL-2 for reasons other than acute GvHD and received an abbreviated course of IL-2, only one developed grade II acute GvHD after stopping cytokine therapy (Figure 2B). This supports the view that lack of clinical efficacy was not negatively influenced by attrition of IL-2 treated patients. Conversely, this suggests an abbreviated course of IL-2 with SIR/TAC may be clinically beneficial. The cumulative incidence of all chronic GvHD and moderate-severe disease was respectively 61.5% (95%CI: 36.1-79.3) and 39.4% (95%CI: 11.2-67.2) at a median follow up of 470 days (Figure 2C). Compared to published data with SIR/TAC alone (moderate-severe chronic GvHD: 24% (95%CI: 7%-47%),6 the addition of IL-2 does not appear to reduce chronic GvHD. Receiving a full versus an abbreviated course of IL-2 therapy did not significantly impact the onset of chronic GvHD [59.4% (95%CI: 27.7-81.0) vs. 68.8% (95%CI: 24.5-90.6), P=0.5483] (Figure 2D). Two (10%) patients developed late acute GvHD, one in skin, and one in the GI tract (GI stage 1) on days +113 and +117.

CD25+ CD127+ Tconv with IL-2/SIR/TAC were similar to SIR/TAC alone at day +30. However, the amount of CD4+ CD25+ CD127+ Tconv at day +30 was significantly increased among IL-2/SIR/TAC patients who developed grade II-IV acute GvHD (Figure 3A and B). The amount of circulating Treg at day +30 was comparable among those with or without acute GvHD by day +100 (Figure 3C and D). As such, the Treg:Tconv ratio at day +30 was significantly higher among IL-2/SIR/TAC patients who never acquired acute GvHD (Figure 3E). Acute GvHD did not correlate with the number of activated Tconv or the Treg:Tconv ratio among the cohort of SIR/TAC controls (Online Supplementary Figure S2C and D).

STAT3 phosphorylation correlates with Tconv reconstitution and is not controlled by IL-2/SIR/TAC As expected, IL-2/SIR/TAC suppressed mTOR signal transduction30 as indicated by a decrease in downstream protein S6 phosphorylation among CD4+ T cells (Figure 4A). Work from our lab and others suggested that STAT5 polarization of donor T cells with IL-2 could reciprocally impair STAT3 phosphorylation which is implicated in acute GvHD pathogenesis.16,31 Conversely, IL-2/SIR/TAC failed to abolish IL-6 signal transduction in circulating CD4+ T cells (Figure 4B). Moreover, the frequency of pSTAT3+ CD4+ T cells at day +30 was significantly increased among IL-2/SIR/TAC patients who developed acute GvHD (Figure 4C). This confirms our earlier observation among patients who predominately received SIR/TAC prophylaxis, which showed that increased pSTAT3+ CD4+ T cells was significantly associated with acute GvHD risk.4 Therefore, adding IL-2 to SIR/TAC did not mitigate this liability as hypothesized. The incidence of grade II-IV acute GvHD was significantly greater among IL-2/SIR/TAC patients with increased pSTAT3+ CD4+ T cells (>48% pSTAT3+) at day +30 (Figure 4D), substantiating our previously determined cut-off point. Patients with elevated pSTAT3+ T cells also developed acute GvHD 20 days earlier than those with less (Figure 4D). Here we observed a strong positive correlation between the amount of pSTAT3+ CD4+ T cells and CD4+ CD25+ CD127+ Tconv reconstitution at day +30 (Figure 3E). The proportion of pS6+ or pSTAT5+ CD4+ T cells at day +30 was similar among those who did and those that did not develop acute GvHD (Figure 4F and G).

Relapse and survival At a median follow up of 470 days, the cumulative incidence of relapse was 35.2% (95%CI: 9.2-63.3) at one year (Figure 2E). Those on IL-2/SIR/TAC were permitted to receive either reduced intensity or myeloablative conditioning. The treated patients also represent a wide range of hematologic malignancies and risk by the Disease Risk Index.27 Of the relapse events, 86% had a pre-HCT Disease Risk Index of intermediate to high with only one low-risk patient. The 1-year overall survival was 77.1% (95%CI: 49.3-90.7) (Figure 2F), with a non-relapse mortality of 5.0% (95%CI: 0.0-65.6) (Figure 2G).

Activated CD25+ CD127+ Tconv are increased among IL-2/SIR/TAC recipients who develop grade II-IV acute GvHD Interleukin-2/SIR/TAC exerts significant suppression over activated Tconv late after allogeneic HCT. Conversely, the frequency and absolute numbers of CD4+ 952

Discussion This phase II trial demonstrates that low-dose IL-2 added to SIR/TAC significantly increases Treg reconstitution early after allogenic HCT compared to SIR/TAC alone6 (Day +30% Treg 23.8% vs. 16.0%, P=0.0016; #Treg 0.052 k/uL vs. 0.037 k/uL, P=0.0163). Replacing broadly suppressive MTX with SIR appears to be beneficial in this regard as the reported Treg frequency at day +30 with IL2/MTX/TAC was 11.1%.11 Despite early Treg reconstitution, the cumulative incidence of grade II-IV acute GvHD with IL-2/SIR/TAC was similar to our published data with SIR/TAC alone.6 Activated CD4+ Tconv from IL2/SIR/TAC patients were significantly increased at day +30 among those who developed acute GvHD. The proportion of Treg to activated Tconv at day +30 was also significantly higher among patients who remained free of acute GvHD. Adding IL-2 to SIR/TAC largely eliminated haematologica | 2017; 102(5)


IL-2 facilitates Treg reconstitution

the activated Tconv by day +90 compared to SIR/TAC, though this presumably occurred too late to be useful in controlling acute GvHD. Confirming our prior findings,4 the frequency of IL-6 activated pSTAT3+ CD4+ T cells at day +30 were signifi-

A

B

C

D

E

cantly increased among those who developed grade II-IV acute GvHD. Moreover, we identified a strong correlation between the amount of pSTAT3+ CD4+ T cells and activated Tconv early after HCT. We previously demonstrated that polarizing the ratio of pSTAT5 to pSTAT3 in CD4+ T

F

G Figure 2. The addition of prolonged IL-2 therapy with sirolimus/tacrolimus (SIR/TAC) does not further reduce acute or chronic graft-versus-host disease. (A) Cumulative incidence of acute graft-versus-host disease (GvHD) is shown for grade II-IV disease (solid) and severe grade III-IV acute GvHD (dashed line) by day +100 (P=0.1177). The incidence of grade II-IV acute GvHD with IL-2/SIR/TAC [40% (95%CI: 15.8%-63.4%)] is not significantly different from our published data with SIR/TAC [43% (95%CI: 27%-59%)] alone. The acute GvHD characteristics, including organ stage, are detailed in Table 2. (B) Cumulative incidence of grade II-IV acute GvHD among patients who received a full course of IL-2 versus those who prematurely stopped (abbreviated course) IL-2 [16.7% (95%CI: 0.001-77.7) vs. 50% (95%CI: 27.8-77.1), P=0.1475]. (C) Cumulative incidence of chronic GvHD (median follow up of 470 days) based on any chronic GvHD (solid black line) or moderate to severe disease (dashed line) (P=0.2698). (D) Cumulative incidence of chronic GvHD among patients who received a full course of IL-2 versus those who prematurely stopped (abbreviated course) IL-2 [59.4% (95%CI: 27.7-81.0) versus 68.8% (95%CI: 24.5-90.6), P=0.5483]. (E-G) Cumulative incidence of relapse, overall survival, and non-relapse mortality among those treated with IL-2/SIR/TAC.

haematologica | 2017; 102(5)

953


B.C. Betts et al. Table 3. Acute graft-versus-host disease characteristics.

cells favored Treg expansion and reciprocally impaired Tconv.16 In that work, a direct STAT3 inhibitor was used to shift STAT5 over STAT3 activity.16 Here we tested the alternative hypothesis that supplementing IL-2 could similarly promote STAT5 phosphorylation in donor T cells, with STAT3 otherwise unchecked. Though IL-2 increases the amount of pSTAT5+ CD4+ T cells and Tregs at day +30, IL-2 does not reciprocally prevent T-cell STAT3 phosphorylation. The early Treg expansion with IL-2/SIR/TAC was shortlived and significantly declined by day +90 despite ongoing cytokine therapy. Understanding that this trial was not powered for acute GvHD prevention, we postulate that the lack of clinical efficacy was greatly influenced by the profound loss of Tregs. In the setting of chronic GvHD treatment, Treg proliferation peaked after four weeks of IL-2 then waned while cytokine therapy continued.9,10,29 While Treg expansion in response to extended IL-2 therapy for chronic GvHD diminished over time, the amount of Tregs after two years of treatment was still greater than baseline, pre-IL-2 levels.29 Conversely, the contraction of Tregs early after allogeneic HCT with low-dose IL-2 and SIR/TAC observed during our study was unexpected and long lasting. It has recently been shown that soluble IL-2 receptor is released into the periphery when low-dose IL2 is administered for GvHD therapy.29 Moreover, the decline in Tregs while receiving IL-2 coincides with the

A

D

B

N (%) Skin stage 0 1 2 3 4 GI stage 0 1 2 3 4 Liver stage 0 1 2 3 4 Overall grade 0 I II III IV

14 (70%) 2 (10%) 2 (10%) 2 (10%) 0 13 (65%) 5 (25%) 0 2 (10%) 0 20 (100%) 0 0 0 0 10 (50%) 2 (10%) 6 (30%) 2 (10%) 0

GI: gastrointestinal.

C

E

Figure 3. CD4+ CD25+ CD127+ conventional T cells (Tconv) are increased among interleukin-2/sirolimus/tacrolimus (IL-2/SIR/TAC) recipients who develop grade II-IV acute graft-versus-host disease (GvHD). (A and B) Mean amount of CD4+ CD25+ CD127+ Tconv (% and absolute #, ±SE) at day +30 among those who did or did not develop grade II-IV acute GvHD by day +100 (Mann-Whitney test). (C and D) Mean amount of CD4+ CD25+ CD127– Treg (% and absolute #) at day +30 among those who did or did not develop grade II-IV acute GvHD by day +100 (Mann-Whitney test). (E) Mean ratio (±SE ) of regulatory T cell (Treg):Tconv at day +30 among those who did or did not develop grade II-IV acute GvHD by day +100 (Mann-Whitney test). *P<0.05, **P=0.001-0.01. NS: not significant.

954

haematologica | 2017; 102(5)


IL-2 facilitates Treg reconstitution

rise of soluble IL-2 receptor in the plasma of these patients.29 The binding affinity of soluble IL-2 receptor is approximately 10-fold less than its membrane-bound form,32 but it can significantly reduce the biological activity of IL-2.33 While we did not measure soluble IL-2 receptor levels during this trial, we hypothesize that IL-2 sequestration by soluble IL-2 receptor greatly contributed to progressive Treg loss. This is further supported by data showing soluble IL-2 receptor is spontaneously shed immediately after allogeneic HCT.34 Therefore, administering low-dose IL-2 early after transplant may intensify plasma concentrations of soluble IL-2 receptor and exacerbate Treg contraction. Therefore, the lack of quantified

A

B

haematologica | 2017; 102(5)

C

E

D

F

soluble IL-2 receptor levels during our clinical trial is an important limitation. Though we surmise that the loss of Tregs on IL2/SIR/TAC was mediated by soluble IL-2 receptor, alternative mechanisms were considered and investigated. We verified that T-cell IL-2 receptor signal transduction remained intact even while the Treg numbers declined. We also confirmed that IL-2 protects Tregs from apoptosis (Online Supplementary Figure S3A),10 and therefore activation-induced cell death is an unlikely explanation for the reduced Tregs. While anti-aldesleukin antibodies can occur among IL-2-treated patients, the incidence of neutralizing antibodies is rare.35,36 Moreover, the STAT5 phos-

G

Figure 4. Interleukin-2/sirolimus/tacrolimus (IL-2/SIR/TAC) suppresses mTOR signal transduction but not STAT3 phosphorylation in CD4+ T cells. (A and B) Mean %pS6+ (mTOR signal transduction) and %pSTAT3+ CD4+ T cells [without (closed circle) or with (open circle) pulse of IL-6 for 15 minutes)] at pretransplant, day +30, and day +90 after allogeneic hematopoietic stem cell transplantation (HCT) (ANOVA). (C) Mean amount of %pSTAT3+ CD4+ T cells on day +30 is significantly increased among patients who develop grade II-IV acute graft-versus-host disease (GvHD) by day +100 (Mann-Whitney test). (D) The previously identified cut-off point of 48% pSTAT3+ CD4+ T cells on day +30 significantly stratified acute GvHD incidence among those on IL-2/SIR/TAC (P<0.0001). The Gray method was used to evaluate the difference in incidence rates between the 2 groups. (E) Correlation between %pSTAT3+ CD4+ T cells versus %CD4+ CD25+ CD127+ conventional T cells. Log-Transformation was performed per the normality assumption test. Pearson correlation coefficient and P-value are shown. (F and G) Percentage of pS6+ and pSTAT5+ CD4+ T cells at day +30 was similar regardless of acute GvHD diagnosis by day +100. *P<0.05, **P=0.001-0.01, ***P=0.0001-0.001.

955


B.C. Betts et al.

phorylation experiments were performed in serum-free culture conditions in vitro and this approach mitigates any potential interference from neutralizing antibodies. The primary mechanism driving contraction of Tregs with IL2/SIR/TAC remains to be determined; we speculate that the reduced Treg responsiveness to IL-2 may be explained by peripheral cytokine sequestration via soluble receptor rather than impaired receptor signal transduction, antibody-mediated neutralization, or Treg apoptosis. While fleeting, it is noteworthy that IL-2/SIR/TAC provided early and robust Treg engraftment. Strategies to overcome inhibitory effects by soluble IL-2 receptor are feasible, and may facilitate durable Treg responses to IL2. In non-human primates, an IgG-IL-2 fusion protein demonstrated enhanced half-life and bioavailability compared to recombinant IL-2.37 The fusion protein also induced sustained Treg proliferation, 4-fold greater than recombinant IL-2.37 Chimeric IL-2/caspase-3 molecules also eliminate alloreactive T cells,38 a source of soluble IL2 receptor,39 and promotes Treg expansion in vivo.38 The duration and frequency of IL-2 dosing may be optimized to mitigate soluble IL-2 receptor shedding after allogeneic HCT.29 For example, the approach successfully demonstrated in rodents used a very brief course of IL-2 (essentially 3-7 days) with SIR.14 This restricted exposure of IL2 with SIR significantly expanded a durable population of Tregs and reduced GvHD mortality.14 Though limited by small numbers, IL-2/SIR/TAC patients who prema-

References 1. Zeiser R, Leveson-Gower DB, Zambricki EA, et al. Differential impact of mammalian target of rapamycin inhibition on CD4+CD25+Foxp3+ regulatory T cells compared with conventional CD4+ T cells. Blood. 2008;111(1):453-462. 2. Seddiki N, Santner-Nanan B, Martinson J, et al. Expression of interleukin (IL)-2 and IL-7 receptors discriminates between human regulatory and activated T cells. J Exp Med. 2006;203(7):1693-1700. 3. Zorn E, Nelson EA, Mohseni M, et al. IL-2 regulates FOXP3 expression in human CD4(+)CD25(+) regulatory T cells through a STAT-dependent mechanism and induces the expansion of these cells in vivo. Blood. 2006;108(5):1571-1579. 4. Betts BC, Sagatys EM, Veerapathran A, et al. CD4+ T cell STAT3 phosphorylation precedes acute GVHD, and subsequent Th17 tissue invasion correlates with GVHD severity and therapeutic response. J Leukoc Biol. 2015;97(4):807-819. 5. Alho AC, Kim HT, Chammas MJ, et al. Unbalanced recovery of regulatory and effector T cells after allogeneic stem cell transplantation contributes to chronic GVHD. Blood. 2016;127(5):646-657. 6. Pidala J, Kim J, Jim H, et al. A randomized phase II study to evaluate tacrolimus in combination with sirolimus or methotrexate after allogeneic hematopoietic cell transplantation. Haematologica. 2012;97(12):18821889. 7. Cutler C, Logan B, Nakamura R, et al.

956

8.

9.

10.

11.

12.

13.

turely discontinued IL-2 before day +90 (n=6) actually showed a trend toward less acute GvHD compared to protocol-adherent individuals (n=14) [16.7% (95%CI: 0.001-77.7) vs. 50% (95%CI: 27.8-77.1), P=0.1475]. We chose to use intermittent injections of low-dose IL-2 based on the favorable pediatric experience in GvHD prophylaxis.11 Besides the concern for IL-2 sequestration, Treg longevity in adult allogeneic HCT recipients may conversely benefit from more consistent IL-2 dosing strategies, such as daily administration or continuous infusion of the cytokine. Our data support the view that IL-2/SIR/TAC can transiently increase IL-2 signal transduction and early Treg reconstitution after allogeneic HCT. However, IL2/SIR/TAC does not mitigate STAT3-mediated GvHD and its impact on Treg expansion is not durable. We propose that strategies to optimize IL-2 dosing and allay cytokine neutralization by soluble IL-2 receptor may facilitate lasting Treg recovery. However, such an approach must still consider the impact of unconstrained STAT3 activity even when IL-2 is replenished post transplant. Funding This work was supported by the Miles for Moffitt Milestone Award (BCB).We thank Prometheus Labs for providing aldesleukin. The Moffitt Cancer Center Flow Cytometry and Biostatistics Cores (P30-CA076292) contributed to the completion of this trial.

Tacrolimus/sirolimus vs. tacrolimus/ methotrexate as GVHD prophylaxis after matched, related donor allogeneic HCT. Blood. 2014;124(8):1372-1377. Zorn E, Mohseni M, Kim H, et al. Combined CD4+ donor lymphocyte infusion and lowdose recombinant IL-2 expand FOXP3+ regulatory T cells following allogeneic hematopoietic stem cell transplantation. Biol Blood Marrow Transplant. 2009;15(3):382388. Koreth J, Matsuoka K, Kim HT, et al. Interleukin-2 and regulatory T cells in graftversus-host disease. N Engl J Med. 2011;365(22):2055-2066. Matsuoka K, Koreth J, Kim HT, et al. Lowdose interleukin-2 therapy restores regulatory T cell homeostasis in patients with chronic graft-versus-host disease. Sci Transl Med. 2013;5(179):179ra143. Kennedy-Nasser AA, Ku S, Castillo-Caro P, et al. Ultra low-dose IL-2 for GVHD prophylaxis after allogeneic hematopoietic stem cell transplantation mediates expansion of regulatory T cells without diminishing antiviral and antileukemic activity. Clin Cancer Res. 2014;20(8):2215-2225. Scotta C, Fanelli G, Hoong SJ, et al. Impact of immunosuppressive drugs on the therapeutic efficacy of ex vivo expanded human regulatory T cells. Haematologica. 2016; 101(1):91-100. Lim DG, Koo SK, Park YH, et al. Impact of immunosuppressants on the therapeutic efficacy of in vitro-expanded CD4+CD25+Foxp3+ regulatory T cells in allotransplantation. Transplantation. 2010; 89(8):928-936.

14. Shin HJ, Baker J, Leveson-Gower DB, Smith AT, Sega EI, Negrin RS. Rapamycin and IL-2 reduce lethal acute graft-versus-host disease associated with increased expansion of donor type CD4+CD25+Foxp3+ regulatory T cells. Blood. 2011;118(8):2342-2350. 15. Johnston L, Florek M, Armstrong R, et al. Sirolimus and mycophenolate mofetil as GVHD prophylaxis in myeloablative, matched-related donor hematopoietic cell transplantation. Bone Marrow Transplant. 2012;47(4):581-588. 16. Betts BC, Veerapathran A, Pidala J, Yu XZ, Anasetti C. STAT5 polarization promotes iTregs and suppresses human T-cell alloresponses while preserving CTL capacity. J Leukoc Biol. 2014;95(2):205-213. 17. Ho VT, Cutler C, Carter S, et al. Blood and marrow transplant clinical trials network toxicity committee consensus summary: thrombotic microangiopathy after hematopoietic stem cell transplantation. Biol Blood Marrow Transplant. 2005;11(8):571575. 18. McDonald GB, Hinds MS, Fisher LD, et al. Veno-occlusive disease of the liver and multiorgan failure after bone marrow transplantation: a cohort study of 355 patients. Ann Intern Med. 1993;118(4):255-267. 19. Przepiorka D, Weisdorf D, Martin P, et al. 1994 Consensus Conference on Acute GVHD Grading. Bone Marrow Transplant. 1995;15(6):825-828. 20. Pavletic SZ, Lee SJ, Socie G, Vogelsang G. Chronic graft-versus-host disease: implications of the National Institutes of Health consensus development project on criteria for clinical trials. Bone Marrow Transplant.

haematologica | 2017; 102(5)


IL-2 facilitates Treg reconstitution

2006;38(10):645-651. 21. Liu W, Putnam AL, Xu-Yu Z, et al. CD127 expression inversely correlates with FoxP3 and suppressive function of human CD4+ T reg cells. J Exp Med. 2006;203(7):1701-1711. 22. Heninger AK, Theil A, Wilhelm C, et al. IL-7 abrogates suppressive activity of human CD4+CD25+FOXP3+ regulatory T cells and allows expansion of alloreactive and autoreactive T cells. J Immunol. 2012;189(12): 5649-5658. 23. Touil S, Rosenzwajg M, Landau DA, et al. Depletion of T regulatory cells through selection of CD127-positive cells results in a population enriched in memory T cells: implications for anti-tumor cell therapy. Haematologica. 2012;97(11):1678-1685. 24. Samarasinghe S, Mancao C, Pule M, et al. Functional characterization of alloreactive T cells identifies CD25 and CD71 as optimal targets for a clinically applicable allodepletion strategy. Blood. 2010;115(2):396-407. 25. Krutzik PO, Hale MB, Nolan GP. Characterization of the murine immunological signaling network with phosphospecific flow cytometry. J Immunol. 2005;175(4): 2366-2373. 26. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(446):496-509. 27. Armand P, Kim HT, Logan BR, et al.

haematologica | 2017; 102(5)

28.

29.

30.

31.

32.

33.

Validation and refinement of the Disease Risk Index for allogeneic stem cell transplantation. Blood. 2014;123(23):3664-3671. Sorror M, Maris M, Baron F, et al. A modified hematopoietic cell transplantation (HCT)-specific-comorbidity index. Blood. 2004;104(11):324a-325a. Koreth J, Kim HT, Jones KT, et al. Efficacy, durability, and response predictors of lowdose interleukin-2 therapy for chronic graftversus-host disease. Blood. 2016;128(1):130137. Hara K, Yonezawa K, Weng QP, Kozlowski MT, Belham C, Avruch J. Amino acid sufficiency and mTOR regulate p70 S6 kinase and eIF-4E BP1 through a common effector mechanism. J Biol Chem. 1998;273(23): 14484-14494. Yang XP, Ghoreschi K, Steward-Tharp SM, et al. Opposing regulation of the locus encoding IL-17 through direct, reciprocal actions of STAT3 and STAT5. Nat Immunol. 2011;12(3):247-254. Baran D, Korner M, Theze J. Characterization of the soluble murine IL2R and estimation of its affinity for IL-2. J Immunol. 1988;141(2):539-546. Jacques Y, Le Mauff B, Boeffard F, Godard A, Soulillou JP. A soluble interleukin 2 receptor produced by a normal alloreactive human T cell clone binds interleukin 2

34.

35.

36.

37.

38.

39.

with low affinity. J Immunol. 1987;139(7): 2308-2316. Foley R, Couban S, Walker I, et al. Monitoring soluble interleukin-2 receptor levels in related and unrelated donor allogenic bone marrow transplantation. Bone Marrow Transplant. 1998;21(8):769-773. Allegretta M, Atkins MB, Dempsey RA, et al. The development of anti-interleukin-2 antibodies in patients treated with recombinant human interleukin-2 (IL-2). J Clin Immunol. 1986;6(6):481-490. Granelli-Piperno A, Andrus L, Reich E. Antibodies to interleukin 2. Effects on immune responses in vitro and in vivo. J Exp Med. 1984;160(3):738-750. Bell CJ, Sun Y, Nowak UM, et al. Sustained in vivo signaling by long-lived IL-2 induces prolonged increases of regulatory T cells. J Autoimmun. 2015;5666-5680. Yarkoni S, Prigozhina TB, Slavin S, Askenasy N. IL-2-targeted therapy ameliorates the severity of graft-versus-host disease: ex vivo selective depletion of hostreactive T cells and in vivo therapy. Biol Blood Marrow Transplant. 2012;18(4):523535. Rubin LA, Kurman CC, Fritz ME, et al. Soluble interleukin 2 receptors are released from activated human lymphoid cells in vitro. J Immunol. 1985;135(5):3172-3177.

957


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Cell Therapy & Immunotherapy

Ferrata Storti Foundation

Haematologica 2017 Volume 102(5):958-966

Improved survival after acute graft-versus-host disease diagnosis in the modern era

Hanna J. Khoury,1 Tao Wang,2,3 Michael T. Hemmer,2 Daniel Couriel,4 Amin Alousi,5 Corey Cutler,6 Mahmoud Aljurf,7 Joseph H. Antin,6 Mouhab Ayas,7 Minoo Battiwalla8, Jean-Yves Cahn,9 Mitchell Cairo,10 Yi-Bin Chen,11 Robert Peter Gale,12 Shahrukh Hashmi,13,14 Robert J. Hayashi,15 Madan Jagasia,16 Mark Juckett,17 Rammurti T. Kamble,18 Mohamed Kharfan-Dabaja,19 Mark Litzow,13 Navneet Majhail,20 Alan Miller,18 Taiga Nishihori,19 Muna Qayed,21 Helene Schoemans,22 Harry C. Schouten,23 Gerard Socie,24 Jan Storek,25 Leo Verdonck,26 Ravi Vij,27 William A. Wood,28 Lolie Yu,29 Rodrigo Martino,30 Matthew Carabasi,31 Christopher Dandoy,32 Usama Gergis,33 Peiman Hematti,17 Melham Solh,34 Kareem Jamani,25 Leslie Lehmann,35 Bipin Savani,16 Kirk R. Schultz,36 Baldeep M. Wirk,37 Stephen Spellman,38 Mukta Arora39 and Joseph Pidala19

Department of Hematology and Medical Oncology, Winship Cancer Institute of Emory University, Atlanta, GA, USA; 2CIBMTR (Center for International Blood and Marrow Transplant Research), Department of Medicine, Medical College of Wisconsin, Milwaukee, WI, USA; 3Division of Biostatistics, Institute for Health and Society, Medical College of Wisconsin, Milwaukee, WI, USA; 4Utah Blood and Marrow Transplant ProgramAdults, Salt Lake City, UT, USA; 5Department of Stem Cell Transplantation, Division of Cancer Medicine, The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 6Center for Hematologic Oncology, Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, USA; 7Department of Pediatric Hematology Oncology, King Faisal Specialist Hospital Center & Research, Riyadh, Saudi Arabia; 8Hematology Branch, National Heart, Lung and Blood Institute-NIH, Bethesda, MD, USA; 9Department of Hematology, University Hospital, Grenoble, France; 10Division of Pediatric Hematology, Oncology and Stem Cell Transplantation, Department of Pediatrics, New York Medical College, Valhalla, NY, USA; 11Division of Hematology/Oncology, Massachusetts General Hospital, Boston, MA, USA; 12Hematology Research Centre, Division of Experimental Medicine, Department of Medicine, Imperial College London, UK; 13Department of Internal Medicine, Mayo Clinic Rochester, MN, USA; 14Department of Oncology, King Faisal Specialist Hospital Center & Research, Riyadh, Saudi Arabia; 15Division of Pediatric Hematology/Oncology, Department of Pediatrics, Washington University School of Medicine in St. Louis, MO, USA; 16Division of Hematology/Oncology, Vanderbilt University Medical Center, Nashville, TN, USA; 17Division of Hematology/Oncology/Bone Marrow Transplantation, Department of Medicine, University of Wisconsin Hospital and Clinics, Madison, WI, USA; 18Division of Hematology and Oncology, Center for Cell and Gene Therapy, Baylor College of Medicine, Houston, TX, USA; 19Department of Blood and Marrow Transplantation, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, USA; 20Blood & Marrow Transplant Program, Cleveland Clinic Taussig Cancer Institute, Cleveland, OH, USA; 21Department of Pediatrics, Emory University School of Medicine, Atlanta, GA, USA; 22University Hospital of Leuven, Belgium; 23Department of Hematology, Academische Ziekenhuis, Maastricht, the Netherlands; 24Department of Hematology, Hopital Saint Louis, Paris, France; 25Department of Medicine, University of Calgary, AB, Canada; 26Isala Clinics Zwolle, the Netherlands; 27Division of Hematology and Oncology, Washington University School of Medicine, St. Louis, MO, USA; 28Division of Hematology/Oncology, Department of Medicine, University of North Carolina, Chapel Hill, NC, USA; 29Division of Hematology/Oncology & HSCT, The Center for Cancer and Blood Disorders, Children’s Hospital/Louisiana State University Medical Center, New Orleans, LA, USA; 30Division of Clinical Hematology, Hospital de la Santa Creu I Sant Pau, Barcelona, Spain; 31Thomas Jefferson University Hospital, Philadelphia, PA, USA; 32 Cincinnati Children’s Hospital Medical Center, OH, USA; 33Hematologic Malignancies & Bone Marrow Transplant, Department of Medical Oncology, New York Presbyterian Hospital/Weill Cornell Medical Center, NY, USA; 34The Blood and Marrow Transplant Group of Georgia, Northside Hospital, Atlanta, GA, USA; 35Dana Farber Cancer Institute/ Boston Children’s Hospital, MA, USA; 36Department of Pediatric Hematology, Oncology and Bone Marrow Transplant, British Columbia’s Children’s Hospital, The University of British Columbia, Vancouver, BC, Canada; 37Division of Bone Marrow Transplant, Seattle Cancer Care Alliance, WA, USA; 38CIBMTR (Center for International Blood and Marrow Transplant Research), National Marrow Donor Program/Be the Match, Minneapolis, MN, USA and 39Division of Hematology, Oncology, and Transplantation, Department of Medicine, University of Minnesota Medical Center, Minneapolis, MN, USA 1

Correspondence: joseph.pidala@moffitt.org

Received: September 16, 2016. Accepted: March 7, 2017. Pre-published: March 16, 2017. doi:10.3324/haematol.2016.156356 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/5/958 ©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.

958

ABSTRACT

A

cute graft-versus-host disease remains a major threat to a successful outcome after allogeneic hematopoietic cell transplantation. While improvements in treatment and supportive care have occurred, it is unknown whether these advances have resulted in haematologica | 2017; 102(5)


Outcomes of acute GvHD over time

improved outcome specifically among those diagnosed with acute graft-versus-host disease. We examined outcome following diagnosis of grade II-IV acute graft-versus-host disease according to time period, and explored effects according to original graft-versus-host disease prophylaxis regimen and maximum overall grade of acute graft-versus-host disease. Between 1999 and 2012, 2,905 patients with acute myeloid leukemia (56%), acute lymphoblastic leukemia (30%) or myelodysplastic syndromes (14%) received a sibling (24%) or unrelated donor (76%) blood (66%) or marrow (34%) transplant and developed grade II-IV acute graft-versus-host disease (n=497 for 1999-2001, n=962 for 2002-2005, n=1,446 for 2006-2010). The median (range) follow-up was 144 (4-174), 97 (4-147) and 60 (8-99) months for 19992001, 2002-2005, and 2006-2010, respectively. Among the cohort with grade II-IV acute graft-versus-host disease, there was a decrease in the proportion of grade III-IV disease over time with 56%, 47%, and 37% for 1999-2001, 2002-2005, and 2006-2012, respectively (P<0.001). Considering the total study population, univariate analysis demonstrated significant improvements in overall survival and treatmentrelated mortality over time, and deaths from organ failure and infection declined. On multivariate analysis, significant improvements in overall survival (P=0.003) and treatment-related mortality (P=0.008) were only noted among those originally treated with tacrolimus-based graft-versus-host disease prophylaxis, and these effects were most apparent among those with overall grade II acute graft-versus-host disease. In conclusion, survival has improved over time for tacrolimus-treated transplant recipients with acute graft-versus-host disease.

Introduction

Methods

Graft-versus-host disease (GvHD) remains one of the most significant barriers to the success of allogeneic hematopoietic stem cell transplantation (HCT). Current pharmacological approaches fail to prevent acute GvHD completely.1-3 Furthermore, durable resolution of acute GvHD following primary steroid-based therapy is infrequent,4-6 and long-term success of systemic immunosuppressive therapies given after steroid therapy is elusive.7 Consequently, acute GvHD and its allied infectious complications and organ failure contribute greatly to early mortality after HCT. However, advances in supportive care over time, in particular improvements in the coverage and potency of antimicrobial agents (e.g. antibacterial, antiviral, and antifungal drugs), hold promise to improve survival for those patients affected by acute GvHD. Previous large analyses have demonstrated improvement in survival outcome over time for HCT recipients, despite evolution in overall practices including increasing HCT recipient age, use of unrelated donors, and peripheral blood stem cells as the predominant graft type.8 More recent transplants have been shown to have improved overall mortality, as well as reducing grade III-IV acute GvHD, major organ injury, and life-threatening infections through the early post-HCT period.9 Further insight is needed, however, as studies have not examined change in survival outcome over time exclusively in an acute GvHD-affected population, survival outcome according to maximal acute GvHD grade, or effects according to specific GvHD prophylaxis. Additionally, the studies have not examined outcomes of HCT procedures conducted after 2007. Accordingly, we conducted a large registry analysis to determine whether survival outcome after a diagnosis of acute GvHD has improved significantly over time.

Data source

haematologica | 2017; 102(5)

The Center for International Blood and Marrow Transplant Research (CIBMTR) is a voluntary working group of more than 450 transplant centers worldwide that contribute detailed data on consecutive allogeneic HCT to the Statistical Center at the Medical College of Wisconsin in Milwaukee or the National Marrow Donor Program (NMDP) Coordinating Center in Minneapolis. Approximately two-thirds of all active transplantation centers worldwide report data to the registry. The registry database includes information on 40-45% of all patients who have received an allotransplant since 1970, with annual updates. Compliance is assessed by periodic audits and accuracy of data is ensured by computerized record checks, physician review of submitted data and on-site audits. Observational studies conducted by the CIBMTR are done with a waiver of informed consent and in compliance with Health Insurance Portability and Accountability Act regulations as determined by the Institutional Review Board and Privacy Officer of the Medical College of Wisconsin.

Selection of patients and definitions The CIBMTR population of patients consisted of first postmyeloablative sibling or unrelated donor, blood or marrow allogeneic HCT recipients with acute myeloid leukemia, acute lymphoblastic leukemia or myelodysplastic syndromes transplanted between 1999 and 2012 who developed grade II-IV acute GvHD within 100 days after HCT. Only those with acute GvHD onset within 100 days after HCT were included, as this analysis is focused on classic acute GvHD, not late acute GvHD. The maximal severity of acute GvHD for each subject was used in all analyses.10 A total of 2,905 eligible cases with complete research data available were identified. This final study population was generated from the total (n=17,244) number of first US allogeneic HCT procedures for acute myeloid leukemia, acute lymphoblastic leukemia and myelodysplastic syndromes from 1999959


H.J. Khoury et al.

2012 after exclusion of umbilical cord blood transplants (n=2,748), donor types other than matched sibling or unrelated donors (n=654), in-vivo or ex-vivo T-cell-depleted grafts (n=4,149), reduced intensity or non-myeloablative conditioning regimens (n=2,133), those with grade 0-1 acute GvHD or missing GvHD information (n=4,002), those surviving patients with <100 days follow up (n=18), relapse prior to acute GvHD diagnosis (n=176), development of acute GvHD more than 100 days after HCT (n=88) or within 7 days of HCT (n=113), or other consent or research form related exclusions (n=258). Overall, molecular typing was available for 1,810 cases (83%) with >99.9% concordance between the 8/8 and well-matched and ≤7/8 and partially or mismatched groups.

Study endpoints The primary endpoint was overall survival (OS), defined as death from any cause. Secondary endpoints considered were the following: disease-free survival (DFS), defined as survival in continuous complete remission; transplant-related mortality (TRM), defined as death in continued remission; malignancy relapse, defined as recurrence of the malignancy for which HCT was performed; and chronic GvHD.11 All endpoints were estimated from the date of acute GvHD onset, as the exact date of maximal grade II-IV acute GvHD was not collected.

Statistical analysis Included patients were divided into three cohorts based on the year of transplant (1999-2001, 2002-2005, 2006-2012). Univariate analysis compared the outcomes of OS, DFS, TRM, relapse and chronic GvHD incidence between these three cohorts. Multivariate analyses were performed using Cox proportional hazards regression models. All the clinical variables were tested for the affirmation of the proportional hazards assumption. Factors violating the proportional hazards assumption were adjusted through stratification. A stepwise model selection procedure was then used to identify clinical variables that were associated with each particular outcome with a threshold of 0.05 for both entry and stay in the model. Interactions between the main variable ‘Year of transplant’ and the adjusted covariates were tested, and no significant interactions between the ‘Year of transplant’ and the adjusted covariates were detected at the significance level of 0.01 in any of the models. A significant center effect was detected for the outcomes of OS and TRM for patients with grade II-IV acute GvHD only among those treated with cyclosporine A, and adjustment for this effect was performed. Potential covariates included patient’s age (by decades), sex and race (Caucasian, African American, Other), Karnofsky performance status (<90%, ≥90%), time from diagnosis to HCT (<2 weeks, 2 weeks – 1 month, 2

Table 1. Transplantation (patient, donor, and transplant) and graft-versus-host disease characteristics.

Age, median (range), years Male sex, n (%) Race Caucasian, n (%) Karnofsky score > 90% at transplant, n (%) Diagnosis, n (%) Acute myeloid leukemia Acute lymphoblastic leukemia Myelodysplastic syndromes Disease status at transplant*, n (%) Early Intermediate Advanced Unknown HLA-identical sibling donor age, years, median (range) Unrelated donor age, years, median (range) Donor/recipient sex match, n (%) Female/male Other Donor/recipient CMV status, n (%) +/+ +/-/+ -/Missing Donor/recipient HLA match**, n (%) HLA-identical sibling URD well-matched URD partially matched URD mismatched URD missing Time from diagnosis to transplant, months, median (range)

1999-2001 (n=497)

Year of transplant 2002-2005 (n=962)

2006-2012 (n=1446)

P-Value

34 (1 - 63) 284 (57) 429 (86) 320 (64)

37 (<1 - 67) 583 (61) 799 (83) 609 (63)

43 (<1 - 70) 805 (56) 1174 (81) 915 (63)

<0.001 0.06 0.05 <0.001 <0.001

257 (52) 167 (34) 73 (15)

502 (52) 343 (36) 117 (12)

879 (61) 364 (25) 203 (14)

166 (33) 158 (32) 173 (35) 0 40 (<1 - 67) 35 (19 - 59)

417 (43) 272 (28) 267 (28) 6 (<1) 43 (1 - 71) 35 (18 - 59)

766 (53) 298 (21) 370 (26) 12 (<1) 47 (2 - 75) 33 (18 - 61)

103 (21) 393 (79)

209 (22) 753 (78)

285 (20) 1160 (80)

91 (18) 61 (12) 129 (26) 194 (39) 22 (4 )

232 (24) 99 (10) 310 (32) 296 (31) 25 (3 )

386 (27) 177 (12) 435 (30) 415 (29) 33 (2 )

107 (22) 171 (34) 149 (30) 70 (14) 0 9 (<1 - 238)

166 (17) 497 (52) 221 (23) 77 (8) 1 (<1) 7 (<1 - 197)

424 (29) 764 (53) 207 (14) 25 (2) 26 (2 ) 6 (<1 - 279)

<0.001

0.001 <0.001 0.50

<0.001

<0.001

<0.001 continued on the next page

960

haematologica | 2017; 102(5)


Outcomes of acute GvHD over time

months – 100 days), disease (acute myeloid leukemia, acute lymphoblastic leukemia, myelodysplastic syndromes), disease status at time of HCT (early, intermediate, advanced), donor type and age (HLA-identical sibling, unrelated well-matched 18-32 years, unrelated well-matched 33-49 years, unrelated wellmatched 50+ years, unrelated partially or mismatched 18-32 years, unrelated partially or mismatched 33-49 years, unrelated partially or mismatched 50+ years),12 donor-recipient sex match, donor-recipient cytomegalovirus serological status, use of total body irradiation (TBI) in the regimen, graft source (bone marrow, peripheral blood mobilized stem cells), and GvHD prophylaxis (cyclosporine ± others, and tacrolimus ± others). The product limit estimator proposed by Kaplan Meier was used to estimate the median and range of the follow-up time. The probabilities of OS and DFS for all patients were calculated using the Kaplan Meier estimator, with the variance estimated by the Greenwood formula. Cumulative incidence estimates were calculated for other endpoints to account for competing risks. Non-relapse death was a competing risk in the estimation of malignancy relapse, death without chronic GvHD was a competing risk for estimation of chronic GvHD, and relapse was a competing risk for estimation of TRM. SAS version 9.4 (SAS Institute, Cary, NC, USA) was used for all the analyses.

Results Patients Between 1999 and 2012, 2,905 un-manipulated blood (66%) or marrow (34%) sibling (24%) or unrelated donor (76%) allograft recipients with acute myeloid leukemia (56%), acute lymphoblastic leukemia (30%) or myelodysplastic syndromes (14%) developed grade II-IV acute GvHD. Conditioning was myeloablative in all cases. Among the 1,759 patients who received TBI-based conditioning, 1326 (75%) received cyclophosphamide/TBI, 210 (12%) received cyclophosphamide/TBI/others, and 153 (9%) received TBI/etoposide. Among the 1,146 patients who did not receive TBI as a part of their conditioning regimen, 691 (60%) received busulfan/cyclophosphamide, 114 (10%) received busulfan/cyclophosphamide/others, and 295 (26%) received busulfan/fludarabine as their regimen. Donor, recipient, and transplant characteristics are summarized in Table 1. The median follow-up for surviving patients was 144 (4-174), 97 (4-147) and 60 (8-99) months for those transplanted in 1999-2001, 2002-2005, and 2006-2012, respectively. Of the 1,077 patients who received cyclosporine-based

continued from the previous page

TBI-based conditioning regimen, n (%) Marrow graft, n (%) Blood graft, n (%) GvHD prophylaxis, n (%) Cyclosporine-based Tacrolimus-based Acute GvHD grade II III IV Acute GvHD-effected organs Skin + Gut + Liver Skin + Gut Skin + Liver Gut + Liver Skin only Gut only Liver only Acute GvHD treatment, n (%) Steroids + ATG + others Steroids + MAB + others Steroids only Missing Time from transplant to acute GvHD grade II-IV onset, days, median (range) < 2 weeks, n (%) 2-4 weeks, n (%) 1-2 months, n (%) 2 months – 100 days, n (%) Follow-up of survivors, months, median (range)

371 (75) 313 (63) 184 (37)

640 (67) 346 (36) 616 (64)

748 (52) 329 (23) 1117 (77)

359 (72) 132 (27)

473 (49) 472 (49)

245 (17) 1163 (80)

219 (44) 190 (38) 88 (18)

507 (53) 310 (32) 145 (15)

904 (63) 368 (25) 174 (12)

133 (27) 125 (25) 70 (14) 21 (4 ) 85 (17) 56 (11) 7 (1 )

189 (20) 289 (30) 72 (7) 42 (4 ) 201 (21) 153 (16) 16 (2 )

193 (13) 562 (39) 63 (4 ) 42 (3 ) 224 (15) 345 (24) 17 (1 )

47 (9 ) 51 (10) 380 (76) 19 (4 ) 23 (8-94)

53 (6) 133 (14) 739 (77) 37(4 ) 24 (8-100)

43 (3 ) 159 (11) 1172 (81) 72 (5 ) 26 (7-100)

123 (25) 221 (44) 128 (26) 25 (5 ) 144 (4-174)

238 (25) 416 (43) 255 (27) 53 (6 ) 97 (4-147)

259 (18) 647 (45) 407 (28) 133 (9 ) 60 (8-99)

<0.001 <0.001 <0.001

<0.001

<0.001

<0.001

<0.001

CMV: cytomegalovirus; HLA: human leukocyte antigen; URD: unrelated; TBI: total body irradiation; ATG: antithymocyte globulin; MAB: monoclonal antibodies. *Disease status is categorized as follows: Early: acute myeloid leukemia/acute lymphoblastic leukemia (first complete remission); myelodysplastic syndromes (refractory anemia, refractory anemia with ringed sideroblasts/pre-HCT marrow blasts <5%); Intermediate: acute myeloid leukemia/acute lymphoblastic leukemia (second complete remission or beyond); Advanced: acute myeloid leukemia/acute lymphoblastic leukemia (relapse/primary induction failure) myelodysplastic syndromes (refractory anemia with excess blasts, refractory anemia with excess blasts in transformation, chronic myelomonocytic leukemia or marrow blasts ≥5%). **Donor-recipient matching definitions as previously defined by Weisdorf, et al. 2008.12

haematologica | 2017; 102(5)

961


H.J. Khoury et al.

GvHD prophylaxis, 922 (86%) also received methotrexate while 56 (5%) received mycophenolate mofetil (MMF); 28 of these received both methotrexate and mycophenolate mofetil. The remainder mostly received cyclosporine alone Âą steroids (n=87, or 8%). Of the 1,767 patients who received tacrolimus-based GvHD prophylaxis, 1,376 (78%) also received methotrexate while 229 (13%) received mycophenolate mofetil; 63 of these received both methotrexate and mycophenolate mofetil. The remainder mostly received tacrolimus with sirolimus (n=99, or 6%). Other infrequent approaches were seen in both groups, in the order of <1-2% of the total each. Trends in GvHD prophylaxis, in this patient population that developed acute GvHD, favored predominant use of tacrolimus-based approaches in the later time period. When compared to cyclosporine-based GvHD prophylaxis, recipients of tacrolimus-based GvHD prophylaxis were comparable, but were more likely to have had a well-matched unrelated donor across all time periods (P<0.001), and to have received peripheral blood stem cells in recent years (P<0.001) (data not shown).

Characteristics of the acute graft-versus-host disease Overall, the proportion of grade III-IV acute GvHD decreased over time, being 56%, 47%, and 37% for the

periods 1999-2001, 2002-2005, and 2006-2012, respectively (P<0.001). Time of onset of acute GvHD remained unchanged over time. The proportion of patients with skin, gut and liver (concurrent 3-organ involvement) acute GvHD decreased (P=0.0014), while gut acute GvHD with or without skin involvement increased with time (P<0.0001 and P<0.0001, respectively). The majority of patients in all time cohorts were treated with steroids only as primary acute GvHD therapy. Approximately 1520% received antithymocyte globulin or other monoclonal antibodies in addition to corticosteroids (Table 1).

Outcomes Univariate analyses summarized in Table 2 showed improvements in OS as well as a reduction in TRM at 100 days, 6 months, and 1, 2, and 3 years for patients who developed grade II or grade III-IV acute GvHD. Proportionally, relapse of the primary disease, GvHD/infection and organ failure were the causes of the majority of deaths overall; however, trends over time supported reductions in idiopathic pneumonia, organ failure, hemorrhage, and infectious mortality (Table 3). No significant differences in relapse or chronic GvHD were observed across time period cohorts for those with overall maximal grade II acute GvHD. Alongside a marked

Table 2. Univariate probabilities of transplant outcomes, among allogeneic transplant recipients who developed acute graft-versus-host disease grade II or grade III-IV.

Outcomes

N. evaluated

Acute GvHD grade II Overall survival 100-day 6-month 1-year 2-year 3-year Relapse 100-day 6-month 1-year 2-year 3-year Transplant-related mortality 100-day 6-month 1-year 2-year 3-year Disease-free survival 100-day 6-month 1-year 2-year 3-year Chronic GvHD 100-day 6-month 1-year 2-year 3-year

1999-2001 Probability (95%CI)

219

2002-2005 N. evaluated Probability (95%CI) 507

85 (80-90)% 74 (68-79)% 58 (51-65)% 46 (39-52)% 41 (34-47)% 214

502

875

502

214

875

502

4 (3-6)% 7 (6-9)% 12 (10-14)% 17 (15-20)% 20 (17-23)% 875

80 (76-83)% 67 (63-71)% 55 (50-59)% 45 (41-50)% 43 (38-47)% 498

31 (25-37)% 44 (37-50)% 55 (48-61)% 56 (50-63)% 57 (50-63)%

14 (12-16)% 22 (20-25)% 28 (25-31)% 32 (29-35)% 35 (32-38)%

7 (5-10)% 12 (9-15)% 17 (14-20)% 21 (17-25)% 23 (19-26)%

76 (70-81)% 63 (57-70)% 51 (44-58)% 40 (34-47)% 38 (32-45)% 218

91 (89-93)% 83 (80-85)% 70 (67-73)% 60 (56-63)% 54 (51-57)%

13 (10-16)% 21 (17-24)% 28 (25-32)% 34 (30-38)% 35 (31-39)%

14 (9-18)% 19 (14-25)% 25 (20-31)% 29 (23-36)% 32 (26-38)%

2006-2012 Probability (95%CI)

904 86 (83-89)% 76 (72-79)% 63 (59-67)% 52 (48-57)% 47 (42-51)%

11 (7-15)% 17 (13-23)% 24 (18-30)% 30 (24-37)% 30 (24-37)% 214

N. evaluated

82 (79-84)% 71 (68-74)% 60 (57-63)% 51 (47-54)% 45 (42-49)% 892

29 (25-33)% 47 (42-51)% 58 (54-63)% 61 (57-65)% 62 (57-66)%

22 (19-24)% 41 (38-44)% 57 (54-60)% 61 (58-64)% 62 (58-65)%

P-value 0.02 0.02 <0.001 <0.001 <0.001 <0.001 0.24 0.48 0.27 0.34 0.63 0.41 0.02 <0.001 <0.001 <0.001 0.001 0.003 0.38 0.14 0.10 0.02 0.01 0.16 0.44 0.001 0.11 0.68 0.42 0.39

continued on the next page

962

haematologica | 2017; 102(5)


Outcomes of acute GvHD over time

reduction in TRM for those with overall maximal grade III-IV acute GvHD over time, relapse incidence increased.

Prognostic factors Two major multivariate modeling approaches were implemented to examine the effect of time period on OS and TRM following the diagnosis of acute GvHD. First, the effect of time period on outcome among cases of grade II-IV acute GvHD was examined separately for the tacrolimus-based GvHD prophylaxis group and cyclosporine-based prophylaxis group (Table 4). Reasons for this analysis include a recent trend to use tacrolimus rather than cyclosporine-based acute GvHD prophylaxis. These analyses demonstrated a significant improvement in OS and reduction in TRM with more recent HCT recipients among those who received tacrolimus-based GvHD prophylaxis. In contrast, these effects were not observed among those who received cyclosporine-based GvHD prophylaxis. These conclusions remained in secondary analyses that: (i) re-categorized time period cohorts (1999-2003, 2004-2008, 2009-2012), and (ii) excluded patients with upper gastrointestinal tract acute GvHD. Given the significant impact of overall acute GvHD grade on OS and TRM, a second modeling approach sep-

arately examined outcomes for the subgroups with either overall grade II or grade III-IV acute GvHD (Table 4). Similar reductions in hazards for OS and TRM were observed for the tacrolimus-based prophylaxis group in both overall grade II and grade III-IV acute GvHD, however these effects were only significant in the overall grade II group. Adjusted OS and TRM plots for the cyclosporine- and tacrolimus-based GvHD prophylaxis groups separately are presented in Figure 1.

Discussion This analysis provides new insight into trends in acute GvHD severity over time and determinants of mortality after the diagnosis of acute GvHD, and clarifies the relative impact of advances in GvHD outcomes according to overall acute GvHD grade and GvHD prophylaxis groups. We report several key findings. These data demonstrate a shift in maximal grade of acute GvHD over time, such that the relative contribution of grade III-IV disease among those affected by acute GvHD in the period of 2006-2012 is 20% lower than that in 1999-2001 and 10% lower than that in 2002-2005. This reduction in grade IIIIV disease is associated with a marked shift in GvHD pro-

continued from the previous page

Acute GvHD grade III-IV Overall survival 100-day 6-month 1-year 2-year 3-year Relapse 100-day 6-month 1-year 2-year 3-year Transplant-related mortality 100-day 6-month 1-year 2-year 3-year Disease-free survival 100-day 6-month 1-year 2-year 3-year Chronic GvHD 100-day 6-month 1-year 2-year 3-year

278

455 55 (49-60)% 36 (31-42)% 27 (22-32)% 21 (16-26)% 19 (15-24)%

277

542 54 (49-58)% 42 (37-46)% 32 (28-36)% 25 (21-29)% 22 (19-26)%

449 9 (6-12)% 13 (9-17)% 14 (10-18)% 16 (12-20)% 16 (12-21)%

277

531 11 (8-14)% 15 (12-19)% 18 (15-22)% 21 (17-24)% 21 (18-25)%

449 43 (37-48)% 55 (49-61)% 61 (55-67)% 64 (59-70)% 65 (60-71)%

277

531

449

29 (25-33)% 35 (31-39)% 41 (37-45)% 45 (41-49)% 47 (43-51)% 531

47 (42-51)% 36 (32-41)% 28 (24-32)% 22 (19-26)% 20 (17-24)% 451

16 (12-21)% 29 (24-35)% 33 (27-38)% 34 (29-40)% NE*

15 (12-19)% 20 (17-24)% 25 (21-29)% 28 (24-32)% 29 (25-33)%

43 (38-47)% 49 (44-53)% 54 (49-59)% 57 (53-62)% 58 (54-63)%

49 (43-55)% 32 (27-38)% 25 (20-30)% 20 (15-25)% 18 (14-23)% 276

65 (61-69)% 52 (48-56)% 40 (36-44)% 31 (27-35)% 26 (23-30)%

56 (52-60)% 45 (40-49)% 34 (30-38)% 27 (23-31)% 24 (21-28)% 532

14 (11-17)% 24 (20-28)% 32 (28-36)% 33 (29-38)% 33 (29-38)%

16 (13-19)% 27 (23-31)% 37 (33-41)% 40 (36-44)% 40 (36-44)%

0.001 <0.001 <0.001 <0.001 0.004 0.06 <0.001 0.009 0.02 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.010 0.01 <0.001 0.01 0.06 0.12 0.09 0.62 0.23 0.22 0.07 0.07

*:fewer than 15 patients eligible to be evaluated for chronic GvHD at 3 years after transplantation.

haematologica | 2017; 102(5)

963


H.J. Khoury et al.

phylaxis delivered over time, with tacrolimus-based approaches comprising 80% of cases in the period 20062012. With that, mortality has decreased over time, and deaths from organ failure and infection have decreased. These data provide current benchmarks for further analyses and counseling of patients for expected current HCT outcomes using these approaches. Multivariate analyses demonstrated significant improvements in OS and TRM over time but, importantly, refined these conclusions further. The effects appear to be limited to those treated with tacrolimus-based GvHD prophylaxis, and most apparent in those with overall grade II acute GvHD. Given these findings, we further examined GvHD characteristics, GvHD therapy, and causes of death among the cyclosporine and tacrolimus groups restricted to overall grade II disease (Online Supplementary Tables S1 and S2). Among grade II cases, the cyclosporine group had greater and more persistent representation of three-organ involvement (skin+liver+gastrointestinal) across the time period cohorts compared to the tacrolimus group. This GvHD

Table 3. Causes of death. Number of deaths (%) Primary disease New malignancy GvHD Idiopathic pneumonia Infection Organ failure Hemorrhage Vascular Othera Missing

1999-2001

2002-2005

2006-2012

381 115 (30) 6 (2) 62 (16) 29 (8) 78 (20) 57 (15) 15 (4) 2 (<1) 7 (2) 10 (3)

690 234 (34) 10 (1) 148 (21) 48 (7) 104 (15) 95 (14) 8 (1) 4 (<1) 15 (2) 24 (3)

890 370 (42) 6 (<1) 183 (21) 27 (3) 137 (15) 108 (12) 6 (<1) 4 (<1) 16 (2) 33 (4)

a Graft rejection (n=6), other HCT-related cause (n=5), accidental death (n=4), prior malignancy (n=4), seizure (n=3), cerebral edema (n=2), diffuse alveolar damage (n=2), drug overdose (n=2), acute myocardial infarction/myocardial infarction (n=2), encephalopathy (n=1), bone marrow myelofibrosis and necrosis (n=1), hypovolemic shock (n=1), leukoencephalopathy (n=1), numerous transient ischemic attacks (n=1), post-transplant lymphoproliferative disease (n=1), Stevens-Johnson syndrome (n=1), thrombotic thrombocytopenic purpura (n=1).

Table 4. Multivariate analysis results: effect of time cohort on overall survival and transplant-related mortality.

Overall survival Cyclosporine A

HR

CI

1

Year of transplant 1999-2001 2002-2005 2006-2012

Tacrolimus

1.00 1.02

0.82-1.21 0.80-1.29

0.98 1.00 0.99 0.88

HR

CI

P-value

3

Year of transplant 1999-2001 2002-2005 2006-2012

0.0028 1.00 1.01 0.80

0.80-1.29 0.63-1.01

Transplant-related mortality HR CI

P-value

0.91 0.06

2

Year of transplant 1999-2001 2002-2005 2006-2012

0.97 1.00 1.00 0.97

0.81-1.24 0.75-1.26

0.99 0.83

HR

CI

P-value

4

Year of transplant 1999-2001 2002-2005 2006-2012

P-value

0.0079 1.00 0.83 0.66

0.61-1.12 0.49-0.90

0.22 0.007

Acute GvHD grade II

Overall survival Tacrolimus

HR

CI

5

Year of transplant 1999-2001 2002-2005 2006-2012 Acute GvHD grade III-IV

1.00 1.02 0.80

0.0494

Tacrolimus

HR

0.69-1.51 0.55-1.16

Transplant-related mortality HR CI

P-value

0.91 0.25

6

Year of transplant 1999-2001 2002-2005 2006-2012

Overall survival CI

5

Year of transplant 1999-2001 2002-2005 2006-2012

0.19 1.00 0.98 0.82

0.71-1.37 0.59-1.14

0.92 0.24

0.0397 1.00 0.60 0.51

0.34-1.04 0.30-0.87

Transplant-related mortality HR CI

P-value 6

Year of transplant 1999-2001 2002-2005 2006-2012

P-value

0.071 0.013

P-Value 0.0785

1.00 0.90 0.70

0.62-1.30 0.48-1.03

0.56 0.07

1 Significant factors (P<0.01) in the overall survival model among cyclosporine A recipients: acute GvHD grade (II vs. III-IV; P<0.0001),acute GvHD-affected organs (P<0.0001), acute GvHD treatment (P=0.0001), disease status (P<0.0001), Karnofsky performance score (P=0.0014). Model stratified by disease, sex match, graft type. 2Significant factors in the transplant-related mortality model among cyclosporine A recipients: acute GvHD grade (II vs. III-IV; P<0.0001), age (P=0.0016), acute GvHD-affected organs (P<0.0001), acute GvHD treatment (P=0.0001), Karnofsky performance score (P=0.0020). Model stratified by sex match, graft type. 3Significant factors in the overall survival model among tacrolimus recipients: acute GvHD grade (II vs. III-IV; P<0.0001), age (P<0.0001), acute GvHD-affected organs (P<0.0001), acute GvHD treatment (P=0.0001), disease status (P<0.0001), donor type & age (P=0.0002), Karnofsky performance score (P=0.0003). Model stratified by disease, sex match, graft type. 4Significant factors in the transplant-related mortality model among tacrolimus recipients: acute GvHD grade (II vs. III-IV; P<0.0001), age (P=0.0007), acute GvHD-affected organs (P<0.0001), acute GvHD treatment (P=0.0001), donor type & age (P<0.0001). Model stratified by sex match, graft type. 5Overall survival models for tacrolimus recipients stratified by disease, sex match, graft type. 6The transplant-related mortality models for tacrolimus recipients were stratified by sex match, graft type.

964

haematologica | 2017; 102(5)


Outcomes of acute GvHD over time

phenotype (skin+liver+gastrointestinal) was associated with greater mortality on multivariate analysis. Additionally, re-categorization of the acute GvHD subjects in the cyclosporine and tacrolimus groups according to the refined Minnesota risk classification demonstrated that a greater and relatively fixed proportion of overall grade II acute GvHD subjects in the cyclosporine group had high-risk disease over the respective time period cohorts (Online Supplementary Table S3).13 High-risk patients by this classification have been shown to have inferior treatment response and increased TRM. 13,14 Otherwise, we found no major differences in acute GvHD therapy between the cyclosporine and tacrolimus-based prophylaxis groups to explain differential survival outcome. Analysis of cause of death information suggested a decrease in deaths related to infections and organ failure in the tacrolimus group, but not in the cyclosporine group. We acknowledge that prior major randomized trials examining tacrolimus versus cyclosporine (both in combination with methotrexate)

have demonstrated no major difference in survival outcome.2,3 This current retrospective analysis differs in that it is exclusively focused on survival estimation from time of acute GvHD onset among those with acute GvHD, and includes patients transplanted in the decade following these major trials. While these effects were most apparent in overall grade II acute GvHD, we note that similar effects were seen in the grade III-IV GvHD group that had received tacrolimusbased prophylaxis. These findings are further supported by a recent analysis that examined mortality after grade III-IV acute GvHD among 427 patients, in whom significant improvements in OS and TRM were observed in the comparison of a 2007-2012 group versus a 1997-2006 group.15 These findings were predominantly identified in the grade IV acute GvHD subgroup in this analysis. While important differences between the two study populations limit direct comparison, these findings support the concept that HCT outcome has improved for patients with grade III-IV acute GvHD also in the current era.

B

A

Figure 1. The adjusted probability of overall survival following a diagnosis of grade II-IV acute graft-versus-host disease. Patients treated with tacrolimus-based GVHD prophylaxis are detailed in (A) and those with cyclosporinebased GVHD prophylaxis are detailed in (B).

B

A

Figure 2. The adjusted probability of transplant-related mortality following a diagnosis of grade II-IV acute graftversus-host disease. Patients treated with tacrolimus-based GVHD prophylaxis are detailed in (A) and those with cyclosporine-based GVHD prophylaxis are detailed in (B).

haematologica | 2017; 102(5)

965


H.J. Khoury et al.

We note the following limitations to this analysis. First, we conducted analyses based on the reported maximal acute GvHD severity, but survival estimation was done from the time of the onset of the acute GvHD. Next, limitations in data on cyclosporine and tacrolimus therapeutic drug levels, prednisone dose information, and type/extent of second-line therapy make detailed analyses of treatment intensity on outcome not possible. Additionally, detailed acute GvHD treatment response data are not available, so we cannot characterize the burden of steroid-refractory disease across groups. Furthermore, based on the stated inclusion criteria for the analysis, these findings cannot be extended to recipients of reduced intensity conditioning. Finally, insufficient data on antimicrobial agents limit analysis of changes in these practices over time as a potential determinant of improved outcome. In conclusion, this analysis has shown that survival has improved over time in tacrolimus-based acute GvHD prophylaxis HCT recipients who develop acute GvHD. Acknowledgments CIBMTR Support List The CIBMTR is supported by Public Health Service Grant/Cooperative Agreement 5U24-CA076518 from the National Cancer Institute (NCI), the National Heart, Lung and Blood Institute (NHLBI) and the National Institute of Allergy and Infectious Diseases (NIAID); a Grant/Cooperative Agreement 5U10HL069294 from NHLBI and NCI; a contract HHSH250201200016C with Health Resources and Services Administration (HRSA/DHHS); two Grants N00014-15-1-0848

References 1. Cutler C, Logan B, Nakamura R, et al. Tacrolimus/sirolimus vs tacrolimus/ methotrexate as GVHD prophylaxis after matched, related donor allogeneic HCT. Blood. 2014;124(8):1372-1377. 2. Nash RA, Antin JH, Karanes C, et al. Phase 3 study comparing methotrexate and tacrolimus with methotrexate and cyclosporine for prophylaxis of acute graftversus-host disease after marrow transplantation from unrelated donors. Blood. 2000;96(6):2062-2068. 3. Ratanatharathorn V, Nash RA, Przepiorka D, et al. Phase III study comparing methotrexate and tacrolimus (prograf, FK506) with methotrexate and cyclosporine for graft-versus-host disease prophylaxis after HLA-identical sibling bone marrow transplantation. Blood. 1998;92(7):2303-2314. 4. Alousi AM, Weisdorf DJ, Logan BR, et al. Etanercept, mycophenolate, denileukin, or pentostatin plus corticosteroids for acute graft-versus-host disease: a randomized phase 2 trial from the Blood and Marrow Transplant Clinical Trials Network. Blood.

966

and N00014-16-1-2020 from the Office of Naval Research; and grants from Alexion; *Amgen, Inc.; Anonymous donation to the Medical College of Wisconsin; Astellas Pharma US; AstraZeneca; Be the Match Foundation; *Bluebird Bio, Inc.; *Bristol Myers Squibb Oncology; *Celgene Corporation; Cellular Dynamics International, Inc.; *Chimerix, Inc.; Fred Hutchinson Cancer Research Center; Gamida Cell Ltd.; Genentech, Inc.; Genzyme Corporation; *Gilead Sciences, Inc.; Health Research, Inc. Roswell Park Cancer Institute; HistoGenetics, Inc.; Incyte Corporation; Janssen Scientific Affairs, LLC; *Jazz Pharmaceuticals, Inc.; Jeff Gordon Children’s Foundation; The Leukemia & Lymphoma Society; Medac, GmbH; MedImmune; The Medical College of Wisconsin; *Merck & Co, Inc.; Mesoblast; MesoScale Diagnostics, Inc.; *Miltenyi Biotec, Inc.; National Marrow Donor Program; Neovii Biotech NA, Inc.; Novartis Pharmaceuticals Corporation; Onyx Pharmaceuticals; Optum Healthcare Solutions, Inc.; Otsuka America Pharmaceutical, Inc.; Otsuka Pharmaceutical Co, Ltd. – Japan; PCORI; Perkin Elmer, Inc.; Pfizer, Inc; *Sanofi US; *Seattle Genetics; *Spectrum Pharmaceuticals, Inc.; St. Baldrick’s Foundation; *Sunesis Pharmaceuticals, Inc.; Swedish Orphan Biovitrum, Inc.; Takeda Oncology; Telomere Diagnostics, Inc.; University of Minnesota; and *Wellpoint, Inc. The views expressed in this article do not reflect the official policy or position of the National Institute of Health, the Department of the Navy, the Department of Defense, Health Resources and Services Administration (HRSA) or any other agency of the U.S. Government. *Corporate Members

2009;114(3):511-517. 5. Levine JE, Logan B, Wu J, et al. Graft-versus-host disease treatment: predictors of survival. Biol Blood Marrow Transplant. 2010;16(12):1693-1699. 6. Bolanos-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;quiz 335. 7. Pidala J, Anasetti C. Glucocorticoid-refractory acute graft-versus-host disease. Biol Blood Marrow Transplant. 2010;16(11): 1504-1518. 8. Hahn T, McCarthy PL Jr, Hassebroek A, et al. Significant improvement in survival after allogeneic hematopoietic cell transplantation during a period of significantly increased use, older recipient age, and use of unrelated donors. J Clin Oncol. 2013;31(19):2437-2449. 9. Gooley TA, Chien JW, Pergam SA, et al. Reduced mortality after allogeneic hematopoietic-cell transplantation. N Engl J Med. 2010;363(22):2091-2101. 10. Przepiorka D, Weisdorf D, Martin P, et al. 1994 Consensus Conference on Acute GVHD Grading. Bone Marrow Transplant.

1995;15(6):825-828. 11. Shulman HM, Sullivan KM, Weiden PL, et al. Chronic graft-versus-host syndrome in man. A long-term clinicopathologic study of 20 Seattle patients. Am J Med. 1980;69(2):204-217. 12. Weisdorf D, Spellman S, Haagenson M, et al. Classification of HLA-matching for retrospective analysis of unrelated donor transplantation: revised definitions to predict survival. Biol Blood Marrow Transplant. 2008;14(7):748-758. 13. MacMillan ML, Robin M, Harris AC, et al. A refined risk score for acute graft-versushost disease that predicts response to initial therapy, survival, and transplant-related mortality. Biol Blood Marrow Transplant. 2015;21(4):761-767. 14. MacMillan ML, DeFor TE, Weisdorf DJ. What predicts high risk acute graft-versushost disease (GVHD) at onset?: identification of those at highest risk by a novel acute GVHD risk score. Br J Haematol. 2012;157(6):732-741. 15. El-Jawahri A, Li S, Antin JH, et al. Improved treatment-related mortality and overall survival of patients with grade IV acute GVHD in the modern years. Biol Blood Marrow Transplant. 2016;22(5):910-918.

haematologica | 2017; 102(5)


Haematologica, volume 102, issue 5  
Read more
Read more
Similar to
Popular now
Just for you