Haematologica, Volume 104, Issue 2

haematologica Journal of the Ferrata Storti Foundation

Editor-in-Chief Luca Malcovati (Pavia)

Managing Director Antonio Majocchi (Pavia)

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

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

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

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

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

haematologica Journal of the 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 2019 are as following: Print edition

Institutional Euro 700

Personal Euro 170

Advertisements. Contact the Advertising Manager, Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, e-mail: marketing@haematologica.org).

Disclaimer. Whilst every effort is made by the publishers and the editorial board to see that no inaccurate or misleading data, opinion or statement appears in this journal, they wish to make it clear that the data and opinions appearing in the articles or advertisements herein are the responsibility of the contributor or advisor concerned. Accordingly, the publisher, the editorial board and their respective employees, officers and agents accept no liability whatsoever for the consequences of any inaccurate or misleading data, opinion or statement. Whilst all due care is taken to ensure that drug doses and other quantities are presented accurately, readers are advised that new methods and techniques involving drug usage, and described within this journal, should only be followed in conjunction with the drug manufacturer’s own published literature.

Direttore responsabile: Prof. Edoardo Ascari; Autorizzazione del Tribunale di Pavia n. 63 del 5 marzo 1955.

Printing: Press Up, zona Via Cassia Km 36, 300 Zona Ind.le Settevene - 01036 Nepi (VT)

haematologica Journal of the Ferrata Storti Foundation

Table of Contents Volume 104, Issue 2: February 2019

Cover Figure Bone marrow smear showing an erythroblastic island in a girl treated with chemotherapy for acute lymphoblastic leukemia. Courtesy of Prof. Rosangela Invernizzi.

Editorials 215

Antithymocyte globulin and cyclosporin: standard of care also for older patients with aplastic anemia Andrea Bacigalupo

216

Milk and the “Grandmother Effect”– a new contribution to the legacy of Ray Owen William J. Burlingham

219

Another step forward in the 20-year history of IGHV mutations in chronic lymphocytic leukemia Ilaria Del Giudice and Robin Foà

222

The stepchild in myeloma treatments: is allogeneic transplantation not so bad after all?

Antonia M.S. Müller et al.

Review Articles 226

CD30-positive primary cutaneous lymphoproliferative disorders: molecular alterations and targeted therapies Lucia Prieto-Torres et al.

236

Pathogenic immune response to therapeutic factor VIII: exacerbated response or failed induction of tolerance?

Aditi Varthaman et al.

Articles Hematopoiesis

245

Rheb1 loss leads to increased hematopoietic stem cell proliferation and myeloid-biased differentiation in vivo Xiaomin Wang et al.

Bone Marrow Failure

256

Aplastic anemia in the elderly: a nationwide survey on behalf of the French Reference Center for Aplastic Anemia Adrien Contejean et al.

Red Cell Biology & its Disorders

263

Exposure to non-inherited maternal antigens by breastfeeding affects antibody responsiveness Henk Schonewille et al.

Phagocyte Biology & its Disorders

269

Treatment outcomes and prognostic factors in adult patients with secondary hemophagocytic lymphohistiocytosis not associated with malignancy Jae-Ho Yoon et al.

Acute Myeloid Leukemia

277

Comprehensive genetic diagnosis of acute myeloid leukemia by next-generation sequencing Elisabeth K.M. Mack et al.

288

A novel deep targeted sequencing method for minimal residual disease monitoring in acute myeloid leukemia Esther Onecha et al.

Haematologica 2019; vol. 104 no. 2 - February 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation 297

Targeted RNA-sequencing for the quantification of measurable residual disease in acute myeloid leukemia Laura W. Dillon et al.

305

Persistent IDH1/2 mutations in remission can predict relapse in patients with acute myeloid leukemia Chi Young Ok et al.

Acute Lymphoblastic Leukemia

312

Prognostic implications of additional genomic lesions in adult Philadelphia chromosome-positive acute lymphoblastic leukemia Anna Lucia Fedullo et al.

319

Hypersensitivity reactions to asparaginase in mice are mediated by anti-asparaginase IgE and IgG and the immunoglobulin receptors FcÎľRI and FcgRIII Sanjay Rathod et al.

Non-Hodgkin Lymphoma

330

JUNB, DUSP2, SGK1, SOCS1 and CREBBP are frequently mutated in T-cell/histiocyte-rich large B-cell lymphoma Bianca Schuhmacher et al.

338

T-cell inflamed tumor microenvironment predicts favorable prognosis in primary testicular lymphoma Suvi-Katri Leivonen et al.

347

miR-497 suppresses cycle progression through an axis involving CDK6 in ALK-positive cells Coralie Hoareau-Aveilla et al.

Chronic Lymphocytic Leukemia

360

Tailored approaches grounded on immunogenetic features for refined prognostication in chronic lymphocytic leukemia Panagiotis Baliakas et al.

Plasma Cell Disorders

370

Allogeneic transplantation of multiple myeloma patients may allow long-term survival in carefully selected patients with acceptable toxicity and preserved quality of life Christine Greil et al.

380

Long-term follow up of tandem autologous-allogeneic hematopoietic cell transplantation for multiple myeloma Enrico Maffini et al.

Stem Cell Transplatation

392

The TLR7 ligand R848 prevents mouse graft-versus-host disease and cooperates with anti-interleukin-27 antibody for maximal protection and regulatory T-cell upregulation MĂŠlanie Gaignage et al.

Blood Transfusion

403

Anti-HLA antibodies with complementary and synergistic interaction geometries promote classical complement activation on platelets Maaike Rijkers et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/104/2.toc

e47

Endothelial protein C receptor (EPCR) expression marks human fetal liver hematopoietic stem cells Agatheeswaran Subramaniam et al. http://www.haematologica.org/content/104/2/e47

e51

Prevalence and management of iron overload in pyruvate kinase deficiency: report from the Pyruvate Kinase Deficiency Natural History Study Eduard J. van Beers et al. http://www.haematologica.org/content/104/2/e51

Haematologica 2019; vol. 104 no. 2 - February 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

e54

Bone marrow mesenchymal stem/stromal cells from risk-stratified acute myeloid leukemia patients are anti-inflammatory in in vivo preclinical models of hematopoietic reconstitution and severe colitis Rafael Diaz de la Guardia et al. http://www.haematologica.org/content/104/2/e54

e59

Relationship between CD33 expression, splicing polymorphism, and in vitro cytotoxicity of gemtuzumab ozogamicin and the CD33/CD3 BiTE® AMG 330 George S. Laszlo et al. http://www.haematologica.org/content/104/2/e59

e63

Continuous high dosing of lenalidomide in relapsed, refractory or older newly diagnosed acute myeloid leukemia patients not suitable for other treatment options - results from a phase I study Marie Luise Hütter-Krönke et al. http://www.haematologica.org/content/104/2/e63

e65

ABVD plus rituximab versus ABVD alone for advanced stage, high-risk classical Hodgkin lymphoma: a randomized phase 2 study Paolo Strati et al. http://www.haematologica.org/content/104/2/e65

e68

Efficacy of venetoclax monotherapy in patients with relapsed, refractory mantle cell lymphoma after Bruton tyrosine kinase inhibitor therapy Toby A. Eyre et al. http://www.haematologica.org/content/104/2/e68

e72

Long non-coding RNA NEAT1 shows high expression unrelated to molecular features and clinical outcome in multiple myeloma Elisa Taiana et al. http://www.haematologica.org/content/104/2/e72

Case Reports to the Editor Case Reports are available online only at www.haematologica.org/content/104/2.toc

e77

Unrelated cord blood transplantation and post-transplant cyclophosphamide Andrea Bacigalupo et al. http://www.haematologica.org/content/104/2/e77

e79

MPI-CDG with transient hypoglycosylation and antithrombin deficiency María Eugenia de la Morena-Barrio et al. http://www.haematologica.org/content/104/2/e79

e83

Disappearance of a strong triple positivity for antiphospholipid antibodies after treatment with anakinra Erik Arnaud et al. http://www.haematologica.org/content/104/2/e83

Comments Comments are available online only at www.haematologica.org/content/104/2.toc

e85

Prevention of serious infections in hereditary hemorrhagic telangiectasia: roles for prophylactic antibiotics, the pulmonary capillaries- but not vaccination Claire Shovlin et al. http://www.haematologica.org/content/104/2/e85

Haematologica 2019; vol. 104 no. 2 - February 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

Ancient Greek

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

Scientific Latin

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

Scientific Latin

haematologicus (adjective) = related to blood

Modern English

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

2017 JCR impact factor = 9.090

EDITORIALS Antithymocyte globulin and cyclosporin: standard of care also for older patients with aplastic anemia Andrea Bacigalupo Fondazione Policlinico Universitario A. Gemelli IRCCS, Universita’ Cattolica del Sacro Cuore, Rome, Italy E-mail: apbacigalupo@yahoo.com doi:10.3324/haematol.2018.207167

T

reatment of aplastic anemia (AA) is challenging because cytopenia exposes patients to infectious and hemorrhagic complications, and immunosuppressive agents may further increase the risk in the short term: this is particularly true in older patients, in whom age, per se, is a risk factor.

For this reason, I believe the study published in this issue on behalf of the French Cooperative Study Group for Aplastic Anemia,1 on 88 patients over the age of 60 years, given immunosuppressive therapy (IST), is clinically relevant and useful on a practical level. The authors identified 184 treatments, including the standard combination of anti-thymocyte globulin and cyclosporine-A (ATG-CsA). Responses were significantly better in those that had followed an ATG-CsA regimen; overall response rate was 70% after first-line treatment and the overall 3-year survival was an excellent 74%. The Authors conclude that age, per se, is not a limiting factor for standard immunosuppressive treatment of patients with AA. Treatment of severe aplastic anemia (SAA) patients over the age of 60 is problematic, and transplant studies claiming good results in older patients2 contain only very few patients over the age of 60: in the study by Shin and coworkers, which suggests that transplant outcome is comparable in patients younger or older than 40 years, the maximum patient age was 63 years.2 The cyclophosphamide (CY) dose-finding study by Anderlini and coworkers on patients receiving an alternative donor transplant, which showed over 90% one-year survival for the low dose CY group, had only a handful of patients over the age of 40, and the median age overall was 21 years.3 More recently, a European Group for Blood and Marrow Transplantation (EBMT) study showed that the outcome of

transplant for SAA patients over the age of 40 remains unsatisfactory4 and has not improved in the past decade. Also, in patients with leukemia, age is a predictive factor, but transplant mortality is much lower than in SAA patients; a recent study in patients over the age of 70 years reported a transplant mortality of 20%.5 Therefore, age has a stronger impact on survival after transplantation in SAA than leukemia, for reasons which are not clear. Currently, transplantation over the age of 60 is rarely taken into consideration by transplant centers: in the EBMT study 2010-2015, 95 grafts there performed on patients over the age of 60 (16 transplants/year in Europe).4 Immunosuppressive treatment is also strongly age-dependent: in a prospective randomized EBMT study, the 6-year survival was 100% for patients <20 years old, 92% for patients aged 20-40 years, 71% for those aged 40-60, and 56% for patients older than 60 years.6 Two recent retrospective analyses have addressed the issue of first-line treatment in a large number of SAA patients with either horse ATG-CsA (in 465 patients)7 or rabbit ATG-CsA (in 955 patients).8 The one year cumulative incidence of response was 65% for both horse or rabbit ATG, but patients over 60 years of age had lower response rates.7,8 As for survival, the horse ATG-CsA study shows a 5-year survival of 70% for patients over the age of 40, compared to 90% for young patients.7 The rabbit ATG-CsA study shows a similar 5-year survival for patients aged 40-60 (80%), but over the age of 60, survival at 5 years is in the order of 50%, compared to 90% for younger patients.8 Therefore, also after ATG-CsA, age remains one of the major predictors of response and survival.

The French study by Contejean and coworkers1 compares

Figure 1. Proportion of patients responding to immunosuppressive therapy (IST). hATG: horse anti-thymocyte globulin; rATG: rabbit ATG; CsA : cyclosporine A.

haematologica | 2019; 104(2)

215

Editorials

favourably with these two large international reports: response rate to ATG+CsA at 6 months was 70% compared to 35% for patients receiving CsA alone, 22% for eltrombopag alone, and 21% for androgens alone (Figure 1).

Overall response rates in patients receiving these different treatment options is outlined in Figure 1, which highlights a superior response rate with the ATG+CsA combination.

Figure 1 also shows there was no difference in response for patients in this older patient population receiving horse (hATG) versus rabbit ATG (rATG), which is in keeping with two large recently published real-life studies.7,8 Also, survival was comparable in patients receiving horse or rabbit ATG. The higher response rate of patients receiving ATG+CsA did not translate to a significantly improved survival compared to other regimens, as already shown in an EBMT study;9 possibly due to second treatment or improved supportive care. Further support to long-term survival also for non-responders comes from the rabbit ATG study; it showed that the 10-year survival of AA patients, classified at 6 months as non-responders to a first course of rATG+CsA, was comparable whether patients were then allografted (64% survival) or not (60% survival).8 Late responses, improved supportive care, and second treatments could possibly explain the outcome of non-allografted patients.

In conclusion, ATG+CsA remains the treatment of choice for patients with AA, also for those over the age of 60; it should be preferred over the administration of CsA with or without androgens or eltrombopag, with or without CsA. Whether the addition of eltrombopag to ATG-CsA first line will further improve the outcome, as recently suggested,10 will be determined by an ongoing prospective randomized trial (RACE trial, EBMT). This study also confirms that in real-life analysis, horse or rabbit ATG produce almost identical response rates and survival, also in older AA patients.

References 1. Contejean A, Resche-Rigon M, Tamburini J et al. Aplastic anemia in the elderly: a nationwide survey on behalf of the French Reference Center for Aplastic Anemia. Haematologica, 2018;104 (2):256-262.

2. Shin SH, Jeon YW, Yoon JH, et al. Comparable outcomes between younger (⩽40 years) and older (>40 years) adult patients with severe aplastic anemia after HLA-matched sibling stem cell transplantation using fludarabine-based conditioning. Bone Marrow Transplant.

2016;51(11):1456-1463 3. Anderlini P, Wu J , Gersten I, et al. Cyclophosphamide conditioning in patients with severe aplastic anaemia given unrelated marrow transplantation: a phase 1–2 dose de-escalation study. Lancet Haematol.

2015;2(9):e367-75.

4. Giammarco S, Peffault de Latour R, Sica S, et al. European Group for Blood and Marrow Transplantation Severe Aplastic Anemia Working Party. Transplant outcome for patients with acquired aplastic anemia over the age of 40: has the outcome improved? Blood.

2018;131(17):1989-1992 5. Al Marki MM, Nathwani N, Yang D, et al. Melphalan based reduced intensity conditioning is associated with favorable disease control and acceptable toxicities in patients older than 70 years with hematologic malignancies, undergoing allogeneic stem cell transplantation. Biol Blood and Marrow Transpl. 2018;24(9):1828-1835 6. Tichelli G, Schrezenmeier H, Socie G, et al. A randomized controlled study in patients with newly diagnosed severe a aplastic anemia receiving antithymocyte globulin (ATG), cyclosporine, with or without GCSF: study of the SAA Working Party of the European Group for Blood and Marrow Transplantation. Blood. 2011;117(17):4434-4441 7. Peffault de la Tour R, Tabrizi R, Marcais A, et al. Nationwide survey on the use of horse antithymocyte globulins (ATGAM) in patients with acquired aplastic anemia: A report on behalf of the French Reference Center for Aplastic Anemia. Am J Hematol. 2018;93(5):1-8 8. Bacigalupo A, Oneto R, Schrezenmeier H, et al. First line treatment of aplastic anemia with thymoglobuline in Europe and Asia: Outcome of 955 patients treated 2001-2012. Am J Hematol. 2018;93(5):643-648 9. Marsh J, Schrezenmeier H, Marin P, et al. Prospective randomized multicenter study comparing cyclosporin alone versus the combination of antithymocyte globulin and cyclosporin for treatment of patients with nonsevere aplastic anemia: a report from the European Blood and Marrow Transplant (EBMT) Severe Aplastic Anaemia Working Party.

Blood. 1999;93(7):2191-2195.

10. Townsley D, Scheinberg P, Winkle T, et al. Eltrombopag added to standard immunosuppression for aplastic anemia. New Engl J Med. 2017; 376(16):1540-1545.

Milk and the “Grandmother Effect”– a new contribution to the legacy of Ray Owen William J. Burlingham University of Wisconsin, Madison, WI, USA E-mail: burlingham@surgery.wisc.edu doi:10.3324/haematol.2018.207340

T

he article by Schonewille et al., in the September 2018 edition of Haematologica,1 tests the theory originally proposed by Owen et al. [PNAS, 19542] that a “Grandmother Effect” can protect the offspring from hemolytic disease of the fetus and newborn (HDFN) [see illustration, Figure 1]. This disease is caused by maternal antibodies formed against the baby’s red blood cells (rbc).

One should note here that, thanks to the research of Owen and others in the 1950s, anti-Rhesus D (RhD) antibody (aka “Rhogam”) prophylaxis began to be practiced for all pregnant women who are RhD- bearing an RhD+ fetus. The absence of such therapy prior to 1954 enabled Owen and colleagues to determine that, when the grandmother had RhD, the mother who was RhD- and had thereby already been “naturally” exposed to this antigen in utero and as a

216

newborn via breast-feeding, was rendered incapable (for the most part) of producing antibody to an RhD encountered in her adult life as the mother of a child sired by her RhD+ husband. Subsequent to the routine application of Rhogam (anti-RhD prophylaxis) in the 1960’s, the incidence of hemolytic disease of the newborn was greatly decreased, although not completely eliminated. Briefly, Rh- mothers are given a bolus of Rhogam at 28 weeks of pregnancy. If the baby that is born is Rh+(a 50:50 chance, if her husband is RhD heterozyogous), she will receive a second dose, 72 hours after giving birth. This procedure has revolutionized obstetrics and made it possible for healthy Rh+ babies to be born from multiple pregnant Rh- women.

While a boon to Ob/Gyn medicine, this highly effective clinical practice made it nearly impossible for others to haematologica | 2019; 104(2)

Editorials

Figure 1. The Grandmother Effect. Grandmother [x -/+] gives birth to a baby girl, who grows into a women [x -/-] whose immune system bears the memory of exposure to a red blood cell (rbc) antigen (X) that she did NOT inherit but remains in her body at low levels due to microchimerism (Mc; the situation where foreign cells--in this case the grandmother's-- although rare, persist in blood and tissues). Mother(x -/-) then conceives a child by a father whose rbc expresses the very same non-inherited maternal antigen (NIMA;=x). If the newly-conceived child inherits this gene from the father (x encodes an inherited paternal antigen, or IPA), then the mother, during pregnancy, recalls her exposure to this "grandmother" antigen and is less likely to form an anti-x antibody response, thus protecting the fetus.

repeat the exact test that Ray Owen had used to develop the Grandmother Effect hypothesis (even though Ray specifically invited others to do so2). A torrent of publications followed his, which were not able to confirm his observations of 1954. Meanwhile, other societal changes were occurring, particularly the advent of formula-fed babies, no longer exposed for long periods to breast milk.

Schonewille et al. therefore used a novel approach to the problem, based on the breast-feeding history of the mother provided by the grandmother. The hypothesis was that breast-feeding would condition a baby girl such that antibody response to any non-inherited erythrocyte antigens, including RhD, but also others for which prophylaxis was not done, would be inhibited. When encountered in later life, as she becomes pregnant, they hypothesized that these breast milk exposures from the grandmother would diminish her chances of developing specific antibodies to her husband’s (inherited paternal) antigens (IPAs) expressed by her child’s rbc.

To test whether there is indeed a cross-generational effect of the grandmother, preventing the daughter from producing antibodies to her baby’s rbc and causing hemolytic disease of the newborn, Schonewille et al. chose to study preghaematologica | 2019; 104(2)

nancies in cases where mothers had already developed the potential for hemolytic disease. In these women, in the period 1987-2008, their as yet unborn children were being treated by intrauterine transfusion, and their pregnancies managed with the most up-to-date techniques. The investigators studied the antibody responses of the mothers to inherited paternal antigens (IPA) expressed by their babies in relation to various factors, including the breast-feeding history of the maternal grandmother. This approach took the focus off of in utero exposure only and put it onto the combination of in utero exposure and breast-feeding.

Schonewille and colleagues made two critically important and novel observations: 1) that there was indeed a Grandmother Effect limiting the mother’s ability to produce anti IPA antibodies to a variety of erythrocyte antigens; and 2) that the capacity for a Grandmother Effect was timedependent: i.e., it developed only after at least two months of breast-feeding of the mother by the grandmother. The power of the study was considerable, since the study was done on 125 3-generational Dutch families (grandmothers, mothers, and babies) exposed to 330 non-D rbc antigens and involved the measurement of antibody responses to each of these. Both highly as well as weakly immunogenic 217

Editorials

antigens were studied so as to develop a more balanced model of what the Grandmother Effect might encompass.

Out of the 549 rbc antigens not expressed by the mothers (after exclusion for antigens with no exposure from grandmother, or where antigen exposure was unknown), there were 330 known rbc antigen exposures in these women.

The differences became meaningful in that, after at least two months of oral NIMA exposure via breast-feeding (i.e., per the maternal grandmother), the odds ratio of the mother (now an adult) forming an antibody to her baby’s rbc was very low (0.12). Although continuation of breast-feeding of the grandmother beyond two months did not add additional protections, it was clear that breast-feeding was associated with the protections gained by the NIMA-exposed mother who had been re-exposed to the same antigen as an IPA during her pregnancy.

Discussion The observations of Schonewille et al.1 add to literature on the NIMA effect in important ways. The original observation of Ray Owen was challenged by a variety of individuals in the Ob/Gyn and blood transfusion fields.

However, there was no adequate breast-feeding history of the mothers involved in Ray’s study nor of the mothers whose results were claimed to have refuted Owen et al.2 However, it was clear from a study by Frans Claas and colleagues in 1988, published in Science,3 that in patients awaiting kidney transplant who were highly sensitized to HLA (>90% reactivity to a random panel), the existence of a NIMA among the different HLA types of cells tested by cytotoxicity assay tended to identify the rare cells to which the highly sensitized patient was unable to make an antibody. Similarly, in 1998 our lab published a paper4 showing that the presence of a NIMA HLA haplotype on a sibling kidney allograft donor greatly increased longterm graft survival in the recipient. This was despite the fact that there was an increased rate of early transplant rejection episodes when the NIMA haplotype was present on the haplo-mismatched kidney.4 This emphasized the split tolerance nature of the NIMA effect, where certain aspects of cellular immunity, particularly the so-called direct pathway of cellular immunity, were increased by reexposure to the NIMA, whereas the indirect pathway, which controls antibody responses in long-term kidney allograft recipients, was impaired. Some 10 years later, Jeff Mold, Mike McCune and colleagues published a couple of papers, one in 20085 and one in 2010,6 outlining in human studies strong support for the NIMA effect in utero, based on the development of NIMA-specific fetal T reg populations, and developing a new category for the “normal” CD4 T cell during in utero life—i.e., as one that has “T reg-like” qualities. Finally, Kinder et al.,7 working in Sing Sing Way’s lab, published a mouse study in 2015 showing exactly how such cross-generational tolerance occurs, and identifying a mechanism whereby NIMA exposure from the grandmother protects the daughter’s later pregnancy from fetal loss due to a Listeria infection.

The elegant study by Schonewille and colleagues raises

218

certain questions for the new era of “surrogate” motherhood. First, the surrogate mother is in a completely separate generational lineage, having been exposed by her own mother to certain non-inherited antigens that may or may not match with the antigens of the fetus which she carries.

One wonders about the records of successful term pregnancies vs. premature delivery or fetal loss in these “two HLA haplotype mismatched” (fully allogeneic) intrauterine transplants. A second question arising from the study by Schonewille et al.1 would be, “Does the surrogate mom breast-feed the baby or not, and if indeed she is breastfed by the surrogate mom, is a baby girl born to the surrogate mom now going to have outcomes of pregnancy that are related to the surrogate mom exposure?” This question can only be answered by further research on these individuals as they become adults. Another question regarding the implications of “milk-kinship” in certain Islamic and Native American societies (en.wikipedia.org/wiki/Milk_kinship), wherein a child is nursed by a woman who is not his/her biologic mother, might now be re-examined in light of the tolerogenic effects of breast-feeding described by Schonewille et al.,1 particularly in those cases where the period of breast-feeding by the “wet nurse” is ≥2 months duration.

Summary Overall, the paper by Schonewille et al.1 is an important contribution to the literature on the NIMA effect. It is fitting that one of the co-authors of this paper is Jon van Rood, who passed away in 2017. Jon would have been very happy to see this paper published. He was an early champion of the NIMA hypothesis, who shared in the discovery of HLA as a key element in human transplantation, even though the Nobel prize for this was awarded to Jean Dausset. Whether or not the paper promotes a recommendation from Ob/Gyn specialists for two or more months of breast-feeding, it certainly supports such a strategy for female babies, in order to insure long-term protection of babies in the next generation from HDFN.

References 1. Schonewille H, van Rood JJ, Verduin EP, et al. Exposure to non-inherited maternal antigens by breastfeeding affects antibody responsiveness.

Haematologica 2018;104(2):259-264.

2. Owen RD, Wood HR, Foord AG, Sturgeon P, Baldwin LG. Evidence for actively acquired tolerance to Rh antigens. Proc Natl Acad Sci U S A. 1954;40(6):420-424.

3. Claas FH, Gijbels Y, van Der Velden-de Munck J, van Rood JJ. Induction of B cell unresponsiveness to noninherited maternal HLA antigens during fetal life. Science. 1988;241(4874):1815-1817.

4. Burlingham WJ, Grailer AP, Heisey DM, et al. The effect of tolerance to noninherited maternal HLA antigens on the survival of renal transplants from sibling donors. N Engl J Med. 1998;339(23):1657-1664.

5. Mold JE, Michaelsson J, Burt TD, et al. Maternal alloantigens promote the development of tolerogenic fetal regulatory T cells in utero.

Science. 2008;322(5907):1562-1565.

6. Mold JE, Venkatasubrahmanyam S, Burt TD, et al. Fetal and adult hematopoietic stem cells give rise to distinct T cell lineages in humans.

Science. 2010;330(6011):1695-1699.

7. Kinder JM, Jiang TT, Ertelt JM, et al. Cross-generational reproductive fitness enforced by maternal cells. Cell. 2015;162(3):1-11.

haematologica | 2019; 104(2)

Editorials

Another step forward in the 20-year history of IGHV mutations in chronic lymphocytic leukemia Ilaria Del Giudice and Robin Foà Hematology, Department of Translational and Precision Medicine, ‘Sapienza’ University, Rome, Italy E-mail: rfoa@bce.uniroma1.it doi:10.3324/haematol.2018.207399

C

hronic lymphocytic leukemia (CLL) derives from the expansion of clonal and antigen-experienced mature B lymphocytes, whose accumulation results from a dynamic imbalance between cell death and proliferation. The former is impaired by the overexpression of the anti-apoptotic protein BCL2 by CLL cells and the latter is mainly driven by the B-cell receptor (BCR), the key molecule to elucidate the pathogenic and evolution mechanisms of the disease.

Chronic lymphocytic leukemia is a highly heterogeneous disease both in terms of biological landscape and clinical course, including the interval from diagnosis to first progression requiring treatment (time-to-first-treatment, TTFT), the degree and duration of response to therapy, overall survival (OS), and risk of transformation into an aggressive lymphoma (Richter’s syndrome). The prognosis of CLL patients can be accurately defined by combining clinical and biological parameters that include BCR features, cytogenetic lesions, immunophenotypic markers, and gene mutations. Some biomarkers are also useful predictors of response to therapy.

Mutations of the genes codifying for the immunoglobulin heavy chain variable region (IGHV) of the BCR represent one of the most robust prognostic biomarkers, and, indeed, was one of the first to be identified. IGHV mutations never change over time, and thus represent the fingerprint of the disease. Back in 1999, it was reported that CLL patients with mutated IGHV genes (M-CLL) (i.e. <98% cut-off of IGHV identity to the germline counterpart) display a longer TTFT and a longer survival than CLL with unmutated IGHV genes (U-CLL) (≥98%).1,2 The subsequent identification in about 30% of CLL of stereotyped BCRs was even more intriguing.3,4 Stereotyped BCRs (namely those with a nearly identical length of the HCDR3 region, shared amino acids in key positions and the non-stochastic pairing of IGHV and light chain genes) identify subgroups defined “subsets”. More frequent in U-CLL (40%) than in M-CLL (10%) in Caucasians, CLL subsets display distinctive clinico-biological associations: subset #4, mostly M-CLL, is associated to a young age at diagnosis and an indolent disease; subset #1, U-CLL, to a very aggressive clinical course; subset #8, U-CLL, to a higher risk of developing Richter’s syndrome; subset #2 to a poor prognosis regardless of the percentage of IGHV mutations.5 Although the IGHV gene usage and the frequency of BCR subsets can vary across populations with a different incidence of CLL (i.e.

Caucasian vs. Chinese), it is interesting that these clinicobiological associations hold true across all ethnic groups.6 In 2015, the value of the IGHV status in predicting the outcome after chemoimmunotherapy also emerged, since M-CLL patients have a significantly longer progression-free survival (PFS), particularly when devoid of poor-risk fluohaematologica | 2019; 104(2)

rescence in situ hybridization (FISH) lesions.7 In contrast, it became apparent that the IGHV status does not influence the efficacy of the BTK inhibitor ibrutinib.8,9 Thus, it has been suggested that both the IGHV status and TP53 deletions/mutations should be investigated at the time of disease progression in order to guide the first-line therapeutic choice between chemoimmunotherapy and novel agents.10 Given the clinical implications, the European Research Initiative on CLL (ERIC) group has conducted an international harmonization process across labs for the analysis and reporting of IGHV and TP53 genes in CLL, and this has led to the recently up-dated recommendations.11,12 Although the pathogenic mechanisms operational in CLL are far from being fully elucidated, the oncogenic function of the BCR is indirectly demonstrated by the high antileukemic efficacy of kinase inhibitors that block BCR signaling (i.e. ibrutinib, idelalisib, acalabrutinib, duvelisib). On the one hand, in CLL, unlike other lymphoproliferative diseases, the BCR is capable of generating a cell-autonomous signaling driven by the interactions between HCDR3 of near BCR (BCR-BCR) on the cell surface.13 On the other hand, the quality of BCR signaling is heterogeneous: U-CLL are more responsive in vitro to IgM ligation in terms of modulation of the gene expression profile, advance in the cell cycle, and increase in proliferation compared to M-CLL.14 As for a commonly accepted model, U-CLL show a weak autonomous BCR-BCR signaling, a low affinity binding to auto-antigens, an increased BCR responsiveness, and an aggressive clinical course, while M-CLL patients show a strong autonomous BCR-BCR signaling that leads to an anergic state, a lower proliferative response after BCR stimulus, and an overall indolent course.15,16 This model conciliates a shared pathogenic mechanism with the biological and clinical heterogeneity of CLL. In addition, BCR stereotyping likely supports the role of an antigenic pressure in the selection of the leukemic clone.3-5 Among the various factors that contribute to modulate the BCR responsiveness, the microenvironment certainly has a relevant role, since the CLL cells within the lymph node show an upregulation of genes involved in BCR signaling and NfKB activation, at variance from the circulating CLL cells from the same individual.17 Although the mechanisms driving the heterogeneity of CLL genetics are currently unknown, since leukemia kinetics and genetic complexity are usually closely related, it would seem that the BCR can play a role in maintaining genetic stability or in acquiring genetic instability in CLL. Indeed, U-CLL and M-CLL display a variable proportion of the different genetic lesions, as well as of the BCR subsets; U-CLL is enriched with biomarkers with an adverse prognostic significance, although not exclusively.

In the present issue of Haematologica, on behalf of the 219

Editorials

Figure 1. The 20-year history of IGHV mutations in chronic lymphocytic leukemia (CLL). IGHV: immunoglobulin heavy chain variable region; BCR: B-cell receptor; M: mutated; U: unmutated; CIT: chemoimmunotherapy; PFS: progression-free survival.

ERIC group, out of a large cohort of 2366 CLL cases, Baliakas et al.18 focused on 1900 stage A CLLs that were divided into the two main CLL immunogenetic subgroups: U-CLL and M-CLL. Considering each of them separately, they analyzed the relative weight of different prognostic markers in determining the TTFT.18 These prognostic markers included: age, sex, CD38, FISH lesions, TP53, SF3B1, NOTCH1, BIRC3, MYD88 gene mutations, and the major BCR subsets #1, #2 and #4.

This IGHV-based prognostic approach suggests that, among stage A M-CLL, cases with trisomy 12 and stereotyped subset #2 show a short TTFT at five and ten years after diagnosis, similar to those with TP53 abnormalities; within stage A U-CLL, cases with del(11q) and/or SF3B1 mutations experience a TTFT as short as cases with TP53 abnormalities. These markers are almost entirely mutually exclusive. The validity of the model is confirmed also in an external validation cohort of 649 Binet A CLL. Interestingly, male sex comes out as a determinant of disease course within U-CLL, a recurrent observation that has never been truly addressed.19 220

Overall, considering patients in all stages, within IGHV M-CLL, the lowest risk category (stage A/non-TP53 abnormalities/+12/subset #2), representing 73% of all M-CLL, shows a TTFT of 12% and 25% at five and ten years, respectively; this means that only 1 out of 4 patients has required treatment after ten years from diagnosis. The intermediate-risk category (stage A/TP53 abnormalities/+12/subset #2), representing 14% of all M-CLL, shows a TTFT of 40% and 55% at five and ten years, implying that 1 out of 2 patients still remains untreated after ten years from diagnosis.

On the other hand, within IGHV U-CLL, the very lowrisk category (stage A/female/non-SF3B1 mutation/del11q), representing 13% of all U-CLL, shows a TTFT of 45% and 65% at five and ten years, respectively, much shorter (median TTFT 6.1 years) than low-risk M-CLL patients (median TTFT not reached). The U-CLL low-risk (stage A/male/non-SF3B1 mutation/del11q) and intermediate-risk (stage A/SF3B1 mutation/del11q) patients, representing 19% and 24% of all U-CLL, respectively, show a median TTFT of 3.6 and 2.1 years. haematologica | 2019; 104(2)

Editorials

Although the retrospective nature of the study and the heterogeneity of the treatments administered prevent the authors from assessing OS in this series, these data identify important differences between the two main CLL immunogenetic subgroups and may help in building a novel CLL patient risk stratification where BCR plays a major role as the core of the prognostic model. For example, trisomy 12, associated to an intermediate prognosis in CLL analyzed as a whole, now has a better definition, having a detrimental prognostic impact on TTFT within M-CLL but not within U-CLL. Moreover, subset #2 should be identified and reported since it has an independent impact on TTFT within M-CLL. The CLL prognostic algorithms proposed so far have always included IGHV status. However, in six different scoring systems (CLL-IPI, MDACC 2007, MDACC 2011, GCLLSG, Barcelona-Brno, O-CLL1), the outcome prediction fails in a proportion ranging from 22% to 35% of early phase CLL.20 Especially within low-risk CLL, there is still a sizable proportion of patients experiencing “early” progression. A prognostic algorithm built within the U-CLL and MCLL IGHV categories may overcome this limitation. This has clinical implications for patients’ counseling, follow-up planning, and potential enrollment in trials exploring early treatment approaches for stage A CLL. It is clear that, in an era of new drugs, the value of all the algorithms proposed so far in terms of prediction of OS will need to be re-evaluated, since the BCR/BCL2 inhibitors have the power to overcome the prognostic impact of the IGHV mutational status. For the time being, besides TP53 deletions/mutations, it is advisable to consider IGHV gene sequencing in all CLL patients requiring first-line treatment as a guide to choose between conventional and innovative approaches. In this respect, the ERIC initiative on the IGHV and TP53 harmonization project and the certification system, with external quality controls for the accreditation of the laboratories performing IGHV/TP53 analysis, is a very timely effort.

Acknowledgments The authors wish to thank Associazione Italiana per la Ricerca sul Cancro (AIRC) 5 x 1000, Milan, Italy (Special Program Molecular Clinical Oncology MCO-10007; Metastases Special Program, n. 21198). They thank Dr. Sara Raponi and Dr. Luca Vincenzo Cappelli for their contribution in the design of Figure 1.

References 1. Damle RN, Wasil T, Fais F, et al. IgV gene mutation status and CD38 expression as novel prognostic indicators in chronic lymphocytic leukemia. Blood. 1999;94(6):1840-1847.

2. Hamblin TJ, Davis Z, Gardiner A, Oscier DG, Stevenson FK. Unmutated Ig V(H) genes are associated with a more aggressive form of chronic lymphocytic leukemia. Blood. 1999;94(6):1848-1854.

3. Messmer BT, Albesiano E, Efremov DG, et al. Multiple distinct sets of stereotyped antigen receptors indicate a role for antigen in promoting chronic lymphocytic leukemia. J Exp Med. 2004;200(4):519-525.

4. Tobin G, Thunberg U, Karlsson K, et al. Subsets with restricted immunoglobulin gene rearrangement features indicate a role for antigen selection in the development of chronic lymphocytic leukemia.

Blood. 2004;104(9):2879-2885.

haematologica | 2019; 104(2)

5. Stamatopoulos K, Agathangelidis A, Rosenquist R, Ghia P. Antigen receptor stereotypy in chronic lymphocytic leukemia. Leukemia.

2017;31(2):282-291.

6. Marinelli M, Ilari C, Xia Y, et al. Immunoglobulin gene rearrangements in Chinese and Italian patients with chronic lymphocytic leukemia.

Oncotarget. 2016;7(15):20520-20531.

7. Rossi D, Terzi-di-Bergamo L, De Paoli L, et al. Molecular prediction of durable remission after first-line fludarabine-cyclophosphamide-rituximab in chronic lymphocytic leukemia. Blood. 2015;126(16):19211924.

8. Brown JR, Hillmen P, O'Brien S, et al. Extended follow-up and impact of high-risk prognostic factors from the phase 3 RESONATE study in patients with previously treated CLL/SLL. Leukemia. 2018;32(1):83-91.

9. O'Brien S, Furman RR, Coutre S, et al. Single-agent ibrutinib in treatment-naïve and relapsed/refractory chronic lymphocytic leukemia: a 5year experience. Blood. 2018;131(17):1910-1919.

10. Hallek M, Cheson BD, Catovsky D, et al. iwCLL guidelines for diagnosis, indications for treatment, response assessment, and supportive management of CLL. Blood. 2018;131(25):2745-2760.

11. Rosenquist R, Ghia P, Hadzidimitriou A, et al. Immunoglobulin gene sequence analysis in chronic lymphocytic leukemia: updated ERIC recommendations. Leukemia. 2017;31(7):1477-1481.

12. Malcikova J, Tausch E, Rossi D, et al; European Research Initiative on Chronic Lymphocytic Leukemia (ERIC) - TP53 network. ERIC recommendations for TP53 mutation analysis in chronic lymphocytic leukemia-update on methodological approaches and results interpretation. Leukemia. 2018;32(5):1070-1080.

13. Dühren-von Minden M, Übelhart R, Schneider D, et al. Chronic lymphocytic leukaemia is driven by antigen-independent cell-autonomous signalling. Nature. 2012;489(7415):309-312.

14. Guarini A, Chiaretti S, Tavolaro S, et al. BCR ligation induced by IgM stimulation results in gene expression and functional changes only in IgV H unmutated chronic lymphocytic leukemia (CLL) cells. Blood.

2008;112(3):782-792.

15. Iacovelli S, Hug E, Bennardo S, et al. Two types of BCR interactions are positively selected during leukemia development in the Eμ-TCL1 transgenic mouse model of CLL. Blood. 2015;125(10):1578-1588.

16. Minici C, Gounari M, Übelhart R, et al. Distinct homotypic B-cell receptor interactions shape the outcome of chronic lymphocytic leukaemia. Nat Commun. 2017;8:15746.

17. 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.

18. Baliakas P, Moysiadis T, Hadzidimitriou A, et al. Tailored approaches grounded on immunogenetic features for refined prognostication in chronic lymphocytic leukemia. Haematologica. 2018 Sep 27. [Epub ahead of print] 19. Catovsky D, Wade R, Else M. The clinical significance of patients' sex in chronic lymphocytic leukemia. Haematologica. 2014;99(6):10881094.

20. Molica S, Giannarelli D, Mirabelli R, et al. Reliability of six prognostic models to predict time-to-first-treatment in patients with chronic lymphocytic leukaemia in early phase. Am J Hematol. 2017;92(6):E91-E93.

21. Fais F, Ghiotto F, Hashimoto S, et al. Chronic lymphocytic leukemia B cells express restricted sets of mutated and unmutated antigen receptors. J Clin Invest. 1998;102(():1515-1525.

22. Klein U, Tu Y, Stolovitzky GA, et al. Gene expression profiling of B cell chronic lymphocytic leukemia reveals a homogeneous phenotype related to memory B cells. J Exp Med. 2001;194(11):1625-1638.

23. Rosenwald A, Alizadeh AA, Widhopf G, et al. Relation of gene expression phenotype to immunoglobulin mutation genotype in B cell chronic lymphocytic leukemia. J Exp Med. 2001;194(11):1639-1647.

24. Ghia P, Stamatopoulos K, Belessi C, et al. Geographic patterns and pathogenetic implications of IGHV gene usage in chronic lymphocytic leukemia: the lesson of the IGHV3-21 gene. Blood. 2005;105(4):16781685.

25. Fischer K, Bahlo J, Fink AM, et al. Long-term remissions after FCR chemoimmunotherapy in previously untreated patients with CLL: updated results of the CLL8 trial. Blood. 2016;127(2):208-215.

26. Thompson PA, Tam CS, O’Brien SM, et al. Fludarabine, cyclophosphamide, and rituximab treatment achieves long-term disease-free survival in IGHV-mutated chronic lymphocytic leukemia. Blood.

2016;127(3):303-309.

27. Byrd JC, Furman RR, Coutre SE, et al. Targeting BTK with ibrutinib in relapsed chronic lymphocytic leukemia. N Engl J Med. 2013;369(1):3242.

221

Editorials

The stepchild in myeloma treatments: is allogeneic transplantation not so bad after all?

Antonia M.S. Müller,1 Shaji K. Kumar2 and Benedetto Bruno3 1

Department of Medical Oncology and Hematology, University Hospital Zurich, Switzerland; 2Division of Hematology, Mayo Clinic, Rochester, MN, USA and 3Department of Oncology/Hematology, A.O.U. Città della Salute e della Scienza di Torino, Presidio Molinette, Torino, Italy E-mail: benedetto.bruno@unito.it doi:10.3324/haematol.2018.206987

I

n this edition of Haematologica, two papers from the US and Germany report on the long-term follow up of patients with multiple myeloma (MM) treated with reduced intensity conditioning (RIC) and allogeneic hematopoietic cell transplantation (HCT).1,2 Given the armamentarium of highly effective agents and a pipeline of novel therapeutic options, including chimeric antigen receptor T (CAR-T) cells, many hemato-oncologists believe that there is no longer a role for allogeneic HCT in the treatment of MM. Despite remarkable improvements in outcomes with novel drugs, these advances have not yet translated into cure of the disease, even when combined with highdose chemotherapy and autologous HCT. Patients eventually relapse and die of their underlying disease. The two reports by Maffini et al. and Greil et al. both show that longterm survival, and potentially a cure, can actually be achieved in a proportion of myeloma patients by an allogeneic HCT, a treatment outcome that so far has not been observed with any other therapeutic strategy.

Maffini et al. present the long-term clinical outcomes of 244 patients who underwent allogeneic HCT following non-myeloablative conditioning with fludarabine and total body irradiation (TBI) between 1998 and 2016. In this study, more than half the patients had a chemotherapybased induction with vincristine-doxorubicin-dexamethasone (VAD), whereas after 2006 mostly immunomodulatory / proteasome inhibitor triplet regimens were given. The majority of patients (86%) received tandem autologousallogeneic treatment upfront, while 14% had failed previous autologous HCT. After high-dose melphalan and autologous HCT, 26% of patients were in a complete remission (CR), 19% in a very good partial remission (VGPR), 38% in a partial remission (PR), and 17% had progressive disease.

Best responses following allogeneic HCT were: CR in 46%, VGPR in 17%, PR in 20% of patients [overall response rate (ORR) 83%], and 17% failed to achieve a response. With a median follow up of 8.3 years (range, 1.0-18.1), 5-year overall survival (OS) and progression-free survival (PFS) rates were 54% and 31%, respectively, and 10-year OS and PFS rates were 41% and 19%, respectively. Non-relapse mortality was low with 2% at day +100 and 14% at five years, The rate of acute graft-versus-host disease (GvHD) was acceptable (33% grade II, 11% grade III / IV), while the cumulative incidence of chronic GvHD was 46%. The key findings of this study were: i) that patients with disease that was refractory to induction and those with high-risk biological features experienced shorter OS and PFS, while among standard-risk patients the median OS was not reached, and the median PFS was 6.5 years. High-risk patients experienced a median OS of 8.4 years with a PFS of 2.5 years; ii) patients who proceeded to tandem HCT after a previously

222

failed autologous HCT had poor outcomes with a median OS of 1.2 years and a median PFS of 0.4 years; iii) those patients who achieved negativity for minimal residual disease (MRD) had a significantly lower relapse rate as compared with MRD-positive (MRD+) patients, indicating that marrow sampling for MRD assessment post HCT is an important tool to guide treatment decisions.

Similar observations were made by Greil et al. who report on their single center experience of 109 consecutive patients who received fludarabine-based RIC preparation followed by allogeneic HCT between 2000 and 2017. With a median follow up of 71.5 months (6 years), the authors observed a high ORR of 70% (CR rate 42%), with a median PFS of 14.2 months (1.2 years) and a median OS of 39.2 months (3.3 years). Consistent with the findings of Maffini et al., survival was better in patients with sufficient response to induction therapy with a median OS of 65 months (5.4 years vs. 11.5 months in non-responders) and best in those undergoing allogeneic-HCT within first-line treatment (median OS not reached vs. 21.6 months in relapsed/refractory patients). Accordingly, the cumulative incidence of relapse was considerably lower in patients transplanted in first-line with 11% within the first year as compared to 50.3% after HCT for relapsed /refractory myeloma. Most relapses occurred within the first two years post HCT. Beyond five years, the survival curves appear to have reached a plateau with late relapses rarely occurring (10year OS 28.4% and 10-year PFS 24%). Also in this cohort, the rate of high-grade GvHD was moderate and the nonrelapse mortality low (8.4% within the first year, 12.4% at 10 years).

Both the studies by Maffini et al. and Greil et al. provide evidence that allogeneic HCT can induce graft-versusmyeloma activity that enables long-term disease control and survival (potentially even a cure) in selected myeloma patients. The following observations stand out and require further reflection and consideration.

1) Induction treatment in both trials was mostly chemotherapy-based. The majority of patients achieved “only” a partial response prior to autologous HCT. Both studies showed that the depth of response prior to HCT is critical, and that outcomes in patients who respond to induction are better when compared with non-responders, also after allogeneic HCT. Depth of response is a wellknown parameter that predicts clinical outcomes in myeloma, and in recent years MRD testing by PCR or flow cytometry has become available at many centers. A hallmark study underlining the importance of MRD monitoring was presented in a report on the long-term outcomes of molecular monitoring after a tandem “auto-non-myeloablative allo” approach.3 Twenty-six patients were prospectivehaematologica | 2019; 104(2)

Editorials

Figure 1. Proposed indication for allogeneic (allo)-hematopoietic cell transplantation (HCT) based on studies by Greil et al.,1 Maffini et al.,2 and other authors. 4,7,17 ASCT: autologous stem cell transplantation; DLI: donor lymphocyte infusion, CNI: calcineurin inhibitors; GVHD: graft-versus-host disease; VGPR: very good partial remission; CAVE: possible adverse effect.

ly evaluated by PCR. At a remarkable median follow up of 12.1 years, median OS and EFS were not reached in patients who achieved nested-PCR negativity while they were 3.3 and 1.5 years, respectively, in the remaining patients.

Besides the intuitive recognition that â&#x20AC;&#x153;responders do better than non-respondersâ&#x20AC;?, the biological relevance of achieving a deep response prior to allogeneic HCT, and maintaining this response for a certain time post HCT, is that the establishment of graft-versus-myeloma activity requires time. In both studies published in this edition of Haematologica, relapses mostly occurred early, during the first 1-2 years post allogeneic HCT, reflecting the ongoing balance between disease biology (aggressiveness) and effectiveness of graft-versus-myeloma activity. The implementation of novel agents not only prior to HCT, but also post HCT, likely results in deeper responses prior to HCT and can help to maintain these responses post allogeneic HCT. Such strategies may translate into lower relapse rates during the first year, and thereby provide time for graft-versusmyeloma effects to establish, particularly once patients are off immunosuppression.

2) Both trials clearly indicate that there is anti-myeloma activity in some patients, as more than 20% were long-term survivors beyond ten years post HCT. Patients who relapsed post HCT appeared to have a relatively long survival, which is consistent with two recently published reports on long-term follow up post autologous versus autologous-allogeneic tandem HCT, in which patients post allografting had a significantly longer OS compared with those with relapse post autologous HCT.4,5 The question not answered by the studies published here is whether and how anti-myeloma activity of donor T cells could be enhanced in vivo, e.g. by donor lymphocyte infusions (DLI), haematologica | 2019; 104(2)

immunomodulatory drugs (IMIDs), proteasome inhibitors, monoclonal antibodies, etc. In both studies, not all patients had been treated with novel myeloma drugs as induction or post HCT, in the latter context, mostly for treatment of active disease. Moreover, at relapse, many patients received drugs outside the context of clinical trials depending on which one was readily available at that time. Defined subgroups in available studies were thus too small to provide statistically significant results regarding the impact of novel therapeutics in general or the influence of a specific drug.

Newer data show that the application of post-HCT bortezomib is feasible, safe, and effective even in heavily pretreated, poor-risk patients.6,7 IMIDs, in contrast, resulted in higher toxicity, acute GvHD, and early discontinuation in one trial,8 whereas in other trials, lenalidomide was given at lower doses and tolerability was good.9,10 3) The current major cause of treatment failure after allografting is disease relapse, not treatment-related mortality.

In contrast to early trials, today, under appropriate standard care, transplant procedures are associated with low toxicity and GvHD rates are acceptable. In the early 2000s, a number of trials introduced the tandem approach with an autograft for tumor debulking followed by reduced-intensity or non-myeloablative allogeneic HCT, a strategy that was able to lower the toxicity of the regimens. Yet, even today, allogeneic HCTs still bear the negative connotation of high toxicity, morbidity, treatment-related mortality, and a substantial negative impact on the quality of life. Interestingly, using the revised Myeloma Comorbidity Index (R-MCI), an established tool and prognostic instrument for risk prediction in myeloma patients that evaluates renal and lung function, Karnofksy Performance Status impairment, frailty, and age,11 Greil et al. showed, that over time, the R-MCI 223

Editorials

declined through treatment, indicating that performance status, and accordingly quality of life, was improved by treating the underlying disease. In those patients whose condition deteriorated, such a deterioration was from decreasing renal function and increasing age, and only in a minority was this due to complications from the allogeneic HCT, such as chronic GvHD. Six prospective trials examined the role of allografting compared with autologous HCT alone.12-21 Substantial differences in inclusion criteria and treatment schemas partly contributed to conflicting outcomes. While most of these trials demonstrated an improved PFS in the allogeneic cohort, in only two studies did this response also translate into a longer OS. Similarly, a meta-analysis of published clinical trials containing 1192 newly diagnosed patients who received tandem auto-auto and 630 who underwent tandem auto-non-myeloablative allogeneic HCT showed that the CR rates were higher in the auto-allo group, but there was no survival advantage in the first three years.22 Of note, the survival advantage in the auto-allo group, reported in two of the published comparative studies, became statistically significant after a follow up of at least three years. All these studies were conducted prior to the routine implementation of novel drugs into induction therapy; treatment of relapsed disease following HCT varied and was not taken into consideration when analyzing survival rates.

Today, there is remarkable heterogeneity in the use of allogeneic HCTs for patients with myeloma among different countries, and even institutions, and few ongoing clinical trials are studying how to improve allogeneic HCT strategies in myeloma or clarify its role. Trends in the use of HCT in myeloma were published in a large evaluation of the European Group of Blood and Marrow Transplantation (EBMT) over a 25-year period that examined a total of 3405 myeloma patients given allogeneic HCT either as an upfront treatment, within an upfront autologous-allogeneic HCT concept or as salvage treatment after a failed autologous HCT.23 It was demonstrated that allogeneic HCT is currently mostly used as salvage therapy for myeloma patients after at least one autograft rather than within an intensified upfront induction-auto-allo approach. Similar to the studies by Maffini et al. and Greil et al., the EBMT analysis demonstrated that OS rates in the first few years following allogeneic HCT were comparable with novel induction strategies involving new drugs and high-dose chemotherapy with autologous HCT; however, also in the EBMT dataset, long-term survival was observed in more than 20% of patients given allogeneic HCT. Again, patients receiving allogeneic HCT within an upfront auto-allo tandem approach achieved better outcomes, but even later transplants, usually in progression or relapse following autologous HCT, resulted in encouraging long-term outcomes with 25% survivor rates at ten years.

In the light of the studies by Maffini et al. and Greil et al., and those by other groups, the abandonment of allografting, as some have suggested, appears rather premature even in newly diagnosed myeloma patients. The two reports draw particular attention to the long-term follow up with potential cure in subsets of patients. Reasons to re-examine the role of allogeneic HCT for patients with myeloma in controlled clinical trials are as follows. â&#x20AC;˘ Today, validated tools are available to identify patients 224

at high risk in whom OS and PFS are very poor even in the era of new drugs, including the R-MCI,11 the revised International Staging System by the International Myeloma Working Group,24 cytogenetics, etc. For this patient subset, new effective treatments are urgently needed. The negative prognostic impact of high-risk cytogenetics appeared to be partly neutralized by graft-versus-myeloma activity in two recent studies.15,25 â&#x20AC;˘ Patient preparation, including the achievement of a deep response prior to HCT, has significantly improved over the past decade by incorporating second and third generation proteasome inhibitors, IMIDs, monoclonal antibodies, and histone deacetylase inhibitors. Patients may, therefore, be in better condition (as many of them have not been exposed to toxic chemotherapy), have improved organ function, and be in deeper remission of the disease prior to HCT. These factors all optimize the baseline setting for an allogeneic HCT, reducing treatment-related toxicity, morbidity, mortality, and providing time for graft-versus-myeloma activity to be established. â&#x20AC;˘ The role of the combination of new drugs with graftversus-myeloma activity has never been systematically explored in this incurable disease. This may partly be attributed to the limited current interest on cell therapy strategies in myeloma. New drugs and graft-versus-myeloma activity are not mutually exclusive and their synergy has clearly been shown in relapsed patients.

In conclusion, despite the recent dramatic improvement in survival, the overwhelming majority of myeloma patients invariably relapse. Given the potentially curative effect of graft-versus-myeloma activity, the role of allografting should become a matter of sound scientific debate in the myeloma community. Combinations of allografts with potent anti-myeloma agents pre- and post-HCT should be examined in young high-risk and/or early relapsed patients for whom life expectancy is currently very poor. Modern MRD monitoring tools may guide individual treatment decisions and thereby further improve long-term outcomes.

References 1. Greil C, Engelhardt M, Ihorst G, et al. Allogeneic transplantation of multiple myeloma patients may allow long-term survival in carefully selected patients with acceptable toxicity and preserved quality of life.

Haematologica. 2018;104(2):370-379.

2. Maffini E, Storer BE, Sandmaier BM, et al. Long term follow-up of tandem autologous-allogeneic hematopoietic cell transplantation for multiple myeloma. Haematologica. 2018;104(2):380-391.

3. Ladetto M, Ferrero S, Drandi D, et al. Prospective molecular monitoring of minimal residual disease after non-myeloablative allografting in newly diagnosed multiple myeloma. Leukemia. 2016;30(5):1211-1214.

4. Giaccone L, Evangelista A, Patriarca F, et al. Impact of New Drugs on the Long-Term Follow-Up of Upfront Tandem Autograft-Allograft in Multiple Myeloma. Biol Blood Marrow Transplant. 2018;24(1):189193.

5. Htut M, D'Souza A, Krishnan A, et al. Autologous/Allogeneic Hematopoietic Cell Transplantation versus Tandem Autologous Transplantation for Multiple Myeloma: Comparison of Long-Term Postrelapse Survival. Biol Blood Marrow Transplant. 2018;24(3):478485.

6. Caballero-Velazquez T, Lopez-Corral L, Encinas C, et al. Phase II clinical trial for the evaluation of bortezomib within the reduced intensity conditioning regimen (RIC) and post-allogeneic transplantation for high-risk myeloma patients. Br J Haematol. 2013;162(4):474-482.

7. Green DJ, Maloney DG, Storer BE, et al. Tandem autologous/allogeneic hematopoietic cell transplantation with bortezomib maintenance therapy for high-risk myeloma. Blood Adv. 2017;1(24):2247-2256.

haematologica | 2019; 104(2)

Editorials

8. Kneppers E, van der Holt B, Kersten MJ, et al. Lenalidomide maintenance after nonmyeloablative allogeneic stem cell transplantation in multiple myeloma is not feasible: results of the HOVON 76 Trial.

Blood. 2011;118(9):2413-2419.

9. Alsina M, Becker PS, Zhong X, et al. Lenalidomide maintenance for high-risk multiple myeloma after allogeneic hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2014;20(8):1183-1189.

10. Kroger N, Zabelina T, Klyuchnikov E, et al. Toxicity-reduced, myeloablative allograft followed by lenalidomide maintenance as salvage therapy for refractory/relapsed myeloma patients. Bone Marrow Transplant. 2013;48(3):403-407.

11. Engelhardt M, Domm AS, Dold SM, et al. A concise revised Myeloma Comorbidity Index as a valid prognostic instrument in a large cohort of 801 multiple myeloma patients. Haematologica. 2017;102(5):910-921.

12. Bjorkstrand B, Iacobelli S, Hegenbart U, et al. Tandem autologous/reduced-intensity conditioning allogeneic stem-cell transplantation versus autologous transplantation in myeloma: long-term follow-up. J Clin Oncol. 2011;29(22):3016-3022.

13. Bruno B, Rotta M, Patriarca F, et al. Nonmyeloablative allografting for newly diagnosed multiple myeloma: the experience of the Gruppo Italiano Trapianti di Midollo. Blood. 2009;113(14):3375-3382.

14. Bruno B, Rotta M, Patriarca F, et al. A comparison of allografting with autografting for newly diagnosed myeloma. N Engl J Med. 2007;356(11):1110-1120.

15. Gahrton G, Iacobelli S, Bjorkstrand B, et al. Autologous/reduced-intensity allogeneic stem cell transplantation vs autologous transplantation in multiple myeloma: long-term results of the EBMT-NMAM2000 study. Blood. 2013;121(25):5055-5063.

16. Garban F, Attal M, Michallet M, et al. Prospective comparison of autologous stem cell transplantation followed by dose-reduced allograft (IFM99-03 trial) with tandem autologous stem cell transplantation (IFM99-04 trial) in high-risk de novo multiple myeloma. Blood.

2006;107(9):3474-3480.

17. Giaccone L, Storer B, Patriarca F, et al. Long-term follow-up of a comparison of nonmyeloablative allografting with autografting for newly

haematologica | 2019; 104(2)

diagnosed myeloma. Blood. 2011;117(24):6721-6727.

18. Krishnan A, Pasquini MC, Logan B, et al. Autologous haemopoietic stem-cell transplantation followed by allogeneic or autologous haemopoietic stem-cell transplantation in patients with multiple myeloma (BMT CTN 0102): a phase 3 biological assignment trial.

Lancet Oncol. 2011;12(13):1195-1203.

19. Moreau P, Garban F, Attal M, et al. Long-term follow-up results of IFM99-03 and IFM99-04 trials comparing nonmyeloablative allotransplantation with autologous transplantation in high-risk de novo multiple myeloma. Blood. 2008;112(9):3914-3915.

20. Rosinol L, Perez-Simon JA, Sureda A, et al. A prospective PETHEMA study of tandem autologous transplantation versus autograft followed by reduced-intensity conditioning allogeneic transplantation in newly diagnosed multiple myeloma. Blood. 2008;112(9):3591-3593.

21. Rotta M, Storer BE, Sahebi F, et al. Long-term outcome of patients with multiple myeloma after autologous hematopoietic cell transplantation and nonmyeloablative allografting. Blood. 2009;113(14):3383-3391.

22. Armeson KE, Hill EG, Costa LJ. Tandem autologous vs autologous plus reduced intensity allogeneic transplantation in the upfront management of multiple myeloma: meta-analysis of trials with biological assignment. Bone Marrow Transplant. 2013;48(4):562-567.

23. Sobh M, Michallet M, Gahrton G, et al. Allogeneic hematopoietic cell transplantation for multiple myeloma in Europe: trends and outcomes over 25 years. A study by the EBMT Chronic Malignancies Working Party. Leukemia. 2016;30(10):2047-2054.

24. Scott EC, Hari P, Kumar S, et al. Staging Systems for Newly Diagnosed Myeloma Patients Undergoing Autologous Hematopoietic Cell Transplantation: The Revised International Staging System Shows the Most Differentiation between Groups. Biol Blood Marrow Transplant.

2018;24(12):2443-2449.

25. Kroger N, Badbaran A, Zabelina T, et al. Impact of high-risk cytogenetics and achievement of molecular remission on long-term freedom from disease after autologous-allogeneic tandem transplantation in patients with multiple myeloma. Biol Blood Marrow Transplant.

2013;19(3):398-404.

225

REVIEW ARTICLE Ferrata Storti Foundation

CD30-positive primary cutaneous lymphoproliferative disorders: molecular alterations and targeted therapies

Lucia Prieto-Torres,1 Socorro M. Rodriguez-Pinilla,2,3 Arantza Onaindia,4 Mariano Ara,5 Luis Requena1 and Miguel Á. Piris2,3

Haematologica 2019 Volume 104(2):226-235

Departments of 1Dermatology, 2Pathology, Hospital Universitario Fundación Jiménez Díaz, Madrid; 3 Hospital Universitario Fundación Jiménez Díaz, Madrid, CIBERONC, Madrid, 4 Pathology, Hospital Universitario Marques de Valdecilla, Santander and 5Dermatology Hospital Clínico Universitario Lozano Blesa, Zaragoza, Spain

ABSTRACT

P

rimary cutaneous CD30-positive T-cell lymphoproliferative disorders are the second most common subgroup of cutaneous T-cell lymphomas. They include two clinically different entities with some overlapping features and borderline cases: lymphomatoid papulosis and primary cutaneous anaplastic large cell lymphoma. Molecular studies of primary cutaneous anaplastic large cell lymphoma reveal an increasing level of heterogeneity that is associated with histological and immunophenotypic features of the cases and their response to specific therapies. Here, we review the most significant genetic, epigenetic and molecular alterations described to date in primary cutaneous CD30-positive T-cell lymphoproliferative disorders, and their potential as therapeutic targets.

Correspondence:

Introduction

lucia14_prie@msn.com

Received: September 19, 2018.

Accepted: December 7, 2018.

Pre-published: January 10, 2019.

doi:10.3324/haematol.2018.197152 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/226 ©2019 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.

226

CD30+ primary cutaneous T-cell lymphoproliferative disorders are the second most common subgroup of cutaneous T-cell lymphomas after mycosis fungoides (MF), accounting for approximately 30% of cases.1 These cutaneous lymphomas have customarily been classified on the basis of their clinical presentation into lymphomatoid papulosis (LyP), primary cutaneous anaplastic large cell lymphoma (pcALCL) and borderline cases. In recent years, genomic analysis has become important for the diagnosis and clinical management of patients affected by systemic and cutaneous hematologic malignancies.2 Systemic anaplastic large cell lymphoma (ALCL) is defined by mutually exclusive rearrangements of ALK, DUSP22/IRF4 and TP63, which have prognostic and survival implications and must be taken into account in the management of patients.2 pcALCL have histopathological and immunophenotypic similarities with systemic ALCL, but have a more indolent behavior. Chromosomal translocations described in systemic ALCL can also be seen in pcALCL, although they have different clinical implications. Thus, there are some ALK+ pcALCL, and some cases of LyP and pcALCL share a DUSP22-IRF4 locus translocation. Bearing all this in mind, we have reviewed the molecular alterations in CD30+ primary cutaneous T-cell lymphoproliferative disorders, describing the various molecular alterations and considering their clinical and therapeutic implications.

Lymphomatoid papulosis LyP is an enigmatic disease that follows the course of a chronic skin condition and has the histology of a lymphoma. It typically has a recurrent, self-healing course, with an excellent prognosis.3 Clinical features of all types of LyP are similar and consist of papular, papulonecrotic and/or nodular skin lesions at different stages of evolution. The number of lesions is, however, highly variable, ranging from only a few lesions to hundreds. Likewise, there is great variability in the duration of lesions, which may be present for a few weeks or persist for decades. Lyp haematologica | 2019; 104(2)

CD30+ primary cutaneous LPD

is seen more frequently in adult patients, but children can also be affected.4 Customarily, on the basis of its extremely variable histopathology, LyP has been divided into five types with similar prognosis, although distinguishing them is important for the differential diagnosis from more aggressive types of lymphoma.5 Although more descriptive terms have been proposed, in 2017 the World Health Organization (WHO) categorized LyP using consecutive alphabetical letters.6 Type A is the most frequent form of LyP, accounting for 80% of cases. Tumor cells are typically CD4+ and CD30+ and appear scattered or in small clusters, accompanied by numerous inflammatory cells, including neutrophils, eosinophils and small lymphocytes. The main differential diagnoses include reactive lesions, such as insect bites, and pityriasis lichenoides et varioliformis acuta (PLEVA).7 Type B is uncommon, accounting for 5% of cases, and has the same CD4+, CD8– immunophenotype.7 It has a histology similar to that of plaque-stage MF with an epidermotropic infiltrate of small, atypical CD30+ cells, which is its main differential diagnosis; less frequently it must be distinguished from cutaneous epidermotropic gamma/delta lymphoma.5 Type C makes up around 10% of LyP cases and has a histology very similar to that of pcALCL, with a nodular cohesive infiltrate of large CD30+, CD4+, CD8– pleomorphic and anaplastic tumor cells featuring mitotic figures and abundant cytoplasm.7 Apart from pcALCL, other entities, such as transformed MF, peripheral T-cell lymphoma not otherwise specified, and adult T-cell lymphoma/leukemia, may have a similar histology.5 Types D and E have only been described relatively recently, and are usually characterized by a cytotoxic phenotype, with CD8+ and CD30+ lymphocytes.

Biopsies from patients with type D LyP show prominent epidermotropism of atypical small-to-medium-sized pleomorphic cells. There may be deep dermal and perivascular infiltrates. This variant accounts for about 5% of cases and needs to be differentiated from pagetoid reticulosis, a peculiar CD8+ form of MF, from more aggressive lymphomas such as primary cutaneous aggressive epidermotropic CD8+ cytotoxic T-cell lymphoma, and from cutaneous gamma/delta lymphoma.8 Accounting for fewer than 5% of cases, type E LyP shows more extensive necrosis and ulceration due to angiocentric and angiodestructive infiltrates of mostly medium-sized, pleomorphic CD8+ and CD30+ lymphocytes with hemorrhage, vascular occlusion and thrombi, admixed with some eosinophils.9 Although clinically indolent, the histology can be confused with that of extranodal NK/T-cell lymphoma, nasal type, cutaneous gamma/delta lymphoma or ALCL (primary cutaneous or systemic form) with angiocentric and angiodestructive growth. It is important to highlight that histological differential diagnoses of LyP (such as aggressive epidermotropic CD8+ cytotoxic T-cell lymphoma or MF) must be excluded by clinicopathological correlation based on characteristic clinical grounds with the typical "waxing and waning" presentation of LyP. Recently, the detection of rearrangements of the DUSP22-IRF4 locus on chromosome 6p25.3 has enabled the identification of a new molecular-based type of LyP with a characteristic histological pattern.10

Lymphomatoid papulosis with 6p25.3 rearrangements This molecular alteration at the DUSP22-IRF4 locus is less frequent in LyP than in pcALCL and accounts for haematologica | 2019; 104(2)

fewer than 5% of cases of LyP. Typically, patients are older than those with other forms of LyP, and their lesions are characterized by a biphasic histological pattern showing, on the one hand, extensive epidermotropism with CD30+ small-to-medium-sized T-lymphocytes that simulate pagetoid reticulosis lesions and, on the other, a dermal neoplastic infiltrate composed of large T cells with strong CD30 positivity.11

T-cell receptor clonality Detection of T-cell receptor clonality differs significantly between LyP types.12 This difference might be related to the number of tumor cells present in the infiltrate, which is lower in type A than in type C, hindering the detection technique in paraffin-embedded tissues. LyP presents atypical evolution for a peripheral T-cell lymphoma, exhibiting self-healing lesions and multiple outbreaks. Tcell receptor clonality studies have been performed in various LyP lesions to investigate the potential role of foreign antigens and the relationship between LyP and other cutaneous T-cell lymphomas in cases with overlapping histological findings. Chott et al. showed that multiple skin lesions and associated T-cell lymphomas (MF and ALCL) were clonally related in most LyP patients, which suggests that a non-random genetic event initiates the disease.13 Shared clonality has been confirmed for LyP and MF lesions occurring in the same patients by de la Garza Bravos et al.14 T-cell receptor-γ expression is considered a feature of primary cutaneous gamma-delta T-cell lymphoma, and is rare in other types of primary cutaneous lymphoma.

However, there are cases of type D LyP with a cytotoxic phenotype and T-cell receptor-γ expression that, unlike in primary cutaneous gamma-delta T-cell lymphoma, is not associated with worse prognosis.15

Other findings in lymphomatoid papulosis SATB1 (special AT-rich sequence-binding protein 1) is an important thymocyte nuclear protein, a chromatin organizer that is crucial to the development of T-lymphocytes.16,17 SATB1 plays a role in inducing resistance to the death of Sézary cells and has been implicated in the pathogenesis of the leukemic form of cutaneous T-cell lymphoma-Sézary syndrome.18 Sun et al. investigated its expression in a large cohort of patients with CD30+ lymphoproliferative disorders, and studied the potential for its use in classifying CD30+ lymphoproliferative disorders with differential clinicopathological behaviors and prognoses.19 They identified SATB1 expression in anaplastic T cells in 91.7% of LyP cases and in 38.1% of pcALCL cases.

SATB1 cases showed T-helper 17 polarization with expression of T-helper 17 cytokines and repressed Thelper 1-related genes.19 They also described a better response to methotrexate and interferon treatment in cases with high levels of SATB1 expression. Furthermore, these cases showed more prominent epidermal hyperplasia and granulocytic infiltration.19 Sun et al. postulated that the variability of SATB1 expression in CD30+ lymphoproliferative disorders could be related to the extent of DNA methylation.19 Tumor necrosis factor receptor (TNFR)-associated factor 1 (TRAF1) is involved in intracellular signal transduction of a range of TNFR, including CD30 and those associated with nuclear factor-kB activation.20 Assaf et al. studied the expression of TRAF1 using one antibody that recognized 227

L. Prieto-Torres et al.

a formalin-resistant epitope (Ber-TRAF1A). They found strong TRAF1 expression in the tumor cells in most LyP cases, in contrast to tumor cells of primary and secondary pcALCL, in which TRAF1 expression was much more restricted.21 Benner et al. also studied this marker in conjunction with MUM1, Bcl2 and CD15, but found it to be of no prognostic or diagnostic utility.22 We review the role of apoptosis in LyP pathogenesis in greater depth in the section on pcALCL, below.

Mahtas et al. studied gene deregulation and spatial genome reorganization near the ALCL translocation breakpoint in ALCL. They described the aberrant expression of Fra2 and of Id2 in LyP, the latter being consistent with the findings of Cotta et al.23,24 They found a lower level of expression in LyP compared with systemic ALCL and argued that gene dosage could be involved in the invasiveness and progression of LyP.23 P53 mutations have rarely been found in LyP. Kapur et al. reported two cases of LyP with P53 mutations from an analysis of 11 exons of the P53 gene. They found P53 mutations to be infrequent in cutaneous CD30+ lymphoproliferative disorders, and the two patients with LyP who harbored the mutation did not show any changes in the clinical behavior of the disease.25 Notch1 expression has been identified in LyP tumor cells, in which it is associated with the expression of the Notch1 ligand Jagged1, but not of the Delta1 ligand, which was expressed at low or negligible levels.26 The levels of expression of Notch1 and Jagged1 were higher in pcALCL samples, a finding that is discussed in greater detail in the section on pcALCL, below.

To date, no ALK fusions have been reported in LyP, in contrast to pcALCL, and no TP63 rearrangements have been found in these patients.25,27,28

Primary cutaneous anaplastic large cell lymphoma pcALCL is, by definition, a CD30+ large T-cell neoplasm composed of large cells with an anaplastic, pleomorphic or immunoblastic morphology. The CD30 antigen is expressed in more than 75% of tumor cells. pcALCL resembles other forms of ALCL but arises primarily in the skin.1 The clinical course of pcALCL differs from that of systemic forms of ALCL, both ALK+ and ALK-, which explains why it has been classified as a distinct category.29 However, there is some overlap between systemic and primary cutaneous forms of ALCL, whereby they share some molecular alterations, suggesting that other genetic and biological differences are likely to exist and therefore need to be identified.30 Primary cutaneous large T-cell lymphoma may also be the result of MF tumor progression. Thus, in patients with pcALCL, a current or previous diagnosis of MF must be excluded. The differential diagnosis between pcALCL and transformed CD30+ MF may be challenging, except when there is a clinical presentation with a previous or simultaneous patch-plaque stage MF lesion. Genetic differences between pcALCL and transformed MF have been found using array-based comparative genomic hybridization.31 From a clinical point of view, patients with pcALCL typically present with solitary or localized nodules or tumors or, more unusually, papules, with frequent ulceration and rapid evolution in some cases that may simulate aggressive lymphomas. The presence of multiple lesions in 20% 228

of cases can hinder the differential diagnosis with type C LyP, which features borderline lesions.7 This cutaneous lymphoma occurs predominantly in males (male:female ratio 3-2:1). It is more frequent in people in their sixth decade, but may also appear in childhood.32 It has been reported that pcALCL is a common form of cutaneous lymphoma in immunosuppressed patients, such as individuals infected with human immunodeficiency virus and organ transplant recipients.33,34 However, in contrast to what usually happens in patients with B-cell lymphomas with CD30+ large cells, especially in immunosuppressed patients, expression of Epstein-Barr virus by the tumor cells is extremely rare or absent in pcALCL with the T/null cell phenotype.32 When it does appear, it is essential to rule out a diagnosis of a B-cell lymphoma with T-cell markers and CD30 expression, such as plasmablastic lymphoma or primary effusion lymphoma.

There is extracutaneous involvement in about 10% of cases, usually with infiltration of locoregional lymph nodes. In these cases it is important to establish the sequence of presentation in order to rule out cutaneous involvement by systemic ALCL, which has an entirely different prognosis. Characteristically, locoregional lymph node involvement is not related to bad prognosis in pcALCL. Clinical presentation with extensive skin lesions on legs or arms is the only risk factor associated with a statistically significantly worse prognosis.7,35,36 The classic histological pattern described in pcALCL consists of a circumscribed nodular infiltrate that is mostly dermal, composed of arranged large lymphoid cells and usually with absent or subtle epidermotropism5 (Figure 1).

However, several variants of this histological pattern have been described, some of which are related to molecular findings described later in this review.5 Neutrophils and eosinophils are usually scattered in classic forms, although rich variants, which are usually more common in immunosuppressed patients, have been reported. The presence of this rich granulocytic infiltrate may be explained by the release of interleukin-8, whose levels are elevated in cultured tumor cells and in the serum of these patients.37 Other unusual histological presentations of pcALCL have been reported, including angiocentric or angiodestructive forms,38,39 subcutaneous and keratoacanthoma-like forms,40 sarcomatoid variants with prominent spindle-cell morphology,41 small cell variants that are rarer than in systemic forms of ALK+ ALCL,42 and an intravascular ALCL that may involve the skin and must be distinguished from the more common intralymphatic spread of tumor cells in pcALCL, which seems to have no prognostic implications.43,44 pcALCL cells usually carry an activated T-cell phenotype in which CD30 is expressed in more than 75% of tumor cells (the hallmark of the disease) (Figure 1), frequently accompanied by CD3, CD4 and CD45RO expression and varying degrees of loss of CD5 and CD2.

The expression of other activation markers, such as CD71, HLA-DR and CD25 (IL-2R), has been noted in approximately half of the cases. Cytotoxic markers (TIA1, perforin and granzyme B) and cutaneous lymphocyte antigen (CLA, HECA-452) may also be found in around half of the patients.45 In contrast to systemic ALKâ&#x20AC;&#x201C; ALCL, in which approximately 43% of cases are EMA+,46 EMA is often negative or only focally positive in pcALCL. Interestingly, pcALCL shares with transformed MF and SĂŠzary syndrome the expression of KIR3DL2 haematologica | 2019; 104(2)

CD30+ primary cutaneous LPD

(CD158k) by neoplastic cells, a potential target reviewed in the next section.47 Altered expression of the T-cell receptor/CD3 complex, T-cell receptor-associated transcription factors and signal transduction molecules is a common feature of systemic and cutaneous CD30+ lymphoproliferations, a finding that may be of potential diagnostic utility.48 Monoclonal rearrangement of the TCR gene occurs in 65% to 90% of cases of pcALCL in contrast with a much lower frequency in LyP types A and B.12,49

Genetic variants of primary cutaneous anaplastic large cell lymphoma pcALCL seems to carry the same molecular alterations as systemic ALCL, but with dramatic differences in frequencies. From the molecular point of view, most pcALCL cases would currently be classified as triple-negative (ALK-, DUSP22-, TP63-negative), since they do not carry any of these precise underlying molecular alterations. However, these lymphomas have an entirely different prognosis from their systemic counterparts, with some of them

A

B

D

ALK-negative primary cutaneous anaplastic large cell lymphoma with DUSP22 rearrangements DUSP22 rearrangements in systemic ALK– ALCL have been associated with a better prognosis, similar to that of ALK+ systemic ALCL, compared with TP63-translocated or triple-negative cases.50,51 About 20% of pcALCL cases harbor the DUSP22-IRF4 translocation, with no correlation with MUM1/IRF4 protein expression.11,52,53 First reported in LyP, a particular biphasic histological pattern has been linked to the presence of DUSP22 rearrangements in pcALCL (Figure 2). This pattern is characterized

A

B

C

D

E

F

C

E

Figure 1. “Classic”, ALK-, DUSP22-, TP63- (triple negative) primary cutaneous anaplastic large cell lymphoma. (A) Clinical picture showing two adjacent tumoral erythematous nodules located in the scapular region simulating dermatofibrosarcoma protuberans. (B,C) (x40), Hematoxylin & eosin stain showing the dermal infiltration consisting of a circumscribed infiltrate, composed of arranged large lymphoid cells with absent or subtle epidermotropism. (D,E) (x40), CD30 stain showing positivity in the membrane and Golgi of the tumoral large cells.

haematologica | 2019; 104(2)

showing spontaneous regression and an absence of progression, with long-term survival in the vast majority of the cases (Figure 1). The molecular alterations present in pcALCL are, in descending order of frequency: (i) DUSP22 rearrangements (Figure 2);11,50-61 (ii) ALK translocations (Figure 3);62-65 (iii) TP63 rearrangements;27,66 and (iv) NPM1TYK2 gene fusion.67-69 ALK translocations are much less frequent than in systemic lymphomas, having been described in isolated cases or very small series.63-65 TP63 rearrangements and NPM1-TYK2 gene fusion are exceptional and only isolated cases have been reported.27,67-69

Figure 2. ALK-negative primary cutaneous anaplastic large cell lymphoma with DUSP22 rearrangements. (A) Clinical picture of a large ulcerated tumor in the left malar region of an elderly woman showing (B) the typical CD30 staining of a case of pcALCL with DUSP22 rearrangement (Ref. 58, courtesy of Onaindia et al.), evidencing extensive epidermotropism with CD30+ small-to-mediumsized T-lymphocytes that simulate pagetoid reticulosis lesions. (C) Hematoxylin & eosin-stained panoramic view of the case in (E) and (F) showing the tumor nodule with profuse dermal involvement. (D) Detail, stained with hematoxylin & eosin, of another case exemplifying the biphasic epidermal and dermal lymphocytic infiltrate. (E,F) (x40), CD3 stain highlights the intraepidermic neoplastic T cells in another case with DUSP22 rearrangement.

229

L. Prieto-Torres et al.

by the simultaneous presence of dermic CD30+ large cells and intraepidermal infiltrate by CD30+ small cells with a pagetoid reticulosis pattern.10,54 The clinical behavior and prognosis of cases with DUSP22 rearrangements is similar to that of cases without such rearrangements.55 Currently, clinical presentation and staging remain essential to differentiate cases of pcALCL with DUSP22 rearrangement from LyP with this molecular alteration, because the two entities have similar histological features.54 DUSP22 rearrangements have also been described in rare cases of CD30-rich tumoral MF.56 Expression of the chemokine receptor gene CCR8 is associated with DUSP22 rearrangements in ALCL.57 It has been proposed that the higher level of expression of this skinhoming chemokine receptor may explain the lower tendency of pcALCL to disseminate to extracutaneous sites.58 The pathways activated after DUSP22-IRF4 rearrangements in peripheral T-cell lymphomas have recently been investigated by Mélard et al.59 They found a tumor suppressor function for DUSP22, with the restored expression of DUSP22 promoting apoptosis and impairing soft agar clonogenicity.59 Negative regulation of the interleukin6/leukemia inhibitory factor/signal transduction and activator of transcription (STAT)-3 pathway has been found in

experimental studies.60 Consistent with these findings, DUSP22 (also known as JKAP) knockout mice develop inflammation and autoimmunity.61

ALK-positive primary cutaneous anaplastic large cell lymphoma Cutaneous involvement in cases of systemic ALK+ ALCL is generally indicative of a bad prognosis (relative risk: 2) and occurs in 26% of pediatric ALK+ ALCL cases.62 In contrast, ALK+ pcALCL cases seem to have a more favorable outcome. Only a few cases of ALK+ pcALCL have been reported, the largest series being that described by Oschlies et al.,63 in which six pediatric patients with ALK+ pcALCL had a favorable clinical course, comparable to that of patients with ALK– pcALCL. Other ALK+ primary cutaneous cases occurred in adults and had peculiar features, such as cytoplasmic ALK expression and a lack of ALK translocation.64 To date, no ALK fusion partners exclusive to the primary cutaneous cases have been reported. It is important to stage patients with ALK+ pcALCL carefully because cutaneous lesions triggered by insect bites can be the first manifestation of systemic ALK+ ALCL.65 The authors postulated that insect bite-associated antigens may attract T lymphocytes to the skin, some of which bear the ALK translocation t(2;5). The subsequent release of different cytokines at the site of the bite could act as a “second hit” and activate T cells, which may express the oncogenic NPM-ALK protein and initiate deregulated growth.65

Primary cutaneous anaplastic large cell lymphoma with TP63 rearrangements

A

B

C

D

In 2012, Vasmatzis et al. described two of 19 pcALCL that carried TP63 rearrangements,27 both of which had an unusual, aggressive clinical behavior, analogous to that of systemic ALCL with TP63 rearrangements.51 Subsequent work by the same group also identified a case of aggressive MF with a TP63 translocation.66 In both series, TP63 translocations were associated with strong TP63 protein expression detectable by immunohistochemistry, but this was not a specific finding since the protein was also expressed in cases without a TP63 translocation. Schrader et al. reviewed a series of 17 cases of aggressive LyP and pcALCL. They found no cases with a TP63 translocation, and confirmed the lack of specificity of the TP63 immunohistochemistry.28

NPM1-TYK2 gene fusion and oncogenic STAT3 activation

E

G

F

H

Figure 3. ALK-positive primary cutaneous anaplastic large cell lymphoma. (A) Clinical picture of a large ulcerated tumor located on the back. (B) Fluorescence in situ hybridization image showing an ALK reciprocal translocation with gain of one or two copies of the ALK gene. (C,D) Hematoxylin & eosin stain showing the morphology of large T cells (E,F) (x40), with CD30 positivity and (G,H) (x40), nuclear and cytoplasmic ALK positivity in tumor T cells.

230

Mutations on the JAK1/STAT3 pathway are common in systemic ALK– ALCL.67,68 However, these mutations have been found in only 5% of pcALCL, which is further evidence of the distinct molecular pathogenesis of the two entities. Whole-transcriptome sequencing done in pcALCL revealed the NPM1-TYK2 gene fusion, which encodes a protein containing an intact catalytic domain in TYK2 and an oligomerization domain of NPM. It was present in two (1 LyP, 1 pcALCL) of the cases of primary cutaneous CD30+ lymphoproliferative disorders studied (n=47) and was not found in other mature T-cell neoplasms (n=151). Both cases showed nuclear STAT5 expression. This defective kinase activates the STAT signaling pathway (STAT1/3/5), implying that TYK2 could be a therapeutic target in this subset of patients.69

Epigenetic abnormalities in primary cutaneous anaplastic large cell lymphoma Epigenetic abnormalities, including histone tail posttranslational modifications and DNA methylation, are a haematologica | 2019; 104(2)

CD30+ primary cutaneous LPD

hallmark of cancer and a potential target for therapy.70 DNA methylation profiling studies have not been done specifically in cutaneous ALCL; data derived from the analysis of systemic ALCL show an ALK-independent methylation signature, characteristic of and different from that of other peripheral T-cell lymphomas, establishing a relation with the promoter DNA methylation of early Tcell stages.71 The enhancer of the zeste homolog 2 (EZH2) constitutes the core catalytic subunit of polycomb repressive complex2 (PRC2) and has a “canonical” function as a PRC2-dependent lysine 27 of the histone H3 (H3K27) methylator, which can also methylate a number of nonhistone proteins, thereby promoting transcriptional activation, and making it a potential therapeutic target.72 Yi et al. found EZH2 overexpression in pcALCL and large cell transformed cutaneous T-cell lymphomas, in which it regulates apoptosis, cell-cycling in the neoplastic cells, and the interaction between the tumor and its microenvironment.73 Non-coding RNA play an important epigenetic regulatory role that has implications for cell development and cancer, above all in the pathogenesis of T-cell lymphomas.74 MicroRNA (miRNA, miR) are small, non-coding RNA molecules that regulate gene expression at the post-transcriptional level by targeting the 3’-untranslated regions of messenger RNA to promote their degradation or decrease their translation.75 Several studies have identified different miRNA signatures for CD30+ pcALCL.62,76-78 Benner et al. showed upregulation of an oncogenic miRNA signature in pcALCL comprising miR-155, miR-27b, miR-30c and miR29.77 Sandoval’s study confirmed the upregulation of miR155 and identified upregulation of miR-21, miR-1423p/5p, let-7i, miR-424, miR-431, miR-542-5p, miR-29b-1, miR-342-p, and miR-484, with downregulation of some interesting tumor-suppressor miRNA such as miR23b/miR-27b, miR-203, miR-205 and miR-125b.78 This miRNA profile reported for pcALCL differs considerably from those reported in systemic ALCL, which suggests a different underlying pathogenic mechanism or reflects differences in the microenvironment.79,80 In this setting, epigenetic therapy (histone deacetylase inhibitors, such as romidepsin or belinostat, and methylation inhibitors) has shown some benefit in the treatment of ALCL and cutaneous T-cell lymphomas.81,82 In the future, more precise identification of miRNA profiling in pcALCL could allow us to develop specific diagnostic and progression markers and may lead to the use of more specific targeted therapies.

NOTCH signaling in primary cutaneous anaplastic large cell lymphoma The deregulation of Notch signaling in hematopoietic cells has been linked to the development of several hematologic malignancies, including acute lymphoblastic T-cell leukemia, B-cell chronic lymphocytic leukemia, multiple myeloma, acute myeloid leukemia, Hodgkin lymphoma and systemic ALCL.83-88 Increased expression of Notch signaling molecules has been described in primary cutaneous CD30+ lymphoproliferative disorders, as previously mentioned.26 The same group found that pcALCL cells increased their expression of the intracellular domains of Notch receptors Notch1, Notch2, Notch3 and Notch4, as well as of the Notch ligand Delta and the product HES1.89 In addition, it was demonstrated that the inhibition of the haematologica | 2019; 104(2)

Notch pathway through inhibition of gamma-secretase with gamma-secretase inhibitors induces apoptosis and decreases cell viability in pcALCL cell lines, findings that identify the Notch pathway as a potential therapeutic target in pcALCL.89

CDKN2A-CDKN2B losses in primary cutaneous anaplastic large cell lymphoma The CDKN2A-CDKN2B locus on the 9p21 chromosomal band encodes the proteins p14ARF, p16INK4A and p15INK4B. p14ARF-mdm2-p53 and p16INK4A/ p15INK4B-Rb1 pathways are important for controlling the cell cycle, especially progression between G1 and S phases.90 Combined CDKN2A-CDKN2B deletion has been linked to aggressive behavior in cutaneous T-cell lymphomas,91-95 but more rarely in pcALCL, as confirmed by two research groups.95,96

Other cytogenetic abnormalities in primary cutaneous anaplastic large cell lymphoma Additional chromosomal alterations, such as gains of 7q31 and losses on 6q16-6q21, 6q27 and 13q34 regions, are recurrently present in pcALCL.31,62,97 Interestingly, these alterations in pcALCL occur mainly in telomeric and centromeric regions. Genomic imbalances have been found in chromosomal regions coding for FGFR1 (8p11), NRAS (1p13.2), MYCN (2p24.1), RAF1 (3p25) and others.98

Targeted therapies for primary cutaneous CD30-positive lymphoproliferative disorders Currently, complete surgical excision and local radiotherapy are the recommended first-line therapies for solitary or grouped localized pcALCL lesions.99 However, the most appropriate treatment in the relapsed/refractory setting has not been clearly identified. Multiagent chemotherapy is customarily indicated for extracutaneous tumor spread beyond locoregional lymph nodes. In recent years, brentuximab vedotin, an anti-CD30 antibody-drug conjugate has been proposed as one of the best options for achieving complete remission with low toxicity in these patients.100 ALCANZA, an international, open-label, randomized, phase 3, multicenter clinical trial, has shown significant improvement in objective responses lasting at least 4 months with brentuximab vedotin compared with the physician’s choice of methotrexate or bexarotene in CD30+ cutaneous T-cell lymphomas, especially MF and pcALCL (having excluded Sézary syndrome and LyP from the study, since these are entities in which brentuximab vedotin is successful).101 The mean duration of response to this drug in pcALCL is 7.6 months. Peripheral neuropathy and fatigue are the most commonly reported adverse events (in 57.2% and 35.6% of cases, respectively).102 Recently the US Food and Drug Administration and the European Medicines Agency approved the use of brentuximab vedotin in cutaneous T-cell lymphomas.

Besides the use of anti-CD30 molecules in the treatment of primary cutaneous CD30+ lymphoproliferative disorders, molecular data generated in the study of pcALCL offer multiple opportunities for targeted therapies (Figure 4).

The recognition of convergent mutations and kinase fusions leading to STAT3 activation in a subgroup of pcALCL could allow the use of JAK1/2/3 inhibitors in the 231

L. Prieto-Torres et al.

Figure 4. New targeted therapies in advanced primary cutaneous anaplastic large cell lymphoma. Brentuximab vedotin is a drug composed of a chimeric anti-CD30 antibody linked to four monomethyl auristatin (MMAE) molecules (on average) (1). This antibody-drug conjugate first binds to CD30 on the surface of pcALCL cells (2). It is then internalized by a receptor-mediated endocytosis process (3 and 4). The resultant vesicle fuses with lysosomes (4), leading to proteolytic cleavage of the dipeptide linker and the presence of free MMAE molecules, which inhibit tubulin polymerization of the cellular cytoskeleton and arrest growth of pcALCL tumor cells.

Gamma-secretase (γ-secretase) inhibitors prevent release of intracellular NOTCH1 (ICN1) from membrane-tethered heterodimeric NOTCH1 protein. This causes the downregulation of the tumor cell nuclear factor-kB (NFKB) pathway and the inactivation of survival genes. JAK1/2/3 inhibitors are effective in vitro for controlling pcALCL cell growth. This mechanism involves oncogenic JAK1 or STAT3 mutations associated with the hyperactive pSTAT3 shown in pcALCL with an NPM1-TYK2 gene fusion and oncogenic STAT3 activation. In addition, anti-ALK molecules such as crizotinib, alectinib, and ceritinib in pcALCL patients with ALK rearrangements could downregulate the STAT3 pathway, ultimately inducing tumor-cell death. IPH4102 is a humanized monoclonal antibody directed against the cellular receptor KIR3DL2 (CD158K). This receptor has been shown to be aberrantly expressed in advanced pcALCL. IPH4140 targets KIR3DL2 in tumor cells and promotes cell lysis after linking to the CD16 activating receptor through antibody-dependent, cell-derived cytotoxicity mediated by NK cells. At the epigenetic level, histone deacetylase (HDAC) inhibitors and demethylating agents have demonstrated a degree of effectiveness at inducing cell-cycle arrest, differentiation and/or apoptosis of tumor cells.

treatment of this subgroup.68,103 In addition, patients with the rare subtype of ALK+ pcALCL could benefit from therapy with the ALK kinase inhibitor crizotinib, and other new-generation inhibitors such as alectinib or ceritinib.104 KIR3DL2 (CD158K) is a killer cell immunoglobulin-like receptor normally expressed by minor subsets of circulating CD8+ lymphocytes and natural killer cells.105 In contrast, tumor cells of pcALCL and CD30+ lymphoproliferative disorders, which are derived from cell lines Mac2a and Mac2b, express KIR3DL2 strongly.105 In Sézary syndrome, the expression of KIR3DL2 in malignant blood cells has been proposed as a marker of blood tumor burden.106 The novel agent IPH4102 is a monoclonal antibody directed against KIR3DL2 that has proven effective in in vitro cell lines of advanced pcALCL.47 The NOTCH pathway is activated in pcALCL as a consequence of various mutations.89 Deregulated activity of the NOTCH pathway can be inhibited using gamma secretase inhibitors, as has been shown in an in vitro experiment.89 232

Finally, therapies acting at the epigenetic level, based on the mutations and epigenetic alterations described above, could be useful in pcALCL. More specifically, romidepsin and vorinostat, inhibitors of histone deacetylase, derived from the bacterium Chromobacterium violaceum are effective at inducing apoptosis with an antitumor effect in cutaneous T-cell lymphomas, alone and in combination.107,108 Recently, researchers from the Yale School of Medicine, using cell lines derived from patients with advanced MF and Sézary syndrome (MyLa, Sez4, HH, Hut78), observed in vitro that bromodomain and extraterminal (BET) protein inhibitors are synergistically potentiated by BCL2 inhibitors (e.g., venetoclax) and histone deacetylase inhibitors (e.g., vorinostat and romidepsin) in cutaneous T-cell lymphomas with MYC oncogene amplification.109 Patients with cutaneous CD30+ lymphoproliferative disorders with MYC-induced Id2 overexpression could also be candidates for this therapy.24 Recently, it was proposed that miR-155 inhibition could be used in combination with apoptotic treatments haematologica | 2019; 104(2)

CD30+ primary cutaneous LPD

and checkpoint inhibitors in MF patients.110 MiR-155 is also upregulated in pcALCL,77 providing an additional therapeutic opportunity for these tumors.

Acknowledgments Amaia Alzelai, Salma Machan, Victor Alegre and David LapeĂąa provided cases and inspiration.

References 1. Swerdlow SH, Campo E, Pileri SA, et al.

The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390.

2. Taylor J, Xiao W, Abdel-Wahab O. Diagnosis and classification of hematologic malignancies on the basis of genetics. Blood.

2017;130(4):410-423.

3. Macaulay WL. Lymphomatoid papulosis. A continuing self-healing eruption, clinically benign--histologically malignant. Arch Dermatol. 1968;97(1):23-30.

4. Nijsten T, Curiel-Lewandrowski C, Kadin ME. Lymphomatoid papulosis in children: a retrospective cohort study of 35 cases. Arch Dermatol. 2004;140(3):306-312.

5. Kempf W. A new era for cutaneous CD30positive T-cell lymphoproliferative disorders. Semin Diagn Pathol. 2017;34(1):22-35.

6. Kempf W, Mitteldorf C, Karai LJ, et al.

Lymphomatoid papulosis - making sense of the alphabet soup: a proposal to simplify terminology. J Dtsch Dermatol Ges. 2017;15(4):390-394.

7. Bekkenk MW, Geelen FA, van Voorst Vader PC, et al. Primary and secondary cutaneous CD30(+) lymphoproliferative disorders: a report from the Dutch Cutaneous Lymphoma Group on the long-term followup data of 219 patients and guidelines for diagnosis and treatment.

Blood.

2000;95(12):3653-3661.

8. Saggini A, Gulia A, Argenyi Z, et al. A variant of lymphomatoid papulosis simulating primary cutaneous aggressive epidermotropic CD8+ cytotoxic T-cell lymphoma.

Description of 9 cases. Am J Surg Pathol.

2010;34(8):1168-1175.

9. Kempf W, Kazakov DV, Scharer L, et al.

Angioinvasive lymphomatoid papulosis: a new variant simulating aggressive lymphomas. Am J Surg Pathol. 2013;37(1):1-13.

10. Karai LJ, Kadin ME, Hsi ED, et al.

Chromosomal rearrangements of 6p25.3 define a new subtype of lymphomatoid papulosis. Am J Surg Pathol. 2013;37(8): 1173-1181.

11. Wada DA, Law ME, Hsi ED, et al.

Specificity of IRF4 translocations for primary cutaneous anaplastic large cell lymphoma: a multicenter study of 204 skin biopsies. Mod Pathol. 2011;24(4):596-605.

12. Greisser J, Palmedo G, Sander C, et al.

Detection of clonal rearrangement of T-cell receptor genes in the diagnosis of primary cutaneous CD30 lymphoproliferative disorders. J Cutan Pathol. 2006;33(11):711-715.

13. Chott A, Vonderheid EC, Olbricht S, et al.

The dominant T cell clone is present in multiple regressing skin lesions and associated T

haematologica | 2019; 104(2)

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

Funding This work was supported by grants from the Instituto de Salud Carlos III (ISCIII) of the Spanish Ministry of Economy and Competence (MINECO, RTICC ISCIII and CIBERONC) (SAF2013-47416-R, RD06/0020/0107-RD012/0036/0060 and Plan Nacional I+D+I: PI17/2172, PI16/01294 and PIE15/0081), AECC and the Madrid Autonomous Community.

cell lymphomas of patients with lymphomatoid papulosis. J Invest Dermatol.

1996;106(4):696-700. de la Garza Bravo MM, Patel KP, Loghavi S, et al. Shared clonality in distinctive lesions of lymphomatoid papulosis and mycosis fungoides occurring in the same patients suggests a common origin. Hum Pathol.

2015;46(4):558-569.

Rodriguez-Pinilla SM, Ortiz-Romero PL, Monsalvez V, et al. TCR-gamma expression in primary cutaneous T-cell lymphomas.

Am J Surg Pathol. 2013;37(3):375-384.

Yasui D, Miyano M, Cai S, et al. SATB1 targets chromatin remodelling to regulate genes over long distances. Nature.

2002;419(6907):641-645.

Alvarez JD, Yasui DH, Niida H, et al. The MAR-binding protein SATB1 orchestrates temporal and spatial expression of multiple genes during T-cell development. Genes Dev. 2000;14(5):521-535.

Wang Y, Su M, Zhou LL, et al. Deficiency of SATB1 expression in Sezary cells causes apoptosis resistance by regulating FasL/CD95L transcription. Blood. 2011;117 (14):3826-3835.

Sun J, Yi S, Qiu L, et al. SATB1 defines a subtype of cutaneous CD30(+) lymphoproliferative disorders associated with a T-helper 17 cytokine profile. J Invest Dermatol.

2018;138(8):1795-1804.

Durkop H, Hirsch B, Hahn C, et al.

Differential expression and function of A20 and TRAF1 in Hodgkin lymphoma and anaplastic large cell lymphoma and their induction by CD30 stimulation. J Pathol.

2003;200(2):229-239.

Assaf C, Hirsch B, Wagner F, et al.

Differential expression of TRAF1 aids in the distinction of cutaneous CD30-positive lymphoproliferations. J Invest Dermatol.

2007;127(8):1898-1904.

Benner MF, Jansen PM, Meijer CJ, et al.

Diagnostic and prognostic evaluation of phenotypic markers TRAF1, MUM1, BCL2 and CD15 in cutaneous CD30-positive lymphoproliferative disorders. Br J Dermatol.

2009;161(1):121-127.

Mathas S, Kreher S, Meaburn KJ, et al. Gene deregulation and spatial genome reorganization near breakpoints prior to formation of translocations in anaplastic large cell lymphoma. Proc Natl Acad Sci USA. 2009;106 (14):5831-5836.

Cotta CV, Leventaki V, Atsaves V, et al. The helix-loop-helix protein Id2 is expressed differentially and induced by Myc in T-cell lymphomas. Cancer. 2008;112(3):552-561.

Kapur S, Tiemann M, Menke MA, et al. The role of p53 and anaplastic lymphoma kinase genes in the progression of cutaneous CD30(+) lymphoproliferative diseases.

Indian J Med Res. 2005;121(1):46-54.

26. Kamstrup MR, Ralfkiaer E, Skovgaard GL, et al. Potential involvement of Notch1 signalling in the pathogenesis of primary cutaneous CD30-positive lymphoproliferative disorders. Br J Dermatol. 2008;158(4):747753.

27. Vasmatzis G, Johnson SH, Knudson RA, et al. Genome-wide analysis reveals recurrent structural abnormalities of TP63 and other p53-related genes in peripheral T-cell lymphomas. Blood. 2012;120(11):2280-2289.

28. Schrader AM, Chung YY, Jansen PM, et al.

No TP63 rearrangements in a selected group of primary cutaneous CD30+ lymphoproliferative disorders with aggressive clinical course. Blood. 2016;128(1):141-143.

29. Fornari A, Piva R, Chiarle R, et al. Anaplastic large cell lymphoma: one or more entities among T-cell lymphoma? Hematol Oncol.

2009;27(4):161-170.

30. Xing X, Feldman AL. Anaplastic large cell lymphomas: ALK positive, ALK negative, and primary cutaneous. Adv Anat Pathol.

2015;22(1):29-49.

31. Laharanne E, Oumouhou N, Bonnet F, et al.

Genome-wide analysis of cutaneous T-cell lymphomas identifies three clinically relevant classes. J Invest Dermatol. 2010;130(6): 1707-1718.

32. Stein H, Foss HD, Durkop H, et al. CD30(+) anaplastic large cell lymphoma: a review of its histopathologic, genetic, and clinical features. Blood. 2000;96(12):3681-3695.

33. Kerschmann RL, Berger TG, Weiss LM, et al. Cutaneous presentations of lymphoma in human immunodeficiency virus disease.

Predominance of T cell lineage. Arch Dermatol. 1995;131(11):1281-1288.

34. Yurtsever H, Kempf W, Laeng RH. Posttransplant CD30+ anaplastic large cell lymphoma with skin and lymph node involvement. Dermatology. 2003;207(1): 107-110.

35. Benner MF, Willemze R. Applicability and prognostic value of the new TNM classification system in 135 patients with primary cutaneous anaplastic large cell lymphoma.

Arch Dermatol. 2009;145(12):1399-1404.

36. Woo DK, Jones CR, Vanoli-Storz MN, et al.

Prognostic factors in primary cutaneous anaplastic large cell lymphoma: characterization of clinical subset with worse outcome. Arch Dermatol. 2009;145(6):667-674.

37. Burg G, Kempf W, Kazakov DV, et al.

Pyogenic lymphoma of the skin: a peculiar variant of primary cutaneous neutrophilrich CD30+ anaplastic large-cell lymphoma.

Clinicopathological study of four cases and review of the literature. Br J Dermatol.

2003;148(3):580-586.

38. Massone C, El-Shabrawi-Caelen L, Kerl H, et al. The morphologic spectrum of primary cutaneous anaplastic large T-cell lymphoma: a histopathologic study on 66 biopsy speci-

233

L. Prieto-Torres et al.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

234

mens from 47 patients with report of rare variants. J Cutan Pathol. 2008;35(1):46-53.

Kempf W, Kazakov DV, Paredes BE, et al.

Primary cutaneous anaplastic large cell lymphoma with angioinvasive features and cytotoxic phenotype: a rare lymphoma variant within the spectrum of CD30+ lymphoproliferative disorders. Dermatology. 2013; 227(4):346-352.

Resnik KS, Kutzner H. Of lymphocytes and cutaneous epithelium: keratoacanthomatous hyperplasia in CD30+ lymphoproliferative disorders and CD30+ cells associated with keratoacanthoma.

Am J Dermatopathol. 2010;32(3):314-315.

Wang J, Sun NC, Nozawa Y, et al.

Histological and immunohistochemical characterization of extranodal diffuse largecell lymphomas with prominent spindle cell features. Histopathology. 2001;39(5):476481.

Kinney MC, Collins RD, Greer JP, et al. A small-cell-predominant variant of primary Ki-1 (CD30)+ T-cell lymphoma. Am J Surg Pathol. 1993;17(9):859-868.

Wang L, Li C, Gao T. Cutaneous intravascular anaplastic large cell lymphoma. J Cutan Pathol. 2011;38(2):221-226.

Ferrara G, Ena L, Cota C, et al.

Intralymphatic spread is a common finding in cutaneous CD30+ lymphoproliferative disorders. Am J Surg Pathol. 2015;39(11): 1511-1517.

Felgar RE, Macon WR, Kinney MC, et al.

TIA-1 expression in lymphoid neoplasms.

Identification of subsets with cytotoxic T lymphocyte or natural killer cell differentiation. Am J Pathol. 1997;150(6):1893-1900.

Savage KJ, Harris NL, Vose JM, et al. ALKanaplastic large-cell lymphoma is clinically and immunophenotypically different from both ALK+ ALCL and peripheral T-cell lymphoma, not otherwise specified: report from the International Peripheral T-Cell Lymphoma Project. Blood. 2008;111(12): 5496-5504.

Battistella M, Janin A, Jean-Louis F, et al.

KIR3DL2 (CD158k) is a potential therapeutic target in primary cutaneous anaplastic large-cell lymphoma. Br J Dermatol. 2016; 175(2):325-333.

Geissinger E, Sadler P, Roth S, et al. Disturbed expression of the T-cell receptor/CD3 complex and associated signaling molecules in CD30+ T-cell lymphoproliferations.

Haematologica. 2010;95(10):1697-1704.

Macgrogan G, Vergier B, Dubus P, et al.

CD30-positive cutaneous large cell lymphomas. A comparative study of clinicopathologic and molecular features of 16 cases. Am J Clin Pathol. 1996;105(4):440450.

Feldman AL, Dogan A, Smith DI, et al.

Discovery of recurrent t(6;7)(p25.3;q32.3) translocations in ALK-negative anaplastic large cell lymphomas by massively parallel genomic sequencing. Blood. 2011;117(3): 915-919.

Parrilla Castellar ER, Jaffe ES, Said JW, et al.

ALK-negative anaplastic large cell lymphoma is a genetically heterogeneous disease with widely disparate clinical outcomes. Blood. 2014;124(9):1473-1480.

Feldman AL, Law M, Remstein ED, et al.

Recurrent translocations involving the IRF4 oncogene locus in peripheral T-cell lymphomas. Leukemia. 2009;23(3):574-580.

Kiran T, Demirkesen C, Eker C, et al. The significance of MUM1/IRF4 protein expression and IRF4 translocation of CD30(+) cutaneous T-cell lymphoproliferative disorders:

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

a study of 53 cases. Leuk Res. 2013;37 (4):396-400.

Onaindia A, Montes-Moreno S, RodriguezPinilla SM, et al. Primary cutaneous anaplastic large cell lymphomas with 6p25.3 rearrangement exhibit particular histological features. Histopathology. 2015;66(6):846855.

Fauconneau A, Pham-Ledard A, Cappellen D, et al. Assessment of diagnostic criteria between primary cutaneous anaplastic large-cell lymphoma and CD30-rich transformed mycosis fungoides; a study of 66 cases. Br J Dermatol. 2015;172(6):1547-1554.

Pham-Ledard A, Prochazkova-Carlotti M, Laharanne E, et al. IRF4 gene rearrangements define a subgroup of CD30-positive cutaneous T-cell lymphoma: a study of 54 cases. J Invest Dermatol. 2010;130(3):816825.

Xing X, Flotte TJ, Law ME, et al. Expression of the chemokine receptor gene, CCR8, is associated With DUSP22 rearrangements in anaplastic large cell lymphoma. Appl Immunohistochem Mol Morphol. 2015;23 (8):580-589. van Kester MS, Tensen CP, Vermeer MH, et al. Cutaneous anaplastic large cell lymphoma and peripheral T-cell lymphoma NOS show distinct chromosomal alterations and differential expression of chemokine receptors and apoptosis regulators. J Invest Dermatol. 2010;130(2):563-575.

Melard P, Idrissi Y, Andrique L, et al.

Molecular alterations and tumor suppressive function of the DUSP22 (Dual specificity phosphatase 22) gene in peripheral T-cell lymphoma subtypes.

Oncotarget.

2016;7(42):68734-68748.

Sekine Y, Tsuji S, Ikeda O, et al. Regulation of STAT3-mediated signaling by LMWDSP2. Oncogene. 2006;25(42):5801-5806.

Li JP, Yang CY, Chuang HC, et al. The phosphatase JKAP/DUSP22 inhibits T-cell receptor signalling and autoimmunity by inactivating Lck. Nat Commun. 2014;5:3618.

Le Deley MC, Reiter A, Williams D, et al.

Prognostic factors in childhood anaplastic large cell lymphoma: results of a large European intergroup study. Blood.

2008;111(3):1560-1566.

Oschlies I, Lisfeld J, Lamant L, et al. ALKpositive anaplastic large cell lymphoma limited to the skin: clinical, histopathological and molecular analysis of 6 pediatric cases.

A report from the ALCL99 study.

Haematologica. 2013;98(1):50-56.

Kadin ME, Pinkus JL, Pinkus GS, et al.

Primary cutaneous ALCL with phosphorylated/activated cytoplasmic ALK and novel phenotype: EMA/MUC1+, cutaneous lymphocyte antigen negative. Am J Surg Pathol.

2008;32(9):1421-1426.

Lamant L, Pileri S, Sabattini E, et al.

Cutaneous presentation of ALK-positive anaplastic large cell lymphoma following insect bites: evidence for an association in five cases. Haematologica. 2010;95(3):449455.

Chavan RN, Bridges AG, Knudson RA, et al.

Somatic rearrangement of the TP63 gene preceding development of mycosis fungoides with aggressive clinical course. Blood Cancer J. 2014;e253.

Ohgami RS, Ma L, Merker JD, et al. STAT3 mutations are frequent in CD30+ T-cell lymphomas and T-cell large granular lymphocytic leukemia. Leukemia 2013;27: 22442247.

Crescenzo R, Abate F, Lasorsa E, et al.

Convergent mutations and kinase fusions

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

87.

lead to oncogenic STAT3 activation in anaplastic large cell lymphoma. Cancer Cell.

2015;27(4):516-532.

Velusamy T, Kiel MJ, Sahasrabuddhe AA, et al. A novel recurrent NPM1-TYK2 gene fusion in cutaneous CD30-positive lymphoproliferative disorders. Blood. 2014;124(25): 3768-3771.

Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144 (5):646-674.

Hassler MR, Pulverer W, Lakshminarasimhan R, et al. Insights into the pathogenesis of anaplastic large-cell lymphoma through genome-wide DNA methylation profiling. Cell Rep. 2016;17(2):596608.

Kim KH, Roberts CW. Targeting EZH2 in cancer. Nat Med. 2016;22(2):128-134.

Yi S, Sun J, Qiu L, et al. Dual role of EZH2 in cutaneous anaplastic large cell lymphoma: promoting tumor cell survival and regulating tumor microenvironment. J Invest Dermatol. 2018;138(5):1126-1136.

Allan RS, Nutt SL. Deciphering the epigenetic code of T lymphocytes. Immunol Rev. 2014;261(1):50-61.

Persson JL. miRNA in mycosis fungoides and skin inflammation. APMIS. 2013;121 (11):1017-1019.

Ralfkiaer U, Hagedorn PH, Bangsgaard N, et al. Diagnostic microRNA profiling in cutaneous T-cell lymphoma (CTCL). Blood.

2011;118(22):5891-5900.

Benner MF, Ballabio E, van Kester MS, et al.

Primary cutaneous anaplastic large cell lymphoma shows a distinct miRNA expression profile and reveals differences from tumorstage mycosis fungoides. Exp Dermatol.

2012;21(8):632-634.

Sandoval J, Diaz-Lagares A, Salgado R, et al.

MicroRNA expression profiling and DNA methylation signature for deregulated microRNA in cutaneous T-cell lymphoma. J Invest Dermatol. 2015;135(4):1128-1137.

Lawrie CH. MicroRNA expression in lymphoid malignancies: new hope for diagnosis and therapy? J Cell Mol Med. 2008;12(5a): 1432-1444.

Liu C, Iqbal J, Teruya-Feldstein J, et al.

MicroRNA expression profiling identifies molecular signatures associated with anaplastic large cell lymphoma. Blood.

2013;122(12):2083-2092.

Hassler MR, Klisaroska A, Kollmann K, et al.

Antineoplastic activity of the DNA methyltransferase inhibitor 5-aza-2'-deoxycytidine in anaplastic large cell lymphoma.

Biochimie. 2012;94(11):2297-2307.

Wong HK. Novel biomarkers, dysregulated epigenetics, and therapy in cutaneous T-cell lymphoma. Discov Med. 2013;16(87):71-78.

Ellisen LW, Bird J, West DC, et al. TAN-1, the human homolog of the Drosophila notch gene, is broken by chromosomal translocations in T lymphoblastic neoplasms. Cell. 1991;66(4):649-661.

Weng AP, Ferrando AA, Lee W, et al.

Activating mutations of NOTCH1 in human T cell acute lymphoblastic leukemia.

Science. 2004;306(5694):269-271.

Jundt F, Anagnostopoulos I, Forster R, et al.

Activated Notch1 signaling promotes tumor cell proliferation and survival in Hodgkin and anaplastic large cell lymphoma. Blood.

2002;99(9):3398-3403.

Jundt F, Probsting KS, Anagnostopoulos I, et al. Jagged1-induced Notch signaling drives proliferation of multiple myeloma cells.

Blood. 2004;103(9):3511-3515.

Rosati E, Sabatini R, Rampino G, et al.

haematologica | 2019; 104(2)

CD30+ primary cutaneous LPD

88.

89.

90.

91.

92.

93.

94.

95.

96.

Constitutively activated Notch signaling is involved in survival and apoptosis resistance of B-CLL cells. Blood. 2009;113(4):856-865.

Tohda S, Nara N. Expression of Notch1 and Jagged1 proteins in acute myeloid leukemia cells. Leuk Lymphoma. 2001;42(3):467-472.

Kamstrup MR, Biskup E, Gniadecki R. Notch signalling in primary cutaneous CD30+ lymphoproliferative disorders: a new therapeutic approach? Br J Dermatol.

2010;163(4):781-788.

Sharpless NE, DePinho RA. The INK4A/ARF locus and its two gene products. Curr Opin Genet Dev. 1999;9(1):22-30.

Scarisbrick JJ, Woolford AJ, Calonje E, et al.

Frequent abnormalities of the p15 and p16 genes in mycosis fungoides and sezary syndrome. J Invest Dermatol. 2002;118(3):493499.

Navas IC, Algara P, Mateo M, Martinez P, Garcia C, Rodriguez JL, et al. p16(INK4a) is selectively silenced in the tumoral progression of mycosis fungoides. Lab Invest.

2002;82(2):123-132.

Navas IC, Ortiz-Romero PL, Villuendas R, et al. p16(INK4a) gene alterations are frequent in lesions of mycosis fungoides. Am J Pathol.

2000;156(5):1565-1572.

Gallardo F, Esteller M, Pujol RM, et al.

Methylation status of the p15, p16 and MGMT promoter genes in primary cutaneous T-cell lymphomas. Haematologica.

2004;89(11):1401-1403.

Laharanne E, Chevret E, Idrissi Y, et al.

CDKN2A-CDKN2B deletion defines an aggressive subset of cutaneous T-cell lymphoma. Mod Pathol. 2010;23(4):547-558.

Nicolae-Cristea AR, Benner MF, Zoutman WH, et al. Diagnostic and prognostic signif-

haematologica | 2019; 104(2)

icance of CDKN2A/CDKN2B deletions in patients with transformed mycosis fungoides and primary cutaneous CD30-positive lymphoproliferative disease. Br J Dermatol. 2015;172(3):784-788.

97. Szuhai K, van Doorn R, Tensen CP, et al.

Array-CGH analysis of cutaneous anaplastic large cell lymphoma. Methods Mol Biol.

2013;973:197-212.

98. Mao X, Orchard G, Lillington DM, et al.

Genetic alterations in primary cutaneous CD30+ anaplastic large cell lymphoma.

Genes Chromosomes Cancer. 2003;37(2): 176-185.

99. Kempf W, Pfaltz K, Vermeer MH, et al.

EORTC, ISCL, and USCLC consensus recommendations for the treatment of primary cutaneous CD30-positive lymphoproliferative disorders: lymphomatoid papulosis and primary cutaneous anaplastic large-cell lymphoma. Blood. 2011;118 (15):4024-4035.

100. Vu K, Ai W. Update on the treatment of anaplastic large cell lymphoma. Curr Hematol Malig Rep. 2018;13(2):135-141.

101. Prince HM, Kim YH, Horwitz SM, et al.

Brentuximab vedotin or physician's choice in CD30-positive cutaneous T-cell lymphoma (ALCANZA): an international, openlabel, randomised, phase 3, multicentre trial.

Lancet. 2017;390(10094):555-566.

102. Enos TH, Feigenbaum LS, Wickless HW. Brentuximab vedotin in CD30(+) primary cutaneous T-cell lymphomas: a review and analysis of existing data. Int J Dermatol.

2017;56(12):1400-1405.

103. Nairismagi M, Gerritsen ME, Li ZM, et al.

Oncogenic activation of JAK3-STAT signaling confers clinical sensitivity to PRN371, a

novel selective and potent JAK3 inhibitor, in natural killer/T-cell lymphoma. Leukemia.

2018;32(5):1147-1156.

104. Mosse YP, Lim MS, Voss SD, et al. Safety and activity of crizotinib for paediatric patients with refractory solid tumours or anaplastic large-cell lymphoma: a Children's Oncology Group phase 1 consortium study.

Lancet Oncol. 2013;14(6):472-480.

105. Moretta A, Parolini S, Castriconi R, et al.

Function and specificity of human natural killer cell receptors. Eur J Immunogenet.

1997;24(6):455-468.

106. Poszepczynska-Guigne E, Schiavon V, D'Incan M, et al. CD158k/KIR3DL2 is a new phenotypic marker of Sezary cells: relevance for the diagnosis and follow-up of Sezary syndrome. J Invest Dermatol. 2004;122(3): 820-823.

107. Ott CJ, Wu CJ. HDAC inhibitors finally open up: chromatin accessibility signatures of CTCL. Cancer Cell. 2017;32(1):1-3.

108. Schcolnik-Cabrera A, Dominguez-Gomez G, Duenas-Gonzalez A. Comparison of DNA demethylating and histone deacetylase inhibitors hydralazine-valproate versus vorinostat-decitabine incutaneous t-cell lymphoma in HUT78 cells. Am J Blood Res. 2018;8(2):5-16.

109. Kim SR, Lewis JM, Cyrenne BM, et al. BET inhibition in advanced cutaneous T cell lymphoma is synergistically potentiated by BCL2 inhibition or HDAC inhibition.

Oncotarget. 2018;9(49):29193-29207.

110. Moyal L, Yehezkel S, Gorovitz B, et al.

Oncogenic role of microRNA-155 in mycosis fungoides: an in vitro and xenograft mouse model study. Br J Dermatol. 2017; 177(3):791-800.

235

REVIEW ARTICLE Ferrata Storti Foundation

Pathogenic immune response to therapeutic factor VIII: exacerbated response or failed induction of tolerance?

Aditi Varthaman1,2,3,4 and Sébastien Lacroix-Desmazes2,3,4

MRC Human Immunology Unit, Weatherall Institute of Molecular Medicine, University of Oxford, UK; 2INSERM, UMR S 1138, Centre de Recherche des Cordeliers, Paris, France; 3 Sorbonne Université, UMR S 1138, Centre de Recherche des Cordeliers, Paris, France and 4Université Paris Descartes, Sorbonne Paris Cité, UMR S 1138, Centre de Recherche des Cordeliers, France 1

Haematologica 2019 Volume 104(2):236-244

ABSTRACT

T

Correspondence: Sebastien.Lacroix-Desmazes@crc.jussieu.fr

Received: October 4, 2018.

Accepted: November 23, 2018.

Pre-published: December 4, 2018

doi:10.3324/haematol.2018.206383 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/236

herapeutic factor VIII is highly immunogenic. Despite intensive research in the last decades, the reasons why 5-30% of patients with hemophilia A (of all severities) develop inhibitory anti-factor VIII antibodies (inhibitors) following replacement therapy remain an enigma. Under physiological conditions, endogenous factor VIII is recognized by the immune system. Likewise, numerous observations indicate that, in hemophilia A patients without inhibitors, exogenous therapeutic factor VIII is immunologically assessed and tolerated. A large part of the research on the immunogenicity of therapeutic factor VIII is attempting to identify the ‘danger signals’ that act as adjuvants to the deleterious anti-factor VIII immune responses. However, several of the inflammatory assaults concomitant to factor VIII administration initially hypothesized as potential sources of danger signals (e.g., bleeding, infection, and vaccination) have been disproved to be such signals. Conversely, recent evidence suggests that cells from inhibitor-negative patients are able to activate anti-inflammatory and tolerogenic mechanisms required to suppress deleterious immune responses, while cells from inhibitor-positive patients are not. Based on the available observations, we propose a model in which all hemophilia A patients develop anti-factor VIII immune responses during replacement therapy irrespective of associated danger signals. We further postulate that the onset of clinically relevant factor VIII inhibitors results from an inability to develop counteractive tolerogenic responses to exogenous factor VIII rather than from an exacerbated activation of the immune system at the time of factor VIII administration.

A better understanding of the pathogenesis of neutralizing anti-factor VIII antibodies will have repercussions on the clinical management of patients and highlight new strategies to achieve active immune tolerance to therapeutic factor VIII.

©2019 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.

236

Introduction Hemophilia A is a rare X-linked hemorrhagic disorder that results from insufficient levels of pro-coagulant factor VIII (FVIII). Patients with hemophilia A constitute a heterogeneous group of individuals. Three severities of hemophilia A are distinguished depending on the levels of circulating endogenous FVIII. They reflect the diversity in the mutations in the gene encoding for FVIII: patients with a severe form of the disease have undetectable FVIII activity in plasma, while patients with mild and moderate hemophilia A have more than 1% of the normal levels of FVIII activity. Patients with severe hemophilia A are further differentiated according to the presence or absence of a non-functional FVIII protein (FVIII:Ag). For instance, among patients with severe hemophilia A, those with the V634M missense mutahaematologica | 2019; 104(2)

Immune response to FVIII in hemophilia patients

tion have normal levels of FVIII:Ag, although the protein is non-functional,1 while patients with large deletion/intron inversions have no circulating protein.2 Due to such differences in protein expression, patients are also heterogeneous as far as the education of their immune system against endogenous FVIII is concerned.

To date, the prevention or treatment of bleeds in hemophilia A patients relies on the intravenous administration of therapeutic FVIII. Therapeutic FVIII is purified from pools of plasma from healthy blood donors or originates from recombinant technology. While differences exist between plasma-derived and recombinant FVIII products, as well as among recombinant products, in terms of structure, glycosylation pattern,3 ability to bind von Willebrand factor (VWF),4 the endogenous chaperone for FVIII, all the available products share the property of inducing neutralizing immunoglobulin G (IgG), termed ‘FVIII inhibitors’, in a substantial number of patients. The occurrence of FVIII inhibitors following replacement therapy is a serious clinical problem that complicates patients’ management and reduces their quality of life, as well as being a major society issue owing to the high costs associated with the treatment of bleeding when FVIII cannot be used.5 Several factors have been identified as increasing the risk of a patient developing FVIII inhibitors, in particular genetic risk factors such as a family history of inhibitor development,6 the type of gene abnormality causing the hemophilia A and the ensuing severity of the disease,7,8 HLA-DR haplotypes9,10 and polymorphisms in a restricted set of immune genes.11-14 Nevertheless, it is, to date, impossible to predict with certainty whether a given patient will develop FVIII inhibitors.

Over the last 20 years, a large body of the research dedicated to deciphering the immunogenicity of FVIII has been based on the ‘danger theory’ proposed by Polly Matzinger almost 25 years ago.15 Researchers have attempted to elucidate the nature of the danger signals that are adjuvants of the immune response to exogenous FVIII in 5-30% of patients with hemophilia A (including all severities of the disease) following replacement therapy. Here, we review the evidence on the presence of danger signals associated with FVIII administration to question the notion that developing an immune response to FVIII requires danger signals. Furthermore, we challenge the idea that developing an immune response to FVIII is unequivocally pathogenic and propose that the development of FVIII inhibitors in a substantial number of patients results from the inability of the immune system to mount a counteractive tolerogenic response.

signals concomitant with antigen uptake to mature appropriately, activate naïve T cells and induce the immune response. In the context of the anti-FVIII immune response, severe/recurrent bleeds, trauma, surgery, infection or vaccination have been proposed as potential sources of danger signals that may trigger the immune system.22 In recent years, experimental investigations performed in FVIII-deficient mice, a model of severe hemophilia A, and clinical observations in hemophilia A patients have challenged most of these hypotheses, as discussed below.

Factor VIII dose matters The dose of administered FVIII is one of the few parameters that has shown consistent association with the development of an anti-FVIII immune response. Seminal work in FVIII-deficient mice clearly demonstrated that increasing the dose of intravenously injected FVIII results in a proportional increase in the intensity of the immune response.23 The experimental data were confirmed by the RODIN study in which the intensity of FVIII treatment and mean dose of administered FVIII correlated with the incidence of inhibitor development in patients with severe hemophilia.24 Conversely, the reduction in the amount of FVIII internalized by antigen-presenting cells and, consequently, presented to CD4+ T cells has been hypothesized as a mechanism by which von Willebrand factor may play an immuno-protective role towards therapeutic FVIII.25 On the other hand, the daily administration of high-dose FVIII to inhibitor-positive patients undergoing immune tolerance induction protocols results in the eradication of FVIII inhibitors in two-thirds of the cases. This suggests that the antigen dose may not be as important in determining immunogenicity.

An enticing recent concept proposes that the immunogenicity of a protein depends on its discontinuous presence in the organism. This ‘discontinuity theory’ first proposed by Pradeu et al. states that “the key to the induction of an immune response is antigenic difference in a timedependent context”. It is hypothesized that the intermittent appearance of a foreign antigen triggers strong and long-lasting immune responses, whereas the persistence of the same foreign antigen over time or its progressive introduction over a long period leads to tolerance.26 This theory is particularly suitable in the case of therapeutic FVIII, a molecule with a short half-life (<15 hours) that is administered either on-demand or every 2 to 3 days in patients under prophylactic treatment, and thus intermittently appears and disappears from the circulation along cycles of intravenous injections.

Danger signals as adjuvants of the anti-factor VIII immune response in patients with hemophilia A

Factor VIII is not an alarmin

The immune response to therapeutic FVIII is believed to be a classical immune response against an exogenous antigen, wherein some of the intravenously administered FVIII transiently accumulates in secondary lymphoid organs, as observed in spleens of FVIII-deficient mice,16,17 is internalized by antigen-presenting cells18,19 and presented to naïve FVIII-specific CD4+ T cells. Upon activation, FVIII-specific T cells proliferate and provide help to naïve FVIII-specific B cells that differentiate into memory B cells or plasmocytes secreting inhibitory anti-FVIII IgG. According to the “infectious non-self model” introduced

by C Janeway20 and later the “danger theory” coined by P Matzinger,21 antigen-presenting cells need to sense danger

Danger signals are provided by endogenous (damageassociated molecular patterns, DAMP) or exogenous (pathogen-associated molecular patterns, PAMP) triggers, collectively referred to as ‘alarmins’, that, upon binding to pattern-recognition receptors, lead to the maturation of antigen-presenting cells which is crucial for the activation of naïve T cells. Early studies on FVIII immunogenicity investigated the possibility that FVIII directly induces the maturation of antigen-presenting cells. While the first domain of hemagglutinin (HA1) was found to induce significantly higher levels of anti-HA1 IgG in intravenously injected mice when fused to the FVIII light chain,27 coincubation of full-length FVIII with Toll-like receptor 2-

haematologica | 2019; 104(2)

237

A. Varthaman et al.

transfected cells or with immature human dendritic cells failed to trigger Toll-like receptor signaling28 or to induce dendritic cell maturation.29 The engagement of FVIII in the coagulation cascade induces a burst of thrombin production that leads to the generation of fibrin and to the consolidation of the platelet clot. Thrombin not only cleaves fibrinogen into fibrin but also cleaves proteinase-associated receptors, which in turn induce the release of pro-inflammatory cytokines and chemokines.30 Initial experiments elegantly suggested that FVIII-mediated thrombin-dependent activation of proteinase-associated receptors provides adjuvant signals for an anti-ovalbumin immune response in FVIII-deficient mice.31 However, later work using non-active FVIII with the V634M mutation in FVIII-deficient mice or normal FVIII in wild-type mice with impaired coagulation ruled out the possibility of FVIII being not only the target but also the trigger of the anti-FVIII immune response.1,32 The inability of FVIII to activate the immune system directly suggests that FVIII is not involved in the generation of danger signals required to mount efficient primary immune responses.

Factor VIII inhibitor development: a case for inflammatory assaults?

Major bleeds, surgery, infections or vaccination at the time of FVIII administration have been proposed as potential inflammatory assaults that could predispose hemophilia A patients to develop allo-antibodies to exogenous FVIII.22 Counter-intuitively, experiments in FVIII-deficient mice recently demonstrated that exposure to the liveattenuated measles-mumps-rubella vaccine at the time of FVIII administration has no influence on the incidence or intensity of immune responses to therapeutic FVIII.33 The experimental data were confirmed by retrospective clinical observations from the PEDNET registry showing that pediatric vaccination given in close proximity to the administration of FVIII is not associated with an increased risk of FVIII inhibitor development.34 Interestingly, influenza vaccination concurrent to FVIII treatment significantly reduced the incidence of anti-FVIII immune responses in mice.33 Chronic inflammation associated with recurrent bleeding35 as well as acute hemarthrosis following knee injury36 in FVIII-deficient mice failed to increase the immune response to exogenous FVIII. In fact, the induction of heme-oxygenase 1 (HO-1), a stress-inducible enzyme with potent anti-inflammatory activity, following injection of heme,35 hemolyzed blood or following knee puncture-associated hemolysis (unpublished data), was shown to reduce the immune response to therapeutic FVIII. Interestingly, these observations find resonance in hemophilia A patients. Indeed, a clinical trial was initiated in which the exposure of previously untreated patients to immunological danger signals was avoided during treatment with FVIII:37,38 the first treatment with FVIII was avoided if the patient was bleeding or had an infection; surgery was avoided during the first 20 days of exposure to FVIII; vaccinations were not given on the same day as FVIII treatment; and bleeds were treated as early as possible to shorten the time of potential tissue damage. Such a drastic protocol led to an inhibitor incidence of 40% provoking the early termination of the EPIC study (ClinicalTrials.gov Identifier: NCT01376700).

Taken together, the large majority of the investigations 238

performed over the last 10 years failed to identify candidate danger signals the control of which would unequivocally reduce the immunogenicity of therapeutic FVIII, reflecting the possibility that danger signals may not be as important as first thought for the initiation of the antiFVIII immune response. Alternatively, the background level of activation of the innate/adaptive immune system may be sufficient to allow activation of naĂŻve FVIII-specific T cells without the need for overt danger signals, provided that a sufficient amount of FVIII is internalized, processed by antigen-presenting cells and presented to T cells. In line with this, we recently observed that the physiological baseline activation of complement C3 (i.e., spontaneous complement C3 tick-over)39,40 increases endocytosis of FVIII by immature dendritic cells without inducing their maturation. The mere in vivo elimination of C3 allowed a drastic reduction of the immunogenicity of therapeutic FVIII in FVIII-deficient mice.41

Self-recognition: the original sin of the immune system The recognition of self, and in particular FVIII, is part of normal immune homeostasis.42 It is clear today that both B and T lymphocytes undergo positive and negative selection processes, in the bone marrow and thymus, respectively, wherein highly autoreactive lymphocytes are eliminated and poorly autoreactive ones are retained. In the case of T lymphocytes, thymic T cells with intermediate affinities for self-antigens may differentiate into natural regulatory T cells that suppress autoreactive T-cell responses.43 Regulatory T cells may also develop in the periphery upon encounter of antigens under non-inflammatory conditions.44 Self-recognition may thus be considered as the original sin of the adaptive immune system: the functionality of the emerging lymphocytes and of their antigen-specific receptors is in essence based on the recognition of self-proteins, which lays the ground for the selection of potential harmful autoreactivity. In this context, physiological immune homeostasis relies on recognition of self, balanced by a tight control of the autoreactive lymphocytes. This is particularly true for the recognition of FVIII under physiological conditions (Figure 1).

Recognition of endogenous factor VIII by T cells under physiological conditions The existence of FVIII-reactive T cells in healthy subjects was first suggested by the seminal work of the group led by B Conti-Fine. In vitro assays revealed proliferation of CD4+ T cells in the presence of the FVIII protein and overlapping peptides spanning the different domains of FVIII in more than 50% of healthy donors.45,46 Recently, preexisting FVIII-specific CD4+ T cells in healthy individuals were quantified accurately following cycles of stimulation of CD4+ T cells by FVIII-loaded mature autologous monocyte-derived dendritic cells.47 The frequency of preexisting FVIII-specific T cells was evaluated to be about 2 cells per million CD4+ T cells.47 This value is in the same range as that of CD4+ T cells specific for foreign proteins such as ovalbumin or keyhole limpet hemocyanin, i.e., 1.3 and 20 CD4+ T cells per million, respectively.48 In marked contrast, it is up to one log higher than that of preexisting T cells specific for immunogenic therapeutic proteins such as adalimumab, rituximab, infliximab and erythropoietin (0.1 to 1 cell/million CD4+ T cells),48,49 and two logs greater than that of preexisting T cells specific for the nonimmunogenic humanized therapeutic antibody haematologica | 2019; 104(2)

Immune response to FVIII in hemophilia patients

trastuzumab (0.1 cell/million CD4+ T cells).50 Taken together, these findings suggest inefficient negative selection of FVIII-reactive T cells in the thymus, despite the reported expression of FVIII mRNA in medullary thymic epithelial cells from healthy individuals.51 An additional, striking finding is that half of the detected FVIII-specific CD4+ T cells have a memory phenotype,47 suggesting that immune responses to endogenous FVIII are ongoing under physiological conditions and are regulated to avoid deleterious autoreactivity. Indeed, the detection of regulatory T cells concomitant to that of FVIII-reactive T cells had been suggested in two independent studies, though the specificity of the regulatory T cells for FVIII was not demonstrated. Hu et al. reported increased ratios of transforming growth factor-beta-producing T helper 3 regulatory cells over interferon-gammaproducing T helper 1 cells or interleukin-4-producing T helper 2 cells among CD4+ T cells from heathy donors, as compared to CD4+ T cells from patients with FVIII inhibitors.45 In parallel, Kamate et al. observed that elimi-

nation of natural CD4+CD25+ regulatory T cells from peripheral blood cells of healthy individuals increases the otherwise poorly detectable FVIII-induced activation of CD4+ T cells in vitro.52

Recognition of endogenous factor VIII by immunoglobulins under physiological conditions The presence of natural anti-FVIII IgG in the blood of healthy donors was described for the first time by the group of MD Kazatchkine.53 In their seminal work, IgGdependent FVIII neutralization was detected in the plasma of 17% of the donors. Subsequently, Reipertâ&#x20AC;&#x2122;s group confirmed the presence of anti-FVIII antibodies in about 19% of healthy individuals, characterized by a highest prevalence of IgG1, IgG3 and IgA subclasses with low apparent binding equilibrium affinities for FVIII in the nanomolar range.54,55 Interestingly, and reminiscent of the situation in the T-cell compartment of the immune system, the recognition of endogenous FVIII by immunoglobulins is subject to regulation by anti-idiotypic antibodies in normal plas-

Figure 1. Immune recognition of factor VIII in health and disease. Healthy donors. At the humoral level, tolerance to factor VIII (FVIII) under physiological conditions relies on an equilibrium between the recognition of FVIII by naturally occurring potentially inhibitory anti-FVIII antibodies and their control by blocking anti-idiotypic antibodies. Blocking anti-idiotypic antibodies may also regulate B-cell clones that secrete FVIII-specific autoantibodies. At the T-cell level, natural FVIII-reactive T cells may be down-regulated by natural regulatory T cells (Tregs; i.e., CD4+CD25+FoxP3+ Tregs) and/or by induced transforming growth factor-beta-secreting Tregs. In rare cases, and in particular in aging individuals, tolerance may fail and a neutralizing immune response to endogenous FVIII may develop - a condition referred to as acquired hemophilia. Patients with congenital hemophilia A. Based on the available evidence, we propose that a similar equilibrium between FVIII-reactive and FVIIIprotective immune elements guarantees tolerance to therapeutic FVIII in inhibitor-negative patients, either upon spontaneous induction of immune tolerance at the time of initiation of FVIII treatment (i.e., in 70-80% of patients with the severe form of the disease), or after a drastic boost of the immune system in patients undergoing successful immune tolerance induction (ITI) (i.e., in about 70% of inhibitor-positive patients with severe hemophilia A undergoing ITI). As is the case in healthy individuals, immune tolerance to FVIII may be lost upon aging.71 Of note, 10% of the patients (30% of the inhibitor-positive patients with severe hemophilia A in whom ITI fails) are not able to develop immune tolerance to FVIII, even after intensive desensitization protocols. Inh+: inhibitor-positive; Inh-: inhibitor-negative; ITI: immune tolerance induction; anti-Id IgG: anti-idiotypic immunoglobulin G; aFVIII T cell: factor VIII-reactive T cell.

haematologica | 2019; 104(2)

239

A. Varthaman et al.

ma.56,57 Thus, IgG purified from the plasma of healthy individuals bind to murine monoclonal anti-FVIII IgG,56 and inhibit the binding of mouse monoclonal and human polyclonal anti-FVIII IgG to FVIII.58 Therapeutic preparations of pooled normal IgG, intravenous immunoglobulins, contain both anti-FVIII IgG with inhibitory potential toward FVIII59 and anti-idiotypic antibodies capable of blocking disease-associated anti-FVIII antibodies.60 This suggests that ‘natural’ protective anti-idiotypic antibodies are produced spontaneously in healthy individuals and exert a tight control on potentially inhibitory ‘natural’ anti-FVIII IgG. In conclusion, endogenous FVIII is not merely ignored by the immune system under physiological conditions.

On the contrary, the homeostasis of FVIII recognition by the healthy immune system relies on an equilibrium between FVIII-specific T and B cells and FVIII-binding antibodies, and a counteractive tolerogenic response mediated by regulatory T cells and blocking anti-idiotypic antibodies (Figure 1).

Recognition of exogenous therapeutic factor VIII in patients with hemophilia A FVIII inhibitors develop in 5 to 30% of patients depending on the severity of hemophilia A. The abnormality in the F8 gene responsible for the disease is the strongest predictor of alloimunization against therapeutic FVIII. In particular, patients with severe hemophilia A, and among them patients lacking circulating FVIII:Ag, have the highest incidence of FVIII inhibitor development.7 Conversely, patients with missense mutations have the lowest risk of developing FVIII inhibitors. This correlation illustrates the importance of the degree of education, either centrally in the thymus or at the periphery, of the immune system of the patients towards endogenous FVIII: the more the endogenously produced FVIII resembles the exogenously administered therapeutic FVIII, the lower the risk of developing a neutralizing immune response.

The situation is not, however, so simple. For instance, some missense mutations responsible for mild/moderate hemophilia A are associated with rates of inhibitor development as elevated as those seen in FVIII:Ag-negative patients wih severe hemophilia A.61 This highlights the existence of confounding parameters such as the impaired secretion of some forms of mutated FVIII consecutive to their retention in the endoplasmic reticulum,62 thus leading to poor antigen presentation resulting in absence of tolerance induction. Alternatively, missense mutations may create T-cell epitopes in the mutated endogenous FVIII which differ from those of therapeutic FVIII.63-67 In the latter case, the education of the immune system does not “fit” with the exogenously administered FVIII and regulatory elements cannot be engaged in an adequate manner. More relevant to this review is the possibility that active tolerance to FVIII in treated patients may be implicated irrespective of the severity of the disease and irrespective of the presence of the endogenous (mutated) FVIII protein. This hypothesis is supported by several lines of evidence.

Immunological assessment of therapeutic factor VIII in patients with hemophilia A About half of all FVIII inhibitors that form have a low inhibitory titer, are clinically insignificant and usually disappear spontaneously.68 This is illustrated by the recent 240

SIPPET study in which 27% of the included patients with severe hemophilia A developed transient FVIII inhibitors in the initial months following first exposure to therapeutic FVIII.69 Immune recognition of FVIII in patients with severe hemophilia A is thus not necessarily pathogenic and can be controlled. In agreement, the presence of nonneutralizing anti-FVIII IgG has been described in inhibitornegative patients as well as in patients after successful immune tolerance induction therapy.54 While the risk of inhibitor development in patients with severe hemophilia A is highest during the first 20 cumulated exposure days (i.e., within the first 3-4 years of life),70 an increased incidence of FVIII inhibitors at an older age has been reported.71 Incidentally, with age, tolerance mechanisms weaken and there is an increased risk of developing autoimmune manifestations.72 This observation pleads in favor of the existence of active tolerance to therapeutic FVIII in inhibitor-negative patients, which can be lost with age. Importantly, immune tolerance induction therapy successfully eradicates FVIII inhibitors in 60-70% of inhibitor-positive patients. Although the induction of active tolerance to FVIII during immune tolerance induction therapy has, to our knowledge, never been demonstrated formally, the disappearance of detectable inhibitors and restoration of normal FVIII pharmacokinetics are associated in some patients with the persistence of FVIII-specific T cells,73 anti-FVIII IgG with neutralizing potential and concomitant induction of blocking anti-idiotypic antibodies.74 Lastly, the possible involvement of CD4+CD25+ regulatory T cells during an ongoing inhibitory anti-FVIII immune response was suggested in a patient with hemophilia A.64 Thus, tolerance to therapeutic FVIII does not merely rely on the elimination of FVIII-specific immune cells/antibodies. Collectively, these findings suggest the existence of an interplay between allogenic FVIIIspecific immune cells/molecules and regulatory elements of the immune system, which is reminiscent of the homeostatic immune assessment of endogenous FVIII that prevails under physiological conditions (illustrated in Figure 1 for patients with severe hemophilia A).

Exacerbated immune response or failed tolerance?

We observed a few years ago that a GT repeat polymorphism in the promoter of the gene encoding the antiinflammatory enzyme heme-oxygenase-1 (HO-1) is associated with inhibitor development in patients with severe hemophilia A.75 Thus, in a retrospective cohort of 300 patients, those who developed FVIII inhibitors had a significantly increased prevalence of longer GT stretches in the hmox1 gene promoter, known to impair the capacity of the cells to turn on expression of the gene.75 Conversely, patients with shorter GT repeats had a lower incidence of FVIII inhibitors. More recently, Matino et al. elegantly demonstrated that dendritic cells from patients who have developed FVIII inhibitors are less prone to express indoleamine 2,3-dioxygenase 1 (IDO1), following in vitro stimulation by CpG.76 IDO1 is a key regulatory enzyme involved in the degradation of tryptophan which supports regulatory T-cell functions and peripheral tolerance in adult life. Taken together with the observation that polymorphisms in promoters of the pro/anti-inflammatory genes encoding tumor necrosis factor-alpha, interleukin-10 and cytotoxic T-lymphocyte antigen-4 are associated with FVIII inhibitor development,11-14 these findings suggest that the capacity of patients to induce the expression of the endogehaematologica | 2019; 104(2)

Immune response to FVIII in hemophilia patients

nous anti-inflammatory machinery and/or regulatory pathways plays a key role in the ability of their immune system to control the response to the administered therapeutic FVIII. In order to test our hypothesis, a re-evaluation of existing cohorts of patients with hemophilia A would help to investigate whether patients who have developed a neutralizing anti-FVIII immune response have a generally more potent capacity to mount immune responses after vaccination, or are more resistant to common viral or bacterial infections, than inhibitor-negative patients.

Reconsidering the immune response to therapeutic factor VIII In the last 15 years, attempts to elucidate the reason(s) for which 5-30% of patients with hemophilia A develop an immune response to therapeutic FVIII, while 70-95% of them do not, and to decipher the nature of the danger

signals that are adjuvants to the inhibitory anti-FVIII immune response have remained largely inconclusive.

Observations over the years demonstrate that endogenous FVIII is not ignored by the immune system under physiological conditions but is immunologically assessed, leading to its homeostatic recognition based on counteracting reactive and tolerogenic adaptive immune effectors. Nonneutralizing FVIII-reactive antibodies are found in virtually all patients with hemophilia A; clinically irrelevant inhibitory antibodies are transiently detected in a substantial number of patients after their first exposures to therapeutic FVIII; and FVIII inhibitors may arise at an age when immune regulatory mechanisms are compromised (Figure 1). Furthermore, the development of FVIII inhibitors is associated with a hampered ability of the organism to activate HO-1 or IDO1-dependent anti-inflammatory/ tolerogenic mechanisms.

Figure 2. Immune response to factor VIII in patients with severe hemophilia A. (A) Classically, deciphering the immunogenicity of therapeutic factor VIII (FVIII) in patients with severe hemophilia A is addressed by investigating the reasons for which 20-30% of the patients develop an immune response to exogenous FVIII, while the remaining patients do not (adapted from Gouw et al.).24 (B) We propose that the immune system of all patients treated with FVIII reacts to the clotting factor. In 70-80% of the cases, this immunological assessment of therapeutic FVIII results in the establishment of active immune tolerance that relies on an equilibrium between FVIII-specific immune effector and regulatory elements. However, for reasons that are yet to be deciphered, the remaining patients fail to turn on appropriate anti-inflammatory/tolerogenic mechanisms. Anti-FVIII IgG: anti-factor VIII immunoglobulin G; aFVIII T cell: factor VIII-reactive T cell; Treg: regulatory T cell; anti-Id IgG: anti-idiotypic immunoglobulin G.

haematologica | 2019; 104(2)

241

A. Varthaman et al.

Collectively, the accumulated evidence favors the hypothesis that all hemophilia A patients treated with FVIII develop an immune response to therapeutic FVIII: mounting an immune response to exogenous FVIII is part of the normal recognition and assessment of an innocuous antigen. In this context, the immunological assessment of therapeutic FVIII does not require the presence of overt danger signals. The maturation of antigen-presenting cells leading to the activation of naïve T cells specific for therapeutic FVIII is triggered in a bystander manner by a dynamic immune system confronted with an everchanging environment. Furthermore, we propose that the onset of neutralizing antibodies to FVIII in a substantial number of patients results from these patients’ inability to develop a counteractive antigen-specific tolerogenic response, rather than from an exacerbated activation of the immune system at the time of FVIII administration.

Consequently, instead of investigating why 5-30% of the patients develop an inhibitory response to FVIII, the scientific challenge is to understand why 5-30% of the patients fail to develop active immune tolerance to FVIII (depicted in Figure 2 for patients with the severe form of the disease). Such a change of perspective will have repercussions on the clinical management of patients. For instance, in inhibitor-positive patients the use of immunosuppressive drugs that favor the onset of T-cellmediated tolerance77,78 should be privileged over that of drugs that simply shut off adaptive immunity. It will also foster the development of strategies to achieve FVIII-specific immuno-tolerance. Of importance, prophylaxis with emicizumab, a recently marketed FVIII-bypassing agent, in previously untreated patients79 or in inhibitor-positive patients80 allows correction of coagulation while avoiding stimulation of the immune system with FVIII. It thus inherently prevents the immunological (re-)assessment of therapeutic FVIII and establishment of active tolerance.

Consequences on the incidence of inhibitor development when the patients must receive therapeutic FVIII in the future will have to be monitored.

References 1. Meeks SL, Cox CL, Healey JF, et al. A major determinant of the immunogenicity of factor VIII in a murine model is independent of its procoagulant function. Blood. 2012;120 (12):2512-2520.

2. Levinson B, Kenwrick S, Gamel P, Fisher K, Gitschier J. Evidence for a third transcript from the human factor VIII gene. Genomics.

1992;14(3):585-589.

3. Kannicht C, Ramstrom M, Kohla G, et al.

Characterisation of the post-translational modifications of a novel, human cell linederived recombinant human factor VIII. Thromb Res. 2013;131(1):78-88.

4. Lin Y, Yang X, Chevrier MC, et al.

Relationships between factor VIII:Ag and factor VIII in recombinant and plasmaderived factor VIII concentrates.

Haemophilia. 2004;10(5):459-469.

5. Gringeri A, Mantovani LG, Scalone L, Mannucci PM. Cost of care and quality of life for patients with hemophilia complicated by inhibitors: the COCIS Study Group.

242

Conclusions Studies in the past 15 years have focused on deciphering the signals that lead to the development of a pathogenic anti-FVIII immune response in 5-30% of patients with hemophilia A, as compared to the therapeutic FVIII being ‘ignored’ in the remaining patients receiving therapy. To our knowledge, there is no evidence supporting the hypothesis that therapeutic FVIII is ignored by the immune system in previously treated inhibitor-negative patients. Based on the available experimental findings and clinical observations, we propose that therapeutic FVIII is not ignored but is assessed by the immune system of all the patients exposed to this treatment. In most patients, immune FVIII assessment is accompanied by the induction of antigen-specific counteractive tolerance mechanisms, which ensure that neutralizing anti-FVIII responses are not incited. In a subset of patients, however, the inability to develop such tolerance leads to the onset of clinically relevant levels of neutralizing anti-FVIII antibodies. A better understanding of the counteractive tolerogenic response in play and of the genetic/environmental factors that prevent its induction in inhibitor-positive patients will have repercussions on the clinical management of patients and highlight new strategies to achieve active immune tolerance towards therapeutic FVIII. Acknowledgments This work was supported by INSERM, Centre National de la Recherche Scientifique and Sorbonne Université (Paris, France) and by the Innovative Medicines Initiative Joint Undertaking ABIRISK (Anti-Biopharmaceutical Immunization Risk) project under grant agreement #115303, the resources of which comprise financial contribution from the European Union's Seventh Framework Program (FP7/2007-2013) and in-kind contributions from EFPIA companies. AV was a recipient of an individual fellowship from the Marie Sklodowska-Curie Actions (grant number: 656757 - Tolltum - MSCA-IF-EF-ST). We thank Dr Kathleen Pratt for her critical reading of the manuscript.

Blood. 2003;102(7):2358-2363.

6. Astermark J, Oldenburg J, Escobar M, White GC 2nd, Berntorp E. The Malmo International Brother Study (MIBS). Genetic defects and inhibitor development in siblings with severe hemophilia A. Haematologica. 2005;90(7):924-931.

7. Oldenburg J, El-Maarri O, Schwaab R. Inhibitor development in correlation to factor VIII genotypes. Haemophilia. 2002;8 Suppl 2:23-29.

8. Gouw SC, van den Berg HM, Oldenburg J, et al. F8 gene mutation type and inhibitor development in patients with severe hemophilia A: systematic review and meta-analysis. Blood. 2012;119(12):2922-2934.

9. Oldenburg J, Picard JK, Schwaab R, Brackmann HH, Tuddenham EG, Simpson E. HLA genotype of patients with severe haemophilia A due to intron 22 inversion with and without inhibitors of factor VIII. Thromb Haemost. 1997;77(2):238-242.

10. Hay CR, Ollier W, Pepper L, et al. HLA class II profile: a weak determinant of factor VIII inhibitor development in severe haemophilia A. UKHCDO Inhibitor Working Party.

Thromb Haemost. 1997;77(2):234-237.

11. Astermark J, Oldenburg J, Carlson J, et al.

Polymorphisms in the TNFA gene and the risk of inhibitor development in patients with hemophilia A. Blood. 2006;108(12): 3739-3745.

12. Astermark J, Oldenburg J, Pavlova A, Berntorp E, Lefvert AK. Polymorphisms in the IL10 but not in the IL1beta and IL4 genes are associated with inhibitor development in patients with hemophilia A. Blood. 2006; 107(8):3167-3172.

13. Astermark J, Wang X, Oldenburg J, Berntorp E, Lefvert AK. Polymorphisms in the CTLA4 gene and inhibitor development in patients with severe hemophilia A. J Thromb Haemost. 2007;5(2):263-265.

14. Pavlova A, Delev D, Lacroix-Desmazes S, et al. Impact of polymorphisms of the major histocompatibility complex class II, interleukin-10, tumor necrosis factor-alpha and cytotoxic T-lymphocyte antigen-4 genes on inhibitor development in severe hemophilia A. J Thromb Haemost. 2009;7(12):20062015.

15. Matzinger P. The danger model: a renewed

haematologica | 2019; 104(2)

Immune response to FVIII in hemophilia patients

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

sense of self. Science. 2002;296(5566):301305.

Navarrete A, Dasgupta S, Delignat S, et al.

Splenic marginal zone antigen-presenting cells are critical for the primary allo-immune response to therapeutic factor VIII in hemophilia A. J Thromb Haemost. 2009;7(11): 1816-1823.

Lai JD, Cartier D, Hartholt RB, et al. Early cellular interactions and immune transcriptome profiles in human factor VIII-exposed hemophilia A mice. J Thromb Haemost.

2018;16(3):533-545.

Dasgupta S, Navarrete AM, Bayry J, et al. A role for exposed mannosylations in presentation of human therapeutic self-proteins to CD4+ T lymphocytes. Proc Natl Acad Sci U S A. 2007;104(21):8965-8970.

Herczenik E, van Haren SD, Wroblewska A, et al. Uptake of blood coagulation factor VIII by dendritic cells is mediated via its C1 domain. J Allergy Clin Immunol. 2012;129 (2):501-509.

Janeway CA, Jr. Approaching the asymptote? Evolution and revolution in immunology. Cold Spring Harb Symp Quant Biol.

1989;54 (Pt 1):1-13.

Matzinger P. Tolerance, danger, and the extended family. Annu Rev Immunol.

1994;12:991-1045.

Astermark J, Altisent C, Batorova A, et al.

Non-genetic risk factors and the development of inhibitors in haemophilia: a comprehensive review and consensus report.

Haemophilia. 2010;16(5):747-766.

Reipert BM, Ahmad RU, Turecek PL, et al.

Characterization of antibodies induced by human factor VIII in a murine knockout model of hemophilia A. Thromb Haemost.

2000;84(5):826-832.

Gouw SC, van den Berg HM, Fischer K, et al.

Intensity of factor VIII treatment and inhibitor development in children with severe hemophilia A: the RODIN study.

Blood. 2013;121(20):4046-4055.

Dasgupta S, Repesse Y, Bayry J, et al. VWF protects FVIII from endocytosis by dendritic cells and subsequent presentation to immune effectors. Blood. 2007;109(2):610612.

Pradeu T, Jaeger S, Vivier E. The speed of change: towards a discontinuity theory of immunity? Nat Rev Immunol. 2013;13(10): 764-769.

Ing M, Gupta N, Teyssandier M, et al.

Immunogenicity of long-lasting recombinant factor VIII products. Cell Immunol.

2016;301:40-48.

Teyssandier M, Andre S, Gupta N, et al.

Therapeutic factor VIII does not trigger TLR1.2 and TLR2.6 signalling in vitro.

Haemophilia. 2013;19(3):399-402.

Pfistershammer K, Stockl J, Siekmann J, Turecek PL, Schwarz HP, Reipert BM. Recombinant factor VIII and factor VIII-von Willebrand factor complex do not present danger signals for human dendritic cells.

Thromb Haemost. 2006;96(3):309-316.

Cirino G, Cicala C, Bucci MR, Sorrentino L, Maraganore JM, Stone SR. Thrombin functions as an inflammatory mediator through activation of its receptor. J Exp Med. 1996;183(3):821-827.

Skupsky J, Zhang AH, Su Y, Scott DW. A role for thrombin in the initiation of the immune response to therapeutic factor VIII. Blood. 2009;114(21):4741-4748.

Gangadharan B, Delignat S, Ollivier V, et al.

Role of coagulation-associated processes on factor VIII immunogenicity in a mouse model of severe hemophilia A. J Thromb

haematologica | 2019; 104(2)

Haemost. 2014;12(12):2065â&#x20AC;&#x201C;2069.

33. Lai JD, Moorehead PC, Sponagle K, et al.

Concurrent influenza vaccination reduces anti-FVIII antibody responses in murine hemophilia A. Blood. 2016;127(26):34393449.

34. Platokouki H, Fischer K, Gouw SC, et al.

Vaccinations are not associated with inhibitor development in boys with severe haemophilia A. Haemophilia. 2018;24(2): 283-290.

35. Dimitrov JD, Dasgupta S, Navarrete AM, et al. Induction of heme oxygenase-1 in factor VIII-deficient mice reduces the immune response to therapeutic factor VIII. Blood.

2010;115(13):2682-2685.

36. Peyron I, Dimitrov JD, Delignat S, et al.

Haemarthrosis and arthropathy do not favour the development of factor VIII inhibitors in severe haemophilia A mice.

Haemophilia. 2015;21(1):e94-98.

37. Kurnik K, Bidlingmaier C, Engl W, Chehadeh H, Reipert B, Auerswald G. New early prophylaxis regimen that avoids immunological danger signals can reduce FVIII inhibitor development. Haemophilia.

2009;16(2):256-262.

38. Auerswald G, Bidlingmaier C, Kurnik K. Early prophylaxis/FVIII tolerization regimen that avoids immunological danger signals is still effective in minimizing FVIII inhibitor developments in previously untreated patients--long-term follow-up and continuing experience. Haemophilia. 2012;18(1): e18-20.

39. Bexborn F, Andersson PO, Chen H, Nilsson B, Ekdahl KN. The tick-over theory revisited: formation and regulation of the soluble alternative complement C3 convertase (C3(H2O)Bb). Mol Immunol. 2008;45(8): 2370-2379.

40. Zewde N, Gorham RD, Jr., Dorado A, Morikis D. Quantitative modeling of the alternative pathway of the complement system. PLoS One. 2016;11(3):e0152337.

41. Rayes J, Ing M, Delignat S, et al.

Complement C3 is a novel modulator of the anti-factor VIII immune response.

Haematologica. 2018;103(2):351-360.

42. Lacroix-Desmazes S, Navarrete AM, Andre S, Bayry J, Kaveri SV, Dasgupta S. Dynamics of factor VIII interactions determine its immunologic fate in hemophilia A. Blood.

2008;112(2):240-249.

43. Sakaguchi S, Vignali DA, Rudensky AY, Niec RE, Waldmann H. The plasticity and stability of regulatory T cells. Nat Rev Immunol.

2013;13(6):461-467.

44. Wang X, Terhorst C, Herzog RW. In vivo induction of regulatory T cells for immune tolerance in hemophilia. Cell Immunol.

2016;301:18-29.

45. Hu GL, Okita DK, Diethelm-Okita BM, Conti-Fine BM. Recognition of coagulation factor VIII by CD4+ T cells of healthy humans. J Thromb Haemost. 2003;1(10): 2159-2166.

46. Reding M, Wu H, Krampf M, et al.

Sensitization of CD4+ T cells to coagulation factor VIII: response in congenital and acquired hemophilia patients and in healthy subjects. Thromb Haemost. 2000;84(4):643652.

47. Meunier S, Menier C, Marcon E, LacroixDesmazes S, Maillere B. CD4 T cells specific for factor VIII are present at high frequency in healthy donors and comprise naĂŻve and memory cells. blood advances.

2017;1(21):1842-1847.

48. Delluc S, Ravot G, Maillere B. Quantitative analysis of the CD4 T-cell repertoire specific

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

to therapeutic antibodies in healthy donors.

FASEB J. 2011;25(6):2040-2048.

Delluc S, Ravot G, Maillere B. Quantification of the preexisting CD4 T-cell repertoire specific for human erythropoietin reveals its immunogenicity potential. Blood.

2010;116(22):4542-4545.

Hamze M, Meunier S, Karle A, et al.

Characterization of CD4 T cell epitopes of infliximab and rituximab identified from healthy donors. Front Immunol. 2017;8:500.

Pinto S, Michel C, Schmidt-Glenewinkel H, et al. Overlapping gene coexpression patterns in human medullary thymic epithelial cells generate self-antigen diversity. Proc Natl Acad Sci USA. 2013;110(37):E34973505.

Kamate C, Lenting PJ, van den Berg HM, Mutis T. Depletion of CD4+/CD25high regulatory T cells may enhance or uncover factor VIII-specific T-cell responses in healthy individuals.

J Thromb Haemost.

2007;5(3):611-613.

Algiman M, Dietrich G, Nydegger U, Boieldieu D, Sultan Y, Kazatchkine M. Natural antibodies to factor VIII (anti-hemophilic factor) in healthy individuals. Proc Natl Acad Sci USA. 1992;89(9):3795-3799.

Whelan SF, Hofbauer CJ, Horling FM, et al.

Distinct characteristics of antibody responses against factor VIII in healthy individuals and in different cohorts of hemophilia A patients. Blood. 2013;121(6):1039-1048.

Hofbauer CJ, Whelan SF, Hirschler M, et al.

Affinity of FVIII-specific antibodies reveals major differences between neutralizing and nonneutralizing antibodies in humans.

Blood. 2015;125(7):1180-1188.

Gilles JG, Saint-Remy JM. Healthy subjects produce both anti-factor VIII and specific anti-idiotypic antibodies. J Clin Invest.

1994;94(4):1496-1505.

Sakurai Y, Shima M, Tanaka I, Fukuda K, Yoshida K, Yoshioka A. Association of antiidiotypic antibodies with immune tolerance induction for the treatment of hemophilia A with inhibitors. Haematologica. 2004;89(6): 696-703.

Dietrich G, Algiman M, Sultan Y, Nydegger UE, Kazatchkine MD. Origin of anti-idiotypic activity against anti-factor VIII autoantibodies in pools of normal human immunoglobulin G (IVIg). Blood. 1992;79 (11):2946-2951.

Moreau A, Lacroix-Desmazes S, Stieltjes N, et al. Antibodies to the FVIII light chain that neutralize FVIII procoagulant activity, are present in the plasma of non-responder patients with severe hemophilia A and healthy blood donors. Blood. 2000;95(11): 3435-3441.

Rossi F, Sultan Y, Kazatchkine MD. Antiidiotypes against autoantibodies and alloantibodies to Factor VIII:C (anti-haemophilic factor) are present in therapeutic polyspecific normal immunoglobulins. Clin Exp Immunol. 1988;74(2):311-316.

Eckhardt CL, van Velzen AS, Peters M, et al.

Factor VIII gene (F8) mutation and risk of inhibitor development in nonsevere hemophilia A. Blood. 2013;122(11):1954-1962.

Malhotra JD, Miao H, Zhang K, et al.

Antioxidants reduce endoplasmic reticulum stress and improve protein secretion. Proc Natl Acad Sci U S A. 2008;105(47):1852518530.

Jacquemin M, Vantomme V, Buhot C, et al.

CD4+ T-cell clones specific for wild-type factor VIII: a molecular mechanism responsible for a higher incidence of inhibitor formation in mild/moderate hemophilia A.

243

A. Varthaman et al.

Blood. 2003;101(4):1351-1358.

64. James EA, Kwok WW, Ettinger RA, Thompson AR, Pratt KP. T-cell responses over time in a mild hemophilia A inhibitor subject: epitope identification and transient immunogenicity of the corresponding selfpeptide. J Thromb Haemost. 2007;5(12): 2399-2407.

65. Ettinger RA, James EA, Kwok WW, Thompson AR, Pratt KP. HLA-DR-restricted T-cell responses to factor VIII epitopes in a mild haemophilia A family with missense substitution A2201P. Haemophilia. 2010;16 (102):44-55.

66. James EA, van Haren SD, Ettinger RA, et al.

T-cell responses in two unrelated hemophilia A inhibitor subjects include an epitope at the factor VIII R593C missense site. J Thromb Haemost. 2011;9(4):689-699.

67. Pashov AD, Calvez T, Gilardin L, et al. In silico calculated affinity of FVIII-derived peptides for HLA class II alleles predicts inhibitor development in haemophilia A patients with missense mutations in the F8 gene. Haemophilia. 2013;20(2):176-184.

68. Hemophilia.org. https://http://www.hemophilia.org/Researchers-HealthcareProviders/Medical-and-Scientific-AdvisoryCouncil-MASAC/MASAC-Recommendations/MASAC-Recommendation-

244

69.

70.

71.

72.

73.

74.

On-SIPPET-Survey-of-Inhibitors-in-PlasmaProduct-Exposed-Toddlers.

Peyvandi F, Mannucci PM, Garagiola I, et al.

A randomized trial of factor VIII and neutralizing antibodies in hemophilia A. N Engl J Med. 2016;374(21):2054-2064.

Gouw SC, van der Bom JG, van den Berg HM. Treatment-related risk factors of inhibitor development in previously untreated patients with hemophilia A: the CANAL cohort study. Blood. 2007;109(11):46484654.

Hay CR, Palmer B, Chalmers E, et al.

Incidence of factor VIII inhibitors throughout life in severe hemophilia A in the United Kingdom. Blood. 2011;117(23): 6367-6370.

Jagger A, Shimojima Y, Goronzy JJ, Weyand CM. Regulatory T cells and the immune aging process: a mini-review. Gerontology.

2014;60(2):130-137.

Pautard B, D'Oiron R, Li Thiao Te V, et al.

Successful immune tolerance induction by FVIII in hemophilia A patients with inhibitor may occur without deletion of FVIII-specific T cells. J Thromb Haemost. 2011;9(6):11631170.

Gilles JG, Desqueper B, Lenk H, Vermylen J, Saint-Remy JM. Neutralizing antiidiotypic antibodies to factor VIII inhibitors after

75.

76.

77.

78.

79.

80.

desensitization in patients with hemophilia A. J Clin Invest. 1996;97(6):1382-1388.

Dimitrov JD, Roumenina LT, Plantier JL, et al. A human FVIII inhibitor modulates FVIII surface electrostatics at a VWF-binding site distant from its epitope. J Thromb Haemost.

2010.

Matino D, Gargaro M, Santagostino E, et al.

IDO1 suppresses inhibitor development in hemophilia A treated with factor VIII. J Clin Invest. 2015;125(10):3766-3781.

Moghimi B, Sack BK, Nayak S, Markusic DM, Mah CS, Herzog RW. Induction of tolerance to factor VIII by transient co-administration with rapamycin. J Thromb Haemost. 2011;9(8):1524-1533.

Hassan S, Fijnvandraat K, van der Bom JG, Gouw SC. Preventing or eradicating factor VIII antibody formation in patients with hemophilia A: what can we learn from other disorders? Semin Thromb Hemost. 2018;44 (6):531-543.

Mahlangu J, Oldenburg J, Paz-Priel I, et al.

Emicizumab prophylaxis in patients who have hemophilia A without inhibitors. N Engl J Med. 2018;379(9):811-822.

Oldenburg J, Mahlangu JN, Kim B, et al.

Emicizumab prophylaxis in hemophilia A with inhibitors. N Engl J Med. 2017;377(9): 809-818.

haematologica | 2019; 104(2)

ARTICLE

Hematopoiesis

Rheb1 loss leads to increased hematopoietic stem cell proliferation and myeloid-biased differentiation in vivo

Ferrata Storti Foundation

Xiaomin Wang,1* Yanan Gao,1* Juan Gao,1* Minghao Li,1 Mi Zhou,2 Jinhong Wang,1 Yakun Pang,1 Hui Cheng,1 Chase Yuan,3 Yajing Chu,1 Yu Jiang,4 Jianfeng Zhou,2 Hongbo R. Luo,1,5 Zhenyu Ju,6 Tao Cheng1 and Weiping Yuan1

State Key Laboratory of Experimental Hematology, Institute of Hematology and Blood Diseases Hospital, Center for Stem Cell Medicine, Chinese Academy of Medical Sciences and Peking Union Medical College, Tianjin, China; 2Department of Hematology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, Hubei, China; 3College of Arts and Sciences, University of North Carolina at Chapel Hill, NC, USA; 4Department of Pharmacology and Chemical Biology, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA; 5Department of Pathology, Harvard Medical School, Dana-Farber/Harvard Cancer Center, Boston, MA, USA and 6Institute of Aging, Hangzhou Normal University, Hangzhou, China 1

*

XM, YG and JG contributed equally to this work.

Haematologica 2019 Volume 104(2):245-255

ABSTRACT

H

ematopoietic stem cells constitute a unique subpopulation of blood cells that can give rise to all types of mature cells in response to physiological demands. However, the intrinsic molecular machinery that regulates this transformative property remains elusive. In this paper, we demonstrate that small GTPase Rheb1 is a critical regulator of proliferation and differentiation of hematopoietic stem cells in vivo. Rheb1 deletion led to increased phenotypic hematopoietic stem cell/hematopoietic progenitor cell proliferation under a steady state condition. Over-proliferating Rheb1-deficient hematopoietic stem cells were severely impaired in functional repopulation assays, and they failed to regenerate the blood system when challenged with hematopoietic ablation by sublethal irradiation. In addition, it was discovered that Rheb1 loss resulted in a lack of maturation of neutrophils / caused neutrophil immaturation by reducing mTORC1 activity, and that activation of the mTORC1 signaling pathway by mTOR activator 3BDO partially restored the maturation of Rheb1-deficient neutrophils. Rheb1 deficiency led to a progressive enlargement of the hematopoietic stem cell population and an eventual excessive myeloproliferation in vivo, including an overproduction of peripheral neutrophils and an excessive expansion of extramedullary hematopoiesis. Moreover, low RHEB expression was correlated with poor survival in acute myeloid leukemia patients with normal karyotype. Our results, therefore, demonstrate a critical and unique role for Rheb1 in maintaining proper hematopoiesis and myeloid differentiation.

Introduction Hematopoietic stem cell (HSC) proliferation and differentiation are regulated by the networks of signaling pathways.1 Under stress conditions, HSCs quickly respond to a variety of proliferative stimuli, exit the quiescent phase, and undergo a period of self-renewal and differentiation to restore hematopoietic homeostasis.2 Fine-tuning is required to adequately control cell growth and quiescence during this process, and multiple signaling pathways are involved in maintaining the correct balance. Interestingly, studies have suggested that common or similar mechanisms regulate stem cell properties in both HSCs and leukemia stem cells (LSCs), suggesting that LSCs may originate from HSCs. The mammalian target of the rapamycin (mTOR) signaling pathway is a key node in these pathways, and plays an essential role in regulating normal and malignant hematopoiesis.3,4 Ras homolog enriched in brain (Rheb), a small GTPase, is known to directly actihaematologica | 2019; 104(2)

Correspondence: wpyuan@ihcams.ac.cn or chengtao@ihcams.ac.cn or zhenyuju@163.com Received: April 7, 2018.

Accepted: September 21, 2018.

Pre-published: September 27, 2018. doi:10.3324/haematol.2018.194811 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/245 Š2019 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.

245

X. Wang et al.

vate mTORC1.5,6 Rheb binds to mTORC1 at the lysosome and activates it, while TSC1-TSC2-TBC1D7, a GTPase-activating protein, inhibits mTORC1 activity by reducing GTP-bound Rheb.7-9 mTORC1 is a serine/threonine (Ser/Thr)-protein kinase complex that responds to multiple signals, regulates cell biological activities, and maintains homeostasis.10 Dysregulation of the mTORC1 signaling pathway results in myeloproliferative neoplasms (MPN) and, in some cases, acute leukemia, making mTORC an important target for therapeutic treatments.10 Perturbation of mTORC1 in adult mouse HSCs by Raptor deletion results in the depletion of the long-term reconstitution ability of HSCs.11 Deletion of TSC1, an upstream negative regulator of mTORC1, causes defective HSC cycling and adversely affects HSC function in mice due to enhanced mTORC1 activity,12 indicating that the level of mTORC1 activity needs to be precisely regulated.

Changes in mTORC1 activity in the hematopoietic system under various conditions alters HSC function and homeostasis.

Rheb1 can also regulate cell proliferation and apoptosis via interaction with many other signaling pathways, such as B-Raf, Notch, small Rho GTPase and Akt signaling pathways. Knockdown of Rheb1 in a TSC2-null, angiomyolipoma-derived cell line decreased Notch activity, suggesting that Notch is a downstream target of Rheb1.

However, Notch activation could not be blocked by rapamycin, the mTORC1 inhibitor.13 Rheb1 was also reported to directly interact with the B-Raf kinase in a rapamycin-resistant manner and inhibit its function, resulting in interference with H-Ras-induced transformation in NIH3T3 cells.14 In addition, Rheb1 interacts with FKBP38 and regulates apoptosis in a rapamycin-insensitive, but amino acid- and serum-sensitive manner.15 We have previously reported that Rheb1 plays a crucial role in myeloid development. The expression of Rheb1 is high in myeloid progenitor, and is down-regulated during granulocyte differentiation. Rheb1 deletion interferes with myeloid progenitor progression and gene expression.16 However, ongoing studies have not directly addressed the specific regulatory role of Rheb1 in hematopoietic stem cells.

In this study, we observed that Rheb1 is an essential regulator of hematopoietic development. Rheb1-deficient mice showed increased phenotypic HSCs, immature neutrophils in bone marrow (BM), and splenomegaly. These phenotypes are reminiscent of the hematopoiesis seen in MPNs. Rheb1-deficient HSCs were defective in their ability to reconstitute the blood tissue and differentiate into normal neutrophils. Interestingly, low Rheb expression was associated with poor survival in acute myeloid leukemia (AML) patients. Thus, our data indicate that Rheb is critical for HSC function and may be involved in the initiation of myeloid proliferation-related diseases or MPN-like disorders.

Methods Mice and genotyping Rheb1fl/fl mice were kindly provided by Dr. Bo Xiao.17 Transgenic mice expressing Cre recombinase under the control of the Vav1 promoter (Vav1 Cre) were purchased from the Jackson Lab. The Rheb1fl/fl mice were crossed with Vav1-Cre mice to generate specific deletion of Rheb1 in the hematopoietic system. All animal pro246

tocols were approved by the Institutional Animal Care and Use Committee (IACUC), the Institute of Hematology, and Blood Diseases Hospital (CAMS/PUMC). All surgery was performed under sodium pentobarbital anesthesia, and every effort was made to minimize mouse suffering.

Flow cytometry analysis Peripheral blood (PB) was obtained from either the tail veins or retro-orbital bleeding of mice. Red blood cells (RBCs) were lysed by ammonium chloride-potassium bicarbonate buffer before staining. BM cells were flushed out from tibias, femurs, and ilia by a 25-gauge needle with PBS supplemented with 2% fetal bovine serum (FBS) and 20 mM EDTA (abbreviated as PBE). Cells were stained with antibodies purchased from either eBioscience or BD Bioscience. To analyze intracellular proteins, 3x106 BM cells were labeled with surface antibodies, fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X100, then washed 2 times with 1 mL cold PBE. Finally, the cells were resuspended with cold PBS supplemented with 25% FBS, and intracellularly stained with antibodies: p-S6 (Ser24/244), p-4EBP1 (Thr37/46). Cells were analyzed

by BD Canto II flow cytometer. FlowJo software was used to analyze the results.

LKS transplant and analysis Whole BM cells (WBMCs) were obtained and Lin– cells were sorted using mouse lineage cell depletion kit (Miltenyi Biotec) according to the manufacturer’s instruction. LKSs were stained as mentioned above and sorted by BD Influx flow cytometer; 200 LKSs (CD45.1) together with 5x105 whole BM cells (WBMCs) (CD45.2) were injected intravenously into lethally irradiated recipient mice (CD45.2). The reconstitution of PB cells was analyzed every four weeks post transplantation. The recipient mice were sacrificed at four months after transplantation. The self-renewal and differentiation capacities of donor-derived HSCs derived from BM were then analyzed.

Competitive bone marrow transplantation and analysis Whole BM cells were isolated from the tibias, femurs and ilia of 8-week old Rheb1fl/fl (CD45.1) or Rheb1Δ/Δ mice (CD45.1). 5x105 Rheb1Δ/Δ WBMCs (CD45.1) together with 5x105 WBMCs (CD45.2) were intravenously injected into the lethally irradiated recipient mice (CD45.2). Then, the reconstituted PB cells were analyzed every four weeks after transplantation.

Lineage– cell homing assay Whole BM cells were obtained, and LKS+ cells (CD45.1) were sorted by flow cytometry. LKS+ cells were cultured with CFSE at 37°C for 8 minutes (min). The reaction was then terminated with 10％ FBS at 4°C for 2 min and washed two times with cold PBS. LKS+ cells (2x106) were intravenously injected into lethally irradiated (9.5 Gy) recipient mice (CD45.2). The recipient mice were sacrificed at 17 hours (h) or 24 h after transplantation. CFSE+ cells in BM of recipient mice were analyzed by FACS.

Histological and pathological analysis To examine the histology of the BM neutrophils, the neutrophils were sorted with CD11b and Ly-6G from BM, then cytospun and stained with Wright-Giemsa solution. For pathological analysis, BM, spleen, liver or lung were fixed with 4% formalin, embedded in paraffin, sectioned, and stained with hematoxylin and eosin.

In vitro bacterial killing assay Escherichia coli (strain 19138; ATCC, Manassas, VA, USA) were cultured overnight, suspended in PBS at an OD600 of 0.10, and haematologica | 2019; 104(2)

Rheb1 regulates HSC proliferation & differentiation

opsonized with 10% mouse serum for 30 min at 37ºC. Neutrophils and bacteria were then incubated together at a 1:5 ratio for 30 and 120 min at 37ºC with intermittent shaking, and then 100 g/mL gentamicin was added for an additional 30 min to eliminate extracellular bacteria. The cells were then lysed by adding distilled water, and subsequently diluted aliquots were spread on LB agar plates. The CFU was counted after incubating the plates overnight at 37ºC. A bacterial suspension without any cells was used as an input control.

Mouse colony-forming cell assay and 3BDO treatment 3BDO was dissolved in DMSO at 60 mM. WBMCs were treated with 3BDO (60 nM) or DMSO for 30 min.18 Then, GMP/CMPs (2x104 cells) were sorted and cultured in MethoCult® media (Catalog #03231) in the presence of SCF (50 ng/mL), IL3 (10 ng/mL), and GM-CSF (25 ng/mL). Colonies were counted after 5-7 days.

Serial colony-forming cell assay Whole BM cells were treated with 3BDO (60 nM) or DMSO for 30 min, then cells (1x105) were isolated and cultured in MethoCult® media (Catalog #03434). Colonies were counted seven days after plating, and serially replating. Colonies are defined as colony forming unit monocyte (CFU-M), granulocytemacrophage (CFU-GM), and granulocyte-eythroid-macrophage (CFU-GEM).

Cobble stone area-forming cell assay Mesenchymal stem cells (MSC) were obtained from mouse BM cells, as previously described.19 MSCs were cultured in long-term culture medium M5300 (Stem Cell Technologies, Vancouver, BC, Canada) for two weeks, and half of the medium was changed every 3-4 days. Cells were trypsinized, irradiated with 15 Gray, and plated at 5000 cells/well in 96-well plates. LKS+ cells (500 cells/well) were inoculated on the irradiated cell layers and incubated at 33°C. Colonies (Cobblestone areas) were counted five weeks after plating.

curves were created using GraphPad Prism 5. All data are presented as mean±Standard Error of Mean (SEM); n ≥3.

Results Rheb1 deletion induced hematopoietic stem cell / hematopoietic progenitor cells expansion in steady state condition To clarify the role of Rheb1 in hematopoiesis, Vav1cre;Rheb1fl/fl (Rheb1Δ/Δ) mice were generated and the cellular composition of the peripheral blood (PB) and BM was first analyzed in Rheb1fl/fl and Rheb1Δ/Δ mice. There was no significant difference in normalized values of white blood cells (WBC), RBCs, hemoglobin (Hgb), hematocrit (Hct) and platelets (PLT) in the PB between Rheb1fl/fl and Rheb1Δ/Δ mice, while absolute numbers of neutrophils were substantially higher in Rheb1Δ/Δ mice than those in Rheb1fl/fl mice (Online Supplementary Figure S1A). The absolute numbers of BM cells did not alter between Rheb1fl/fl and Rheb1Δ/Δ mice (Online Supplementary Figure S1B). The percentage and absolute numbers of LKS+ and LKS– cells were increased in Rheb1Δ/Δ mice (Online Supplementary Figure S1C and D). In the HSC-enriched LKS+ subsets, Rheb1 loss significantly increased the absolute numbers of CD150+CD48–LKS+ cells (Figure 1A and Online Supplementary Figure S1E), while in the hematopoietic progenitor cells (HPC)-enriched LKS– subsets, Rheb1 loss resulted in an increase in the absolute number of CD16/32–CD34+LKS– cells and CD16/32–CD34+LKS– cells (GMP and CMP) (Figure 1B and C, and Online Supplementary Figure S1F). More Rheb1Δ/Δ LKS+ cells were in G2/S/M phase of cell cycle than that of the control (Figure 1D and Online Supplementary Figure S1G), although the percentage of Annexin V-positive cells was not altered in Rheb1Δ/Δ LKS+ (Online Supplementary Figure S1H). Thus, Rheb1 deletion led to an expansion of murine HSC/HPCs.

Gene expression profiling and patient data analysis LKS+ cells were sorted from WBMCs by BD Influx flow cytometer. The sorted cells were treated with Trizol, and then sent to the Shanghai Biotechnology Corporation for microarray analysis (Agilent, mouse 4*44K). Analysis of gene expression data from Rheb1Δ/Δ and Rheb1fl/fl LKSs revealed 2515 differentially expressed genes, with 922 up-regulated and 1593 down-regulated genes in Rheb1-deficient LKSs. The differentially expressed genes were filtered as P<0.05 and fold change greater than 3. The gene-set enrichment data were analyzed using GSEA and GO (FDR <0.25, P<0.05, -Lg p = -Log10 p). The microarray data were deposited with the GEO under the accession number GSE79538. For the GSEA analysis of patient data, we examined AML-related data using Molecular Signatures Database (MSigDB) (http://software.broadinstitute.org/gsea/msigdb/search.jsp).

Quantitative real-time PCR (qRT-PCR) Total RNA from BM samples was extracted using the RNeasy Mini Kit (Qiagen). First-strand cDNA was synthesized with an oligo-(dT) primer according to the manufacturer’s instructions.

The mRNA expression was determined by qRT-PCR (Stepone Fast Real-Time PCR system; Applied Biosystems) using FastStart Universal SYBR Green PCR Master mix (Roche). GAPDH was used as endogenous controls for gene expression assays.

Statistical analyses Experimental results were analyzed using Student t-test. P<0.05 was considered significant for all tests. Kaplan-Meier survival haematologica | 2019; 104(2)

Rheb1 deletion caused neutrophil immaturity in steady condition Interestingly, the absolute numbers and percentages of myeloid cells (CD11b+) were increased in BM of Rheb1Δ/Δ mice (Online Supplementary Figure S2A and B), while no significant changes were found in B (B220+) and T (CD3+) cells in the BM of Rheb1fl/fl and Rheb1Δ/Δ mice (Online Supplementary Figure S2C). We then analyzed neutrophils by flow cytometry (FACS) with CD11b and Ly-6G antibodies that have been used as neutrophil subpopulation markers for the identification of myelocytes/promyelocytes, as well as immature and mature neutrophils. The CD11b+Ly-6G+ subpopulation of Rheb1Δ/Δ neutrophils shifted to the left in the PB and BM (Figure 1E). We divided neutrophils into three distinct subpopulations, indicated as the CD11blowLy-6Glow population (P1), the CD11bhigh Ly-6Glow population (P2), and the CD11b+ Ly-6Ghigh population (P3). We found that the percentages of the P1 and P2 subpopulations were increased, while the percentage of the P3 subpopulation was decreased both in the PB and BM of Rheb1Δ/Δ mice (Figure 1F and G). Morphological analysis of Rheb1fl/fl neutrophils from BM showed that the P1 subpopulation was comprised mainly of myeloblasts with oval-shaped nuclei and a wider, less basophilic cytoplasm. The segmentation of the nuclei became gradually evident in the P2 subpopulation while the P3 subpopulation was mainly composed of cells with doughnut-shaped 247

X. Wang et al.

A

B

D

E

H

C

F

I

G

J

Figure 1. Rheb1 deletion increases the number of hematopoietic stem cell/hematopoietic progenitor cells and immature neutrophils in bone marrow (BM) of Rheb1Δ/Δ mice. Whole BM cells were isolated from BM of tibias, femurs and ilia of 8-week old mice. (A) The absolute number of CD48–CD150+LKS+ population in the BM of Rheb1fl/fl and Rheb1Δ/Δ mice; n=3. (B) The absolute cell number of GMP, CMP and MEP populations in the BM of Rheb1fl/fl and Rheb1Δ/Δ mice, n=3. (C) FACS analysis of LKS+ cells and LKS– in the BM; n=3. (D) Cell cycle status of Rheb1fl/fl or Rheb1Δ/Δ LKS+; n=3. (E) FACS analysis of neutrophils in Rheb1fl/fl and Rheb1Δ/Δ PB and BM, n=3. (F and G) The percentages of neutrophil subpopulations in Rheb1fl/fl and Rheb1Δ/Δ PB and BM; n=3. (H) Representative images of Wright-Giemsa staining of the sorted P1-P3 subpopulations of neutrophils from BM. (I) qRT-PCR analysis for the mRNA expression of granule proteins in neutrophils. Elane: neutrophil elastase; Ltf: lactotransferrin precursor; Mpo: myeloperoxidase; n=3. (J) The E. coli survival rates after co-culturing with neutrophils; n=3. Data are presented as mean±Standard Error of Mean. *P<0.05; **P<0.01; ***P<0.001.

nuclei. Interestingly, the P3 subpopulation of Rheb1Δ/Δ neutrophils was composed of cells with butterfly-shaped nuclei (Figure 1H and Online Supplementary Figure S2D).

The expression levels of Ltf and Elane (encoding granule proteins) were greatly reduced in Rheb1Δ/Δ neutrophils from BM (Figure 1I), further indicating the reduced maturity of Rheb1Δ/Δ neutrophils. The bacterial survival assay showed that Rheb1Δ/Δ neutrophils killed less than 10% of the bacteria, while Rheb1fl/fl neutrophils killed more than 60% of the bacteria (Figure 1J). These results suggest that Rheb1 deficiency caused neutrophil immaturity.

Rheb1 deletion induced extramedullary hematopoiesis in the spleen

Physical examinations revealed that Rheb1Δ/Δ mice displayed splenomegaly as demonstrated by the increase of their spleen size and weight (Figure 2A and B).

Histopathological examination showed evident extramedullary hematopoiesis, including clustered

248

megakaryocytes (Figure 2C, arrowheads) and hematopoietic islands (Figure 2C, arrows) in Rheb1Δ/Δ mice. However, liver and lung examination showed no significant differences between Rheb1fl/fl and Rheb1Δ/Δ mice (Online Supplementary Figure S2E and F). FACS analysis confirmed splenic extramedullary hematopoiesis, characterized by an increase in the percentages of LKS+ (HSCs) and LKS– (HPCs) cells (Figure 2D and E) and a marked increase in the CD34+LKS– (GMP/CMPs) populations (Figure 2F). The proportion of myeloid cells was also increased, while those of T and B cells were reduced in Rheb1Δ/Δ spleens when compared to those of the control spleens (Figure 2G). Moreover, FACS showed an increase in the percentage of neutrophils in Rheb1Δ/Δ spleens (Figure 2H), in agreement with granulocytic expansion in the PB in comparison with the controls. Collectively, our results demonstrated that Rheb1 deficiency induces extramedullary hematopoiesis in the spleen to compensate the hematopoiesis defect in the BM. haematologica | 2019; 104(2)

Rheb1 regulates HSC proliferation & differentiation

A

B

C

D

E

G

H

F

Figure 2. Rheb1Δ/Δ mice display extramedullary hematopoiesis in the spleen. (A and B) Size and weight of spleens in 8-week old Rheb1fl/fl and Rheb1Δ/Δ mice. Data are presented as mean±Standard Error of Mean (SEM); n>6. (C) Hematoxylin & eosin-stained spleen sections of Rheb1fl/fl and Rheb1Δ/Δ mice. Arrowheads indicate megakaryocytes; arrows point to hematopoietic islands. (D-F) Percentages of LKS+ subsets, LKS– subsets and corresponding subpopulations in Rheb1fl/fl and Rheb1Δ/Δ spleens; n=3. (G) Percentages of B cells (B220+), T cells (CD3+) and myeloid cells (CD11b+) in Rheb1fl/fl and Rheb1Δ/Δ spleens; n=3. (H) Percentages of neutrophils in Rheb1fl/fl and Rheb1Δ/Δ spleens; n=3. Data are presented as mean±SEM. *P<0.05; **P<0.01; ***P<0.001.

Rheb1 deletion impaired HSC regeneration ability in transplant assay To assess the role of Rheb1 in HSC function during hematopoiesis, we transplanted Rheb1Δ/Δ BM cells (CD45.1) with competitive cells (CD45.2) into lethally irradiated recipient mice (CD45.2) to examine the role of Rheb1 in adult HSC reconstitution. All lineages derived from Rheb1Δ/Δ cells were significantly reduced in the PB and the BM after transplantation (Figure 3A and B, and Online Supplementary Figure S3A), while the homing capacity of transplanted Rheb1Δ/Δ CFSE+ labeled cells was equivalent to that of Rheb1fl/fl cells (Figure 3C).

In addition, Rheb1fl/fl and Rheb1Δ/Δ mice were subjected to 4 Gy of X-ray irradiation and we examined the effects of Rheb1 deficiency on HSCs under hematopoietic stress. Interestingly, all Rheb1fl/fl mice survived while most Rheb1Δ/Δ mice died, probably due to impaired recovery of the BM and thymic cellularity (Figure 3D-F) and impaired B lymphopoiesis in the BM and spleen (Figure 3G). The absolute number of Rheb1Δ/Δ HS/PCs was reduced comhaematologica | 2019; 104(2)

pared with Rheb1fl/fl HS/PCs at 21 days after 4 Gy irradiation (Online Supplementary Figure S3B). Thus, a delicate balance of developmental decisions for HSC homeostasis, including stem cell quiescence, self-renewal and differentiation, was maintained under steady state conditions in Rheb1Δ/Δ mice but was broken under hematopoietic stress.

We further examined the self-renewal and differentiation capacities of HSCs using LKS+ transplantation. A total of 200 LKS+ cells isolated from Rheb1fl/fl and Rheb1Δ/Δ mice (CD45.1) together with 5x105 WBMCs (CD45.2) were intravenously injected into lethally irradiated recipient mice. The chimerism in PB was analyzed every four weeks post transplantation. The repopulating capacity of Rheb1Δ/Δ LKS+ cells in the PB was significantly lower in recipient mice than the control LKS+ cells (Figure 3H). All lineages derived from Rheb1Δ/Δ HSCs were significantly reduced in the PB and the BM after transplantation (Online Supplementary Figure S3C). The recipient mice were sacrificed at four months after transplantation and 249

X. Wang et al.

A

E

I

B

F

C

D

G

J

H

K

L

Figure 3. Rheb1 deletion impaired hematopoietic stem cell (HSC) regeneration upon transplantation. (A) Donor-derived-cell chimerism in peripheral blood (PB) of mice transplanted with Rheb1fl/fl or Rheb1Δ/Δ whole bone marrow cells (WBMC); n=6. (B) Percentage of donor-derived-cells in the BM four months after mice were transplanted with Rheb1fl/fl or Rheb1Δ/Δ WBMCs; n=6. (C) Homing of CFSE+ 7-AAD- LKS+ cells to the BM 24 hours (h) post transplantation; n=3. (D) Kaplan-Meier survival curve of Rheb1fl/fl and Rheb1Δ/Δ mice that underwent sublethal irradiation; n=12. (E and F) Absolute number of cells (CD45.1+) in BM and thymus of Rheb1fl/fl and Rheb1Δ/Δ mice under sublethal irradiation; n=3. (G) Percentage of B cells (B220+ ) from the spleens and BM of mice under sublethal irradiation; n=3. (H) Donorderived-cell chimerism in the PB of mice transplanted with Rheb1fl/fl or Rheb1Δ/Δ LKS+ cells; n=6. (I) Percentage of donor-derived cells in the BM four months after mice were transplanted with Rheb1fl/fl and Rheb1Δ/Δ LKS+ cells; n=6. (J) Absolute number of donor-derived LKS+ cells in the BM per mouse; n=6. (K and L) Self-renewal and differentiation quotients of donor-derived LKS+ cells; n=6. Data are presented as mean±Standard Error of Mean. *P<0.05; **P<0.01;

their donor-derived HSCs in BM were analyzed. We found the donor-derived Rheb1Δ/Δ cells decreased significantly in the BM at four months after transplantation (Figure 3I). To quantify the function of HSCs, we calculated the LKS+ cell amplification and differentiation ability four months after transplantation.20 We found that the number of Rheb1Δ/Δ LKS+-derived HSCs was substantially lower than that of the control (Figure 3J). The selfrenewal quotient (the number of donor-derived HSCs recovered at the end of the transplant per original input HSC) was also significantly reduced in mice receiving Rheb1Δ/Δ LKS+ cells (Figure 3K), while there was no change in the differentiation quotient (WBC count/mL blood x the percentage test-cell blood chimerism/number of donor-derived HSCs) when compared with the control (Figure 3L). Rheb1 deficiency thus impaired the ability of HSCs to repopulate under hematopoietic stress due to reduced self-renewal capability.

250

Aged Rheb1Δ/Δ mice develop myeloproliferative disorders Since adult mice lacking Rheb1 have impaired HSC function and splenic extramedullary hematopoiesis, we went on to perform a detailed phenotypic analysis to investigate whether Rheb1 deletion leads to progressive hematopoiesis defects in aged Rheb1Δ/Δ mice up to two years of age. We found the survival time of aged Rheb1Δ/Δ mice was significantly shorter than that of littermate controls (Figure 4A). Furthermore, Rheb1Δ/Δ mice, but not Rheb1fl/fl littermates, had evident progressive leukocytosis.

Neutrophil count in PB was significantly increased in Rheb1Δ/Δ mice (Figure 4B). Peripheral blood analysis revealed that the percentage of neutrophils was also increased in Rheb1Δ/Δ mice (Online Supplementary Figure S4A). The number of RBC and Hgb was normal in Rheb1Δ/Δ mice while the number of PLTs was decreased in Rheb1Δ/Δ mice (Online Supplementary Figure S3B). Myeloproliferative haematologica | 2019; 104(2)

Rheb1 regulates HSC proliferation & differentiation

disorders were apparent in both BM and PB as defined by myeloid left shift (presence of blasts, promyelocytes, myelocytes, or metamyelocytes) (Figure 4C and D). In addition, Rheb1 deletion led to an expansion of HSCs and myeloid progenitors in BM (Figure 4E-G and Online Supplementary Figure S4C and D). Furthermore, we isolated BM cells and performed serial colony forming unit assays in vitro. We found the number of colonies formed by Rheb1Δ/Δ BM cells was increased when compared with

A

B

C

D

F

G

I

Rheb1fl/fl BM cells both in first plating experiment and second serial replating experiment (Figure 4H and Online Supplementary Figure S4E). These data demonstrated that loss of Rheb1 increased proliferation ability of hematopoietic progenitor.

Aged Rheb1Δ/Δ mice had enlarged spleens in comparison to the littermate controls. Histopathological analysis revealed that Rheb1Δ/Δ mice developed splenomegaly, consistent with significant myeloproliferation disorder

E

H

J

K

Figure 4. Rheb1 deletion leads to progressive myeloproliferation in bone marrow (BM) and extramedullary hematopoiesis in aged Rheb1Δ/Δ mice. (A) Survival curves of Rheb1fl/fl and Rheb1Δ/Δ mice (n=22). (B) The absolute number of different cell populations in peripheral blood (PB) of 2-year old Rheb1fl/fl and Rheb1Δ/Δ mice; n=3.

(C and D) FACS analysis of granulocytes according to the expression levels of CD11b and Ly-6G in PB and bone marrow (BM); n=3. (E) FACS analysis of LKS+ cells and LKS– in the BM; n=3. (F) The absolute number of LKS+ cells and LKS– in the BM; n=3. (G) The absolute number of GMP, CMP and MEP populations in the BM of 2-year old Rheb1fl/fl and Rheb1Δ/Δ mice; n=3. (H) Whole BM cells were isolated from 2-year old mice and plated in M3434 methylcellulose. Colonies were counted seven days after plating, and serially replating; n=3. (I) Hematoxylin & eosin-stained BM, spleen and liver sections of 2-year old Rheb1fl/fl and Rheb1Δ/Δ mice. (J) FACS analysis of LKS+ cells and LKS– in the spleen. (K) FACS analysis of granulocytes in the spleen. Data are presented as mean±Standard Error of Mean. *P<0.05; **P<0.01.

haematologica | 2019; 104(2)

251

X. Wang et al.

A

B

C

Figure 5. The low expression of RHEB is correlated with poor survival of acute myeloid leukemia (AML) patients with normal karyotype. (A) The percentage of lossof-function mutations of RHEB in AML patients (http://www.cbioportal.org/). (B) The expression of RHEB in two sets of high-risk versus low-risk AML patients (http://bioinformatica.mty.itesm.mx/SurvExpress; GSE5122 and GSE12417-GPL97). (C) Cumulative (Cum) overall survival of AML patients with different expression levels of RHEB (http://www.abren.net/PrognoScan-cgi/; GSE12417). Data are presented as mean±Standard Error of Mean. ***P<0.001.

(Online Supplementary Figure S4F). Histopathological analysis revealed that Rheb1Δ/Δ mice had not only prominent myeloproliferative features including extramedullary hematopoiesis in the spleen, but also infiltration of liver and neutrophilia in BM (Figure 4I). FACS analysis of the spleen revealed a significant enlargement of both the HSC (LKS+) and HPC populations (Figure 4J and Online Supplementary Figure S4G), and with granulocytic expansion (Figure 4K). Collectively, these data demonstrate that Rheb1 loss leads to progressive myeloproliferation disorder in vivo.

Low expression of RHEB was correlated with poor survival in AML patients with normal karyotype The phenotypes observed in Rheb1 loss and progressive myeloproliferative disorder prompted us to investigate if RHEB also plays a role in human leukemia or myeloproliferative diseases. We used a curated database (http://www.cbioportal.org/), which provides large-scale cancer genomics data sets, to analyze the mutations and copy number alteration (CNA) of RHEB in leukemia patients. It was found that deep deleted mutations in RHEB were about 3% (6 of 188) in an AML patient cohort (Figure 5A).

We also analyzed the mutation types in ICGA (international cancer genome consortium; http://dcc.icgc.org/) and found that the percentage of loss-of-function mutations in RHEB was 1.7% (2 of 117) in AML patients and 0.92% (2 of 218) in CLL patients (Online Supplementary Figure S5A).

We then compared the expression of RHEB in AML patients with high-risk versus AML patients with low risk in SurvExpress (http://bioinformatica.mty.itesm.mx/Surv 252

Express). We found the expression of RHEB was significantly lower in AML patients with high risk than that of the low risk in two GSE set analyses (GSE5122 and GSE12417-GPL97; P<0.0001) (Figure 5B). We then collected prognostic values for 163 AML patients with normal karyotype in PrognoScan database (http://www.abren.net/ PrognoScan-cgi/, GSE12417), and divided these patients into two groups for analysis: patients with RHEB expression above median and patients with RHEB expression below median (Online Supplementary Figure S5B). The survival curve showed that lower RHEB expression was associated with poor survival in AML patients with normal karyotype (log-rank test; P=0.034) (Figure 5C). These results indicated that loss of RHEB was correlated with AML progression.

3BDO partially rescued the defect of Rheb1Δ/Δ neutrophils Intrigued by the underlying phenotypic plasticities observed in Rheb1Δ/Δ HSCs, we further explored the potential transcriptional changes associated with the phenotypes by microarray analysis. Differentially expressed genes between Rheb1Δ/Δ and Rheb1fl/fl HSCs were significantly enriched in pathways involved in cell adhesion and cell development (Figure 6A, and Online Supplementary Tables S1 and S2), suggesting that complete loss of Rheb1 in hematopoietic stem cells affected the expression of genes involved in HSCs engraftment and differentiation.

Gene set enrichment analysis (GSEA) of pairwise comparisons revealed that Rheb1Δ/Δ HSCs had a significant increase in the expression of genes associated with AML incidence haematologica | 2019; 104(2)

Rheb1 regulates HSC proliferation & differentiation

and development (Figure 6B). The analysis also suggests that Rheb1Δ/Δ HSCs adopt a transcriptional program similar to AML cells due to loss of Rheb1.

Furthermore, mTORC1 signaling pathway-related genes were also down-regulated due to Rheb1 deletion (Figure 6B). S6 ribosomal protein (Ser240/244) and 4E-BP1 (Thr37/46) are typical downstream targets of mTORC1.

We found that p-S6 was reduced significantly in Rheb1Δ/Δ LKS+ cells while p-4E-BP1 did not change by phosphorylation-flow analysis (Figure 6C). To determine whether reduced p-S6 causes HSCs differentiation to immature neutrophils, we cultured Rheb1fl/fl and Rheb1Δ/Δ LKS+ cells with mTORC1 activator 3BDO in CFC assays to analyze

the myeloid development. Rheb1fl/fl and Rheb1Δ/Δ BM cells were treated with 3BDO (60 nM) or DMSO for 30 min, then analyzed for phosphorylation level of S6 and sorted for LKS+ cells seeding in M3231 medium. As expected, the p-S6 level in Rheb1-deficient BM cells was lower than that of wild-type (WT) BM cells. However, it was markedly increased with treatment of 3BDO in Rheb1Δ/Δ BM cells (Figure 6D). p-S6 level was also restored in Rheb1Δ/Δ Lin– cells with the treatment of 3BDO measured by phosphoflow analysis (Figure 6E). Interestingly, 3BDO addition in Rheb1Δ/Δ LKS+ cell culture resulted in a decrease in the percentage of immature neutrophils when compared with non-treatment controls (Figure 6F and G). These data sug-

B

A

C

D

F

G

E

Figure 6. 3BDO restores the maturation of Rheb1Δ/Δ neutrophils. (A) Gene ontology (GO) enrichment analysis of genes down-regulated and up-regulated in the absence of Rheb1. (B) Enrichment plots of selected gene sets using Gene Set Enrichment analysis. (C) Relative fluorescence intensity of p-S6 and p-4E-BP1 in Rheb1fl/fl and Rheb1Δ/Δ LKS+; n=3. (D) Protein level of p-S6 in Rheb1fl/fl and Rheb1Δ/Δ BM cells after 3BDO (60 nM) or DMSO treatment for 30 minutes (min). (E) Relative fluorescence intensity of p-S6 in Rheb1fl/fl and Rheb1Δ/Δ Lin– cells after 3BDO (60 nM) or DMSO treatment for 30 min; n=3. (F and G) FACS analysis of neutrophils and percentage of immature neutrophils in CFC assay after 3BDO (60 nM) or DMSO treatment for six days; n=3. Data are presented as mean±Standard Error of Mean. *P<0.05; **P<0.01.

haematologica | 2019; 104(2)

253

X. Wang et al.

gest that Rheb1 may regulate neutrophil differentiation partially through mTORC1 signaling pathway.

To investigate the function effects of 3BDO on Rheb1Δ/Δ HSCs, we performed a long-term culture assay in vitro and found that the CAFC (cobblestone area) formed by Rheb1Δ/Δ HSCs was similiar to that of the control in both 3BDO and DMSO groups (Online Supplementary Figure S6A and B).We further treated Rheb1Δ/Δ and Rheb1fl/fl BM cells with 3BDO or DMSO. Then, we transplanted Rheb1Δ/Δ and Rheb1fl/fl BM cells (CD45.1) together with competitive cells (CD45.2) to lethally irradiated mice. The chimerism was analyzed every four weeks after transplantation. We found that the repopulating capacity of Rheb1Δ/Δ BM cells in the PB was significantly lower when compared with Rheb1fl/fl cells, both in 3BDO and DMSO groups (Online Supplementary Figure S6C and D). These data indicated that Rheb1 may regulate HSCs engraftment independently of the mTORC1 signaling pathway.

Discussion Although Rheb1 has been known to regulate TSC1/2 upstream of mTORC1, its role in the regulation of hematopoiesis is still not fully understood. The somatic loss-function mutation in the RHEB gene in AML patients also provides genetic evidence that mutational inactivation of RHEB might be a pathogenic event in myeloid malignancies (Figure 5). Notably, we showed here that AML patients with low RHEB expression had shorter survival time than AML patients with high RHEB expression, indicating that RHEB could be a prognosticator for leukemia patient survival. In a Rheb1 conditional deletion mouse model, we determined that loss of Rheb1 caused defective HSCs and an increased number of immature neutrophils in BM, accompanied by excessive extramedullary hematopoiesis in the spleen. In addition, Rheb1 loss leads to progressive myeloproliferation in aged Rheb1Δ/Δ mice. These data suggest that Rheb1 participates in proliferation and differentiation in myeloproliferative disease.21 Our gene transcriptional expression data also reinforce the idea that Rheb1 loss leads to an increased expression of myeloid leukemia-related gene expression programs (Figure 6A and B). Further studies are needed to functionally dissect the downstream targets of Rheb1 that confer enhanced proliferation ability for stem/progenitor cells or LSCs. In the steady-state condition, the majority of HSCs are in a quiescent state.22 However, when mice are under hematopoietic stresses, such as transplantation, HSCs actively proliferate to generate progenitors that enable rapid hematologic regeneration.23 Successful reconstitution upon transplantation of HSCs depends on multiple parameters, including correct homing to the BM, residing and proliferating successfully in the BM niche, the right cellcycle status of the transplanted cells, and the adequate rate of apoptosis. Here, although Rheb1Δ/Δ mice could survive with increased HSCs and HPCs in BM and spleen to compensate for the loss of Rheb1 under steady-state conditions, Rheb1Δ/Δ HSCs were unable to reconstitute adult BM in transplantation. However, CAFC assay showed the proliferation capacity of Rheb1Δ/Δ HSCs was similar to that of Rheb1fl/fl HSCs in vitro (Online Supplementary Figure S6A and B). Our results thus suggested that the overproliferation of HSCs in Rheb1Δ/Δ mice may be caused by multiple factors including extrinsic factors such as BM microenvironment.

254

Indeed, our gene transcriptional expression data showed that Rheb1 loss leads to a decrease in expression of adhesion-related gene expression programs (Figure 6A and B).

Rheb1 may regulate HSC regeneration through noncanonical signaling pathways rather than being fully dependent on mTORC1 signaling pathway. This is consistant with the findings of Peng et al. that the proliferation ability of HSCs was impaired in Raptor-deficient HSCs (Raptor is a component of mTORC1 downstream of Rheb1), but not in Rheb1-deficient HSCs upon transplantation.24 Interstingly, in Peng et al.’s model of TAM injection at eight weeks post transplantation of Rheb1fl/fl ;RosaCreERT2 BM cells into wild-type (WT) recipient mice, no significant difference was observed in the regeneration of donor cells when compared to WT competitor cells. This is also consistant with the idea that Rheb1 deficiency does not affect HSC proliferation.24 It will be important to determine the contribution of Rheb in canonical and non-canonical signaling pathways in future studies.

Neutrophils are specially developed cells to provide a defense against bacterial infection and are essential for host survival. Rheb1Δ/Δ mice showed severe neutrophilia in PB with an increased percentage of immature neutrophils in the BM. The augmented GMP/CMPs may be the reason for the neutrophilia, as the absolute number of GMP/CMPs significantly increased in the BM and spleen of Rheb1Δ/Δ mice (Figures 1B and 2F). Although the number of neutrophils was higher in the PB and BM in Rheb1Δ/Δ mice when compared to the control (Online Supplementary Figure S1A), Rheb1Δ/Δ neutrophils could not kill bacteria effectively in vitro (Figure 1J). It has been reported that Raptor deficiency resulted in decreased numbers of Gr-1+Mac-1+ (Gr1+CD11b+) granulocytes.25 However, our results showed that Rheb1 deletion led to an increase in the number of Ly6G+CD11b+ granulocytes. This discrepancy may be due to Ly-6G marking only a subset of Gr-1 cells, while Ly-6G is used for separating granulocytes from Mac-1+ myeloid cells.

Furthermore, we found mTORC1 signaling was inhibited in Rheb1Δ/Δ progenitor cells, as evidenced by a reduced level of p-S6 (Figure 6C and D). A specific mTORC1 activator, 3BDO could only partially rescue Rheb1 deficiencyinduced immature neutrophils, indicating Rheb1 regulates neutrophil development at least partially via the mTORC1 pathway. It is possible that Rheb1 regulates neutrophil differentiation via other signaling pathways. This possibility still needs to be explored in future studies.

We have shown that mutations or reduced expression of RHEB are associated with leukemia. Interestingly, although Rheb1 deficiency leads to an MPN-like disorder in aged Rheb1Δ/Δ mice, loss of Rheb1 shortens the life of Rheb1Δ/Δ mice but does not lead to spontaneous leukemia in Rheb1Δ/Δ mice in our observations (Figures 4 and 5). It is possible that the reduced survival of aged Rheb1Δ/Δ mice may be due to ineffective myelopoiesis-related inflammation. Although Rheb1-deficiency was not a key factor for leukemogenesis, low expression of Rheb1 does associate with the initiation of myeloid proliferation-related diseases, such as MPN. We predict that additional mutations and/or pro-leukemia development environment factors are needed to initiate leukemogenesis in Rheb1Δ/Δ hematopoietic cells.

Acknowledgments The authors thank Dr. Bo Xiao for providing the Rheb1fl/fl mice, Dr. Junying Miao for providing the 3BDO reagent, and Ms. Xiaohuan Mu and Yuemin Gong for technical assistance. haematologica | 2019; 104(2)

Rheb1 regulates HSC proliferation & differentiation

Funding This work was supported by the funds from the Ministry of Science and Technology of China (2018YFA0107801, 2017YFA0103402, 2017YFA0103302, 2016YFA0100600), National Nature Science Foundation (81870088, 81470280,

References 1. Rathinam C, Flavell RA. The hematopoiesis paradigm: clarity or ambiguity? Blood. 2008;112(9):3534-3535.

2. Feng CG, Weksberg DC, Taylor GA, Sher A, Goodell MA. The p47 GTPase Lrg-47 (Irgm1) Links Host Defense and Hematopoietic Stem Cell Proliferation. Cell Stem Cell. 2008;2(1):83-89.

3. Hoshii T, Matsuda S, Hirao A. Pleiotropic roles of mTOR complexes in haemato-lymphopoiesis and leukemogenesis. J Biochem.

2014;156(2):73-83.

4. Guertin DA, Stevens DM, Thoreen CC, et al. Ablation in Mice of the mTORC Components raptor, rictor, or mLST8 Reveals that mTORC2 is required for signaling to Akt-FOXO and PKCÎą, but not S6K1. Dev Cell. 2006;11(6):859-871.

5. Patel PH, Thapar N, Guo L, et al.

Drosophila Rheb GTPase is required for cell cycle progression and cell growth. J Cell Sci. 2003;116(17):3601-3610.

6. Saucedo LJ, Gao X, Chiarelli DA, Li L, Pan D, Edgar BA. Rheb promotes cell growth as a component of the insulin/TOR signalling network. Nat Cell Biol. 2003;5(6):566-571.

7. Neuman NA, Henske EP. Non-canonical functions of the tuberous sclerosis complex-Rheb signalling axis. EMBO Mol Med. 2011;3(4):189-200.

8. Inoki K, Li Y, Xu T, Guan KL. Rheb GTPase is a direct target of TSC2 GAP activity and regulates mTOR signaling. Genes Dev. 2003;17(15):1829-1834.

9. Zhang Y, Gao X, Saucedo LJ, et al. Rheb is

haematologica | 2019; 104(2)

10.

11.

12.

13.

14.

15.

16.

17.

81300436, 81770105, 81420108017, 81525010), Chinese Academy of Medical Sciences (CAMS) Initiative for Innovative Medicine CAMS-I2M (2017-I2M-3-015, 2016-I2M-1-017), and PUMC Youth Grant (3332016141, 3332016092, 3332018199).

a direct target of the tuberous sclerosis tumour suppressor proteins. Nat Cell Biol.

2003;5(6):578-581.

Ma XM, Blenis J. Molecular mechanisms of mTOR-mediated translational control. Nat Rev Mol Cell Biol. 2009;10(5):307-318.

Kalaitzidis D, Sykes SM, Wang Z, et al. mTOR complex 1 plays critical roles in hematopoiesis and Pten-loss-evoked leukemogenesis. Cell Stem Cell. 2012;11(3):429439.

Gan B, Sahin E, Jiang S, et al. mTORC1dependent and -independent regulation of stem cell renewal, differentiation, and mobilization. Proc Natl Acad Sci USA. 2008;105(49):19384-19389.

Karbowniczek M, Zitserman D, Khabibullin D, et al. The evolutionarily conserved TSC/Rheb pathway activates Notch in tuberous sclerosis complex and Drosophila external sensory organ development. J Clin Invest. 2010;120(1):93-102.

Karbowniczek M, Robertson GP, Henske EP. Rheb inhibits C-raf activity and B-raf/Craf heterodimerization. J Biol Chem. 2006; 281(35):25447-25456.

Ma D, Bai X, Zou H, Lai Y, Jiang Y. Rheb GTPase controls apoptosis by regulating interaction of FKBP38 with Bcl-2 and BclXL. J Biol Chem. 2010;285(12):8621-8627.

Wang X, Li M, Gao Y, et al. Rheb1mTORC1 maintains macrophage differentiation and phagocytosis in mice. Exp Cell Res. 2016;344(2):219-228.

Zou J, Zhou L, Du XX, et al. Rheb1 is required for mTORC1 and myelination in postnatal brain development. Dev Cell.

2011;20(1):97-108.

18. Lu W, Han L, Su L, et al. A 3'UTR-associated RNA, FLJ11812 maintains stemness of human embryonic stem cells by targeting miR-4459. Stem Cells Dev. 2015; 24(9):1133-1140.

19. Zhu H, Guo ZK, Jiang XX, et al. A protocol for isolation and culture of mesenchymal stem cells from mouse compact bone. Nat Protoc. 2010;5(3):550-560.

20. Challen GA, Sun D, Mayle A, et al. Dnmt3a and Dnmt3b Have Overlapping and Distinct Functions in Hematopoietic Stem Cells. Cell Stem Cell. 2014;15(3):350-364.

21. Orazi A, Germing U. The myelodysplastic/myeloproliferative neoplasms: myeloproliferative diseases with dysplastic features. Leukemia. 2008;22(7):1308-1319.

22. Rumman M, Dhawan J, Kassem M. Concise Review: Quiescence in Adult Stem Cells: Biological Significance and Relevance to Tissue Regeneration. Stem Cells. 2015; 33(10):2903-2912.

23. Wilson A, Laurenti E, Oser G, et al.

Hematopoietic stem cells reversibly switch from dormancy to self-renewal during homeostasis and repair. Cell. 2008;135(6): 1118-1129.

24. Peng H, Kasada A, Ueno M, et al. Distinct roles of Rheb and Raptor in activating mTOR complex 1 for the self-renewal of hematopoietic stem cells. Biochem Biophys Res Commun. 2018;495(1):1129-1135.

25. Fiedler K, Sindrilaru A, Terszowski G, et al. Neutrophil development and function critically depend on Bruton tyrosine kinase in a mouse model of X-linked agammaglobulinemia. Blood. 2011;117(4): 1329-1339.

255

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(2):256-262

Bone Marrow Failure

Aplastic anemia in the elderly: a nationwide survey on behalf of the French Reference Center for Aplastic Anemia

Adrien Contejean,1,2,3 Matthieu Resche-Rigon,4 Jérôme Tamburini,2,3 Marion Alcantara,3,5,6 Fabrice Jardin,6 Etienne Lengliné,1,7,8 Lionel Adès,8,9 Didier Bouscary,2,3 Ambroise Marçais,3,10 Delphine Lebon,11 Cécile Chabrot,12 Louis Terriou,13 Fiorenza Barraco,14 Anne Banos,15 Lucile Bussot,16 Jean-Yves Cahn,16 Pierre Hirsch,17 Natacha Maillard,18 Laurence Simon,19 Luc-Matthieu Fornecker,20 Gerard Socié,1,8,21,22 Regis Peffault de Latour1,8,21 and Flore Sicre de Fontbrune1,21

French Reference Center for Aplastic Anemia, CHU Saint Louis, Paris; 2Hematology department, CHU Cochin, Paris; 3Paris Descartes University, Sorbonne Paris Cité; 4 Medical informatics and biostatistics department, CHU Saint-Louis, Paris; 5Department of biological hematology, CHU Necker, Paris; 6Hematology department, Centre Henri Becquerel, Rouen; 7Hematology department, CHU Saint-Louis, Paris; 8Paris Diderot University; 9Senior hematology department, CHU Saint-Louis, Paris; 10Hematology department, CHU Necker, Paris; 11Hematology department, CHU Amiens; 12Hematology department, CHU Estaing, Clermont-Ferrand; 13Clinical immunology department, CHU Lille; 14Hematology department, CHU Lyon-Sud; 15Hematology department, CH Côte Basque, Bayonne; 16Hematology department, CHU Grenoble; 17Biological hematology department, CHU Saint-Antoine; 18Hematology department, CHU La Miletrie, Poitiers; 19 Hematology department, CHU Pitié-Salpêtrière, Paris; 20Hematology department, CHU Strasbourg; 21Bone-marrow transplantation department, CHU Saint-Louis, Paris; 22 Inserm UMR 1160, CHU Saint Louis, Paris, France 1

ABSTRACT

Correspondence: regis.peffaultdelatour@aphp.fr

Received: May 24, 2018.

Accepted: September 24, 2018.

Pre-published: September 27, 2018. doi:10.3324/haematol.2018.198440 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/256 ©2019 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.

256

A

plastic anemia is a rare but potentially life-threatening disease that may affect older patients. Data regarding the treatment of aplastic anemia in this ageing population remains scarce. We conducted a retrospective nationwide multicenter study in France to examine current treatments for aplastic anemia patients over 60 years old. Our aims were to evaluate efficacy and tolerance, and to analyze predictive factors for response and survival. Over the course of a decade, 88 patients (median age 68.5 years) were identified in 19 centers, with a median follow up of 2.7 years; 21% had very severe and 36% severe aplastic anemia. We analyzed 184 treatment lines, mostly involving the standard combination of anti-thymocyte globulin and cyclosporine-A (33%), which was also the most frequent first-line treatment (50%). After first-line therapy, 32% of patients achieved a complete response, and 15% a partial response.

Responses were significantly better in first line and in patients with good performance status, as well as in those that had followed an anti-thymocyte globulin and cyclosporine-A regimen (overall response rate of 70% after first-line treatment). All treatments were well tolerated by patients, including over the age of 70. Three-year survival was 74.7% (median 7.36 years). Age, Charlson comorbidity index and very severe aplastic anemia were independently associated with mortality. Age, per se, is not a limiting factor to aplastic anemia treatment with anti-thymocyte globulin and cyclosporine-A; this regimen should be used as a first-line treatment in elderly patients if they have a good performance status and low comorbidity index score.

Introduction Aplastic anemia (AA) affects two to seven individuals per million annually with a higher incidence in Asia than in the west.1–3 Its incidence varies with age, occurring most frequently over the age of 60.4–6 Members of this patient population share specific epidemiological characteristics of the disease, suggesting a different haematologica | 2019; 104(2)

Aplastic anemia in the elderly

pathophysiology to that seen in younger patients. The majority of AA cases in the elderly are idiopathic, since hardly any genetic or viral etiologies are found in this population.7,8 However, distinguishing AA from hypocellular myelodysplasia or acute myeloid leukemia (AML) is clearly essential at this age.9 The gender ratio is also different: although the majority of AA patients under 40 are men, six in 10 patients over the age of 60 are women.7 From a biological perspective, more mutations are found in AA among the elderly that are potentially associated with adverse outcomes.10 Conversely, cytogenetic abnormalities are more frequent in younger patients.11 Moreover, age over 60 years has been identified as an independent risk factor for mortality in AA.4,10 Various therapeutic options for AA can be proposed.

Immunosuppressive regimens â&#x20AC;&#x201C; with anti-thymocyte globulin (ATG) associated with cyclosporine-A (CsA), or CsA alone â&#x20AC;&#x201C; have been used to treat AA in the elderly, but have been little studied in this population.7,12,13 ATG-CsA regimen is recommended as first line in patients over the age of 60 with severe or very severe AA.14,15 Other options may include androgens, eltrombopag, alemtuzumab, supportive care alone, monitoring with no active treatment, and palliative care; however, none of these has been studied specifically in the elderly. Likewise, data on combined treatments with CsA remains scarce. Eltrombopag has been added to the most recent recommendations for treating AA in the elderly, and may be used as a standalone second-line treatment.15 Androgens, with or without CsA, are recommended as a first-line treatment for non-severe AA.14 Allogeneic hematopoietic stem-cell transplantation (AHSCT) with a matched sibling donor is recommended as front-line treatment for patients under the age of 40.16 Recent study suggests that matched sibling donor AHSCT with reduced intensity conditioning may be safely performed after the age of 40 with encouraging outcome.17 However, in older patients, AHSCT is associated with higher risks, as well as treatment-related mortality and toxicity, even when sibling donors are used.16,18,19 Consequently, AHSCT in patients over the age of 50 is usually only undertaken in very rare and specific cases.16,18 Our study therefore aimed to describe the epidemiology and treatments administered to elderly AA patients in France, as well as to address the factors associated with treatment response and outcome in this population.

Population Patients who were diagnosed with AA by a bone marrow biopsy at the age of 60 or over were included in the study. All histologic diagnoses were made or confirmed in the French lymphopath network, on an expert basis analysis. Patients with toxic or chemotherapy-related aplasia, hypocellular myelodysplasia or AML according to previously published criteria,9 or paroxysmal nocturnal hemoglobinuria (PNH) without AA, were all excluded.

Definitions Severe AA (SAA) was defined according to the Camitta criteria20 as a bone marrow hypoplasia (<30%) shown by a biopsy with two of the three following criteria: polymorphonuclear neutrophils (PMN) <0.5x109/L, platelets <20x109/L, or reticulocytes <20x109/L. Very severe AA (vSAA) was defined using the same three criteria, but with PMN <0.2x109/L.21 Mild AA was defined by medullar hypoplasia and two associated cytopenias, but no criterion for SAA. Complete remission (CR) was declared if the patient had the following blood counts without any transfusion: PMN >1x109/L, hemoglobin >10g/dL, and platelets >100x109/L. Partial response (PR) was defined by the absence of any criterion for either SAA or CR. For patients with mild AA, PR was defined by a significant and stable improvement in the blood counts and no more transfusion requirements, along with no criterion for CR. All other situations corresponded to treatment failure. Responses were considered to be assessable if the treatment line had been initiated more than three months previously, or if a response had occurred before this time.

Treatment lines corresponded to the period between the date of diagnosis for the first line or date of treatment initiation for the following lines and either the date of the next treatment line or the date of patient death. Erythropoietin (EPO) and granulocytecolony stimulating factor (G-CSF) were considered to be supportive care treatments rather than treatment lines.

PNH clones were assessed using flow cytometry analysis on PMNs and monocytes. PNH clone size detection was considered to be positive if above 5%. At inclusion, the Charlson comorbidity index was assessed.22,23 Performance status was evaluated at the beginning of each treatment line. Adverse events were assessed using the CTCAEV4.0 scoring system.

Statistical analysis Statistical analysis is described in the Online Supplementary Material.

Results Methods Study Design and Case Detection We conducted a retrospective study in 19 centers in France on behalf of the French Reference Center for Aplastic Anemia. Cases that fitted the inclusion criteria were identified within a 10-year period (1/1/2007 to 12/31/2016). We started by including all patients whose symptoms and treatments were reviewed monthly in the French Reference Center for Aplastic Anemia. We then asked the pathologist of each center to send us reports concerning patients with aplasia or hypoplasia in their bone marrow biopsy analysis, among whom we identified patients with AA. Before inclusion, all cases were reviewed (AC, FSF, RPL) and data were anonymously collected in a case report form. In accordance with French law, written consent was not required for this retrospective non-interventional study. Patients provided a non-opposition statement. haematologica | 2019; 104(2)

We identified 88 patients from the 19 participating centers who fulfilled the inclusion criteria during the study period. Median follow up was 32.1 months (979 days).

Our population had a median age of 68.5 years (range 6089), with 43 women (49%). At diagnosis, the severity of aplastic anemia was mild for 38 patients (43%), severe for 32 (36%), and very severe for 18 (21%); the median Charlson comorbidity index was 2 (range 0-6), with a median performance status of 1 (range 0-3). Population characteristics and main baseline biological results are shown in Table 1. Median blood count values were as follows: PMN 0.705x109/L (range 0-7.4); lymphocytes 1.3x109/L (0.031-3.94); hemoglobin 8.25g/dL (3-13.6); mean corpuscular volume (MCV) 94fL (80-115); reticulocytes 23.5x109/L (0-90); and platelets 10.5x109/L (1-82).

Bone marrow aspirates were taken from all patients at 257

A. Contejean et al.

Table 1. Population characteristics at diagnosis (n=88).

n(%) or median [IQR] (min-max) Male Age (years) Weight (kg) Charlson comorbidity index Performance status Blood count PMN (x109/L) Lymphocytes (x109/L) Hemoglobin (g/dL) MCV (fL) Reticulocytes (x109/L) Platelets (x109/L) AA severity Mild Severe Very severe Myelogram cellularity Markedly increased Increased Average Reduced Poor Unknown or failure PNH clone â&#x2030;Ľ 5% Bone marrow biopsy performed Cytogenetic abnormalities& (n=78)

45 (51%) 68.5 [63.75;74] (60-89) 69 [61;80] (45-110) 2 [1;2.25] (0-6) 1 [0;1] (0-3) 0.705 [0.335;1.2] (0-7.4) 1.3 [0.979;1.792] (0.031-3.94) 8.25 [6.98;9.2] (3-13.6) 95 [89;101] (80-115) 23.5 [13;42] (0-90) 10.5 [5;19.8] (1-82) 38 (43%) 32 (36%) 18 (21%) 1 (1%) 2 (2%) 18 (21%) 44 (50%) 17 (19%) 6 (7%) 7 (9%) 88 (100%) 6 (8%)

& Cytogenetic analysis was performed in 80 patients, showing normal results in 80% of cases and failure in 11%. Six patients had cytogenetic abnormalities: 2 delY, 1 del13p, 1 del4q, 1 split of IgH locus and 1 tri8. For one patient, the result was unknown. IQR: Inter-Quartile Range; MCV: Mean Corpuscular Volume; PMN: Polymorphonuclear Neutrophils; PNH: Paroxysmal Nocturnal Hemoglobinuria.

diagnosis. In 17 cases (21%), we found minimal signs of dysplasia without fulfilling the criteria for hypoplastic myelodysplastic syndrome.9 In seven cases (9%), PNH clones were above 5%. Bone marrow biopsies were taken from 88 patients (100%) at diagnosis. Cytogenetic analysis was performed for 78 patients, showing normal results in 80% of these cases, and failure in 11%. Six of these 78 patients (8%) presented cytogenetic abnormalities (Table 1).

Overall, the 88 study patients received 184 treatment lines, with a median of two lines per patient (range 1-5).

Treatment lines comprised ATG-CsA (33%, including 72% with horse ATG), CsA alone (14%), androgen alone (14%), eltrombopag alone (10%), CsA associated with androgen or eltrombopag (9%), two AHSCTs (1%), and other treatments (19%). For patients who received CsA, median residual dosing of CsA between one and two months after treatment initiation was 200ng/mL (range 42-560). During the first month following initiation of each treatment line, patients received a median of four red blood packs (RBPs) (range 0-26) and four platelet transfusions (range 0-18). Iron chelators were administered in 19% of treatment lines, EPO in 21%, and G-CSF in 23%.

First-line treatment was ATG-CsA for 50% of patients (73% received horse ATG, and 11% a combination of 258

Figure 1. Overall survival. Causes of death were as follows: nine infections (38%), four hemorrhagic complications (17%), five deaths in palliative care or after active treatment had finished (21%), two cases involving unknown etiologies (8%), one case of clonal evolution to acute myeloid leukemia, one case of multi-metastatic breast cancer, one case of hypercalcemia, and one cardiac arrest. The survival curve (solid line) was obtained using the Kaplan Meier estimator. Dashed lines represent confidence intervals (CI95%).

ATG, CsA and eltrombopag), CsA alone for 20%, androgens for 9%, eltrombopag for 3%, and other treatments in 18% of cases. Comparisons of patients treated with ATGCsA, CsA or other treatments (Table 2) revealed that patients receiving ATG-CsA were significantly younger (66 years, vs. 71.5 vs. 71.5, respectively, P=0.007), more frequently female (61%, vs. 50% vs. 27%, respectively, P=0.02) and had a lower platelet count (8x109/L, vs.

12x109/L vs. 15x109/L, respectively, P=0.025). Interestingly, we found no difference with respect to weight, Charlson comorbidity index score, performance status, disease severity, PNH clones, or myelogram cellularity.

After the 181 assessable treatment lines, 19% achieved CR and 19% a PR. Median time until best response was 151 days (4.9 months). After first-line treatment, 32% of patients achieved CR and 15% a PR. The overall response rate (ORR) was 62% with ATG-CsA (70% as a first-line treatment), 35% with CsA alone (39% as first line), 22% with eltrombopag, and 21% with androgen. The ORR after first line was 69% in patients who received horse ATG and 75% in patients who received rabbit ATG (P=0.40). Strikingly, the ORR in patients over 70 years receiving ATG-CsA (n=16) was 81% (50% achieving a CR, 31% a PR). Treatment line and poor performance status at treatment initiation were associated with poorer treatment responses, whereas sex, age, weight, high Charlson comorbidity index score or disease severity were not (Online Supplementary Table). After adjustment for treatment line, disease severity and performance status, we found that, when compared with all other treatments, ATG-CsA regimen was significantly associated with a better response rate (OR 4.18 (1.93;9.09), P=0.0003) as shown in Table 3. Moreover, in a multivariable analysis using ATG-CsA as a baseline, we found that CsA alone (OR haematologica | 2019; 104(2)

Aplastic anemia in the elderly Table 2. Population characteristics according to first-line treatment (n=88).

n(%) or median [IQR] Male Age (years) Weight (kg) Charlson comorbidity index Performance status Blood count PMN (x109/L) Lymphocytes (x109/L) Hemoglobin (g/dL) MCV (fL) Reticulocytes (x109/L) Platelets (x109/L) AA severity Mild Severe Very severe Myelogram cellularity Markedly increased Increased Average Reduced Poor Unknown or failure PNH clone â&#x2030;Ľ 5%

ATG-CsA (n=44)

CsA alone (n=18)

Others (n=26)

P

17 (39%) 66 [62;70] 66.5 [61;78] 1.5 [1;2.25] 1 [0;1]

9 (50%) 71.5 [65;74] 73 [61;79.5] 2 [1;2] 1 [1;1]

19 (73%) 71.5 [68.5;76] 69.5 [65.75;80] 1.5 [1;2.75] 1 [0;1.75]

0.02 0.007 0.62 0.96 0.26

0.51 [0.195;1.1] 1.5 [0.999;1.95] 7.9 [6.98;8.65] 92 [88;98] 20 [11.92;41.25] 8 [4;13]

0.915 [0.25;1.43] 1.29 [1;1.585] 8.65 [7.23;9.2] 92.5 [89;101.2] 23 [16;40] 12 [6;16.75]

0.8 [0.5;1.513] 1.2 [0.8;1.5] 8.85 [6.98;10.1] 100.3 [95.5;104] 28 [19.5;43.5] 15 [7.25;37.75]

0.12 0.46 0.11 0.095 0.48 0.025

16 (36.5%) 16 (36.5%) 12 (27%)

7 (39%) 6 (33%) 5 (28%)

15 (58%) 10 (38%) 1 (4%)

0.11

1 (2%) 1 (2%) 11 (25%) 18 (41%) 10 (23%) 3 (7%) 3 (7%)

0 (0%) 0 (0%) 4 (22%) 11 (61%) 3 (17%) 0 (0%) 3 (17%)

0 (0%) 1 (4%) 3 (12%) 15 (58%) 4 (15%) 3 (11%) 1 (4%)

0.88

0.33

AA: Aplastic Anemia; ATG: Anti-Thymocyte Globulin; CsA: Cyclosporine-A; IQR: Inter-Quartile range; MCV: Mean Corpuscular Volume; PMN: Polymorphonuclear Neutrophils; PNH: Paroxysmal Nocturnal Hemoglobinuria.

Table 3. Impact of treatment on outcome (overall response and mortality) after adjustment for treatment line, disease severity and performance status.

Overall response OR (CI95%) P ATG-CsA CsA alone Eltrombopag alone Androgens alone EPO G-CSF

4.18 (1.93;9.09) 0.73 (0.28;1.91) 0.34 (0.08;1.36) 0.44 (0.14;1.33) 2.05 (0.91;4.6) 1.32 (0.59;2.93)

0.0003 0.52 0.13 0.15 0.082 0.5

Mortality HR (CI95%)

P

0.44 (0.13;1.45) 2.84 (0.93;8.61) 0.80 (0.08;7.71) 1.33 (0.36;4.88) 1.12 (0.35;3.58) 1.28 (0.50;3.29)

0.18 0.066 0.85 0.67 0.84 0.61

In analyses, EPO and G-CSF were considered to be supportive care treatments rather than treatment lines. ATG: Anti-thymocyte Globulin; CI: Confidence Interval; CsA: Cyclosporine-A; EPO: Erythropoietin; G-CSF: Granulocyte-Colony Stimulating Factor; HR: Hazard Ratio; OR: Odds Ratio.

0.35 (0.13;0.96), P=0.042), eltrombopag (OR 0.12 (0.03;0.54), P=0.0057) and androgens (OR 0.17 (0.05;0.58), P=0.0047) were all individually associated with lower response rates than an ATG-CsA regimen (Table 4).

The main treatment-associated complications were infections, with 35% of treatment lines complicated by at least one grade-III/IV infection, including 9 deaths (5%), and renal issues, with grade-III/IV acute kidney failure after 29% of treatment lines. Other grade-III/IV complications were hepatic or digestive (11%), cardiovascular (9%), hemorrhagic (9%, including 4 deaths (2%)) and neurological (2%). In comparison with other treatments, ATG-CsA was associated with significantly more infectious (72% vs. 24%, P<0.0001) and cardiovascular complications (32% vs. 15%, P=0.01) and acute kidney failure haematologica | 2019; 104(2)

(43% vs. 22%, P=0.003). Interestingly, patients aged 70 and over receiving ATG-CsA did not seem to experience any more complications than younger patients, and only one patient in this subgroup died during ATG-CsA treatment (Table 5). Four clonal evolutions were recorded: two abnormal karyotypes (1 monosomy of chromosome 7 and 1 t(3;4)), one case of acute myeloid leukemia, and one case of myelodysplastic syndrome with 17% excess blasts.

Overall survival at three years was 74.7%, with median survival of 7.36 years (Figure 1) and a total of 24 patient deaths from the following causes: nine infections (38%), five deaths in palliative care or after active treatment had finished (21%), four hemorrhagic complications (17%), and six miscellaneous causes. Univariate analysis showed that mortality was significantly higher in patients with 259

A. Contejean et al.

vSAA (HR 3.02 (1.12;8.12), P=0.029), or with a high baseline Charlson comorbidity index score (HR 1.38 (1.11;1.72), P=0.0043) or performance status (HR 1.77 (1.11;2.83), P=0.017). We also found a trend pointing to an association between age and mortality (HR per year 1.05 (0.99;1.11), P=0.079). Due to the small number of events, our multivariable analysis only included the three most relevant factors. We consequently found that age (OR 1.07 (1.01;1.14), P=0.03), Charlson comorbidity index (HR 1.34 (1.07;1.67), P=0.01) and vSAA (HR 3.67 (1.51;8.91), P=0.004) were independently associated with mortality (Table 6). Regarding the impact of treatment on mortality, Table 3 shows that after adjustment for treatment line, disease severity and performance status, the benefits observed in terms of response with ATG-CsA were not associated with reduced mortality (HR 0.44 (0.13;1.45), P=0.18). Likewise, we found no statistical difference in overall survival between patients who received first-line horse ATG in comparison with rabbit ATG (HR 4.7 (0.4;52.6), P=0.2) or in mortality between patients over 70 years old who received first-line ATG-CsA in comparison with other treatments (HR 0.17 (0.02;1.28), P=0.08).

Discussion This study focused on current real-life management of AA in the elderly in France. This is, to our knowledge, the largest cohort specifically addressing the question of AA management in the elderly.

Our population comprised patients with a median age of 68.5 years (maximum 89 years) with an even sex ratio, although previous epidemiological studies suggested a feminine predominance in older patients with AA.7 Although patients mostly had SAA or vSAA (57%), nine patients did not require transfusion during the first months following treatment initiation. They had few comorbidities (median Charlson comorbidity index of 2) and were fit (median performance status 1). Of particular notice, although PMN count or hemoglobin may be normal in some patients at diagnosis, all patients had thrombocytopenia (maximum platelets count 82x109/L).

We reported 8% of cytogenetic abnormalities at diagnosis, close to that described in literature (4-5%).11,24,25 We found two Y deletions, an abnormality frequently encountered in the elderly. Others were: one trisomy 8, one split of IgH locus, one deletion 13p and one deletion 4q.

Trisomy 8 is known to be seen in AA,25,26 and may be the most frequent cytogenetic aberration in this pathology.11,27 Abnormalities of IgH locus have not been reported in AA, but chromosome 14 aberrations can be seen in AA.27

Abnormalities of chromosome 13 have been reported but affecting more often the long arm (13q).28 Finally, deletion 4q has not before been described in this pathology.

This observational study highlights the great heterogeneity in the choice of treatment in this population. In first line, half of the patients received ATG-CsA, including five with eltrombopag, although combined treatments with eltrombopag in first line is not recommended yet.14 This proportion is consistent with previously reported series.13 One fifth of the patients received CsA alone.

Patients receiving ATG-CsA in front line were significantly younger and mostly women. We could not explain this difference in sex ratio. Conversely, other parameters, which could be decisive in treatment choice, like performance status, Charlson comorbidity index or disease severity, did not differ between groups. Analysis of all treatment lines lets us see that most of those were immunosuppressive treatments (58%), mainly ATG-CsA combinations. The majority of ATG-CsA were performed in first line (73%). Patients receiving ATG-CsA after the first line (n=16) were younger (maximum 70 years). Six of them had already received this treatment in first line; while 10 patients had previously received another first line therapy.

CsA represented 14% of treatment lines, as androgens.

Eltrombopag was given as a monotherapy in 9% of cases.

As reported in literature,13 AHSCTs were marginal in this population since two patients received this treatment: the first one for a clonal evolution with 7q- who died from invasive fungal infection; and the second who is alive 36 months after AHSCT and was refractory to two lines before graft including ATG-CsA. Despite no recommendations, 21% of treatment lines went along with EPO and 23% with G-CSF. EPO and G-CSF were considered to be supportive care treatments and not treatment lines.

The global response rate was 38%, all lines combined.

Time to best response was long: 151 days (4.9 months).

We observed that in first line, 47% of patients reached response, with 32% of CR. Only 6 patients obtained CR Table 4. Impact of treatment on overall response, after adjustment for treatment line, disease severity and performance status (multivariable analysis, with ATG-CsA as a baseline).

ATG-CsA CsA alone Eltrombopag alone Androgens alone Other

P

1 0.35 (0.13;0.96) 0.12 (0.03;0.54) 0.17 (0.05;0.58) 0.24 (0.09;0.63)

1 0.042 0.0057 0.0047 0.0038

ATG: Anti-thymocyte Globulin; CI: Confidence Interval; CsA: Cyclosporine-A; OR: Odds Ratio.

Table 5. Treatment tolerance comparing ATG-CsA regimens with others.

Infectious complications Cardiovascular complications Acute kidney failure Hemorrhagic complications Hepatic or digestive complications Neurological complications

OR (CI95%)

ATG-CsA (n=60)

Others (n=124)

P

ATG-CsA â&#x2030;Ľ70 years (n=16)

72% 32% 43% 18% 23% 7%

24% 15% 22% 15% 19% 4%

<0.0001 0.01 0.003 0.62 0.56 0.48

88% 31% 50% 19% 31% 13%

no: ATG: Anti-Thymocyte Globulin; CsA: Cyclosporine-A.

260

haematologica | 2019; 104(2)

Aplastic anemia in the elderly Table 6. Impact of baseline characteristics on mortality.

Univariable analysis HR (CI95%) P Male Age (per year) Charlson comorbidity index (for each one-point increase) Performance status (for each one-point increase) Weight (for each one-kg increase) SAA vSAA PNH clone â&#x2030;Ľ 5%

1.74 (0.76;3.99) 1.05 (0.99;1.11) 1.38 (1.11;1.72) 1.77 (1.11;2.83) 1.03 (1;1.07) 1.24 (0.45;3.45) 3.02 (1.12;8.12) 0.66 (0.09;5)

0.19 0.079 0.0043 0.017 0.059 0.68 0.029 0.69

Multivariable analysis HR (CI95%) P

1.07 (1.01;1.14) 1.34 (1.07;1.67) 3.67 (1.51;8.91) -

0.03 0.01 0.004 -

CI: Confidence Interval; HR: Hazard Ratio; PNH: Paroxysmal Nocturnal Hemoglobinuria; SAA: Severe Aplastic Anemia; vSAA: Very Severe Aplastic Anemia.

after first line. The global response rate was higher with ATG-CsA (62% all lines combined, 70% in first line).

These results are similar to results in younger patients (global responses of 60% to 80%).29,30 In this study, we found no difference in first line ORR between horse and rabbit ATG. Older patients (>70 years) responded well to ATG-CsA, with a global response rate of 81% of (50% CR and 31% PR). Global response after a CsA course was 35%, all lines combined, and 39% in first line, as in a younger population.31 Responses to eltrombopag (22%) or androgen (21%) were lower. Although defined with different criteria and in a broader population, Lengline et al. found a higher response rate with eltrombopag than us.32 Baseline Charlson comorbidity index and performance status were associated with a lower response rate. After adjustment for treatment line, performance status and disease severity, we found that ATG-CsA was associated with a better response. Conversely, multivariable analysis showed that patients treated with CsA alone, eltrombopag or androgens were worse responders in comparison with ATG-CsA. Treatments were well tolerated in this population, with very little treatment-related mortalities, although this should be interpreted with caution due to our retrospective design. Complications were mostly infectious, particularly after immunosuppressive treatments, and especially ATG-CsA, followed by acute kidney failure, almost exclusively seen in patients receiving CsA. Conversely, eltrombopag carries a good tolerance profile in our study. Among patients who died from hemorrhagic complications, none were treated with ATG-CsA, and only 3 in 9 patients who died from infectious complications were under an ATGCsA regimen. Thus, this treatment, although exposing to more infectious events, rarely conducted to death from infectious origin. Interestingly, patients over 70 years tolerated ATG-CsA as well as younger patients, with one death from sepsis.

Survival was noticeably better than in recently published data,13,33 with a median survival of 7.36 years and three-year survival of 74.7%. Some of the baseline characteristics were associated with mortality: age, Charlson comorbidity index and vSAA in comparison to SAA and mild AA. The positive impact of ATG-CsA on response rates did not translate into better survival after adjustment for treatment line, performance status and disease severity. Likewise, overall survival did not differ between patients over 70 years who received ATG-CsA in first line in comparison with other treatments as well as in patients receiving horse ATG in comparison with rabbit ATG in haematologica | 2019; 104(2)

first line. It is possible that our cohort was too small to reach significance in survival for patients receiving ATGCsA and in the other subgroups of our study.

Nevertheless, supportive care optimization of hematological patients with cytopenias might also explain the observed survival even in patients with refractory disease.

Moreover, death in older patients may be due to causes unrelated to the underlying disease, even in the AA population.7 Our study has a number of limitations. Firstly, its retrospective design and case identification strategy may bring some bias, particularly a lack of exhaustivity of our data or patient identifications. We also may have selected more complicated cases by using the data of the French Reference Center for Aplastic Anemia. However, the systematic identification of cases using data from pathologists from each center may have limited this bias.

Secondly, the great heterogeneity of treatments in our population did not allow us to conclude on specific regimens like CsA-androgen or CsA-eltrombopag. Specific studies assessing these combinations in this population should be conducted. Thirdly, we found a low rate of clonal evolution with four occurrences. This is less than previously reported,7 and may be due to the relatively short median follow up of our cohort. Fourthly, multivariable analyses were limited in the number of adjustment variables, to fit with the number of events. However, these variables were selected as the most pertinent and susceptible to bias the univariate analyses results. For the same reason, we were not able to adjust the response or mortality predicting factors on age, although age was associated with survival. We chose to adjust on treatment line, disease severity and performance status because these factors were, for us, the most relevant in this context. Lastly, the retrospective design and data collection of files may limit interpretation of tolerance data, since less serious adverse events may have been neglected in patient reports. Nevertheless, it is probable that most of the serious adverse events have been well reported.

In conclusion, our study shows that current therapeutic choices in elderly patients with AA are very heterogeneous in France, and are not based on objective selection criteria, such as performance status or comorbidities. The impact of the first-line treatment regimen choice in terms of response underlies the need for strong recommendations in this population. Our analysis stresses that some of the patient characteristics are associated with mortality: age, comorbidities, and very severe AA. Given that survival in the elderly is naturally shorter than in younger 261

A. Contejean et al.

patients, overall mortality of AA in elderly is very encouraging. Age, per se, is not a limiting factor to AA treatment with ATG-CsA; this regimen should be used as first-line treatment in elderly patients if they have a good performance status and low comorbidity index score. Among patients with adverse performance status, or comorbidities contra-indicating the use of ATG, CsA alone or in combination may be safely used. Other treatment strategies might be reserved for later courses of treatments. In addition, supportive care may have a great impact on survival in this population. These results require confirmation through prospective data specifically collected in this population with extended follow up.

References 1. Young NS, Kaufman DW. The epidemiology of acquired aplastic anemia.

Haematologica. 2008;93(4):489-492.

2. Modan B, Segal S, Shani M, Sheba C. Aplastic anemia in Isreal: evaluation of the etiological role of chloramphenicol on a community-wide basis. Am J Med Sci. 1975;270(3):441-445.

3. Kaufman DW, Kelly JP, Jurgelon JM, et al.

Drugs in the aetiology of agranulocytosis and aplastic anaemia. Eur J Haematol Suppl. 1996;6023-30.

4. Montane E, Ibanez L, Vidal X, et al.

Epidemiology of aplastic anemia: a prospective multicenter study.

Haematologica. 2008;93(4):518-523.

5. Incidence of aplastic anemia: the relevance of diagnostic criteria. By the International Agranulocytosis and Aplastic Anemia Study. Blood. 1987;70(6):1718-1721.

6. Mary JY, Baumelou E, Guiguet M. Epidemiology of aplastic anemia in France: a prospective multicentric study. The French Cooperative Group for Epidemiological Study of Aplastic Anemia.

Blood. 1990;75(8):1646-1653.

7. Tichelli A, Socié G, Henry-Amar M, et al.

Effectiveness of immunosuppressive therapy in older patients with aplastic anemia.

Ann Intern Med. 1999;130(3):193-201.

8. Brodsky RA, Jones RJ. Aplastic anaemia.

Lancet Lond Engl. 2005;365(9471):16471656.

9. Bennett JM, Orazi A. Diagnostic criteria to distinguish hypocellular acute myeloid leukemia from hypocellular myelodysplastic syndromes and aplastic anemia: recommendations for a standardized approach.

Haematologica. 2009;94(2):264-268.

10. Yoshizato T, Dumitriu B, Hosokawa K, et al. Yoshizato T, Dumitriu B, Hosokawa K, et al. Somatic mutations and clonal hematopoiesis in aplastic anemia. N Engl J Med. 2015;373(1):35-47.

11. Kim S-Y, Lee J-W, Lee S-E, et al. The characteristics and clinical outcome of adult patients with aplastic anemia and abnormal cytogenetics at diagnosis. Genes Chromosomes Cancer. 2010; 49(9):844850.

12. Kao SY, Xu W, Brandwein JM, et al.

Outcomes of older patients (>60 years) with acquired aplastic anaemia treated

262

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

Acknowledgments Authors would like to thank HPN-France association for helping fund this study, David Williams for editorial assistance and all who participated to data collection: Audrey Aleme, Marie Balsat, Claire Blanchard-Delaunay, Julie Bruneau, Axelle Canard, Cécile Chabrot, Sylvain Chantepie, Régis Costello, Luc Darnige, Eric Deconinck, Stéphane Girault, Gaëlle Guillerm, Mathilde Hunault, Dominique Jacomy, Ingrid Lafon, Marielle Le Goff, Ollivier Legrand, Richard Lemal, Bruno Lioure, Marius Moldovan, Valérie Rebeix, Isabelle Roche-Lachaise, Célia Salanoubat, Andreea Sandu, Jose Miguel Torrerosa-Diaz and Aliénor Xhaard.

with immunosuppressive therapy. Br J Haematol. 2008;143(5):738-743.

Vaht K, Göransson M, Carlson K, et al.

Incidence and outcome of acquired aplastic anemia: real-world data from patients diagnosed in Sweden from 2000-2011.

Haematologica. 2017;102(10):1683-1690.

Tichelli A, Marsh JCW. Treatment of aplastic anaemia in elderly patients aged> 60 years.

Bone Marrow Transplant.

2013;48(2):180-182.

Killick SB, Bown N, Cavenagh J, et al.

Guidelines for the diagnosis and management of adult aplastic anaemia. Br J Haematol. 2016;172(2):187-207.

Bacigalupo A. How I treat acquired aplastic anemia. Blood. 2017;129(11):1428-1436.

Shin SH, Jeon YW, Yoon JH, et al.

Comparable outcomes between younger ( 40 years) and older (>40 years) adult patients with severe aplastic anemia after HLA-matched sibling stem cell transplantation using fludarabine-based conditioning.

Bone Marrow Transplant.

2016; 51(11):1456-1463.

Horowitz MM. Current status of allogeneic bone marrow transplantation in acquired aplastic anemia. Semin Hematol. 2000; 37(1):30-42.

Devillier R, Dalle J-H, Kulasekararaj A, et al. Unrelated alternative donor transplantation for severe acquired aplastic anemia: a study from the French Society of Bone Marrow Transplantation and Cell Therapies and the EBMT Severe Aplastic Anemia Working Party. Haematologica.

2016;101(7):884-890.

Camitta BM, Thomas ED, Nathan DG, et al. Severe aplastic anemia: a prospective study of the effect of early marrow transplantation on acute mortality. Blood.

1976;48(1):63-70.

Bacigalupo A. Guidelines for the treatment of severe aplastic anemia. Working Party on Severe Aplastic Anemia (WPSAA) of the European Group of Bone Marrow Transplantation (EBMT). Haematologica.

1994;79(5):438-444.

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.

Quan H, Li B, Couris CM, et al. Updating and validating the Charlson comorbidity

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol. 2011; 173(6):676-682.

Appelbaum FR, Barrall J, Storb R, et al.

Clonal cytogenetic abnormalities in patients with otherwise typical aplastic anemia. Exp Hematol 1987; 15(11):11341139.

Hosokawa K, Sugimori N, Katagiri T, et al.

Increased glycosylphosphatidylinositolanchored protein-deficient granulocytes define a benign subset of bone marrow failures in patients with trisomy 8. Eur J Haematol. 2015;95(3):230-238.

Haase D. Cytogenetic features in myelodysplastic syndromes.

Ann Hematol. 2008;87(7):515-526.

Mikhailova N, Sessarego M, Fugazza G, et al. Cytogenetic abnormalities in patients with severe aplastic anemia.

Haematologica. 1996;81(5):418-422.

Maciejewski JP, Risitano A, Sloand EM, Nunez O, Young NS. Distinct clinical outcomes for cytogenetic abnormalities evolving from aplastic anemia. Blood. 2002; 99(9):3129-3135.

Scheinberg P, Young NS. How I treat acquired aplastic anemia. Blood. 2012; 120(6):1185-1196.

Dufour C, Svahn J, Bacigalupo A. Frontline immunosuppressive treatment of acquired aplastic anemia. Bone Marrow Transplant. 2013;48(2):174-177.

Gluckman E, Esperou-Bourdeau H, Baruchel A, et al. Multicenter randomized study comparing cyclosporine-A alone and antithymocyte globulin with prednisone for treatment of severe aplastic anemia.

Blood. 1992;79(10):2540-2546.

Lengline E, Drenou B, Peterlin P, et al.

Nationwide survey on the use of eltrombopag in patients with severe aplastic anemia: a report on behalf of the French Reference Center for Aplastic Anemia.

Haematologica. 2018;103(2):212-220.

Tichelli A, Schrezenmeier H, Socie G, et al.

A randomized controlled study in patients with newly diagnosed severe aplastic anemia receiving antithymocyte globulin (ATG), cyclosporine, with or without GCSF: a study of the SAA Working Party of the European Group for Blood and Marrow Transplantation. Blood. 2011; 117(17):4434-4441.

haematologica | 2019; 104(2)

ARTICLE

Red Cell Biology & its Disorders

Exposure to non-inherited maternal antigens by breastfeeding affects antibody responsiveness

Ferrata Storti Foundation

Henk Schonewille,1,2 Jon J. van Rood,3† Esther P. Verduin,1,2,3 Leo M.G. van de Watering,1,2 Geert W. Haasnoot,3 Frans H.J. Claas,3 Dick Oepkes,4 Enrico Lopriore5 and Anneke Brand1,3

Center for Clinical Transfusion Research, Sanquin Research, Leiden; 2Jon J van Rood Center for Clinical Transfusion Research, Sanquin-Leiden University Medical Center, Amsterdam; 3Department of Immunohematology and Blood Transfusion, Leiden University Medical Center; 4Division of Fetal Medicine, Department of Obstetrics, Leiden University Medical Center; 5Division of Neonatology, Department of Pediatrics, Leiden University Medical Center, the Netherlands 1

Haematologica 2018 Volume 104(2):263-268

ABSTRACT

T

he observation, by Ray Owen and colleagues in 1954, that D-negative women were less likely to form anti-D antibodies against their D-positive fetus if their mother possessed the D-antigen, was not found in all later studies. We hypothesized that breastfeeding, received by the mother, may affect her immunity against non-inherited maternal red blood cell antigens. We studied a cohort of 125 grandmother-mother-child combinations, from a follow-up study of mothers after intrauterine transfusion of the fetus for alloimmune hemolytic disease.

For mismatched red blood cell antigens the mother was exposed to, whether or not antibodies were formed, we determined whether her mother, the grandmother, carried these antigens. The duration for which the mothers were breastfed was estimated by way of a questionnaire.

Using multivariate logistic regression analyses, the interaction term (noninherited maternal antigen exposure by categorized breastfeeding period) showed that a longer breastfeeding period was associated with decreased alloimmunization against non-inherited maternal antigens (adjusted odds ratio 0.66; 95% confidence interval 0.48-0.93). Sensitivity analysis with dichotomized (shorter versus longer) breastfeeding periods showed that this lower risk was reached after two months (aOR 0.22; 95% CI 0.07-0.71) and longer duration of breastfeeding did not seem to provide additional protection. These data suggest that oral neonatal exposure to non-inherited maternal red blood cell antigens through breastfeeding for at least two months diminishes the risk of alloimmunization against these antigens when encountered later in life.

Introduction In utero, all humans are exposed to non-inherited maternal antigens (NIMAs), however only pregnant women encounter inherited paternal antigens (IPAs).

NIMAs and IPAs can involve the same antigens (Figure 1). Maternal antibodies against IPAs expressed on red blood cells (RBC) such as Rh and K antigens, can cause severe hemolytic disease of the fetus and newborn (HDFN).

In 1954, Owen and colleagues found that D-negative mothers were less likely to form anti-D against a D-positive fetus if they had previously been exposed to the D blood group as a NIMA, a phenomenon referred to as the “grandmother effect”.1 These findings seemed in accordance with the concept of neonatal tolerance in mice, published a year earlier by Billingham and collegues.2 However, subsequent investigations did not confirm the grandmother theory.3-5 Some studies even reported that a D-negative child may develop anti-D against the D NIMA.6,7 As a result, the grandmother concept was almost forgotten.

Decades later, in 1988, Claas and colleagues observed that hyper-immunized dialysis patients awaiting renal allograft, formed antibodies against non-inherited haematologica | 2019; 104(2)

Correspondence: h.schonewille@sanquin.nl

Received: June 14, 2018.

Accepted: September 11, 2018.

Pre-published: September 27, 2018. doi:10.3324/haematol.2018.199406 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/263 ©2019 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.

263

H. Schonewille et al.

paternal HLA antigens (NIPAs) significantly more often than against NIMAs.8 However, a protective effect of HLA NIMAs exposure on later renal and stem cell transplant outcome was not confirmed in all further studies.9-11 Studies in mice models showed that a maximum immune tolerance to NIMA is obtained when in utero exposure to NIMA is followed by breastfeeding (BF).12,13 One study in humans showed a superior graft survival of maternal and sibling renal transplants when the recipient was breastfed.14 Other studies in humans also showed that the duration of BF was associated with autoimmune diseases later in life.15,16 Therefore, the controversial results on the role of exposure to NIMAs on later immunity when challenged by pregnancy, transfusion or transplantation, may - among other factors - be due to different BF habits. Breast milk contains soluble molecules such as HLA, immunoglobulins and extracellular vesicles, as well as viable cells, the latter already observed by Antoni van Leeuwenhoek in the 17th century.17–19 Despite ante- and postnatal anti-D immunoprophylaxis since 1998, Rhesus D antibodies are still the most frequent cause of severe HDFN. We previously showed that, yearly, about 15 pregnancies complicated by anti-D, four by anti-K and one by anti-c required intra-uterine transfusions (IUT).20 RhD immunoprophylaxis however hampers investigation of the effect of D NIMA exposure in utero and by BF on the anti-D response towards a D-positive child.

Severe HDFN is nowadays successfully treated with IUT. Unfortunately, such IUTs expose the mother to RBC antigens of the fetus and IUT donors, often leading to the induction of additional RBC antibodies.21 In the present study we investigated the hypothesis, that BF may affect immunity against non-inherited maternal red blood cell antigens, when encountered later in life through pregnancy or by transfusion, in a cohort of mothers whose fetuses were treated with IUT because of HDFN.

1. Identification of non-D RBC antigens (C, c, E, e, K, Fya, Fyb, Jk , Jkb, M, S and s) expressed by the child or IUT donor(s) but not by the mother i.e., mismatched antigens.

2. The presence or absence of maternal antibodies against each of these mismatched antigens.

3. For each mismatched antigen, whether the grandmother carried the antigen as a NIMA. a

Statistical analyses Univariate logistic regression was used to calculate odds ratio (OR) and 95% confidence intervals (CIs). The presence of antibodies was used as the dependent variable. BF duration was analysed categorized as 0, 1, 2, 3, 4-6 and >6 months and in a sensitivity analysis dichotomized (≤ or > 0, 1, 2, 3, 4 and 6 months). Adjusted odds ratio (aOR) was calculated in the final multivariate logistic regression model. The following variables were considered potential confounders for RBC antibodies: ABO compatibility between mother and child, maternal HLA-DRB1*15 genotype,25 number of IUTs (categorized as 1, 2, 3, 4 and >4), number of pregnancies (categorized as ≤2, 3 and >3), year of IUT (categorized in 5-yearblocks; 1988-93, 1994-98, 1999-03 and 2004-08) and RBC antigen immunogenicity (high: C, c, E, e and K and low: Fya, Fyb, Jka, Jkb, M, S and s antigens).

The associations between the duration of BF and the induction of antibodies were adjusted for potential confounders (i.e., P-values <0.2 in univariate analyses). To test for effect modification, two interaction terms (NIMA by months of BF and NIMA by antigen immunogenicity) were added to the model. The variables NIMA, months of BF, and the interaction term (NIMA by months of BF) were forced into the model.

All statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS Inc, Chicago, IL, USA). A 95% CI not overlapping the null value 1.00 for OR was regarded statistically significant.

Results Study population

Methods Study design A cohort study of 125 grandmother-mother-child combinations, participating in the LOTUS (LOng Term follow Up after intrauterine transfusionS) study. In short, all women with children who were treated with IUT for HDFN from 1987-2008 were eligible.

Details of the population and the methods adopted have been published previously22 (see Online Supplementary Appendix for details). All participating women were asked to invite their mothers to participate. Grandmothers were asked to complete a questionnaire on duration of breastfeeding (regardless of exclusivity).

The study was approved by the ethics committee of the Leiden University Medical Center (P08.080).

Data collection and intra-uterine transfusion policy All participants and IUT donors were typed for relevant RBC antigens (see Online Supplementary Appendix for details). The mothers were screened for RBC antibodies as previously described.23 Maternal transfusion history including date, number and donor of each IUT and number of pregnancies were collected.

Over the years the transfusion policy changed with increasing degree of extended RBC antigen matching between mother and IUT donor and also procedural technique (see Online Supplementary Appendix for details).20,24 The following was determined: 264

A total of 125 grandmother-mother-child pairs participated in the study. The mothers (median age at follow up, 40 years; range 24-52) gave birth to 399 children of which 143 were treated for HDFN with a total of 405 IUTs. The median birth order of the first IUT-treated child was three; there were three cases with an affected first child. The antibodies primarily responsible for HDFN were anti-D (n=93), anti-K (n=19), anti-c (n=11) and anti-Cw and -Kpa, one each (see Table 1 for additional characteristics).

A total of 549 RBC antigens - other than D - were not expressed by the mothers. For 171 of these antigens, there was no exposure by child or donor, and due to incomplete child and/or donor RBC typing, exposure was not known for 48 antigens. Consequently, analyses were restricted to the remaining 330 antigen exposures (median 3; range 1-5 per mother), of which 123 were NIMA re-exposures and 207 not previously exposed to as NIMA. These resulted in 158 antibodies, 54 (44%) against NIMA of which six antibodies caused the HDFN, and 104 (50%) against antigens not exposed to as NIMA, of which 26 caused the HDFN (Figure 2).

Breastfeeding duration and antibody responses The median period that the 125 mothers were breastfed for was two months (range 0-12); 45 mothers were not haematologica | 2019; 104(2)

Breastfeeding reduces anti-NIMA immune response

breastfed, 80 mothers for a median period of three months (Table 1).

Univariate analyses revealed that antigen immunogenicity (OR 17.3; 95% CI 10.0-29.9), number of IUTs (OR 1.22; 95 % CI 1.04-1.43) and IUT-year (OR 0.85; 95% CI 0.68-1.06) were associated (all P<0.2) with antibody formation. ABO compatibility, maternal HLA-DRB1*15 genotype and number of pregnancies were regarded not associated (all P>0.5) (Table 2).

Multivariate analysis showed that, high compared to low antigen immunogenicity, increasing number of IUTs and antigen exposure as NIMA compared to exposure not as NIMA were associated with an increased risk of having antibodies (aOR for the latter 3.28; 95% CI 1.38-7.80). A longer BF period showed a trend towards a higher risk for antibodies (OR 1.17; 95% CI 0.97-1.43). The interaction term (NIMA by categorized months BF) revealed that a longer BF period was associated with a decreased risk of antibody formation against NIMA (aOR 0.66; 95% CI 0.48-0.93) compared to antigens not exposed to as NIMA (Table 3).

Sensitivity analyses with dichotomized periods of BF showed that BF for at least two months compared to a shorter period, and NIMA exposure were associated with an increased risk for antibodies (aOR 2.39; 95% CI 1.125.13 and aOR 3.34; 95% CI 1.55-7.24, respectively).

Again, the interaction term showed a decreased risk for antibodies against NIMA after at least two months BF (aOR 0.12; 95% CI 0.03-0.42). These associations remained when comparing at least three months BF to a shorter period. Continuation of BF beyond three months did not seem to provide additional protection against the formation of NIMA antibodies (Table 3).

For all analyses, antigen immunogenicity and number of IUTs were associated with an increased risk for antibodies, while IUT-year and the interaction term NIMA by antigen immunogenicity were not associated with antibody risk.

Discussion The role of exposure to RBC-NIMAs during fetal or neonatal life on later development of (non-D) RBC antibodies after re-exposure to the antigens during pregnancy

as IPAs or by transfusions was investigated in 125 threegeneration families with a (grand)child with HDFN treated with IUT. In contrast to women who were breastfed for less than two months, BF for at least two months was associated with a significantly lower incidence of alloantibodies to NIMAs compared to the same antigens not previously encountered as NIMA. The relevance of BF in the induction of tolerance against NIMA has mainly been studied in the context of the major histocompatibility complex (MHC) antigens in mice models.12,26,27 These studies clearly showed that a maximal immunoregulatory effect to NIMA is obtained after exposure both in utero and by BF.12,13 However, the underlying mechanism(s) remain elusive, and several immunological consequences of fetal/maternal interactions have been proposed both in mice and humans. BF in mice induced Foxp3+CD25+ T regulator cells potentially capable of regulating anti-maternal MHC immune responses.13,28 BF

Table 1. Characteristics of the 125 mothers.

Characteristics

N (%)*

Age at follow up, median (range), years 40 (24–52) BF duration in months, median (range) 2 (0-12) No breastfeeding 45 (36) One month 17 (14) Two months 15 (12) Three months 17 (14) Four to six months 18 (14) More than six months 13 (10) Birth order of first IUT treated child, median (range) 3 (1–7) Number of IUTs, median (range) 3 (1–10) Maternal-fetal major ABO compatibility – N/N tested (%)101/112 (90) HLA-DRB1*15 positive 38 (30) Antibody causing HDFN: Anti-D 93 (74) Anti-K 19 (15) Anti-c 11 (9) Anti-Kpa and -Cw 2 (2) BF: breastfeeding; IUT: intra-uterine transfusion; HDFN: hemolytic disease of the fetus and newborn. *: Data presented as number (%) of women unless stated otherwise.

Table 2. Variables associated with RBC antibody formation after mismatched antigen exposures, univariate analysis.

Variables

OR

95% CI

P

RBC antigen immunogenicity (low** [non Rh, K] vs. high [Rh, K]) Number of IUTs (1, 2, 3, 4 and >4) IUT year (5-year blocks) ABO compatibility** versus incompatibility HLA-DRB1*15 genotype absent** versus present Number of pregnancies (≤2, 3 and >3) Exposure not as a NIMA** vs. NIMA re-exposure Breastfeeding months (0, 1, 2, 3, 4-6 and >6) Antibodies after NIMA-mismatched exposures Antibodies after NIMA-matched exposures

17.3 1.22 0.85 1.26 1.10 0.98 0.78 0.98 1.08 0.82

10.0-29.9 1.04-1.43 0.68-1.06 0.62-2.57 0.69-1.75 0.75-1.27 0.50-1.21 0.87-1.10 0.93-1.25 0.67-1.01

<0.001 0.014 0.148 0.531 0.689 0.877 0.265 0.721 0.320 0.062

RBC: red blood cell; OR: odds ratio; CI: confidence interval; IUT: intra-uterine transfusion; NIMA: non-inherited maternal antigen; [non Rh, K]: Fya, Fyb, Jka, Jkb, M, S and s antigens; [Rh, K]: C, c, E, e and K antigens; **: reference.

haematologica | 2019; 104(2)

265

H. Schonewille et al.

A

B

C

Figure 1. Non-inherited maternal antigens (NIMA) and anti-IPA (inherited paternal antigens) immunity. A. The X-/- child (= “mother” in cohort) encounters X as NIMA during pregnancy and nursing from her heterozygous X-/+ mother (=”grandmother”), resulting in immunity or regulation against the X-antigen. B. During her (=”mother”) pregnancy of a X-/+ child this can determine whether she will form antibodies against the fetal X-IPA. C. During her (=”mother”) pregnancy of a X-/+ child she re-encounters X as a fetal (f) -IPA.

implies extended exposure to soluble and cellular maternal antigens and, in addition, extracellular vescicles (EVs) which can exert several immune effects. EVs are derived from a variety of cells including RBC, and several clinically relevant Rhesus, Kell, Duffy and Kidd blood group antigens are expressed.29,30 Erythrocyte derived EVs have immunomodulating properties.31 Up to 6 months after delivery, human colostrum and breast milk contains exosomes capable of increasing Foxp3+CD4+CD25+ allospecific T regulator cells in vitro.19 In the postnatal period, when the newborn is developing immunity against environmental threats, prolonged exposure to the maternal antigens may be perceived as auto-antigens requiring regulation.

Moreover, oral exposure to antigens and antigen-antibody complexes was reported to induce regulation in neonates as well as in adults.32,33 In D-sensitized women pregnant of a subsequent D-positive child, oral administration of Dantigen inhibited an expected boost in anti-D titer.34 Finally, breast milk provides an additional source of viable maternal cells that may increase maternal 266

microchimerism, which is considered a driving force of regulation leading to maintenance of tolerance for NIMAs.18,35,36 A different history of maternal BF may explain the discrepancy of finding a grandmother effect in various studies. Until manufactured baby milk became available, BF was part of our immunological education. Several, also cultural aspects, such as wealth and women’s emancipation, often replaced BF by cow milk. This may however have consequences that are still unknown. Another aspect of BF is the transmission of maternal cells which seem to play a role in immunity against malignant diseases in the child as shown by Amatay and colleagues, based on a meta-analysis suggesting that BF may strengthen antileukemic immunity in progeny.37 Our observations on the effect of oral exposure to NIMAs by BF may not just explain the historical controversies regarding the grandmother theory. Why only a minority of individuals form RBC alloantibodies after pregnancy and/or blood transfusions (responders), while haematologica | 2019; 104(2)

Breastfeeding reduces anti-NIMA immune response

Table 3. Variables associated with red blood cell antibody formation after mismatched antigen exposures, by categorized and dichotomized breastfeeding periods, multivariate analysis.*

BF duration Categorized 0, 1, 2, 3, 4-6, >6 Dichotomized 0 vs.>0 ≤1 vs.>1 ≤2 vs.>2 ≤3 vs >3 ≤4 vs. >4 ≤6 vs. >6 <2 vs. >3 2 and 3 vs. >3

Breastfeeding aOR 95% CI

NIMA exposure** aOR 95% CI

NIMA by months BF aOR 95% CI

1.17

0.97-1.43

3.28

1.38-7.80

0.66

0.48-0.93

1.11 1.15 2.39 1.95 1.34 1.08 2.37 1.11

0.54-2.26 0.56-2.33 1.12-5.13 0.83-4.59 0.53-3.43 0.35-3.32 0.93-6.05 0.41-2.96

1.81 1.78 3.34 2.49 1.77 1.55 3.57 1.15

0.68-4.79 0.71-4.41 1.55-7.24 1.23-5.06 0.92-3.39 0.83-2.92 1.47-8.70 0.36-3.69

0.77 0.78 0.12 0.41 0.37 0.86 0.11 0.35

0.23-2.53 0.24-2.48 0.03-0.42 0.03-0.59 0.07-1.98 0.13-5.61 0.02-0.51 0.07-1.80

*The complete table 3, including results from the variables, Immunogenicity, Number of IUTs, Year of IUT and the interaction term NIMA by Immunogenicity are in the Online Supplementary Appendix. BF: breastfeeding; NIMA: non-inherited maternal antigen; aOR: adjusted odds ratio; CI: confidence interval; **The reference is incidence of antibodies after exposure not as a NIMA.

Figure 2. Study flowchart of 125 mothers exposed to 330 non-D RBC antigens and antibody response. RBC: red blood cell; NIMA: non-inherited maternal antigen; HDFN: hemolytic disease of the fetus and newborn.

others do not, despite multiple exposures to mismatched RBC antigens (nonresponders), has yet to be unravelled and considered to be multifactorial. Factors such as antigen immunogenicity and dose, route of exposure, race, age, co-existence of inflammation during RBC antigen exposure, genetic factors (such as HLA and immunoregulatory gene polymorphisms), and medical (immunosuppressive) conditions have all been reported to enhance or ameliorate RBC alloantibody formation.38-42 The duration of exposure to NIMA by BF adds a new aspect to this already complex question.

Limitations Firstly, our study cohort - similar to the HLA immunized cohort on the renal transplant waiting list, mentioned before8 - consisted of highly immunized individuals, already possessing at inclusion at least one strong RBC antibody causing HDFN. We previously showed that once individuals produced an RBC antibody, there is over a 20% risk of forming additional antibodies upon subsequent exposures.21,43 The results in these extensively allohaematologica | 2019; 104(2)

exposed young women, of which 80% formed antibodies against multiple RBC antigens after delivery, may not be generalizable for first RBC antigen encounter and for immune-compromised patients. Secondly, unidentified or unknown antigen exposures that had not resulted in RBC antibodies may have been missed; however, it is likely that these were equally distributed in the distinct BF periods. Thirdly, Owen and colleagues evaluated anti-D formation in relation to the D-NIMA. As a result of RhD immunoprophylaxis, we could not repeat Owen’s design, however we did find support for a grandmother effect to non-D NIMAs after oral exposure by more than two months BF. Fourthly, the duration of BF was assessed through a questionnaire. BF was given decades before this study, and memory may not be exactly accurate. Lastly, in agreement with Owen, our study is hypothesis-generating and needs corroboration before drawing any definitive conclusion.

In conclusion, in women with HDFN-affected children, prolonged oral exposure to non-D NIMA by BF was associated with a significantly lower incidence of antibody production when later challenged to these NIMAs com267

H. Schonewille et al.

pared to exposure to the same antigens not previously exposed to as NIMAs. Ray Owen and colleagues concluded their 1954 publication on the grandmother theory with these words: “Presentation of this hypothesis here, on the basis of admittedly limited data, is justified by the hope that others in a position to test it will be encouraged to do so.”1

References 1. Owen RD, Wood HR, Foord AG, Sturgeon P, Baldwin LG. Evidence for actively acquired tolerance to Rh antigens. Proc Natl Acad Sci USA. 1954;40(6):420-424.

2. Billingham RE, Brent L, Medawar PB. Actively acquired tolerance of foreign cells.

Nature. 1953;172(4379):603-606.

3. Booth RD, Dunsford J, Grant J, Murray S. Haemolytic disease in first-born infants. Brit Med J. 1953;2(4826):41-42.

4. Ward HK, Walsh RJ, Kooptzoff O. Rh antigens and immunological tolerance. Nature.

1957;179(4574):1352-1353.

5. Mayeda K. The self-marker concept as applied to the Rh blood group system. Am J Hum Genet. 1962;14(3):281-289.

6. Taylor JF. Sensitization of Rh-negative daughters by their Rh-positive mothers. N Engl J Med. 1967;276(10):547-551.

7. Hattevig G, Jonsson M, Kjellman B, Khellman H, Messeter L, Tibblin E. Screening of Rh-antibodies in Rh-negative female infants with Rh-positive mothers.

Acta Paediatr Scand. 1981;70(4):541-545.

8. Claas FH, Gijbels Y, van der Velden-de Munck J, van Rood JJ. Induction of B cell unresponsiveness to noninherited maternal HLA antigen during fetal life. Science.

1988;241(4874):1815-1817.

9. Burlingham WJ, Grailer AP, Heisey DM, et al. The effect of tolerance to noninherited maternal HLA antigens on the survival of renal transplants from sibling donors. N Engl J Med. 1998;339(23):1657-1664.

10. Panajotopoulos N, Ianhez LE, Neumann J, Sabbaga E, Kalil J. Immunological tolerance in human transplantation. The possible existence of a maternal effect. Transplantation.

1990;50(3):443-445.

11. Pohanka E, Cohen N, Colombe BW, Lou C, Salvatierra O Jr, Garovoy MR. Non-inherited maternal HLA antigens and protection against sensitization. Lancet. 1990;336 (8722):1025-1028.

12. Andrassy J, Kusaka S, Jankowska-Gan E, et al. Tolerance to noninherited maternal MHC antigens in mice. J Immunol. 2003;171(10): 5554-5561.

13. Aoyama K, Koyama M, Matsuoka K, et al.

Improved outcome of allogeneic bone marrow transplantation due to breastfeedinginduced tolerance to maternal antigens.

Blood. 2009;113(8):1829-1833.

14. Kois WE, Campbell DA Jr, Lorber MI, Sweeton JC, Dafoe DC. Influence of breast feeding on subsequent reactivity to a related renal allograft. J Surg Res. 1984;37(2):89-93.

15. Kindgren E, Fredrikson M, Ludvigsson J. Early feeding and risk of juvenile idiopathic arthritis: a case control study in a prospective birth cohort. Pediatr Rheumatol 2017;15(1):46.

16. Conradi S, Malzahn U, Paul F, et al.

Breastfeeding is associated with lower risk for multiple sclerosis. Mult Scler.

2013;19(5):553-558

268

Acknowledgments The authors express their special thanks to all families who participated in the LOTUS study.

The LOTUS study group is a consortium of Sanquin and the Leiden University Medical Center departments of Immunohematology and Blood transfusion, Obstetrics and Neonatology.

17. Van Leeuwenhoek A. Arcana naturae detecta delphis batavorum. Apud Henricum a Krooneveld 1695.

18. Hassiotou F, Beltran A, Chetwynd E, et al.

Breastmilk is a novel source of stem cells with multilineage differentiation potential.

Stem Cells. 2012;30(10):2164-2174.

19. Admyre C, Johansson SM, Qazi KR, et al.

Exosomes with immune modulatory features are present in human breast milk. J Immunol. 2007;179(3):1969-1978.

20. Schonewille H, Prinsen-Zander KJ, Reijnart M, et al. Extended matched intrauterine transfusions reduce maternal Duffy, Kidd, and S antibody formation. Transfusion.

2015;55(12):2912-2919.

21. Schonewille H, Klumper FJ, van de Watering LM, Kanhai HH, Brand A. High additional maternal red cell alloimmunization after Rhesus- and K-matched intrauterine intravascular transfusions for hemolytic disease of the fetus. Am J Obstet Gynecol.

2007;196(2):e143-146.

22. Verduin EP, Lindenburg IT, Smits-Wintjens VE, et al. LOng-Term follow up after intraUterine transfusionS: the LOTUS study.

BMC Pregnancy Childbirth. 2010;10:77.

23. Verduin EP, Schonewille H, Brand A, et al.

High anti-HLA response in women exposed to intrauterine transfusions for severe alloimmune haemolytic disease is associated with mother-child HLA triplet mismatches, high anti-D titer, and new red blood cell antibody formation.

Transfusion.

2013;53(5):939-947.

24. Zwiers C, Lindenburg ITM, Klumper FJ, de Haas M, Oepkes D, Van Kamp IL. Complications of intrauterine intravascular blood transfusion: lessons learned after 1678 procedures. Ultrasound Obstet Gynecol.

2017;50(2):180-186 25. Verduin EP, Brand A, van de Watering LM, et al. The HLA-DRB1*15 phenotype is associated with multiple red blood cell and HLA antibody responsiveness. Transfusion.

2016;56(7):1849-1856 26. Molitor-Dart ML, Andrassy J, Haynes LD, Burlingham WJ. Tolerance induction or sensitization in mice exposed to noninherited maternal antigen (NIMA). Am J Transplant.

2008;8(11):2307-2315.

27. Opiela SJ, Levy RB, Adkins B. Murine neonates develop vigorous in vivo cytotoxic and Th1/Th2 responses upon exposure to low doses of NIMA-like alloantigens. Blood.

2008;112(4):1530-1538.

28. Matsuoka K, Ichinohe T, Hashimoto D, Asakura S, Tanimoto M, Teshima T. Fetal tolerance to maternal antigens improves the outcome of allogeneic bone marrow transplantation by a CD4+ CD25+ T-cell-dependent mechanism. Blood. 2006;107(1):404409.

29. Jaime-Pérez JC, Gómez-Almaguer D. Immunoreactivity of common red blood cell antigens in cytoskeleton-free red blood cell microvesicles. Arch Med Res. 2000;30(2): 169-171 30. Canellini G, Rubin O, Delobel J, Crettaz D,

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

Lion N, Tissot JD. Red blood cell microparticles and blood group antigens: an analysis by flow cytometry. Blood Transfus. 2012;10 (suppl 2):s39-45 Sadallah S, Eken C, Schifferli JA. Erythrocyte-derived ectosomes have immunosuppressive properties. J Leukoc Biol. 2008;84(5):1316-1325.

Hostmann A, Meyer T, Maul J, et al.

Preexisting antigen-specific immune responses are modulated by oral KLH feeding in humans. Eur J Immunol. 2015;45(7): 1991-1996.

Mosconi E, Rekima A, Seitz-Polski B, et al.

Breast milk immune complexes are potent inducers of oral tolerance in neonates and prevent asthma development. Mucosal Immunol. 2010;3(5):461-474.

Bierme SJ, Blanc M, Abbal M, Fournie A. Oral Rh treatment for severely immunised mothers. Lancet. 1979;1(8116):604-605.

Zhou L, Yoshimura Y, Huang Y, et al. Two independent pathways of maternal cell transmission to offspring: through placenta during pregnancy and by breast-feeding after birth. Immunology. 2000;101(4):570580.

Molitor ML, Haynes LD, Jankowska-Gan E, Mulder A, Burlingham WJ. HLA class I noninherited maternal antigens in cord blood and breast milk. Hum Immunol.

2004;65(3):231-239.

Amitay EL, Keinan-Boker L. Breastfeeding and childhood leukemia incidence: A metaanalysis and systematic review. JAMA Pediatr. 2015;169(6):e151025.

Karafin MS, Westlake M, Hauser RG, et al; NHLBI Recipient Epidemiology and Donor Evaluation Study-III (REDS-III).Risk factors for red blood cell alloimmunization in the Recipient Epidemiology and Donor Evaluation Study (REDS-III) database. Br J Haematol. 2018;181(5):672-681.

Schonewille H, Doxiadis II, Levering WH, Roelen DL, Claas FH, Brand A. HLA-DRB1 associations in individuals with single and multiple clinically relevant red blood cell antibodies. Transfusion. 2014;54(8):1971-1980.

Tatari-Calderone Z, Tamouza R, Le Bouder GP, et al. The association of CD81 polymorphisms with alloimmunization in sickle cell disease. Clin Dev Immunol. 2013;2013: 937846 Sippert EÂ, Visentainer JE, Alves HV, et al.

Red blood cell alloimmunization in patients with sickle cell disease: correlation with HLA and cytokine gene polymorphisms.

Transfusion. 2017;57(2):379-389 Meinderts SM, Sins JWR, Fijnvandraat K, et al. Non-classical FCGR2C haplotype is associated with protection from red blood cell allo-immunization in sickle cell disease.

Blood. 2017;130(19):2121-2130 Schonewille H, van de Watering LMG, Brand A. Additional RBC alloantibodies after blood transfusion in a non-hematological alloimmunized patient cohort. Is it time to take precautionary measures?

Transfusion. 2006;46(4):630-635.

haematologica | 2019; 104(2)

ARTICLE

Phagocyte Biology & its Disorders

Treatment outcomes and prognostic factors in adult patients with secondary hemophagocytic lymphohistiocytosis not associated with malignancy

Ferrata Storti Foundation

Jae-Ho Yoon, Sung-Soo Park, Young-Woo Jeon, Sung-Eun Lee, Byung-Sik Cho, Ki-Seong Eom, Yoo-Jin Kim, Hee-Je Kim, Seok Lee, Chang-Ki Min, Seok-Goo Cho and Jong Wook Lee Department of Hematology, Catholic Blood and Marrow Transplantation Center, Seoul St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul, Korea

Haematologica 2019 Volume 104(2):269-276

ABSTRACT

H

emophagocytic lymphohistiocytosis is an overwhelming systemic inflammatory process that is life-threatening if not treated appropriately. We analyzed prognostic factors in patients with secondary hemophagocytic lymphohistiocytosis excluding malignancy.

In this retrospective study, we analyzed 126 adult cases between 2001 and 2017. Treatment was based on dexamethasone with or without etoposide and cyclosporine. Patients who achieved a complete response by 4 weeks were defined as early stable responders, those who failed to achieve a complete response but showed continuous improvement until 8 weeks were defined as late responders, and those whose conditions waxed and waned until 8 weeks were defined as unstable responders.

Patients with hemophagocytic lymphohistiocytosis caused by EpsteinBarr virus had a worse 5-year overall survival compared to those whose disease was secondary to autoimmune disease, other infections, or unknown causes (25.1% versus 82.4%, 78.7% and 55.5%, respectively; P<0.001). We observed that the overall response rate at 4 weeks was similar, but decreased at 8 weeks in the Epstein-Barr virus subgroup from 75.5% to 51.0%, and finally decreased to 30.6%. Multivariate analysis revealed that 8-week treatment response was the most relevant factor for overall survival. Excluding 8-week response, the presence of EpsteinBarr virus, old age, hyperferritinemia, and thrombocytopenia were associated with poor survival. We established a prognostic model with the parameters: low-risk (score 0-1), intermediate-risk (score 2), and highrisk (score ≥3). These groups had 5-year overall survival rates of 92.1%, 36.8%, and 18.0%, respectively (P<0.001). We found that 8-week treatment response was a good predictor for overall survival, and that Epstein-Barr virus, old age, thrombocytopenia, and hyperferritinemia were associated with poor survival outcomes. Physicians should take care to identify high-risk patients for appropriate treatment strategies.

Introduction Secondary hemophagocytic lymphohistiocytosis (HLH) without a family history or known genetic predisposition is considered a rare clinical condition and its underlying pathophysiological mechanisms have not yet been well elucidated. The distinction between primary HLH and secondary HLH is not clear because of the emergence of new gene defects, and the diagnosis is difficult because the symptoms and signs of HLH overlap with those of other severe conditions such as sepsis, multi-organ failure, and progression of malignancies or rheumatologic disorders.1,2 The incidence and prevalence of HLH may, therefore, be under- or overestimated if clinicians lack suspicion of and specific knowledge about HLH, especially in adult patients.3 HLH is a life-threatening inflammatory disease, and therefore early diagnosis and urgent treatment, including dexamethasone, cyclosporine, and etoposide, are haematologica | 2019; 104(2)

Correspondence: jwlee@catholic.ac.kr

Received: May 27, 2018.

Accepted: September 11, 2018.

Pre-published: September 13, 2018. doi:10.3324/haematol.2018.198655 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/269 ©2019 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.

269

J.H. Yoon et al.

important for survival.4-6 However, even when specific therapies are administered promptly, treatment response and overall survival rates remain poor, especially when the condition is associated with malignancies.7,8 Poor survival and frequent relapse are common not only in patients with malignancy-associated HLH, but also in those with active Epstein-Barr virus (EBV) infection and in some highrisk patients with HLH of unknown cause.9-11 For relapsed or refractory HLH patients, several salvage treatments12-18 followed by allogeneic hematopoietic cell transplantation (HCT) are now considered,19-21 but many of these patients die of rapid deterioration due to severe sepsis and multiorgan failure. It is, therefore, important for physicians to identify high-risk patients earlier in the course of treatment, and to apply second-line therapy and prepare the patient immediately for allogeneic HCT. However, there are few data identifying relevant prognostic factors and enabling risk-stratification for adult patients with secondary HLH.8,11 In this study, we initially analyzed treatment response and survival outcomes among 126 adult patients with non-malignancy-associated secondary HLH according to the underlying causes of the disease. We then assessed factors predictive of treatment response and survival outcome. Finally, we constructed a prognostic model using risk-stratification according to predictive factors in order to identify high-risk patients who should be considered for early second-line therapies, including salvage regimens and allogeneic HCT.

sented with fever, some criteria of HLH2004,5,22 and increased histiocytosis with active hemophagocytosis in bone marrow. Of these 165 patients, we excluded 12 who did not satisfy the full diagnostic criteria and 27 with malignancy-associated HLH (Figure 1 and Online Supplementary Methods for Identification of the causes of HLH). Ultimately, we analyzed 126 patients (median age 45 years, range 15-85 years) with non-malignancy-associated secondary HLH due to several causes. This research was conducted in accordance with Institutional Review Board and Ethics Committee guidelines of the Catholic Medical Center (KC15RISI0863) and the principles of the Declaration of Helsinki.

Parameters associated with hemophagocytic lymphohistiocytosis

Methods

All parameters included in the diagnostic guidelines for HLH were reviewed: fever, splenomegaly, cytopenia affecting at least two lineages, triglycerides, fibrinogen, ferritin, and natural killer (NK)-cell cytotoxic activity. Laboratory findings were serially accessed and the lowest levels were captured for complete blood cell counts, fibrinogen, albumin, and prothrombin time before treatment. The highest levels within 4 weeks after initial treatment were used for ferritin, aspartate aminotransferase, alanine aminotransferase, total bilirubin, lactate dehydrogenase, and triglycerides in order to assess clinical outcomes. We have analyzed NK-cell cytotoxicity by a flow cytometry-based assay since 2012.23 We did not check levels of soluble CD25. Thus, the majority of patients enrolled in the current study satisfied at least five out of six criteria for the diagnosis of HLH, excluding the soluble CD25 and NK cytotoxicity criteria. An association with EBV was evaluated by both serology and EBV DNA real-time quantitative polymerase chain reaction (RQ-PCR) analysis (See the Online Supplementary Methods for Identification of the causes of HLH).

Patients

Treatments

We retrospectively reviewed records for 165 adult patients who were diagnosed with HLH from 2001 to 2017. All patients pre-

Treatment strategies were based on HLH-94 protocols, which consisted mainly of dexamethasone, etoposide, and additional

Figure 1. Consort diagram of enrolled patients. A total of 126 patients with secondary hemophagocytic lymphohistiocytosis without malignancy were enrolled. HLH: hemophagocytic lymphohistiocytosis; EBV: Epstein-Barr virus.

270

haematologica | 2019; 104(2)

Prognostic factors in secondary adult HLH

cyclosporine.5,6,24 Final treatment response was evaluated after 8 weeks of induction according to the HLH-94 protocol. Patients exhibiting a complete response at 8 weeks were given maintenance therapy with low-dose steroids and cyclosporine for 2 to 4 more weeks, while patients who did not achieve a complete response or who were positive for EBV according to RQ-PCR (> 3-log) were considered for continuation therapy. If the EBV RQPCR level increased by at least 1-log with features of relapsing HLH, rituximab or alemtuzumab was given. Five patients who relapsed or had refractory disease were treated with allogeneic HCT using fludarabine (30 mg/m2/day for 5 days), busulfan (3.2 mg/kg/day for 2 days), and antithymocyte globulin (ATG, Thymoglobulin®, 2.5 mg/kg for 2 days).

Response assessment Complete response was defined as resolution of all clinical signs and symptoms, as well as recovery of the complete blood count and normalization of abnormal laboratory findings associated with HLH. Progressive disease was identified when both cytopenia and abnormal laboratory findings remained and partial response was defined when patients achieved either complete blood count recovery alone without normalization of laboratory findings or improvement of HLH-related laboratory findings alone without complete blood count recovery. We evaluated treatment response at 4 and 8 weeks after treatment, and also evaluated dynamic responses according to the response time and disease progression within 8 weeks. Patients with stable complete response at both 4 weeks and 8 weeks were classified as early stable responders, patients who failed to achieve a complete response by 4 weeks but showed continuous response until 8 weeks were classified as late stable responders, patients who showed only transient response and eventually progressed until 8 weeks were defined as unstable responders, and patients exhibiting no response at all were classified as non-responders. Details on the statistical analysis are provided in the Online Supplementary Methods.

Results Baseline characteristics of the patients The causes of HLH in the 126 patients enrolled in this study were categorized as EBV-associated (n=49), infection other than EBV (n=24), and autoimmune disease (n=17). Among the 49 cases of EBV-associated HLH, five were concomitantly related with an autoimmune disease.

We were unable to identify related causes in 36 (28.6%) patients (Figure 1). The baseline characteristics of the patients and their responses to treatment are presented in Table 1. Fever was observed in all patients and 118 (93.6%) patients presented with splenomegaly. Infections other than EBV included hepatitis A virus (n=5), cytomegalovirus (n=2), Staphylococcus aureus (n=3), mycoplasma (n=3), parvovirus (n=3), human immunodeficiency virus (n=2), mycobacterium (n=2), malaria (n=2), amebiasis (n=1), anaplasmosis (n=1), Hantaan virus (n=1), herpes simplex virus (n=1), influenza virus (n=1), and enterococcus (n=1). Patients with Kaposi sarcoma caused by human immunodeficiency virus were considered to have malignancy-associated HLH. Autoimmune diseases and similar features included systemic lupus erythematosus (n=8), Kikuchi disease (n=6), Behçet syndrome (n=2), Sjögren syndrome (n=2), Graves disease (n=1), rheumatoid arthritis (n=1), autoimmune hemolytic anemia (n=1), and ulcerative colitis (n=1). Among the 35 EBV-associated haematologica | 2019; 104(2)

HLH patients with available serology results, seven showed a recent primary infection, five reactivation, and 23 a past infection. In addition, initial bone marrow findings revealed that 36 out of 49 patients (73.5%) with EBVassociated HLH had active and frequent hemophagocytosis, while 17 out of the remaining 77 patients (22.1%) showed active hemophagocytosis.

Treatment outcomes according to causes of hemophagocytic lymphohistiocytosis After excluding four patients (2 with hepatitis A and an unknown cause of HLH who showed a self-limiting disease course, and 2 cases with progressive multi-organ failure before treatment), 122 patients were treated. The HLH-94 protocol was applied in 81 (64.3%) patients: 23 (63.8%) with HLH of unknown cause, 40 (81.6%) with EBV-HLH, 13 (54.2%) with an infection other than EBV, and five (29.4%) with an autoimmune disease).

Dexamethasone plus cyclosporine treatment was admin-

Table 1. Baseline characteristics and treatment outcome of patients with non-malignancy-associated secondary HLH (n=126).

Gender, male; n(%) 64 (50.8%) Age, years; median (range) 45.0 (15-85) Time from symptom to diagnosis, days; median (range) 13 (0 – 82) Fever; n(%) 126 (100%) Splenomegaly; n(%) 118 (93.6%) Lowest level prior to therapy; median (range) Leukocytes (x109/L) 2.39 (0.14-31.9) Neutrophils (x109/L) 1.08 (0.01-24.1) Hemoglobin (g/dL) 9.3 (3.7-16.3) Platelets (x109/L) 44.0 (5-307) Fibrinogen (mg/dL) 215.0 (4 – 513) Albumin (g/dL) 2.8 (1.7-4.1) Prothrombin time (%) 69.0 (10-145) Highest level within 4 weeks after treatment; median (range) Ferritin (ng/mL) 4702 (278-214000) Aspartate aminotransferase (U/L) 196 (12-4990) Alanine aminotransferase (U/L) 134 (5-5117) Total bilirubin (mg/dL) 1.17 (0.07-19.9) Lactate dehydrogenase (U/L) 1670 (285-30000) Triglyceride (mg/dL) 173.5 (7-813) C-reactive protein (mg/dL) 4.9 (0.02-34.0) Causes of HLH; n(%) EBV-associated 49 (38.9%) Median highest level (log) 4.5 (2.7-7.5) ≥ 5.0-log 20 (41.0%) < 5.0-log 29 (59.0%) Infection other than EBV 24 (19.0%) Autoimmune disease 17 (13.5%) Unknown cause 36 (28.6%) Treated patients; n(%) 122 (96.8%) HLH94-based therapy 81 (64.3%) Steroid + cyclosporine 18 (14.3%) Steroid alone 23 (18.2%) Complete remission achieved after treatments; n(%) 81 (64.3%) Treatment response at 8 weeks; n(%) Early stable responder 42 (34.4%) Late stable responder 36 (29.5%) Unstable responder 19 (15.6%) Non-responder 25 (20.5%) HLH: hemophagocytic lymphohistiocytosis; EBV: Epstein-Barr virus.

271

J.H. Yoon et al.

istered to 18 (14.3%) patients, while 23 (18.2%) patients were given dexamethasone alone. Patients who had mild symptoms and improved rapidly were treated with dexamethasone alone and further therapy was not needed. In contrast, we were unable to administer cyclosporine and etoposide to patients with significant deterioration of organ function or a poor general condition. Of the 81 (64.3%) patients who achieved a complete response after treatments, 43 achieved this response by 4 weeks after treatment and the other 38 achieved it by 8 weeks or later.

Of the 27 patients who relapsed, one relapsed between 4 and 8 weeks and 26 relapsed after 8 weeks during the course of treatment. After relapsing, most of the patients were treated with the HLH-94 protocol again. For relapsed EBV-HLH, we considered rituximab (n=6) or alemtuzumab (n=2) concomitantly, and two patients were treated with allogeneic HCT. With regards to dynamic responses, 42 (34.4%) patients were early stable responders, 36 (29.5%) were late stable responders, 19 (15.6%) were unstable responders, and 25 (20.5%) were primary refractory non-responders (Table 1). Detailed responses according to causes of HLH are as follows. In patients with EBV-HLH, the overall response rate (complete plus partial responses) at 4 weeks was 75.5% (complete response in 16, partial response in 21) but decreased to 51.0% (complete response in 17, partial response in 8) at 8 weeks, and finally decreased to 30.6% (complete response in 15). The overall response rates at 4 weeks and 8 weeks for patients with HLH associated with infections other than EBV were 62.5% (complete response in 10, partial response in 5) and 75% (complete response in 14, partial response in 4), respectively, for those with autoimmune diseases, 82.3% (complete response in 9, partial response in 5) and 82.3% (complete response in 12, partial response in 2), respectively, and for those with HLH of unknown cause, 66.7% (complete response in 8, partial response in 16) and 66.7% (complete response in 19, partial response in 5), respectively. Patients with HLH associated with an autoimmune disease or infection other than EBV had superior 5-year overall survival rates (82.4% and 78.7%, respectively) and lower cumulative incidences

of progression (29.4% and 22.7%, respectively) compared to those of patients with EBV-associated HLH. The estimated 5-year overall survival rate and cumulative incidence of progression of patients with EBV-associated HLH were 25.1% and 77.5%, respectively (Figure 2).

A

B

Survival outcomes according to treatment response Estimated 5-year overall survival rates of patients who achieved a complete response by 4 weeks (n=43) and 8 weeks (n=62) were 82.8% and 82.7%, respectively, while the overall survival rate of patients who achieved a partial response until 8 weeks (n=19) was 68.4% (Figure 3A,B).

As we recognized four subgroups defined according to dynamic treatment response, there were 42 patients in the early stable responder group, 36 in the late stable responder group, 19 in the unstable responder group, and 25 in the non-responder group. Figure 3C,D presents the overall survival and cumulative incidence of progression curves of patients divided according to dynamic treatment response. Early stable responders and late stable responders had 5-year overall survival rates of 80.7% and 73.7%, respectively, and 5-year cumulative incidences of progression of 20.2% and 43.1%, respectively, while almost all patients in the unstable and non-responder groups died of disease progression (P<0.001).

Prognostic factors and risk stratification We found that most parameters associated with HLH disease activity were predictive of survival outcomes, and many were correlated with each other. Age was correlated with albumin and alanine aminotransferase levels, EBVassociation with triglyceride and fibrinogen levels, anemia with albumin levels, thrombocytopenia with prothrombin time, total bilirubin and erythrocyte sedimentation rate, and ferritin with prothrombin time, fibrinogen, erythrocyte sedimentation rate, and lactate dehydrogenase levels.

Multivariate analyses were performed using age, EBVassociation, thrombocytopenia, ferritin, and dynamic treatment response after excluding correlated parameters (Table 2). Remaining an unstable responder or nonresponder at 8 weeks was the most powerful predictive

Figure 2. Survival outcomes according to the cause of secondary hemophagocytic lymphohistiocytosis. (A) Comparison of 5-year overall survival rates according to the cause of secondary hemophagocytic lymphohistiocytosis. (B) Comparison of cumulative incidences of progression after treatment according to the cause of secondary hemophagocytic lymphohistiocytosis. OS: overall survival; HLH: hemophagocytic lymphohistiocytosis; EBV: Epstein-Barr virus.

272

haematologica | 2019; 104(2)

Prognostic factors in secondary adult HLH

factor for poor overall survival; age >45 years and low platelet counts (<35x109/L) were also significant predictive factors for poor overall survival. When we excluded treatment response as a parameter, age >45 years, EBV-association, thrombocytopenia (platelets <35x109/L), and hyperferritinemia (>20,000 ng/mL) were associated with poor overall survival. All of these factors were also associated with higher cumulative incidences of progression and poor 8-week response (data not shown). We ultimately selected four factors and weighted them based on the hazard ratio: age >45 years (2 points), EBV-association (1 point), low platelet count (<35x109/L) (1 point), and hyperferritinemia (>20,000 ng/mL) (1 point). Using these variables, we performed risk scoring and stratification. The 5year overall survival rates of low-risk (score 0-1), intermediate-risk (score 2), and poor-risk (score â&#x2030;Ľ3) patients were 92.1%, 36.8%, and 18.0%, respectively, while the corresponding 5-year cumulative incidences of progression were 18.5%, 59.4% and 87.0% (both P<0.001) (Figure 4A,B). In the EBV-HLH subgroup, the 5-year overall survival rate was 59.2% in early stable responders, 44.4% in the late responders, and 0.0% in non-responders.

A

B

C

D

Multivariate analysis in the EBV-HLH subgroup revealed that maximal EBV RQ-PCR level greater than 5-log, age >45 years, and hyperbilirubinemia were associated with poor overall survival

Discussion Causes of HLH and the proportions of each identified in the current study differed from those of previous studies, in which the most common cause of secondary HLH was a hematologic malignancy.3,7,8,25 Some patients with malignancy-associated HLH die early from rapid deterioration without a suspicion of HLH and their prognosis depends on the outcome of the malignant disease. In the current study we, therefore, analyzed treatment response and prognostic factors for adult secondary HLH after excluding those with malignancy-associated HLH. To the best of our knowledge, this is the largest data set used to analyze prognostic factors in adult patients with secondary HLH without malignancies.

Excluding malignancies, EBV was the most common

Figure 3. Survival outcomes according to treatment response. (A, B) Five-year overall survival according to treatment response at (A) 4 weeks and (B) 8 weeks. (C,D).

Five-year (C) overall survival and (D) cumulative incidence progression according to dynamic treatment response at 8 weeks. OS: overall survival; CIP: cumulative incidence of progression.

haematologica | 2019; 104(2)

273

J.H. Yoon et al.

cause of secondary HLH in this study. The prognosis of patients with EBV-associated HLH is expected to be better than that of those with malignancies or primary HLH, but if patients are not treated appropriately, their survival outcomes are dismal.26,27 Our data revealed that patients with EBV-associated HLH had similar responses to treatment in the early period but many relapsed afterward, which finally resulted in these patients having significantly inferior overall survivals and higher cumulative incidences of progression compared to patients with HLH related to other causes. It is, therefore, important to decide on early salvage treatment, including allogeneic HCT, for poor responders with EBV-associated HLH. Previous reports suggested a role of acyclovir in the management of EBVassociated disease,28,29 but no relevant data are available on the additional effects of antiviral therapy.9 Salvage treatments include rituximab12,17,30 and alemtuzumab31 followed by allogeneic HCT.32,33 Although we used rituximab in six patients and alemtuzumab in two in the setting of relapsed disease, none of the patients showed a response to additional therapy suggesting that incorporation of rituximab into the HLH protocol might be an alternative approach as initial treatment in EBV-associated HLH. Of the three patients with relapsed/refractory EBV-HLH treated with allogeneic HCT, one died of relapse, one died of septic pneumonia, and one is alive without relapse.

We calculated survival outcomes of responders and nonresponders at 4 weeks and 8 weeks, respectively, and

found that patients who did not achieve at least a stable partial response at 8 weeks had very high mortality. A recent prospective study investigating the effects of a salvage regimen using liposomal doxorubicin applied salvage chemotherapy for refractory patients who were defined as non-responsive to initial therapy at 2 weeks.18 However, our data show that patients who failed to achieve a complete response by 4 weeks after treatment could still achieve long-term survival if they exhibited a continuous and stable response until 8 weeks. The HLH-94 protocol recommends continuation therapy when disease is active at 8 weeks after initial therapy, and our study population included patients who proceeded to continuation therapy.

Among our 21 patients who were given continuation therapy, 16 additional patients ultimately achieved a complete response and two of these patients were treated with allogeneic HCT. There are no data comparing survival between patients receiving continuation therapy and other salvage regimens, or on the proper timing for allogeneic HCT, especially in patients with EBV-HLH. However, as our data indicate that the course of EBV-HLH after relapse is very progressive, urgent preparation for allogeneic HCT is recommended. We also identified that the risk of relapse among patients with EBV-HLH was higher in those with an age >45 years, maximal EBV RQPCR higher than 5-log, and hyperbilirubinemia.

Apart from non-responders at 8 weeks, who have a particularly poor outcome, we further identified high-risk

Table 2. Univariate and multivariate analyses for parameters affecting overall survival in non-malignancy-associated HLH.

Univariate analysis

Age > 45 years old** (n=61, 49.4%) EBV-associated** (n=49, 45.9%) Hemoglobin <9.0 g/dL (n=47, 42.5%) Platelet count <35 x109/L ** (n=47, 36.8%) Alanine aminotransferase >165 U/L (n=55, 32.2%) Bilirubin >1.6 mg/dL (n=51, 43.7%) Albumin <3.0 g/dL (n=76, 49.4%) Triglycerides >200 mg/dL (n=50, 43.7%) Ferritin >20,000 ng/mL** (n=26, 50.6%) Fibrinogin <150 mg/dL (n=45, 35.6%) Dynamic treatment response** Early stable responder (n=42) Late responder (n=36) Unstable or non-responder (n=44)

5-year OS

P

26.8% (vs. 74.9%) 25.1% (vs. 69.3%) 38.7% (vs. 75.6%) 32.8% (vs. 64.7%) 67.7% (vs. 40.6%) 30.1% (vs. 68.5%) 34.2% (vs. 80.8%) 38.1% (vs. 61.4%) 23.1%% (vs. 60.1%) 33.4% (vs. 63.3%)

< 0.001*

87.0% 73.7% 0.0%

Multivariate analysis Treatment response included Treatment response excluded HR P HR P 2.54 (1.4-4.7)

0.003*

< 0.001*

3.38 (1.9-6.1) 2.16 (1.2-3.8)

< 0.001*

2.20 (1.3-3.8)

0.004*

1.99 (1.1-3.6)

0.021*

0.006*

< 0.001* < 0.001*

1.99 (1.1-3.5)

0.016*

0.0055* < 0.001* < 0.001* 0.0256* < 0.001* 0.002*

< 0.001*

1.0 1.73 (0.6-5.2) 17.3 (6.5-45.9)

0.326 < 0.001*

OS: overall survival; HLH: hemophagocytic lymphohistiocytosis; HR: hazard ratio; EBV: Epstein-Barr virus; *P < 0.05. **Variables used in multivariate analyses were selected after excluding significant correlations: age, EBV-association, thrombocytopenia, hyperferritinemia, dynamic treatment response

274

haematologica | 2019; 104(2)

Prognostic factors in secondary adult HLH

A

B

Figure 4. Survival outcomes according to hemophagocytic lymphohistiocytosis risk score. (A) Overall survival according to risk score. (B) Cumulative incidence of progression according to risk score. Low risk (score 0-1), intermediate risk (score 2), high risk (score â&#x2030;Ľ3). OS: overall survival; CIP: cumulative incidence of progression.

patients according to variables such as age, thrombocytopenia, EBV-association, and high ferritin level, which were significant factors for poor overall survival in multivariate analysis. As we observed a high probability of relapse or refractory disease in high-risk patients, early intervention with salvage therapy and allogeneic HCT should be considered in selected patients. In contrast to high-risk or relapsed patients, we also observed patients who showed good responses to initial steroid therapy alone. Response to steroid alone was expected in some patients with an infection other than EBV, an autoimmune disease, or HLH of unknown cause. We may, therefore, consider delaying etoposide in some HLH patients without EBV and those with an autoimmune disease.

Although the distinction between primary and secondary HLH is obvious, several genetic abnormalities found in adult patients support the hypothesis that many cases of secondary HLH have an underlying genetic predisposition.34-36 This issue will remain challenging unless we can assess patients for the presence of genetic defects in every suspicious case, and more genes associated with cytotoxic function may be identified in the future.37 We were unable to check for all possible genetic abnormalities, except for some patients with negative results for UNC13D and PRF1 mutations, and we lack reference data regarding gene defects in adult patients. We, therefore, hypothesized that all enrolled patients without gene study data were cases of secondary HLH unless they did not satisfy HLH-2004 criteria.5 In addition, all patients enrolled in this study were diagnosed based on increased histiocytosis

References 1. Janka GE. Familial and acquired hemophagocytic lymphohistiocytosis. Eur J Pediatr. 2007;166(2):95-109.

2. Janka GE. Familial and acquired hemophagocytic lymphohistiocytosis. Annu Rev Med. 2012;63:233-246.

haematologica | 2019; 104(2)

with active hemophagocytosis in a bone marrow study, although previous reports showed that 30% of HLH cases may not present with hemophagocytosis in an initial bone marrow study.3,38 Conversely, severely ill patients not satisfying the diagnostic criteria for HLH sometimes present with bone marrow hemophagocytosis.39 As we initially identified HLH patients based on bone marrow results, it is possible that we missed some HLH patients without evidence of bone marrow hemophagocytosis or who did not undergo a bone marrow examination.

HLH is a rare disease for which well-organized prospective studies are lacking: analysis of prognostic factors for secondary HLH is, therefore, difficult and few data are available for risk-stratification. Although our data originated from a retrospective study, patients were consistently treated and managed, and the number of patients was sufficient for the final analysis. As we excluded cases of malignancy-associated HLH, treatment response to protocols and survival outcomes were relevant for risk-stratification. Our prognostic risk model should be validated by larger studies. Combinations of several predictive factors at diagnosis and 8-week treatment response can allow clinicians to identify patients at high risk of disease progression, and second-line therapy including allogeneic HCT should be considered earlier in the course of treatment.

Acknowledgments This study was supported by a grant from the National R&D Program for Cancer Control, Ministry for Health and Welfare, Republic of Korea (1020370).

3. Riviere S, Galicier L, Coppo P, et al. Reactive hemophagocytic syndrome in adults: a retrospective analysis of 162 patients. Am J Med. 2014;127(11):1118-1125.

4. Lee JS, Kang JH, Lee GK, Park HJ. Successful treatment of Epstein-Barr virus-associated hemophagocytic lymphohistiocytosis with HLH-94 protocol. J Korean Med Sci. 2005;20(2):209-214.

5. Henter JI, Horne A, Arico M, et al. HLH2004: diagnostic and therapeutic guidelines for hemophagocytic lymphohistiocytosis.

Pediatr Blood Cancer. 2007;48(2):124-131.

6. Trottestam H, Horne A, Arico M, et al.

Chemoimmunotherapy for hemophagocytic lymphohistiocytosis: long-term results of the HLH-94 treatment protocol. Blood.

2011;118(17):4577-4584.

275

J.H. Yoon et al.

7. Takahashi N, Chubachi A, Kume M, et al. A clinical analysis of 52 adult patients with hemophagocytic syndrome: the prognostic significance of the underlying diseases. Int J Hematol. 2001;74(2):209-213.

8. Li J, Wang Q, Zheng W, et al.

Hemophagocytic lymphohistiocytosis: clinical analysis of 103 adult patients. Medicine.

2014;93(2):100-105.

9. Imashuku S. Treatment of Epstein-Barr virus-related hemophagocytic lymphohistiocytosis (EBV-HLH); update 2010. J Pediatr Hematol Oncol. 2011;33(1):35-39.

10. Imashuku S, Tabata Y, Teramura T, Hibi S. Treatment strategies for Epstein-Barr virusassociated hemophagocytic lymphohistiocytosis (EBV-HLH). Leuk Lymphoma.

2000;39(1-2):37-49.

11. Kaito K, Kobayashi M, Katayama T, et al.

Prognostic factors of hemophagocytic syndrome in adults: analysis of 34 cases. Eur J Haematol. 1997;59(4):247-253.

12. Milone MC, Tsai DE, Hodinka RL, et al.

Treatment of primary Epstein-Barr virus infection in patients with X-linked lymphoproliferative disease using B-cell-directed therapy. Blood. 2005;105(3):994-996.

13. Henzan T, Nagafuji K, Tsukamoto H, et al.

Success with infliximab in treating refractory hemophagocytic lymphohistiocytosis.

Am J Hematol. 2006;81(1):59-61.

14. Olin RL, Nichols KE, Naghashpour M, et al.

Successful use of the anti-CD25 antibody daclizumab in an adult patient with hemophagocytic lymphohistiocytosis. Am J Hematol. 2008;83(9):747-749.

15. Shiraishi A, Ohga S, Doi T, et al. Treatment choice of immunotherapy or further chemotherapy for Epstein-Barr virus-associated hemophagocytic lymphohistiocytosis.

Pediatr Blood Cancer. 2012;59(2):265-270.

16. Marsh RA, Allen CE, McClain KL, et al.

Salvage therapy of refractory hemophagocytic lymphohistiocytosis with alemtuzumab. Pediatr Blood Cancer. 2013;60(1):101109.

17. Chellapandian D, Das R, Zelley K, et al.

Treatment of Epstein Barr virus-induced haemophagocytic lymphohistiocytosis with rituximab-containing chemo-immunotherapeutic regimens. Br J Haematol. 2013;162 (3):376-382.

18. Wang Y, Huang W, Hu L, et al. Multicenter study of combination DEP regimen as a salvage therapy for adult refractory hemophagocytic lymphohistiocytosis. Blood.

2015;126(19):2186-2192.

276

19. Henter JI, Samuelsson-Horne A, Arico M, et al. Treatment of hemophagocytic lymphohistiocytosis with HLH-94 immunochemotherapy and bone marrow transplantation. Blood. 2002;100(7):2367-2373.

20. Cooper N, Rao K, Gilmour K, et al. Stem cell transplantation with reduced-intensity conditioning for hemophagocytic lymphohistiocytosis. Blood. 2006;107(3):1233-1236.

21. Machaczka M, Nahi H, Karbach H, Klimkowska M, Hagglund H. Successful treatment of recurrent malignancy-associated hemophagocytic lymphohistiocytosis with a modified HLH-94 immunochemotherapy and allogeneic stem cell transplantation. Med Oncol. 2012;29(2): 1231-1236.

22. Henter JI, Elinder G, Ost A. Diagnostic guidelines for hemophagocytic lymphohistiocytosis. The FHL Study Group of the Histiocyte Society. Semin Oncol. 1991;18(1): 29-33.

23. Park KH, Park H, Kim M, Kim Y, Han K, Oh EJ. Evaluation of NK cell function by flowcytometric measurement and impedance based assay using real-time cell electronic sensing system. Biomed Res Int. 2013;2013:210726.

24. Henter JI, Arico M, Egeler RM, et al. HLH94: a treatment protocol for hemophagocytic lymphohistiocytosis. HLH study group of the Histiocyte Society. Med Pediatr Oncol.

1997;28(5):342-347.

25. Parikh SA, Kapoor P, Letendre L, Kumar S, Wolanskyj AP. Prognostic factors and outcomes of adults with hemophagocytic lymphohistiocytosis. Mayo Clin Proc. 2014;89 (4):484-492.

26. Imashuku S, Hibi S, Kuriyama K, et al.

Management of severe neutropenia with cyclosporin during initial treatment of Epstein-Barr virus-related hemophagocytic lymphohistiocytosis. Leuk Lymphoma.

2000;36(3-4):339-346.

27. Imashuku S, Hibi S, Ohara T, et al. Effective control of Epstein-Barr virus-related hemophagocytic lymphohistiocytosis with immunochemotherapy. Histiocyte Society.

Blood. 1999;93(6):1869-1874.

28. Sullivan JL, Woda BA, Herrod HG, Koh G, Rivara FP, Mulder C. Epstein-Barr virus-associated hemophagocytic syndrome: virological and immunopathological studies. Blood.

1985;65(5):1097-1104.

29. Davis CL, Harrison KL, McVicar JP, Forg PJ, Bronner MP, Marsh CL. Antiviral prophylaxis and the Epstein Barr virus-related post-

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

transplant lymphoproliferative disorder.

Clin Transplant. 1995;9(1):53-59.

Balamuth NJ, Nichols KE, Paessler M, Teachey DT. Use of rituximab in conjunction with immunosuppressive chemotherapy as a novel therapy for Epstein Barr virusassociated hemophagocytic lymphohistiocytosis. J Pediatr Hematol Oncol.

2007;29(8):569-573.

Strout MP, Seropian S, Berliner N. Alemtuzumab as a bridge to allogeneic SCT in atypical hemophagocytic lymphohistiocytosis. Nat Rev Clin Oncol. 2010;7(7):415420.

Gratama JW, Oosterveer MA, Zwaan FE, Lepoutre J, Klein G, Ernberg I. Eradication of Epstein-Barr virus by allogeneic bone marrow transplantation: implications for sites of viral latency. Proc Natl Acad Sci USA. 1988;85(22):8693-8696.

Ohga S, Kudo K, Ishii E, et al. Hematopoietic stem cell transplantation for familial hemophagocytic lymphohistiocytosis and Epstein-Barr virus-associated hemophagocytic lymphohistiocytosis in Japan. Pediatr Blood Cancer. 2010;54(2):299-306.

Zhang K, Jordan MB, Marsh RA, et al.

Hypomorphic mutations in PRF1, MUNC13-4, and STXBP2 are associated with adult-onset familial HLH. Blood.

2011;118(22):5794-5798.

Jordan MB, Allen CE, Weitzman S, Filipovich AH, McClain KL. How I treat hemophagocytic lymphohistiocytosis.

Blood. 2011;118(15):4041-4052.

Zhang K, Chandrakasan S, Chapman H, et al. Synergistic defects of different molecules in the cytotoxic pathway lead to clinical familial hemophagocytic lymphohistiocytosis. Blood. 2014;124(8):1331-1334.

Filipovich AH. Hemophagocytic lymphohistiocytosis (HLH) and related disorders.

Hematology Am Soc Hematol Educ Program. 2009;127-131.

Gupta A, Tyrrell P, Valani R, Benseler S, Weitzman S, Abdelhaleem M. The role of the initial bone marrow aspirate in the diagnosis of hemophagocytic lymphohistiocytosis. Pediatr Blood Cancer. 2008;51(3):402404.

Strauss R, Neureiter D, Westenburger B, Wehler M, Kirchner T, Hahn EG. Multifactorial risk analysis of bone marrow histiocytic hyperplasia with hemophagocytosis in critically ill medical patients--a postmortem clinicopathologic analysis. Crit Care Med. 2004;32(6):1316-1321.

haematologica | 2019; 104(2)

ARTICLE

Acute Myeloid Leukemia

Comprehensive genetic diagnosis of acute myeloid leukemia by next-generation sequencing

Ferrata Storti Foundation

Elisabeth K.M. Mack,1 André Marquardt,1* Danny Langer,1* Petra Ross,1 Alfred Ultsch,2 Michael G. Kiehl,3 Hildegard I.D. Mack,4 Torsten Haferlach,5 Andreas Neubauer1 and Cornelia Brendel1

Department of Hematology, Oncology and Immunology, Philipps-University Marburg, and University Hospital Gießen and Marburg, Marburg, Germany; 2Databionics, Department of Mathematics and Informatics, Philipps-University Marburg, Germany; 3 Department of Internal Medicine, Frankfurt (Oder) General Hospital, Frankfurt/Oder, Germany; 4Institute for Biomedical Aging Research, Leopold-Franzens-University Innsbruck, Austria and 5MLL, Munich Leukemia Laboratory, Germany 1

Haematologica 2019 Volume 104(2):277-287

*These authors contributed equally to this work.

ABSTRACT

D

ifferential induction therapy of all subtypes of acute myeloid leukemia other than acute promyelocytic leukemia is impeded by the long time required to complete complex and diverse cytogenetic and molecular genetic analyses for risk stratification or targeted treatment decisions. Here, we describe a reliable, rapid and sensitive diagnostic approach that combines karyotyping and mutational screening in a single, integrated, next-generation sequencing assay. Numerical karyotyping was performed by low coverage whole genome sequencing followed by copy number variation analysis using a novel algorithm based on in silico-generated reference karyotypes. Translocations and DNA variants were examined by targeted resequencing of fusion transcripts and mutational hotspot regions using commercially available kits and analysis pipelines. For the identification of FLT3 internal tandem duplications and KMT2A partial tandem duplications, we adapted previously described tools. In a validation cohort including 22 primary patients’ samples, 9/9 numerically normal karyotypes were classified correctly and 30/31 (97%) copy number variations reported by classical cytogenetics and fluorescence in situ hybridization analysis were uncovered by our next-generation sequencing karyotyping approach.

Predesigned fusion and mutation panels were validated exemplarily on leukemia cell lines and a subset of patients’ samples and identified all expected genomic alterations. Finally, blinded analysis of eight additional patients’ samples using our comprehensive assay accurately reproduced reference results. Therefore, calculated karyotyping by low coverage whole genome sequencing enables fast and reliable detection of numerical chromosomal changes and, in combination with panel-based fusionand mutation screening, will greatly facilitate implementation of subtype-specific induction therapies in acute myeloid leukemia.

Introduction Acute myeloid leukemia (AML) has recently been the object of thorough investigations at a molecular level, including whole genome next-generation sequencing (NGS) studies.1 According to current guidelines from the European Leukemia Network, recommended genetic testing in adult AML is predominantly directed towards risk stratification in order to identify an appropriate strategy for consolidation therapy.2 Genetic markers comprise t(8;21)(q22;q22.1)/RUNX1-RUNX1T1, inv(16)(p13.1q22) or t(16;16)(p13.1;q22)/CBFB-MYH11, t(15;17)/PML-RARA, t(9;11)(p21.3;q23.3)/MLLT3-KMT2A, other translocations involving the KMT2A haematologica | 2019; 104(2)

Correspondence: brendelc@staff.uni-marburg.de

Received: May 26, 2018.

Accepted: September 5, 2018.

Pre-published: September 6, 2018. doi:10.3324/haematol.2018.194258 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/277 ©2019 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.

277

E.K.M. Mack et al.

gene, t(6;9)(p23;q34.1)/DEK-NUP214, inv(3)(q21.3;q26.2) or t(3;3)(q21.3;q26.2)/GATA2, MECOM, loss of chromosome 5/5q, 7, or 17/17p, mutations in CEPBA (biallelic), NPM1, RUNX1, ASXL1 and TP53, and internal tandem duplications (ITD) in the FLT3 gene.2 Additionally, AML with IDH2R172 mutations alone and AML with mutations in chromatin regulators or splicing factors such as DNMT3A, TET2, SRSF2 and SF3B1 have been proposed recently as distinct genomic subclasses of AML.3 In spite of the considerable genetic heterogeneity of the disease, chemotherapy with cytarabine and anthracyclines has been the backbone of induction treatment for adults with AML throughout the last 30 years.4,5 Only acute promyelocytic leukemia with the hallmark translocation t(15;17)/PML-RARA has been shown to be highly curable by all-trans retinoic acid and arsenic trioxide.6 Immediate chemotherapy-free first-line treatment of acute promyelocytic leukemia is possible because this specific entity can be diagnosed within just a few hours by peripheral blood smear or bone marrow cytology and targeted reverse transcriptase polymerase chain reaction (PCR) analysis for PML-RARA. In contrast, discrimination of all other AML subtypes requires multiple molecular and cytogenetic analyses. In particular, AML karyotyping often requires shipping of samples to specialized laboratories, thus precluding completion within 5-7 days, as recommended.2 Here, we developed and evaluated an integrated NGS platform for numerical karyotyping and focused screening for translocations and mutations in AML-related genes which enables fast identification of the majority of genetic alterations in AML with prognostic and therapeutic implications.

Methods Patients and cell lines The 33 patients’ samples analyzed in this study were obtained from the repository of the Clinic for Hematology, Oncology and Immunology or the Munich Leukemia Laboratory with the patients’ informed consent. Cell lines were purchased from DSMZ. This study was approved by the clinical ethics committee at the Faculty of Medicine, Philipps-University Marburg (N. 38/16).

PTD identifier) was developed. Details are given in the Online Supplementary Methods.

Results Design of an integrated next-generation sequencing platform for comprehensive genetic analyses in acute myeloid leukemia Our NGS platform for comprehensive genetic characterization of AML samples was designed to enable fast and reliable detection of genetic aberrations that are of critical importance for diagnosis, prognosis and therapy in adult AML.1–3,9 The complete workflow including three NGSlibrary preparations and data analysis by five different algorithms can be completed within 5 days (Figure 1A).

In order to limit the required sequencing resources to the capacities of a benchtop sequencing device, we addressed AML-relevant translocations on the level of RNA using anchored multiplex PCR10 for targeted enrichment of chimeric transcripts. RNA-based detection of common gene fusions in AML and DNA-based mutational screening are already available through predesigned commercial kits (Online Supplementary Tables S1-S4) with associated analysis software. Thus, we included numerical karyotyping into our platform by a strategy that does not require specific target enrichment. In particular, we performed low coverage whole genome sequencing (lc-WGS), which has been shown previously to enable robust detection of CNV.11,12 For data analysis, we developed novel algorithms for the detection of CNV (Figure 1B) and KMT2A-PTD (CAI[N] and PTDi, respectively). Moreover, we added ITD-seek7 for the identification of FLT3-ITD to the available bioinformatics pipelines for fusion- or mutation calling. Depending on the size of the mutation panel, one or two samples can be analyzed at a time on a standard (maximum 15 x106 reads) flowcell (Figure 1C). Operational costs per sample are comparable to the total expenses for conventional cytogenetics, fluorescence in situ hybridization (FISH) and mutation analysis. Taken together, our integrated NGS approach rapidly and economically delivers clinically meaningful insights into AML genomes, opening up the possibility to inform treatment decisions early based on molecular features and calculated cytogenetic information.

Library preparation and sequencing Whole genome libraries were constructed using the NEBNext® Ultra II kit (New England Biolabs, Ipswich, MA, USA). Fusion libraries were prepared using the FusionPlex® Heme v1 or v2 panels (ArcherDX, Boulder, CO, USA). Variant libraries were generated using the TruSight® Myeloid panel (Illumina, San Diego, CA, USA) or the QIAseq™ Human Myeloid Neoplasms panel (Qiagen, Hilden, Germany). Sequencing was performed on an Illumina MiSeq instrument. The experimental procedures are detailed in the Online Supplementary Methods.

Bioinformatics Copy number variations (CNV) were analyzed using a proprietary algorithm designated CAI[N] (chromosomal aberration identifier [numerical]) which was implemented in Python. Fusion calling was performed using the Archer® Analysis pipeline (version 4.1). For variant calling, the BaseSpace® TruSight® Myeloid App (Illumina) and ITD-seek7 or, respectively, smCounter8 (Qiagen) were used as appropriate. For identification of KMT2A-partial tandem duplications (PTD) in amplicon libraries, a novel Python algorithm (PTDi: 278

Principles of the CAI[N] algorithm and stability of in silico-generated reference karyotypes In order to facilitate clinical interpretation, we modified the concept of “virtual” or “digital” karyotypes13,14 and constructed “calculated karyotypes” from NGS data which resemble cytogenetic karyotypes. CNV in the range from cytogenetic bands to whole chromosomes are conveniently identified using a read depth approach15 and require only 5-10% genome coverage for detection with >90% sensitivity and specificity,16,17 corresponding to 1-2x106 reads. CAI[N] compares read frequencies in 1 Mb fixed genomic windows to in silico-generated normal reference karyotypes and maps amplified/deleted regions to cytogenetic bands so that chromosomal gains or losses can be reported using cytogenetic notation (Figures 1B and 2A).

Centromeres are not covered by our NGS karyotyping method as they include repetitive sequences that prevent unique alignment of sequencing reads.

To examine the stability of in silico reference karyotypes, haematologica | 2019; 104(2)

NGS for genetic diagnosis of AML

A

B

C

Figure 1. Comprehensive genetic diagnosis of acute myeloid leukemia by next-generation sequencing. (A) Outline of the workflow: each sample is subjected to preparation of three sequencing libraries. Libraries are indexed separately for sequencing on the same flowcell. Data are analyzed using five distinct algorithms for the detection of CNV, fusions, and DNA variants. The whole workflow can be completed within 5 days if performed by one person; times to perform individual steps of the composite assay are indicated on the right. (B) Outline of the CAI[N] algorithm for CNV analysis. Reads are mapped to 1 Mb fixed genomic windows and read distributions are compared to the average of more than 2,500 normal karyotypes (Nfemale=2,819, Nmale=2,605) generated by random sampling of 150-250 bp reads from the reference genome. A region is called amplified or deleted if the observed read number in a window differs significantly (P<0.003) from the average of in silico-generated karyotypes. (C) Flow cell occupancy by three sequencing libraries. Two samples can be analyzed in parallel in one sequencing run on a standard MiSeq v2 flowcell when libraries are sequenced with the read numbers indicated in (A).

we analyzed read distributions on whole chromosomes and in 1 Mb windows for random normal female and male karyotypes. Read frequencies showed very narrow variances and more reads mapped to autosomes in male karyotypes than in female ones, consistent with fewer reads mapping to the Y chromosome compared to a second X (Figure 2B). Of note, the Y chromosome appears smaller than its actual size, also due to repetitive sequences. To further investigate whether lc-WGS data resemble the results of in silico random experiments, we sequenced two libraries from healthy female donors at 14 x106 reads. Read distribution patterns matched the in silico reference at all read depths examined (Figure 2C).

These results confirm that lc-WGS can be accurately simulated computationally, allowing us to use random normal karyotypes as a stable reference for CNV analyses.

Detection of chromosomal gains and losses by copy number variation karyotyping After evaluation of CAI[N] for consistency with normal karyotypes, we determined its capacity to detect numerical aberrations. First, we examined an individual with Down syndrome (T21) and the benign meningioma cell line BENMEN-118 by lc-WGS and CAI[N] analysis. Both trisomy 21 haematologica | 2019; 104(2)

in the T21 proband and loss of chromosome 22 in BENMEN-1 cells were identified correctly (Figure 3).

Next, we investigated deletions or additions of chromosome parts in three AML patientsâ&#x20AC;&#x2122; samples exhibiting loss of the long arm of chromosome 5 (Table 1, Online Supplementary Figures S1-S3). CAI[N] recovered 5q deletions with different breakpoints that closely matched reference laboratory results (Figure 4A, Table 1). Moreover, a gain of chromosome 1p was detected in patient AML-1, consistent with a previously reported partial trisomy 1p (Figure 4B, Online Supplementary Figure S1, Table 1).

Finally, to test the capability of our approach to identify chromosomal gains or losses that are not readily detected by cytogenetic banding, we performed CNV karyotyping on two AML cell lines, HL-60 and NB-4. We observed complex patterns of copy number alterations in both cell lines, including massive overrepresentation of 8q24.21 (containing the MYC locus) with loss of the remaining parts of chromosome 8 (Online Supplementary Figures S4A,B, S5 and S6, Online Supplementary Tables S5 and S6), as described previously.19â&#x20AC;&#x201C;26 Amplification of MYC was validated by quantitative PCR in HL-60 and NB-4 cells and in patient AML-2, in whom a copy number gain of 8q24 had not been observed by cytogenetics. Similarly, quantitative 279

E.K.M. Mack et al.

Table 1. Patientsâ&#x20AC;&#x2122; samples and karyotypes.

Pat. ID

Age

FAB Sample Blasts type (Cyto)

AML-11

68

M2

BM

n.a.

AML-2

19

M2 (after ALL)

BM

AML-3

71

M6

AML-42

53

AML-5 AML-63 ML-7 AML-8 AML-9 AML-10

Blasts (FACS)

Reference karyotype

CNV karyotype

n.a.

45,XX, +der(1)t(1;11)(q21;q13); del(5)(q13q33), del(7)(q22q36), -17,del(17)(p12)[26] FISH: partial trisomy 1 (1p), trisomy 11q

90%

n.a.

BM

60%

26%

M2

PB

34 36 82

M4 M4 M5

BM PB BM

100% CD34 -pos. fraction >90% 47% n.a.

30% 30% 40%

45,Y,-X, del(5)(q15q31), del(9)(p24p22), del(17)(p13p11.2), -20, +mar [10]; 46,XY,idem, del(7)(q22q36)[11] 46-49,XX,+X,+X der(3)t(3;17)(p12;?), del(5)(q13q33), +del(5)(q13q33), der(7)t(3;7)(?;p22), +10, der(11)amp(11q)t(11;20), der(17;22)(q10;q10), -18, der(20)t(20;22)[cp15] n.a

47,XX, add(1)(p36.32p21.3), del(5)(q14.3q34), del(7)(q22.1q36.3), +11, del(15)(q11.2q25.1), del(17)(p13.3p11.2), del(17)(q11.1q21.33) 46,XY, del(5)(q14.3q35.3), add(8)(q24), del(9)(p22.1p21.2), del(17)(p13.3p11.1), del(20)(q11.22q13.33)

57 65 61

M0 M1 M2

BM PB BM

80% 90% 80%

83% 86% 91%

AML-11a 74

M4

BM

65%

76%

AML-11b 74 AML-124 72

M4 M5

PB BM

90% 30%

20% 20%

AML-13 AML-14

43 53

M4eo M5a (t-AML)

PB PB

59% n.a.

44% n.a.

AML-15

70

M2

BM

90%

98%

AML-16

58

M2

BM

50%

60%

AML-17

52

M1/ M4/ M2

BM

n.a

51%

47,XX,+X, del(3)(p21.33p14.3), del(5)(q14.3q34), add(11)(q13.4q25), del(17)(p13.3p13.1), -18, der(20)del(p12.3p12.1) add(p11.23q11.22) del(q11.23q13.31), +22 46,XX, del(7)(q11.22q36.3) 46,XX,t(6;11)(q27;q23)[19]; 46,XX[1] 46,XX n.a 46,XX 46,XX,t(X;16)(p11;q21)[19]; 46,XX 46,XX [1] 46,XX [15] 46,XX 46,XY[20] 46,XY 46,XX, 46,XX der(10)t(10;11)(p12;q13) inv(11)(q13q23), der(11)t(10;11)(p12;q13) [20]; nuc ish 11q23(5'MLLx2,3'MLLx2) (5'MLL sep 3'MLLx1) [60], nuc ish 11q23(5'MLLx2,3'MLLx2) (5'MLL sep 3'MLLx0) [40] 46,XX, 47,XX,+4 t(8;21)(q22;q22) [4]; 47,XX, +4, t(8;21)(q22;q22)[2] n.a 47,XX,+4 n.a 46,XY, del(7)(q21.1q36.1), add(13)(q11q14.2) 46,XY, inv(16)(p13q22)[20] 46,XX n.a.

46,XY (female pat. after allo-HSCT from male donor; chimerism 100%) n.a 46,XY (NPM1_A, FLT3-ITD) 46,XX,t(8;21)(q22;q22) [6], 45,X,-X 45,X,-X,t(8;21)(q22;q22) [12], 46,XX [2] 46,XX [21] 46,XX

Total reads

Mappable reads

4,994,165

3,958,833

1,290,882

1,026,834

4,974,979

3,840,179

1,537,585

1,216,416

2,078,039 2,079,920 2,652,806

1,634,883 1,646,231 2,113,075

1,180,039 1,980,007 1,741,918

933,365 1,554,048 1,379,261

1,722,984

1,363,640

1,690,417 1,798,644

1,339,197 1,408,808

2,170,329 1,627,409

1,699,480 1,273,816

1,622,690

1,274,059

2,053,046

1,629,435

2,394,150

1,886,189

continued on the next page

280

haematologica | 2019; 104(2)

NGS for genetic diagnosis of AML continued from the previous page

AML-185 AML-19

45 68

M5 M3v

PB BM

89% 80%

97% 90%

AML-20

61

n.a

BM

100%

98%

AML-21

18

M2

PB

n.a.

>30%

AML-22

31

M5

BM

n.a.

58%

AML-23

48

M4eo

PB

80%

63%

AML-24

41 RAEB-t

BM

10%

11%

AML-25 AML-26

57 M1 72 M0/M1

PB BM

70% n.a.

60% n.a.

AML-27

66

M4

BM

n.a.

n.a.

AML-28 CML-1

54 74

M5a CML

BM PB

n.a. n.a.

n.a. n.a.

HES-1 ALL-1

73 HES BM 62 Common BM B-ALL

n.a.

80%

n.a.

70%

n.a 46XY, t(15;17)(q22;q12)[2]; 46XY,t(1;6;7) (p36;p21;q22), t(15;17)(q22;q12)[11] n.a (FLT3-ITD, NPM1_D, DNMT3A R882H, IDH1 R132H) 46,XX, del(7q),t(8;21)(q22;q22) 48-49,XY, +19, +der(21)X1-2[cp4]; 46,XY[7] 46,XY [6]; 46,XY, inv. (16)(p13q22) [18] 45,XY, -7 [7], 46, XY, -7 +21 [9], 46, XY [5] 46,XX [20] nuc ish 5p15(hTERTx3), 5q31(CDC25C,EGR1x3), 8cen(D8Z2x4), 17p13(TP53x3), 17q11(NF1x3) 45,XY,-7[9]; 45,XY,-7.ish del(12)(p13p13) (3'ETV6-)[3]; 45,XY,-7.ish del(12)(p13p13) (3'ETV6-,5'ETV6-)[6], 46,XY[4] 46,XY,t(9;11)(p22;q23) 46,XX, t(9;22)(q34;q11)[20] 46, XY [15] 47,XX, t(1;9)(q22;q34),+17 [3], 46,XX,t(1;9)(q22;q34),der(6) t(6;17)(p23;q21)[2]

46,XY 46,XY

2,037,251 2,241,454

1,604,338 1,772,271

46,XY

3,603,346

,2838,440

46,XX,del(7)(q33q36.2)

1,104,847

875,403

48,XY,+19,+21

2,042,269

1,602,693

46,XY

1,940,532

1,523,373

46,XY,-7,+21

1,661,351

1,297,049

46,XX poor DNA quality; evidence for del(5p), del(5)(q31), add(8p), +17 poor DNA quality/ too few reads; evidence for, -7, del(12p13.2)

2,326,960 920,209

1,848,467 643,682

590,092

472,009

47,XY,+19 n.d

867,272 n.a.

687,922 n.a.

46,XY n.d.

2,072,387 n.a.

1,636,629 n.a.

Sample AML-1: chromosome 15 was not covered by FISH analyses performed by the reference laboratory (probes included WCP1/WCP18,WCP7, EGR1,WCP5,ATM, p53, IGH BCL2).

Sample AML-4: cytogenetic analyses were not performed at the time the sample was taken for NGS karyotyping: 12 months earlier (BM): 45,X,der(X;7)(q10;p10)[7]; 46,XX,add(7)(q21)[6], 46,XX [3], 46,XY[4]. nuc ish Xp11.1-q11.1(DXZ1x2),Yq12(DYZ1x0)[50]/ Xp11.1-q11.1(DXZ1x1),Yq12(DYZx1)[50],7cen(D7Z1x2), 7q31(DS486x1)[31/100]; 3 months later (pB): 45,der(X;7)(q10;p10)[2], 46,XX,der(7)t(X;7)(?q27;q11)[3], 46,XX,der(7)t(7;18)(p13;q22)t(X;7)(q27;q11),der(18)t(7;18)(p13;q22)[13],46,XX[3]. nuc ish 7cen(D7Z1x2), 7q31(DS486x1)[99/100]; Read counts (total/mapped) are for the 100% CD34-positive fraction, see Online Supplementary Figure S7. 3Sample AML-6: the sample was taken after third-line treatment, which led to a partial remission. Karyotype before treatment (presumed persisting clones in bold; see. also Online Supplementary Table S7): 46,XX,inv(3)(q21q26)[1]/idem,der(5)t(5;16)(q13;q21), der(12)t(12;15)(p12;q21),-15,-16[11]/46,idem,t(12;22)(p12;q12)[8]. 4Sample AML-12: cytogenetic analyses were not performed at the time the sample was taken for NGS karyotyping: approximately 6 weeks earlier: 46,XY [5], nuc ish 5p15(hTERTx2),5q31(CDC25C,EGR1x2)[100], 7cen(D7Z1x2), 7q31(D7S486x2)[100], 13q14(DLEUx3),17p13(TP53x2)[20/100]. 5Sample AML-18: cytogenetic analyses were not performed at the time the sample was taken for NGS karyotyping (relapse): approximately 15 months earlier at primary diagnosis: normal karyotype 46,XY, NPM1 mut, FLT3-ITD positive. BM: bone marrow; PB: peripheral blood.

1 2

PCR analysis of three loci on chromosome 7q (ARHGEF5, PIK3CG, VKORC1L1) confirmed that this region was amplified in NB-4 cells and not lost in AML-2 (Online Supplementary Figures S2 and S4C, Table 1).

In summary, our results demonstrate that lc-WGS followed by CAI[N] analysis correctly identifies copy number changes with high resolution and allows specific genes to be linked directly to amplified or deleted regions.

Sensitivity of CAI[N] copy number variation karyotyping To test the sensitivity of our karyotyping approach, we performed lc-WGS and CAI[N] analysis on a dilution series haematologica | 2019; 104(2)

of BEN-MEN-1 DNA in healthy donor DNA. Moreover, we investigated samples with different blast contents which were prepared after enrichment of CD34-positive cells from the peripheral blood of patient AML-4 by magnetic bead separation.27 Loss of chromosome 22 was readily detectable in mixtures containing as little as 10% BEN-MEN-1 DNA (Figure 5). Deletion of chromosome 7q was recovered by CNV karyotyping for blast contents â&#x2030;Ľ20% with almost identical breakpoints (Online Supplementary Figure S7, Table 1). As loss of chromosome 7q was not present in all cells in the CD34-positve blast population (Table 1), the detection limit for this aberration was slightly higher than for monosomy 281

E.K.M. Mack et al.

22, which is found in all BEN-MEN-1 cells. These findings indicate that the sensitivity of CAI[N]-CNV karyotyping is sufficient to detect highly prevalent chromosomal aberrations in AML samples with a blast count of at least 20% without prior enrichment of the blast population.

Fusion gene and DNA variant detection Our composite assay relies on predesigned amplicon panels for the detection of fusion transcripts and DNA variants. Amplification strategies implemented in these kits10 or, respectively, specific panels, have already been extensively evaluated.7,28 Thus, we focused our studies on the detection of all subclass-defining translocations and all major types of clinically relevant DNA variants in adult AML. To investigate coverage of important fusion genes, we analyzed cell lines and patients’ samples harboring or lacking frequent chimeric transcripts. All expected fusions were identified in KASUMI-1,29 ME-1,30 NB-4,21 AML-5, AML-6 and CML-1, including two variants31 of BCR-ABL1 in the last sample. On the other hand, no fusions were detected in HL-6019 cells and in a patient with hypere-

osinophilic syndrome (HES-1), as reported by the reference laboratory (Online Supplementary Figure S8A, Online Supplementary Table S7). In a pool of the four cell lines, all fusions were recovered, but in a 1:25 dilution thereof, only the RUNX1-RUNX1T1 fusion transcript was identified.

This finding underlines that RNA-based fusion detection is expression-dependent, so that the sensitivity of the assay varies for different samples and fusions.

Moreover, we exemplarily tested the TruSight® Myeloid panel (Illumina) and the QIASeq™ Myeloid Neoplasms panel (Qiagen), which incorporates molecular barcodes for PCR-error correction,32 as screening tools to identify short DNA variants in AML genomes. All single nucleotide variants detected in HL-60, NB-4, ME-1, MV4-11 and SKNO-1 cells by the TruSight® Myeloid panel were consistent with COSMIC33 data or confirmed by Sanger sequencing, and the QiaSeq™ panel uncovered all reported mutations in samples from two patients with AML (Online Supplementary Table S8). Sequencing a dilution series of MV4-11 DNA revealed detection limits for the two p53 mutations of 1% and 10% with the TruSight® and

A

B

C

282

Figure 2. Calculated chromosome banding and in silicogenerated reference karyotypes. (A) Calculated chromosome banding by CAI[N] analysis of lc-WGS data. Read distributions in genomic windows along chromosome 9 of an AML patient (AML-2, Table 1) are indicated (left: cytogenetic bands, right: 1 Mb windows.

(B) Frequencies of uniquely mapped reads on whole chromosomes for in silico-generated normal karyotypes. RF: random female (N=2,819), RM: random male (N=2,605). Error bars represent the standard deviation (below visibility; <0.01%). Note that in (A) the centromere of a chromosome is not covered and in (B) the Y chromosome appears smaller than its actual size because of repetitive DNA sequences, which prevent unique alignment of sequencing reads. (C) Scalability of the CAI[N] algorithm: Four whole genome libraries from two healthy female donors were sequenced with different read numbers in multiplexed sequencing runs (right panel).

Healthy F1.1-4: four runs of the same library.

haematologica | 2019; 104(2)

NGS for genetic diagnosis of AML

QiaSeq™ panels, respectively (Online Supplementary Figure S8B). Therefore, amplicon sequencing enables variant detection with analytical sensitivities that, in well-covered regions, are equivalent to those of Sanger sequencing, the reference method for clinical mutation testing.34,35 Larger DNA sequence variants such as FLT3-ITD and KMT2A-PTD require special attention in NGS data analysis as they may be missed by common variant calling tools.36,37 Thus, previous authors have added ITD-seek to the TruSight® Myeloid analysis pipeline.7 Applying this tool to our MV4-11 dilution series, we identified a 34 bp insertion in FLT3 exon 14, which closely matched published results,38 with an analytical sensitivity of 10% (Online Supplementary Figure S9A). The FLT3-ITD was not detected in the 10% MV4-11 sample using the QiaSeq™ panel with smCounter analysis, presumably because of suboptimal coverage achieved in our sequencing runs.

However, given that our work with commercial kits aimed at clarifying their principal applicability for diagnostic purposes, we did not repeat these experiments.

For the identification of KMT2A-PTD in amplicon sequencing data, we developed PTDi by adapting a tool that had been used previously with a capture-based targeted sequencing approach.36 PTDi analysis revealed amplification of exons 3-8 in patient AML-7 with a known e3e9 KMT2A-PTD (Online Supplementary Figure S9B, Online

Supplementary Table S9). Thus, not only are short DNA variants detectable by amplicon sequencing, but also difficult ITD and PTD. Taken together, our results clearly confirm that combining RNA- and DNA-based amplicon panels allows all major translocations and all types of clinically relevant mutations in AML to be uncovered by NGS.

Evaluation of the comprehensive next-generation sequencing platform for the diagnosis of acute myeloid leukemia in a clinical setting After testing the performance of all sequencing modules and bioinformatics procedures, we evaluated the clinical utility of our platform as a diagnostic tool. As our karyotype studies in cell lines clearly showed that NGS detects a higher number of numerical aberrations than chromosome banding (Online Supplementary Tables S5 and S6), we first investigated a potential need for manual review of the raw CAI[N] output in order to avoid overestimation of karyotype complexity and enable appropriate risk stratification (Table 1, Online Supplementary Tables S10 and S11).

We performed lc-WGS on additional patients' samples and reconstructed CNV karyotypes by cross comparison of CAI[N] results and known cytogenetic findings. All noncomplex karyotypes, including three samples with presumably normal karyotypes in which cytogenetic analysis had failed, were identified correctly (AML-5, -7-11a/b, -13,

A

B

Figure 3. Detection of whole chromosome gains and losses by copy number variation karyotyping. Whole genome libraries from (A) an individual with Down syndrome (T21) and (B) the BEN-MEN-1 cell line were sequenced with low coverage and analyzed by CAI[N]. RF: random female (N=2,819), RM: random male (N=2,605). Error bars represent the standard deviation (below visibility).

haematologica | 2019; 104(2)

283

E.K.M. Mack et al.

A

B

Figure 4. Detection of partial chromosome losses and gains by copy number variation karyotyping. Whole genome libraries from three AML patients’ samples were sequenced with low coverage and analyzed by CAI[N]. (A) Region plots for chromosome 5. (B) Region plot of chromosome 1 for patient AML-1. Read numbers in 1 Mb windows were normalized to 1 x106 total reads. RF: random female (n=2,819), RM: random male (n=2,605). See also Online Supplementary Figures S1-S3.

-16, -17, -19, HES-1 / AML-14, -15, -20). In two samples that had not been characterized extensively by FISH, CNV karyotyping apparently identified marker chromosomes or detected additional aberrations (AML-2 / AML-1).

Moreover, CAI[N] revealed copy number changes for five of six chromosomes involved in translocations in a highly complex karyotype (AML-3), but missed loss of chromosome 10, which had been detected in <10% of cells by interphase FISH. In four patients, for whom cytogenetics had not been performed when samples were taken, NGS recovered at least a subset of aberrant clones or a normal karyotype as reported at initial diagnosis (AML-4, -6, -12 / AML-18, respectively). Taken together, overall karyotype complexity was determined correctly in all cases and 13/13 samples without risk-defining translocations were accurately assigned to prognostic groups based on CNV karyotyping alone (Online Supplementary Table S10). Thus, we did not specifically validate discordant results between conventional and NGS karyotyping.

Next, to study patients’ samples in an unbiased manner, a group of four of the authors performed blinded analysis of CNV, fusion genes and mutations on eight additional AML samples and one acute lymphocytic leukemia sample (AML-21–28, ALL-1;39 Table 1, Online Supplementary Tables S7-S9). Two samples were excluded before unblinding because of insufficient read coverage resulting from poor DNA quality (AML-26, -27). In the remaining samples, CNV karyotyping uncovered all expected numerical 284

changes and one additional aberration, gain of chromosome 19 in patient AML-28. Fusion analysis identified all translocations previously described in these patients.

Variant analysis revealed no single nucleotide variants, insertions, deletions or ITD in this series of samples, in agreement with reference results. These findings further underscore the potential diagnostic value of our assay for the clinical management of AML.

Discussion Here we developed and validated an integrated diagnostic platform that exclusively utilizes state-of-the-art NGS technology to obtain comprehensive clinically relevant insights into AML genomes. Our targeted approach covers all genetic features that define subclasses of AML with recurrent genetic abnormalities and/or prognostic groups as well as potentially actionable mutations2,9 and thus can serve as a potential alternative for diverse classical cytogenetics and molecular biology assays in certain laboratory settings. In contrast to a previously described single-run NGS assay for AML diagnosis,40 we did not include detection of copy number neutral loss of heterozygosity,41,42 as its prognostic value in AML43 is not fully established.2 We examine numerical aberrations, translocations, short sequence variants, FLT3-ITD and KMT2A-PTD in a rapid, robust and reliable composite assay. Our platform uses a haematologica | 2019; 104(2)

NGS for genetic diagnosis of AML

A

B

Figure 5. Sensitivity of copy number variation karyotyping. Genomic DNA from the BEN-MEN-1 cell line (monosomy 22) was diluted in healthy donor DNA (Healthy F1, Figure 2) in different ratios and subjected to lc-WGS and CAI[N] analysis. (A) Region plots for chromosome 22. The range ±3 standard deviations around the mean is indicated in pale red. (B) CNV decision plots. Read numbers in 1 Mb windows were normalized to 1x106 total reads. RF: random female (N=2,819). Color coding in (B) as in (A).

benchtop sequencing device and commercially available reagents and kits, thus facilitating implementation even at smaller diagnostic centers that currently cannot offer the full spectrum of molecular and cytogenetic analyses required for a complete genetic work-up of AML. NGS results differ from diagnostic reports obtained by current standard techniques in two major aspects: first, NGS karyotyping by lc-WGS and fusion transcript analysis does not reveal insights into the potential clonal heterogeneity of the disease, and second, DNA-based mutation testing yields variant allele frequencies rather than allelic ratios.

While the observed frequency of cytogenetic abnormalities in a certain number of metaphases is not relevant for risk stratification, mutational burden is highly prognostically relevant for FLT3-ITD. Allelic ratios calculated from fragment analysis are not equivalent to variant allele frequencies, but based on our data presented here we cannot propose a conversion from variant allele frequencies to the established risk stratification parameter.

In the method described here, sequencing resources are most effectively reduced by fusion detection on the level of RNA using anchored multiplex PCR for target enrichment.

Anchored multiplex PCR requires previous knowledge on only one fusion partner, which is targeted by a gene-specific primer included in the sequencing panel,10 so that even novel translocations involving commonly rearranged genes can be detected using a relatively low-complexity primer pool. Moreover, if a translocation does not result in a fusion transcript, but rather in transcriptional activation of haematologica | 2019; 104(2)

a target gene, such as MECOM overexpression in AML with inv(3)(q21.3;q26.2) or t(3;3)(q21.3;q26.2),44 the underlying chromosomal rearrangement may be deduced from relative expression analyses via molecular barcode quantification. In contrast, translocation detection on the level of DNA needs to cover large intronic breakpoint regions and therefore requires higher throughput sequencing equipment or limitation of the assay to a subset of AMLrelevant translocations.40,45,46 Most importantly, our approach provides an easy to use tool for “numerical” karyotyping, that – differently from single nucleotide polymorphism-karyotyping methods40 – enables CNV analysis in a completely unbiased manner without the need for specific capture probes. Notably, the CAI[N] algorithm does not uncover absolute ploidy, but classifies gains and losses even in strongly altered hypo- or hyperdiploid cases (e.g. HL-60 and NB-4 cells).

Despite mapping to chromosome bands, CNV karyotyping by lc-WGS and CAI[N] analysis identifies chromosomal gains and losses at higher resolution than conventional cytogenetics (1 Mb versus 5-10 Mb).

Subcytogenetic CNV have been observed previously by array-based comparative genomic hybridization in both complex and normal karyotype AML, including recurrent aberrations with potential prognostic impact, such as gain of 8q24.32,41,47 Moreover, NGS karyotyping does not require short-term culture of the sample material and therefore eliminates a major technical challenge of classical cytogenetics and potential biases in clonally heteroge285

E.K.M. Mack et al.

neous populations. For example, gain of chromosome 19 has been reported to be missed frequently by conventional banding techniques,48 as observed here in patient AML28. Nevertheless, we did not verify this finding by an alternative detection method. Given that NGS technology is increasingly used by diagnostic laboratories, the lcWGS strategy described here opens up the possibility of performing high-resolution numerical karyotyping of AML samples on a routine basis. On the other hand, unlike classical chromosome banding or FISH analyses, CNV karyotyping combined with targeted resequencing of chromosomal fusions will not identify balanced translocations that are not covered by the sequencing panel. These features include some abnormalities defining AML with myelodysplasia-related changes, which is not, however, exclusively a genetic diagnosis.9 Hence, larger cohorts of AML patients will have to be examined in order to develop a new risk stratification system that incorporates the gain of information obtained by lc-WGS (or array-based comparative genomic hybridization) compared to cytogenetics and potential gaps of knowledge that may emerge with a particular CNV karyotyping strategy. Future studies also need to include more extensive validation of the sensitivity of our CNV karyotyping

References 1. Ley TJ, Miller C, Ding L, et al. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):2059–2074.

2. Dohner H, Estey E, Grimwade D, et al.

Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129 (4):424-447.

3. Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic classification and prognosis in acute myeloid leukemia. N Engl J Med. 2016;374(23):2209–2221.

4. Dohner H, Weisdorf DJ, Bloomfield CD. Acute myeloid leukemia. N Engl J Med. 2015;373(12):1136–1152.

5. Grimwade D, Ivey A, Huntly BJP, et al.

Molecular landscape of acute myeloid leukemia in younger adults and its clinical relevance. Blood. 2016;127(1):29–41.

6. Lo-Coco F, Avvisati G, Vignetti M, et al.

Retinoic acid and arsenic trioxide for acute promyelocytic leukemia. N Engl J Med. 2013;369(2):111–121.

7. Au CH, Wa A, Ho DN, Chan TL, Ma ES. Clinical evaluation of panel testing by nextgeneration sequencing (NGS) for gene mutations in myeloid neoplasms. Diagn Pathol.

2016;11:11.

8. Xu C, Nezami Ranjbar MR, Wu Z, DiCarlo J, Wang Y. Detecting very low allele fraction variants using targeted DNA sequencing and a novel molecular barcode-aware variant caller. BMC Genomics. 2017;18(1):5.

9. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016;127 (20):2391–2405.

10. Zheng Z, Liebers M, Zhelyazkova B, et al.

Anchored multiplex PCR for targeted nextgeneration sequencing.

Nat Med. 2014;20(12):1479–1484.

286

assay to detect chromosomal gains and losses in regions that have not been specifically investigated here.

In summary, to the best of our knowledge, our work represents the fastest and most comprehensive analysis platform for diagnosing AML developed so far.

Combining molecular genetics and cytogenetics in one NGS run will pave the way for differentiated management of AML patients not only in clinical trials, but also in a standard-of-care setting, as is currently emerging with midostaurin as the first targeted agent in induction therapy for patients with FLT3 alterations.49 Funding This work was supported by grants from Rhön Klinikum AG (RKA n. 64 to EKMM, CB, MGK), Deutsche Forschungsgemeinschaft KFO210 (BR2232/2 to CB; Ne310/181 to AN), Deutsche José Carreras Leukämie-Stiftung (AH06-01 to AN, 18R/2016 to EKMM) and Stiftung PE Kempkes (06/2014 to EKMM).

Acknowledgments We thank Lisa-Marie Koch, Ute Niebergall and Kathleen Stabla for excellent technical assistance and Prof. Edison T. Liu for helpful advice.

11. Liu S, Song L, Cram DS, et al. Traditional karyotyping vs copy number variation sequencing for detection of chromosomal abnormalities associated with spontaneous miscarriage. Ultrasound Obstet Gynecol.

2015;46(4):472–477.

12. Heitzer E, Ulz P, Belic J, et al. Tumor-associated copy number changes in the circulation of patients with prostate cancer identified through whole-genome sequencing.

Genome Med. 2013;5(4):30.

13. Hagenkord JM, Parwani AV, Lyons-Weiler MA, et al. Virtual karyotyping with SNP microarrays reduces uncertainty in the diagnosis of renal epithelial tumors. Diagn Pathol. 2008;3:44.

14. Wang T-L, Maierhofer C, Speicher MR, et al.

Digital karyotyping. Proc Natl Acad Sci U S A 2002;99(25):16156–16161.

15. Pirooznia M, Goes FS, Zandi PP. Wholegenome CNV analysis: advances in computational approaches.

Front Genet.

2015;6:138.

16. Xie C, Tammi MT. CNV-seq, a new method to detect copy number variation using highthroughput sequencing.

BMC Bioinformatics. 2009;10:80.

17. Li X, Chen S, Xie W, et al. PSCC: sensitive and reliable population-scale copy number variation detection method based on low coverage sequencing.

PLoS One. 2014;9(1):e85096.

18. Puttmann S, Senner V, Braune S, et al.

Establishment of a benign meningioma cell line by hTERT-mediated immortalization.

Lab Invest. 2005;85(9):1163–1171.

19. Gallagher R, Collins S, Trujillo J, et al.

Characterization of the continuous, differentiating myeloid cell line (HL-60) from a patient with acute promyelocytic leukemia.

Blood. 1979;54(3):713-733.

20. Karnan S, Tsuzuki S, Kiyoi H, et al.

Genomewide array-based comparative genomic hybridization analysis of acute promyelocytic leukemia.

Genes

Chromosom Cancer. 2006;45(4):420–425.

21. Lanotte M, Martin-Thouvenin V, Najman S, Balerini P, Valensi F, Berger R. NB4, a maturation inducible cell line with t(15;17) marker isolated from a human acute promyelocytic leukemia (M3). Blood. 1991;77(5):10801086.

22. Shipley J, Weber-Hall S, Birdsall S. Loss of the chromosomal region 5q11-q31 in the myeloid cell line HL-60: characterization by comparative genomic hybridization and fluorescence in situ hybridization. Genes Chromosom Cancer. 1996;15(3):182–186.

23. Mozziconacci MJ, Rosenauer A, Restouin A, et al. Molecular cytogenetics of the acute promyelocytic leukemia-derived cell line NB4 and of four all-trans retinoic acid-resistant subclones. Genes Chromosom Cancer.

2002;35(3):261–270.

24. Liang JC, Ning Y, Wang RY, et al. Spectral karyotypic study of the HL-60 cell line: detection of complex rearrangements involving chromosomes 5, 7, and 16 and delineation of critical region of deletion on 5q31.1.

Cancer Genet Cytogenet.

1999;113(2):105–109.

25. Collins S, Groudine M. Amplification of endogenous myc-related DNA sequences in a human myeloid leukaemia cell line.

Nature. 1982;298(5875):679–681.

26. Von Hoff DD, Forseth B, Clare CN, Hansen KL, VanDevanter D. Double minutes arise from circular extrachromosomal DNA intermediates which integrate into chromosomal sites in human HL-60 leukemia cells. J Clin Invest. 1990;85(6):1887–1895.

27. Hoffmann JC, Stabla K, Burchert A, et al.

Monitoring of acute myeloid leukemia patients after allogeneic stem cell transplantation employing semi-automated CD34+ donor cell chimerism analysis. Ann Hematol 2014;93(2):279–285.

28. Luthra R, Patel KP, Reddy NG, et al. Nextgeneration sequencing-based multigene mutational screening for acute myeloid

haematologica | 2019; 104(2)

NGS for genetic diagnosis of AML

29.

30.

31.

32.

33.

34.

35.

leukemia using MiSeq: applicability for diagnostics and disease monitoring.

Haematologica. 2014;99(3):465–473.

Asou H, Tashiro S, Hamamoto K, Otsuji A, Kita K, Kamada N. Establishment of a human acute myeloid leukemia cell line (Kasumi-1) with 8;21 chromosome translocation. Blood. 1991;77(9):2031–2036.

Yanagisawa K, Horiuchi T, Fujita S. Establishment and characterization of a new human leukemia cell line derived from M4E0. Blood. 1991;78(2):451–457.

Meissner R de V, Covas DT, Dias PMB, Job F, Leite M, Nardi NB. Analysis of mRNA transcripts in chronic myeloid leukemia patients. Genet Mol Biol. 1999;22:475–479.

Rücker FG, Bullinger L, Schwaenen C, et al.

Disclosure of candidate genes in acute myeloid leukemia with complex karyotypes using microarray-based molecular characterization. J Clin Oncol. 2006;24(24):3887– 3894.

Forbes SA, Tang G, Bindal N, et al. COSMIC (the Catalogue of Somatic Mutations in Cancer): a resource to investigate acquired mutations in human cancer. Nucleic Acids Res. 2010;38(Database issue):D652–D657.

Katsanis SH, Katsanis N. Molecular genetic testing and the future of clinical genomics.

Nat Rev Genet. 2013;14(6):415–426.

Strom SP. Current practices and guidelines for clinical next-generation sequencing oncology testing. Cancer Biol Med. 2016;13 (1):3–11.

haematologica | 2019; 104(2)

36. Conte N, Varela I, Grove C, et al. Detailed molecular characterisation of acute myeloid leukaemia with a normal karyotype using targeted DNA capture.

Leukemia.

2013;27(9):1820–1825.

37. Spencer DH, Abel HJ, Lockwood CM, et al.

Detection of FLT3 internal tandem duplication in targeted, short-read-length, next-generation sequencing data. J Mol Diagn.

2013;15(1):81–93.

38. Quentmeier H, Reinhardt J, Zaborski M, Drexler HG. FLT3 mutations in acute myeloid leukemia cell lines. Leukemia.

2003;17(1):120–124.

39. Frech M, Jehn LB, Stabla K, et al. Dasatinib and allogeneic stem cell transplantation enable sustained response in an elderly patient with RCSD1-ABL1-positive acute lymphoblastic leukemia. Haematologica.

2017;102(4):e160-e162.

40. McKerrell T, Moreno T, Ponstingl H, et al.

Development and validation of a comprehensive genomic diagnostic tool for myeloid malignancies. Blood. 2016;128(1):e1-9.

41. Akagi T, Ogawa S, Dugas M, et al. Frequent genomic abnormalities in acute myeloid leukemia/myelodysplastic syndrome with normal karyotype.

Haematologica.

2009;94(2):213–223.

42. Walter MJ, Payton JE, Ries RE, et al.

Acquired copy number alterations in adult acute myeloid leukemia genomes. Proc Natl Acad Sci U S A. 2009;106(31):12950–12955.

43. Gronseth CM, McElhone SE, Storer BE, et al.

44.

45.

46.

47.

48.

49.

Prognostic significance of acquired copyneutral loss of heterozygosity in acute myeloid leukemia. Cancer. 2015;121(17): 2900–2908.

Wang H-Y, Rashidi HH. The new clinicopathologic and molecular findings in myeloid neoplasms with inv(3)(q21q26)/ t(3;3)(q21;q26.2). Arch Pathol Lab Med. 2016;140(12):1404–1410.

Duncavage EJ, Abel HJ, Szankasi P, Kelley TW, Pfeifer JD. Targeted next generation sequencing of clinically significant gene mutations and translocations in leukemia.

Mod Pathol. 2012;25(6):795–804.

He J, Abdel-Wahab O, Nahas MK, et al.

Integrated genomic DNA/RNA profiling of hematologic malignancies in the clinical setting. Blood. 2016;127(24):3004–3014.

Tyybakinoja A, Elonen E, Piippo K, Porkka K, Knuutila S. Oligonucleotide array-CGH reveals cryptic gene copy number alterations in karyotypically normal acute myeloid leukemia. Leukemia. 2007;21(3): 571–574.

Alvarez S, MacGrogan D, Calasanz MJ, Nimer SD, Jhanwar SC. Frequent gain of chromosome 19 in megakaryoblastic leukemias detected by comparative genomic hybridization. Genes Chromosom Cancer.

2001;32(3):285–293.

Stone RM, Mandrekar SJ, Sanford BL, et al.

Midostaurin plus chemotherapy for acute myeloid leukemia with a FLT3 mutation. N Engl J Med. 2017;377(5):454–464.

287

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(2):288-296

Acute Myeloid Leukemia

A novel deep targeted sequencing method for minimal residual disease monitoring in acute myeloid leukemia Esther Onecha,1,2 Maria Linares,1,2 Inmaculada Rapado,1,2,3 Yanira Ruiz-Heredia,1,2 Pilar Martinez-Sanchez,1 Teresa Cedena,1,2,3,4 Marta Pratcorona,5 Jaime Perez Oteyza,6 Pilar Herrera,7 Eva Barragan,4,8 Pau Montesinos,4,8 Jose Antonio Garcia Vela,9 Elena Magro,10 Eduardo Anguita,11 Angela Figuera,12 Rosalia Riaza,13 Pilar Martinez-Barranco,14 Beatriz Sanchez-Vega,1,2 Josep Nomdedeu,5 Miguel Gallardo,2* Joaquin Martinez-Lopez1,2,3,4* and Rosa Ayala1,2,3,4*

1 Hematology Department, Hospital Universitario 12 de Octubre, Madrid; 2Hematological Malignancies Clinical Research Unit, CNIO, Madrid; 3Centro de Investigación Biomédica en Red Cáncer (CIBERONC), Madrid; 4Complutense University, Madrid; 5Hematology Department, Hospital Santa Creu i Sant Pau, Barcelona; 6Hematology Department, Hospital Universitario Sanchinarro, Madrid; 7Hematology Department, Hospital Universitario Ramon y Cajal, Madrid; 8 Hematology Department, Hospital Universitario La Fe, Valencia; 9Department of Hematology, Hospital Universitario de Getafe, Madrid; 10Hematology Department, Hospital Universitario Principe de Asturias, Madrid; 11Hematology Department, Hospital Clínico San Carlos, IdISSC, UCM, Madrid; 12Hematology Department, Hospital Universitario de la Princesa, Madrid; 13 Hematology Department, Hospital Universitario Severo Ochoa, Madrid and 14Hematology Department, Hospital Universitario Fundación Alcorcón, Madrid, Spain

*MG, JM-L and RA contributed equally to this work.

ABSTRACT

Correspondence: rosam.ayala@salud.madrid.org

Received: April 2, 2018.

Accepted: August 8, 2018.

Pre-published: August 9, 2018. doi:10.3324/haematol.2018.194712 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/288 ©2019 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.

288

A

high proportion of patients with acute myeloid leukemia who achieve minimal residual disease negative status ultimately relapse because a fraction of pathological clones remains undetected by standard methods. We designed and validated a high-throughput sequencing method for minimal residual disease assessment of cell clonotypes with mutations of NPM1, IDH1/2 and/or FLT3-single nucleotide variants. For clinical validation, 106 follow-up samples from 63 patients in complete remission were studied by sequencing, evaluating the level of mutations detected at diagnosis. The predictive value of minimal residual disease status by sequencing, multiparameter flow cytometry, or quantitative polymerase chain reaction analysis was determined by survival analysis. The sequencing method achieved a sensitivity of 10-4 for single nucleotide variants and 10-5 for insertions/deletions and could be used in acute myeloid leukemia patients who carry any mutation (86% in our diagnostic data set).

Sequencing–determined minimal residual disease positive status was associated with lower disease-free survival (hazard ratio 3.4, P=0.005) and lower overall survival (hazard ratio 4.2, P<0.001). Multivariate analysis showed that minimal residual disease positive status determined by sequencing was an independent factor associated with risk of death (hazard ratio 4.54, P=0.005) and the only independent factor conferring risk of relapse (hazard ratio 3.76, P=0.012). This sequencing-based method simplifies and standardizes minimal residual disease evaluation, with high applicability in acute myeloid leukemia. It is also an improvement upon flow cytometry- and quantitative polymerase chain reaction-based prediction of outcomes of patients with acute myeloid leukemia and could be incorporated in clinical settings and clinical trials.

Introduction Cytogenetic and molecular alterations at diagnosis and response to treatment are the most useful criteria for predicting the relative risk of relapse in acute myeloid leukemia (AML), and for guiding the choice between chemotherapy and haematologica | 2019; 104(2)

Deep sequencing method for MRD monitoring in AML

hematopoietic stem cell transplantation in first complete remission.1 The definition of complete remission for AML includes criteria for the identification of patients with poor prognosis using cytomorphological methods,2 but these studies do not have a good predictive value because most of the patients in complete remission relapse within 3 years of diagnosis.3 Assessment of minimal residual disease (MRD) is critical in monitoring patients in morphological remission, to inform decisions about further therapy.1 Indeed, several studies have reported MRD status as a stronger predictor of relapse, because patients who are MRD negative have a better prognosis than those who are MRD positive.4,5 In support of this, recent non-randomized studies from prospective multicenter trials suggested better outcomes when leukemia therapy was selected based on the results of MRD assessment.6-8 AML is, nevertheless, a biologically complex and heterogeneous disease, which makes MRD testing more challenging in this condition than in other hematologic neoplasms such as acute lymphoblastic leukemia or multiple myeloma. The detection of very low levels of MRD by conventional methods such as quantitative (q) polymerase chain reaction (PCR) or multiparameter flow cytometry (MFC) provides powerful independent prognostic information. Unfortunately, as described for cytomorphological complete remission, many patients who achieve MRD negative status relapse as a result of the progression of undetected leukemic cells. The most common method for MRD detection is MFC, which has intermediate applicability (70–80%) and limited sensitivity.9,10 However, there is no consensus on multi-antibody panels with regards to inter-laboratory performance, and the technique requires a high level of expertise. The other principal method for monitoring MRD, qPCR, has good sensitivity (10-4-10-6), but its applicability is limited to the approximately 40% of patients who present with molecular alterations (RUNX1RUNX1T1, CBFβ-MYH11 or NPM1) at diagnosis.11 For the above reasons, new methods with higher sensitivity, specificity, applicability and performance are needed for MRD assessment in AML. Against this background, next-generation sequencing (NGS) and digital PCR (dPCR) have recently emerged as potentially promising platforms for the assessment of MRD.12 Here, we optimized and clinically validated a new deep targeted NGS-based method, supported with dPCR technical validation, for the detection and quantification of MRD [both small insertion/deletions (indels) and single nucleotide variants (SNV)] in AML patients, in an attempt to improve and/or complement the current techniques for MRD evaluation, and to establish its potential as a predictor of patients’ outcome.

sis, and availability of at least one follow-up genomic (g)-DNA sample.

The MRD approach was applied to 51 (48%) follow-up samples taken after induction therapy and 55 (52%) taken after consolidation, corresponding to 63 patients diagnosed between 2006 and 2016 (for selection criteria see Online Supplement 6 and Supplementary Table S1). Patients were treated according to PETHEMA (Programa Español de Tratamientos en Hematología) or CETLAM (Grupo cooperativo de Estudio y Tratamiento de Leucemias Agudas y Mielodisplasias) protocols. The study was conducted according to Spanish law 14/2007 on biomedical research, and was approved by the Research Ethics Board of each participating institution. All patients provided informed consent. The main clinical characteristics of the patients are summarized in Table 1.

All patients achieved complete remission according to cytomorphological criteria after induction therapy (<5% of bone marrow blasts).

To construct calibration curves, commercial (Horizon Discovery, UK) reference standard gDNA was used for somatic SNV in IDH1 (R132C) and IDH2 (R172K). As a further source of gDNA, we used the OCI-AML3 cell line (ACC 582, DSMZ, Germany) with the NPM1 type A mutation (c.863_864insCCTG) to examine indels. As OCI-AML3 cells also present a SNV in DNMT3A (R882C), this was included only for technical optimization.

Deep targeted sequencing workflow The sequencing workflow included a first study at diagnosis and a second study at follow-up. Mutational profile screening at diagnosis was done with a customized NGS myeloid panel of 32 genes frequently mutated in myeloid diseases,13 (Online Supplementary Table S2) and NPM1 analysis was carried out with qPCR.14 The specific mutations detected at diagnosis were studied at follow-up. We first tested a variety of experimental steps to define optimal conditions (Online Supplement 1). We established an optimal protocol (Figure 1) that included DNA amplification, library preparation and sequencing as experimental steps (Online Supplement 2).

Libraries were sequenced on the Ion Proton System platform (Life Technologies, Thermo Fisher Scientific Inc.) with an estimated depth ≥1,000,000 of reads, generating .fastq files. These files were analyzed using a customized bioinformatic pipeline; which leads from the .fastq file and a .csv file that contains information about name identifier, run and barcode identifier, chromosomal position and the variant detected in the diagnosis to be evaluated in the follow-up sample. Through Ensembl Perl API,15 the aligned mutated sequence and the aligned wild-type (wt) sequence are presented in FASTA format (sequences of 40 bp). Finally, we obtained a .csv file containing the name identifier, run and barcode identifier, chromosomal position, the variant, the specific target sequence in FASTA format (mutated forward, mutated reverse, wt forward and wt reverse), the counts of each and the ratio (mutated/wt) in absolute values.

Methods More detailed information can be found in the Online Supplementary Data (1–6).

Patients and samples One hundred and ninety patients with de novo or secondary non–M3 AML were included in mutational profile screening at diagnosis. We performed a new selection for retrospective MRD assessment using the following criteria: presence of the NPM1 type A mutation, or SNV in FLT3, IDH1 and/or IDH2 at diagnohaematologica | 2019; 104(2)

Results A high percentage of acute myeloid leukemia patients could benefit from deep sequencing minimal residual disease assessment In total, 211 (80%) SNV and 46 (20%) indels were detected in the 190 patients analyzed at diagnosis using the customized NGS panel. We detected one variant (SNV or indel) in 48 (25%) cases, two or more variants in 116 289

E. Onecha et al.

(61%) cases and no variants in 26 (14%) cases. In addition, we detected the NPM1 type A mutation in 53 (28%) patients by qPCR. Genes (TET2, ASXL1, or DNMT3A) with evidence of an association with clonal hematopoiesis of indeterminate potential (CHIP) were excluded from the analysis.11 Consequently, 82% of patients in our cohort could benefit from this approach.

Based on those genes reported as potential markers for monitoring MRD,16 and also the availability of follow-up samples, we focused on IDH1/2 and FLT3-SNV. We identified at diagnosis IDH1 mutations in 13 patients (7%), IDH2 mutations in 27 patients (14%) and FLT3-SNV mutations (18%) in 34 patients.

Deep sequencing minimal residual disease assessment has a sensitivity of 10-4 for single nucleotide variants and 10-5 for insertions-deletions To establish the limit of quantification (LOQ) of the method, we used 10-fold serial dilutions of mixed mutated and control DNA. To study prototype indels, we used gDNA from OCI-AML3 cells (NPM1 type A) and to study prototype SNV, we used both gDNA from OCI-AML3 cells (DNMT3A) and commercial reference gDNA (IDH1/IDH2). As a control, we used a pool of gDNA from ten individuals without somatic mutations in these chromosomal regions. In all cases, initial allele frequency was 50% and a total of six dilutions were tested to construct a calibration curve, covering a theoretical dynamic range from 10-1 to 10-7.

As shown in Figure 2A,B, MRD NGS testing of NPM1 (indel) could quantify one mutated cell in the order of 10-5 normal ones and in the case of SNV (IDH1, IDH2 and DNMT3A) the LOQ was 10-4, which was reproducible for all SNV tested.

Next-generation sequencing is more sensitive than digital polymerase chain reaction analysis for minimal residual disease assessment We compared the sensitivity of the sequencing method with that of dPCR using the same LOQ dilution protocol.

Clone frequency expressed as target concentration (mutated copies/mL in wt copies/mL) gradually decreased with each dilution, reaching a LOQ of 10-3 for NPM1, IDH1 and IDH2 (Figure 2C,D). While both methods showed similar detection limits and good linearity, the LOQ for the sequencing method was one order of magnitude higher than that for dPCR (IDH1 and IDH2), and two orders of magnitude higher for indels (NMP1).

Minimal residual disease status assessed by sequencing has prognosis value in acute myeloid leukemia The median depth coverage was 401,300 aligned reads (interquartile range 195,100–825,700) for the 88 NPM1 and 18 SNV (9 IDH1, 7 IDH2, and 2 FLT3) follow-up samples evaluated. We detected no mutated sequence in 13 (12%) samples, one to five mutated sequences in 19 (18%) samples, and more than ten in 74 (70%) samples. The ratio of mutated sequences to wt sequences defined MRD levels.

Considering MRD levels from the 106 samples evaluated we established the optimal cutoff to classify MRD status (positive versus negative) by receiver operating characteristic curves (Online Supplementary Figure S1) at each check-point of MRD evaluation [post-induction (n=51), post-consolidation (n=55), or both together (n=106)].

290

Survival analysis revealed that positive MRD status (MRD levels >0.1%) after induction (n=35) was associated with a significantly lower rate of overall survival [33% versus 78%; hazard ratio (HR): 3.5; 95% confidence interval (CI): 1.1–10.7; P=0.019], but a non-significant lower rate of disease-free survival (58% versus 78%; HR: 2.18; 95% CI: 0.63–7.5; P=0.208) (Figure 3A,B). In postconsolidation samples (n=28), MRD positive status

Table 1. Main characteristics of the patients with acute myeloid leukemia included in the minimal residual disease study.

Patients (n = 63) Follow-up sample type Bone marrow Peripheral blood Sex Male Female Age at diagnosis, median

21 (33%) 42 (67%) 54 (IQR, 41.5–66.0)

Blasts at diagnosis, median count

69 (IQR, 51.0–81.0)

58 (92%) 5 (8%)

Leukocytes at diagnosis 15.7 (IQR,12.2–20.24) median count (×109/L) Secondary AML No 59 (94%) Yes 4 (6%) Cytogenetic risk Favorable 25 (40%) Intermediate 36 (57%) Adverse 2 (3%) FLT3-ITD FLT3 negative 49 (78%) FLT3 positive 14 (22%) FLT3-TKD FLT3 negative 60 (95%) FLT3 positive 3 (5%) NPM1 NPM1 negative 6 (10%) NPM1 positive 57 (90%) Hematopoietic stem cell transplantation No 42 (67%) Allogeneic 7 (11%) Autologous 14 (22%) Relapse No 42 (67%) Yes 21 (33%) Death No 40 (63%) Yes 23 (37%) Treatment* 3+7 regimen 50 (80%) Flugaza 8 (13%) Mylotarg 2 (3%) Panobidara 3 (4%) AML: acute myeloid leukemia; ITD: internal tandem duplications; TKD: tyrosine kinase domain; *3+7 regimen of chemotherapy: one or two induction cycles of cytarabine and idarubicin for 7 and 3 days, respectively; and two or three consolidation cycles of high doses of cytarabine, twice a day for 3 alternate days followed by allogeneic or autologous hematopoietic stem cell transplantation. The remainder of the patients were included in other clinical trials (Mylotarg, NTC0104104; Flugaza, NCT02319135; Panobidara, NCT00840346). Clinical data were collected in the following Spanish AML epidemiological registries: NCT01700413, NCT02006004, NCT00464217, NCT02607059, NCT01041040 and NCT01296178.

haematologica | 2019; 104(2)

Deep sequencing method for MRD monitoring in AML

(MRD levels >0.025%) was associated with both significantly shorter overall survival (33% versus 81%; HR: 6.0; 95% CI: 1.3–28.7; P<0.001) and significantly shorter disease-free survival (17% versus 94%; HR: 19.6; 95% CI: 2.5–155.6; P<0.001) (Figure 3C,D). Survival outcomes were also analyzed combining post-induction and postconsolidation (n=63) tests, in order to compare survival analysis with MFC and qPCR data sets. We observed that positive MRD status (MRD levels >0.035%) was associated with a higher risk of relapse (48% versus 81%; HR: 3.4; 95% CI: 1.4–8.5; P=0.005) and death (37% versus 81%; HR: 4.2; 95% CI: 1.6–10.7; P<0.001) (Figure 3E,F). In order to test the power of NPM1 and SNV as independent predictive markers, we performed the analysis separately. Evaluating NPM1 as an MRD marker (n=54), we found that MRD positive status was associated with both significantly shorter overall survival (43% versus 78%; HR: 3.3; 95% CI: 1.2–8.8; P=0.011), and

shorter disease-free survival (57% versus 85%; HR: 2.9; 95% CI: 0.9–7.6; P=0.052). Similar results were found when we evaluated IDH1, IDH2 or FLT3-SNV as MRD markers (n=11). Accordingly, MRD positive status was associated with both significantly shorter overall survival (17% versus 100%; HR: not applicable; P=0.041), and shorter disease-free survival (17% versus 75%; HR: 6.3; 95% CI:0.7–54; P=0.058).

In univariate Cox analysis (Table 2A), the risk of death was significantly higher with increasing age (HR: 1.04; P=0.013), in patients with FLT3-ITD (HR: 3.45; P=0.007), and in those with MRD positive status as determined by NGS (HR: 4.22; P=0.002). The risk of relapse was significantly higher only in those patients with MRD positive status determined by NGS (HR: 3.4; P=0.008). In multivariate analysis (Table 2B), the risk of death was significantly higher with increasing age (HR: 1.05; P=0.004), in patients with mutated FLT3-ITD (HR: 8.87; P=0.001), and

Figure 1. Workflow of the next-geneartion sequencing – minimal residual disease method. DNA amplification, library preparation and sequencing experimental workflow. Genomic DNA (gDNA) is amplified by quantitative (q) polymerase chain reaction (PCR) using specific primers. Libraries are prepared in four steps: end repair, adaptor ligation, size selection, and PCR amplification. The libraries are then sequenced. A customized bioinformatic pipeline analyzes the sequences obtained. The results are expressed as a ratio of mutated sequences (mut) among wild-type (wt) sequences.

haematologica | 2019; 104(2)

291

E. Onecha et al.

in those with MRD positive status determined by NGS (HR: 4.54; P=0.005). The risk of relapse was higher only in patients who were MRD positive as determined by NGS (HR: 3.76; P=0.012).

Minimal residual disease assessment by sequencing predicts overall survival and disease-free survival better than multiparameter flow cytometry or quantitative polymerase chain reaction analysis A positive correlation was found when comparing MRD assessment by NGS versus MFC (r=0.47, P=0.005, n=75), and NGS versus qPCR (r=0.62, P<0.001, n=80) (Online Supplementary Figure S2). There were differences between positive MRD and negative MRD groups of patients tested by MFC, but they were not significant for either overall survival (P=0.193) or disease-free survival (P=0.117) (n=46) (Figure 4A). Similarly, differences were observed between positive MRD and negative MRD groups defined by qPCR of NPM1, although statistical significance was not reached for either overall survival (P=0.212) or disease-free survival (P=0.086) (n=46) (Figure 4B).

A

C

Discussion We have optimized and validated a high sensitivity NGS method for the detection and quantification of NPM1, IDH1, IDH2 and FLT3-SNV mutated sequences at very low allele frequency in follow-up gDNA samples.

NGS has demonstrated prognostic value for pre-treatment status in patients with AML,17 and may also be a useful tool for detecting MRD.18,19 We first studied the mutational profile of patients with AML using a customized NGS panel to ensure a high applicability (82% of patients). This approach is also a useful screening method for detecting all potential MRD markers and choosing those most relevant. The combination of several markers is possible and recommended to overcome limitations of MRD assessment due to sub-clonal heterogeneity of AML and CHIP.11 Accordingly, our method has the capacity to evaluate multiple markers simultaneously and, considering that 61% of patients in our cohort had two or more genetic alterations, this approach is sufficiently robust to monitor MRD even in patients with clonal evolution.

B

D

Figure 2. Calibration curve of minimal residual disease in serial dilutions. (A,B) Ten-fold dilution curves for the assessment of the sensitivity of next-generation sequencing (NGS) in (A) insertions-deletions (InDel), using OCI-AML3 gDNA with 50% NPM1 type A mutation (R2 = 0.98); and (B) single nucleotide variabts (SNV), using OCI-AML3 gDNA with 50% mutated DNMT3A (R2 = 0.98), and gDNA with 50% mutated IDH1 or IDH2 from a commercial standard (R2 = 0.91 and R2 = 0.98, respectively). (C,D) The same 10-fold dilution curves for the assessment of sensitivity of digital polymerase chain reaction (dPCR) in (C) InDel (R2 = 0.98); and (D) SNV (R2 = 0.91 for IDH1 and R2 = 0.98 for IDH2). The vertical red bars indicate the limit of quantification (LOQ) according to the sample. Clone frequency is expressed as target concentration as mutated copies/mL in wild-type copies/mL. Negative controls are included in the calibration curves and had levels below the corresponding LOQ values.

292

haematologica | 2019; 104(2)

Deep sequencing method for MRD monitoring in AML

Reported variants associated with CHIP are frequently located in DNMT3A, TET2 or ASXL1 genes, and are detected in the preleukemic phase and during complete AML remission.20-23 Indeed, any gene could carry both CHIP and non-CHIP variants, and these should be evaluated for each patient. Moreover, studies have shown that genes related to CHIP (IDH1/2) are useful for predicting

A

B

C

D

E

haematologica | 2019; 104(2)

F

prognosis because in these cases the genetic alterations have been acquired in the leukemic clone and not before.24 The sensitivity of this method equates to one mutated cell per 100,000 cells (LOQ 10-5) for NPM1 and one mutated cell per 10,000 cells (LOQ 10-4) for IDH1, IDH2 and FLT3-SNV. This difference in sensitivity is related to the fact that the NPM1 type A mutation (insCCTG) is rarely

Figure 3. Analysis of overall survival and disease-free survival in patients with acute myeloid leukemia stratified according to minimal residual disease levels determined by sequencing.

Analysis of overall survival for (A) the induction data set, (C) the consolidation data set, and (C) both together. Analysis of disease-free survival for (B) the induction data set, (D) the consolidation data set, and (F) both together. The cutoff used for overall and disease-free survival was 0.001 at the postâ&#x20AC;&#x201C;induction check-point (n=35), 0.00026 at the post-consolidation checkâ&#x20AC;&#x201C;point (n=28) and 0.00035 for both check-points (all data set) (n=63). The numbers of censored patients with respect to the stratified groups and the numbers at risk are indicated. Statistically significant values: *P<0.05, **P<0.01.

293

E. Onecha et al.

Table 2. Cox regression analyses.

A.

HR (95%CI) Sex (female vs. male) Age per year Blasts at diagnosis (%) Leukocytes at diagnosis (×109/L) Favorable vs. adverse ELN risk Intermediate vs. adverse ELN risk Mutated FLT3-ITD Allo-HSCT vs. intensive chemotherapy Allo-HSCT vs. auto-HSCT MRD+ by MFC MRD+ by qPCR MRD+ by NGS

Risk of death P value

1.20 (0.50–2.83) 1.04 (1.00–1.07) 1.00 (0.99–1.02) 1.01 (0.99–1.01) 0.67 (0.08–5.43) 1.03 (0.13–7.86) 3.45 (1.40-8.52) 1.35 (0.40–4.57) 0.29 (0.05–1.74) 2.10 (0.67–6.62) 2.51 (0.56–11.2) 4.22 (1.66–10.7)

0.682 0.013 * 0.667 0.418 0.714 0.976 0.007 * 0.634 0.176 0.203 0.228 0.002 **

HR (95%CI) 0.94 (0.37–2.44) 1.03 (0.99–1.06) 1.01 (0.99–1.03) 1.00 (0.99–1.01) 0.75 (0.09–6.00) 1.02 (0.13–7.82) 2.37 (0.86–6.51) 1.78 (0.41–7.78) 0.64 (0.11–3.77) 2.40 (0.77–7.46) 5.01 (0.64–38.8) 3.41 (1.37–8.48)

Risk of relapse P value 0.906 0.069 0.532 0.508 0.786 0.988 0.095 0.44 0.629 0.130 0.123 0.008 **

B.

HR (95%CI) Age per year Sex (female vs. male) Leukocytes at diagnosis (×109/L) Favorable vs. adverse ELN risk Intermediate vs. adverse ELN risk Mutated FLT3-ITD MRD+ by NGS

Risk of death P value

1.05 (1.02–1.09) 0.84 (0.33–2.17) 1.01 (0.99–1.03) 13.75 (0.84–226.1) 11.22(0.82–154.2) 8.87 (2.54–30.95) 4.54 (1.58–13.03)

0.004 * 0.720 0.219 0.067 0.071 0.001 ** 0.005 **

HR (95%CI) 1.03 (0.99–1.07) 1.25 (0.44–3.52) 1.07 (0.99–1.02) 7.09 (0.37–134.15) 5.86 (0.39–86.84) 4.18 (1.11–15.69) 3.76 (1.34–10.54)

Risk of relapse P value 0.061 0.671 0.481 0.192 0.203 0.034* 0.012*

Cox regression analyses of prognostic factors influencing the risk of relapse and risk of death of patients with acute myeloid leukemia. (A) Univariate Cox regression analysis of each prognostic factor. (B) Multivariate Cox regression analysis evaluating the most relevant factors detected in the univariate analyses. HR: hazard ratio; 95% CI: 95% confidence interval; ITD: internal tandem duplication; ELN: European LeukemiaNet; allo-HSCT: allogeneic hematopoietic stem cell transplantation; auto-HSCT: autologous hematopoietic stem cell transplantation; MFC, multiparametric flow cytometry; MRD, minimal residual disease; NGS, next-generation sequencing. Statistically significant values: *P<0.05, **P<0.01.

generated erroneously by NGS, and the quantification is precise.

Our method, as with any NGS method, has an intrinsic error rate that limits its sensitivity for most SNV to 1–2% of all reads. This limitation can nevertheless be overcome by virtue of the scalable nature of NGS.16 Thus, we boosted NGS sensitivity by increasing the amount of DNA by PCR prior to sequencing, which increased the depth of coverage to one million reads. By also optimizing the bioinformatic analysis, we focused the search for the precise variant in order to eliminate random sequencing errors, enhancing the specificity of the technique and reducing the computational time. To the best of knowledge, our NGS method presents possibly the highest sensitivity reported for NGS in AML.18,19,24-27 dPCR is a relatively novel technique for precise and absolute quantification of nucleic acids, which is based on limiting partitions of the PCR volume and Poisson statistics.28 It is also an extremely sensitive technique, with a high specificity due to the detection of mutant alleles.29 However, when we compared the same standard dilutions in NGS and dPCR, NGS afforded a 2-log increment in LOQ for indels (NPM1) and a 1-log increment for SNV (IDH1/2), with the sensitivity of dPCR for indels being similar to that reported in a previously published study 294

(10-2).30 Compared with NGS, dPCR is a faster measurement technique but, as it is focused, it requires allele-specific primers that can complicate the experimental procedure, and a high number of parallel experiments are needed to raise the sensitivity, which increases the cost of the assay. Additionally, although it is possible to multiplex dPCR, unfortunately only a few targets can be monitored simultaneously within each sample.29 Another advantage of NGS technology is that it does not require calibration curves in each assay, and the results are reported in absolute values, facilitating its standardization.

The NGS method described here showed comparable sensitivities (10-4 for SNV and 10-5 for indels) to those of MFC methods in those cases with immunophenotyphically aberrant populations.10,31 Although our method had a similar sensitivity to that of qPCR, it does not require oligonucleotides that hybridize specifically to a particular sequence, so all nucleotides in the amplified region can be studied. Consequently, the NGS test is capable of detecting all NPM1 subtype mutations in the same assay.

We found positive correlations when MRD levels were evaluated by NGS versus MFC and versus qPCR, but not with the expected results. In the case of MFC, this could be explained, in part, by the fact that NPM1 mutations are haematologica | 2019; 104(2)

Deep sequencing method for MRD monitoring in AML

A

B

C

D

Figure 4. Analysis of overall survival and disease-free survival in patients with acute myeloid leukemia stratified according to minimal residual disease levels determined by conventional methods. Kaplan-Meier plots of (A) overall survival and (B) disease-free survival according to minimal residual disease (MRD) assessment by multiparametric flow cytometry (MRC) and (C) overall survival and (D) disease-free survival according to MRD assessment by quantitative polymerase chain reaction (qPCR) analysis. The numbers of censored patients with respect to each stratified group and numbers at risk are indicated. Statistically significant values: *P<0.05, **P<0.01.

usually associated with monocytic subtype-AML, which frequently presents more difficulties for identifying MRD

by MFC. Indeed, Salipante et al.27 described that the level of success of MFC depends greatly on the immunophenotype of the abnormal blasts and how to discriminate them from background regenerative blasts. Moreover, due to the lack of standardization, MFC shows substantial variability across laboratories, including that of sample processing, instrument configuration, number of events, and training of pathologists.32 The lack of a strong correlation between NGS and qPCR could be explained by the nature of the sample (sequencing uses gDNA whereas qPCR uses cDNA). Although RNA overexpression allows a higher sensitivity of detection, RNA levels do not correlate with the number of tumor cells, in contrast to mutated DNA. haematologica | 2019; 104(2)

Accordingly, mutated DNA is more representative of the tumor burden than is overexpression of mutated RNA.35 It should be noted that the prediction of survival and progression of AML using MRD NGS was better than that of the other methodologies employed, at least in the cohorts evaluated.

Finally, survival analysis showed that MRD positive status determined by NGS was associated with a higher risk of relapse and death and that MRD negative status in postconsolidation ssmples was associated with longer overall and disease-free survival, in accordance with recently published studies.23 Supporting these findings, previous studies reported that an MRD check-point after consolidation could be the best moment for analysis because it afforded better prediction.8,34-37 Cox regression multivariate analyses 295

E. Onecha et al.

confirmed that MRD positive status determined by sequencing was the only statistically significant predictor of the risk of relapse (P=0.012).

In conclusion, we have optimized a new targeted sequencing method with high sensitivity for MRD evaluation with applicability in a high percentage of AML patients, improving the capacity, over MFC or qPCR, to predict the survival outcomes of the AML patients in our cohort.

References 1. Dohner H, Estey E, Grimwade D, et al.

Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel.

Blood.

2017;129(4):424-447.

2. Cheson BD, Bennett JM, Kopecky KJ, et al.

Revised 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. Dohner H, Weisdorf DJ, Bloomfield CD. Acute myeloid leukemia. N Engl J Med. 2015;373(12):1136-1152.

4. Ivey A, Hills RK, Simpson MA, et al.

Assessment of minimal residual disease in standard-risk AML. N Engl J Med. 2016;374(5):422-433.

5. Kronke J, Schlenk RF, Jensen KO, et al.

Monitoring of minimal residual disease in NPM1-mutated acute myeloid leukemia: a study from the German-Austrian Acute Myeloid Leukemia Study Group. J Clin Oncol. 2011;29(19):2709-2716.

6. Rubnitz JE, Inaba H, Dahl G, et al. Minimal residual disease-directed therapy for childhood acute myeloid leukaemia: results of the AML02 multicentre trial. Lancet Oncol.

2010;11(6): 543-552.

7. Balsat M, Renneville A, Thomas X, et al.

Postinduction minimal residual disease predicts outcome and benefit from allogeneic stem cell transplantation in acute myeloid leukemia with NPM1 mutation: a study by the Acute Leukemia French Association Group. J Clin Oncol. 2017;35(2):185-193.

8. Zhu HH, Zhang XH, Qin YZ, et al. MRDdirected risk stratification treatment may improve outcomes of t(8;21) AML in the first complete remission: results from the AML05 multicenter trial.

Blood.

2013;121(20):4056-4062.

9. Freeman SD, Virgo P, Couzens S, et al.

Prognostic relevance of treatment response measured by flow cytometric residual disease detection in older patients with acute myeloid leukemia. J Clin Oncol.

2013;31(32):4123-4131.

10. Terwijn M, van Putten WL, Kelder A, et al.

High prognostic impact of flow cytometric minimal residual disease detection in acute myeloid leukemia: data from the HOVON/SAKK AML 42A study. J Clin Oncol. 2013;31(31):3889-3897.

11. Schuurhuis GJ, Heuser M, Freeman S, et al.

Minimal/measurable residual disease in AML: a consensus document from the European LeukemiaNet MRD Working Party. Blood. 2018;131(12):1275-1291.

12. Grimwade D, Freeman SD. Defining minimal

296

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

Acknowledgments This study was supported by the Subdirecci贸n General de Investigaci贸n Sanitaria (Instituto de Salud Carlos III, Spain) grants PI13/02387 and PI16/01530, and the CRIS against Cancer foundation grant 2014/0120. ML holds a postdoctoral fellowship of the Spanish Ministry of Economy and Competitiveness (FPDI-201316409). PRP holds a postdoctoral fellowship of the Spanish Instituto de Salud Carlos III: Contrato Predoctoral de Formaci贸n en Investigaci贸n en Salud i-PFIS (IFI 14/00008).

residual disease in acute myeloid leukemia: which platforms are ready for "prime time"?

Hematology Am Soc Hematol Educ Program.

2014;2014(1):222-233.

Cedena MT, Rapado I, Santos-Lozano A, et al. Mutations in the DNA methylation pathway and number of driver mutations predict response to azacitidine in myelodysplastic syndromes. Oncotarget. 2017;8(63):106948106961.

Gorello P, Cazzaniga G, Alberti F, et al.

Quantitative assessment of minimal residual disease in acute myeloid leukemia carrying nucleophosmin (NPM1) gene mutations.

Leukemia. 2006;20(6):1103-1108.

McLaren W, Gil L, Hunt SE, et al. The ensembl variant effect predictor. Genome Biol. 2016;17(1):122.

Tomlinson B, Lazarus HM. Enhancing acute myeloid leukemia therapy - monitoring response using residual disease testing as a guide to therapeutic decision-making.

Expert Rev Hematol. 2017;10(6):563-574.

Patel JP, Gonen M, Figueroa ME, et al.

Prognostic relevance of integrated genetic profiling in acute myeloid leukemia. N Engl J Med. 2012;366(12):1079-1089.

Thol F, Kolking B, Damm F, et al. Next-generation sequencing for minimal residual disease monitoring in acute myeloid leukemia patients with FLT3-ITD or NPM1 mutations. Genes Chromosomes Cancer.

2012;51(7):689-695.

Kohlmann A, Nadarajah N, Alpermann T, et al. Monitoring of residual disease by nextgeneration deep-sequencing of RUNX1 mutations can identify acute myeloid leukemia patients with resistant disease.

Leukemia. 2014;28(1):129-137.

Metzeler KH, Herold T, RothenbergThurley M, et al. Spectrum and prognostic relevance of driver gene mutations in acute myeloid leukemia. Blood. 2016; 128(5):686698.

Bullinger L, Dohner K, Dohner H. Genomics of acute myeloid leukemia diagnosis and pathways. J Clin Oncol. 2017;35(9):934-946.

Koeffler HP, Leong G. Preleukemia: one name, many meanings. Leukemia. 2017; 31(3):534-542.

Jongen-Lavrencic M, Grob T, Hanekamp D, et al. Molecular minimal residual disease in acute myeloid leukemia. N Engl J Med. 2018;378(13):1189-1199.

Klco JM, Miller CA, Griffith M, et al.

Association between mutation clearance after induction therapy and outcomes in acute myeloid leukemia. JAMA. 2015;314 (8):811-822.

Getta BM, Devlin SM, Levine RL, et al.

Multicolor flow cytometry and multigene next-generation sequencing are complementary and highly predictive for relapse in acute myeloid leukemia after allogeneic transplantation. Biol Blood Marrow

Transplant. 2017;23(7):1064-1071.

26. Debarri H, Lebon D, Roumier C, et al.

IDH1/2 but not DNMT3A mutations are suitable targets for minimal residual disease monitoring in acute myeloid leukemia patients: a study by the Acute Leukemia French Association. Oncotarget. 2015;6(39): 42345-42353.

27. Salipante SJ, Fromm JR, Shendure J, et al.

Detection of minimal residual disease in NPM1-mutated acute myeloid leukemia by next-generation sequencing. Mod Pathol.

2014;27(11):1438-1446.

28. Brunetti C, Anelli L, Zagaria A, et al. Droplet digital PCR is a reliable tool for monitoring minimal residual disease in acute promyelocytic lseukemia. J Mol Diagn. 2017;19(3): 437-444.

29. Roloff GW, Lai C, Hourigan CS, et al.

Technical advances in the measurement of residual disease in acute myeloid leukemia. J Clin Med. 2017;6(9).

30. Minervini A, Francesco Minervini C, Anelli L, et al. Droplet digital PCR analysis of NOTCH1 gene mutations in chronic lymphocytic leukemia. Oncotarget. 2016;7(52): 86469-86479.

31. Ossenkoppele G, Schuurhuis GJ. MRD in AML: does it already guide therapy decision-making? Hematology Am Soc Hematol Educ Program. 2016;2016(1):356-365.

32. Cruz NM, Mencia-Trinchant N, Hassane DC, et al. Minimal residual disease in acute myelogenous leukemia. Int J Lab Hematol.

2017;39 (Suppl 1):53-60.

33. Duployez N, Nibourel O, Marceau-Renaut A, et al. Minimal residual disease monitoring in t(8;21) acute myeloid leukemia based on RUNX1-RUNX1T1 fusion quantification on genomic DNA. Am J Hematol. 2014;89(6): 610-615.

34. 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.

35. Hourigan CS, Gale RP, Gormley NJ, et al.

Measurable residual disease testing in acute myeloid leukaemia. Leukemia. 2017;31(7): 1482-1490.

36. 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.

37. Jourdan E, Boissel N, Chevret S, et al.

Prospective evaluation of gene mutations and minimal residual disease in patients with core binding factor acute myeloid leukemia. Blood. 2013;121(12): 2213-2223.

haematologica | 2019; 104(2)

ARTICLE

Acute Myeloid Leukemia

Targeted RNA-sequencing for the quantification of measurable residual disease in acute myeloid leukemia

Ferrata Storti Foundation

Laura W. Dillon,1 Sheida Hayati,1,2 Gregory W. Roloff,1 Ilker Tunc,3 Mehdi Pirooznia,3 Antonina Mitrofanova2 and Christopher S. Hourigan1

Laboratory of Myeloid Malignancies, Hematology Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD; 2Department of Health Informatics, Rutgers School of Health Professions, Rutgers, The State University of New Jersey, Newark, NJ and 3Bioinformatics and Computational Biology Core Facility, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD, USA

1

Haematologica 2019 Volume 104(2):297-304

ABSTRACT

G

reat effort is spent on developing therapies to improve the dire outcomes of those diagnosed with acute myeloid leukemia. The methods for quantifying response to therapeutic intervention have however lacked sensitivity. Patients achieving a complete remission as defined by conventional cytomorphological methods therefore remain at risk of subsequent relapse due to disease persistence.

Improved risk stratification is possible based on tests designed to detect this residual leukemic burden (measurable residual disease). However, acute myeloid leukemia is a genetically diverse set of diseases, which has made it difficult to develop a single, highly reproducible, and sensitive assay for measurable residual disease. Here we present the development of a digital targeted RNA-sequencing-based approach designed to overcome these limitations by detecting all newly approved European LeukemiaNet molecular targets for measurable residual disease in acute myeloid leukemia in a single standardized assay. Iterative modifications and novel bioinformatics approaches resulted in a greater than 100-fold increase in performance compared with commercially available targeted RNA-sequencing approaches and a limit of detection as low as one leukemic cell in 100,000 cells measured, which is comparable to quantitative polymerase chain reaction analysis, the current gold standard for the detection of measurable residual disease. This assay, which can be customized and expanded, is the first demonstrated use of high-sensitivity RNA-sequencing for measurable residual disease detection in acute myeloid leukemia and could serve as a broadly applicable standardized tool.

Introduction Despite the achievement of a “complete remission” following therapy, patients with acute myeloid leukemia (AML) remain at risk of relapse because of the persistence of disease that is not detected by conventional cytomorphological methods.1 Risk stratification of such patients is possible based on tests designed to detect this residual leukemic burden (termed measurable residual disease; MRD).16 The importance of MRD testing in AML for prognostic risk stratification has become increasingly evident, such that the AML response criteria were substantively updated in 2017 by the introduction of the category of complete remission without MRD.4 A variety of technologies, such as real-time quantitative polymerase chain reaction (qPCR) analysis and flow cytometry, focusing on the identification of recurrent molecular abnormalities or leukemia-associated immune phenotypes, have been developed for MRD detection in AML.7 The mutational and clonal heterogeneity of AML provides a wide variety of targets for molecular MRD tracking, haematologica | 2019; 104(2)

Correspondence: hourigan@nih.gov

Received: July 26 2018.

Accepted: August 28, 2018.

Pre-published: August 31, 2018. doi:10.3324/haematol.2018.203133 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/297 ©2019 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.

297

L.W. Dillon et al. et al.

however not all targets can reliably serve as a surrogate of disease burden. For example, mutations found in preleukemic founder clones can persist at significant levels even during complete remission8-11 and have also been observed in healthy individuals without hematologic malignancies with increasing age.12-15 To help harmonize AML MRD detection efforts, the European LeukemiaNet (ELN) recently released consensus guidelines for MRD detection in AML, including recommended markers.6 Despite advances in the field, it has remained difficult to develop a single MRD detection assay which is highly reproducible and sensitive, has limited operator dependence, and is capable of quantifying multiple targets simultaneously. To overcome these limitations, we developed a multi-gene, targeted RNA-sequencing-based method for the sensitive detection and quantification of MRD in AML, which we present here. This novel assay has a minimal sequencing budget requirement and outperforms a commercially available, conventional myeloid-targeted RNA-sequencing assay. It has a demonstrated limit of detection as low as one leukemic cell in 100,000 cells measured, so its performance equals that of current goldstandard, single-target qPCR MRD assays.

Gene Q 5plex HRM (Qiagen). A standard curve for each gene was obtained from the plasmid serial dilutions and used to calculate the copy number for each sample. A patient’s sample was considered positive for NPM1 mutA or RUNX1-RUNX1T1 if the copy number detected was contained within the standard curve and the water control was negative.

Digital droplet polymerase chain reaction CBFB-MYH11 type A expression was determined with digital droplet PCR (ddPCR) by converting the established European Against Cancer (EAC) assay for CBFB-MYH11 type A to the Raindance platform (RainDance Technologies, Billerica, MA, USA), as described in the Online Supplementary Appendix.

ArcherDx Myeloid FusionPlex library preparation and analysis Anchored multiplex PCR-based enrichment RNA-sequencing libraries were generated from 250 ng of RNA using the ArcherDx Myeloid FusionPlex assay for Illumina (ArcherDx, Boulder, CO, USA) and analyzed using Archer Analysis software version 5.1.3, as described in the Online Supplementary Appendix.

Statistical analysis Data were analyzed using Prism statistical software (v.7.0b, GraphPad software, La Jolla, CA, USA).

Methods Cell lines and clinical samples Leukemia cell lines positive for fusion genes targeted by the AML MRD panel were cultured according to the supplier’s guidelines. Peripheral blood samples were collected from a healthy adult donor and from a 46-year old female with monocytic AML who underwent myeloablative matched related donor allogeneic stem cell transplantation (NHLBI protocol # 07H-0113). Additional samples from patients were collected in local institutional review board-approved biobanking protocols by collaborators. The samples were processed and RNA isolated as described in the Online Supplementary Appendix.

Preparation and analysis of targeted RNA-sequencing libraries for acute myeloid leukemia measurable residual disease detection Targeted RNA-sequencing libraries were prepared and sequenced as described in the Online Supplementary Appendix. In short, unique molecular identifier (UMI) assignment and complementary DNA (cDNA) generation were performed on 250 ng of RNA using the SuperScript IV First Strand Synthesis System (Invitrogen, Carlsbad, CA, USA) and a pool of 100 nM of each barcoded (BC) primer. The barcoded cDNA was subjected to eight cycles of amplification with a pool of 100 nM of each limited amplification (LA) primer and 600 nM RS2 primer, followed by final library amplification and sample indexing. Single-end 150 bp sequencing was performed on a Miseq or Hiseq 2500 platform (Illumina) using the QIAseq Read 1 Primer 1 custom primer (Qiagen, Germany). Raw sequencing FASTQ files were processed, aligned, and panel targets called as outlined in the Online Supplementary Appendix.

Real-time quantitative polymerase chain reaction Expression of the “type A” mutation of the nucleophosmin gene (NPM1 mutA) and the RUNX1-RUNX1T1 fusion gene was determined for cell dilutions and patient samples by qPCR using the ipsogen NPM1 mutA MutaQuant kit (Qiagen, cat# 677513) or ipsogen RUNX1-RUNX1T1 kit (Qiagen, cat# 675013), respectively, according to the manufacturer’s instructions, and the Rotor298

Results Targeted RNA-sequencing panel target selection for acute myeloid leukemia measurable residual disease detection Targets for this multi-gene RNA-sequencing panel were chosen from those recommended for AML MRD detection by the ELN6 and for which standardized qPCR assays for MRD detection have already been well established5,16-19 (Figure 1A). These targets include: (i) recurrent chimeric fusion transcripts PML-RARA, CBFB-MYH11, RUNX1RUNX1T1, and BCR-ABL1, present in 20.8% of AML patients in The Cancer Genome Atlas (TCGA) dataset;20 (ii) the recurrent insertion site in exon 12 of NPM1, which is found in an additional 27.2% of patients; and lastly (iii) aberrant expression of WT1 and PRAME transcripts, which may be used for MRD detection for up to another 20.2% of patients not covered by the preferred fusion or mutated NPM1 targets. Based on the TCGA AML cohort, WT1 and PRAME could also serve as a secondary tracking target, for orthogonal validation, in over 80% of those patients who also express fusion or NPM1 mutant transcripts. Similar to the established qPCR assays, wild-type ABL1 transcript expression was included as a normalizing control.16

Development of the targeted RNA-sequencing method for acute myeloid leukemia measurable residual disease detection In developing a targeted RNA-sequencing strategy that would be optimal for MRD detection, we took into account several important factors including: (i) the need to optimize target capture efficiency while minimizing the amount of RNA required, as this resource is often limited when dealing with patient specimens; (ii) the incorporation of strategies for the digital quantification of transcripts; (iii) the need to minimize sequencing burden; and (iv) the generation of a workflow that is simple, efficient, and easy to adopt. haematologica | 2019; 104(2)

Targeted RNA-sequencing for MRD in AML

A

B

Figure 1. The targeted RNA-sequencing assay for measurable residual disease detection can be applied to over two-thirds of patients with acute myeloid leukemia.

(A) The acute myeloid leukemia (AML) measurable residual disease (MRD) RNA-sequencing assay comprises eight targets, including insertion mutations, fusion transcripts, and wild-type transcript expression. The frequency of each target was determined using AML cases in The Cancer Genome Atlas (TCGA) for whom clinical information, mutation analysis, and RNA-sequencing data were available (n=173). The expression of insertions and fusion transcripts was evaluated first. For the remaining patients, WT1 and PRAME wild-type transcript overexpression was evaluated and defined as overexpressed if greater than the mean of the entire cohort. (B) The AML MRD RNA-sequencing assay begins with targeted reverse transcription utilizing a pool of primers consisting of a gene-specific region (GSP1), random 12-nucleotide unique molecular identifier (UMI), and conserved sequence region (RS2) per target. The resulting barcoded complementary DNA (cDNA) is subjected to limited amplification using a reverse primer complementary to the RS2 sequence and a pool of forward primers consisting of a gene-specific region (GSP2) and conserved sequence region (FS2) per target. The targeted amplicons are then subjected to amplification, library construction, and sample indexing for Illumina sequencing.

The final assay design addresses these factors by utilizing a pool of target-specific primers containing 12 nucleotide UMIs (Online Supplementary Table S1) which capture and individually tag RNA molecules of interest during reverse transcription, followed by targeted PCR of the barcoded cDNA, and library construction (Figure 1B).

Unlike most targeted RNA-sequencing approaches, this simplified design reduces protocol steps while maximizing utilization of the RNA input by performing target enrichment during the reverse transcription step, as opposed to after cDNA generation. The concurrent addition of molecular barcodes during this first step also allows for a digital output, increasing the accuracy of transcript quantification. Additionally, the amplicon-based enrichment design for fusion detection limits the sequencing requirements, since only fusion transcripts and not wild-type transcripts, will be amplified. Finally, with a total of only three steps, the hands-on time is minimized, allowing for the entire protocol to be completed in less than a day.

Sequencing files were processed by extracting the UMI from each read, alignment to the human genome, and clustering of sequences which correspond to the intended panel targets (Online Supplementary Figure S1). Targets were quantified by the number of unique UMIs, with a library-specific UMI cutoff value established to eliminate background due to sequencing errors (Online Supplementary Figure S2).

The expression of all assay targets exhibited significant correlation (linear regression r2 â&#x2030;Ľ0.97) and leukemic cells could be detected at a level as low as one leukemic cell in 100,000 healthy donor cells (Figure 2). For the fusion and mutated NPM1 transcripts, detection sensitivity was between 1:10,000 to 1:100,000. Many of the cell lines expressing fusion transcripts also displayed aberrant WT1 and/or PRAME transcript expression, which was also highly correlated and for which the assay showed varying degrees of sensitivity, ranging from 1:1,000 to 1:100,000.

Detection of all targets was highly reproducible across replicates.

Validation of assay performance and limit of detection

Determination of assay sequencing requirements

To assess the sensitivity and dynamic range of the AML MRD RNA-sequencing panel, cell lines expressing fusion transcripts or patient cells positive for the NPM1 mutA insertion mutation (94% blasts) were serially diluted (1:10 to 1:100,000) into healthy adult donor peripheral blood mononuclear cells and RNA was isolated. A total of 250 ng of RNA from each dilution was subjected to targeted RNA-sequencing library preparation and sequencing.

Variations in sequencing depth depending on the leukemic burden present in a sample and baseline error rates between sequencing platforms are important elements which can affect assay performance and reliability.

Additionally, sequencing read requirements play an important role in the feasibility of assay adaptation into practice.

To address these factors, we first examined the impact of sequencing depth and platform on assay detection met-

haematologica | 2019; 104(2)

299

L.W. Dillon et al. et al.

A

B

C

D

E

F

G

H

I

Figure 2. Evaluation of assay performance by target. Cell lines or patient cells positive for assay targets (A) PML-RARA, (B) CBFB-MYH11, (C) RUNX1-RUNX1T1, (D) BCR-ABL1 p210 isoform, (E) BCR-ABL1 p190 isoform, (F) PRAME, (G) WT-1, and (H) NPM1 mutA insertion mutation were serially diluted in peripheral blood mononuclear cells from a normal healthy individual and RNA was isolated. The resulting RNA was subjected to acute myeloid leukemia (AML) measurable residual disease (MRD) RNA-sequencing library preparation and sequencing. The data are presented as a ratio of the number of target copies/ABL1 copies (log10) on the y-axis and the cell dilution (-log10) on the x-axis. Each data point represents two replicates, except NPM1 mutA (n=1), and error bars represent the standard deviation. A linear regression was calculated for each dilution series (excluding the 100% leukemia samples and any samples for which the target was not detected) and the coefficient of determination (r2) indicated in the graph area. NPM1 mutA data were normalized to either (H) ABL1 copies or (I) wild-type NPM1 copies, both of which exhibited similar results and a strong correlation. If background values were detected in the normal healthy individual, an average of the two libraries sequenced is represented as a dashed line.

rics. A set of libraries generated on a serial dilution of K562 cells (positive for BCR-ABL1 p210 fusion transcript, WT1, and PRAME expression) in normal healthy peripheral blood mononuclear cells were subjected to sequencing on both the Illumina MiSeq and HiSeq 2500 platforms at an average depth of 3.6 million and 38.6 million reads, respectively. Regardless of platform or sequencing depth, the ratio of each target relative to ABL1 expression was highly reproducible and significantly correlated (Pearson correlation r = 1.00) (Online Supplementary Figure S3), confirming the robustness of the bioinformatics pipeline across different depths and sequencers.

Next, we determined the minimum number of reads necessary to retain assay sensitivity across the dynamic range. To do so, libraries obtained from the serial dilution of K562 cells or NPM1 mutA patient cells in healthy donor peripheral blood mononuclear cells were down-sampled stepwise and separately subjected to the bioinformatics pipeline (Online Supplementary Figure S4). We determined that as few as 1 million reads are needed for the detection 300

of fusion transcripts and 3 million reads for the detection of NPM1 insertion mutations (due to high wild-type NPM1 transcript expression). Thus, with a minimal sequencing read requirement equivalent to a micro or nano flow cell on the MiSeq platform, this assay is able to retain up to a 5-log dynamic range.

Comparison of the performance of the study assay with that of current gold-standard, single-target methodologies Next, we compared the performance of the AML MRD RNA-sequencing panel to that of the current gold-standard qPCR assays for single-mutation MRD detection.

The same RNA from the serial dilution of positive control cell lines used to generate targeted RNA-sequencing libraries was assessed by qPCR or ddPCR for NPM1 mutA (Figure 3A), RUNX1-RUNX1T1 (Figure 3B), or CBFBMYH11 type A (Figure 3C) mutations. Across all three mutations, a comparable limit of detection was observed for both qPCR and ddPCR (right graph) compared to tarhaematologica | 2019; 104(2)

Targeted RNA-sequencing for MRD in AML

A

B Figure 3. Comparison of the performance of the acute myeloid leukemia measurable residual disease targeted RNA-sequencing assay with that of current gold-standard, single-target methodologies. Gold-standard quantitative real-time polymerase chain reaction (qPCR) or digital droplet polymerase chain reaction (ddPCR) (right) techniques performed on the serial dilution samples for (A) NPM1 mutA, (B) RUNX1-RUNX1T1, and (C) CBFB-MYH11 exhibited comparable performance to that of the acute myeloid leukemia (AML) measurable residual disease (MRD) targeted RNAsequencing (RNA-seq) technique (left). The data are presented as a ratio of the number of target copies/ABL1 copies (log10) on the y-axis and the cell dilution (-log10) on the x-axis. A linear regression was calculated for each dilution series (excluding the 100% leukemia samples and any samples for which the target was not detected) and the coefficient of determination (r2) indicated in the graph area.

C

geted RNA-sequencing (left graph), thus confirming that the AML MRD RNA-sequencing assay on its own serves as an adequate replacement for current, single-target MRD detection assays.

Comparison of the performance of the study assay with that of a commercially available myeloid targeted RNA-sequencing assay The Myeloid FusionPlex assay from ArcherDx is a commercially available targeted RNA-sequenced assay which also utilizes molecular barcodes and is designed to detect and identify fusions, point mutations, and expression level changes in a panel of 84 genes associated with malignancies of myeloid origin. In contrast to the assay presented here, the FusionPlex technology utilizes an anchored multiplex-PCR enrichment (AMP-E) approach, which is intended for fusion discovery in a bulk tumor setting.

The performance of the Myeloid FusionPlex assay for the detection of MRD was assessed for NPM1 mutA and CBFB-MYH11 mutations using 250 ng of the same RNA haematologica | 2019; 104(2)

as used in the AML MRD RNA-sequencing assay. We confirmed that the Myeloid FusionPlex assay can detect these mutations, but at varying frequencies (Online Supplementary Table S2). Compared to the Myeloid FusionPlex assay, our AML MRD RNA-sequencing assay exhibits a 1,000-fold increased limit of detection for NPM1 mutA and 100-fold increased limit of detection for CBFB-MYH11 (Figure 4), indicating that our assay outperforms a commercially available targeted RNA-sequencing assay for MRD-level detection in AML.

Validation of assay performance in patient specimens Finally, we examined the performance of the AML MRD RNA-sequencing assay applied to patient samples.

RNA was isolated from diagnostic peripheral blood or bone marrow samples of patients clinically annotated to harbor mutations included in our assay and assessed the samples by AML MRD targeted RNA-sequencing. In all cases, the annotated mutation was detectable using our assay (Table 1). For NPM1 mutA and RUNX1-RUNX1T1301

L.W. Dillon et al. et al.

Figure 4. Comparison of the performance of the acute myeloid leukemia measurable residual disease targeted RNA-sequencing assay to that of the commercially available Myeloid FusionPlex assay. NPM1 mutA and CBFB-MYH11 serial dilution samples were analyzed using the ArcherDx Myeloid FusionPlex assay. The ratio of target copies/ABL1 copies (log10) is plotted on the y-axis and the cell dilution (-log10) on the x-axis for both the National Institutes of Health acute myeloid leukemia measurable residual disease targeted RNA-sequencing assay (NIH AML MRD RNA-Seq) (blue) and the Myeloid FusionPlex assay (red). The shaded areas indicate the cell dilution frequencies in which the mutation was detectable by both methodologies.

Table 1. Validation of mutation detection in diagnostic patient samples by acute myeloid leukemia measurable residual disease targeted RNAsequencing.

Patient A B C D E F G H I J

Sample type

Annotated mutation

Mutation detected by RNA-seq

Gene expression detected by RNA-seq

PB BM PB PB BM BM BM BM BM BM

NPM1 mut NPM1 mut NPM1 mut RUNX1-RUNX1T1 PML-RARA PML-RARA PML-RARA BCR-ABL1 p210 BCR-ABL1 p210 CBFB-MYH11

NPM1 mutA NPM1 mutA NPM1 mutA RUNX1-RUNX1T1 PML-RARA bcr1 PML-RARA bcr1 PML-RARA bcr3 BCR-ABL1 p210 BCR-ABL1 p210 CBFB-MYH11 type D

WT1, PRAME WT1 WT1, PRAME WT1, PRAME WT1, PRAME WT1, PRAME WT1 WT1, PRAME WT1 WT1, PRAME

RNA-seq: RNA-sequencing; PB: peripheral blood; BM: bone marrow.

positive patient samples, we also confirmed the detectability of these mutations by qPCR (Online Supplementary Table S3). Furthermore, using our AML MRD RNA-sequencing assay we were able to specifically define the isoform present in each sample, including the rarer CBFB-MYH11 type D isoform (Online Supplementary Figure S5), and to identify the overexpression of WT1 or PRAME as potential secondary tracking targets (Table 1).

To test the ability of this assay to track disease burden in a patient and detect relapse relative to standard clinical assessments, we performed AML MRD targeted RNAsequencing on RNA isolated from peripheral blood mononuclear cells of an NPM1 mutA-positive AML patient at various time-points during treatment. NPM1 mutA transcripts were detected at the time of pathological cytomorphological complete remission prior to allogenic stem cell transplantation. While NPM1 mutA transcript expression was not detected immediately after transplantation, an increase in NPM1 mutA transcript was detected at two subsequent time-points, up to 3 months prior to clinical relapse (Figure 5).

Collectively, these findings confirm that the AML MRD targeted RNA-sequencing assay can identify all of the included targets in both peripheral blood and bone mar302

row from patients harboring these mutations and is capable of detecting MRD in patient samples during clinical complete remission.

Discussion Here we present the development of a targeted RNAsequencing method for the detection and measurement of MRD in AML. This assay, which can cover over twothirds of patients in a single standardized assay (Figure 1), is highly specific, sensitive, and resilient to variations in sequencing depth and platform used during data collection. MRD status is now integrated into response criteria for AML,4 with ELN consensus guidelines for measurement now available.6 The assay presented here detects all of the molecular targets included in these guidelines.

To date, molecular MRD detection in AML has primarily focused on single target qPCR assays.18 The use of nextgeneration sequencing for AML MRD detection has begun to emerge but has primarily focused on DNA as the starting material.21-27 Several features of the AML MRD RNAsequencing assay presented here in theory make it ideal for MRD detection, including: (i) the use of next-generahaematologica | 2019; 104(2)

Targeted RNA-sequencing for MRD in AML

Figure 5. Evaluation of assay performance in serial samples from a patient. RNA isolated from peripheral blood mononuclear cells of a leukemia patient positive for the NPM1 mutA insertion mutation at serial time-points was subjected to acute myeloid leukemia (AML) measurable residual disease (MRD) RNA-sequencing library preparation and sequencing. The data are presented as the ratio of NPM1 mutA copies/ABL1 copies (log10) on the y-axis and the date of sample collection relative to allogeneic stem cell transplantation (in days) on the x-axis. Important clinical time-points are depicted above the graph. Clinical relapse occurred on day 175 while MRD was detectable by targeted RNA-sequencing on days 87 and 119. The background ratio of NPM1 mutA copies/ABL1 copies (log10) detected in the normal donor is represented as a dashed line. Pre-SCT: before stem cell transplantation; Path CR: pathological cytomorphological complete remission; Allo-SCT: allogeneic stem cell transplantation.

tion sequencing allows for both target multiplexing and flexibility in identifying mutations that could vary between patients (variations in fusion breakpoints or insertion sequences, etc.); (ii) use of primers targeting recurrent AML abnormalities greatly increases the sensitivity of the assay over bulk RNA- or DNA-sequencing; (iii) RNA as the starting material allows for the simultaneous examination of mutations/fusions and changes in transcript expression; (iv) RNA increases the limit of detection over that provided by DNA if the transcript is expressed at a level greater than the genomic equivalent per cell; (v) UMIs allow for the absolute quantification of target levels; and (vi) targeted primers during reverse transcription increase the limit of detection and allow for efficient use of the starting material. However, it is conceivable that this RNA-sequencing assay could be complemented in the future by the use of a DNA-based AML MRD next-generation sequencing approach (tracking, for example, somatic mutations in TP53, IDH1, IDH2, FLT3, etc.) which together would allow coverage of almost all cases of AML. Utilizing cell lines and patient samples expressing the targets included in our assay, we demonstrated that the AML MRD RNA-sequencing assay has a sensitivity for residual disease detection down to as low as 1 in 100,000 cells (Figure 2). Importantly, this detection limit is well below the threshold of 1 in 1,000 cells currently suggested by the ELN MRD consortium6 and is comparable to that of the gold standard single-target molecular techniques (Figure 3). Furthermore, due to the MRD-focused design of our assay, it can achieve up to a 1,000-fold greater sensitivity than that of the most similar targeted RNAsequencing assay for myeloid malignancies available on the commercial market (Figure 4). However, since RNA expression levels can vary from one patient to another, it is important to note that sensitivity levels can vary and a detection limit of 1 in 100,000 cells may not always be attainable. This is a general difficulty affecting all RNAbased methods of MRD detection, including qPCR, and is reflected in the recommendation that molecular relapse be haematologica | 2019; 104(2)

determined by the progression of trends across multiple time-points in an individual patient rather than by a single landmark assessment.6 Several important considerations for the clinical utility of an MRD assay include sample input requirements, cost, time, and ease of assay adaptation. Each of these factors were considered in the assay design. The use of targeted primers and the addition of UMIs during the reverse transcription step allow for maximal usage of the starting material while simplifying the workflow (Figure 1B), which can be completed in a single day and is easy to adopt and/or automate. Additionally, with a minimal sequencing requirement of only 1-3 million reads (Online Supplementary Figure S4), a single patient sample can be assessed for immediate results or multiplexed on larger scale runs to minimize costs (Online Supplementary Table S4). Importantly, this multiplex assay and analytic workflow is both flexible and expandable. The design allows for all targets to be screened at diagnosis, with the ability to tailor analysis to a subset of patient-specific targets at later time-points to further minimize sequencing costs. Additional targets can easily be added to the assay design, potentially allowing for the detection of other AML MRD markers, chimerism,28,29 and HLA loss30 for those poor-risk AML patients who are not optimally covered by this assay but who will often undergo allogeneic stem cell transplantation.

While we confirmed the feasibility of use of this test in patient blood and bone marrow (Table 1, Figure 5), future work is needed to test the utility of this technique in large cohorts of patients and to determine the specific impact of MRD detection on AML patient outcomes in this setting.

Overall, we believe that this UMI-based RNA-sequencing assay provides a high-throughput, reproducible, and broadly applicable tool for standardized detection of residual disease in patients with AML. Acknowledgments This work was supported by the Intramural Research Program of the National Heart, Lung, and Blood Institute (NHLBI) of the National Institutes of Health (NIH). GWR is 303

L.W. Dillon et al. et al.

supported by the Medical Research Scholars Program, a publicprivate partnership supported jointly by the NIH, and by generous contributions to the Foundation for the NIH by the Doris Duke Charitable Foundation (Grant #2014194). The authors thank Samuel J. Rulli and Melanie Hussong of Qiagen, Inc. for their insight and advice during the assay development. The authors also thank Yuesheng Li, Yan Luo, and Patrick Burr of

References 1. Hourigan CS, Gale RP, Gormley NJ, Ossenkoppele GJ, Walter RB. Measurable residual disease testing in acute myeloid leukaemia. Leukemia. 2017;31(7):14821490.

2. 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.

3. Buckley SA, Wood BL, Othus M, et al.

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

Haematologica. 2017;102(5):865-873.

4. Dohner H, Estey E, Grimwade D, et al.

Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel.

Blood.

2017;129(4):424-447.

5. 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.

6. Schuurhuis GJ, Heuser M, Freeman S, et al.

Minimal/measurable residual disease in AML: a consensus document from the European LeukemiaNet MRD Working Party. Blood. 2018;131(12):1275-1291.

7. Zhou Y, Wood BL. Methods of detection of measurable residual disease in AML. Curr Hematol Malig Rep. 2017;12(6):557-567.

8. Grimwade D, Ivey A, Huntly BJ. Molecular landscape of acute myeloid leukemia in younger adults and its clinical relevance.

Blood. 2016;127(1):29-41.

9. Corces-Zimmerman MR, Hong WJ, Weissman IL, Medeiros BC, Majeti R. Preleukemic mutations in human acute myeloid leukemia affect epigenetic regulators and persist in remission. Proc Natl Acad Sci USA. 2014;111(7):2548-2553.

10. Jan M, Snyder TM, Corces-Zimmerman MR, et al. Clonal evolution of preleukemic hematopoietic stem cells precedes human acute myeloid leukemia. Sci Transl Med.

304

the NHLBI DNA Sequencing and Genomics Core for providing next-generation sequencing services and troubleshooting issues that arose during data collection. Finally, the authors also thank A. John Barrett of the National Heart, Lung and Blood Institute, Paul Liu of the NHLBI, and Gabriel Ghiaur of the Sidney Kimmel Cancer Center at Johns Hopkins for kindly providing patient samples or cell lines.

2012;4(149):149ra118.

11. Shlush LI, Zandi S, Mitchell A, et al.

Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia. Nature.

2014;506(7488):328-333.

12. Genovese G, Kahler AK, Handsaker RE, et al. Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. N Engl J Med. 2014;371(26):2477-2487.

13. Jaiswal S, Fontanillas P, Flannick J, et al. Agerelated clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371(26):2488-2498.

14. Steensma DP, Bejar R, Jaiswal S, et al. Clonal hematopoiesis of indeterminate potential and its distinction from myelodysplastic syndromes. Blood. 2015;126(1):9-16.

15. Young AL, Challen GA, Birmann BM, Druley TE. Clonal haematopoiesis harbouring AML-associated mutations is ubiquitous in healthy adults. Nat Commun.

2016;7: 12484.

16. Beillard E, Pallisgaard N, van der Velden VH, et al. Evaluation of candidate control genes for diagnosis and residual disease detection in leukemic patients using 'real-time' quantitative reverse-transcriptase polymerase chain reaction (RQ-PCR) - a Europe Against Cancer program. Leukemia. 2003;17(12): 2474-2486.

17. Cilloni D, Renneville A, Hermitte F, et al.

Real-time quantitative polymerase chain reaction detection of minimal residual disease by standardized WT1 assay to enhance risk stratification in acute myeloid leukemia: a European LeukemiaNet study. J Clin Oncol. 2009;27(31):5195-5201.

18. Gabert J, Beillard E, van der Velden VH, et al. Standardization and quality control studies of 'real-time' quantitative reverse transcriptase polymerase chain reaction of fusion gene transcripts for residual disease detection in leukemia - a Europe Against Cancer program. Leukemia. 2003;17(12): 2318-2357.

19. Ivey A, Hills RK, Simpson MA, et al.

Assessment of minimal residual disease in standard-risk AML. N Engl J Med. 2016; 374(5):422-433.

20. Cancer Genome Atlas Research N, Ley TJ, Miller C, et al. Genomic and epigenomic

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):20592074.

Gaksch L, Kashofer K, Heitzer E, et al.

Residual disease detection using targeted parallel sequencing predicts relapse in cytogenetically normal acute myeloid leukemia.

Am J Hematol. 2018;93(1):23-30.

Klco JM, Miller CA, Griffith M, et al.

Association between mutation clearance after induction therapy and outcomes in acute myeloid leukemia. JAMA. 2015;314 (8):811-822.

Levis MJ, Perl AE, Altman JK, et al. A nextgeneration sequencing-based assay for minimal residual disease assessment in AML patients with FLT3-ITD mutations. Blood Adv. 2018;2(8):825-831.

Malmberg EB, Stahlman S, Rehammar A, et al. Patient-tailored analysis of minimal residual disease in acute myeloid leukemia using next-generation sequencing. Eur J Haematol. 2017;98(1):26-37.

Morita K, Kantarjian HM, Wang F, et al.

Clearance of somatic mutations at remission and the risk of relapse in acute myeloid leukemia. J Clin Oncol. 2018;36(18):17881797.

Roloff GW, Lai C, Hourigan CS, Dillon LW. Technical advances in the measurement of residual disease in acute myeloid leukemia.

J Clin Med. 2017;6(9).

Rothenberg-Thurley M, Amler S, Goerlich D, et al. Persistence of pre-leukemic clones during first remission and risk of relapse in acute myeloid leukemia. Leukemia.

2018;32(7):1598-1608.

Aloisio M, Licastro D, Caenazzo L, et al. A technical application of quantitative next generation sequencing for chimerism evaluation. Mol Med Rep. 2016;14(4):2967-2974.

Bornhauser M, Oelschlaegel U, Platzbecker U, et al. Monitoring of donor chimerism in sorted CD34+ peripheral blood cells allows the sensitive detection of imminent relapse after allogeneic stem cell transplantation.

Haematologica. 2009;94(11):1613-1617.

Ahci M, Toffalori C, Bouwmans E, et al. A new tool for rapid and reliable diagnosis of HLA loss relapses after HSCT. Blood.

2017;130(10):1270-1273.

haematologica | 2019; 104(2)

ARTICLE

Acute Myeloid Leukemia

Persistent IDH1/2 mutations in remission can predict relapse in patients with acute myeloid leukemia

Chi Young Ok,1 Sanam Loghavi,1 Dawen Sui,2 Peng Wei,2 Rashmi KanagalShamanna,1 C. Cameron Yin,1 Zhuang Zuo,1 Mark J. Routbort,1 Guilin Tang,1 Zhenya Tang,1 Jeffrey L. Jorgensen,1 Rajyalakshmi Luthra,1 Farhad Ravandi,3 Hagop M. Kantarjian,3 Courtney D. DiNardo,3 L. Jeffrey Medeiros,1 Sa A. Wang1 and Keyur P. Patel1 Department of Hematopathology, 2Department of Biostatistics and 3Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA

1

Ferrata Storti Foundation

Haematologica 2019 Volume 104(2):305-311

ABSTRACT

P

ersistence of IDH1 or IDH2 mutations in remission bone marrow specimens of patients with acute myeloid leukemia has been observed, but the clinical impact of these mutations is not well known. In this study, we evaluated 80 acute myeloid leukemia patients with known IDH1 R132 or IDH2 R140/R172 mutations and assessed their bone marrow at the time of remission to determine the potential impact of persistent IDH1/2 mutations. Approximately 40% of acute myeloid leukemia patients given standard treatment in this cohort had persistent mutations in IDH1/2. Patients with an IDH1/2 mutation had an increased risk of relapse after 1 year of follow-up compared to patients without a detectable IDH1/2 mutation (59% versus 24%; P<0.01). However, a persistent mutation was not associated with a shorter time to relapse. High IDH1/2 mutation burden (mutant allelic frequency â&#x2030;Ľ10%) did not correlate with relapse rate (77% versus 86% for patients with a low burden, i.e., mutant allelic frequency <10%; P=0.66).

Persistent mutations were also observed in NPM1, DNMT3A and FLT3 during remission, but IDH1/2 mutations remained significant in predicting relapse by multivariate analysis. Flow cytometry was comparable and complementary to next-generation sequencing-based assay for predicting relapse. Monitoring for persistent IDH1/2 mutations in patients with acute myeloid leukemia in remission can provide information that could be used to justify early interventions, with the hope of facilitating longer remissions and better outcomes in these patients.

Correspondence: cok@mdanderson.org or KPPatel@mdanderson.org Received: February 12, 2018.

Accepted: August 24, 2018.

Pre-published: August 31, 2018. doi:10.3324/haematol.2018.191148

Introduction Acute myeloid leukemia (AML), defined as more than 20% of myeloblasts in blood and/or bone marrow, is heterogeneous and complex at the genomic level.

Data from The Cancer Genome Atlas show that many genes are recurrently mutated in patients with AML, including NPM1, FLT3, DNMT3A, IDH1/2, and KRAS/NRAS.1 IDH1 and IDH2 mutations are found in 6-16% and 8-19% of AML patients, respectively.2-7 Collectively, IDH1/2 mutations are observed in 16-20% of AML patients and are enriched (25-30%) in cases of AML with a normal karyotype.6,8,9 IDH1/2 mutations are acquired early in the natural history of AML and can be present in the founding clone.10 There are known mutational hot spots in these genes: codon 132 (Arg) in IDH1 and codons 140 (Arg) and 172 (Arg) in IDH2. IDH2 R140 mutations occur more commonly than R172 mutations in AML.5 IDH1 and IDH2 mutations can also infrequently occur together at presentation.11 The presence of an IDH1/2 mutation alone is not sufficient for the development of AML.12 IDH2 mutations can be associated with clonal hematopoiesis of indeterminate potential (CHIP) in the older population.13 Moreover, IDH1/2 mutations occur haematologica | 2019; 104(2)

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/305 Š2019 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.

305

C.Y. Ok et al.

together with mutations of other genes, at frequencies that depend on the IDH allele, suggesting that additional genomic insults are needed for AML to develop fully. For example, IDH2 R140 mutations are strongly associated with NPM1 mutations.10 IDH1 and IDH2 encode NADP+-dependent isocitrate dehydrogenases, converting isocitrate to α-ketoglutarate, while reducing NADP+ to NADPH with the production of CO2. IDH1 is present in the cytoplasm and peroxisome, whereas IDH2 resides in mitochondria and is a component of the Krebs’ cycle.14 IDH1 R132 mutation and IDH2 R140/R172 mutations reduce α-ketoglutarate to the oncometabolite D-2-hydroxyglutarate (also known as R2-hydroxyglutarate).14,15 D-2-hydroxyglutarate has structural similarities to α-ketoglutarate and can competitively inhibit enzymes dependent on α-ketoglutarate, such as the TET enzyme family and histone lysine demethylases, and indeed IDH1/2 mutations in AML are associated with global DNA hypermethylation and impaired hematopoietic differentiation.16,17 Persistent IDH1 or IDH2 mutations have been observed in AML patients at the time of clinical and morphological remission.18,19 Debarri et al. reported that persistent IDH1/2 mutations in AML at the time of remission could predict relapse.19 However, their study cohort was small with only eight patients in complete remission with persistent IDH1/2 mutations, precluding a definitive conclusion. In this study, we explored the utility of mutant IDH1 and IDH2 as minimal residual disease markers in predicting relapse in a large cohort of AML patients.

Methods Patients We searched the database of The University of Texas MD Anderson Cancer Center from November 1, 2012 to December 31, 2017 and identified 80 newly diagnosed AML patients with IDH1 R132 or IDH2 R140/R172 mutations who achieved complete remission (CR) or CR with incomplete hematologic recovery (CRi), according to the 2017 European LeukemiaNet (ELN) recommendations for the diagnosis and management of AML,20 in bone marrow at any time-point of their treatment. To investigate the effect of predominant and well-established mutant IDH1/2 clones in AML, only cases with a mutant allelic frequency (MAF) ≥10% in a pre-treatment sample were included. All cases were collected consecutively and classified according to the 2017 World Health Organization (WHO) classification system.21 Patients with therapy-related AML were excluded from this study. Clinical, laboratory and cytogenetic data were collected from the patients’ electronic medical records. This study was approved by the Institutional Review Board at The University of Texas MD Anderson Cancer Center (Houston, TX, USA) and was conducted in accordance with the Declaration of Helsinki.22

IDH1/2 sequencing IDH1/2 sequencing was performed on all patients as a part of clinically validated next-generation sequencing-based (NGS) assay (a 53-gene panel, a 28-gene panel or an 81-gene panel) as described previously.23 The limit of detection was 1% for the NGS assay. A sequencing library was prepared using 250 ng of genomic DNA and respective sequencing libraries were subjected to a MiSeq sequencer (Illumina Inc.). NGS data were analyzed using MiSeq Reporter (TruSeq) or SureCall (Haloplex). The Integrative Genomics Viewer (IGV, Broad Institute) was used to visualize read 306

alignment and confirm variant calls.24 A custom-developed, inhouse software package (OncoSeek) was used to annotate sequence variants and to interface the data with the IGV. Nomenclature of genetic variants was designated following the Human Genome Variation Society recommendations.25

FLT3 analysis The presence of internal tandem duplications or point mutations at codon 835 or 836 in FLT3 was determined as described previously.26

Cytogenetic analysis Conventional chromosome analysis (karyotyping) was performed on G-banded metaphase cells prepared from unstimulated 24-hour and 48-hour bone marrow cultures as described previously.27 Twenty metaphases were analyzed in most cases, but fewer than 20 metaphases were analyzed in some cases when inadequate metaphases were available for complete analysis. The results were reported using the current International System for Human Cytogenetic Nomenclature.28 Cytogenetic risk stratification was assessed in each patient using the United Kingdom Medical Research Council (UKMRC) system.29

Statistical analysis A Fisher exact test was used when comparing categorical variables. Mann-Whitney and Kruskal-Wallis tests were used when comparing numerical variables in two groups or three or more groups, respectively. The cumulative incidence rate of relapse was determined using the competing risk method. The association between an IDH1/2 mutation and the cumulative incidence outcome was determined using a proportional subdistribution hazards regression model (Fine and Gray regression model).30 Differences in the cumulative incidence among patients with different mutations were assessed using the Gray test.31 Time to relapse was calculated from the date of morphological remission to the date of relapse. All variables with a P value <0.05 (twotailed) were considered to be statistically significant. Statistical analyses were performed using GraphPad Prism 6 (GraphPad Software, Inc., La Jolla, CA, USA) and SAS 9.4 for Windows (SAS Institute Inc., Cary, NC, USA).

Results Patients The study group included 80 patients (37 men and 43 women) with a median age of 59 years (range, 31 to 90) at diagnosis. The median hemoglobin concentration, white blood cell count and platelet count were 9.0 g/dL (range, 6.3 to 13.9), 7.9x109/L (range, 0.4 to 263.1x109/L) and 58x109/L (range, 1 to 1,069x109/L), respectively (Table 1).

The median bone marrow blast count was 62% (range, 21 to 95%). Among 76 patients with cytogenetic information available, 88% (n=67) and 12% (n=9) had intermediate and adverse cytogenetic risk, respectively. There were no patients with favorable cytogenetic risk. A diploid karyotype was seen in 52 (68%) patients. Various frontline therapies were administered to this cohort of patients, but no patients received an IDH inhibitor as frontline therapy.

All patients younger than 60 years of age (n=41) were treated with intensive chemotherapy including 7+3 (idarubicin and cytarabine), CIA (clofarabine, idarubicin and cytarabine), FIA (fludarabine, idarubicin and cytarabine), or CLIA (cladribine, idarubicin and cytarabine with or without sorafenib). The patients over 60 years old (n=39) haematologica | 2019; 104(2)

IDH1/2 mutations in remission predict AML relapse

were treated with intensive chemotherapy (n=10), hypomethylating agents (n=23) or low-dose cytarabine with or without nucleoside analogs (n=6). The median clinical follow-up was 17.5 months (range, 4.9 to 77.5 months).

IDH mutations in pretreatment samples All 80 patients harbored IDH1 and/or IDH2 mutations: 78 patients had a single IDH1 or IDH2 mutation and two patients had both IDH1 and IDH2 mutations. The two patients who had two different IDH1/2 mutations had major (20~34%) and minor (1~14%) clones represented

by MAF. As a single mutation, IDH2 R140 mutations were most common (n=46), followed by IDH1 R132 (n=24) and IDH2 R172 (n=7) mutations. IDH2 R172_H173delinsSA was found in one patient. Detailed information regarding the IDH1/2 mutations is presented in Table 1. The median MAF of IDH1/2 mutations in pretreatment samples was 43.8% (range, 12.3% to 62.7%). The median MAF of the IDH1 R132 mutation (39.2%) was similar to that of IDH2 R140 (44.1%) and IDH2 R172 (42.5%) mutation (P=0.31).

There were no significant differences in median bone marrow blast count among the three groups (P=0.54).

Persistent IDH mutations in complete remission or complete remission with incomplete hematologic remission The mutational status of IDH1/2 was available in first CR (n=36) or CRi (n=44) for all patients. In 51 patients treated with intensive chemotherapy, analysis of IDH1/2 was performed after the first, second and third or beyond cycles of therapy in 21, 16 and 14 patients, respectively. In 23 patients treated with hypomethylating agents, analysis of IDH1/2 was performed before and after the fourth cycle of therapy in nine and 15 patients, respectively. The latter also included six patients who had received six or more cycles of therapy. In patients treated with low-dose cytarabine with or without additional drugs (n=6), one, one and four patients were tested for IDH1/2 after, respectively, their first, second and fourth or beyond cycles of therapy. A total of 31 (39%) patients had persistent IDH1/2 mutations in CR or CRi (CRIDH+). Among the patients in CR (n=36), 12 (33%) had persistent IDH1/2 mutations. Similarly, among patients in CRi (n=44), 19 (43%) had persistent IDH1/2 mutations (P=0.49). IDH1 R132, IDH2 R140 and IDH2 R172 mutations were observed in, respectively, ten (38.5%), 19 (41.3%), and two (25%) patients with mutations in a pretreatment bone marrow specimen (P=0.68) (Figure 1A). Compared to the MAF values in pretreatment samples, the MAF of IDH1/2 mutations in CR or CRi were reduced in all patients (median MAF: 10.2%, range, 1% to 34.3%) (Figure 1B). CRIDH+ was not correlated with cytogenetic abnormalities. CRIDH+ was observed in patients with diploid karyotype and those with any cytogenetic abnormalities with similar frequency (40.4% and 37.5%, respectively; P=0.99). Of 24 patients with cytogenetic abnormalities in pretreatment samples, only two had persistent cytogenetic abnormalities.

IDH mutations in remission are associated with an increased risk of relapse The cumulative incidence rate of relapse in patients with CRIDH+ was 59% at 12 months and 80% at 24 months. The cumulative incidence rate was significantly haematologica | 2019; 104(2)

higher in patients with CRIDH+ than in patients without a detectable IDH1/2 mutation in remission (CRIDH-) (Figure 2A). Using the Fine and Gray regression model, the risk of relapse at 1 year of follow-up was higher for patients with CRIDH+ than patients in the CRIDH- group (59% versus 24%; hazard ratio, 3.89; 95% confidence interval: 1.98 to 7.62; P<0.01) (Table 2). Regarding mutation type, 90%, 74% and 100% of patients with persistent IDH1 R132, IDH2 R140 and IDH2 R172 clones relapsed, respectively (P=0.44). There were no differences regarding relapse between patients treated with intensive chemotherapy and hypomethylating agents (47.1% and 43.5%, respectively) (P=0.77). The median time to relapse in patients with CRIDH+ was not significantly different from that of CRIDH- patients (median: 8.1 and 6.9 months, respectively) (P=0.71).

IDH1/2 mutation burden does not correlate with relapse Among patients with CRIDH+, the MAF values for CRIDH+ were not significantly different between relapsed (median MAF: 10.0%) and non-relapsed patients (median MAF: 20.5%) (P=0.19). To further evaluate the correlation between IDH1/2 mutation burden and relapse, we arbitrarily divided patients according to whether they had a high MAF (â&#x2030;Ľ10%) or low MAF (<10%). Among 17 patients with a high MAF, 13 (77%) patients relapsed, whereas 12 of 14 (86%) patients with a low MAF relapsed (P=0.66). The difference in the median time to relapse Table 1. Laboratory, cytogenetic and IDH1/2 mutation data of patients with acute myeloid leukemia (n=80).

Age (years) Hemoglobin (g/dL) White blood cells, x109 L Platelets, x109 L Bone marrow blasts (%)

Cytogenetic risk Favorable Intermediate Adverse Treatment Intensive chemotherapy Hypomethylating agent Low-dose cytarabine IDH1/2 mutation Single p.R132C p.R132H p.R132S p.R132L p.R132G p.R140Q p.R140L p.R140W p.R172K p.R172G p.R172_H173delinsSA Double p.R132C and p.R140Q

Median

Range

59 9.0 7.9 58 62

31 to 90 6.3 to 13.9 0.4 to 263.1 1 to 1,069 21 to 95

Number of patients

(%)

0 67 9

0 88 12

51 23 6

64 29 7

13 5 3 2 1 44 1 1 6 1 1

16.7 6.4 3.8 2.6 1.3 56.4 1.3 1.3 7.7 1.3 1.3

2

NA 307

C.Y. Ok et al.

between patients in the high and low MAF groups (6.2 and 9.9 months, respectively) was not statistically significant (P=0.18). We also assessed different MAF as cutoffs for high versus low mutation burden (5%, 20% and 30%), but correlation with relapse was not observed using any of these cutoffs (data not shown).

Other gene mutations were detected in 89% of patients who had IDH1/2 mutations in a pre-treatment sample.

NPM1 (n=38) was the most frequently co-mutated gene followed by DNMT3A R882 (n=25), FLT3-ITD (internal tandem duplication) (n=22), and KRAS/NRAS (n=12). Few (<5) patients had mutations in ASXL1, BRAF, CEBPA, JAK2, RUNX1, TET2 or TP53. CRIDH+ was significantly more common (39%) than CRFLT3+ (n=3, 14%), CRNPM1+ (n=4, 11%), and CRKRAS/NRAS+ (n=0, 0%) in CR or CRi (P>0.05). CRDNMT3A+ was present with a similar frequency, being seen in 36% of patients (n=9). We assessed the effect of CRIDH+ in the context of co-mutations using the Fine and Gray regression model. By univariate analysis,

A

B

IDH mutations in remission predict relapse in the context of co-mutations

Figure 1. Persistent IDH1/2 mutations in remission and changes in mutant allelic frequencies in pretreatment and remission samples. (A) Percentages of persistent IDH1/2 mutations in patients with acute myeloid leukemia in remission. Persistent mutations occur at similar frequencies for the different mutations. (B) Mutant allelic frequency (MAF) of IDH mutations present in pretreatment samples and remission samples. The median mutant allelic frequency is shown for each occasion.

A

B

Figure 2. Cumulative incidence rates of relapse in patients with persistent IDH1/2 mutations in remission. (A) Cumulative incidences of relapse in patients with persistent IDH1/2 mutation and patients without detectable IDH1/2 mutation. The cumulative incidence rate is significantly higher in patients with persistent IDH1/2 mutation in remission (CRIDH+) than in patients without detectable IDH1/2 mutation in remission (CRIDH-). (B) Cumulative incidence rate of relapse in patients with respect to mutational status in IDH1/2 and flow cytometrically determined presence of minimal residual disease in remission. The cumulative incidence of relapse was significantly higher in patients who were positive in both molecular and flow cytometry tests compared to patients with a positive result in either of the tests or negative in both. CRIDH+; persistent IDH1/2 mutation in remission, CRIDH-; non-detectable IDH1/2 mutation in remission, FC+; positive for minimal residual disease by flow cytometry, FCâ&#x20AC;&#x201C;; negative for minimal residual disease by flow cytometry.

308

haematologica | 2019; 104(2)

IDH1/2 mutations in remission predict AML relapse Table 2. Risk of relapse according to the presence of persistent mutations in different genes (Fine and Gray regression model).

Age ≥ 60 years Female gender CRIDH+ CRFLT3+ CRNPM1+ CRDNMT3A+

HR

Univariate 95% CI

P value

HR

Multivariate 95% CI

P value

0.99 1.23 3.89 12.6 1.38 2.01

0.96 to 1.03 0.56 to 2.67 1.98 to 7.62 1.66 to 95.1 0.25 to 7.75 0.92 to 4.39

0.9 0.61 <0.01 0.01 0.71 0.08

4.45 20.2

2.15 to 9.19 3.99 to 102

<0.01 <0.01

HR; hazard ratio, CI; confidence interval, CRIDH+; persistent IDH1/2 mutation in CR or CRi, CRFLT3+; persistent FLT3-ITD mutation in CR or CRi, CRNPM1+; persistent NPM1 mutation in CR or CRi, CRDNMT3A+; persistent DNMT3A mutation in CR or CRi.

CRFLT3+ also demonstrated an increased risk of relapse. By multivariate analysis, CRIDH+ and CRFLT3+ remained significant for an increased risk of relapse.

We also assessed the dynamic changes of clonal architecture in 25 CRIDH+ patients who relapsed. Comparing mutational profiles at CR/CRi and relapse, four patients acquired novel mutations at relapse. These mutations were ERBB2 p.R784C (MAF: 12.3%), FLT3 p.D835Y (1.5%), TP53 p.G245D (3.8%) and WT1 p.K467fs (12.2%).

These mutations showed subclonal fraction patterns compared with the MAF of IDH1/2 mutation at relapse.

Flow cytometry is comparable to next-generation sequencing in predicting relapse We compared molecular test results to those of multiparametric flow cytometry (FC) immunophenotyping in CR/CRi bone marrow specimens. Flow cytometric results were available for a total of 79 patients. Minimal residual disease (MRD) determination by FC has been described previously.32,33 The sensitivity of the flow cytometry was validated to 0.1% - 0.01% depending on the leukemic cell phenotype. According to FC, 19 (26%) patients had MRD, 55 (76%) were MRD-negative and results were indeterminate in five patients. Among 74 patients in whom both FC and molecular MRD tests were performed, the results were concordant in 61 (82%) patients and discordant in 13 (18%) patients with a statistically significant association (P<0.01). Of the 13 patients with discordant FC and molecular testing MRD results, ten patients were positive

by FC only and three patients were positive by molecular testing only. Four of the five patients for whom FC was indeterminate with regards to MRD had a persistent IDH2 p.R140Q mutation with various MAF values (median 12.5%; range, 1 to 28.5%) and all of them relapsed.

Similarly to CRIDH+ patients, those who were positive for MRD assessed by flow cytometry (FC+) showed an increased risk of relapse compared to patients who were negative for MRD by flow cytometry (FC–) after 1 year of follow-up (63% versus 27%; hazard ratio, 4.24; 95% confidence interval: 2.22 to 8.13; P<0.01). Patients with positive results in both methods (CRIDH+/FC+) had a significantly higher risk of relapse compared to those with discordant results (CRIDH+/FC– or CRIDH-/FC+) or negative results in both methods (CRIDH-/FC–) (Figure 2 and Table 3).

Discussion IDH1 and IDH2 mutations are not uncommon in AML. In addition, IDH1/2 mutations can be found in the prehaematologica | 2019; 104(2)

Table 3. Risk of relapse with respect to presence or absence of a persistent IDH1/2 mutation and flow cytometry determined minimal residual disease status (Fine and Gray regression model).

Group IDH+

HR +

CR /FC (n=16) CRIDH+/FC- or CRIDH-/FC+ (n=13) CRIDH-/FC- (n=45)

95% CI

P value

0.15 to 0.87 0.08 to 0.34

0.02 <0.01

Reference 0.36 0.17

CRIDH+; persistent IDH1 or IDH2 mutation in remission, CRIDH-; no detectable IDH1 or IDH2 mutation in remission, FC+; positive for minimal residual disease by flow cytometry, FC-; negative for minimal residual disease by flow cytometry.

leukemic clone in individuals without pathology-proven AML or even in healthy individuals as a sign of age-related clonal hematopoiesis.13,34-36 For these reasons, in this study, we only selected AML patients with an IDH1/2 mutation as a predominant clone (MAF >10%) in pretreatment samples to minimize the effect on our analysis of subclonal IDH1/2 mutations.

In patients with AML associated with IDH1/2 mutations, karyotyping is not a preferred method for monitoring persistent aberrancy during remission for at least two reasons: cytogenetic analysis is less sensitive and IDH1/2 mutations are enriched in AML with a normal karyotype.5,8,11,37 In support of this statement, 68% of patients had a diploid karyotype at diagnosis. Out of 24 patients who had cytogenetic abnormalities in pretreatment samples, only two had persistent cytogenetic abnormalities in the remission sample. Therefore, follow-up with more sensitive methods is necessary to monitor for MRD. In this cohort of patients, IDH2 mutations (69%) were more common than IDH1 mutations (31%). In the IDH2 group, mutations at codon 140 were far more frequent than R172 mutations in an approximately 5.8 to 1 ratio.

IDH1 mutations were present in almost one-third of patients. These frequency data are consistent with those from other studies of AML in the literature.5,11,37 The median MAF of an IDH1/2 mutation was 43.8%, indicating that the IDH1/2 mutation was the predominant clone in most patients. The median MAF of the IDH1 R132 mutation (39.2%) was slightly lower than that of IDH2 mutations (44.1% and 42.5% for IDH2 R140 and R172 mutations, respectively).

Persistent IDH1/2 mutations in patients with AML who are in complete remission (CRIDH+) have been reported by others.18,19 We observed that approximately 40% of AML patients in remission had persistent IDH1/2 mutations with decreased MAF regardless of IDH mutation subtype (IDH1 R132, IDH2 R140 and IDH2 R172) or treatment 309

C.Y. Ok et al.

type (intensive chemotherapy versus hypomethylating agents). Approximately 50% of patients with CRIDH+ had MAF below the assay sensitivity of the Sanger sequencing (<10%). This indicates that a NGS-based approach is necessary to monitor persistent IDH1/2 mutations. CRIDH+ was associated with an increased risk of relapse (hazard ratio, 3.89; 95% confidence interval: 1.98-7.62; P<0.01) compared to patients with CRIDH-. However, CRIDH+ was not associated with a shorter time to relapse (median 8.1 months versus 6.9 months in patients with CRIDH-; P=0.71).

Interestingly, high mutation burden did not correlate with relapse in this study because patients with lower IDH1/2 mutation burden (MAF <10%) relapsed with a similar frequency as patients with a higher mutation burden (MAF ≥10%) (77% and 86%, respectively, P=0.66). Accordingly, these data suggest that presence of persistent IDH1/2 mutation in remission is per se associated with relapse in AML patients and that mutation burden does not have an additive predictive effect.

Focusing on a single event (IDH1/2 mutation) as a predictive marker of relapse in AML is potentially problematic because of frequent co-mutations in other genes including FLT3, NPM1 and DNMT3A. Indeed, co-mutations in other genes were detected in the majority of patients (89%) in this study cohort. However, persistent mutation in other genes in remission was rare, except for DNMT3A. By univariate analysis, CRFLT3+ also showed an increased risk of relapse. This result might not be reliable because only a few patients (n=3) had persistent FLT3 mutation in remission.

We noticed that a few AML patients acquired novel mutations at relapse, but at a relatively low burden (MAF range: 1.5% – 12.2%). These mutations occurred in genes in the activated signaling (FLT3 or KRAS) and tumor suppressor (TP53 and WT1) classes, apparently providing either proliferative or survival signals to the IDH-mutated clone.

FC is a powerful tool for detecting residual leukemic cells and can be used to predict relapse in patients with AML. The concordance rate for detecting MRD between FC and molecular methods was 82%, similar to earlier studies.38,39 Positivity for MRD determined by FC was also

References 1. Cancer Genome Atlas Research Network.

Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):2059-2074.

2. Chou WC, Hou HA, Chen CY, et al. Distinct clinical and biologic characteristics in adult acute myeloid leukemia bearing the isocitrate dehydrogenase 1 mutation. Blood.

2010;115(14):2749-2754.

3. Mardis ER, Ding L, Dooling DJ, et al.

Recurring mutations found by sequencing an acute myeloid leukemia genome. N Engl J Med. 2009;361(11):1058-1066.

4. Schnittger S, Haferlach C, Ulke M, Alpermann T, Kern W, Haferlach T. IDH1 mutations are detected in 6.6% of 1414 AML patients and are associated with intermediate risk karyotype and unfavorable prognosis in adults younger than 60 years and unmutated NPM1 status. Blood.

2010;116(25):5486-5496.

310

associated with an increased risk of relapse in this cohort.

Interestingly, patients with positive results according to both methods (CRIDH+/FC+) had a significantly higher risk of relapse compared to those with discordant results (CRIDH+/FC– or CRIDH-/FC+) or negative results by both methods (CRIDH-/FC–). These findings suggest that mutational analysis and FC are complementary methods useful for predicting relapse in AML patients in CR or CRi. The data we present are in accordance with those of a recent study by Jongen-Lavrencic et al., who investigated 430 patients with AML or refractory anemia with excess blasts treated according to the clinical protocol of either the HOVON or SAKK with achievement of either CR or CRi after two cycles of induction chemotherapy.39 Mutational screening was performed at the time of diagnosis and at CR/CRi using a targeted, 54-gene NGS panel (limit of detection: ≤1% of mutant allele). Their study showed that persistent mutation in genes other than DNMT3A, TET2 and ASXL1 in CR/CRi was an independent risk factor for relapse. IDH1 and IDH2 mutations were included in their study and showed a similar frequency of persistent mutation in CR/CRi (28%). However, the authors did not focus on particular genes with respect to the increased risk of relapse.

To the best of our knowledge, our study is the largest cohort (n=80) investigating the impact of persistent IDH1/2 mutations in CR/CRi in AML patients. However, our study does have some limitations: (i) the time point of IDH1/2 analysis in remission was not uniform, and (ii) our cohort was not sufficiently large to reliably investigate the effect of co-mutations in remission. Larger-scale studies are necessary to reproduce the results of our study.

In summary, approximately 40% of AML patients with an IDH1/2 mutation at initial diagnosis will have persistent mutations after therapy in remission bone marrow samples. A persistent IDH1/2 mutation is associated with an increased risk of relapse. Monitoring IDH1/2 mutations in AML patients during remission using a highly sensitive NGS-based assay may provide useful information to guide early interventions with the aim of achieving longer remissions and better outcomes.

5. Im AP, Sehgal AR, Carroll MP, et al.

DNMT3A and IDH mutations in acute myeloid leukemia and other myeloid malignancies: associations with prognosis and potential treatment strategies.

Leukemia.2014;28(9):1774-1783.

6. Paschka P, Schlenk RF, Gaidzik VI, et al.

IDH1 and IDH2 mutations are frequent genetic alterations in acute myeloid leukemia and confer adverse prognosis in cytogenetically normal acute myeloid leukemia with NPM1 mutation without FLT3 internal tandem duplication. J Clin Oncol. 2010;28(22):3636-3643.

7. Patel KP, Ravandi F, Ma D, et al. Acute myeloid leukemia with IDH1 or IDH2 mutation: frequency and clinicopathologic features. Am J Clin Pathol. 2011;135(1):3545.

8. Medeiros BC, Fathi AT, DiNardo CD, Pollyea DA, Chan SM, Swords R. Isocitrate dehydrogenase mutations in myeloid malignancies. Leukemia. 2017;31(2):272-281.

9. Rakheja D, Konoplev S, Medeiros LJ, Chen

10.

11.

12.

13.

14.

15.

W. IDH mutations in acute myeloid leukemia. Hum Pathol. 2012;43(10):15411551.

Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic classification and prognosis in acute myeloid leukemia. N Engl J Med. 2016;374(23):2209-2221.

DiNardo CD, Ravandi F, Agresta S, et al.

Characteristics, clinical outcome, and prognostic significance of IDH mutations in AML. Am J Hematol. 2015;90(8):732-736.

Sasaki M, Knobbe CB, Munger JC, et al.

IDH1(R132H) mutation increases murine haematopoietic progenitors and alters epigenetics. Nature. 2012;488(7413):656-659.

Jaiswal S, Fontanillas P, Flannick J, et al. Agerelated clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371 (26):2488-2498.

Reitman ZJ, Parsons DW, Yan H. IDH1 and IDH2: not your typical oncogenes. Cancer Cell. 2010;17(3):215-216.

Cairns RA, Mak TW. Oncogenic isocitrate dehydrogenase mutations: mechanisms,

haematologica | 2019; 104(2)

IDH1/2 mutations in remission predict AML relapse

16.

17.

18.

19.

20.

21.

22.

23.

models, and clinical opportunities. Cancer Discov. 2013;3(7):730-741.

Xu W, Yang H, Liu Y, et al. Oncometabolite 2hydroxyglutarate is a competitive inhibitor of alpha-ketoglutarate-dependent dioxygenases. Cancer Cell. 2011;19(1):17-30.

Figueroa ME, Abdel-Wahab O, Lu C, et al.

Leukemic IDH1 and IDH2 mutations result in a hypermethylation phenotype, disrupt TET2 function, and impair hematopoietic differentiation. Cancer Cell. 2010;18(6):553-567.

Chou WC, Peng KY, Lei WC, et al.

Persistence of mutant isocitrate dehydrogenase in patients with acute myeloid leukemia in remission.

Leukemia.

2012;26(3):527-529.

Debarri H, Lebon D, Roumier C, et al.

IDH1/2 but not DNMT3A mutations are sui targets for minimal residual disease monitoring in acute myeloid leukemia patients: a study by the Acute Leukemia French Association. Oncotarget. 2015;6(39):4234542353.

Dohner H, Estey E, Grimwade D, et al.

Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129 (4):424-447.

Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood.

2016;127(20):2391-2405.

World Medical Association. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA. 2013;310(20):21912194.

Ok CY, Patel KP, Garcia-Manero G, et al.

TP53 mutation characteristics in therapyrelated myelodysplastic syndromes and

haematologica | 2019; 104(2)

24.

25.

26.

27.

28.

29.

30.

31.

32.

acute myeloid leukemia is similar to de novo diseases. J Hematol Oncol. 2015;8:45-52.

Robinson JT, Thorvaldsdottir H, Winckler W, et al. Integrative genomics viewer. Nat Biotechnol. 2011;29(1):24-26. den Dunnen JT, Dalgleish R, Maglott DR, et al. HGVS Recommendations for the description of sequence variants: 2016 update. Hum Mutat. 2016;37(6):564-569.

Chen W, Jones D, Medeiros LJ, Luthra R, Lin P. Acute myeloid leukaemia with FLT3 gene mutations of both internal tandem duplication and point mutation type. Br J Haematol.

2005;130(5):726-728.

Tang Z, Medeiros LJ, Yin CC, et al. Sex chromosome loss after allogeneic hematopoietic stem cell transplant in patients with hematologic neoplasms: a diagnostic dilemma for clinical cytogeneticists. Mol Cytogenet.

2016;9:62-68.

McGowan-Jordan J, Simons A, Schmid M: ISCN: An International System for Human Cytogenomic Nomenclature (2016). Basel, Switzerland: Karger, 2016.

Grimwade D, Hills RK, Moorman AV, et al.

Refinement of cytogenetic classification in acute myeloid leukemia: determination of prognostic significance of rare recurring chromosomal abnormalities among 5876 younger adult patients treated in the United Kingdom Medical Research Council trials.

Blood. 2010;116(3):354-365.

Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. Am Stat Assoc. 1999;94(446): 496509.

Gray RJ. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat. 1988;16(3):1141-1154.

Ouyang J, Goswami M, Peng J, et al.

Comparison of multiparameter flow cytom-

33.

34.

35.

36.

37.

38.

39.

etry immunophenotypic analysis and quantitative RT-PCR for the detection of minimal residual disease of core binding factor acute myeloid leukemia. Am J Clin Pathol.

2016;145(6):769-777.

Xu J, Jorgensen JL, Wang SA. How do we use multicolor flow cytometry to detect minimal residual disease in acute myeloid leukemia? Clin Lab Med. 2017;37(4):787802.

Xie M, Lu C, Wang J, et al. Age-related mutations associated with clonal hematopoietic expansion and malignancies.

Nat Med. 2014;20(12):1472-1478.

Genovese G, Kahler AK, Handsaker RE, et al. Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. N Engl J Med. 2014;371(26):2477-2487.

Corces-Zimmerman MR, Hong WJ, Weissman IL, Medeiros BC, Majeti R. Preleukemic mutations in human acute myeloid leukemia affect epigenetic regulators and persist in remission. Proc Natl Acad Sci U S A. 2014;111(7):2548-2553.

Marcucci G, Maharry K, Wu YZ, et al. IDH1 and IDH2 gene mutations identify novel molecular subsets within de novo cytogenetically normal acute myeloid leukemia: a Cancer and Leukemia Group B study. J Clin Oncol. 2010;28(14):2348-2355.

Getta BM, Devlin SM, Levine RL, et al.

Multicolor flow cytometry and multigene next-generation sequencing are complementary and highly predictive for relapse in acute myeloid leukemia after allogeneic transplantation. Biol Blood Marrow Transplant. 2017;23(7):1064-1071.

Jongen-Lavrencic M, Grob T, Hanekamp D, et al. Molecular minimal residual disease in acute myeloid leukemia. N Engl J Med. 2018;378(13):1189-1199.

311

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(2):312-318

Acute Lymphoblastic Leukemia

Prognostic implications of additional genomic lesions in adult Philadelphia chromosomepositive acute lymphoblastic leukemia Anna Lucia Fedullo,1* Monica Messina,1* Loredana Elia,1 Alfonso Piciocchi,2 Valentina Gianfelici,1 Alessia Lauretti,1 Stefano Soddu,2 Maria Cristina Puzzolo,1 Clara Minotti,1 Felicetto Ferrara,3 Bruno Martino,4 Patrizia Chiusolo,5 Valeria Calafiore,6 Stefania Paolini,7 Marco Vignetti,1,2 Antonella Vitale,1 Anna Guarini,8 Robin Foà1* and Sabina Chiaretti1*

Hematology, Department of Translational and Precision Medicine, Sapienza University, Rome; 2GIMEMA Data Center, Rome; 3Division of Hematology and Stem Cell Transplantation Unit, Cardarelli Hospital, Naples; 4Hematology Unit, Azienda Ospedaliera Bianchi Melacrino Morelli, Reggio Calabria; 5Institute of Hematology, Catholic University, Rome; 6Division of Hematology, AOU Policlinico, University of Catania; 7"L. and A. Seràgnoli" Institute of Hematology, University of Bologna and 8Department of Molecular Medicine, Sapienza University, Rome, Italy 1

*These authors contributed equally to this work

ABSTRACT

T

Correspondence: chiaretti@bce.uniroma1.it or rfoa@bce.uniroma1.it Received: April 26, 2018.

Accepted: August 30, 2018.

Pre-published: September 6, 2018. doi:10.3324/haematol.2018.196055 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/312 ©2019 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.

312

o shed light onto the molecular basis of Philadelphia chromosome-positive acute lymphoblastic leukemia and to investigate the prognostic role of additional genomic lesions, we analyzed copy number aberrations using the Cytoscan HD Array in 116 newly diagnosed adult patients with Philadelphia chromosome-positive acute lymphoblastic leukemia enrolled in four different GIMEMA protocols, all based on a chemotherapy-free induction strategy. This analysis showed that patients with Philadelphia chromosome-positive acute lymphoblastic leukemia carry an average of 7.8 lesions/case, with deletions outnumbering gains (88% versus 12%). The most common deletions were those targeting IKZF1, PAX5 and CDKN2A/B, which were detected in 84%, 36% and 32% of cases, respectively. Patients carrying simultaneous deletions of IKZF1 plus CDKN2A/B and/or PAX5 had a significantly lower disease-free survival rate (24.9% versus 43.3%; P=0.026).

The only IKZF1 isoform affecting prognosis was the dominant negative one (P=0.003). Analysis of copy number aberrations showed that 18% of patients harbored MEF2C deletions, which were of two types, differing in size: the longer deletions were associated with the achievement of a complete molecular remission (P=0.05) and had a favorable impact on disease-free survival (64.3% versus 32.1% at 36 months; P=0.031). These findings retained statistical significance also in multivariate analysis (P=0.057). KRAS deletions, detected in 6% of cases, were associated with the achievement of a complete molecular remission (P=0.009).

These results indicate that in adults with Philadelphia chromosome-positive acute lymphoblastic leukemia a detailed evaluation of additional deletions - including CDKN2A/B, PAX5, IKZF1, MEF2C and KRAS - has prognostic implications and should be incorporated in the design of more personalized treatment strategies.

Introduction The Philadelphia (Ph) chromosome derives from the t(9;22)(q34;q11) and leads to a BCR-ABL1 rearrangement.1 The incidence of this chromosomal change in acute lymphoblastic leukemia (ALL) increases with age, being detected in 25% of adults and in about 50% of elderly patients.2 Prior to the advent of tyrosine kinase inhibitors, the outcome of Ph+ ALL patients was extremely poor,3-5 and the only haematologica | 2019; 104(2)

Additional genomic lesions in adult Ph+ ALL

possibility of a cure was allogeneic stem cell transplantation (HSCT), when feasible.6,7 The introduction of tyrosine kinase inhibitors, administered with low doses or without chemotherapy during induction, followed by consolidation chemotherapy and HSCT has markedly improved the management and outcome of adult Ph+ ALL patients, with survival rates at 5 years now approaching 50%.8-17 Different biological features - the type of fusion transcript (i.e. p190 or p210),18 the persistence and/or reappearance of minimal residual disease (MRD),19,20 additional genomic deletions (particularly IKZF1, and to a lesser extent CDKN2A/B and PAX521-24) - and the presence of mutations at relapse are associated with a worse outcome.25-27 However, a broad and refined biological algorithm that could help to optimize treatment strategies and define better whether some patients could be spared intensive treatment, including HSCT, has so far not been proposed.

To this end, in the present study we investigated copy number aberrations (CNA) in 116 newly diagnosed adult Ph+ ALL patients to identify additional molecular lesions with the aim of improving patients’ stratification and management.

Methods Experimental strategy Bone marrow and/or peripheral blood samples from 116 patients (Table 1) with newly diagnosed Ph+ ALL enrolled in four GIMEMA (Gruppo Italiano Malattie EMatologiche dell'Adulto) trials were analyzed (Online Supplementary Table S1). The study was carried out in four phases (Online Supplementary Figure S1): (i) CNA analysis of 116 samples by Cytoscan; (ii) multiplex ligation-dependent probe amplification analysis; (iii) validation of MEF2C deletions by digital droplet (dd) polymerase chain reaction (PCR); and (iv) MEF2C and KRAS mutational screening.

This study was approved in the context of an Associazione Italiana per la Ricerca sul Cancro (AIRC) project (10007) with Institutional Review Board number 2182/16.06.2011.

Copy number aberration analysis CNA were analyzed using CytoScan® HD Arrays (Affymetrix, Santa Clara, CA, USA) and Chromosome Analysis Suite (ChAS) software. Germline material from five paired samples was also evaluated. Recurrent deletions were validated with the Salsa MLPA P335 ALL-IKZF1 kit (MRC-Holland, Amsterdam, the Netherlands)28,29 (Online Supplementary Data). Statistical analyses on clinical correlates are described in the Online Supplementary Data.

Digital droplet polymerase chain reaction assays MEF2C deletions were validated by ddPCR using the QX200TM Droplet DigitalTM PCR System (BioRad, Hercules, CA, USA) and QuantaSoft Analysis Pro software according to the manufacturer's instructions (Online Supplementary Data).

Mutational screening Sanger sequencing of PCR products for MEF2C and KRAS exons (Online Supplementary Table S2) was performed with the ABI-Prism 3500 sequencer (Applied Biosystem, Life Technologies, Foster City, CA, USA) (Online Supplementary Data).

Results Copy number aberration analysis CNA analysis revealed 7.8 aberrations/patient (range: 028), the majority being losses (88%) with only 12% gains, both spanning from whole chromosomes to focal lesions;22,23,30-32 no differences were recorded among trials (Figure 1A).

Gross chromosomal lesions were found in 42% of cases: the majority were losses of chromosome 7 (18.1%), followed by monosomy of chromosome 9 or 9p deletion (9%) and gain of 1q (7.7%) (Figure 1B, Online Supplementary Table S3). Smaller deletions - limited to one to three genes and defined as focal - were found in 56% of cases.

The most frequently deleted region involved the 7p12 cytoband comprising IKZF122,23,33,34 which was detected in 97 cases (84%).

PAX5 was deleted in 43 patients (36.2%), while 37 (31.9%) had deletions of CDKN2A/B. MLLT3, BTG1, BTLA, CD200 and RB1 were deleted in 30, 27, 21, 17 and 16 cases, respectively (25.9%, 23%, 18.1%, 17.2%, 14.6%, and 13.8%) (Figure 1C).

IKZF1 deletions (ΔIKZF1) occurred together with CDKN2A/B and/or PAX5 deletions in 45/97 cases (46.4%) and are defined as ΔIKZF1+CDKN2A and/or PAX5 (Figure 1D): this subset displayed similar lesions to those recently described by Stanulla and colleagues.35 With regard to potential interactions, we found a significant association between IKZF1 and PAX5 deletions (P=0.01), but not with CDKN2A. Multiplex ligation-dependent probe amplification confirmed IKZF1, PAX5, CDKN2A, BTG1, EBF1, ETV6 and RB1 lesions, and allowed evaluation of IKZF1 isoforms.

These isoforms were grouped into four classes:24,36 wildtype, dominant-negative (comprising Δ4-7 cases, 29.8%), haploinsufficient (including all cases harboring a deletion that involves exon 2 - i.e. Δ2-7, Δ2-8, Δ2-3, Δ1-3 - or the whole gene, 57.7%) and miscellaneous (remaining cases, 11.3%).

Table 1. Patients’ clinical features.

Patients (n=116) Gender (male/female) Age, years (range) Median white cell count x 109/L (range) Median hemoglobin g/dL (range) Median platelet count x 109/L (range) Fusion transcript (p190/p210/p190-210)¥ Complete molecular response* yes/no

55/61 51.1 (18.9-89) 25.4 (1.7-597) 9.6 (4-16.3) 50 (4-333) 70/29/16 17/99

*As per protocol definition; ¥information missing for one patient.

haematologica | 2019; 104(2)

Identification of novel lesions CNA analysis highlighted additional genomic lesions (Table 2, Online Supplementary Table S4). We focused in particular on MEF2C and KRAS deletions since these had prognostic significance (see below). MEF2C deletions were detected in 21 patients (18.1%) and differed in size.

According to the length of intron 1-2 losses, deletions were grouped into two categories. One category - detected in 14 cases (67% of MEF2C deleted cases) - was characterized by a larger minimal common region (6.2 Kb) involving introns 1-2 and exon 2 (the first codifying exon), 313

A.L. Fedullo et al. defined ΔMEF2C-long. The other category, detected in seven patients, was smaller (5.4 Kb) and involved only exon 2, and was called ΔMEF2C-short (Figure 2A). ddPCR confirmed MEF2C lesions in all cases. No MEF2C mutations were identified.

KRAS deletions (ΔKRAS) were detected in seven cases (6%); the focal lesion of KRAS started in the 5’ untranslated region and ended in intron 1-2, involving the first noncodifying exon (Figure 2B). The minimal common region consisted of 135 Kb. KRAS was not affected by mutations.

overall survival (62.6% versus 40.2%; P=0.02) (Figure 3D).

The presence of ΔMEF2C-long was associated with a higher rate of CMR achievement (P=0.05); this effect was not influenced by the protocol or the tyrosine kinase inhibitor used (imatinib or dasatinib). Furthermore, ΔMEF2C-long cases were also associated with a signifi-

Table 2. Minimal common region (MCR) of identified focal lesions.

Deleted gene

Impact of known and novel deletions on complete molecular response achievement and disease-free survival

FOCAD CDK6 PTPRD MEF2C BTLA JAK2 ADD3 SLX4IP CD200 HBS1L ATP10A TOX KRAS ARHGAP24 EBF1 LEF1 MDM2 TCF12 ERG

We did not find significant differences between patients with ΔIKZF1 and IKZF1 wild-type cases with regard to achievement of complete molecular response (CMR) or disease-free survival (DFS) (Online Supplementary Figure S2). Further stratification according to IKZF1 isoforms showed that patients with the dominant-negative isoform had a lower DFS rate (23.3%; P=0.039) compared to patients with the other isorforms, particularly wild-type (53.3%; P=0.016) and haploinsufficient cases (40.3%; P=0.015); the DFS rate of the miscellaneous group (34.1%) did not differ significantly from that of the dominant-negative cases (Figure 3A). These differences were not statistically significant in the overall survival analysis (Figure 3B).

We also investigated the outcome of ΔIKZF1+CDKN2A and/or PAX5 cases. The CMR rate did not differ between ΔIKZF1+CDKN2A and/or PAX5 and ΔIKZF1-only cases; contrariwise, ΔIKZF1+CDKN2A and/or PAX5 patients had a significantly worse DFS than ΔIKZF1-only cases (43.3% versus 24.9%; P=0.026) (Figure 3C) and an inferior

N. of patients (%)

MCR (hg19)

29 (25) 24 (20.7) 21 (18.1) 21 (18.1) 21 (18.1) 20 (17.2) 18 (15.5) 17 (14.6) 17 (14.6) 16 (13.7) 14 (12) 8 (6.9) 7 (6) 7 (6) 6 (5.1) 5 (4.3) 5 (4.3) 4 (3.4) 2 (1.7)

chr9: 20685149 - 20759956 chr7: 92456635 - 92266647 chr9: 8153932 - 8854489 chr5:88122179 - 88127630 chr3:112154702 - 112217769 chr9: 5123013 - 5234403 chr10: 111795029 - 111853667 chr20: 10417444 - 10451891 chr3:112054292 - 112217769 chr6: 135375338 - 135418257 chr15: 26055568 - 26103185 chr8:60028851 - 60110235 chr12: 25402194 - 25537468 chr4:86493655 - 86436188 chr5: 158440156 - 158164599 chr4:109034392 - 109084557 chr12:69159076 - 69205287 chr15:57294905 - 57399047 chr21:39772775 - 39788683

A

B

C

D Figure 1. Overall load and incidence of genetic lesions in Philadelphia chromosome-positive acute lymphoblastic leukemia. (A) Distribution of copy number aberrations in the whole cohort and across different protocols. (B) Percentages of gross chromosomal aberrations. (C) Percentages of deletions of known genes in the whole cohort (n=116) and in the different studies analyzed. (D) Heatmap of IKZF1, CDKN2A/B, and PAX5 deletions in the whole cohort.

314

haematologica | 2019; 104(2)

Additional genomic lesions in adult Ph+ ALL

cantly better DFS (64.3% versus 32.1%; P=0.031) (Figure 4A) and overall survival (77.9% versus 48.4%; P=0.036) (Figure 4B).

Lastly, ΔKRAS was more frequently found in patients who obtained a CMR (24% versus 3%; P=0.009), but this finding did not have an impact on DFS.

A

Prognostic impact of known and novel genomic lesions in univariate and multivariate analyses

In univariate analysis, ΔMEF2C-long and ΔKRAS had an impact on CMR achievement, while ΔMEF2C-long and ΔIKZF1+CDKN2A and/or PAX5 influenced DFS (Table 3).

In multivariate analysis for CMR, performed taking into

B

Figure 2. Representation of ΔMEF2C and ΔKRAS. (A) Representation of ΔMEF2C for each patient. Lesions are ordered according to their size: one case had a deletion of the whole gene, one had a deletion that involved only exon 1 spanning from intron 1-2 to the 5’ untranslated region (5’UTR), four had deletions starting from intron 2-3 and ending at 5’-UTR, thereby involving both exons 2 and 1 (the latter being an untranslated exon), 13 had lesions spanning from intron 2-3 to intron 12, therefore involving exon 2 (the first codifying exon), with six of them harboring a longer intron 1-2 deletion. Lastly, two cases had deletions that involved only intron 1-2. The first 14 cases were considered as ΔMEF2C-long and the remaining as ΔMEF2C-short. (B) Representation of ΔKRAS for each patient. Lesions are ordered according to their size: in four cases, the deletion encompassed only KRAS itself, whereas in three it involved the short arm of chromosome 12. INTR: intron; EX: exon; 5’UTR: 5’ untranslated region.

A

B

C

D

Figure 3. Survival probability curves according to IKZF1 status. (A) Disease-free survival and (B) overall survival at 36 months of patients divided according to IKZF1 isoform. (C) Disease-free survival and (D) overall survival at 36 months of ΔIKZF1-only and ΔIKZF1+ CDKN2A and/or PAX5 patients. WT: wild-type; DFS: disease-free survival; OS: overall survival; CR: complete remission.

haematologica | 2019; 104(2)

315

A.L. Fedullo et al.

account white blood cell count, age, tyrosine kinase inhibitor use and the genomic lesions described above, the only factor that retained statistical significance was ΔKRAS (P=0.01); a trend was noted for ΔMEF2C-long deletions (P=0.075) (Table 3).

In multivariate analysis for DFS, considering ΔMEF2Clong, ΔIKZF1+CDKN2A and/or PAX5, white blood cell count and CMR as variables, the factors that had a negative impact were ΔMEF2C-long (P=0.057) and white blood cell count (P=0.05), while a trend towards a worse DFS was observed for ΔIKZF1+CDKN2A and/or PAX5 (P=0.089) (Table 3). HSCT did not affect the prognostic role of the above-mentioned lesions.

Discussion The management of adults with Ph+ ALL currently relies on the use of first,8-10,13-16 second11,12 and third37 generation tyrosine kinase inhibitors, either alone9-12 or in combination with chemotherapy,8,13-16,37 followed - if feasible and necessary - by HSCT. These approaches have greatly improved the outcome of Ph+ ALL: nowadays, virtually all patients - independent of age - achieve a complete hematologic remission, coupled to a CMR in a variable proportion of cases. Nonetheless, in all reported studies the long-term outcome is in the range of 50% at 5 years; thus, additional prognosticators capable of better stratify-

Table 3. Summary of univariate and multivariate analyses for complete molecular response and disease-free survival for the factors identified.

MEF2C deletions KRAS deletions White blood cell count Age Imatinib vs. dasatinib Fusion protein (p190 vs. p210 and p190/210) ΔIKZF1+ CDKN2A and/or PAX5

MEF2C deletions ΔIKZF1+ CDKN2A and/or PAX5 White blood cell count Age CMR Fusion protein (p190 vs. p210 and p190/210) Imatinib vs. dasatinib Allogeneic transplant

Univariate analysis for CMR OR (95% CI)

P value

Multivariate analysis for CMR OR (95%CI)

P value

0.288 (0.082, 1007) 0.12 (0.024, 0.601) 0.986 (0.969, 1.003) 1.026 (0.99, 1.063) 0.267 (0.057, 1.248)

0.051 0.009 0.1 0.16 0.093

0.259 (0.058, 1.146) 0.068 (0.009, 0.529) 0.986 (0.966, 1.007) 1.028 (0.985, 1.072) 0.296 (0.054, 1.615)

0.075 0.01 0.188 0.205 0.159

1.247 (0.421, 3.693) 1.600 (0.599, 4.581)

0.691 0.381

Univariate analysis for DFS HR (95% CI)

P value

Multivariate analysis for DFS HR (95%CI)

P value

0.359 (0.144, 0.891) 1.834 (1.148, 2.929) 1.002 (1, 1.004) 1.001 (0.986, 1.017) 0.423 (0.181-0.987)

0.027 0.011 0.065 0.866 0.046

0.417 (0.169-1.028) 1.608 (0.930, 2.781) 1.003 (1, 1.006) 1.01 (0.995-1.028) 0.402 (0.167-0.969)

0.057 0.089 0.050 0.180 0.042

0.939 (0.582, 1.515) 1.305 (0.807, 2.109) 0.682 (0.362, 1.284)

0.797 0.277 0.23

CMR: complete molecular response; OR: odds ratio; 95% CI: 95% confidence interval; DFS: disease-free survival; HR: hazard ratio.

A

B

Figure 4. Survival probability curves according to MEF2C status. (A) Disease-free survival at 36 months of ΔMEF2C versus MEF2C wild-type patients. (B) Overall survival at 36 months of ΔMEF2C versus MEF2C wild-type patients. WT: wild-type; DFS: disease-free survival; OS: overall survival; CR: complete remission.

316

haematologica | 2019; 104(2)

Additional genomic lesions in adult Ph+ ALL

ing patients into high- and low-risk categories are urgently needed to further optimize treatment. Moreover, another unsolved issue is whether all eligible patients should undergo HSCT,7,17 a procedure still associated with shortand long-term side effects, as well as treatment-related mortality. This is particularly important for patients who obtain a CMR. To address these issues we sought to identify additional genomic lesions with prognostic significance in adult Ph+ ALL using high density Cytoscan arrays. We found that adult Ph+ ALL patients carried an average of 7.8 aberrations each, with deletions outnumbering gains, in line with other ALL subsets.22,30,38,39 Macro-aberrations were identified in 48% of cases and micro-aberrations in the majority of patients: among the latter, the most frequent was ΔIKZF1, which was detected in 84% of cases. ΔIKZF1 has been regarded as a poor prognostic marker in both childhood and adult ALL.34,36,39-41 This finding was not, however, confirmed in our cohort: in fact, the presence of ΔIKZF1 alone was not associated with a worse DFS. A further evaluation of the various IKZF1 isoforms showed that only the dominant-negative genotype was deleterious for outcome. In addition, patients with ΔIKZF1+CDKN2A and/or PAX5, accounting for almost half the ΔIKZF1 cases, experienced a significantly inferior DFS (P=0.005) and overall survival (P=0.02), in line with previous reports on ALL in general.28,29,36,39,42,43 ΔIKZF1+CDKN2A and/or PAX5 also had a prognostic impact in multivariate analysis; survival analysis was carried out merging all cases enrolled in the different trials together in order to gain statistical significance.

Recently, studies have been focused on the presence of additional karyotypic aberrations in Ph+ ALL.44-48 These studies have highlighted that a high percentage of Ph+ ALL cases (60-80%) harbor additional chromosomal abnormalities, with the most frequent aberrations involving chromosomes 7, 9, and 14. Patients with additional abnormalities, particularly loss of 9/9p and/or CDKN2A, have a worse outcome. These results point to the importance of screening for other molecular markers, and not only IKZF1, in agreement with our findings on ΔIKZF1+CDKN2A and/or PAX5. At variance from these reports, our study also identified novel lesions that had a favorable impact on

References 1. Nowell PC, Hungerford DA. Chromosome studies on normal and leukemic human leukocytes. J Natl Cancer Inst. 1960;25:85109.

2. Chiaretti S, Vitale A, Cazzaniga G, et al.

Clinico-biological features of 5202 patients with acute lymphoblastic leukemia enrolled in the Italian AIEOP and GIMEMA protocols and stratified in age cohorts.

Haematologica. 2013;98(11):1702-1710.

3. Dombret H, Gabert J, Boiron JM, et al.

Outcome of treatment in adults Philadelphia chromosome-positive acute lymphoblastic leukemia-results of the prospective multicenter LALA-94 trial. Blood. 2002;100(7): 2357-2366.

4. Gleissner B, Gökbuget N, Bartram CR, et al.

Leading prognostic relevance of the BCRABL translocation in adult acute B-lineage

haematologica | 2019; 104(2)

outcome. Among these, it is worth mentioning ΔMEF2C, which occurred in 18.1% of patients and was of two sizes, a long deletion, encompassing introns 1-2 and exon 2, and a second, smaller one, involving only exon 2. MEF2C is a transcription factor involved in B-cell survival and proliferation whose overexpression is associated with an unfavorable prognosis in T-ALL and acute myeloid leukemia.49-52 In our study, the presence of ΔMEF2C-long was associated with achievement of a CMR (P=0.05) and with a significantly better DFS compared to the remaining cases (P=0.031) also in a multivariate model; as for IKZF1 deletions, survival analysis was performed merging the whole cohort because of the sample sizes. ΔMEF2C-long was widely distributed among cases, with no association with white blood cell count, age, type of fusion protein or additional deletions. To our knowledge, this is the first report that correlates ΔMEF2C-long with prognosis in Ph+ ALL: Martinelli et al.40 and Mullighan et al.22,41 described ΔMEF2C in Ph+ ALL, but did not demonstrate a correlation with outcome. Finally, ΔKRAS was associated with a higher rate of CMR achievement upon induction (P=0.01), but not with a better DFS. In conclusion, we show that additional genetic lesions can be found at presentation in adult Ph+ ALL patients and that these lesions have prognostic significance, with the IKZF1 dominant-negative isoform and ΔIKZF1+CDKN2A and/or PAX5 negatively affecting outcome, and ΔMEF2C and ΔKRAS being instead associated with a more favorable prognosis. Screening for these genetic lesions should, therefore, be performed at the time of diagnosis for a more refined prognostic stratification, and for a more personalized and tailored management of Ph+ ALL patients.

Acknowledgments The authors thank Associazione Italiana per la Ricerca sul Cancro (AIRC), Special Program Molecular Clinical Oncology-Extension program, 5 x 1000 (10007), Milan (Italy) for funding RF; Finanziamento per l’avvio alla ricerca 2015 (Sapienza University of Rome) for funding MM; Finanziamento Medi Progetti Universitari 2015 for funding SC (Sapienza University of Rome); and Fondazione Le Molinette Onlus, Turin (Italy).

lymphoblastic leukemia: a prospective study of the German Multicenter Trial Group and confirmed polymerase chain reaction analysis. Blood. 2002;99(5):15361543.

5. Pullarkat V, Slovak ML, Kopecky KJ, et al.

Impact of cytogenetics on the outcome of adult acute lymphoblastic leukemia: results of Southwest Oncology Group 9400 study.

Blood. 2008;111(5):2563-2572.

6. Hunault M, Harousseau JL, Delain M, et al.

Better outcome of adult acute lymphoblastic leukemia after early genoidentical allogeneic bone marrow transplantation (BMT) than after late high-dose therapy and autologous BMT: a GOELAMS trial. Blood.

2004;104(10):3028-3037.

7. Patel JN, Druhan LJ. Genetic effects on hematopoietic stem cell transplant prognosis and outcomes, more than just histocompatibility. Biol Blood Marrow Transplant.

2017;23(8):1227-1228.

8. de Labarthe A, Rousselot P, Huguet-Rigal F, et al. Imatinib combined with induction or consolidation chemotherapy in patients with de novo Philadelphia chromosomepositive acute lymphoblastic leukemia: results of the GRAAPH-2003 study. Blood.

2007;109(4):1408-1413.

9. Vignetti M, Fazi P, Cimino G, et al. Imatinib plus steroids induces complete remissions and prolonged survival in elderly Philadelphia chromosome-positive patients with acute lymphoblastic leukemia without additional chemotherapy: results of the Gruppo Italiano Malattie Ematologiche dell'Adulto (GIMEMA) LAL0201-B protocol. Blood. 2007;109(9):3676-3678.

10. Chiaretti S, Vitale A, Vignetti M, et al. A sequential approach with imatinib, chemotherapy and transplant for adult Ph+ acute lymphoblastic leukemia. Final results of the GIMEMA LAL 0904 study.

Haematologica. 2016;101(12):1544-1552.

317

A.L. Fedullo et al.

11. Foà R, Vitale A, Vignetti M, et al. Dasatinib as first-line treatment for adult patients with Philadelphia chromosome-positive acute lymphoblastic leukemia. Blood. 2011;118 (25):6521-6528.

12. Chiaretti S, Vitale A, Elia L, et al. Multicenter Total Therapy GIMEMA LAL 1509 protocol for de novo adult Ph+ acute lymphoblastic leukemia (ALL) patients. Updated results and refined genetic-based prognostic stratification. Blood. 2015;126 (23):81.

13. Bassan R, Rossi G, Pogliani EM, et al.

Chemotherapy-phased imatinib pulses improve long-term outcome of adult patients with Philadelphia chromosomepositive acute lymphoblastic leukemia: Northern Italy Leukemia Group protocol 09/00. J Clin Oncol. 2010;28(22):3644-3652.

14. Ribera JM, García O, Montesinos P, et al.

Treatment of young patients with Philadelphia chromosome-positive acute lymphoblastic leukaemia using increased dose of imatinib and deintensified chemotherapy before allogeneic stem cell transplantation. Br J Haematol. 2012;159 (1):78-81.

15. Fielding AK, Rowe JM, Buck G, et al.

UKALLXII/ECOG2993: addition of imatinib to a standard treatment regimen enhances long-term outcomes in Philadelphia positive acute lymphoblastic leukemia. Blood.

2014;123(6):843-850.

16. Chalandon Y, Thomas X, Hayette S, et al.

Randomized study of reduced-intensity chemotherapy combined with imatinib in adults with Ph-positive acute lymphoblastic leukemia. Blood. 2015;125(24):3711-3719.

17. Litzow MR, Fielding AK, Luger SM, et al.

The evolving role of chemotherapy and hematopoietic cell transplants in Ph-positive acute lymphoblastic leukemia in adults.

Bone Marrow Transplant. 2017;52(12):15921598.

18. Cimino G, Pane F, Elia L, et al. The role of BCR/ABL isoforms in the presentation and outcome of patients with Philadelphia-positive acute lymphoblastic leukemia: a sevenyear update of the GIMEMA 0496 trial.

Haematologica. 2006;91(3):377-380.

19. Lee S, Kim DW, Cho BS, et al. Impact of minimal residual disease kinetics during imatinib-based treatment on transplantation outcome in Philadelphia chromosome-positive acute lymphoblastic leukemia.

Leukemia. 2012;26(11):2367-2374.

20. Ravandi F, Jorgensen JL, Thomas DA, et al.

Detection of MRD may predict the outcome of patients with Philadelphia chromosomepositive ALL treated with tyrosine kinase inhibitors plus chemotherapy. Blood.

2013;122(7):1214-1221.

21. Martinelli G, Iacobucci I, Storlazzi CT, et al.

IKZF1 (Ikaros) deletions in BCR-ABL1-positive acute lymphoblastic leukemia are associated with short disease-free survival and high rate of cumulative incidence of relapse: a GIMEMA AL WP report. J Clin Oncol.

2009;27(31):5202-5207.

22. Mullighan CG, Miller CB, Radtke I, et al.

BCR-ABL1 lymphoblastic leukaemia is characterized by the deletion of Ikaros. Nature.

2008;453(7191):110-114.

23. Mullighan CG. Genomic profiling of B-progenitor acute lymphoblastic leukemia. Best Pract Res Clin Haematol. 2011;24(4):489503.

24. van der Veer A, Zaliova M, Mottadelli F, et al. IKZF1 status as a prognostic feature in BCR-ABL1-positive childhood ALL. Blood.

2014;123(11):1691-1698.

25. DeBoer R, Koval G, Mulkey F, et al. Clinical impact of ABL1 kinase domain mutations

318

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

and IKZF1 deletion in adults under age 60 with Philadelphia chromosome-positive (Ph+) acute lymphoblastic leukemia (ALL): molecular analysis of CALGB (Alliance) 10001 and 9665. Leuk Lymphoma.

2016;57(10):2298-2306.

Soverini S, Vitale A, Poerio A, et al.

Philadelphia-positive acute lymphoblastic leukemia patients already harbor BCR-ABL kinase domain mutations at low levels at the time of diagnosis. Haematologica.

2011;96(4):552-557.

Soverini S, De Benedittis C, Machova Polakova K, et al. Unraveling the complexity of tyrosine kinase inhibitor-resistant populations by ultra-deep sequencing of the BCRABL kinase domain. Blood. 2013;122(9): 1634-1648.

Messina M, Chiaretti S, Fedullo AL, et al.

Clinical significance of recurrent copy number aberrations in B-lineage acute lymphoblastic leukaemia without recurrent fusion genes across age cohorts. Br J Haematol. 2017;178(4):583-587.

Moorman AV, Enshaei A, Schwab C, et al. A novel integrated cytogenetic and genomic classification refines risk stratification in pediatric acute lymphoblastic leukemia.

Blood. 2014;124(9):1434-1444.

Mullighan CG, Goorha S, Radtke I, et al.

Genome-wide analysis of genetic alterations in acute lymphoblastic leukaemia. Nature.

2007;446(7137):758-764.

Kuiper RP, Schoenmakers EF, van Reijmersdal SV, et al. High-resolution genomic profiling of childhood ALL reveals novel recurrent genetic lesions affecting pathways involved in lymphocyte differentiation and cell cycle progression. Leukemia.

2007;21(6):1258-1266.

Safavi S, Hansson M, Karlsson K, et al.

Novel gene targets detected by genomic profiling in a consecutive series of 126 adults with acute lymphoblastic leukemia.

Haematologica. 2015;100(1):55-61.

Iacobucci I, Storlazzi CT, Cilloni D, et al.

Identification and molecular characterization of recurrent genomic deletions on 7p12 in the IKZF1 gene in a large cohort of BCRABL1-positive acute lymphoblastic leukemia patients: on behalf of Gruppo Italiano Malattie EMatologiche dell’Adulto Acute Leukemia Working Party (GIMEMA AL WP). Blood. 2009;114(10):2159-2167.

Mullighan CG, Su X, Zhang J, et al. Deletion of IKZF1 and prognosis in acute lymphoblastic leukemia. N Engl J Med. 2009; 360(5):470-480.

Stanulla M, Dagdan E, Zaliova M, et al.

IKZF1plus defines a new minimal residual disease-dependent very-poor prognostic profile in pediatric B-cell precursor acute lymphoblastic leukemia. J Clin Oncol. 2018; 36(12):1240-1249.

Boer JM, van der Veer A, Rizopoulos D, et al.

Prognostic value of rare IKZF1 deletion in childhood B-cell precursor acute lymphoblastic leukemia: an international collaborative study. Leukemia. 2016;30(1):32-38.

Jabbour E, Kantarjian H, Ravandi F, et al.

Combination of hyper-CVAD with ponatinib as first-line therapy for patients with Philadelphia chromosome-positive acute lymphoblastic leukaemia: a single-centre, phase 2 study.

Lancet Oncol.

2015;16(15):1547-1555.

Messina M, Chiaretti S, Wang J, et al.

Prognostic and therapeutic role of targetable lesions in B-lineage acute lymphoblastic leukemia without recurrent fusion genes.

Oncotarget. 2016;7(12):13886-13901.

Ribera J, Morgades M, Zamora L, et al.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

Prognostic significance of copy number alterations in adolescent and adult patients with precursor B acute lymphoblastic leukemia enrolled in PETHEMA protocols.

Cancer. 2015;121(21):3809-3817.

Martinelli G, Iacobucci I, Papayannidis C, et al. New targets for Ph+ leukaemia therapy.

Best Pract Res Clin Haematol. 2009;22(3): 445-454.

Mullighan CG, Downing JR. Genome-wide profiling of genetic alterations in acute lymphoblastic leukemia: recent insights and future directions. Leukemia. 2009;23(7): 1209-1218.

Xu N, Li YL, Li X, et al. Correlation between deletion of the CDKN2 gene and tyrosine kinase inhibitor resistance in adult Philadelphia chromosome-positive acute lymphoblastic leukemia. J Hematol Oncol.

2016;9:40.

Pfeifer H, Raum K, Markovic S, et al.

Genomic CDKN2A/2B deletions in adult Ph+ ALL are adverse despite allogeneic stem cell transplantation. Blood. 2018;131(13): 1464-1475.

Heerema NA, Harbott J, Galimberti S, et al.

Secondary cytogenetic aberrations in childhood Philadelphia chromosome positive acute lymphoblastic leukemia are nonrandom and may be associated with outcome.

Leukemia. 2004;18(4):693-702.

Li Y, Qiu L, Zou D, et al. Additional chromosomal abnormalities and their prognostic significance in adult Philadelphia-positive acute lymphoblastic leukemia: with or without imatinib in chemotherapy. Ann Hematol. 2009;88(11):1069-1077.

Short NJ, Kantarjian HM, Sasaki K,et al. Poor outcomes associated with +der(22)t(9;22) and -9/9p in patients with Philadelphia chromosome-positive acute lymphoblastic leukemia receiving chemotherapy plus a tyrosine kinase inhibitor. Am J Hematol.

2017;92(3):238-243.

Seol CA, Cho YU, Jang S, et al. Prognostic significance of recurrent additional chromosomal abnormalities in adult patients with Philadelphia chromosome-positive acute lymphoblastic leukemia. Cancer Genet.

2017;216-217:29-36.

Motlló C, Ribera JM, Morgades M, et al.

Frequency and prognostic significance of additional cytogenetic abnormalities to the Philadelphia chromosome in young and older adults with acute lymphoblastic leukemia. Leuk Lymphoma. 2018;59(1):146154.

Homminga I, Pieters R, Langerak AW, et al.

Integrated transcript and genome analyses reveal NKX2-1 and MEF2C as potential oncogenes in T cell acute lymphoblastic leukemia. Cancer Cell. 2011;19(4):484-497.

Zuurbier L, Gutierrez A, Mullighan CG, et al. Immature MEF2C-dysregulated T-cell leukemia patients have an early T-cell precursor acute lymphoblastic leukemia gene signature and typically have non-rearranged T-cell receptors. Haematologica. 2014;99(1): 94-102.

Laszlo GS, Alonzo TA, Gudgeon CJ, et al.

High expression of myocyte enhancer factor 2C (MEF2C) is associated with adverse-risk features and poor outcome in pediatric acute myeloid leukemia: a report from the Children's Oncology Group. J Hematol Oncol. 2015;8:115.

Colomer-Lahiguera S, Pisecker M, König M, et al. MEF2C-dysregulated pediatric T-cell acute lymphoblastic leukemia is associated with CDKN1B deletions and a poor response to glucocorticoid therapy. Leuk Lymphoma. 2017;58(12):2895-2904.

haematologica | 2019; 104(2)

ARTICLE

Acute Lymphoblastic Leukemia

Hypersensitivity reactions to asparaginase in mice are mediated by anti-asparaginase IgE and IgG and the immunoglobulin receptors FcεRI and FcγRIII

Ferrata Storti Foundation

Sanjay Rathod,1 Manda Ramsey,1 Mary V. Relling,2 Fred D. Finkelman3 and Christian A. Fernandez1

Center for Pharmacogenetics and Department of Pharmaceutical Sciences, University of Pittsburgh, PA; 2Department of Pharmaceutical Sciences, St. Jude Children’s Research Hospital, Memphis, TN; 3Department of Internal Medicine, Division of Immunology, Allergy and Rheumatology, University of Cincinnati College of Medicine and the Division of Immunobiology, Cincinnati Children's Hospital Medical Center, OH, USA. 1

Haematologica 2019 Volume 104(2):319-329

ABSTRACT

A

sparaginase is an important drug for the treatment of leukemias.

However, anti-asparaginase antibodies often develop, which can decrease asparaginase drug levels and increase the risk of relapse. The aim of this study is to identify the immunoglobulin isotypes and receptors responsible for asparaginase hypersensitivities.

Mice immunized with asparaginase developed anti-asparaginase IgG1 and IgE antibodies, and challenging the sensitized mice with asparaginase induced severe hypersensitivity reactions. Flow cytometry analysis indicated that macrophages/monocytes, neutrophils, and basophils bind asparaginase ex vivo through FcγRIII. In contrast, asparaginase binding to basophils was dependent on FcγRIII and IgE. Consistent with the asparaginase binding data, basophil activation by asparaginase occurred via both IgG/FcγRIII and IgE/FcεRI. Depleting >95% of B cells suppressed IgG but not IgE-dependent hypersensitivity, while depleting CD4+ T cells provided complete protection. Combined treatment with either anti-IgE mAb plus a platelet-activating factor receptor antagonist or anti-FcγRIII mAb plus a H1 receptor antagonist suppressed asparaginase hypersensitivity. The observations indicate that asparaginase hypersensitivity is mediated by antigen-specific IgG and/or IgE through the immunoglobulin receptors FcγRIII and FcεRI, respectively. Provided that these results apply to humans, they emphasize the importance of monitoring both IgE- and IgG-mediated asparaginase hypersensitivities in patients receiving this agent.

Correspondence: chf63@pitt.edu

Received: June 8, 2018.

Accepted: September 20, 2018.

Pre-published: September 20, 2018. doi:10.3324/haematol.2018.199448

Introduction L-Asparaginase (ASNase) is given repeatedly during treatment regimens for acute lymphoblastic leukemia (ALL). The non-human enzyme is derived from bacteria and inhibits leukemic cell proliferation by depleting asparagine.1 The most common adverse reaction of ASNase in children results from the production of anti-ASNase antibodies (seen in up to 70% of patients) and the onset of clinical hypersensitivity reactions during treatment.2-7 ASNase-mediated hypersensitivity can occur in 3075% of patients receiving native E. coli ASNase3,8-10 and typically manifest as urticaria, angioedema, bronchospasm, dyspnea, and anaphylaxis.11 Typically, if a patient develops a hypersensitivity reaction to first-line PEG-ASNase, a substitution with Erwinia ASNase is recommended; a subsequent reaction to Erwinia ASNase may necessitate discontinuing ASNase therapy.12 In addition, the development of antiASNase antibodies can increase the risk of relapse by neutralizing ASNase in vivo.13 ASNase-mediated hypersensitivity during ALL treatment is most common upon drug re-exposure,2 this suggests that patients are sensitized to the agent earlier in therapy. However, the mechanism of ASNase hypersensitivity is not clear, as some patients develop reactions in the absence of detectable anti-ASNase IgG antibody.

Moreover, many patients who have circulating anti-ASNase IgG never develop a haematologica | 2019; 104(2)

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/319 ©2019 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.

319

S. Rathod et al. hypersensitivity reaction.2,7 Interestingly, few studies have assessed anti-ASNase IgE levels in patients or have suggested a role of anti-ASNase IgE in the ASNase hypersensitivity reaction,14-16 likely due to a lack of methods available to detect anti-ASNase IgE in the presence of high antiASNase IgG levels.

Pharmacogenomic studies of the immune response to ASNase have identified risk variants in genes involved in antigen presentation17 or T-cell activation,18 supporting the importance to B/T-cell activation during ASNase sensitization and in the production of anti-ASNase antibodies.

Anti-ASNase IgG can elicit a hypersensitivity reaction by forming an immune complex with ASNase, binding to the Fcγ receptor of immune cells (e.g., basophils, mast cells, neutrophils, and/or macrophages), and resulting in the release of platelet activating factor (PAF).19 Alternatively, cells expressing FcεRI, such as mast cells and basophils, can bind anti-ASNase IgE during sensitization and mediate a hypersensitivity upon antigen exposure via the release of histamine.20 Recently, a murine model of ASNase hypersensitivity has been described that recapitulates many of the features of clinical ASNase-mediated hypersensitivity.21 Studies using this model indicate that pretreatment with the antihistamine, triprolidine, and the PAF receptor antagonist, CV-6209, can strongly mitigate the onset of ASNase hypersensitivity, suggesting that both histamine and PAF release play a role in the immune response to ASNase.

The current study uses the murine model of ASNase hypersensitivity to identify the immune cells required for ASNase sensitization and the immunoglobulin isotypes and receptors responsible for the onset of hypersensitivity.

Our results indicate that anti-ASNase IgE plays an important role in ASNase-induced hypersensitivities and that the binding of ASNase to basophils may be predictive of hypersensitivity. In accordance with the importance of binding to basophils, we show that ASNase activates basophils through both FcγRIII- and IgE-dependent mechanisms. Our results suggest that both anti-ASNase IgG and IgE play important roles in the development of ASNase hypersensitivity.

Methods ASNase sensitization 8-week-old, female, C57BL/6 mice were injected intraperitoneally (IP) with 10 mg of E. coli ASNase formulated with 1 mg of aluminum hydroxide adjuvant, on days 0 and 14, as previously described.21 ASNase hypersensitivity reactions were induced in sensitized mice by challenging with a 100 mg IV dose of E. coli ASNase on Day 24 of treatment. All experiments with mice were reviewed and conducted under approved protocol by the University of Pittsburgh Institutional Animal Cares and Use Committee.

Detection of anti-ASNase IgE by flow cytometry

Anti-IgE-biotin (Biolegend, USA) at 1 mg/mL was bound to 3x106 streptavidin-coupled 6-8 mm diameter magnetic particles (Spherotech, USA). Plasma samples diluted to 1:100 in PBS were added to anti-IgE-coated beads for 3060 minutes at room temperature, washed with PBST, and stained with labeled ASNase at 1 IU/mL. The stained samples were analyzed by flow cytometry for ASNase fluorescence.

320

Basophilic activation test (BAT) BAT was performed as previously described.22,23 Briefly, 50 mL of blood was incubated for 15 min at 37°C and further stimulated with EM-95 at 300 ng/mL, 2.4G2 at 300 ng/mL, ASNase at 1 IU/mL, or medium (as a negative control). Samples were further incubated for 2 h at 37°C in 5% CO2, quenched by adding 20 mM EDTA, and incubated on ice for 10 minutes. Cells were blocked with 15% HS in PBS for 30 minutes on ice, washed, and stained with anti-IgE, anti-CD49b, anti-CD200R3, and anti-CD200R1 mAbs for 30-60 minutes at 4°C. The cells were then lysed, washed with 1% BSA in PBS, and analyzed by flow cytometry. The percent change in CD200R1 expression is equal to the mean experimental expression of CD200R1 minus that of the mean expression of the sample stimulated with medium, divided by the mean expression of the sample stimulated with medium. Similarly, the percent change in CD200R3 is the mean expression of the sample stimulated with medium minus the mean experimental expression of CD200R3, divided by the mean expression of the sample stimulated with medium.

In vivo immune cell depletion Anti-CD4 mAb or anti-CD19 mAb were injected IP in mice at 200 mg/mouse three days before each sensitization dose of ASNase. Cell depletions were confirmed by flow cytometry, as described above, where different mAb clones targeting CD19 or CD4 were used for cell depletion and staining. Mice were challenged with E. coli ASNase on Day 24, as described above.

In vivo blocking of ASNase-induced hypersensitivity reactions with anti-IgE or anti-FcγRIIB/III mAb To prevent IgE- or IgG-mediated hypersensitivities, a single 100 µg dose of anti-IgE (EM-95)20 or 500 µg of antiFcγRIIB/III mAb (2.4G2)24 was administered IP 24 hours before the ASNase challenge. Pretreatment medication, as a single agent or in combination, was given before the ASNase challenge in a total volume of 150 mL per injection. The doses of each drug used are based on previous studies.21 66 mg of CV-6209 (PAF receptor antagonist) was given 5 minutes before challenges via IV injection, and 200 mg of antihistamine (triprolidine, an H1 receptor antagonist) was given IP 30 minutes before the ASNase challenge. Additional Methods are included in the Online Supplementary Material.

Results Anti-ASNase IgE plays a role in ASNase-specific recognition The onset of ASNase-mediated hypersensitivity requires a humoral immune response to the agent after its initial use during induction therapy and antigen-specific recognition upon drug re-exposure after sensitization.

Therefore, as expected, staining peripheral blood cells from naïve mice with fluorochrome-labeled ASNase (Figure 1A) revealed no ASNase positive neutrophils, T cells, basophils, or B cells. Yet, consistent with the perceived mechanism of ASNase clearance involving macrophages of the reticuloendothelial system (RES),25 a small percentage of macrophages/monocytes were ASNase positive (Figure 1A). Mice were given 100 mg of fluorochrome-labeled ASNase IV to verify its uptake by macrophages/monocytes. The labeled ASNase was primahaematologica | 2019; 104(2)

ASNase hypersensitivities are IgG/IgE dependent

rily detected in cells obtained from the liver (73.6%), blood (11.4%), and spleen (5.0%) 8 h after administration (Online Supplementary Figure S1A). Among CD45+ leukocytes, ASNase accumulates primarily in macrophages/monocytes of the liver and blood (95.1%, and 93.7%, respectively; Online Supplementary Figure S1A and B), to a lesser extent in neutrophils (1.1% and 2.1%, respectively), but not in other immune cells. Supporting the role of macrophages in ASNase clearance, the uptake of ASNase was assessed using a murine macrophage cell line (i.e., RAW 264.7 cells). ASNase positive RAW cells were detected after 5 minutes of incubation with ASNase

and the percentage of ASNase+ RAW cells plateaued after about 2 h of incubation with the drug (Online Supplementary Figure S2A-B). Therefore, it is likely that the binding of ASNase to naĂŻve macrophages/monocytes is associated with drug clearance rather than antigen-specific recognition.

After sensitizing mice to ASNase, as described previously21 (Online Supplementary Figure S3A) and staining blood samples with ASNase and immune cell surface markers, we found that the drug was bound ex vivo by B cells (53.3%), neutrophils (14.5%), macrophages/monocytes (32.9%), and basophils (69.5%), but not T cells (Figure 1B).

A

B

C

D

E

F

Figure 1. ASNase is recognized ex vivo by B cells, neutrophils, macrophages/monocytes, and basophils after sensitization. (A) Peripheral blood cells were collected from naĂŻve mice, cultured with labeled ASNase for 30 minutes, and analyzed for ASNase positive immune cells by flow cytometry. (B) The ex vivo ASNase-specific recognition by B cells, neutrophils, macrophages/monocytes, basophils, and T cells of the blood were analyzed within CD45+ populations by flow cytometry of sensitized (red data points) and non-sensitized (green data points) mice on Day 23 of the sensitization protocol. (C) ASNase binding to total leukocytes (CD45+) and (D) anti-ASNase IgG antibodies were measured throughout the sensitization protocol. Anti-ASNase IgE antibodies were measured by (E) ELISA and (F) flow cytometry using plasma samples collected on Day 23 of the sensitization protocol. A total of 5 to 20 mice were included in each analysis, as indicated, and P value significance is indicated as * for P<0.05, ** for P<0.01, *** for P<1x10-3, and **** for P<1x10-4.

haematologica | 2019; 104(2)

321

S. Rathod et al.

However, the binding of ASNase to immune cells might have depended on anti-ASNase IgG, primarily the antiASNase IgG1 subclass (Online Supplementary Figure S4), because CD45+ASNase+ cells were solely detected on Day 23 of our protocol (Figure 1C, P<1x10-4), which is when plasma anti-ASNase IgG antibody levels became elevated relative to naïve mice (Figure 1D, P<1x10-4).

Basophils express both the high-affinity IgE receptor, FcεRI, and the low-affinity IgG receptor, FcγRIII, and, therefore, can bind antigen via cell-associated IgE or antigen-specific IgG (i.e., after immune complex formation).

Suggesting that anti-ASNase IgE may play a role in the detected binding to basophils, plasma anti-ASNase IgE antibodies were elevated in the plasma samples of sensitized mice on Day 23 relative to controls by ELISA (Figure 1E). Due to the possible interference of anti-ASNase IgG during the detection of anti-ASNase IgE by ELISA,26 the presence of anti-ASNase IgE in plasma samples was confirmed by capturing IgE using anti-IgE-coated polystyrene beads and staining with labeled ASNase (Figure 1F). Based on the antigen-specific antibody isotypes detected, it is possible that recognition or binding of free ASNase to immune cells after sensitization can occur through cell associated-IgE among cells expressing FcεRI (e.g., basophils, Figure 1B) or via the binding of ASNase-IgG immune complexes to cells expressing FcγRIII (e.g., neutrophils, macrophages/monocytes, basophils, Figure 1B).

Supporting this proposed mechanism of recognition, the binding of ASNase to the basophils of sensitized mice is decreased by blocking antibodies targeting the FcγRIII receptor (2.4G2, Figure 2, P< 1x10-4) or IgE (EM-95, Figure 2, P<1x10-4). The binding of ASNase to the basophils of

sensitized mice is further decreased when both antibodies are used in combination rather than singly (Figure 2, P<1x10-3). Thus, basophils can bind ASNase immune complexes and free ASNase, and both can occur simultaneously. In contrast, the binding of ASNase to IgE- leukocytes of sensitized mice is decreased after blocking with 2.4G2 (Figure 2, P<1x10-3) but binding is not affected by EM-95 (Figure 2, P>0.05). Consistent with requiring immune complex formation prior to ASNase binding to cells expressing the FcγRIII receptor, if soluble IgG is removed during erythrocyte lysing before staining cells, then the binding of ASNase to basophils is decreased relative to non-lysed samples (43% decrease, Figure 2, P<1x10-4).

Blocking lysed cells with 2.4G2 has little to no effect on the binding of ASNase to basophils; in contrast, blocking with EM-95 strongly inhibited binding (Figure 2, P<1x10-4). The results suggest that ASNase binding to immune cells depends on the surface expression of FcγRIII and an IgE receptor, most likely, FcεRI.

The frequencies of blood basophils, B cells, and CD4+ Tregs increase during sensitization to ASNase Upon challenge with ASNase, immune cell binding to ASNase in sensitized mice may lead to a hypersensitivity reaction, yet other cells that do not recognize or bind to ASNase likely play a role during sensitization to the drug.

To identify other cells that may be involved in sensitization, we measured the frequency of B cells, CD4+/8+ T cells, CD4+/8+ Tregs, neutrophils, macrophages/monocytes, and basophils in blood within the CD45+ leukocytes population at five different time points during sensitization (Figure 3A-D; Online Supplementary Figure S5A-D).

Figure 2. Ex vivo ASNase binding to basophils is dependent on FcγRIII and FcεRI. Ex vivo ASNase binding to basophils after immunoglobulin receptor blocking with 2.4G2 (anti-FcγRIIB/III mAB) and/or EM-95 (anti-IgE mAB) suggests that the binding is dependent on both FcγRIII and FcεRI (n = 10). Furthermore, lysing cells and removing soluble IgG antibodies before measuring ASNase binding reduces the frequency of ASNase+ cells relative to non-lysed cells (P<1x10-4). The binding of ASNase to basophils is attenuated by EM-95 but not 2.4G2 after removing soluble IgG. In contrast, ASNase binding to CD45+IgE- cells decreases after blocking with 2.4G2 but not after blocking with EM-95 (n=10). P value significance is indicated as * for P<0.05, ** for P<0.01, *** for P<1x10-3, and **** for P<1x10-4.

322

haematologica | 2019; 104(2)

ASNase hypersensitivities are IgG/IgE dependent

The frequency of macrophages/monocytes increased by approximately 5-10% after each ASNase dose (2 h and Day 14; Figure 3A), possibly reflecting the role of macrophages/monocytes on ASNase clearance; however, no other change in immune cell phenotype was detected among samples collected 2 h after the first sensitization dose or on Days 7 or 14 of the sensitization protocol (Online Supplementary Figure S5A-D). In contrast, frequencies of CD4+ Tregs, B cells, and basophils in the blood increased by Day 23 relative to naïve mice (Figure 3B-D, respectively). The data are consistent with a Th2-mediated response to ASNase and suggest that CD4+ Tregs may limit sensitization to ASNase.

Depletion of CD4+ T cells protects against the development of the anti-ASNase antibodies An immediate drop in rectal temperature was measured in ASNase sensitized mice upon ASNase challenge on Day 24, which was quantified by estimating the area under the temperature versus time curve (AUC, Figure 4A). Plasma concentrations of mMCP-1 were elevated ~10 fold 2 h after the challenge relative to controls (Figure 4B), suggesting that mast cell degranulation is playing a role in the reactions. Sensitized mice had much lower blood ASNase enzyme activity than controls (Figure 4C), indicating rapid, presumably antibody-mediated clearance of the drug. Similar to the pre-challenge data, only the frequency of CD4+ Tregs and basophils of the blood increased relative to naïve controls (Online Supplementary Figure S6A-H).

Furthermore, ASNase bound ex vivo to B cells, neutrophils, macrophages/monocytes and basophils (Figure 4D).

However, there was a sharp drop in the binding of ASNase to B cells (~25%) and a modest increase in ASNase+ basophils (~10%) after the challenge (Figure 4D) relative to one day before the ASNase challenge (Day 23; Figure 1B).

To determine if B-cell or CD4+ T-cell depletion can prevent ASNase sensitization and the onset of hypersensitivity reactions, we depleted B cells or CD4+ T cells using anti-CD19 or anti-CD4 mAb, respectively, three days before each immunization dose (Online Supplementary Figure S3B). Less than 5% of B or CD4+ T cells remained after depletion in the blood (Figure 5A-B). Upon ASNase challenge, mice with depleted B or CD4+ T cells were protected from severe hypersensitivity reactions compared to sensitized mice with no immune cell depletion (Figure 5C; Online Supplementary Figure S7; P<0.01). Surprisingly, B-cell depletion did not completely protect mice from hypersensitivity (Figure 5C, P<0.05), whereas CD4+ T-cell depletion provided full protection (Figure 5C, P>0.05). The stronger effect of the CD4+ T cell-depleting mAb than the B-cell depleting mAb may reflect the ability of anti-CD4 mAb to suppress the function of CD4+ T cells that it fails to deplete,27 while anti-CD19 mAb has a less global suppressive effect on B-cell function.28 Both groups failed to develop detectable levels of antiASNase IgG (Figure 5D, P>0.05), and consistent with the lower anti-ASNase antibodies measured, both had similar ASNase enzyme activity levels relative to naïve mice (Figure 5E, P>0.05). CD4+ T-cell depletion led to similar mMCP-1 levels as naïve mice, whereas mice depleted of B cells had elevated mMCP-1 relative to both naïve and T cell-depleted mice (Figure 5F, P<1x10-4), suggesting that hypersensitivity in B cell-depleted mice was mediated by anti-ASNase IgE and mast cell degranulation. Supporting haematologica | 2019; 104(2)

this interpretation, the plasma anti-ASNase IgE levels after B-cell depletion were similar to sensitized mice and elevated compared to non-sensitized, naïve controls and CD4+ T cell-depleted mice by ELISA (Figure 5G, P<1x10-3) and by flow cytometry (Figure 5H, P<1x10-4). On Day 24 after the challenge, ex vivo binding of ASNase to basophils was detected in mice with depleted B cells (Online Supplementary Figure S8A), but not in mice depleted of CD4+ T cells (Online Supplementary Figure S8B). The ex vivo binding of ASNase to macrophages/monocytes was significantly higher in sensitized mice after CD4+ T-cell or B-

A

B

C

D

Figure 3. Frequencies of CD4+ Tregs, B cells, and basophils increase in blood after ASNase sensitization. The frequency of (A) macrophages/monocytes, (B) CD4+ Treg cells, (C) B cells, and (D) basophils was measured in the blood of mice at various time points during sensitization. A total of 5 or 10 mice were included in each analysis, as indicated, and P value significance is indicated as * for P<0.05, ** for P<0.01, *** for P<1x10-3, and **** for P<1x10-4.

323

S. Rathod et al.

A

B

C

D

Figure 4. ASNase-sensitized mice develop hypersensitivity reactions when challenged. The rectal temperature of sensitized and non-sensitized mice was monitored for 2 hours after the ASNase challenge on Day 24. (A) Sensitized mice experienced a drop in rectal temperature that was quantified by estimating the AUC of the rectal temperature versus time curve. (B) Sensitized mice had elevated levels of mMCP-1 and (C) low ASNase drug levels compared to non-sensitized mice. (D) ASNase binds to B cells, neutrophils, macrophages/monocytes, basophils, and T cells after challenging with ASNase on Day 24. A total of 5 or 20 mice were included in each analysis, as indicated, and P value significance is indicated as * for P<0.05, ** for P<0.01, *** for P<1x10-3, and **** for P<1x10-4.

cell depletion relative to naïve mice (Online Supplementary Figure S8A-B, P<1x10-4), but at about half of the frequency relative to sensitized mice with no cell depletion (Figure 4D). No binding of ASNase to B cells or neutrophils was detected after CD4+ T-cell or B-cell depletion (Online Supplementary Figure S8A-B, P>0.05). The data suggests that binding of ASNase to B cells, neutrophils, or macrophages/monocytes may not correlate with the onset of ASNase hypersensitivity, whereas the binding of ASNase to basophils may be a useful marker of ASNase hypersensitivity.

zation dose of ASNase on Day 9 and challenged the mice with IV ASNase on Day 10 (Online Supplementary Figure S3C). We detected no basophil activation by BAT (Online Supplementary Figure S9A-B) or ex vivo ASNase binding to basophils (Online Supplementary Figure S9C). Consistent with these results, mice did not develop hypersensitivity reactions upon ASNase challenge (Online Supplementary Figure S9D). Taken together, our data suggest that the binding of ASNase to basophils and ASNase-mediated basophil activation correlate with the onset of ASNase hypersensitivity.

A basophil activation test (BAT) can detect IgG and/or IgE ASNase hypersensitivity

In vivo blocking with EM-95 or 2.4G2 suggests multiple mechanisms of ASNase-induced hypersensitivity

A basophil activation test was used to determine if ASNase-induced activation can be mediated and detected via both ASNase immune complex binding to FcγRIII and free ASNase binding to cell-associated IgE. Blood samples from sensitized and non-sensitized mice were collected on Day 23 of the sensitization protocol (Online Supplementary Figure S3A) and incubated with ASNase, RPMI-medium (negative control), EM-95 (positive control for IgE-mediated activation, Figure 6A) or 2.4G2 (positive control for IgGmediated activation, Figure 6B). Samples from naïve mice showed no ASNase-mediated basophil activation (Figure 6C-D); however, ASNase-sensitized mice showed an upregulation of CD200R1 and a downregulation of CD200R3 relative to naïve mice (Figure 6C-D, P<0.01), suggesting that the activation of basophils and possibly the onset of ASNase hypersensitivities may be mediated by both pathways of anaphylaxis. We anticipated that antiASNase IgG may interfere with the binding of ASNase to cell-associated IgE.26 Consistent with that hypothesis, after washing the samples with mouse plasma and removing anti-ASNase IgG antibodies, CD200R3 was no longer downregulated after incubation with ASNase (Figure 6D, P>0.05), whereas a significant upregulation of CD200R1 after removing anti-ASNase IgG was measured relative to sensitized basophils with no antibody removal and to naïve control samples (Figure 6C, P<0.01).

To determine if the BAT can distinguish between sensitized and non-sensitized samples from mice with ASNase exposure, we collected blood samples after a single sensiti-

In vivo blocking experiments with EM-95 or 2.4G2 were performed to determine whether ASNase hypersensitivity occurs via both the FcεRI/IgE- and FcγRIII/IgG-mediated pathways as suggested by our BAT data. Mice were pretreated with either mAb on Day 23 and challenged the next day with ASNase. Mice receiving EM-95 or 2.4G2 were partially protected from hypersensitivity compared to sensitized mice (Figure 7, P<1x10-3). The results suggest that ASNase hypersensitivity occurs via IgG- and IgEmediated mechanisms and that hypersensitivity after pretreatment with EM-95 was due to PAF release and that after 2.4G2 was due to histamine release.20,24 We, therefore, tested this hypothesis by pretreating mice with CV6209 or triprolidine after EM-95 or 2.4G2 administration, respectively, or with each receptor antagonist alone or in combination with no blocking antibody. Consistent with our hypothesis, mice receiving EM-95 or 2.4G2 as well as the appropriate pretreatment medication were completely protected from ASNase hypersensitivity (Figure 7, P>0.05). Pretreating with CV-6209 or triprolidine alone yielded similar results as either blocking antibody (Figure 7, P>0.05), and ASNase hypersensitivity was completely suppressed in sensitized mice pretreated with CV-6209 and triprolidine (Figure 7, P>0.05). Our results indicate that both pathways of hypersensitivity simultaneously play a role in ASNase-mediated reactions and that receptor antagonists of both histamine and PAF are required to completely block the severity of the hypersensitivity reaction in our mouse model.

324

haematologica | 2019; 104(2)

ASNase hypersensitivities are IgG/IgE dependent

A

B

C

D

E

F

G

H

Figure 5. The in vivo depletion of CD4+ T cells but not B cells prevents ASNase-mediated hypersensitivity reactions. (A) B cells or (B) CD4+ T cells were depleted in vivo using anti-CD19 or anti-CD4 mAb three days prior to each ASNase sensitization dose. Less than 5% of B or CD4+ T cells remained after depletion in the blood.

(C) The rectal temperature AUC after B or CD4+ T-cell depletion was determined after challenging with ASNase. (D) Anti-ASNase IgG levels, (E) ASNase activity, (F) mMCP-1 concentrations, and anti-ASNase IgE levels, as assessed by (G) ELISA and (H) flow cytometry, were measured from the plasma samples of mice collected after the challenge. A total of 5 or 10 mice were included in each analysis, as indicated, and P value significance is indicated as * for P<0.05, ** for P<0.01, *** for P<1x10-3, and **** for P<1x10-4.

Discussion ASNase is an essential component of ALL treatment, but the ASNase-neutralizing antibody response to this chemotherapeutic agent can increase the risk of leukemia relapse.13 Little is known about the mechanism of the immune response to ASNase, and currently no clinical test is available to accurately predict which patients will develop hypersensitivity to subsequent doses of ASNase. The results of our study suggest that a possible explanation for the lack of predictive biomarkers is that multiple mechanisms of ASNase hypersensitivity can occur separately or simultaneously. Our data indicate that hypersensitivity to ASNase can be mediated by both anti-ASNase IgG and IgE through the immunoglobulin receptors FcγRIII and FcεRI, respectively, which is consistent with the heterogeneity observed in the clinical immune response to the chemotherapeutic agent.2 We provide the following evidence supporting our proposed mechanisms of hypersensitivity: (1) ASNase binds ex vivo to sensitized cells expressing FcγRIII and/or FcεRI (Figure 1B). Among IgEleukocytes, ASNase binding is inhibited by antibodies targeting FcγRIII but not IgE (Figure 2), whereas, in contrast, the ex vivo binding of ASNase to basophils, which express both FcγRIII and FcεRI, is inhibited by antibodies targeting IgE and FcγRIII (Figure 2). (2) B cell-depleted mice develhaematologica | 2019; 104(2)

oped IgE-mediated hypersensitivity reactions; that is, B cell-depleted mice have no detectable anti-ASNase IgG (Figure 5D), but develop elevated levels of anti-ASNase IgE (Figure 5G and H), experience hypersensitivity reactions when challenged with ASNase (Figure 5C) and have basophils that bind ASNase (Online Supplementary Figure S8). This is consistent with previous evidence that partial Rag deficiency (Omenn syndrome), which leads to B and T-cell oligoclonality, is associated with relative enrichment of IgE responses.29 (3) Markers of ASNase basophil activation also support multiple mechanisms of ASNase hypersensitivity in mice (Figure 6C-D). (4) Pretreating sensitized mice with anti-IgE (EM-95) or anti-FcγRIIB/III (2.4G2) demonstrates that each pathway is playing a substantial role in hypersensitivity reactions (Figure 7). These results are consistent with our current and previous study demonstrating that receptor antagonists of histamine and PAF are required to completely mitigate the severity of the ASNase immune response (Figure 7),21 as well as a previous study of hypersensitivity in mice treated with insulinderived peptides.24 Few clinical studies of the immune response to ASNase have investigated the role of anti-ASNase IgE on the development of hypersensitivity reactions.14-16 Based on our results demonstrating detectable plasma levels of antiASNase IgE (Figure 1E and F), we believe that antigen-spe325

S. Rathod et al.

cific recognition and hypersensitivity reactions may be mediated in part by cell-associated IgE. In support of a role for anti-ASNase IgE in the binding or recognition of free ASNase and not ASNase immune complexes, we found that blocking IgE with EM-95 decreases the binding of ASNase to sensitized basophils (Figure 2). Furthermore, removing soluble IgG before assessing ex vivo binding to ASNase demonstrates that free ASNase binding to basophils depends on cell-associated IgE and not the FcγRIII receptor (Figure 2). These data are in agreement with our BAT experiments, which demonstrate that basophils activation is IgE- but not FcγRIII-dependent after removing soluble anti-ASNase antibody (Figure 6C-D).

Anti-ASNase IgG antibodies are typically measured during ASNase therapy and higher levels associate with the onset of ASNase hypersensitivity in pediatric patients.2 The ex vivo binding of IgE- leukocytes to ASNase (Figure 2) and the downregulation of CD200R3 basophil expression (Figure 6) were dependent on soluble IgG, indicating that the formation of ASNase immune complexes is required to induce IgG-mediated hypersensitivity. Similarly, and

consistent with a murine IgG1 response (Online Supplementary Figure S4),30 IgG-mediated ASNase hypersensitivity was FcγRIII-dependent (Figure 7). Interestingly, the binding of ASNase to B cells and basophils, but not macrophages/monocytes or neutrophils, increased after challenging sensitized mice with ASNase (Figure 1B and 4D). Although this change in ASNase binding is likely due to the presence of ASNase immune complexes that formed during the challenge, it is not clear why the binding to macrophages/monocytes or neutrophils was unaffected, or why the binding to basophils was increased. A possible explanation for the increased ex vivo binding of ASNase to basophils may be that adding additional ASNase after the challenge allowed for free ASNase to bind to cell-associated IgE, whereas before the challenge anti-ASNase IgG may have neutralized the drug before it could bind to basophil-associated anti-ASNase IgE. Cell depletion studies indicate that depletion of CD4+ T cells or B cells can block the development of anti-ASNase IgG (Figure 5D) and drug neutralization (Figure 5E).

However, CD4+ T-cell but not B-cell depletion (both

A

B

C

D

Figure 6. Ex vivo ASNase activates basophils in an IgG- and IgE-dependent manner. (A) IgE-mediated basophil activation by EM-95 (anti-IgE mAb) upregulates CD200R1 basophil expression relative to media control. (B) IgG-mediated basophil activation by 2.4G2 (anti-FcγRIIB/III mAB) downregulates CD200R3 basophil expression relative to media. (C) ASNase exposure among sensitized peripheral blood cells upregulates the CD200R1 expression of basophils in the presence and absence of anti-ASNase IgG (no IgG). (D) Downregulation of basophil CD200R3 expression in response to ASNase in ASNase-sensitized mice requires anti-ASNase IgG. A total of 10 mice were included in each analysis, as indicated, and P value significance is indicated as * for P<0.05, ** for P<0.01, *** for P<1x10-3, and **** for P<1x10-4.

326

haematologica | 2019; 104(2)

ASNase hypersensitivities are IgG/IgE dependent

depleted to below 5%) is sufficient to completely protect mice from the development of anti-ASNase IgE antibodies and the onset of ASNase hypersensitivity (Figure 5C). The development of ASNase hypersensitivity after B-cell depletion is consistent with previous clinical data indicating that patients are depleted of B cells but not T cells during leukemia treatment, but yet develop hypersensitivity reactions to ASNase.31 The presence of anti-ASNase IgE may explain why some patients develop ASNase hypersensitivity in the absence of detectable anti-ASNase IgG,2 suggesting that ASNase hypersensitivity reactions probably also occur via multiple pathways in humans.

Regarding the prevention of ASNase-mediated immune responses, our results suggest that although treatment with rituximab during ALL induction therapy may protect against the inactivation and accelerated clearance of ASNase by antigen-specific IgG antibodies, rituximab-pretreated patients may not necessarily be protected against ASNase hypersensitivity. It is possible that pretreating patients before ASNase administration with rituximab and antihistamine or an agent that can attenuate T-cell activation may mask or mitigate hypersensitivity reactions while allowing ASNase therapeutic drug levels to be achieved.

It is not routine practice to evaluate anti-ASNase IgE antibody levels during ASNase therapy,2,32,33 and while some studies have suggested a role of anti-ASNase IgE during hypersensitivity reactions to ASNase,14-16 the presence or absence of plasma or serum antigen-specific IgE may not necessarily be predictive of FcεRI receptor-bound antigen-specific IgE.34 Likewise, performing skin prick tests,35 basophil activation tests that solely incorporate markers of IgE-mediated hypersensitivity,36 or anti-

ASNase IgG antibody measurements2 cannot accurately explain or predict clinical reactions to ASNase. Therefore, investigating receptors or immune cells that are directly involved in hypersensitivity reactions, rather than relying on biomarkers of sensitization for prediction, may be more useful. We hypothesize that a cell-based approach will distinguish when antigen and antibody levels are sufficient to form immune complexes capable of binding the Fcγ receptor, and also be able to detect antigen binding to cell-associated IgE. Consistent with our hypothesis, the ASNase basophil activation test can detect activation via both the classical and alternative pathway of anaphylaxis (Figure 6C-D). Although the changes in CD200R3 measured were modest (but statistically significant), they are similar to those we observed using a positive control (i.e., 2.4G2) and to those that have been previously described.22,23 ASNase ex vivo binding was simultaneously detected in an FcγRIII- and IgE-dependent mechanism of hypersensitivity (Figure 2), and due to the simultaneous upregulation of CD200R1 and downregulation of CD200R3 by the BAT, we hypothesized that ASNase hypersensitivity was both FcγRIII- and IgE-dependent. Supporting our hypothesis, blocking IgE-mediated hypersensitivity using EM-95 or antihistamine only partially mitigated ASNase hypersensitivity (Figure 7). Similar results were obtained when blocking IgG-mediated hypersensitivity reactions using 2.4G2 or PAF receptor antagonist (Figure 7). However, ASNase hypersensitivity reactions were completely inhibited when both pathways of anaphylaxis were simultaneously blocked using EM-95 and PAF receptor antagonist, 2.4G2 and antihistamine, or PAF receptor antagonist and antihistamine. It is not yet known whether both path-

Figure 7. FcεRI and FcγRIII play a role in ASNase hypersensitivity. Sensitized mice were pretreated with EM-95 (antiIgE mAB), 2.4G2 (anti-FcγRIIB/III mAB), CV-6209 (PAF receptor antagonist), triprolidine (antihistamine), EM-95 and CV-6209, 2.4G2 and triprolidine, or triprolidine and CV-6209 and challenged with ASNase to induce hypersensitivities. The AUC of the temperature vs. time curve was estimated. All mice receiving any pretreatment medication had a significant reduction in the severity of ASNase hypersensitivities relative to sensitized, non-pretreated mice (P<1x10-4). A total of 5 mice were included in each analysis, as indicated, and P value significance is indicated as * for P<0.05, ** for P<0.01, *** for P<1x10-3, and **** for P<1x10-4.

haematologica | 2019; 104(2)

327

S. Rathod et al.

ways of ASNase-induced hypersensitivity are clinically relevant in humans. Furthermore, while our results suggest that ASNase hypersensitivities can be masked by pretreatment with antihistamine and a PAF receptor antagonist, any pretreatment medication to mitigate ASNase hypersensitivity in patients must ensure that adequate drug levels are achieved to avoid the possibility of masking the hypersensitivity without achieving therapeutic ASNase drug levels.

The onset of ASNase hypersensitivities in our model requires two ASNase doses to sensitize mice before hypersensitivity reactions can be induced (Online Supplementary Figure S9D), likely due to the non-detectable levels of anti-ASNase IgG or IgE antibodies measured through Day 23 of our protocol (Figure 1D). Nevertheless, ASNase hypersensitivities were induced 10 days after the last sensitization dose (Day 24, Figure 4A), which correlates with the detection of high anti-ASNase antibody levels (Figure 1D).21 A similar correlation exists between antiASNase IgG levels and the onset of clinical ASNase hypersensitivities.2 Murine hypersensitivity reactions can be monitored via a decrease in core body temperature due to the increased permeability of vascular endothelium that is induced by histamine and other mast cell-produced vasoactive mediators.20,26,37-39 Increased endothelial permeability results in vascular fluid leak and hypovolemia, which causes shock that is most easily detected as hypothermia.40,41 Other markers of anaphylaxis include decreased physical activity, increased plasma levels of degranulation products (e.g., mMCP-1), and hemoconcentration (increased hematocrit levels) due to vascular leakage.20,42 Our study detected hypersensitivities via the development of hypothermia and the release of mMCP-1.

Previous studies on ASNase hypersensitivity have demonstrated a correlation between the dose of ASNase, the severity of ASNase hypersensitivity, and the levels of mMCP-1 released, supporting that our methods accurate-

References 1. Asselin BL, Ryan D, Frantz CN, et al. In vitro and in vivo killing of acute lymphoblastic leukemia cells by L-asparaginase. Cancer Res. 1989;49(15):4363-4368.

2. Liu C, Kawedia JD, Cheng C, et al. Clinical utility and implications of asparaginase antibodies in acute lymphoblastic leukemia. Leukemia. 2012;26(11):23032309.

3. Oettgen HF, Stephenson PA, Schwartz MK, et al. Toxicity of E. coli L-asparaginase in man. Cancer. 1970;25(2):253-278.

4. Panosyan EH, Seibel NL, Martin-Aragon S, et al. Asparaginase antibody and asparaginase activity in children with higher-risk acute lymphoblastic leukemia: Children's Cancer Group Study CCG-1961. J Pediatr Hematol Oncol. 2004;26(4):217-226.

5. Avramis VI, Sencer S, Periclou AP, et al. A randomized comparison of native Escherichia coli asparaginase and polyethylene glycol conjugated asparaginase for treatment of children with newly diagnosed standard-risk acute lymphoblastic leukemia: a Children's Cancer Group study. Blood. 2002;99(6):1986-1994.

328

ly measure the onset, severity, and extent of hypersensitivity.21 Our data support a hypothesis that the mechanism of ASNase-mediated hypersensitivity involves antigen-specific IgG and/or IgE and the immunoglobulin receptors FcγRIII and/or FcεRI. Our results also indicate that both mechanisms can simultaneously or independently contribute to the onset and extent of ASNase hypersensitivity. We show that cells expressing FcγRIII and FcεRI can bind ASNase ex vivo and, therefore, it is likely that multiple cells play a role during the onset of hypersensitivity reactions, including mast cells, which are not present in systemic circulation.

Our study has several possible clinical implications regarding predicting, overcoming, and preventing ASNase hypersensitivities. The ex vivo binding of ASNase to basophils correlates with the onset of ASNase hypersensitivity reactions, suggesting that ASNase binding to basophils may be a useful biomarker of ASNase hypersensitivity regardless of whether anaphylaxis is mediated by the classical and/or alternative pathway. Similarly, the ASNase BAT can detect IgE/FcεRI- or IgG/FcγRIII-dependent basophil activation in our study; however, human markers of IgG/FcγR-mediated basophil activation are not available. In addition, our results suggest that both PAF and histamine are important mediators of ASNase hypersensitivity and that receptor antagonists of these molecules may be able to block the clinical manifestation of ASNase hypersensitivity. Future research will attempt to demonstrate the specificity and sensitivity of both basophil binding and basophil activation by ASNase for predicting ASNase hypersensitivity and verify the possibility of the alternative pathway of ASNase hypersensitivity in humans.

Funding This study was supported by the University of Pittsburgh School of Pharmacy and NIH Grants RO1 CA216815 and CA142665.

6. Woo MH, Hak LJ, Storm MC, et al. Antiasparaginase antibodies following E. coli asparaginase therapy in pediatric acute lymphoblastic leukemia.

Leukemia.

1998;12(10):1527-1533.

7. Woo MH, Hak LJ, Storm MC, et al.

Hypersensitivity or development of antibodies to asparaginase does not impact treatment outcome of childhood acute lymphoblastic leukemia. J Clin Oncol.

2000;18(7):1525-1532.

8. Asselin BL. The three asparaginases.

Comparative pharmacology and optimal use in childhood leukemia. Adv Exp Med Biol. 1999;457:621-629.

9. Appel IM, Kazemier KM, Boos J, et al.

Pharmacokinetic, pharmacodynamic and intracellular effects of PEG-asparaginase in newly diagnosed childhood acute lymphoblastic leukemia: results from a single agent window study.

Leukemia.

2008;22(9):1665-1679.

10. Bowman WP, Larsen EL, Devidas M, et al.

Augmented therapy improves outcome for pediatric high risk acute lymphocytic leukemia: results of Children's Oncology Group trial P9906. Pediatr Blood Cancer.

2011;57(4):569-577.

11. Soyer OU, Aytac S, Tuncer A, Cetin M,

12.

13.

14.

15.

16.

Yetgin S, Sekerel BE. Alternative algorithm for L-asparaginase allergy in children with acute lymphoblastic leukemia. J Allergy Clin Immunol. 2009;123(4):895-899.

Pieters R, Hunger SP, Boos J, et al. Lasparaginase treatment in acute lymphoblastic leukemia: a focus on Erwinia asparaginase. Cancer. 2011;117(2):238-249.

Kawedia JD, Liu C, Pei D, et al.

Dexamethasone exposure and asparaginase antibodies affect relapse risk in acute lymphoblastic leukemia.

Blood.

2012;119(7):1658-1664.

Galindo-Rodriguez G, Jaime-Perez JC, Salinas-Carmona MC, et al. Do immunoglobulin G and immunoglobulin E anti-l-asparaginase antibodies have distinct implications in children with acute lymphoblastic leukemia? A cross-sectional study. Rev Bras Hematol Hemoter.

2017;39(3):202-209.

Tsurusawa M, Chin M, Iwai A, et al. LAsparagine depletion levels and L-asparaginase activity in plasma of children with acute lymphoblastic leukemia under asparaginase treatment. Cancer Chemother Pharmacol. 2004;53(3):204-208.

Killander D, Dohlwitz A, Engstedt L, et al.

Hypersensitive reactions and antibody for-

haematologica | 2019; 104(2)

ASNase hypersensitivities are IgG/IgE dependent

17.

18.

19.

20.

21.

22.

23.

24.

25.

mation during L-asparaginase treatment of children and adults with acute leukemia.

Cancer. 1976;37(1):220-228.

Fernandez CA, Smith C, Yang W, et al.

HLA-DRB1*07:01 is associated with a higher risk of asparaginase allergies. Blood.

2014;124(8):1266-1276.

Fernandez CA, Smith C, Yang W, et al.

Genome-wide analysis links NFATC2 with asparaginase hypersensitivity. Blood.

2015;126(1):69-75.

Tsujimura Y, Obata K, Mukai K, et al.

Basophils play a pivotal role in immunoglobulin-G-mediated but not immunoglobulin-E-mediated systemic anaphylaxis. Immunity. 2008;28(4):581-589.

Strait RT, Morris SC, Yang M, Qu XW, Finkelman FD. Pathways of anaphylaxis in the mouse. J Allergy Clin Immunol.

2002;109(4):658-668.

Fernandez CA, Smith C, Karol SE, et al.

Effect of premedications in a murine model of asparaginase hypersensitivity. The Journal of pharmacology and experimental therapeutics. 2015;352(3):541-551.

Iwamoto H, Matsubara T, Nakazato Y, Namba K, Takeda Y. Evaluation of the antigenicity of hydrolyzed cow's milk protein formulas using the mouse basophil activation test. Toxicol Lett. 2016;242:53-59.

Iwamoto H, Matsubara T, Nakazato Y, Namba K, Takeda Y. Decreased expression of CD200R3 on mouse basophils as a novel marker for IgG1-mediated anaphylaxis.

Immun Inflamm Dis. 2015;3(3):280-288.

Liu E, Moriyama H, Abiru N, et al. Antipeptide autoantibodies and fatal anaphylaxis in NOD mice in response to insulin self-peptides B:9-23 and B:13-23. J Clin Invest. 2002;110(7):1021-1027. van der Meer LT, Terry SY, van Ingen Schenau DS, et al. In vivo imaging of antileukemic drug asparaginase reveals a rapid macrophage-mediated clearance from the bone marrow. J Nucl Med. 2017;58(2):214-220.

haematologica | 2019; 104(2)

26. Strait RT, Morris SC, Finkelman FD. IgGblocking antibodies inhibit IgE-mediated anaphylaxis in vivo through both antigen interception and Fc gamma RIIb cross-linking. J Clin Invest. 2006;116(3):833-841.

27. Wilde DB, Marrack P, Kappler J, Dialynas DP, Fitch FW. Evidence implicating L3T4 in class II MHC antigen reactivity; monoclonal antibody GK1.5 (anti-L3T4a) blocks class II MHC antigen-specific proliferation, release of lymphokines, and binding by cloned murine helper T lymphocyte lines. J Immunol. 1983;131(5):2178-2183.

28. Karnell JL, Dimasi N, Karnell FG, 3rd, et al.

CD19 and CD32b differentially regulate human B cell responsiveness. J Immunol.

2014;192(4):1480-1490.

29. Villa A, Santagata S, Bozzi F, et al. Partial V(D)J recombination activity leads to Omenn syndrome. Cell. 1998;93(5):885896.

30. Bruhns P. Properties of mouse and human IgG receptors and their contribution to disease models. Blood. 2012;119(24):56405649.

31. Caver TE, Slobod KS, Flynn PM, et al.

Profound abnormality of the B/T lymphocyte ratio during chemotherapy for pediatric acute lymphoblastic leukemia.

Leukemia. 1998;12(4):619-622.

32. Wang B, Hak LJ, Relling MV, Pui CH, Woo MH, Storm MC. ELISA to evaluate plasma anti-asparaginase IgG concentrations in patients with acute lymphoblastic leukemia. J Immunol Methods 2000;239(12):75-83.

33. van der Sluis IM, Vrooman LM, Pieters R, et al. Consensus expert recommendations for identification and management of asparaginase hypersensitivity and silent inactivation. Haematologica. 2016;101(3):279-285.

34. Fish SC, Donaldson DD, Goldman SJ, Williams CM, Kasaian MT. IgE generation and mast cell effector function in mice deficient in IL-4 and IL-13. J Immunol.

2005;174(12):7716-7724.

35. Stone HD, Jr., DiPiro C, Davis PC, Meyer CF, Wray BB. Hypersensitivity reactions to Escherichia coli-derived polyethylene glycolated-asparaginase associated with subsequent immediate skin test reactivity to E. coli-derived granulocyte colony-stimulating factor. J Allergy Clin Immunol.

1998;101(3):429-431.

36. Hino M, Shimojo N, Ochiai H, et al.

Expression of CD203c on basophils as a marker of immunoglobulin E-mediated (L)asparaginase allergy. Leuk Lymphoma.

2014;55(1):92-96.

37. Beutier H, Gillis CM, Iannascoli B, et al.

IgG subclasses determine pathways of anaphylaxis in mice. J Allergy Clin Immunol.

2017;139(1):269-280.e267.

38. O'Konek JJ, Landers JJ, Janczak KW, et al.

Nanoemulsion adjuvant-driven redirection of TH2 immunity inhibits allergic reactions in murine models of peanut allergy. J Allergy Clin Immunol. 2018;141(6):21212131.

39. Walker MT, Green JE, Ferrie RP, Queener AM, Kaplan MH, Cook-Mills JM. Mechanism for initiation of food allergy: dependence on skin barrier mutations and environmental allergen costimulation. J Allergy Clin Immunol. 2018;141(5):17111725.e1719.

40. Kind LS. Fall in rectal temperature as an indication of anaphylactic shock in the mouse. J Immunol. 1955;74(5):387-390.

41. Miyajima I, Dombrowicz D, Martin TR, Ravetch JV, Kinet JP, Galli SJ. Systemic anaphylaxis in the mouse can be mediated largely through IgG1 and Fc gammaRIII. Assessment of the cardiopulmonary changes, mast cell degranulation, and death associated with active or IgE- or IgG1dependent passive anaphylaxis. J Clin Invest. 1997;99(5):901-914.

42. Fulton JD, Harris WE, Craft CE. Hematocrit change as indication of anaphylactic shock in the mouse. Proc Soc Exp Biol Med. 1957;95(4):625-627.

329

ARTICLE Ferrata Storti Foundation

Non-Hodgkin Lymphoma

JUNB, DUSP2, SGK1, SOCS1 and CREBBP are frequently mutated in T-cell/histiocyterich large B-cell lymphoma

Bianca Schuhmacher,1 Julia Bein,1 Tobias Rausch,2,3 Vladimir Benes2, Thomas Tousseyn,4 Martine Vornanen,5 Maurilio Ponzoni,6 Lorenz Thurner,7,8 Randy Gascoyne,9 Christian Steidl,9 Ralf Küppers,10,11 Martin-Leo Hansmann1,12,13 and Sylvia Hartmann1,12

Haematologica 2018 Volume 104(2):330-337

Dr. Senckenberg Institute of Pathology, Goethe University, Frankfurt am Main, Germany; Genecore, European Molecular Biology Laboratory (EMBL), Heidelberg, Germany; 3 Genome Biology Unit, European Molecular Biology Laboratory (EMBL), Heidelberg, Germany; 4Department of Pathology, University Hospitals K.U.Leuven, Belgium; 5 Department of Pathology, Tampere University Hospital and University of Tampere, Finland; 6 Unit of Lymphoid Malignancies, Department of Pathology, Scientific Institute San Raffaele, Milan, Italy; 7José Carreras Center for Immuno and Gene Therapy and Internal Medicine I, Saarland University Medical School, Homburg, Saar, Germany; 8Department of Internal Medicine 2, Hospital of the J. W. Goethe University, Frankfurt am Main, Germany; 9 Department of Pathology and Laboratory Medicine and the Centre for Lymphoid Cancer, British Columbia Cancer Agency, University of British Columbia, Vancouver, Canada; 10 Institute of Cell Biology (Cancer Research), Faculty of Medicine, University of DuisburgEssen, Essen, Germany; 11Deutsches Konsortium für Translationale Krebsforschung (DKTK), Germany; 12Reference and Consultant Center for Lymphoma and Lymph Node Diagnostics, Goethe University, Frankfurt am Main, Germany and 13Frankfurt Institute of Advanced Studies, Frankfurt am Main, Germany 1 2

ABSTRACT

Correspondence: s.hartmann@em.uni-frankfurt.de

Received: August 1, 2018.

Accepted: September 7, 2018.

Pre-published: September 13, 2018. doi:10.3324/haematol.2018.203224 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/330 ©2019 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.

330

T

-cell/histiocyte-rich large B-cell lymphoma is a rare aggressive lymphoma showing histopathological overlap with nodular lymphocyte-predominant Hodgkin lymphoma. Despite differences in tumor microenvironment and clinical behavior, the tumor cells of both entities show remarkable similarities, suggesting that both lymphomas might represent a spectrum of the same disease. To address this issue, we investigated whether these entities share mutations. Ultra-deep targeted resequencing of six typical and 11 histopathological variants of nodular lymphocyte-predominant Hodgkin lymphoma, and nine cases of T-cell/histiocyte-rich large B-cell lymphoma revealed that genes recurrently mutated in nodular lymphocyte-predominant Hodgkin lymphoma are affected by mutations at similar frequencies in T-cell/histiocyte-rich large B-cell lymphoma. The most recurrently mutated genes were JUNB, DUSP2, SGK1, SOCS1 and CREBBP, which harbored mutations more frequently in T-cell/histiocyte-rich large B-cell lymphoma and the histopathological variants of nodular lymphocyte-predominant Hodgkin lymphoma than in its typical form. Mutations in JUNB, DUSP2, SGK1 and SOCS1 were highly enriched for somatic hypermutation hotspot sites, suggesting an important role of aberrant somatic hypermutation in the generation of these somatic mutations and thus in the pathogenesis of both lymphoma entities. Mutations in JUNB are generally rarely observed in malignant lymphomas and thus are relatively specific for nodular lymphocyte-predominant Hodgkin lymphoma and T-cell/histiocyte-rich large B-cell lymphoma at such high frequencies (5/17 and 5/9 cases with JUNB mutations, respectively). Taken together, the findings of the present study further support a close relationship between T-cell/histiocyte-rich large B-cell lymphoma and nodular lymphocyte-predominant Hodgkin lymphoma by showing that they share highly recurrent genetic lesions.

haematologica | 2019; 104(2)

JUNB, DUSP2, SGK1 and SOCS1 mutations in THRLBCL

Introduction

Ultra-deep targeted resequencing and identification of somatic variants

T-cell/histiocyte-rich large B-cell lymphoma (THRLBCL) is a rare subtype of diffuse large B-cell lymphoma (DLBCL) characterized by a low fraction of tumor B cells and a cellular background rich in T cells and histiocytes. It has been classified as a separate entity of mature B-cell lymphoma since the fourth edition of the World Health Organization (WHO) classification of lymphoid neoplasms.1,2 Although it has a more aggressive clinical behavior and distinct microenvironmental composition,3,4 THRLBCL shares several clinical and pathological features with nodular lymphocyte-predominant Hodgkin lymphoma (NLPHL), a rare subtype of Hodgkin lymphoma.

The similarities include a predominance of middle-aged male patients5 and a minority of tumor cells derived from germinal center B cells in an abundant microenvironment.6,7 Furthermore, a high similarity of gene expression signatures4,8 and genomic copy number changes in the microdissected tumor cells of NLPHL and THRLBCL were found.9 According to Fan et al.,10 NLPHL can be subdivided into six different histopathological patterns, which include two typical nodular growth patterns (A and B) and four histopathological variants (C-F). The NLPHL variant E, also called THRLBCL-like variant, shows particularly marked similarities with THRLBCL and can only be distinguished from de novo THRLBCL by the presence of typical NLPHL remnants in the same lymph node or in another lymph node simultaneously sampled. In general, histopathological NLPHL variants are associated with an advanced clinical stage and an increased relapse rate.10,11 Data on somatic gene mutations of the tumor cells of THRLBCL are still lacking. Hence, we aimed to elucidate the relationship of THRLBCL and NLPHL through a comparison of recurrently mutated genes to obtain a more comprehensive understanding of the pathogenesis of THRLBCL.

Extraction of genomic DNA from frozen tissue of cases of THRLBCL and NLPHL, ultra-deep targeted resequencing after bait-based enrichment by a custom Haloplex kit (Agilent, Santa Clara, CA, USA) and processing of sequencing reads were carried out as described previously.12 All 62 genes used for the ultra-deep targeted resequencing were selected based on mutated genes from a previous study of two clonally related composite lymphomas consisting of NLPHL and DLBCL (Online Supplementary Table S1).12 Some of these selected genes were already confirmed to be mutated in primary NLPHL without transformation into DLBCL.12 The mean coverage of the 26 cases ranged between 3500 and 8500x (Online Supplementary Table S2). Non-synonymous single nucleotide variants (SNVs) were filtered for allele frequencies based on the expected tumor cell content (0.1-10%), and the presence and somatic origin of selected variants were confirmed in a semi-nested polymerase chain reaction (PCR) approach followed

by Sanger sequencing of the PCR products, as indicated in Online Supplementary Table S3. Details on validation of mutations using microdissected tumor cells and analysis of the somatic hypermutation (SHM) features of SNVs are provided in the Online Supplementary Methods section.

Methods Cases Cases were collected based on the availability of frozen tissue at the Dr. Senckenberg Institute of Pathology, Frankfurt am Main, Germany; the Department of Pathology University Hospitals, K.U. Leuven, Belgium; the Unit of Lymphoid Malignancies, San Raffaele Scientific Institute, Milan, Italy; Tampere University Hospital and University of Tampere, Tampere, Finland; and the Department of Pathology and Laboratory Medicine and the Centre for Lymphoid Cancer, British Columbia Cancer Agency, Vancouver, Canada. The local ethics committees approved the study, and written informed consent from the donors was obtained in accordance with the Declaration of Helsinki. All cases were reviewed on a multi-head microscope by expert hematopathologists (RG, SH, MLH, and TT). Only cases meeting the diagnostic criteria of the current WHO classification for NLPHL and THRLBCL1,2 were included in the study. Coexisting NLPHL was not found in any of the THRLBCL cases. For correlative analysis, NLPHL cases were classified according to Fan et al.10 For this purpose, the major pattern present in the frozen sample was considered. In all cases, the initial biopsies before treatment were analyzed. All cases also presented a typical immunophenotype of the tumor cells (CD20+, BCL6+, OCT2+, CD30–, CD15–, LMP1–). Information on the patients is provided in Table 1.

haematologica | 2019; 104(2)

Laser microdissection and Immunohistochemistry

Frozen sections (5 - 10 mm) of lymph nodes from lymphoma patients were mounted on membrane-covered slides (PALM, Zeiss, Bernried, Germany), then air-dried and fixed in acetone.

Sections were stained with a mouse monoclonal anti-CD20 antibody (clone L26, Dako, Glostrup, Denmark) in 1:200 dilution for 1 h at room temperature. Binding of the primary antibody was visualized with the aid of the Super SensitiveTM Link-Label IHC Detection System (BioGenex, Fremont, CA, USA), and counterstaining with hematoxylin was performed. For PCR analysis, 20 single tumor cells and non-tumor cells were isolated using the PALM laser capture microdissection technique (PALM MicroBeam, Zeiss, Bernried, Germany) and collected in 20 µL PCR buffer without MgCl2 (Expand High Fidelity, Roche, Grenzach, Germany) supplemented with 0.1% Triton X-100.

The immunohistochemical staining for activation-induced cytidine deaminase (AICDA) was performed on an independent series of 15 typical and 11 variant NLPHL as well as 12 THRLBCL with formalin-fixed paraffin-embedded tissue as previously described.13 The anti-AICDA antibody (clone EK2 5G9, Cell Signaling, Danvers, MA, USA) was applied in a dilution of 1:100.

Results T-cell/histiocyte-rich large B-cell lymphoma shares recurrently mutated genes with nodular lymphocyte-predominant Hodgkin lymphoma Ultra-deep targeted resequencing of the coding exons of 62 genes (Online Supplementary Table S1), which were found to be mutated in NLPHL in a previous study,12 was performed in the following three groups: THRLBCL (n=9); typical NLPHL, patterns A/B (n=6); and histopathological NLPHL variants, patterns C/D/E (n=11), yielding a total number of 26 cases. The patients’ characteristics are given in Table 1. The coding exons of selected genes were analyzed for non-synonymous SNVs in order to identify mutations with a potential functional effect.

Of the 62 NLPHL-related genes, 48 (77%) were somatically mutated in at least one of the 26 cases (Online 331

B. Schuhmacher et al.

Table 1. Clinical information on nodular lymphocyte-predominant Hodgkin lymphoma and T-cell/histiocyte-rich large B-cell lymphoma patients.

Case number

Former case number*

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

6 7 8 9 10 11 5

23 24 25 26

12 13 17

Diagnosis THRLBCL THRLBCL THRLBCL THRLBCL THRLBCL THRLBCL THRLBCL THRLBCL THRLBCL Variant NLPHL Variant NLPHL Variant NLPHL Variant NLPHL Variant NLPHL Variant NLPHL Typical NLPHL Typical NLPHL Typical NLPHL Typical NLPHL Typical NLPHL Typical NLPHL Variant NLPHL , part of composite lymphoma Variant NLPHL Variant NLPHL Variant NLPHL Variant NLPHL

Fan pattern#

Estimated tumor cell content

Gender

Age

Stage

pattern E pattern E pattern E pattern E pattern E pattern E pattern A pattern A pattern A pattern A pattern A pattern B pattern D

15% 5% 10% 15% 10% 1% 10% 3% 5% 1% 1% 1% 3% 5% 3% 3% 1% 1% 3% 1% 3% 3%

M M F M M F M M F M M F M M M M M M M M F M

20 47 50 40 50 89 27 31 75 59 46 85 40 23 46 10 72 64 13 52 53 35

IVB IVB IVB IVB n.a.

IVA IVA IVB IIIA IVB IVA IIIB IVB IIIA IVA IIA n.a. n.a.

IA n.a. n.a. n.a.

pattern C pattern C pattern D pattern D

1% 3% 3% 3%

M M M M

64 15 31 75

n.a.

IIA IA n.a.

* These cases were included in a previous study of nodular lymphocyte-predominant Hodgkin lymphoma.12 # Histopathological patterns were determined according to Fan et al.10 THRLBCL:T-cell/histiocyte-rich large B-cell lymphoma; NLPHL: nodular lymphocyte-predominant Hodgkin lymphoma; n.a.: Information on Ann Arbor stage was not available.

Supplementary Table S3). A total of 33 genes (53%) were recurrently affected by mutations in at least two cases (Figure 1A). Specifically in THRLBCL, 31 genes (50%) were mutated, and 13 of these showed recurrent mutations in at least two cases, indicating a considerable mutational overlap between THRLBCL and NLPHL. The genes with the highest overall mutation frequency (≥20% of all 26 cases) were JUNB, DUSP2, SGK1, SOCS1, CREBBP, FN1 and TRRAP (Figure 1A). Mutation frequencies in JUNB, SGK1, CREBBP and TRRAP were higher, albeit not significantly so, in both THRLBCL and the NLPHL variants C/D/E than in typical NLPHL A/B (Figure 1A).

The median number of mutated genes per case was comparable between typical NLPHL (median 3.5; range, 213), histopathological NLPHL variants C/D/E (median 5.0; range, 1-11) and THRLBCL (median 4.0; range, 2-20) (Figure 1B). However, the median number of SNVs per case was slightly increased in THRLBCL (median 12; range, 2-30) when compared with that in typical NLPHL (median 7; range, 2-14) and NLPHL variants C/D/E (median 8; range, 1-20) (Figure 1C). Notably, the number of SNVs per case was specifically increased in the four genes JUNB, DUSP2, SGK1 and SOCS1 (1-5 SNVs/case) when compared to CREBBP, FN1 and TRRAP (1-2 SNVs/case), 332

particularly in regard to the shorter coding sequence of the first four genes (<1.3 kb versus >7.3 kb) (Figure 1D). The higher number of SNVs per case found in THRLBCL was, therefore, related to the higher number of SNVs in these four genes.

SOCS1 was the only gene in which mutations occurred significantly more frequently in THRLBCL than in typical NLPHL A/B (4/9 cases versus 0/6 cases, respectively; P=0.010, χ2-test) and in NLPHL in general (4/9 cases versus 3/17 cases, respectively; P=0.035, χ2-test).

Furthermore, the number of SNVs per case was significantly increased in SOCS1 in THRLBCL (median 5.0; range, 3-6) when compared with that in histopathological NLPHL variants C/D/E (median 1.5; range, 1-2; P=0.048, Mann-Whitney test). JUNB was the gene with the highest number of nonsense mutations, which were present in both NLPHL (2/15 mutations) and THRLBCL (2/12 mutations).

The genes with the highest mutation frequency are targets of aberrant somatic hypermutation Regarding the distribution of mutations in the seven most recurrently mutated genes, SNVs were generally distributed throughout the coding sequence with no differhaematologica | 2019; 104(2)

JUNB, DUSP2, SGK1 and SOCS1 mutations in THRLBCL

A

B

C

D

Figure 1. Genes recurrently affected by mutations and their mutational load in nodular lymphocyte-predominant Hodgkin lymphoma and T-cell/histiocyte-rich large B-cell lymphoma. (A) Frequencies of recurrently mutated genes (â&#x2030;Ľ2 cases) in six cases of NLPHL A/B (blue), 11 cases of NLPHL C/D/E (green) and nine cases of THRLBCL (red) sorted by overall recurrence. (B) Mutated genes per case in NLPHL A/B, NLPHL C/D/E and THRLBCL. (C) Number of SNVs per case in NLPHL A/B, NLPHL C/D/E and THRLBCL. (D) Scatter plot of SNVs per case and kbp of CCDS in the seven most recurrently mutated genes in NLPHL A/B, NLPHL C/D/E and THRLBCL. In (B-D), horizontal lines correspond to medians; P-values by Kruskal-Wallis test are indicated in the case of statistical significance. *P<0.05, **P<0.01, ***P<0.001. NLPHL: nodular lymphocyte-predominant Hodgkin lymphoma; THRLBCL: T-cell/histiocyte-rich large B-cell lymphoma; SNVs: single nucleotide variants; CCDS: Consensus coding sequence.

ence in the mutational distribution between the three groups of lymphoma (Figure 2). Protein domains with functional relevance were affected by mutations, e.g. the SH2 domain of the negative regulator of JAK-STAT signaling SOCS1, the catalytic domain of the protein kinase SGK1 and the histone acetyltransferase domain of the chromatin modifier CREBBP, suggesting that mutations in these genes result in an alteration of the protein function.

In the three relatively small genes (<2.5 kb) JUNB, DUSP2 and SOCS1, SNVs clustered to specific regions in the coding sequence with an enrichment to the first 150 bp downstream of the start codon in JUNB, to exon 2 in DUSP2 and to the first 80 bp of the coding sequence in SOCS1. In CREBBP, FN1 and TRRAP, which were the three larger genes (>75 kb), mutations tended to be diffusely scattered and localized further downstream of the transcriptional start site. Given the fact of ongoing SHM in NLPHL and THRLBCL,6,7 as well as aberrant activity of the SHM machinery in these entities,14 we explored whether mutations in the most recurrently mutated genes were caused by aberrant SHM. Criteria of SHM activity were investigated, as previously described in other reports.14â&#x20AC;&#x201C;17 Mutations in the genes JUNB, DUSP2, SGK1 and SOCS1 were highly enriched in SHM hotspot sites (Figure 3A) and also met other SHM criteria in most SNVs (Figure 3B,C).

Mutations in CREBBP, FN1 and TRRAP occasionally haematologica | 2019; 104(2)

occurred at C:G sites and showed a predominance of transitions over transversions (Figure 3B,C). However, they were clearly located outside the SHM hotspot motifs WRCY/RGYW (Figure 3A). In line with these results of an active SHM machinery and consistent with previous data,13,18 expression of AICDA could be demonstrated in the tumor cells of typical and histopathological NLPHL variants (10/15 and 3/11 cases, respectively) (Figure 3D) as well as THRLBCL (10/12 cases) (Figure 3E) in an independent case series.

Discussion In the present study, we performed ultra-deep targeted resequencing of a set of genes that had been previously identified to be mutated in two cases of composite lymphoma of NLPHL with transformation into DLBCL.12 Four of these target genes, namely JUNB, DUSP2, SGK1 and SOCS1, were also found to be mutated in primary NLPHL cases without transformation. Since the relationship of NLPHL and THRLBCL has been long-discussed, we aimed to determine whether these four genes, among others, are also mutated in THRLBCL. Here, we confirm that mutations in these genes also occur at comparable frequencies in THRLBCL. 333

B. Schuhmacher et al.

Figure 2. Mutation patterns in the most recurrently mutated genes in nodular lymphocyte-predominant Hodgkin lymphoma and T-cell/histiocyte-rich large B-cell lymphoma. Schematic overview of the mutational distribution in the most recurrently mutated genes. For each gene missense (triangles), stop gain (spheres) and splice site (squares) mutations are mapped to coding exons (top) and protein domains (bottom). Mutations are color-coded according to the occurrence in the groups (blue: NLPHL A/B; green: NLPHL C/D/E; red: THRLBCL). Amino acid positions of protein domains are adopted from the Uniprot database (www.uniprot.org) and refer to the canonical sequences (JUNB: P17275, DUSP2: Q05923, SGK1: O00141, SOCS1: O15524, CREBBP: Q92793, FN1: P02751, TRRAP: Q9Y4A5). bZIP: basic leucine zipper motif; DsPc: dual specific phosphatase, catalytic domain; Pkinase: protein kinase domain; AGC-kinase C: AGC-kinase C-terminal domain; Rhod: rhodanese; SH2: Src homology 2 domain; KIR: kinase inhibitory region; CH1/2/3: cysteine/hystidine-rich region; KIX: kinase inducible domain; BROMO: bromodomain; HAT: histone acetyltransferase domain; Q: poly glutamine stretch; FN: fibronectin; TP53: tumor suppressor p53 binding site; FAT: FRAP-ATM-TRRAP domain; FATC: FAT C-terminal; PI3K/PI4K: phosphatidylinositol 3-kinase/ phosphatidylinositol 4-kinase domain; NLPHL: nodular lymphocyte-predominant Hodgkin lymphoma; THRLBCL: T-cell/histiocyte-rich large B-cell lymphoma.

334

haematologica | 2019; 104(2)

JUNB, DUSP2, SGK1 and SOCS1 mutations in THRLBCL

A

B

D

C

E

Figure 3. Characteristics of somatic mutations and activation-induced cytidine deaminase expression in nodular lymphocyte-predominant Hodgkin lymphoma and T-cell/histiocyte-rich large B-cell lymphoma. (A) Analysis of distribution of mutations to SHM hotspot motifs. (B) Ratio of mutations at C:G sites to A:T sites. (C) Ratio of transition to transversion mutations. In (A-C) synonymous SNVs were considered in addition to non-synonymous SNVs. Asterisks denote statistical significance. *P<0.05, **P<0.01, ***P<0.001. P-values by Ď&#x2021;2 test. (D) Expression of AICDA in the LP cells of NLPHL (400x magnification). (E) Expression of AICDA in the tumor cells of THRLBCL (400x magnification). Tumor cells are highlighted by arrows. AICDA: activation-induced cytidine deaminase; NLPHL: nodular lymphocyte-predominant Hodgkin lymphoma; THRLBCL: T-cell/histiocyte-rich large B-cell lymphoma; SHM: somatic hypermutation; SNVs: single nucleotide variants.

SGK1 and SOCS1 are frequently mutated in germinal center B-cell (GCB)-derived lymphomas such as follicular lymphoma and GCB-type DLBCL.18â&#x20AC;&#x201C;21 DUSP2 is less frequently mutated in malignant lymphomas; however, DUSP2 mutations have been described in DLBCL,15,22,23 primary mediastinal B-cell lymphoma24 and chronic lymphocytic leukemia.25 In contrast, to date, JUNB has been reported, to our knowledge, to be relatively specifically mutated in NLPHL.12 The total percentage of samples of hematopoietic and lymphoid neoplasms with mutations in JUNB in the COSMIC database (https://cancer.sanger.ac.uk/cosmic) is 0.09% (3 of 3516 samples tested). Two of these cases had non-synonymous mutations: one DLBCL and one lymphoid neoplasm not otherwise specified (COSMIC ID COSS2121012). The third case, a cutaneous T-cell lymphoma of mycosis fungoides type,26 presented a synonymous JUNB mutation. In the whole exome sequencing study on classical Hodgkin lymphoma by Tiacci et al.27, only one of 34 cases presented a non-synonymous JUNB mutation. Therefore, the high frequency of non-synonymous JUNB mutations, including a relatively frequent occurrence of JUNB nonsense mutations in both NLPHL and THRLBCL, is a hallmark of these lymphoma entities.

Mutations in the four genes JUNB, DUSP2, SGK1 and SOCS1 were more frequent in THRLBCL and the histopathological NLPHL variants than in typical NLPHL. Since we performed ultra-deep targeted resequencing of whole tissue DNA and allowed allele frequencies of mutations to range within the expected tumor cell content (0.1haematologica | 2019; 104(2)

10%), this approach might more sensitively identify subclonal variants in THRLBCL than in NLPHL due to the slightly higher tumor cell content in THRLBCL compared to NLPHL (median 10% versus 3%). However, the tumor cell content did not differ significantly between histopathological NLPHL variants and typical NLPHL (3% versus 2%, respectively) and thus does not explain the higher number of mutations in histopathological NLPHL variants than in typical NLPHL. SOCS1 is a known target of aberrant SHM,18 and SHM of immunoglobulin genes is known to be ongoing in the tumor cells of NLPHL and THRLBCL.6,7,14 The mutation patterns of JUNB, SGK1, DUSP2 and SOCS1 observed in the tumor cells of NLPHL and THRLBCL suggest that mutations in these four genes are most likely the result of aberrant SHM, which is likely contributing importantly to the development of these lymphomas. The genes DUSP2 and SGK1 were previously identified as potential targets of aberrant SHM in GCB-type DLBCL,15 but are less frequently mutated in GCB-type DLBCL than in NLPHL and THRLBCL. SHM activity is reported to start closely downstream from the transcriptional start site extending up to 2 kb into the gene,28 which is in line with an enrichment of SNVs in SHM hotspot motifs in the three relatively small genes JUNB, DUSP2 and SOCS1 (<2.5 kb). The more frequent occurrence of SNVs in SHM hotspot sites in THRLBCL and histopathological NLPHL variants might be related to a longer and/or stronger exposure of the tumor cells to the SHM machinery. Aberrant activity of the SHM machinery may be caused by loss of target specificity and 335

B. Schuhmacher et al.

lead to insertion of mutations into genomic regions showing active transcription and a favorable epigenetic environment.15,29,30 This might also contribute to the more aggressive clinical behavior of these entities if a high rate of aberrant SHM affects further target genes that were not profiled here.

According to our data, SOCS1, previously shown to be mutated in NLPHL,12,31 was mutated in both histopathological NLPHL variants and THRLBCL. SOCS1 was not mutated in the typical NLPHL cases of this study, as observed previously.12 This is likely related to the low number of typical NLPHL cases investigated. On the other hand, one may speculate that SOCS1 may act as a potential driver gene indicating disease progression and may thus represent a progression driver. However, the histopathological growth patterns of the NLPHL cases investigated were not considered in the study by Mottok et al.31 Despite the fact that typical NLPHL represents the majority of NLPHL cases, in our experience most frozen NLPHL samples are acquired from histopathological NLPHL variants. One reason for this may be because the patients with histopathological NLPHL variants present with more advanced disease and are more likely to present in a specialized medical center where frozen tissue can be preserved. A potential role of SOCS1 as a driver towards disease progression is further supported by previous reports, in which SOCS1 missense mutations were found to be related to a more aggressive clinical behavior in a cohort of DLBCL cases treated with CHOP-like regimens.32 The prognostic impact of SOCS1 mutations should therefore be investigated in a larger cohort of NLPHL. Moreover, as the ultra-deep targeted sequencing approach used here is not very reliable in the identification of structural variants, these were not considered here. Given that a high number of SOCS1 aberrations have been reported to be insertions/deletions31,32 we, therefore, likely underestimate the frequency of SOCS1 mutations.

Mutations in the acetyltransferase CREBBP were usually not a result of aberrant SHM. They frequently affected the KIX domain that mediates binding to transcription factors,33 the HAT domain that performs histone acetyltransferase activity and the C-terminal Q-rich domain,

References 1. Swerdlow SH, Campo E, Harris NL, et al.

WHO Classification of Tumors of Haematopoietic and Lymphoid Tissues.

4th edition. Lyon: IARC; 2008.

2. Swerdlow SH, Campo E, Harris NL, et al.

WHO Classification of Tumors of Haematopoietic and Lymphoid Tissues.

Revised 4th edition. Lyon: IARC; 2017.

3. Van Loo P, Tousseyn T, Vanhentenrijk V, et al.

T-cell/histiocyte-rich large B-cell lymphoma shows transcriptional features suggestive of a tolerogenic host immune response.

Haematologica. 2010;95(3):440–448.

4. Hartmann S, Döring C, Jakobus C, et al.

Nodular lymphocyte predominant Hodgkin lymphoma and T cell/histiocyte rich large B cell lymphoma--endpoints of a spectrum of one disease? PloS One. 2013;8(11):e78812.

336

which is part of a transactivation domain,33 consistent with the functional consequences reported for DLBCL and follicular lymphoma.34 Notably, two mutations in the HAT domain of CREBBP detected in DLBCL (nonsense R1341X and R1360X)34 also occurred at the same residue in a patient with NLPHL variant pattern E (R1341X) and a patient with THRLBCL (R1360P), suggesting that the HAT domain is under inactivating pressure in these GCB-cell derived malignancies. Thus, in addition to aberrant SHM, further transforming events are likely required for the development of NLPHL and THRLBCL. Since the gene panel applied in the present study was based on genes previously identified in composite lymphomas of NLPHL and DLBCL, we were not in a position to identify novel genes that are recurrently mutated in THRLBCL but not or only rarely in NLPHL. This is a clear limitation of our study. Thus, genome-wide mutation studies are warranted to comprehensively determine the mutational landscape of THRLBCL. Nevertheless, the fact that the most frequently mutated genes in NLPHL are also recurrently mutated in THRLBCL makes a strong point regarding the close relationship of these malignancies, which had been proposed in earlier studies based on histopathological features, related gene expression profiles, and similar genomic imbalances.4,8,9 Perhaps de novo THRLBCL could principally represent a transformation from NLPHL, and thus share key mutations that were acquired in the earlier NLPHL lymphomagenesis.

Considering the clinical presentation of the patients, one could speculate that typical and histopathological NLPHL variants have a similar relationship to THRLBCL, as is observed in follicular lymphoma grade 1/2 to 3a and transformation into DLBCL. Acknowledgments The authors would like to thank Ralf Lieberz, Smaro Soworka, Nina Becker, Elena Hartung and the EMBL GeneCore sequencing team and the EMBL IT unit for their excellent technical assistance.

Funding This project was supported by the Deutsche Forschungsgemeinschaft (grant HA6145/1-2 and HA6145/3-1).

5. Achten R, Verhoef G, Vanuytsel L, De Wolf-Peeters C. Histiocyte-rich, T-cell-rich B-cell lymphoma: a distinct diffuse large Bcell lymphoma subtype showing characteristic morphologic and immunophenotypic features. Histopathology. 2002;40(1):31–45.

6. Braeuninger A, Küppers R, Strickler JG, Wacker HH, Rajewsky K, Hansmann ML. Hodgkin and Reed-Sternberg cells in lymphocyte predominant Hodgkin disease represent clonal populations of germinal center-derived tumor B cells. Proc Natl Acad Sci U S A. 1997;94(17):9337–9342.

7. Bräuninger A, Küppers R, Spieker T, et al.

Molecular analysis of single B cells from Tcell-rich B-cell lymphoma shows the derivation of the tumor cells from mutating germinal center B cells and exemplifies means by which immunoglobulin genes are modified in germinal center B cells.

Blood. 1999;93(8):2679–2687.

8. Brune V, Tiacci E, Pfeil I, et al. Origin and

pathogenesis of nodular lymphocyte-predominant Hodgkin lymphoma as revealed by global gene expression analysis. J Exp Med. 2008;205(10):2251–2268.

9. Hartmann S, Döring C, Vucic E, et al. Array comparative genomic hybridization reveals similarities between nodular lymphocyte predominant Hodgkin lymphoma and T cell/histiocyte rich large B cell lymphoma.

Br J Haematol. 2015;169(3):415–422.

10. Fan Z, Natkunam Y, Bair E, Tibshirani R, Warnke RA. Characterization of variant patterns of nodular lymphocyte predominant Hodgkin lymphoma with immunohistologic and clinical correlation. Am J Surg Pathol. 2003;27(10):1346–1356.

11. Hartmann S, Eichenauer DA, Plütschow A, et al. The prognostic impact of variant histology in nodular lymphocyte-predominant Hodgkin lymphoma: a report from the German Hodgkin Study Group (GHSG).

Blood. 2013;122(26):4246–4252; quiz 4292.

haematologica | 2019; 104(2)

JUNB, DUSP2, SGK1 and SOCS1 mutations in THRLBCL

12. Hartmann S, Schuhmacher B, Rausch T, et al. Highly recurrent mutations of SGK1, DUSP2 and JUNB in nodular lymphocyte predominant Hodgkin lymphoma.

Leukemia. 2016;30(4):844–853.

13. Greiner A, Tobollik S, Buettner M, et al.

Differential expression of activationinduced cytidine deaminase (AID) in nodular lymphocyte-predominant and classical Hodgkin lymphoma. J Pathol. 2005;205(5): 541–547.

14. Liso A, Capello D, Marafioti T, et al.

Aberrant somatic hypermutation in tumor cells of nodular-lymphocyte-predominant and classic Hodgkin lymphoma. Blood.

2006;108(3):1013–1020.

15. Khodabakhshi AH, Morin RD, Fejes AP, et al. Recurrent targets of aberrant somatic hypermutation in lymphoma. Oncotarget.

2012;3(11):1308–1319.

16. Pasqualucci L, Neumeister P, Goossens T, et al. Hypermutation of multiple proto-oncogenes in B-cell diffuse large-cell lymphomas. Nature. 2001;412(6844):341–346.

17. Mottok A, Woolcock B, Chan FC, et al.

Genomic alterations in CIITA are frequent in primary mediastinal large B cell lymphoma and are associated with diminished MHC class II expression. Cell Rep. 2015;13 (7):1418–1431.

18. Mottok A, Renné C, Seifert M, et al.

Inactivating SOCS1 mutations are caused by aberrant somatic hypermutation and restricted to a subset of B-cell lymphoma entities. Blood. 2009;114(20):4503–4506.

19. Morin RD, Mendez-Lago M, Mungall AJ, et

haematologica | 2019; 104(2)

20.

21.

22.

23.

24.

25.

26.

al. Frequent mutation of histone-modifying genes in non-Hodgkin lymphoma. Nature.

2011;476(7360):298–303.

Zhang J, Grubor V, Love CL, et al. Genetic heterogeneity of diffuse large B-cell lymphoma. Proc Natl Acad Sci U S A. 2013;110(4):1398–1403.

Pasqualucci L, Khiabanian H, Fangazio M, et al. Genetics of follicular lymphoma transformation. Cell Rep. 2014;6(1):130– 140.

Morin RD, Assouline S, Alcaide M, et al.

Genetic landscapes of relapsed and refractory diffuse large B-cell lymphomas. Clin Cancer Res. 2016;22(9):2290–2300.

Mareschal S, Dubois S, Viailly P-J, et al.

Whole exome sequencing of relapsed/refractory patients expands the repertoire of somatic mutations in diffuse large B-cell lymphoma.

Genes Chromosomes Cancer. 2016;55(3):251– 267.

Gunawardana J, Chan FC, Telenius A, et al.

Recurrent somatic mutations of PTPN1 in primary mediastinal B cell lymphoma and Hodgkin lymphoma.

Nat Genet.

2014;46(4):329–335.

Ljungström V, Cortese D, Young E, et al.

Whole-exome sequencing in relapsing chronic lymphocytic leukemia: clinical impact of recurrent RPS15 mutations.

Blood. 2016;127(8):1007–1016.

McGirt LY, Jia P, Baerenwald DA, et al.

Whole-genome sequencing reveals oncogenic mutations in mycosis fungoides.

Blood. 2015;126(4):508–519.

27. Tiacci E, Ladewig E, Schiavoni G, et al.

Pervasive mutations of JAK-STAT pathway genes in classical Hodgkin lymphoma.

Blood. 2018;131(22):2454–2465.

28. Storb U, Peters A, Klotz E, et al. Cis-acting sequences that affect somatic hypermutation of Ig genes. Immunol Rev. 1998;162153–160.

29. Qian J, Wang Q, Dose M, et al. B cell superenhancers and regulatory clusters recruit AID tumorigenic activity.

Cell.

2014;159(7):1524–1537.

30. Wang Q, Oliveira T, Jankovic M, et al.

Epigenetic targeting of activation-induced cytidine deaminase. Proc Natl Acad Sci U S A. 2014;111(52):18667–18672.

31. Mottok A, Renné C, Willenbrock K, Hansmann M-L, Bräuninger A. Somatic hypermutation of SOCS1 in lymphocytepredominant Hodgkin lymphoma is accompanied by high JAK2 expression and activation of STAT6. Blood. 2007;110(9): 3387–3390.

32. Schif B, Lennerz JK, Kohler CW, et al.

SOCS1 mutation subtypes predict divergent outcomes in diffuse large B-cell lymphoma (DLBCL) patients. Oncotarget.

2013;4(1):35–47.

33. Kwok RP, Lundblad JR, Chrivia JC, et al.

Nuclear protein CBP is a coactivator for the transcription factor CREB. Nature.

1994;370(6486):223–226.

34. Pasqualucci L, Dominguez-Sola D, Chiarenza A, et al. Inactivating mutations of acetyltransferase genes in B-cell lymphoma. Nature. 2011;471(7337):189–195.

337

ARTICLE Ferrata Storti Foundation

Non-Hodgkin Lymphoma

T-cell inflamed tumor microenvironment predicts favorable prognosis in primary testicular lymphoma

Suvi-Katri Leivonen,1,2 Marjukka Pollari,1,3 Oscar Brück,4 Teijo Pellinen,5 Matias Autio,1,2 Marja-Liisa Karjalainen-Lindsberg,6 Susanna Mannisto,1,2 Pirkko-Liisa Kellokumpu-Lehtinen,3,7 Olli Kallioniemi,5,8 Satu Mustjoki4,9 and Sirpa Leppä1,2

Haematologica 2019 Volume 104(2):338-346

Research Program Unit, Medical Faculty, University of Helsinki, Finland; 2Department of Oncology, Comprehensive Cancer Center, Helsinki University Hospital, Finland; 3 Department of Oncology, Tampere University Hospital, Finland; 4Hematology Research Unit Helsinki, Department of Clinical Chemistry and Hematology, University of Helsinki, Finland; 5Institute for Molecular Medicine Finland (FIMM), Helsinki, Finland; 6 Department of Pathology, Helsinki University Hospital, Finland; 7University of Tampere, Faculty of Medicine and Life Sciences, Finland; 8Science for Life Laboratory, Karolinska Institutet, Department of Oncology and Pathology, Solna, Sweden and 9Department of Hematology, Comprehensive Cancer Center, Helsinki University Hospital, Finland 1

ABSTRACT

P

Correspondence: sirpa.leppa@helsinki.fi

Received: June 18, 2018.

Accepted: September 19, 2018.

Pre-published: September 20, 2018. doi:10.3324/haematol.2018.200105 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/338 ©2019 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.

338

rimary testicular lymphoma is a rare lymphoid malignancy, most often, histologically, representing diffuse large B-cell lymphoma.

The tumor microenvironment and limited immune surveillance have a major impact on diffuse large B-cell lymphoma pathogenesis and survival, but the impact on primary testicular lymphoma is unknown.

Here, the purpose of the study was to characterize the tumor microenvironment in primary testicular lymphoma, and associate the findings with outcome. We profiled the expression of 730 immune response genes in 60 primary testicular lymphomas utilizing the Nanostring platform, and used multiplex immunohistochemistry to characterize the immune cell phenotypes in the tumor tissue. We identified a gene signature enriched for T-lymphocyte markers differentially expressed between the patients. Low expression of the signature predicted poor outcome independently of the International Prognostic Index (progression-free survival: HR=2.810, 95%CI: 1.228-6.431, P=0.014; overall survival: HR=3.267, 95%CI: 1.406-7.590, P=0.006). The T-lymphocyte signature was associated with outcome also in an independent diffuse large B-cell lymphoma cohort (n=96). Multiplex immunohistochemistry revealed that poor survival of primary testicular lymphoma patients correlated with low percentage of CD3+CD4+ and CD3+CD8+ tumor-infiltrating lymphocytes (P<0.001). Importantly, patients with a high T-cell inflamed tumor microenvironment had a better response to rituximabbased immunochemotherapy, as compared to other patients.

Furthermore, loss of membrane-associated human-leukocyte antigen complexes was frequent and correlated with low T-cell infiltration. Our results demonstrate that a T-cell inflamed tumor microenvironment associates with favorable survival in primary testicular lymphoma. This further highlights the importance of immune escape as a mechanism of treatment failure.

Introduction Primary testicular lymphoma (PTL) is a rare and aggressive lymphoid malignancy; the majority of the cases are diffuse large B-cell lymphomas (DLBCL).1 Patients with PTL have a substantial risk of recurrence and a tendency of lymphoma involvement in additional extranodal sites, such as contralateral testis and the central nervous system (CNS). Most cases are classified as non-germinal center (nonGC) cell-of-origin subtype, which may partially account for the aggressive nature of the disease. The current standard of care consists of rituximab, cyclophoshaematologica | 2019; 104(2)

Immunoprofiling and survival in PTL

phamide, doxorubicin, vincristine, and prednisone (RCHOP) immunochemotherapy, CNS prophylaxis, and locoregional radiotherapy or surgery of the remaining contralateral testicle. While the addition of rituximab to systemic chemotherapy has improved the overall prognosis of DLBCL, treatment responses in PTL have not been equally evident.1,2 The tumor microenvironment (TME) and limited immune surveillance play important roles in the lymphoma pathogenesis and survival.3,4 The TME of B-cell lymphomas consists of immune cells [e.g. T cells, macrophages, natural killer (NK) cells], stromal cells, blood vessels, and extracellular matrix. The TME may confer effective tumor recognition and clearance, but on the other hand, can also provide a protective niche for the tumor cells, and facilitate cell proliferation and survival.5 The interactions between lymphoma cells and the TME contribute to the ability of tumor cells to escape host immune surveillance; these mechanisms include loss of expression of human leukocyte antigen (HLA) class I and II molecules that interferes with correct tumor cell recognition by the cellular immune system and may provide tumor cells with an escape mechanism.6 Several studies have reported loss of HLA I and II expression in B-cell lymphomas, including PTL.7-11 Another immune escape mechanism is the expression of immune checkpoint molecules, such as programmed cell death 1 (PD-1) and the cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4).6 In this study, we aimed to characterize cellular and molecular immunological profiles in PTL, and associate the findings with outcome. We identified a gene signature enriched for T-cell- and NK-cell-related genes, the low expression of which predicts high risk of recurrence and mortality in patients with PTL and DLBCL. Our results emphasize the key role of the TME and immune escape in regulating therapy resistance in aggressive B-cell lymphomas.

Methods Patients' characteristics and samples The study consisted of formalin-fixed paraffin-embedded (FFPE) samples from primary orchiectomy of 60 patients diagnosed with PTL in 1993-2013 (Table 1). Twenty-eight of the patients were diagnosed and treated in the pre-rituximab era with anthracyclin-based chemotherapy, whereas 32 of the patients were treated with rituximab-containing immunochemotherapy.

In addition, 34 patients received CNS prophylaxis, consisting mainly of intravenously administered high-dose methotrexate and/or high-dose cytarabine. Nineteen patients received treatment of the contralateral testicle (radiotherapy or orchiectomy). The cell-of-origin (COO) was determined by immunohistochemistry using the Hans algorithm.12 The study was approved by the Ethics Committees in Helsinki and Tampere, Finland, and by the Finnish National Authority for Medicolegal Affairs, and by the Institutional Review Boards of the institutes involved in the study.

The dataset of primary DLBCL from the Cancer Genome Characterization Initiative (CGCI) (the database of Genotypes and Phenotypes study accession: phs000532.v2.p1; n=96)13 was used for comparison (Table 1).

Immunoprofiling Panel (XT-CSO-HIP1-12, NanoString Technologies, Seattle, WA, USA). The detailed protocol is provided in the Online Supplementary Methods.

Immunohistochemistry Formalin-fixed paraffin-embedded tissue microarray (TMA) was utilized for the immunohistochemical (IHC) analyses. The protocol for HLA-ABC, HLA-DR, and β2 microglobulin staining is provided in the Online Supplementary Methods. Membranous staining in the majority (>90%) of tumor cells was scored as normal (highly positive). Cases with mixed cytoplasmic and membranous staining were scored as moderately positive. Cases with

Table 1. Patients’ characteristics of the primary testicular lymphoma (PTL) and diffuse large B-cell lymphomas (DLBCL).

Characteristics Number of patients Age <60 years ≥60 years Sex Men Women Molecular subgroup GCBa non-GCB nd Stage I-IIb III-IV IPI score 0-2 3-5 Elevated LDH CNS prophylaxis IV-prophylaxis IT-prophylaxis Treated with rituximab-containing regimen Primary CNS involvement Relapses Primary refractoryc CNS relapse/progression Systemic and CNS relapse Systemic relapse/progression Late relapse CNS relapse Systemic and CNS relapse Systemic relapse Deaths Lymphoma-specific Other

PTL cohort DLBCL cohort All n (%) All n (%) 60 (100)

96 (100)

16 (27) 44 (73)

41 (43) 55 (57)

60 (100)

63 (66) 33 (34)

15 (25) 40 (67) 5 (8)

55 (57) 41 (43)

39 (65) 21 (35)

18 (19) 78 (81)

42 (70) 18 (30) 22 (37) 34 (57) 31 (52) 8 (13) 32 (53) 2 (3) 18 (30) 11 (18) 2 (3) 2 (3) 7 (12) 7 (12) 2 (3) 0 (0) 5 (8) 25 (42) 16 (27) 9 (15)

25 (26) 71 (74) n/a n/a n/a n/a 91 (96) n/a 27 (28) n/a n/a n/a n/a n/a n/a n/a n/a 24 (25) 21 (22) 3 (3)

GCB; germinal-center B cell; nd: not determined; LDH: lactate dehydrogenase; n/a: not assigned; CNS: central nervous system; IV: intravenous; IT: intrathecal. bIsolated bilateral testicular involvement was defined as stage IAE. cRelapse/progression less than 12 months after the diagnosis and/or within 6 months from the end of primary treatment. a

Gene expression analysis Gene expression from the FFPE samples of 60 PTL patients was profiled with the Nanostring nCounter Human PanCancer haematologica | 2019; 104(2)

339

S-K. Leivonen et al.

no membranous staining were scored as negative. When determining the triple-positive cases, the highly and moderately positive groups were merged. Scoring was performed independently by MA and SMa. Multiplex immunohistochemistry (mIHC) using a panel with antibodies for CD3 (clone EP449E, Abcam), CD4 (clone EPR6855, Abcam), CD8 (clone C8/144B, Abcam), CD56 (clone MRQ-42, Cell Marque, Rocklin, CA, USA) was performed as

previously described.14 Further details are provided in the Online Supplementary Methods. The digital image analysis platform CellProfiler 2.1.215 was used for cell segmentation, intensity measurements, and immune cell classification. Marker co-localization was computed with the single-cell analysis software FlowJo v.10 (FlowJo LLC.). Spots with less than 5000 cells were excluded from analysis and duplicate spots from the same patient merged.

Table 2. Pathways enriched among T-lymphocyte signature genes.

P

Term h_thelperPathway:T Helper Cell Surface Molecules h_tcytotoxicPathway:T Cytotoxic Cell Surface Molecules h_no2il12Pathway:NO2-dependent IL 12 Pathway in NK cells h_ctlPathway:CTL mediated immune response against target cells h_tcraPathway:Lck and Fyn tyrosine kinases in initiation of TCR Activation h_cskPathway:Activation of Csk by cAMP-dependent Protein Kinase Inhibits Signaling through the T Cell Receptor hsa05150:Staphylococcus aureus infection h_tcapoptosisPathway:HIV Induced T Cell Apoptosis hsa04514:Cell adhesion molecules (CAMs)

9.74E-04 9.74E-04 0.001542 0.003803 0.013718 0.013718 0.022019 0.028495 0.048673

Figure 1. Hierarchical clustering reveals gene signatures differentially expressed in primary testicular lymphoma (PTL) patients. The expression of PanCancer Immune profiling panel genes was assayed by Nanostring nCounter from 60 PTL samples. The data were z-score transformed and visualized by unsupervised hierarchical clustering using Euclidean distance. Relative levels of gene expression are depicted according to the color scale shown. Each row represents a different gene and each column a different sample. Three main gene signatures were discovered: T-lymphocyte signature, and Cytokine I and II signatures.

340

haematologica | 2019; 104(2)

Immunoprofiling and survival in PTL

Statistical analysis Statistical analyses were performed with IBM SPSS v.24.0 (IBM, Armonk, NY, USA). The χ2 test was used to assess differences in categorical variables. Mann-Whitney U and Kruskal-Wallis tests were used to compare differences between two or more groups, respectively. The Kaplan-Meier method was used to estimate survival rates (log-rank test). Cox univariate regression analysis was performed to study the prognostic value of the factors.

Multivariate analyses were performed according to the Cox proportional hazards regression model using categorical data. Overall survival (OS) and disease-specific survival (DSS) were determined as the interval from diagnosis to death from any cause or death due to lymphoma, respectively. Progression-free survival (PFS) was defined as the period between diagnosis and progression or death from any cause. Correlation analyses were performed with Spearman rank analysis. All comparisons were two-tailed, and P<0.05 was considered significant.

Results Unsupervised hierarchical clustering reveals genes differentially expressed between primary testicular lymphoma patients We profiled the expression of 730 immune-associated

genes and 40 housekeeping genes from the FFPE samples of 60 primary testicular DLBCLs utilizing the Nanostring PanCancer Immune Profiling Panel. Unsupervised hierarchical clustering revealed three gene signatures differentially expressed between the patients (Figure 1). We identified a large cluster of genes, which clearly separated the patients. The core signature contained 121 genes (Online Supplementary Table S1), which were enriched for T-cell and NK-cell markers and signaling (e.g. CD3D/E/G, CD4, CD8A/B, ITGB2, PRF1, GZMA/B/H/M/K, KLRB1/G1/K1) (Online Supplementary Tables S1 and S2). This signature was named “T-lymphocyte signature”. In addition, two minor cytokine signatures were recognized. “Cytokine signature I” included 44 genes, enriched for cytokines (e.g.

CSF3, CSF2, IL3, IFNA1, IL5, IL11, IL2) (Online Supplementary Table S2), whereas “Cytokine signature II” included 25 genes, enriched for both cytokines and cytokine receptors (e.g. IL4, IL17A, TNFRSF11A, IL17B, TNFSF11, PPBP, IL9, CXCR1, TNFSF18, CCL28) (Online Supplementary Table S3).

T-lymphocyte signature is associated with survival in patients with PTL To study the association of the gene signatures with survival, we re-clustered samples according to the expression

A

B

C

D

E

F

Figure 2. T-lymphocyte signature is associated with survival. (A) The gene expression data from 60 primary testicular lymphoma (PTL) patients were re-clustered according to the expression of the T-lymphocyte signature genes. The data clustered into three groups: Group 1: low expression; Group 2: intermediate expression; Group 3: high expression. (B) Kaplan-Meier (log-rank test) survival plots depict progression-free survival (PFS) and overall survival (OS) in the three patient groups.

(C and D) Kaplan-Meier plots show the PFS and OS of PTL patients stratified for the treatment with rituximab-containing regimen (C) versus no rituximab (D). (E and F) RNA-seq data from the CGCI cohort with 96 de novo DLBCL cases was clustered based on the T-lymphocyte signature gene expression. This divided the patients into two groups with higher (Group 2) and lower (Group 1) expression (E). Kaplan-Meier plots depict survival differences between the two groups (F).

haematologica | 2019; 104(2)

341

S-K. Leivonen et al.

of the signature genes. Re-clustering the data based on the expression of the 121 genes of the T-lymphocyte signature separated the patients into three distinct groups with high, intermediate and low expression (Figure 2A and Online Supplementary Figure S1A). Interestingly, patients with low expression of the T-lymphocyte signature genes (n=12) had a significantly worse PFS, OS, and DSS in comparison to patients with intermediate (n=32) and high (n=16) gene expression (log-rank P=0.041, P=0.009, and P=0.033, respectively) (Figure 2B and Online Supplementary Figure S2A). In Cox multivariate analysis, low expression of Tlymphocyte signature had prognostic impact on survival independently of the International Prognostic Index (IPI) (PFS: HR=2.810, 95%CI: 0.228-6.431, P=0.014; OS: HR=3.267, 95%CI: 1.406-7.590, P=0.006; and DSS: HR=2.910, 95%CI: 1.004-8.436, P=0.049). This was evident for PFS and OS also with the individual IPI factors (Online Supplementary Table S4). Baseline characteristics, including age, molecular subgroup, stage, and IPI were equally distributed in the subgroups with higher and lower T-lymphocyte signature expression (Online Supplementary Table S5). When the survival analyses were adjusted according to treatment, adverse prognostic impact of the low expression of T-lymphocyte signature on outcome was particularly evident in the patients treated with the rituximab-containing regimen (Figure 2C and D and Online Supplementary Figure S2B).

At the individual gene level, 72 genes from the T-lymphocyte signature were significantly (P<0.05) associated with survival in Cox univariate analysis with continuous variables and had lower expression in the poor prognosis patient group (Online Supplementary Table S6). These included, for example, T-cell surface markers (CD3D/E/G, CD4, CD8A/B), cytolytic factors (PRF1, GZMA/K/M), chemokines (CCL2/3/4/5), and killer-cell lectin-like receptors (KLRB1, KLRG1, KLRK1).

For the Cytokine I signature, no association with survival was found (data not shown). However, when patients were re-clustered according to the expression of the 25 genes of the Cytokine II signature, they could be separated into two groups with higher and lower expression. The group with higher expression of the Cytokine II signature had a shorter 5-year PFS as compared to those with low or no expression of the signature (36% vs. 66%; P=0.005) (Online Supplementary Figure S3). In Cox multivariate analysis with IPI, the Cytokine II signature was also found to be an independent predictor of PFS (HR=3.393, 95%CI: 1.531-7.521; P=0.003). Baseline characteristics, including age, molecular subgroup, stage and IPI were equally distributed in the subgroups with higher and lower expression of the Cytokine II signature (data not shown). In general, the absolute expression levels of the Cytokine II signature genes were low (Online Supplementary Figure S1B).

Low expression of the T-lymphocyte signature is associated with poor outcome in patients with de novo diffuse large B-cell lymphomas Next, we tested whether the signatures could be identified also from other B-cell lymphomas. To this end, we used RNA-sequencing data from 96 primary DLBCL patients from the Cancer Genome Characterization Initiative (CGCI) cohort (Table 1). Following hierarchical clustering of the gene expression of the T-lymphocyte signature, a subgroup of patients with low expression of the signature was identified (Figure 2E and Online 342

Supplementary Figure S1C). This group of patients had shorter survival as compared to patients with higher expression of the signature genes (PFS: P=0.007, OS: P=0.034) (Figure 2F). In a multivariate analysis with IPI, low T-lymphocyte signature remained an independent prognostic factor for PFS (HR=2.560, 95%CI: 1.151-5.695; P=0.021). The baseline characteristics were equally distributed between the patient groups (Online Supplementary Table S5). The results indicate that the T-lymphocyte signature identified from the PTL cohort has prognostic impact on survival also in patients with de novo DLBCL. On the contrary, genes from the cytokine signatures were neither differentially expressed between the patients or associated with survival in DLBCL (data not shown). As in PTL, the absolute expression levels of the cytokine signaTable 3. Univariate and multivariate Cox regression survival analysis of multiplex immunohistochemistry data.

Univariate PFS CD3+ CD3+CD4+ CD3+CD8+ OS CD3+ CD3+CD4+ CD3+CD8+ DSS CD3+ CD3+CD4+ CD3+CD8+ Multivariate PFS CD3+ IPI CD3+CD4+ IPI CD3+CD8+ IPI OS CD3+ IPI CD3+CD4+ IPI CD3+CD8+ IPI DSS CD3+ IPI CD3+CD4+ IPI CD3+CD8+ IPI

HRa

95% CI

P

3.675 3.766 2.738

1.195-11.299 1.225-11.573 0.870-8.614

0.023 0.021 0.085

5.461 4.813 3.957

1.552-19.214 1.355-17.100 1.102-14.212

0.008 0.015 0.035

9.889 9.495 6.512

1.235-79.200 1.186-76.004 0.783-54.148

0.031 0.034 0.083

4.706 4.476 4.114 4.107 2.861 3.620

1.494-14.829 1.969-10.174 1.327-12.755 1.797-9.387 0.906-9.041 1.653-7.928

0.008 <0.001 0.014 0.001 0.073 0.001

7.265 5.639 5.244 4.747 4.024 4.125

2.007-26.302 2.382-13.348 1.463-18.801 2.003-11.250 1.116-14.512 1.816-9.369

0.003 <0.001 0.011 <0.001 0.033 0.001

15.219 16.477 10.150 13.623 6.744 11.576

1.804-128.374 4.782-56.777 1.243-82.889 3.987-46.546 0.799-56.923 3.562-37.616

0.012 <0.001 0.031 <0.001 0.079 <0.001

IPI: International Prognostic Index. aHR: hazard Ratio; IPI: International Prognostic Index. CI: Confidence Interval. PFS: progression-free survival; OS: overall survival; DSS: disease-specific survival;

haematologica | 2019; 104(2)

Immunoprofiling and survival in PTL

A

B

Figure 3. Lower number of tumor-infiltrating T cells is associated with poor survival in primary testicular lymphoma (PTL). (A) Representative images (high, intermediate, low) from mIHC analysis of PTL tumor-associate macrophages probed with a 4-plex panel of T-cell markers. Blue: CD3; red: CD8; white: CD4; green: CD56; gray: DAPI. Scale bars 50 µm (upper panel) and 20 mm (lower panel). Images from individual channels are presented in Online Supplementary Figure S4. (B) Boxplots visualizing the expression of CD3+, CD3+CD4+, and CD3+CD8+ lymphocytes in the three groups based on the T-lymphocyte signature (1: poor prognosis, 2: intermediate, and 3: better prognosis). Statistical significance was determined by Kruskall-Wallis test.

ture genes were low (data not shown), suggesting that they are not clinically relevant.

Low amount of tumor-infiltrating T cells is associated with poor outcome in patients with PTL To study the presence of different T-cell and NK-cell subtypes in the tumor milieu, we performed mIHC with CD3, CD4, CD8, and CD56 cell surface markers on a TMA of 60 PTL patients (Figure 3A and Online Supplementary Figure S4). The relative levels of tumor-infiltrating CD3−CD56+ cells were very low (median 0.3%, range 0.0-52.7%), and there were no significant differences in the amount of CD56+ cells between the good and poor prognosis groups (data not shown). However, mIHC analysis confirmed significantly (P<0.001) lower proportions of CD3+, CD3+CD4+, and CD3+CD8+ lymphocytes in patients with a poor prognosis in comparison to patients with a favorable outcome (Figure 3B). Expression of CD3, CD4, and CD8 at the protein level correlated with the corresponding gene expressions (Spearman P=0.779, P<0.001; P=0.645, P<0.001; and P=0.768, P<0.001, respectively). Furthermore, low amounts of tumor-infiltrating CD3+, CD3+CD4+, and CD3+CD8+ lymphocytes were associated with a poor outcome (Online Supplementary Figure S5), which was independent of IPI (Table 3).

Consistent with the findings on the gene expression level, an adverse prognostic impact of the low tumor-infiltrating CD4+ and CD8+ T-cell counts was particularly evident in patients treated with the rituximab-containing regimen (Online Supplementary Figure S6).

Loss of membrane-associated HLA class I and II expression is frequent in PTL Loss or aberrant expression of HLA molecules may cause tumor cells to escape from immunosurveillance.

Here, the expression of HLA I and HLA II genes was lower in the poor prognosis group of PTL patients (P<0.05) (Online Supplementary Figure S7A). This was evihaematologica | 2019; 104(2)

Table 4. HLA and β2M immunohistochemical staining results.

Negative Moderately positive Highly positive

β2M %

HLA-ABC %

HLA-DR %

31.7 55.0 13.3

18.3 53.3 28.3

53.3 30.0 16.7

dent also in the de novo DLBCL cohort (Online Supplementary Figure S7B).

To study the subcellular localization of HLA class I and II, we analyzed HLA-DR, HLA-ABC and β2M, a component of HLA I, immunohistochemically (Figure 4A). A minority of the PTL cases showed highly positive membrane staining for HLA-DR, HLA-ABC, and β2M (17%, 28%, and 13%, respectively) (Table 4). Interestingly, triple-positive membrane staining of HLA I, HLA II and β2M was associated with a higher number of tumor-infiltrating CD3+, CD3+CD4+, and CD3+CD8+ lymphocytes (Figure 4B).

Discussion The TME plays an important role in lymphoma pathogenesis and patient prognosis.3 In this study, we aimed to characterize the immunological profiles and their association with outcome in patients with PTL. We found that a gene signature enriched for T-cell genes was associated with outcome in patients treated with chemotherapy and immunochemotherapy. We could further demonstrate that lower amounts of tumor-infiltrating CD4+ and CD8+ T cells was associated with a poor prognosis in patients with PTL. The impact was particularly evident among patients treated with rituximab-containing immunochemotherapy. The limitation of our study is the lack of a proper validation cohort due to the rare nature of 343

S-K. Leivonen et al.

A

B

Figure 4. Loss of membrane-associated HLA class I and II expression is frequent in primary testicular lymphoma (PTL) and correlates with low T-cell infiltration.

(A) Representative images of HLA-ABC, HLA-DR, and β2 microglobulin antibody of PTL tissue sections. The samples were scored as negative (neg), moderately positive (pos), and highly positive according to the membrane staining. Scale bar 50 mm. (B) Boxplots visualizing the correlation of HLA-ABC, HLA-DR and β2M triplepositive membrane staining (either moderate or high membrane staining) with T-cell numbers in PTL (n=14 and n=46 in positive and negative groups, respectively).

P-values were determined by Mann-Whitney test.

PTL. However, the T-lymphocyte signature had a prognostic impact in an independent cohort of de novo DLBCL patients treated with immunochemotherapy, demonstrating the importance of the signature genes also in other aggressive B-cell lymphomas. Our data extend previous findings on DLBCL patients treated with CHOP and RCHOP-like regimens.16-18 Together, the results emphasize the important role of the T-cell inflamed TME in regulating therapy resistance in PTL. T lymphocytes, mostly comprising CD4+ and CD8+ T cells, play a major role in cell-mediated immunity.

Lymphoma cells have been shown to escape immunosurveillance due to loss of expression or mislocalization of HLA I and II molecules.7-11,19 We found that reduced membranous staining of HLA I and II molecules and β2M correlated with lower T-cell infiltration, implying that defects in HLA complexes may impair the recruitment of the tumor-infiltrating T-cell subsets.

Indeed, our data suggest that immune escape does not only provide a mechanism for lymphoma pathogenesis, but also plays a role in promoting resistance to immunochemotherapy. We propose that lymphomas with inflammatory profile characterized by high content of tumor-infiltrating CD4+ and CD8+ T cells, “the hot tumors”, display pre-existing antitumor immune response.

In response to therapy, and particularly rituximab-containing regimen, tumor-infiltrating T cells are stimulated further to participate in immune response against lymphoma cells. In contrast, lymphomas that lack T-cell infiltration, “the cold tumors”, reflect the absence of pre-existing antitumor immunity and have a lower likelihood of having an optimal response to therapy. Consistent with our hypothesis, it has been shown that many chemotherapeutic drugs, including cyclophosphamide and doxorubicin, which are the main components in the CHOP regimen, can activate anti-tumor immune response by increasing immunogenicity of malignant cells as well as by directly relieving immunosuppressive networks.20 Rituximab and 344

other therapeutic CD20 antibodies can, in turn, further promote a long-term anti-tumor immune response, called the “vaccinal” effect, which is dependent on the presence of both CD4+ and CD8+ lymphocytes.21-23 Further studies should aim to characterize in more detail the underlying mechanisms for the loss of T-cell trafficking and infiltration. For example, differences in the mutational density between the T-cell inflamed “hot” and non-inflamed “cold” tumors might explain the loss of T cells in a subset of tumors. Additional gene expression profiling studies could provide information as to which genes and molecular pathways are differentially expressed or activated in the T-cell inflamed and non-inflamed tumors, and thus might mediate T-cell exclusion from the TME. For example, in melanoma and bladder cancer, the Wnt/β-catenin pathway has been shown to be causal in preventing T-cell activation and trafficking into the tumor microenvironment.24,25 In addition, it would be interesting to examine the distribution of other cell lineages and their phenotypes, and determine whether PTLs express other immunoregulatory molecules, including PD-L1, LAG-3 or IDO-1 and IDO-2, which can be targeted by novel therapies.

Recently, we showed that tumor-associated macrophages (TAMs) play a major role in PTL.26 High infiltration of PD-L1+CD68+ TAMs was associated with favorable survival and correlated with CD4+ and CD8+ T cells positive for PD1. In addition, both PD-L1+ TAMs and PD-1+ T cells emerged as independent indicators of survival for the patients with PTL. The interaction of PD-L1+ TAMs and PD-1+ T cells may modify the TME in PTL, or otherwise promote an anti-tumor immune response following immunochemotherapy.

The poor outcome associated with a low tumor-infiltrating T-cell content might provide the objective for therapeutic interventions. Cancer immunotherapy, especially protocols aiming to activate T-cell-mediated anti-tumor responses, and T-cell trafficking into the non-inflamed haematologica | 2019; 104(2)

Immunoprofiling and survival in PTL

tumors has indeed gained a lot of attention recently.

Antibodies targeting inhibitory molecules, such as PD-1, PD-L1, and CTLA-4 that regulate T-cell cytotoxicity have achieved impressive clinical responses.27-29 However, despite clear clinical responses, only a fraction of patients respond to treatment. Based on the findings from solid tumors, it has been suggested that the status of the HLA machinery affects the success of immunotherapy. For example, inactivating mutations in the β2M and HLA I complex in lung cancer caused a lack of response to immunotherapy because the cytotoxic cells were unable to find and attack the tumor cells.30 It was suggested that the recurrent inactivation of HLA I is an acquired mechanism for avoiding tumor immune recognition. In lymphomas, novel immunotherapies, including bispecific antibodies, cancer vaccines, and chimeric antigen receptor (CAR)-T cells may help T cells to identify and attack cancer cells.31-36 Another therapeutic implication that could be tested is to recuperate the loss of HLA II expression and the subsequent inability of T cells to recognize lymphoma cells by using histone deacetylase inhibitors.37,38 In conclusion, we have combined immune gene signatures and mIHC data with clinical information in a cohort of patients with PTL. We found three immune signatures, of which the T-lymphocyte signature was associated with survival both in PTL and DLBCL. The patients with T-cell-

References 1. Cheah CY, Wirth A, Seymour JF. Primary testicular lymphoma. Blood. 2014; 123(4):486-493.

2. Deng L, Xu-Monette ZY, Loghavi S, et al.

Primary testicular diffuse large B-cell lymphoma displays distinct clinical and biological features for treatment failure in rituximab era: a report from the International PTL Consortium. Leukemia. 2016; 30(2):361-372.

3. Scott DW, Gascoyne RD. The tumour microenvironment in B cell lymphomas.

Nat Rev Cancer. 2014;14(8):517-534.

4. Shain KH, Dalton WS, Tao J. The tumor microenvironment shapes hallmarks of mature B-cell malignancies. Oncogene.

2015;34(36):4673-4682.

5. Gajewski TF, Schreiber H, Fu YX. Innate and adaptive immune cells in the tumor microenvironment. Nat Immunol. 2013; 14(10):1014-1022.

6. Nicholas NS, Apollonio B, Ramsay AG. Tumor microenvironment (TME)-driven immune suppression in B cell malignancy.

Biochim Biophys Acta. 2016;1863(3):471482.

7. Booman M, Douwes J, Glas AM, et al.

Mechanisms and effects of loss of human leukocyte antigen class II expression in immune-privileged site-associated B-cell lymphoma. Clin Cancer Res. 2006; 12(9):2698-2705.

8. Challa-Malladi M, Lieu YK, Califano O, et al. Combined genetic inactivation of beta2Microglobulin and CD58 reveals frequent escape from immune recognition in diffuse large B cell lymphoma. Cancer Cell.

2011;20(6):728-740.

9. Nijland M, Veenstra RN, Visser L, et al.

HLA dependent immune escape mecha-

haematologica | 2019; 104(2)

10.

11.

12.

13.

14.

15.

16.

17.

inflamed TME had a significantly increased risk of progression and death independently of IPI. Furthermore, reduced membranous staining of HLA I and II correlated with low T-cell infiltration. Taken together, the results presented herein are novel and emphasize the importance of immune escape as a mechanism regulating therapy resistance in patients with PTL. Acknowledgments We thank Dr. Petri Auvinen and Lars Paulin (Institute of Biotechnology, University of Helsinki, Finland) for the Nanostring analyses. Anne Aarnio and Marika Tuukkanen are acknowledged for technical assistance. The DLBCL dataset is part of the Cancer Genomics Characterization Initiative (CGCI), supported by NCI contract N01-C0-12400 (http://cgap.nci.nih.gov/cgci.html).

Funding This work was supported by grants from the Academy of Finland (to SL); Finnish Cancer Foundation (to SL and SMu); Finnish Cancer Institute (to SMu); Sigrid Juselius Foundation (to SL and SMu); Gyllenberg Foundation (to SMu); University of Helsinki (to SL); Helsinki University Hospital (to SL and SMu); Pirkanmaa Cancer Society (to MP); Seppo Nieminen Foundation (to MP); Eino Saarinen Foundation (to MP); and Finnish Oncology Association (to MP).

nisms in B-cell lymphomas: Implications for immune checkpoint inhibitor therapy?

Oncoimmunology. 2017;6(4):e1295202.

Riemersma SA, Jordanova ES, Schop RF, et al. Extensive genetic alterations of the HLA region, including homozygous deletions of HLA class II genes in B-cell lymphomas arising in immune-privileged sites. Blood.

2000;96(10):3569-3577.

Riemersma SA, Oudejans JJ, Vonk MJ, et al.

High numbers of tumour-infiltrating activated cytotoxic T lymphocytes, and frequent loss of HLA class I and II expression, are features of aggressive B cell lymphomas of the brain and testis. J Pathol.

2005;206(3):328-336.

Hans CP, Weisenburger DD, Greiner TC, et al. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004;103(1):275-282.

Morin RD, Mendez-Lago M, Mungall AJ, et al. Frequent mutation of histone-modifying genes in non-Hodgkin lymphoma. Nature.

2011;476(7360):298-303.

Blom S, Paavolainen L, Bychkov D, et al.

Systems pathology by multiplexed immunohistochemistry and whole-slide digital image analysis. Sci Rep. 2017;7(1):15580.

Carpenter AE, Jones TR, Lamprecht MR, et al. CellProfiler: image analysis software for identifying and quantifying cell phenotypes. Genome Biol. 2006;7(10):R100.

Ansell SM, Stenson M, Habermann TM, Jelinek DF, Witzig TE. Cd4+ T-cell immune response to large B-cell non-Hodgkin's lymphoma predicts patient outcome. J Clin Oncol. 2001;19(3):720-726.

Keane C, Gill D, Vari F, Cross D, Griffiths L, Gandhi M. CD4(+) tumor infiltrating lymphocytes are prognostic and independent

18.

19.

20.

21.

22.

23.

24.

of R-IPI in patients with DLBCL receiving R-CHOP chemo-immunotherapy. Am J Hematol. 2013;88(4):273-276.

Keane C, Vari F, Hertzberg M, et al. Ratios of T-cell immune effectors and checkpoint molecules as prognostic biomarkers in diffuse large B-cell lymphoma: a populationbased study.

Lancet Haematol.

2015;2(10):e445-455.

Rimsza LM, Roberts RA, Miller TP, et al.

Loss of MHC class II gene and protein expression in diffuse large B-cell lymphoma is related to decreased tumor immunosurveillance and poor patient survival regardless of other prognostic factors: a follow-up study from the Leukemia and Lymphoma Molecular Profiling Project. Blood.

2004;103(11):4251-4258.

Galluzzi L, Buque A, Kepp O, Zitvogel L, Kroemer G. Immunological Effects of Conventional Chemotherapy and Targeted Anticancer Agents.

Cancer Cell.

2015;28(6):690-714.

Abes R, Gelize E, Fridman WH, Teillaud JL. Long-lasting antitumor protection by antiCD20 antibody through cellular immune response. Blood. 2010;116(6):926-934.

Battella S, Cox MC, La Scaleia R, et al.

Peripheral blood T cell alterations in newly diagnosed diffuse large B cell lymphoma patients and their long-term dynamics upon rituximab-based chemoimmunotherapy. Cancer Immunol Immunother.

2017;66(10):1295-1306.

Hilchey SP, Hyrien O, Mosmann TR, et al.

Rituximab immunotherapy results in the induction of a lymphoma idiotype-specific T-cell response in patients with follicular lymphoma: support for a "vaccinal effect" of rituximab. Blood. 2009;113(16):38093812.

Spranger S, Bao R, Gajewski TF.

345

S-K. Leivonen et al.

25.

26.

27.

28.

29.

346

Melanoma-intrinsic beta-catenin signalling prevents anti-tumour immunity. Nature.

2015;523(7559):231-235.

Sweis RF, Spranger S, Bao R, et al.

Molecular Drivers of the Non-T-cellInflamed Tumor Microenvironment in Urothelial Bladder Cancer. Cancer Immunol Res. 2016;4(7):563-568.

Pollari M, Bruck O, Pellinen T, et al. PD-L1+ tumor-associated macrophages and PD-1+ tumor infiltrating lymphocytes predict survival in primary testicular lymphoma.

Haematologica. 2018;103(11):1908-1914.

Goodman A, Patel SP, Kurzrock R. PD-1PD-L1 immune-checkpoint blockade in Bcell lymphomas. Nat Rev Clin Onc. 2017;14(4):203-220.

Robert C, Schachter J, Long GV, et al.

Pembrolizumab versus Ipilimumab in Advanced Melanoma. New Engl J Med. 2015;372(26):2521-2532.

Wolchok JD, Chiarion-Sileni V, Gonzalez R, et al. Overall Survival with Combined

30.

31.

32.

33.

34.

Nivolumab and Ipilimumab in Advanced Melanoma. New Engl J Med. 2017; 377(14):1345-1356.

Pereira C, Gimenez-Xavier P, Pros E, et al.

Genomic profiling of patient-derived xenografts for lung cancer identifies B2M inactivation impairing immunorecognition.

Clin Cancer Res. 2017;23(12):3203-3213.

Allegra A, Russo S, Gerace D, et al.

Vaccination strategies in lymphoproliferative disorders: Failures and successes. Leuk Res. 2015;39(10):1006-1019.

Brudno JN, Kochenderfer JN. Chimeric antigen receptor T-cell therapies for lymphoma.

Nat Rev Clin Oncol. 2018;15(1):31-46.

Goebeler ME, Knop S, Viardot A, et al.

Bispecific T-Cell Engager (BiTE) Antibody Construct Blinatumomab for the Treatment of Patients With Relapsed/Refractory NonHodgkin Lymphoma: Final Results From a Phase I Study. J Clin Oncol. 2016;34(10): 1104-1111.

Pishko A, Nasta SD. The role of novel

35.

36.

37.

38.

immunotherapies in non-Hodgkin lymphoma. Transl Cancer Res. 2017;6(1):93-103.

Neelapu SS, Locke FL, Bartlett NL, et al.

Axicabtagene Ciloleucel CAR T-Cell Therapy in Refractory Large B-Cell Lymphoma.

New Engl J Med. 2017;377(26):2531-2544.

Viardot A, Goebeler ME, Hess G, et al.

Phase 2 study of the bispecific T-cell engager (BiTE) antibody blinatumomab in relapsed/refractory diffuse large B-cell lymphoma. Blood. 2016;127(11):1410-1416.

Cycon KA, Mulvaney K, Rimsza LM, Persky D, Murphy SP. Histone deacetylase inhibitors activate CIITA and MHC class II antigen expression in diffuse large B-cell lymphoma. Immunology. 2013;140(2):259272.

Jiang Y, Ortega-Molina A, Geng H, et al.

CREBBP Inactivation Promotes the Development of HDAC3-Dependent Lymphomas. Cancer Discov. 2017;7(1):3853.

haematologica | 2019; 104(2)

ARTICLE

Non-Hodgkin Lymphoma

miR-497 suppresses cycle progression through an axis involving CDK6 in ALK-positive cells

Ferrata Storti Foundation

Coralie Hoareau-Aveilla,1,2,3* Cathy Quelen,1,2,3* Annabelle Congras,1,2,3#§ Nina Caillet,1,2,3#§ Delphine Labourdette,4 Christine Dozier,1,2,3 Pierre Brousset,1,2,3,5,6,7§ Laurence Lamant1,2,3,5,6,7§ and Fabienne Meggetto1,2,3,5,6,7§

Inserm, UMR1037 CRCT, F-31000 Toulouse, France; 2Université Toulouse III-Paul Sabatier, UMR1037 CRCT, F-31000 Toulouse, France; 3CNRS, ERL5294 CRCT, F-31000 Toulouse, France; 4Laboratoire d’Ingénierie des Systèmes Biologiques et des Procédés, Université de Toulouse, CNRS, INRA, INSA, Toulouse, France; 5Institut Carnot Lymphome-CALYM, F-31024 Toulouse, France; 6Laboratoire d’Excellence Toulouse Cancer-TOUCAN, F-31024 Toulouse, France and 7European Research Initiative on ALKRelated Malignancies, Cambridge, UK 1

Haematologica 2019 Volume 104(2):347-359

*CH-A, CQ, AC, NC and FB all contributed equally to this work. §Member of the Equipe Labellisée LIGUE 2017.

ABSTRACT

A

naplastic large-cell lymphoma, a T-cell neoplasm, is primarily a pediatric disease. Seventy-five percent of pediatric anaplastic large-cell lymphoma cases harbor the chromosomal translocation t(2;5)(p23;q35) leading to the ectopic expression of NPM-ALK, a chimeric tyrosine kinase. NPM-ALK consists of an N-terminal nucleophosmin (NPM) domain fused to an anaplastic lymphoma kinase (ALK) cytoplasmic domain. Pediatric NPM-ALK+ anaplastic large-cell lymphoma is often a disseminated disease and young patients are prone to chemoresistance or relapse shortly after chemotherapeutic treatment.

Furthermore, there is no gold standard protocol for the treatment of relapses. To the best of our knowledge, this is the first study on the potential role of the microRNA, miR-497, in NPM-ALK+ anaplastic largecell lymphoma tumorigenesis. Our results show that miR-497 expression is repressed in NPM-ALK+ cell lines and patient samples through the hypermethylation of its promoter and the activity of NPM-ALK is responsible for this epigenetic repression. We demonstrate that overexpression of miR-497 in human NPM-ALK+ anaplastic large-cell lymphoma cells inhibits cellular growth and causes cell cycle arrest by targeting CDK6, E2F3 and CCNE1, the three regulators of the G1 phase of the cell cycle. Interestingly, we show that a scoring system based on CDK6, E2F3 and CCNE1 expression could help to identify relapsing pediatric patients. In addition, we demonstrate the sensitivity of NPMALK+ cells to CDK4/6 inhibition using for the first time a selective inhibitor, palbociclib. Together, our findings suggest that CDK6 could be a therapeutic target for the development of future treatments for NPMALK+ anaplastic large-cell lymphoma.

fabienne.meggetto@inserm.fr

Introduction

©2019 Ferrata Storti Foundation

Anaplastic large cell lymphoma (ALCL) is an aggressive form of T-cell nonHodgkin lymphoma (NHL) with a constant membrane expression of the CD30 antigen, a cytokine receptor from the tumor necrosis factor receptor family.

Four distinct entities of ALCL are currently recognized based on the 2016 revised World Health Organization (WHO) lymphoma classification: 1) anaplastic lymphoma kinase (ALK)-positive(+) ALCL; 2) ALK-negative (ALK–) ALCL; 3) primary cutaneous ALCL; and 4) breast implant-associated ALCL.1 ALCL accounts for approximately 10-15% of all NHL in children but is rare in adults (1-2% of adult NHL). Pediatricians have adopted ALCL99 chemotherapy including low cumulative doses of methotrexate, as standard therapy.2 Patients with pediatric ALCL have a favorable prognosis; however, systemic ALCL is haematologica | 2019; 104(2)

Correspondence:

Received: April 9, 2018.

Accepted: September 21, 2018.

Pre-published: September 27, 2018. doi:10.3324/haematol.2018.195131 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/347

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.

347

C. Hoareau-Aveilla et al.

often a disseminated disease and pediatric patients are prone to chemoresistance or to relapse after stopping the treatment. All relapses occur in patients with advanced disease and there is still no gold standard for treating relapses.

In children and adolescents, more than 90% of systemic ALCL cases harbor a chromosomal translocation t(2;5)(p23;q35) that leads to the expression of the NPMALK chimeric tyrosine kinase. This kinase consists of an N-terminal nucleophosmin (NPM) domain fused to the ALK cytoplasmic domain.3,4 ALK is a tyrosine kinase receptor whose expression is normally restricted to neural progenitor cells during development.5 At diagnosis, ALCL patients are grouped into risk groups based on the presence of NPM-ALK transcripts in peripheral blood and/or bone marrow6 and the quantification of anti-ALK antibody in peripheral blood,7 and chemotherapy treatment is adapted accordingly.

In ALCL, oncogenic NPM-ALK signaling is mediated by different pathways, such as RAS/ERK and JNK/STAT pathways, which play major roles in lymphomagenesis by controlling key cellular processes such as cell cycle progression.8,9 Studies showed increased proliferation of Rat1a fibroblasts upon ectopic expression of NPM-ALK as a consequence of accelerated entry into the S phase of the cell cycle. Furthermore, this was associated with a significant upregulation of cyclin A and cyclin D1 and increased levels of the mitogenic proto-oncogenes Jun, Fos, and Myc.10 Moreover, in 293T and Jurkat cells, it has been shown that forced expression of NPM-ALK induces JNK (JUN N-terminal kinase) and Jun phosphorylation, thereby up-regulating the activity of AP1 transcription factors. This results in uncontrolled cell cycle progression and cell growth due to the downregulation of p21 and the concomitant upregulation of cyclin A and cyclin D3.11 The ectopic expression of constitutively active NPM-ALK in Ba/F3 mouse cells leads to FOXO3A phosphorylation by Akt, upregulation of cyclin D2 expression and downregulation of p27 and Bim-1 expression to promote cell survival and cell cycle progression.12 In Ba/F3 mouse cells, NPM-ALK, via the PI3K/Akt pathway, also controls cell division cycle 25 A (Cdc25A), a key regulator of the G1 phase and the G1/S transition.13 Many microRNAs (miRNAs) modulate several major proliferation pathways by controlling critical regulators such as Cyclin-CDK complexes.14 miRNAs are singlestranded small non-coding RNAs that are pivotal in physiological and pathological processes such as development, cell proliferation and apoptosis. In general, by binding to specific targets with distinct degrees of complementarity, miRNAs exhibit a negative regulatory role at the post-transcriptional level through the inhibition of translation and/or degradation of their messenger RNA targets. There is growing evidence to show that differentially expressed miRNAs are associated with tumor types and cancer development.15 Indeed, several miRNAs display defective expression patterns in tumors, consequently altering oncogenic or tumor suppressive targets. miRNAs such as miR-16, miR-17-92, miR-21, miR-26a, miR-29a, miR-96, miR-101, miR135b, miR-146a, miR-150, miR-155 and miR-219 are dysregulated and serve as oncogenes or tumor suppressors in NPM-ALK+ ALCL.16-20 Most of these miRNAs have been found to be down-regulated (miR-16, miR-21, 348

miR-26a, miR-29a, miR-96, miR-101, miR-146a, miR150, miR-155 et miR-219) in NPM-ALK+ ALCL. Our laboratory showed, for the first time, that NPM-ALK+ ALCL cell lines and primary tissues express low levels of several miRNAs mediated by the hypermethylation of their gene promoter.17,21 Both NPM-ALK and STAT3 activities contributed to epigenetic silencing in NPMALK+ ALCL cell lines and biopsy specimens by up-regulating and recruiting DNMT1 to the promoter of miR29a, miR-125b and miR-150.17,19,21 The repressive methylation catalyzed by DNMT1 can be partially reversed by treatment with 5-aza-2â&#x20AC;&#x2122;-deoxycytidine (5-aza-dC, decitabine, Dacogen,ÂŽ SuperGen Inc., Dublin, CA, USA), a DNMT inhibitor. This DNA-demethylating agent has been shown to restore miR-497 expression, which is suppressed in HT29 colorectal cancer cells.22 In addition, miR-497 downregulation has been consistently demonstrated in a variety of solid tumor types such as hepatocellular carcinoma, ovarian cancer, colorectal adenomas, and in multiple myeloma cells.22,23 MiR-497, a highly conserved miRNA encoded by the first intron of the MIR497HG gene on human chromosome 17p13.110 belongs to the miR-15/16 family (miR-15a, miR-15b, miR-16-1/2, miR-195, miR-424 and miR-497) sharing the same seed sequence AGCAGCA.24 Downregulation of miR-497 controls cell cycle progression by regulating cell cycle regulators such as Cyclin A2, Cyclin D1, Cyclin D2, Cyclin D3 and Cdc25a. In a previous study, using microarray miRNA-expression profiling, we showed that miR-195 and miR-497 was differentially expressed in NPM-ALK+ ALCL lymph node primary tissues compared to reactive lymph nodes of healthy donors.21 As miR-195 and miR-497 are encoded as a cluster within the same host gene, MIR497HG (a highly conserved miRNA cluster),25 we sought to simultaneously study the roles of miR-195 and miR-497 in NPMALK+ ALCL tumorigenesis. Accordingly, we measured miR-195 and miR-497 expression in human NPM-ALK+ ALCL primary biopsies and cell lines. First, we studied the biological functions of these miRNAs in human NPM-ALK+ ALCL cells. We showed that overexpression of miR-497 inhibits cellular growth and causes cell cycle arrest. We identified cyclin E1, E2F3 and CDK6 as the main miR-497-targets responsible for the observed phenotype. Several CDK4/6 inhibitors have been developed [PD-0332991/palbociclib (Pfizer), LEE011/ribociclib (Novartis), and LY2835219/abemaciclib (Lilly)] and are currently being tested in clinical trials for patients with solid tumors and B lymphomas.26 Previous studies have demonstrated that palbociclib caused cycle arrest and apoptosis in T-cell leukemia in vitro and delayed disease progression in mouse models of T-cell acute lymphoblastic leukemia.27 However, this class of drugs has not been tested in T-cell lymphomas. Thus, we demonstrate, for the first time, the sensitivity of NPM-ALK+ cell lines to CDK4/6 inhibition using palbociclib.

Furthermore, we show that a scoring system based on CDK6, E2F3 and CCNE1 expression could help identify relapsing patients with a very short survival. Altogether, our results indicate that miR-497 functions as a tumor suppressor and that some of the miR-497 functional downstream targets could be used as predictors of clinical outcomes. Amongst them, CDK6 could be a therapeutic target for the development of future treatments for NPM-ALK+ ALCL. haematologica | 2019; 104(2)

CDK4/CDK6 inhibitors in ALK-positive lymphomas

Methods Human cell lines, tumoral and normal samples The two human NPM-ALK+ ALCL cell lines, KARPAS-299 and SU-DHL1, were obtained from DSMZ (German Collection of Microorganisms and Cell Culture, Braunschweig, Germany).

The COST cell line was established in our laboratory.28 CD4+ cells were stimulated for two days using CD3/CD28 antibodies coupled to magnetic beads (Dynabeads, Invitrogen, Waltham, USA) in RPMI-1640 with 20% fetal calf serum (FCS). All cells were cultured in RPMI-1640 supplemented with 10% FCS (KARPAS-299 and COST) or 15% FCS (SU-DHL1) supplemented with 2 mM L-glutamine, 1 mM sodium pyruvate, and 100 U/mL penicillin/streptomycin (all from Invitrogen, Waltham, USA). Cells were cultured at 37°C with 5% CO2 and maintained in exponential growth phase. Tumor samples from NPM-ALK+ and NPM-ALK– ALCL diagnosed based on morphological and immunophenotypical criteria (according to the last WHO classification) were obtained from our tumor tissue bank. Only cases with at least 50% lymph node involvement (assessed by CD30 staining of frozen biopsies) and good RNA integrity were selected from our tumor bank. Studies were conducted in accordance with the Declaration of Helsinki, and with the approval of the relevant ethics committees. Lymph nodes collected from patients with reactive non-malignant disease, who were considered to be healthy donors, were retrieved from the CRB-Cancer du CHU de Bordeaux n. BRIF BB-0033-00036, a member of the “Cancéropôle Grand Sud-Ouest” network.

Cell treatment Anaplastic large cell lymphoma cell lines were treated with 2 µM decitabine (Sigma-Aldrich, Saint Quentin Fallavier, France) for four days [fresh drug was added at 48 hours (h)] and NPMALK activity was inhibited using 500 nM crizotinib (@rtMolecule, Poitiers, France) for three days. ALCL cell lines were treated with 1 mM of PD-0332991 (Selleck Chemicals, Houston, TX, USA) during 24 h.

MiRNA and siRNA transfections siRNAs (si-CDK6, si-CCNE1, si-CDC25A and si-STAT3) at 0.5 nmol, si-E2F3 (Online Supplementary Table S1) at 1 nmol or miRNA mimics (miR-497-5p: mirVana MC10490 and miR-CTL: mirVana AM17110, Invitrogen, Waltham, USA) were transfected by electroporation according to a previously described procedure.17 In the case of co-transfection of several siRNAs, 0.2 nmol of each si-CDK6 and si-CCNE1 and 1 nmol of si-E2F3 were used. The same amount of duplex siRNA (Eurogentec, Angers, France) was transfected and used as negative control.

Purification of biotinylated-miRNA containing complexes Five million KARPAS-299, COST and SU-DHL1 cells were transfected with 0.25 nmol of biotinylated hsa-miR-497-5p (hsamiR-497 mercury LNA miRNA mimics premium) or biotinylated-irrelevant miRNA (negative control) (Exiqon, Vedbaek, Denmark). Total cell lysates, 24 h post transfection, were prepared in IP-buffer: 25 mM Tris-HCl pH 7.5, 200 mM NaCl, 0.2% Triton-X100, 5 mM MgAc, 1 mM DTT, protease inhibitors (Roche France, Meylan, France) and 0.2 U/mL RNase OUT (Invitrogen, Waltham, USA). The lysates were incubated for 2 h with streptavidin coupled to magnetic beads (Thermofisher, Waltham, USA) previously saturated with 0.5 mg/mL BSA and 0.2 mg/mL yeast tRNA. Purified RNAs were recovered after 5 washes with IP buffer by Trizol reagent-mediated extraction (Invitrogen, Waltham, USA), according to the manufacturer’s haematologica | 2019; 104(2)

instructions. RNA was reverse transcribed using SuperScript III Reverse Transcriptase (Invitrogen, Waltham, USA) and subjected to quantitative PCR. Biotin pull-down efficiency was expressed as a percentage of input. GAPDH amplification was used to remove pull-down background. The specificity of purification was evaluated for each putative target based on the enrichment in the Biotin-miR-497 condition relative to Biotin-CTL.

RNA extraction, reverse transcription and quantitative PCR Total RNA from cell lines and CD4+ cells were extracted using the Trizol reagent (Invitrogen, Waltham, USA), according to the manufacturer’s instructions. MiRNAs were reverse transcribed using the Universal cDNA Synthesis Kit II (Exiqon, Vedbaek, Denmark). The expression of miR-195, miR-497, RNU1A1 and SNORD44 (used as reference genes) was measured by quantitative PCR (qPCR) using a specific Exiqon PCR primer set and SYBR qPCR Premix Ex Taq (Tli RNaseH Plus) Kit (Takara Bio Europe, Saint-Germain-en-Laye, France). Messenger RNAs (mRNAs) were reverse transcribed using Primescript RT reagent kit (Takara Bio Europe, Saint-Germain-en-Laye, France), according to the manufacturer’s recommendations. qPCR was performed using the primers listed in Online Supplementary Table S1.

Protein extraction and Western blotting analyses Total cell lysates were prepared by incubating ALCL cells in Laemmli 1X buffer (Bio-Rad, Marnes-la-Coquette, France) for 10 minutes (min) on ice. Cells were then subjected to sonication using a Bioblock Vibra-cell 72446 apparatus at 35% of its power (Fisher Scientific, Illkirch, France). Proteins were separated by electrophoresis on SDS-polyacrylamide gels and electroblotted onto nitrocellulose membrane (GE Healthcare Europe, VelizyVillacoublay, France). Proteins of interest were detected using primary antibodies [(CDC25A 1/500: sc-7389, CDK6 1/2000: CST#3136, CCNE1 1/2000: sc-247, E2F3 1/2000: GTX102302, CDK4 1/2000: sc-260, Rb 1/2000: OP66, phosphor-Rb (ser780) 1/2000: CST#9307)] and HRP -conjugated anti-mouse (Promega, Madison, USA) or anti-rabbit secondary antibodies (Cell Signaling Technology, Danvers, USA). Densitometric analysis was performed using ImageJ software.29

Cell proliferation assays and cell cycle analysis Cell proliferation was evaluated using CellTiter 96AQueus One Solution (Promega, Madison, USA) and cell counting. For cell cycle analysis, NPM-ALK+ ALCL cells were washed with PBS and fixed overnight at 4°C in 70% ethanol. Cells were then washed with PBS supplemented with 0.1% BSA and incubated for 30 min with 10 mg/mL propidium iodide (Invitrogen, Waltham, USA) and 500 mg/mL of RNAse (Sigma-Aldrich, Saint Quentin Fallavier, France). Thereafter, cells were analyzed with a flow cytometer MACSQUANT VYB (Miltenyi, Paris, France).

Apoptosis luminescent assay The Caspase-Glo® 3/7 Assay Kit (Promega, Madison, WI, USA) was used to assess cell apoptosis, according to the manufacturer’s instructions.

Bisulfite conversion and DNA methylation analysis Genomic DNA was bisulfite converted using the MethylEdge Bisulfite Conversion System (Promega, Madison, USA) following the manufacturer’s protocol. The region of interest was amplified by PCR and methylation level was measured by pyrosequencing using PyromarkQ24 (Qiagen France, Courtaboeuf, France). (Primers are listed in Online Supplementary Table S1).

349

C. Hoareau-Aveilla et al.

A

B

Figure 1. MiR-195 and miR-497 are down-regulated in human NPM-anaplastic lymphoma kinase (ALK)-positive(+) primary lymphoma samples and cell lines. (A) RNA sequencing from NPM-ALK+ KARPAS-299 cells following decitabine or DMSO treatment or transfection with a control siRNA (si-CTL) or siRNA targeting ALK mRNA (si-ALK). A set of 13 miRNAs was found to be up-deregulated >1.5 fold in treated cells in both conditions (si-ALK and decitabine) compared to either negative control siRNA (si-CTL) or drug vehicle alone (DMSO). (B) Quantitative real-time PCR (qRT-PCR) analysis of miR-195 and miR-497 in NPM-ALK+ anaplastic large cell lymphoma patients (n=52). Data were normalized against equivalent miRNA levels from reactive lymph node (RLN) tissue samples (n=19). Data represent means±Standard Error of Mean (bars), ***P<0.0001; unpaired two-tailed Student’s t-test with Welch’s correction.

Xenograft tumor assay Mice were housed under pathogen-free conditions in an animal room at constant temperature (20-22°C), with a 12 h/12 h light:dark cycle and free access to food and water. All animal procedures were performed following the principle guidelines of INSERM, and our protocol was approved by the Midi-Pyrénées Ethics Committee on Animal Experimentation. ALCL cell lines were transfected either with control (miR-CTL) or miR-497 miRNA mimics. Twenty-four hours later, a total of 3x106 COST or KARPAS-299 cells or 2x106 SU-DHL1 cells were injected subcutaneously into both flanks of 7-week old female non-obese diabetic/severe combined immunodeficient (NOD-SCID) or NODSCID Gamma (NSG) mice (Janvier Labs, Le Genest, St Isle, France). Mouse body weight and tumor volumes were measured three times a week with calipers, using the formula “length x width2 x π/6”. At the end of the experiment, mice (6 per group) were humanely sacrificed. Subcutaneous tumors were then excised and sections were fixed in 10% neutral buffered formalin for histochemical analysis.

RNA sequencing RNA sequencing was performed at the LISBP, Toulouse, France, on an Ion Proton (Thermofischer, Waltham, MA, USA) as singleend oriented 95pb-long reads. RNAs co-purified with biotinylated hsa-miR-497-5p or with a biotin-negative control mimic (biotinmiR-CTL) were used for library preparation using the Ion Total RNA-seq Kit V2 and Ion PI HiQ OT2 Kit (Thermofischer, Waltham, MA, USA). TopHat2 (v.2.1.0) was used for sequence alignment on the human genome GRCH38.83 and counting reads performed using HTSeq-Count (v.0.6.1P1).

Statistical analysis Results are presented as mean values±Standard Error of Mean (SEM) from at least 3 independent experiments unless otherwise 350

indicated. Differences between groups were examined using the unpaired two-tailed Student’s t-test, unpaired two-tailed Student’s t-test with Welch’s correction or two-way ANOVA with Bonferroni post-test using GraphPad Prism software version 6.00 for Windows (GraphPad software) (La Jolla, CA, USA). For all tests, P<0.05 (*), P<0.01 (**) or P<0.001 (***) were considered statistically significant. Event-free survival was analyzed using a Receiver Operating Characteristics (ROC) method. A cut-off value was then chosen which would assume a minimum error rate and a limited number of misclassifications within the cohort.

Results The miR-497∼195 cluster is down-regulated in human NPM-ALK+ primary lymphoma samples and cell lines To determine the profile of methylated miRNA genes driven by NPM-ALK oncogene in ALCL cells, we performed a high throughput analysis of miRNA expression (small RNA sequencing) from NPM-ALK+ KARPAS-299 cells following decitabine treatment or transfection with a siRNA targeting ALK mRNA (si-ALK). A set of 13 miRNAs (miR-9-5p, miR-1287-5p, miR-15a-5p, miR-22-3p, miR26a-5p, miR-29c-3p, miR-194-5p, miR-195-5p, miR-199a3p, miR-199a-5p, miR-223-3p, miR-3613-5p and miR-4975p) was found to be up-regulated >1.5-fold in treated cells in both conditions (si-ALK and decitabine) compared either to negative control siRNA (si-CTL) or drug vehicle alone (DMSO) (Figure 1A). Amongst these 13 miRNAs, using microarray data previously published of miRNAexpression of NPM-ALK+ (n=52) and NPM-ALK– (n=5) ALCL patients,21 we observed that miR-195 and miR-497 were down-regulated only in NPM-ALK+ ALCL patients compared to normal reactive lymph node (RLNs, n=3) (Online Supplementary Table S2). To corroborate the haematologica | 2019; 104(2)

CDK4/CDK6 inhibitors in ALK-positive lymphomas

A

B

C

D

E

F

Figure 2. NPM-anaplastic lymphoma kinase (ALK) activity and DNA methylation mediate downregulation of miR-195 and miR-497. Quantitative RT-PCR analysis of miR-195 and miR-497 expression in NPM-ALK-positive lymphoma cell lines, COST, KARPAS-299 (KARPAS) and SU-DHL1, transfected with either an irrelevant siRNA as the negative control (si-CTL) or a siRNA targeting ALK mRNA (si-ALK) (A and D) or treated for 72 hours (h) or not (PBS) with crizotinib (B and E) or treated for 96 h or not (PBS) with decitabine (C and F). SNORD44 was used as an internal control and the relative ratio of miR-195 and miR-497 expression was expressed as 2– ΔΔCt relative to untreated cells. Data represent means±Standard Error of Mean (bars) from 3 independent experiments, **P<0.01; ***P<0.0001; unpaired twotailed Student’s t-test with Welch’s correction.

miRNA array data, we performed qPCR analysis for miR195 and miR-497 on 23 samples from NPM-ALK+ by selecting cases with at least 50% lymph node involvement compared to 12 RLNs (Figure 1B). Significant downregulation of both miR-195 and miR-497 was confirmed in primary biopsies despite the presence of a mixed population of neoplastic and normal cells (Figure 1B).

NPM-ALK activity and DNA methylation mediate silencing of miR-195 and miR-497 in lymphoma cells The downregulation of miR-195 and miR-497 observed in NPM-ALK+ cells suggests that the NPM-ALK+ protein itself might drive this phenomenon. This hypothesis is corroborated by the fact that, in our initial screen, miR195 and miR-497 were up-regulated following NPM-ALK knockdown in KARPAS-299 cells (Figure 1A). To verify whether NPM-ALK is involved in miR-195 and miR-497 silencing, NPM-ALK expression was knocked down in 2 other human NPM-ALK+ ALCL cell lines (COST and SUDHL1) using siRNA directed against ALK mRNA (Figure 2A and D). This was followed by reducing NPM-ALK activity with crizotinib treatment, a pharmacological inhibitor of ALK tyrosine kinase activity (Figure 2B and E). Western blotting analysis (Online Supplementary Figure S1A and B) showed that NPM-ALK was down-regulated and inhibited. Both inhibition of ALK gene expression haematologica | 2019; 104(2)

and its activity induce a significant increase in both miR195 and miR-497 expression in NPM-ALK+ cells (Figure 2A-E). Next, we measured the levels of both miR-195 and miR-497 after decitabine treatment in order to confirm that the observed miR-195 and miR-497 downregulation in NPM-ALK+ lymphoma cells was, indeed, due to DNA methylation (Figure 2C and F). We detected a significant increase in miR-195 and miR-497 expression in DNMT inhibitor-treated cells compared to the drug vehicle alone (DMSO) in all NPM-ALK+ cell lines (Figure 2C and F).

STAT3 appears to be implicated in this mechanism (data not shown). To determine the DNA methylation status of the MIR497HG promoter, the miR-195 and miR-497 host gene on chromosome 17, we carried out bisulfite sequencing analysis of genomic DNA isolated from the three NPM-ALK+ cell lines: KARPAS-299, SU-DHL1 and COST and normal CD4 T lymphocytes expressing CD30 antigen activated using CD3/CD28 antibodies (n=3) (Online Supplementary Figure S2A). The results showed significantly high levels of methylation of all CpGs located in the region upstream of miR-195 and miR-497 promoters in the KARPAS-299 and SU-DHL1, and a few CpGs (CpG1, CpG2 and CpG3) in COST tumor cells. We observed lower levels of CpG methylation in the region upstream of miR-195 and miR-497 promoters in normal cells (Online Supplementary Figure S2B). Decitabine signifi351

C. Hoareau-Aveilla et al.

cantly decreased methylation levels of CpGs in NPMALK+ cancer cells (Online Supplementary Figure S2C).

Collectively, these data suggest that the silencing of both miR-195 and miR-497 is due, at least in part, to NPM-ALK expression and aberrant MIR497HG promoter methylation in ALCL cells.

A

Forced expression of miR-497 suppresses cell cycle progression and prevents the growth of NPM-ALK+ ALCL cells To investigate possible biological functions (potential effects on cell proliferation) of miR-195 and miR-497 in NPM-ALK+ cells, the two miRNAs were ectopically

B

Figure 3. Overexpression of miR-497 reduces in vitro viability and growth of NPM-anaplastic lymphoma kinase (ALK)-positive(+) lymphoma cells. Cell viability determined by MTS assay of NPM-ALK+ lymphoma COST, KARPAS-299 (KARPAS) and SU-DHL1 cells 72 hours (h) after transfection with mimic miR-195 (miR-195) (A) or mimic miR-497 (miR-497) (B). Data represent means±Standard Error of Mean (bars) from 3 independent biological replicates, ***P<0.0001; unpaired two-tailed Student’s t-test with Welch’s correction.

A

B

C

D

Figure 4. Ectopic expression of miR-497 induces cell cycle arrest and reduces growth in vivo of NPM-anaplastic lymphoma kinase (ALK)-positive(+) lymphoma cells.

NPM-ALK+ ALCL COST, KARPAS-299 (KARPAS) and SU-DHL1 cells were transfected with negative control miRNA (miR-CTL) or mimic miR-497 (miR-497). Cell cycle was analyzed (A) and caspase 3/7 activity was measured (B). Data were normalized against the miR-CTL condition. Each experiment was repeated 3 times. (C) NPMALK+ COST, KARPAS-299 (KARPAS) and SU-DHL1 cells transfected either with miR-CTL or miR-497 were injected subcutaneously in the left or right flank of immunodeficient mice, respectively (n=6). Tumor volume was evaluated over time by caliper measurements. Data represent mean±Standard Error of Mean, *P<0.05; **P<0.001; ***P<0.0001; ns: not significant; unpaired two-tailed Student’s t-test with Welch’s correction. (D) Micrographs of hematoxylin & eosin staining of excised miR-CTL or miR-497 tumors (original magnification x40). Arrows indicate cells with phenotypical hallmarks of cellular degeneration.

352

haematologica | 2019; 104(2)

CDK4/CDK6 inhibitors in ALK-positive lymphomas

expressed by transfecting synthetic mimics into KARPAS299, SU-DHL1 and COST cell lines. We observed that upregulation of miR-195 by transfection with miR-195 mimic does not impair cell viability of NPM-ALK+ cells (Figure 3A). By contrast, upregulation of miR-497 significantly affects cell viability of the three cell lines (Figure 3B). The successful overexpression of miR-497 and of miR-195 were confirmed by qPCR (Online Supplementary Figure S3). Taken together, these data suggest that only miR-497 in the miR-497/miR-195 cluster plays a role in cell proliferation of NPM-ALK+ cells.

Since cell proliferation is closely associated with cell cycle control and progression, we performed flow cytometric analysis of cell cycle with propidium iodide DNA staining to examine the effects of miR-497 on cell cycle distribution in all three NPM-ALK+ cell lines.

Accumulation in the G1 phase of the cell cycle was significant in KARPAS-299 cells, 48 h post transfection with miR-497 compared with those transfected with miR-CTL (Figure 4A). In addition, in the same condition, ectopic expression of miR-497 in COST and SU-DHL1 cells led to a significant accumulation of cells in a sub-G1 phase of cell cycle suggesting cell death (Figure 4A). To test whether apoptosis is involved in the growth inhibition caused by miR-497 overexpression, we measured caspase 3 and 7

A

activity. As shown in Figure 4B, compared to that of miRCTL, miR-497 overexpression induced activation of apoptosis pathways in COST and SU-DHL1 cells (2.2- and 4fold, respectively) but not in KARPAS-299 cells. These findings suggest that miR-497 overexpression induces growth-inhibitory effects on cancer cells by impeding cell cycle progression and leads to programmed cell death. We next investigated the effect of miR-497 transfection on tumor growth of NPM-ALK+ cell lines KARPAS-299, SUDHL1 and COST in vivo. MiR-CTL or miR-497-transfected NPM-ALK+ cells were inoculated subcutaneously into the left or right flanks of each immunodeficient mouse, respectively (n = 6 for each condition). MiR-497-transfected NPM-ALK+ cells resulted in significant reduction of tumor growth (Figure 4C) when compared to those transfected with miR-CTL. Morphological analyses showed that forced expression of miR-497 in NPM-ALK+ cells caused phenotypic hallmarks of cellular degeneration (uncommon chromatin condensation, nuclear piknosis, decreased cellular volume and disruption of the nuclear envelope) (Figure 4 D).These results show that miR-497 overexpression drastically reduces the growth of NPMALK+ cells in vivo. Taken together, these experiments suggest a tumor-suppressive function of miR-497 in NPMALK+ cells.

B

C

Figure 5. MiR-497 targets G1/S regulators in anaplastic lymphoma kinase (ALK)-positive(+) lymphoma cells. (A) mRNA targets of miR-497 identification workflow.

NPM-ALK+ KARPAS-299 cells transfected with biotinylated forms of human miR-497 (biotin-miR-497) or an irrelevant biotin-miRNA as negative control (biotin-miRCTL) were used to co-purify associated mRNAs (enrichment ratio >1.5). Validated candidate targets of miR-497 were collected from miRTarBase (n=77) and experimentally validated miR-497 associated with the regulation of cell cycle from miRSystem database (n=16). CCNE1, CDC25A, CDK6, CCND3 and E2F3 were the only genes identified in the two computational analyses. (B) Microarray analysis of CCNE1, CDC25A, CDK6, CCND3 and E2F3 mRNA expression in human NPM-ALK+ (n=56; Daugrois et al., submitted publication). Data were normalized against equivalent miRNA levels from reactive lymph (RLN) node tissue samples (n=3). Data represent means±Standard Error of Mean (bars), *P<0.05; **P<0.001; ***P<0.0001; unpaired two-tailed Student’s t-test with Welch’s correction. (C) Quantitative real-time PCR (qRT-PCR) analysis of CCNE1, CDC25A, CDK6, CCND3 and E2F3 mRNA was performed after pulldown of the biotinylated miRNAs using streptavidin beads. The results are presented as the relative enrichment in miR-497 pulldown compared to miRNA control pulldown (enrichment ratio >2).

haematologica | 2019; 104(2)

353

C. Hoareau-Aveilla et al.

A

B

C

Figure 6. CDK6, CDC25A, CCNE1 and E2F3 are target genes of miR-497 in anaplastic lymphoma kinase (ALK)-positive(+) lymphoma cells. (A) Quantitative real-time (RT)-PCR analysis of CDK6, CDC25A, CCNE1 and E2F3 mRNA expression in NPM-ALK+ lymphoma COST, KARPAS-299 (KARPAS) and SU-DHL1 cells transfected with negative control miRNA (miR-CTL) or a mimic miR-497 (miR-497). GAPDH was used as internal control and the relative ratios of mRNA expression was expressed as 2–ΔΔCt relative to those in cells transfected with miR-CTL. Data represent means±Standard Deviation (bars), **P<0.001; ***P<0.0001; unpaired two-tailed Student’s t-test with Welch’s correction. (B) Western blotting analysis of CDK6, CDC25A, CCNE1 and E2F3 expression in NPM-ALK+ COST, KARPAS-299 (KARPAS) and SU-DHL1 cells transfected by mir-CTL or mir-497. The GAPDH protein served as an internal control to ensure equal loading. Results from one representative experiment are shown. (C) Densitometric analysis was performed using ImageJ software from Wayne Rasband (NIH). The relative levels of the proteins of interest were normalized to GAPDH levels. Data represent means±Standard Error of Mean (bars) from 3 independent experiments, **P<0.001; ***P<0.0001; ns: not significant; unpaired two-tailed Student’s t-test with Welch’s correction.

Effect of miR-497 on the expression of cell cycle regulatory genes in NPM-ALK+ lymphoma cells To determine the mechanism(s) by which miR-497 regulates tumor growth and progression of NPM-ALK+ cancer cells, we searched for mRNA targets of miR-497. To capture miR-497 associated mRNAs, KARPAS-299 cells were transfected with biotinylated human miR-497. We isolated and sequenced RNAs that co-purified with biotin-miR-497 mimic to map miR-497-mRNA interaction. As a negative control, we performed the same experiment with a biotin-negative control mimic (biotin-miRCTL). A large set of mRNAs is regulated by miR-497 (enrichment ratio >1.5). In this data set, 77 validated candidate targets of miR-497 were collected from the miRTarBase (http://mirtarbase.mbc.nctu.edu.tw/, Release 7.0) (Figure 5A). In addition, using the miRSystem database (http://mirsystem.cgm.ntu.edu.tw/), we identified 16 miR497 mRNA targets, experimentally validated and associated with cell cycle regulation (KEGG pathway database) (Figure 5A). We identified 5 mRNA gene targets that were common to both databases: CCNE1, CDC25A, CDK6, CCND3 and E2F3 (Figure 5A). Next, we used gene expression array data (Daugrois et al., 2018, submitted publication) to identify the veracity of these 5 targets captured by biotin-miR-497. Despite the presence of a mixed population of neoplastic and normal cells, CCNE1, CDC25A, CDK6 and E2F3 mRNAs were found to be sig354

nificantly over-expressed in NPM-ALK+ ALCL patients (n=55) compared to reactive lymph node controls (RLN) (Figure 5B). As previously reported by Thompson et al., CCND3 was not over-expressed in NPM-ALK+ ALCL patients.30 Finally, to validate the specificity of the biotin pulldown and applicability to other cell lines, biotinylated forms of miR-497 or negative miRNA-control were used to co-purify CCNE1, CCDN3, CDC25A, CDK6 and E2F3 mRNAs in three KARPAS-299, SU-DHL1 and COST cell lines. Enrichment of target mRNAs by streptavidin-mediated purification (enrichment ratio >2) was measured by qPCR and normalized to miR-CTL enrichment. In the biotin-miR-497 pulldown of COST, SU-DHL1 and KARPAS-299 cells, CDC25A, CCNE1 and E2F3 transcripts but not CCDN3 were captured 24 h after transfection (Figure 5C). Moreover, biotin pulldown enriches CDK6 mRNA in COST and SU-DHL1 cells but not in KARPAS-299 cells (Figure 5C). This result was in accordance with data published by Nagen et al.,31 who reported very weak CDK6 expression in KARPAS-299 cells compared to SU-DHL1 cells (Online Supplementary Figure S5).

Moreover, for the first time, we observed that CDK6 protein is over-expressed in COST cells (similar to SU-DHL1 cells) in comparison to KARPAS-299 cells (Online Supplementary Figure S4). To corroborate the results of miR-497–bound mRNAs capture and transcriptomic analyses, the effect of miR-497 on the endogenous haematologica | 2019; 104(2)

CDK4/CDK6 inhibitors in ALK-positive lymphomas

A

B

C

D

E

F

G

H

I

Figure 7. Silencing of CDK6, CCNE1 and E2F3 genes reduces cell proliferation in anaplastic lymphoma kinase (ALK)-positive(+) lymphoma cells. NPM-ALK+ ALCL SU-DHL1 (A-C), COST (D-F) and KARPAS-299 (KARPAS) (G-I) cells were transfected with either an irrelevant siRNA as the negative control (si-CTL) or a siRNA targeting CDK6 mRNA (si-CDK6) or CDK6, CCNE1 and E2F3 mRNA together (si-Pool). Cell cycle distribution (A, D and G), cell count at three different time points (0, 1, 2 days) (B, E and H) and caspase 3/7 activity (C, F and I) were measured. Data represent means±Standard Error of Mean (bars) from 3 independent experiments, *P<0.05; **P<0.001; ***P<0.0001; ns: not significant; unpaired two-tailed Student’s t-test with Welch’s correction.

expression of CCNE1, CDC25A, CDK6 and E2F3 was subsequently examined using qPCR and Western blotting. Transfection of miR-497 mimic induced a significant decrease of all four mRNAs in all three cell lines compared to cells transfected with miR-CTL (Figure 6A).

Overexpression of miR-497 significantly down-regulated the protein level of CDK6, CDC25A and E2F3 in all three cell lines (Figure 6B). In addition, CCNE1 expression was down-regulated in KARPAS-299 cells but not in COST or SU-DHL1 cells (Figure 6B). Densitometric analysis of the mean relative intensity for each target protein of miR-497 supports the Western blotting results (Figure 6C). Taken together, these results strongly suggest that miR-497 regulates the expression of CCNE1, CDC25A, CDK6 and E2F3 in NPM-ALK+ cells.

Significance of identified miR-497 target genes in NPM-ALK+ lymphoma cells To evaluate whether CCNE1, CDC25A, CDK6 and E2F3 may be functional downstream targets of miR-497 and involved in regulation of NPM-ALK+ cell proliferation, we individually knocked down the expression of these genes using specific siRNAs. In order to check that the knockdown of gene expression had been efficiently achieved, we performed qPCR and Western blotting to detect the mRNA (Online Supplementary Figure S5A) and haematologica | 2019; 104(2)

protein (Online Supplementary Figure S5B) levels, respectively. Densitometric analysis of the mean relative intensity for each target protein of miR-497 supports the Western blot results (Online Supplementary Figure S5C). We observed that, in SU-DHL1 cells, silencing of CDK6 mRNA simultaneously induced a decrease of cell viability and an increase of both subG1 fraction and caspase 3/7 activity, thereby inhibiting cell growth (Figure 7A-C).

Hence, CDK6 silencing alone is sufficient to recapitulate the phenotype observed upon miR-497 ectopic expression in SU-DHL1 cells. These results are similar to those published by Nagen et al.31 showing an addiction to CDK6 expression for proper cell cycle progression in this cell line.

In contrast, even though siRNA directed against CDK6 led to a significant accumulation of COST and KARPAS-299 cells in G0/G1 phase (Online Supplementary Figure S6B and E), it is not sufficient to fully recapitulate miR-497-induced blockage of cell cycle. The same observation was made after E2F3 and CCNE1 mRNA downregulation (Online Supplementary Figure S6A and D, and C and F, respectively). Finally, we observed that CDC25A knockdown did not modify the cell cycle profiles of COST and KARPAS299 cells (data not shown). We also tested the effect of each siRNA on cellular apoptosis in COST and KARPAS-299 cells. Compared to the control condition (si-CTL), CDK6 and CCNE1 silencing induced apoptosis in COST cells 355

C. Hoareau-Aveilla et al.

A

B

C

D

E

F

G

H

I

J

K

L

Figure 8. Anaplastic lymphoma kinase (ALK)-positive(+) lymphoma cells are sensitive to CDK6 inhibitor. NPM-ALK+ ALCL SU-DHL1 (A-D), COST (E-H) and KARPAS299 (KARPAS) (I-L) cells were treated or not (H2O) with palbociclib, an CDK4/6 inhibitor. Cell count at three different time points (0, 1, 2 days) (A, E and I), cell cycle distribution (B, F and J), caspase 3/7 activity (C, G and K) and cell proliferation rate (D, H and L) were measured. Data represent means±SEM (bars) from 3 independent experiments. *P<0.05; **P<0.001; ***P<0.0001; ns: not significant; unpaired two-tailed Student’s t-test with Welch’s correction. Data represent means±Standard Error of Mean (bars) from 3 independent experiments, *P<0.05; **P<0.001; unpaired two-tailed Student’s t-test with Welch’s correction.

but, as expected, not in KARPAS-299 cells (Online Supplementary Figure S7). In addition, the silencing of E2F3 mRNA did not affect caspase activity in both NPM-ALK+ COST and KARPAS-299 cells (Online Supplementary Figure S7). Moreover, regardless of the siRNA transfected, cell viability following RNAi delivery was never significantly modified in NPM-ALK+ cells (data not shown). Thus, individual silencing of three miR-497-targets, CDK6, CCNE1 and E2F3, can reproduce (only to a lesser extent) the miR497-induced phenotype (Figure 4A and B). Consequently, in order to improve the effects observed in COST and KARPAS-299 cells, we decided to pool siRNAs targeting CCNE1, CDK6 and E2F3. We transfected KARPAS-299 or COST cells with siRNAs (pools of three siRNAs/targets) targeting each of the miR-497-down-regulated transcripts.

In order to check whether the knockdown of gene expression had been efficiently achieved, we performed qPCR to detect the mRNA (Online Supplementary Figure S8). We then analyzed transfected cells for their cell cycle distributions. Similar to transfection of miR-497 mimics, siRNA pools induced significant accumulation of KARPAS-299 and COST cells in G0/G1 fraction (Figure 7D and G). Of note, the same result was observed in COST and SUDHL1 cells, 24 h post transfection with the miR-497 mimic (Online Supplementary Figure S9A and B). The reduction in cell proliferation in the knockdown NPM-ALK+ cells was observed by cell counting (Figure 7E and H). In addition, compared to the control condition (si-CTL), CCNE1, CDK6, E2F3 silencing induced apoptosis in COST cells thereby inhibiting cell growth (Figure 7G and 356

K) but, as expected, not in KARPAS-299 cells (Figure 7K and H). Taken together, these findings demonstrate that miR-497 controls multiple targets that collaborate to regulate cell cycle progression and cell growth.

ALK-positive cells are responsive to CDK4/6 inhibitors The new generation of selective CDK4/6 inhibitors, with anti-proliferative activity in Rb-proficient cells, target tumor types in which CDK4/6 has a pivotal role in the G1to-S-phase cell cycle transition with improved effectiveness and fewer adverse effects. Palbociclib is an orally bioavailable CDK4/6 inhibitor recently approved by the US Food and Drug Administration for the treatment of metastatic breast cancer.32 As all three NPM-ALK+ cell lines, SU-DHL1, COST and KARPAS-299 expressed CDK4, CDK6 and Rb (Online Supplementary Figure S4), we evaluated the effect of palbociclib treatment. We observed that all NPM-ALK+ cell lines are sensitive to this molecule (Figure 8A, E and I). In addition, we showed that palbociclib markedly inhibited the proliferation of NPM-ALK+ cells through the induction of G0/G1 cell-cycle arrest (Figure 8B, F and J), increased apoptosis (Figure 8C, G and K) and reduced cell growth (Figure 8D, H and L). As a control of palbociclib efficacy, we observed a decrease in Cdk4/6-specific phosphorylation of the Rb protein as shown by Western blotting (Online Supplementary Figure S10A). Densitometric analysis of the mean relative intensity for pRB and RB corroborate Western blotting results (Online Supplementary Figure S10B). As palbociclib is a selective cyclin D kinase 4/6 inhibitor, NPM-ALK+ ALCL haematologica | 2019; 104(2)

CDK4/CDK6 inhibitors in ALK-positive lymphomas

SU-DHL1, COST and KARPAS-299 cells were transfected with either an irrelevant siRNA as the negative control (siCTL) or a siRNA targeting CDK4 mRNA. In order to check whether the knockdown of gene expression had been efficiently achieved, we performed qPCR to detect the CDK4 mRNA (Online Supplementary Figure S11C, F and I). As silencing of CDK4 does not reduce cell proliferation in ALK+ lymphoma cells (Online Supplementary Figure S11A, D and G), nor induce apoptosis (Online Supplementary Figure S11B, E and H), we suggest that the action of palbiciclib is only CDK6-dependent.

Based on our findings that CDK6, CCNE1 and E2F3 are important for efficient proliferation of NPM-ALK+ ALCL cells, we hypothesized that the highest level of expression of these factors could be the most favorable for tumor development and possibly affect chemotherapy effectiveness. To test this hypothesis, we used qPCR to measure the expression of CDK6, CCNE1 and E2F3 in a cohort of NPM-ALK+ primary ALCL samples (n=55) obtained from chemotherapy-naive patients at diagnosis. Compared to normal reactive lymph nodes (RLNs) (n=19) and despite the presence of a mixed population of neoplastic and normal cells, we observed a significant CDK6, CCNE1 and E2F3 gene overexpression in NPM-ALK+ samples (Online Supplementary Figure S12A). In order to evaluate the clinical significance of CCNE1, CDK6 and E2F3 overexpression, first using linear regression, we derived a model based on microarray data on expression levels of CCNE1, CDK6, E2F3 genes in NPM-ALK+ samples (n=55). The samples included NPM-ALK+ primary biopsies of patients who had experienced early relapse and those without relapse after three years of minimal follow up after chemotherapy cure.

By applying the formulae obtained from linear regression, we derived a ‘score’ for each patient: score = (-0.08378 x CDK6) + (-0.41269 x CCNE1) + (0.37893 x E2F3).

Secondly, a ROC curve was generated from CCNE1, CDK6, E2F3 genes-based score from 44 NPM-ALK+ ALCL samples, including 8 adults and 36 children, for whom event-free survival is known (Online Supplementary Figure S12B). The cut-off value and Air Under Curve (AUC) values were determined as -0.038 and 0.65, respectively.

Based on the optimal cut-off points from the ROC curve, patients were significantly categorized into two groups, high (n=25) and low (n=14) CCNE1, CDK6, E2F3 expression (Online Supplementary Figure S12C and D). NPM-ALK+ ALCL patients with a low expression of CCNE1, CDK6, E2F3 had a poor prognosis (P<0.046) (Online Supplementary Figure S12C). Moreover, this trend can be observed for high (n=14) and low (n=22) expression in NPM-ALK+ pediatric ALCL (Online Supplementary Figure S12D). In addition, for the same pediatric cases, we noted that CCNE1, CDK6, E2F3 gene-based score was increased in relapsing patients (n=23 vs. without relapse n=13) (Online Supplementary Figure S12E). As all patients were treated with chemotherapy following diagnosis, our data suggest that CDK4/6 inhibitors could benefit ALK+ patients harboring resistant tumors.

Discussion MiR-195 and miR-497 are important members of the miR-15/16 family and are aberrantly expressed in multiple diseases, such as solid cancers.23,33 Based on the expression of miRNAs, Steinhilber et al. have identified a gene signahaematologica | 2019; 104(2)

ture including reduced expression of some miR-15/16 family members, such as miR-16, miR-195 and miR-497 that distinguish ALK+ from ALK– ALCL.34 Except miR-16,19 the functional characterization of miR-195 and miR-497, which are clustered on chromosome 17p13.1, has never been undertaken in ALK+ lymphoma cells.

MiR-195 and miR-497 downregulation has been consistently demonstrated in a variety of solid tumor types. miRNA-silencing in cancer is frequently related to abnormal promoter methylation.35 Accordingly, the downregulation of miR-195 and miR-497 has been reported to be linked to promoter-associated hypermethylation in solid cancers; demethylation following 5-aza-CdR treatment resulted in the re-expression of these two miRNAs.36,37 In the present work, we demonstrate for the first time that miR-195 and miR-497 are down-regulated in human NPM-ALK+ ALCL tissues and cell lines through a mechanism involving NPM-ALK and DNA hypermethylation.

These results are highly reminiscent of what we previously observed for three other miRNA genes, MIR29A, MIR125B and MIR150, the levels of which were also reduced in NPM-ALK+ ALCL as a consequence of the promoter hypermethylation.17,20,21 In addition, we observed that only ectopic expression of miR-497 markedly attenuates in vitro cell proliferation of NPM-ALK+ cells.

It is estimated that more than half of human genes are targeted by at least one miRNA. One miRNA may have multiple different mRNA targets and at the same time, one mRNA might be targeted by multiple miRNAs. miR-497 regulates the expression of a plethora of targets, which are involved in cell cycle, apoptosis, proliferation, etc.23 Recent studies have revealed that miR-497 can directly reduce the CCNE1 protein level to suppress tumor growth by inducing G1 arrest in breast cancer38 and hepatocellular carcinoma.39 Indeed, CCNE1 protein binds to and activates cyclindependent kinase 2 (CDK2) to promote a cascade of events required for cell-cycle progression from G1 to S phase. Several other genes encoding cell-cycle regulators, including Cyclin D1/CCND1 and Cyclin D3/CCND3, Cyclin-dependent kinase 4 (CDK4) and CDK6, cell-cycle division factors CDC25A and/or E2F3 transcription factor, are also known targets of miR-497.23,40 The role of miR-497 in human cancer is not yet clear. Several studies have reported that increased miR-497 expression via miR-497 mimic transfection suppressed proliferation and increased apoptosis in various solid cancers, such as adrenocortical carcinoma41 and pancreatic cancer.42 Accordingly, miR-497 overexpression was found to block G0/G1 phase transition in breast cancer cells43 and induce G1/S arrest in gastric cancer cells.37 In our study, we revealed for the first time that enhanced miR-497 expression in human NPMALK+ lymphoma cells could suppress cell proliferation, by inducing cell cycle arrest, and inhibit tumor growth in vivo.

These biological effects are the consequence of simultaneous downregulation of three mir-497 mRNA targets: CCNE1, E2F3 and CDK6, and their subsequent decreased protein level. Contribution of CCNE1 and E2F3 in uncontrolled cell-cycle progression and oncogenesis have never been studied in human NPM-ALK+ ALCL models. Of note, only two publications report an abnormal CDK6 expression in human NPM-ALK+ lymphoma cells. The amplification of the CDK6 locus in NPM-ALK+ SU-DHL1 cells and co-expression of high levels of CDK6 and CD31 (angiogenic marker) in NPM-ALK+ ALCL samples have been reported.31,44 357

C. Hoareau-Aveilla et al.

NPM-ALK+ ALCL is a rare disease which accounts for about 1-2% of adult non-Hodgkin lymphomas (NHL) and 10-15% of childhood NHL. The prognosis of NPM-ALK+ ALCL is better than that of other T-cell lymphomas.

Overall, patients with NPM-ALK+ ALCL who are resistant to primary chemotherapy or who relapse early have the worst prognosis.4,45 We checked the CCNE1, E2F3 and CDK6 overexpression in a cohort of NPM-ALK+ primary ALCL samples obtained from chemotherapy-naive patients at diagnosis. As aberrant growth control is a hallmark of cancer cells, we hypothesized that altered CCNE1, E2F3 and CDK6 expression in NPM-ALK+ patients could affect their response to treatment. Using a NPM-ALK+ cohort, including patients who had experienced early relapse after chemotherapy, we show that the three G1-S regulators together could predict treatment outcome in pediatric patients.

CDK6 protein is a member of the cyclin-dependent kinase family, which includes CDK4, with a 71% shared amino acid identity. These homologs are ubiquitously expressed and have largely overlapping functions. In mammalian cells, cell cycle is regulated by CDK4 and CDK6 in early G1 phase through interactions with cyclins D1, D2 and D3. Cyclin D/CDK4 and cyclin D/CDK6 complexes induce phosphorylation of the retinoblastoma protein (Rb) and the release of E2F3 transcription factor from RB-E2F3 complexes. Cell-cycle components, such as CDK4 and CDK6, are frequently altered in human cancer, reflecting the deregulated growth of transformed cells. In B-cell lymphoid malignancies, DNA hypermethylation-mediated epigenetic silencing of the tumor suppressors miR-124a and miR-29 results in an upregulation of their target CDK6.46 Downregulation of CDK6 was observed following introduction of miR-29a mimics into T-cell acute lymphoblastic leukemia Jurkat cells.47 CDK6 has also been implicated in thymic lymphoma formation in a transgenic mouse model.48 Moreover, Kollmann et al. have observed that NPM-ALK– transgenic mice lacking CDK6 expression developed disease with a significantly prolonged latency.44 At the same time, in a NOD/SCID xenograft human multiple myeloma model, palbociclib, a newly developed small-molecule inhibitor specific to CDK4 and CDK6, can enhance the killing of myeloma cells by dexamethasone.49 In addition, palbociclib inhibits CDK4/6 activity according to the activation status of cycling primary bone marrow

References 1. Swerdlow SH, Campo E, Pileri SA, et al.

The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390.

2. Brugieres L, Le Deley MC, Rosolen A, et al.

Impact of the methotrexate administration dose on the need for intrathecal treatment in children and adolescents with anaplastic large-cell lymphoma: results of a randomized trial of the EICNHL Group. J Clin Oncol. 2009;27(6):897-903.

3. Morris SW, Kirstein MN, Valentine MB, et al. Fusion of a kinase gene, ALK, to a nucleolar protein gene, NPM, in non-Hodgkin's lymphoma. Science. 1994;263(5151):12811284.

358

myeloma cells freshly isolated from both new and relapsed patients.49 A phase II study has shown that progression-free survival was doubled with palbociclib plus standard hormone therapy in advanced hormone receptorpositive, HER2-negative breast cancer.32 Taken together, these findings provide the grounds for new therapeutic strategies in NPM-ALK+ ALCL either targeting the epigenetic regulation of miRNAs and/or directly targeting the CDK6 pathway. We have previously reported that hypomethylating drugs may benefit NPM-ALK+ patients.17 Bonvini et al. previously described the effect of flavopiridol, a pan-CDK inhibitor, on NPM-ALK+ ALCL cells.50 In the present study, as a proof of concept, we have presented in vitro data showing that palbociclib can be the first promising and specific inhibitor for therapeutic targeting of Cdk4/6 in NPM-ALK+ lymphoma cells. Our work also provides a rationale for targeting ALK+ ALCL progression with palbociclib. Indeed, it could be possible to achieve greater synergy by combining palbociclib with other cytotoxic agents or anti-ALK tyrosine kinase inhibitor such as crizotinib, recently approved for the treatment of metastatic and late-stage ALK-rearranged non-small-cell lung cancers (NSCLC). In summary, these findings suggest that CDK4/CDK6 inhibitors could be novel candidates for mechanism-defined combination therapy in ALK+ lymphomas and possibly other ALK+-cell cancers including a proportion of NSCLC. Acknowledgments This work is dedicated to CHA: «A winner is a dreamer who never gives up (Nelson Mandela)». English proofreading was performed by EnPro Language Solutions (www.enprols.fr) and Greenland scientific proofreading.

Funding This work was supported by grants from the “Ligue contre le Cancer”, Fondation ARC pour la Recherche sur le Cancer” (FM), and INCa (projet PAIR Lymphomes) (LL). CHA and AC were supported by a fellowship from the “Labex TOUCAN / Laboratoire d’excellence Toulouse Cancer”. For their technical assistance, the authors thank M. Tosolini (Pôle Technologique du CRCT – Plateau Bioinformatique INSERM-UMR1037); F. Capilla and C. Salon at the histology facility and the staff of the zootechnie Langlade, INSERM/UPS-US006/CREFRE (Toulouse, France).

4. Delsol G, Falini B, Mûller-hermelink HK, Campo E, et al. Anaplastic large cell lymphoma, ALK-positive. Lyon: IARC Press, 2008.

5. Iwahara T, Fujimoto J, Wen D, et al.

Molecular characterization of ALK, a receptor tyrosine kinase expressed specifically in the nervous system.

Oncogene.

1997;14(4):439-449.

6. Damm-Welk C, Busch K, Burkhardt B, et al. Prognostic significance of circulating tumor cells in bone marrow or peripheral blood as detected by qualitative and quantitative PCR in pediatric NPM-ALK-positive anaplastic large-cell lymphoma. Blood.

2007;110(2):670-677.

7. Ait-Tahar K, Damm-Welk C, Burkhardt B, et al. Correlation of the autoantibody response to the ALK oncoantigen in pedi-

8.

9.

10.

11.

atric anaplastic lymphoma kinase-positive anaplastic large cell lymphoma with tumor dissemination and relapse risk. Blood.

2010;115(16):3314-3319.

Chiarle R, Voena C, Ambrogio C, Piva R, Inghirami G. The anaplastic lymphoma kinase in the pathogenesis of cancer. Nat Rev Cancer. 2008;8(1):11-23.

McDonnell SR, Hwang SR, Basrur V, et al.

NPM-ALK signals through glycogen synthase kinase 3beta to promote oncogenesis.

Oncogene. 2012;31(32):3733-3740.

Wellmann A, Doseeva V, Butscher W, et al.

The activated anaplastic lymphoma kinase increases cellular proliferation and oncogene up-regulation in rat 1a fibroblasts.

FASEB J. 1997;11(12):965-972.

Leventaki V, Drakos E, Medeiros LJ, et al.

NPM-ALK oncogenic kinase promotes cell-

haematologica | 2019; 104(2)

CDK4/CDK6 inhibitors in ALK-positive lymphomas

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

cycle progression through activation of JNK/cJun signaling in anaplastic large-cell lymphoma. Blood. 2007;110(5):1621-1630.

Gu TL, Tothova Z, Scheijen B, Griffin JD, Gilliland DG, Sternberg DW. NPM-ALK fusion kinase of anaplastic large-cell lymphoma regulates survival and proliferative signaling through modulation of FOXO3a.

Blood. 2004;103(12):4622-4629.

Fernandez-Vidal A, Mazars A, Gautier EF, Prevost G, Payrastre B, Manenti S. Upregulation of the CDC25A phosphatase down-stream of the NPM/ALK oncogene participates to anaplastic large cell lymphoma enhanced proliferation. Cell Cycle.

2009;8(9):1373-1379.

Bueno MJ, Malumbres M. MicroRNAs and the cell cycle. Biochim Biophys Acta. 2011; 1812(5):592-601.

Garzon R, Calin GA, Croce CM. MicroRNAs in Cancer. Annu Rev Med. 2009;60:167-179.

Hoareau-Aveilla C, Merkel O, Meggetto F. MicroRNA and ALK-positive anaplastic large cell lymphoma. Front Biosci (Schol Ed). 2015;7:217-225.

Hoareau-Aveilla C, Valentin T, Daugrois C, et al. Reversal of microRNA-150 silencing disadvantages crizotinib-resistant NPMALK(+) cell growth. J Clin Invest.

2015;125(9):3505-3518.

Merkel O, Robert G, Hamacher F, Grabner L, Greil R, Kenner L. Subtype specific expression of immnune-modulating miR155 and miR-146a in anaplastic large cell lymphoma. Cancer Res. 2009;72(8): supplemental 1.

Dejean E, Renalier MH, Foisseau M, et al.

Hypoxia-microRNA-16 downregulation induces VEGF expression in anaplastic lymphoma kinase (ALK)-positive anaplastic large-cell lymphomas.

Leukemia.

2011;25(12):1882-1890.

Desjobert C, Renalier MH, Bergalet J, et al.

MiR-29a down-regulation in ALK-positive anaplastic large cell lymphomas contributes to apoptosis blockade through MCL-1 overexpression. Blood. 2011;117(24):6627-6637.

Congras A, Caillet N, Torossian N, et al.

Doxorubicin-induced loss of DNA topoisomerase II and DNMT1- dependent suppression of MiR-125b induces chemoresistance in ALK-positive cells. Oncotarget.

2018;9(18):14539-14551.

He XX, Kuang SZ, Liao JZ, et al. The regulation of microRNA expression by DNA methylation in hepatocellular carcinoma.

Mol Biosyst. 2015;11(2):532-539.

Yang G, Xiong G, Cao Z, et al. miR-497 expression, function and clinical application in cancer. Oncotarget. 2016; 7(34):55900-55911.

Finnerty JR, Wang WX, Hebert SS, Wilfred BR, Mao G, Nelson PT. The miR-15/107

haematologica | 2019; 104(2)

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

group of microRNA genes: evolutionary biology, cellular functions, and roles in human diseases. J Mol Biol. 2010; 402(3):491-509.

Iorio MV, Casalini P, Tagliabue E, Menard S, Croce CM. MicroRNA profiling as a tool to understand prognosis, therapy response and resistance in breast cancer. Eur J Cancer. 2008;44(18):2753-2759.

Pikman Y, Alexe G, Roti G, et al.

Synergistic Drug Combinations with a CDK4/6 Inhibitor in T-cell Acute Lymphoblastic Leukemia. Clin Cancer Res. 2017;23(4):1012-1024.

Sawai CM, Freund J, Oh P, et al.

Therapeutic targeting of the cyclin D3:CDK4/6 complex in T cell leukemia.

Cancer Cell. 2012;22(4):452-465.

Lamant L, Espinos E, Duplantier M, et al.

Establishment of a novel anaplastic largecell lymphoma-cell line (COST) from a 'small-cell variant' of ALCL. Leukemia.

2004;18(10):1693-1698.

Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image analysis. Nat Methods. 2012;9(7):671-675.

Thompson MA, Stumph J, Henrickson SE, et al. Differential gene expression in anaplastic lymphoma kinase-positive and anaplastic lymphoma kinase-negative anaplastic large cell lymphomas. Hum Pathol. 2005;36(5):494-504.

Nagel S, Leich E, Quentmeier H, et al.

Amplification at 7q22 targets cyclin-dependent kinase 6 in T-cell lymphoma.

Leukemia. 2008;22(2):387-392.

Degenhardt T, Wuerstlein R, Eggersmann T, Harbeck N. The safety of palbociclib for the treatment of advanced breast cancer.

Expert Opin Drug Saf. 2018;28:1-6.

He JF, Luo YM, Wan XH, Jiang D. Biogenesis of MiRNA-195 and its role in biogenesis, the cell cycle, and apoptosis. J Biochem Mol Toxicol. 2011;25(6):404-408.

Steinhilber J, Bonin M, Walter M, Fend F, Bonzheim I, Quintanilla-Martinez L. Nextgeneration sequencing identifies deregulation of microRNAs involved in both innate and adaptive immune response in ALK+ ALCL. PLoS One. 2015;10(2):e0117780.

Liu X, Chen X, Yu X, et al. Regulation of microRNAs by epigenetics and their interplay involved in cancer. J Exp Clin Cancer Res. 2013;32:96.

Deng H, Guo Y, Song H, et al. MicroRNA195 and microRNA-378 mediate tumor growth suppression by epigenetical regulation in gastric cancer. Gene. 2013; 518(2):351-359.

Li W, Jin X, Deng X, Zhang G, Zhang B, Ma L. The putative tumor suppressor microRNA-497 modulates gastric cancer cell proliferation and invasion by repressing eIF4E. Biochem Biophys Res Commun.

2014;449(2):235-240.

38. Luo Q, Li X, Gao Y, et al. MiRNA-497 regulates cell growth and invasion by targeting cyclin E1 in breast cancer. Cancer Cell Int. 2013;13(1):95.

39. Furuta M, Kozaki K, Tanimoto K, et al. The tumor-suppressive miR-497-195 cluster targets multiple cell-cycle regulators in hepatocellular carcinoma. PLoS One. 2013; 8(3):e60155.

40. Zhang Y, Zhang Z, Li Z, et al. MicroRNA497 inhibits the proliferation, migration and invasion of human bladder transitional cell carcinoma cells by targeting E2F3.

Oncol Rep. 2016;36(3):1293-1300.

41. Ozata DM, Caramuta S, VelazquezFernandez D, et al. The role of microRNA deregulation in the pathogenesis of adrenocortical carcinoma. Endocr Relat Cancer.

2011;18(6):643-655.

42. Xu JW, Wang TX, You L, et al. Insulin-like growth factor 1 receptor (IGF-1R) as a target of MiR-497 and plasma IGF-1R levels associated with TNM stage of pancreatic cancer. PLoS One. 2014;9(3):e92847.

43. Shen L, Li J, Xu L, et al. miR-497 induces apoptosis of breast cancer cells by targeting Bcl-w. Exp Ther Med. 2012;3(3):475-480.

44. Kollmann K, Heller G, Schneckenleithner C, et al. A kinase-independent function of CDK6 links the cell cycle to tumor angiogenesis. Cancer Cell. 2013;24(2):167-181.

45. Ferreri AJ, Govi S, Pileri SA, Savage KJ. Anaplastic large cell lymphoma, ALK-positive. Crit Rev Oncol Hematol. 2012;83 (2):293-302.

46. Agirre X, Vilas-Zornoza A, JimenezVelasco A, et al. Epigenetic silencing of the tumor suppressor microRNA Hsa-miR124a regulates CDK6 expression and confers a poor prognosis in acute lymphoblastic leukemia. Cancer Res. 2009;69(10): 4443-4453.

47. Oliveira LH, Schiavinato JL, Fraguas MS, et al. Potential roles of microRNA-29a in the molecular pathophysiology of T-cell acute lymphoblastic leukemia. Cancer Sci. 2015;106(10):1264-1277.

48. Hu MG, Deshpande A, Enos M, et al. A requirement for cyclin-dependent kinase 6 in thymocyte development and tumorigenesis. Cancer Res. 2009;69(3):810-818.

49. Baughn LB, Di Liberto M, Wu K, et al. A novel orally active small molecule potently induces G1 arrest in primary myeloma cells and prevents tumor growth by specific inhibition of cyclin-dependent kinase 4/6.

Cancer Res. 2006;66(15):7661-7667.

50. Bonvini P, Zorzi E, Mussolin L, et al. The effect of the cyclin-dependent kinase inhibitor flavopiridol on anaplastic large cell lymphoma cells and relationship with NPM-ALK kinase expression and activity.

Haematologica. 2009;94(7):944-955.

359

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(2):360-369

Chronic Lymphocytic Leukemia

Tailored approaches grounded on immunogenetic features for refined prognostication in chronic lymphocytic leukemia

Panagiotis Baliakas,1* Theodoros Moysiadis,2* Anastasia Hadzidimitriou,1,2 Aliki Xochelli,1,2 Sabine Jeromin,3 Andreas Agathangelidis,2 Mattias Mattsson,1 LesleyAnn Sutton,1 Eva Minga,2 Lydia Scarfò,4 Davide Rossi,5 Zadie Davis,6 Neus Villamor,7 Helen Parker,8 Jana Kotaskova,9 Evangelia Stalika,2,10 Karla Plevova,9 Larry Mansouri,1 Diego Cortese,1 Alba Navarro,7 Julio Delgado,11 Marta Larrayoz,8 Emma Young,1 Achilles Anagnostopoulos,10 Karin E. Smedby,12 Gunnar Juliusson,13 Oonagh Sheehy,14 Mark Catherwood,14 Jonathan C. Strefford,8 Niki Stavroyianni,10 Chrysoula Belessi,15 Sarka Pospisilova,9 David Oscier,6 Gianluca Gaidano,16 Elias Campo,7,17 Claudia Haferlach,3 Paolo Ghia,4 Richard Rosenquist,18 Kostas Stamatopoulos;1,2 on behalf of The European Research Initiative on CLL (ERIC)

Department of Immunology, Genetics and Pathology, Science for Life Laboratory, Uppsala University, Sweden; 2Institute of Applied Biosciences, Center for Research and Technology Hellas, Thessaloniki, Greece; 3MLL Munich Leukemia Laboratory, Munich, Germany; 4Division of Experimental Oncology, IRCCS Istituto Scientifico San Raffaele and Università Vita-Salute San Raffaele, Milan, Italy; 5Oncology Institute of Southern Switzerland, Bellinzona, Switzerland; 6Department of Haematology, Royal Bournemouth Hospital, UK; 7Hemopathology Unit, Hospital Clinic, Barcelona, Spain; 8Cancer Genomics, Academic Unit of Cancer Sciences, Cancer Research UK Centre and Experimental Cancer Medicine Centre, Faculty of Medicine, University of Southampton, UK; 9Central European Institute of Technology, Masaryk University and University Hospital Brno, Czech Republic; 10Hematology Department and HCT Unit, G. Papanicolaou Hospital, Thessaloniki, Greece; 11Hematology Department, Hospital Clinic, Barcelona, Spain; 12Department of Medicine, Solna, Clinical Epidemiology Unit, Karolinska Institutet, Stockholm, Sweden; 13Lund University and Hospital Department of Hematology, Lund Stem Cell Center, Sweden; 14Department of Hemato-Oncology, Belfast City Hospital, UK; 15Hematology Department, Nikea General Hospital, Pireaus, Greece; 16 Division of Hematology, Department of Translational Medicine, Amedeo Avogadro University of Eastern Piedmont, Novara, Italy; 17Department of Pathology, University of Barcelona, Spain and 18Department of Molecular Medicine and Surgery, Karolinska Institutet, Stockholm, Sweden 1

Correspondence: kostas.stamatopoulos@gmail.com

Received: April 12, 2018.

Accepted: September 25, 2018.

Pre-published: September 27, 2018. doi:10.3324/haematol.2018.195032 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/360 ©2019 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.

360

*PB and TM contributed equally to this work as first authors.

ABSTRACT

C

hronic lymphocytic leukemia (CLL) patients with differential somatic hypermutation status of the immunoglobulin heavy variable genes, namely mutated or unmutated, display fundamental clinico-biological differences. Considering this, we assessed prognosis separately within mutated (M-CLL) and unmutated (U-CLL) CLL in 3015 patients, hypothesizing that the relative significance of relevant indicators may differ between these two categories. Within Binet A M-CLL patients, besides TP53 abnormalities, trisomy 12 and stereotyped subset #2 membership were equivalently associated with the shortest time-tofirst-treatment and a treatment probability at five and ten years after diagnosis of 40% and 55%, respectively; the remaining cases exhibited 5-year and 10-year treatment probability of 12% and 25%, respectively.

Within Binet A U-CLL patients, besides TP53 abnormalities, del(11q) and/or SF3B1 mutations were associated with the shortest time-to-firsttreatment (5- and 10-year treatment probability: 78% and 98%, respectively); in the remaining cases, males had a significantly worse prognosis than females. In conclusion, the relative weight of indicators that can accurately risk stratify early-stage CLL patients differs depending on the somatic hypermutation status of the immunoglobulin heavy variable genes of each patient. This finding highlights the fact that compartmentalized approaches based on immunogenetic features are necessary to refine and tailor prognostication in CLL. haematologica | 2019; 104(2)

Refining prognosis in early-stage CLL

Introduction Despite mounting evidence for the existence of distinct biological variants of chronic lymphocytic leukemia (CLL), the 2016 update of the World Health Organization (WHO) classification still considers CLL as a single, homogeneous entity, in contrast to other hematologic malignancies (e.g. diffuse large B-cell lymphoma, DLBCL) which are segregated in different subgroups, based on the integration of genetic, morphological, immunophenotypic and clinical features.1 Since the introduction of the Rai and Binet clinical staging systems in the 1970s,2,3 it has become increasingly evident that the clinical heterogeneity in CLL is linked to and reflects the underlying biological heterogeneity. Hence, several initiatives have focused on identifying biomarkers that would refine prognostication, especially for cases who present with early stage disease, who nowadays represent the great majority of patients (80-85%).4-12 Consequently, numerous prognostic indices have been proposed; however, none has been adopted in every-day clinical practice.13 This is partly due to the fact that different variables have been assessed in each evaluated cohort while the actual routine diagnostic and monitoring practice varies between different institutions. Moreover, most reported cohorts were rather small, thus inherently limited in their capacity to both encompass the remarkable clinico-biological heterogeneity of CLL and reveal possible interactions and interdependencies among the evaluated prognosticators.

The clonotypic B-cell receptor immunoglobulin (BcR IG) is a unique molecular signature for every CLL clone, present from its genesis and remaining unaltered throughout the course of the disease, thus sharply contrasting other tumor-derived biomarkers.14-19 Seminal studies from the late 1990s have established that the somatic hypermutation (SHM) status of the immunoglobulin heavy variable (IGHV) gene expressed by the clonotypic BcR IG is a robust prognostic and predictive biomarker for CLL, stratifying patients into two non-interchangeable categories with different clinical behavior.20,21 More specifically, CLL with a significant SHM load (“mutated” CLL, MCLL) generally follow an indolent clinical course, whereas CLL carrying no or few mutations (“unmutated” CLL, UCLL) generally have an aggressive disease and an overall inferior response to chemoimmunotherapy.22-24 This subclassification into M-CLL and U-CLL reflects fundamental clinico-biological differences extending from the genomic and epigenomic to the transcriptomic and proteomic levels, alluding to distinct ontogeny and evolution patterns, including response to treatment, for the two patient categories.14,24-27 That said, within both M-CLL and U-CLL, a sizeable proportion of cases exhibit a clinico-biological behavior pattern that deviates from the expected, thus highlighting that the heterogeneity of CLL persists even within a given SHM category.28-31 The paradigmatic example is offered by CLL subset #2, defined by the expression of stereotyped IGHV3-21/IGLV3-21 BcR IG, within which M-CLL cases follow an aggressive clinical course similar to U-CLL.30,32,33 Notably, other established prognosticators such as cytogenetic aberrations or recurrent gene mutations are asymmetrically distributed within M-CLL or U-CLL.10,34-36 On these grounds, it is not unreasonable to think that definitive conclusions about the precise clinical implications of haematologica | 2019; 104(2)

any given biomarker should be drawn only after considering the SHM status of the clonotypic BcR IG. In this study, we followed a compartmentalized approach where we assessed the prognostic impact of traditional and novel prognostic parameters separately within M-CLL and U-CLL in a large multi-institutional cohort of well characterized CLL patients, based on the hypothesis that not all variables would carry equal weight within the two SHM categories. Considering that the key challenge at the time of diagnosis is determining if, and consequently when, early stage/asymptomatic patients will require treatment, we focused on identifying a robust prognostication scheme for time-to-first-treatment (TTFT) in these separate disease categories.

Methods Patients’ characteristics Overall, 2366 general practice patients with CLL diagnosed following the 2008 International Workshop on CLL (IWCLL) diagnostic criteria37 from 10 European institutions were included in this multicenter retrospective study (Online Supplementary Table S1). The external validation cohort comprised of 649 Binet A cases evaluated at the Munich Leukemia Laboratory (n=508) and from a Scandinavian population-based study (n=141) (Online Supplementary Table S2). Ethical approval was granted by the local review committees and informed consent was collected according to the Declaration of Helsinki.

Methodologies Detailed information about the methodologies used to analyze the prognostic markers and definitions about stereotyped subsets and the criteria for subset assignment are provided in the Online Supplementary Appendix. Briefly: i) mutational screening was performed for the following genes: NOTCH1, entire exon 34 or targeted analysis for del7544-45/p.P2514Rfs*4; TP53, exons 4–8 but also exons 9–10 for some centers; SF3B1, exons 14–16; BIRC3, exons 6–9 and MYD88, exons 3 and 5 or targeted analysis for p.L265P. The great majority (80%) of cases included in this study were screened for the aforementioned recurrent mutations by Sanger sequencing. In the remaining cases (20%), next generation sequencing (NGS) was applied and only those clones with higher than 10% variant allele frequency (VAF >10%) were considered; ii) fluorescence in situ hybridization (FISH) was performed in 1825 (77%) cases using probes for the 13q14, 11q22, 17p13 regions and trisomy 12 (cut off: 5%) while results were interpreted following Döhner’s hierarchical model;38 and iii) sequence analysis and interpretation of IGHV-IGHD–IGHJ rearrangements (including BcR IG stereotypy) was performed as described previously.39

Statistical analysis We assessed the prognostic impact on TTFT of the following variables: age at the time of diagnosis; sex; CD38 expression; cytogenetic aberrations [del(17p), del(11q), del(13q) and trisomy 12 (+12)]; mutations within the TP53, SF3B1, NOTCH1, BIRC3 and MYD88 genes; and immunogenetic features, including borderline germline identity (GI: 97-97.99%) and assignment to stereotyped subsets #1, #2 and #4.

The following methodology was applied separately for M-CLL and U-CLL patients. The χ2 test was used to assess the homogeneity of all categorical prognostic variables within the different Binet stages (A, B and C). In order to evaluate homogeneity and detect significant differences within each categorical variable, the Bonferroni correction was applied to adjust for multiple compari361

P. Baliakas et al.

A

B

Figure 1. Kaplan-Meier curves for time-to-first-treatment (TTFT). (A) In Binet A, B and C mutated chronic lymphocytic leukemia (M-CLL) patients and (B) Binet A, B and C unmutated chronic lymphocytic leukemia (U-CLL) patients.

son error and the significance level was set to P<0.017. TTFT was evaluated from the diagnostic date until the date of initial treatment; untreated cases were censored at the time of last follow up.

Survival curves were constructed using the Kaplan-Meier method, and the log-rank test was applied to determine statistically significant differences between survival proportions.

The univariable Cox regression model was used to determine the most important prognostic factors. Multivariable Cox regression analysis was subsequently applied to evaluate the simultaneous effect of the important variables on TTFT. For the multivariable Cox analysis, we considered only those cases with available data for all the factors included in the model (n=918 for M-CLL and n=384 for U-CLL) on the grounds that imputing the values of the biomarkers could introduce substantial bias. The same rule was applied for the binary recursive partitioning. That said, no major differences were observed between the entire cohort and the proportion of cases included in the multivariable Cox analysis (Online Supplementary Table S3); this suggests that the patients included in the multivariable analysis were representative of the entire cohort.

The proportional hazard assumption was assessed for both the univariable and multivariable analysis. The stability of the multivariable Cox model was internally validated using a bootstrapping procedure. Harrellâ&#x20AC;&#x2122;s C index was also calculated to further assess the discriminatory power of the multivariable analysis when: a) all important prognostic variables were included in the model; and b) the resultant prognostic index was included in the model as the sole prognostic variable (C=1 indicates perfect discrimination; C=0.5 equates to chance).40 Binary recursive partitioning, based on the development of conditional inference trees, was used to further validate the results of Cox regression analysis.41 The patients were initially divided in subgroups (terminal nodes) with different survival behavior. An amalgamation algorithm was subsequently applied to merge the terminal nodes that did not exhibit a significant difference in survival. All tests were two-sided and P<0.05 was considered statistically significant. Statistical analyses were performed using the Statistica Software 10.0 (StatSoftInc, Tulsa, OK, USA) and R-3.2.2.

Details are provided in the Online Supplementary Appendix.

362

Results Prognostic impact of clinical staging within M-CLL and U-CLL The median follow-up time for the entire cohort was 7.1 years (range, 0.1-33.1) with a median TTFT of 6.4 years (95%CI: 0.01-20.2 years) (Online Supplementary Figure S1). As expected, in both M-CLL and U-CLL, Binet A patients exhibited significantly longer TTFT compared to Binet B and C cases (Figure 1). Most likely, this reflects the sharp clinico-biological differences between Binet A and Binet B/C patients while also underscoring the current indications for treatment that broadly overlap with the criteria for Binet stage B and C assignment (Online Supplementary Tables S4 and S5). In keeping with previous reports, M-CLL cases (n=1364, 58%) exhibited significantly (P<0.0001) longer TTFT compared to U-CLL cases (n=1002, 42%) (Online Supplementary Figure S2A).

Significant differences (P<0.0001) between M-CLL versus U-CLL remained when the analysis was restricted to Binet A cases (Online Supplementary Figure S2B).

Prompted by these findings, and also considering that Binet A cases predominated in both M-CLL and U-CLL (90% and 67%, respectively), we focused our attention on 1900 Binet A cases (M-CLL: n=1224; U-CLL: n=676). We refrained from investigating the impact of biomarkers within Binet B/C cases since the limited number of cases within each subgroup would not allow firm conclusions to be drawn.

Analysis for time-to-first-treatment within early stage mutated chronic lymphocytic leukemia Univariable Cox regression analysis within M-CLL Binet A cases (n=1224) revealed that male sex, CD38 positivity, +12, subset #2, TP53 abnormalities (TP53abn) and borderline IGHV gene germline identity (GI: 97-97.9%) were associated with significantly shorter TTFT (Table 1).

Bordeline GI remained significant in the univariable analysis even when subset #2 cases were excluded. TP53abn, haematologica | 2019; 104(2)

Refining prognosis in early-stage CLL Table 1. Univariable and multivariable analysis for time-to-first-treatment within Binet A mutated chronic lymphocytic leukemia (M-CLL) and unmutated chronic lymphocytic leukemia (U-CLL). Harrellâ&#x20AC;&#x2122;s C-index (standard error in parenthesis) was calculated as 0.745 (se=0.013) and 0.753 (se=0.013) for the M-CLL and the U-CLL, respectively. Internal bootstrap validation for the multivariable analysis. Means of the Hazard Ratio (HR), 95% Confidence Interval (CI) and percentage of selection within 1000 bootstrap samples are displayed per factor.

M-CLL/Binet A Univariable analysis (n=1224) HR 95% CI P Male Age >65 years CD38 expression idel(13q) +12 del(11q) TP53abn NOTCH1 SF3B1 MYD88 GI:97-97.99% IGHV3-23 IGHV3-21* Subset #2 Subset #4

1.352 0.976 2.017 0.761 2.428 2.010 3.347 2.107 1.856 1.551 1.571 1.025 1.466 3.384 0.658

1.063-1.718 0.769-2.237 1.474-2.760 0.573-1.010 1.691-3.487 0.944-4.281 2.175-5.151 0.993-4.447 0.913-3.771 0.680-3.538 1.069-2.308 0.719-1.461 0.605-3.555 1.594-7.181 0.309-1.398

0.013 0.842 <0.0001 0.059 <0.0001 0.071 <0.0001 0.052 0.088 0.296 0.021 0.89 0.396 0.001 0.277

U-CLL/Binet A Univariable analysis (n=672) HR 95% CI P Male Age >65 years CD38 expression del13q** +12 del(11q) TP53abn NOTCH1 SF3B1 BIRC3 GI:100% IGHV1-69 Subset #1

1.341 0.845 1.144 0.902 0.751 1.640 1.613 1.142 1.892 1.206 1.231 1.190 1.382

1.106-1.626 0.701-1-020 0.832-1.575 0.729-1.117 0.568-0.993 1.292-2.082 1.213-2.146 0.874-1.493 1.335-2.682 0.532-2.732 0.993-1.524 0.959-1.457 0.939-2.033

0.028 0.079 0.405 0.347 0.045 <0.0001 0.001 0.328 <0.0001 0.652 0.056 0.112 0.099

Multivariable analysis (n=918) HR 95% CI P 1.354 1.058 2.565 2.685 0.988 3.268 -

1.007-1.820 0.678-1.652 1.665-3.952 1.607-4.485 0.568-1.717 1.155-9.244 -

0.044 0.801 <0.0001 0.0001 0.966 0.025 -

U-CLL/Binet A Multivariable analysis (n=384) HR 95% CI P 1.610 0.898 1.432 2.357 1.624 -

1.229-2.109 0.632-1.274 1.071-1.916 1.548-3.589 1.129-2.337 -

0.0005 0.547 0.0151 <0.0001 0.008 v -

Internal bootstrap validation Factors (mean) Selection HR 95% CI percentage 1.406

1.044-1.894

57%

1.048

0.666-1.648

8.4%

2.669

1.769-4.031

98.2%

3.064

1.864-5.047

95.6%

0.940

0.529-1.679

3.4%

3.885

1.353-12.267

62%

Internal bootstrap validation Factors (mean) Selection HR 95% CI percentage 1.630

1.239-2.146

93.8%

0.912 1.475 1.475

0.640-1.299 1.098-1.983 1.098-1.983

10.4% 69.1% 69.1%

1.682

1.159-2.442

73.4%

CD38 expression: positivity >30%; idel(13q): isolated deletion of chromosome 13q; +12: trisomy 12; del(11q): deletion of chromosome 11q; TP53abn: deletion of chromosome 17p (del(17p)) and/or TP53 mutation; GI: germline identity; IGHV3-23: usage of IGHV3-23; IGHV3-21*: usage of IGHV3-21 non subset #2; subset #2: assignment to stereotyped subset #2; subset #4: assignment to stereotyped subset #4. **Due to the low number of cases with available data for all abnormalities detected by fluorescence in situ hybridization (FISH), deletion of chromosome 13q was used instead of idel(13q) for the U-CLL cohort.

+12, subset #2 membership and male sex retained independent adverse prognostic significance in the multivariable analysis (n=918) (Table 1). Of note, TP53abn, +12 and subset #2 membership identified 3 groups of almost mutually exclusive cases (altogether: n=153, 16% of all Binet A M-CLL) (Figure 2A) with no statistical differences in TTFT (median TTFT for TP53abn, +12, subset #2: 5.4, 7.1 and 4.1 years respectively; P=0.19) (Figure 2B).

To further validate the results of the Cox regression analysis, we applied recursive partitioning models within a conditional inference framework. TP53abn, +12 and subset #2 membership were again identified as the most haematologica | 2019; 104(2)

important predictors since the 3 groups defined by these predictors displayed similar TTFT that was significantly shorter from the remaining Binet A M-CLL (Figure 3A).

Analysis for time-to-first-treatment within early stage unmutated chronic lymphocytic leukemia Univariable Cox regression analysis revealed that TP53abn, SF3B1 mutations, del(11q) and male sex had a shorter TTFT within Binet-A U-CLL (n=676), while +12 had a longer TTFT (Table 1). TP53abn, SF3B1 mutations, del(11q) and male sex retained significance in the multivariable analysis (n=384) (Table 1). TP53abn and/or SF3B1 363

P. Baliakas et al.

A

B

P=0.19

C

D

P=0.47

Figure 2. Subgroups of patients with similar prognosis within mutated chronic lymphocytic leukemia (M-CLL) and unmutated chronic lymphocytic leukemia (U-CLL).

(A) TP53abn, trisomy 12 (+12) and stereotyped subset #2 assignment define three almost non-overlapping groups within early stage M-CLL. (B) Binet A M-CLL cases with TP53abn, +12 or assignment to stereotyped subset #2 display similar time-to-first-treatment (TTFT.) (C) Distribution of TP53abn, SF3B1 mutations and del(11q) in Binet A U-CLL. (D) Binet A U-CLL cases carrying TP53abn, SF3B1 mutations or del(11q) exhibit similar TTFT. N: number of patients.

A

B

Figure 3. Application of binary recursive partitioning in mutated chronic lymphocytic leukemia (M-CLL) and unmutated chronic lymphocytic leukemia (U-CLL). (A) Decision tree for Binet A M-CLL based on binary recursive partitioning and the subsequent application of an amalgamation algorithm. Trisomy 12 (+12), TP53abn and subset #2 membership were found to be the most significant factors as determined by the partitioning algorithm. The Binet A population is split in 4 terminal nodes (4, 5, 6 and 7). The amalgamation algorithm applied subsequently merged 3 of them in a larger terminal node. In particular, +12 was considered as the covariate with the strongest association to time-to-first-treatment (TTFT). Amongst patients lacking +12, TP53abn was the co-variate with the strongest association to TTFT and so on. After applying the amalgamation algorithm, patients with +12 and/or TP53abn and/or assignment to subset #2 were grouped into a larger node, resulting in 2 terminal nodes. The splitting is performed from right to left, following the criterion of strongest factor association with TTFT. The right branch represents the presence of a particular factor and the left branch the absence of that factor. P-value corresponds to a log-rank scores based test. The Kaplan-Meier curves estimate the TTFT of patients within each terminal node and n represents the number of patients per node. (B) Decision tree for Binet A U-CLL based on binary recursive partitioning and the subsequent application of an amalgamation algorithm. Male sex, TP53abn and del(11q) were the most significant factors as determined by the partitioning algorithm. The Binet A population was split into 6 terminal nodes (4, 5, 6, 9, 10 and 11). The amalgamation algorithm applied merged 3 of the terminal nodes into a larger terminal node. Sex was deemed to be the co-variate with the strongest association to TTFT. Amongst male patients, TP53abn was the co-variate with strongest association to TTFT. Amongst female patients, del(11q) was the co-variate with strongest association to TTFT and so on. After applying the amalgamation algorithm, male patients without TP53abn and with del(11q), and female patients with del(11q) and/or TP53abn were grouped into a larger node. The final number of terminal nodes was 4. The splitting is performed from top to bottom, following the criterion of strongest factor association with TTFT. The right branch represents the presence of a particular factor and the left one the absence of that factor. P-value corresponds to a log-rank scores-based test. The Kaplan-Meier curves estimate the TTFT of patients within each terminal node and n represents the number of patients per node.

364

haematologica | 2019; 104(2)

Refining prognosis in early-stage CLL

mutations and/or del(11q) [TP53abn/SF3B1mut/del(11q)] were positive in 146 cases (42% of the evaluated cohort with available data for all 3 parameters) (Figure 2C) and exhibited a similar TTFT (median TTFT for TP53abn, SF3B1mut, del(11q): 1.8, 2.8 and 2.1 years, respectively; P=0.47) (Figure 2D). The co-occurrence of these aberrations was not found to aggravate the prognosis when compared to single aberrations alone (Online Supplementary Figure S3). Male sex was associated with a shorter TTFT (median TTFT: 3.9 years, 95%CI: 0.1-5.9 years) amongst non-TP53abn/SF3B1mut/del(11q) U-CLL cases (Online Supplementary Figure S4).

Similar to M-CLL, we applied recursive partitioning (Figure 3B) which highlighted TP53abn, del(11q) and male sex as the most important variables within Binet A U-CLL.

A

SF3B1 mutations did not reach significance, potentially due to the low number of cases with isolated SF3B1 mutations at the stage of the final nodes. In short, the evaluated U-CLL Binet A cohort was allocated to subgroups with differing TTFT, from shorter to longer, as follows: males with TP53abn; males with del(11q)/females with TP53abn and/or del(11q); males without TP53abn/del(11q); and, females without TP53abn/del(11q).

Two somatic hypermutation categories of chronic lymphocytic leukemia patients, two prognostic indices for time-to-first-treatment Based on the above, we developed two prognostic indices for assessing TTFT, tailored to each SHM category.

Within M-CLL, we defined 4 groups based on their clini-

B

Figure 4. Prognostic index for time-to-first-treatment (TTFT) for mutated chronic lymphocytic leukemia (M-CLL) and unmutated chronic lymphocytic leukemia (UCLL). (A) Prognostic index for TTFT within M-CLL: i) very high risk: Binet C; ii) high risk: Binet B; iii) intermediate risk: Binet A with at least one of the following: TP53abn or +12 or assignment to subset #2 and; iv) low risk: non-TP53abn/+12/#2 Binet A. (B) Prognostic index for TTFT within U-CLL: i) very high risk: Binet C; ii) high risk: Binet B; iii) intermediate risk: Binet A with at least one of the following: TP53abn or +SF3B1 mutations or del(11q) membership; iv) low risk: non-TP53abn/SF3B1mut/del(11q) male Binet A and; v) very low risk: non-TP53abn/SF3B1mut/del(11q) female Binet A.

A

B

Figure 5. Kaplan-Meier curves for time-to-first-treatment (TTFT) in the validation cohort. (A) Within mutated chronic lymphocytic leukemia (M-CLL), Binet A cases positive for TP53abn, trisomy 12 (+12) and stereotyped subset #2 assignment display similar TTFT. (B) No difference regarding TTFT among Binet A and unmutated chronic lymphocytic leukemia (U-CLL) cases carrying TP53abn, SF3B1 mutations or del(11q) in the validation cohort.

haematologica | 2019; 104(2)

365

P. Baliakas et al.

A

B

Figure 6. Kaplan-Meier curves for time-to-first-treatment (TTFT) for mutated chronic lymphocytic leukemia (M-CLL) and unmutated chronic lymphocytic leukemia (U-CLL) cases carrying trisomy 12 (+12). (A) +12 is an unfavorable prognosticator in M-CLL. (B) +12 is associated with a more indolent clinical course in U-CLL.

co-biological profiles at the time of diagnosis: i) very high risk: Binet C with identical 5- and 10-year treatment-probability (TP) of 92%; ii) high risk: Binet B, 5y-TP and 10yTP: 64% and 84%, respectively; iii) intermediate risk: Binet A with one of the following: TP53abn and/or +12 and/or subset #2 membership, 5y-TP and 10y-TP: 40% and 55%, respectively. [Of note, among 18 non-censored cases with no treatment indication for more than 10 years after diagnosis, 5 (30%) carried TP53abn]; and iv) low risk: non-TP53abn/+12/subset #2 Binet A, 5y-TP and 10y-TP: 12% and 25%, respectively (Figure 4A). Harrellâ&#x20AC;&#x2122;s C Index was calculated for the multivariable Cox model with the prognostic index above being the sole predictor and was found to equal 0.745 (se = 0.013).

In U-CLL, we could define 5 groups: i) very high risk: Binet C with 5- and 10-year TP of 100%; ii) high risk: Binet B, 5y-TP and 10y-TP: 90% and 100%, respectively; iii) intermediate risk: Binet A with one of the following: TP53abn and/or SF3B1mut and/or del(11q), 5y-TP and 10y-TP: 78% and 98%, respectively; v) low risk: nonTP53abn/SF3B1mut/del(11q) male Binet A, 5y-TP and 10yTP: 65% and 85%, respectively; and iv) very low risk: non-TP53abn/SF3B1mut/del(11q) female Binet A, 5y-TP and 10y-TP: 45% and 65%, respectively (Figure 4B).

Harrellâ&#x20AC;&#x2122;s C Index was calculated for the multivariable Cox model with the prognostic index being the sole predictor and was found to equal 0.753 (se = 0.013).

Internal validation In order to validate the results mentioned above, a bootstrapping procedure was performed. In early stage M-CLL patients, this showed that the average number of predictors included in the multivariable Cox model was 3.2 with three variables exhibiting selection percentages greater than 60%, i.e. TP53abn, +12 and subset #2 (Table 1). In early stage U-CLL patients, bootstrapping showed that the average number of predictors considered significant in the multivariable Cox model was 3.5. Four variables exhibited selection percentages greater than 60%, i.e.

TP53abn, SF3B1mut, del(11q) and male sex.

366

External validation Application of the above mentioned indices to the validation cohort (n=649) led to the following observations: i) in M-CLL, TP53abn, +12 and subset #2 membership (intermediate risk for M-CLL) exhibited similar TTFT, constituting a group (16% of all Binet cases also in the validation cohort) with almost identical 5y-TP (43%) and 10y-TP (60%) to those observed in the training cohort (40% and 55%, respectively) (Figure 5A); ii) similarly, in U-CLL, no difference was observed in TTFT for Binet A cases carrying TP53abn and/or SF3B1mut and/or del(11q) (intermediate risk for U-CLL) who exhibited similar 5y-TP and 10yTP to that of the training cohort (74% vs. 78% and 92% vs. 98%, respectively; P-values: non significant) (Figure 5B); iii) amongst the remaining nonTP53abn/SF3B1mut/del(11q) U-CLL Binet A cases, the difference between male and female patients did not reach statistical significance (P=0.2) (Online Supplementary Figure S5A and B).

Discussion Chronic lymphocytic leukemia constitutes a rather unique case amongst cancers in that the great majority of patients are asymptomatic at diagnosis and classified as early stage, and thus do not require immediate treatment.42 However, most patients will progress and meet the criteria for treatment initiation albeit at a variable time from diagnosis.37,42 Therefore, accurate prediction of the TTFT is of major importance for both patients and physicians having to address the challenge of living with and managing this (largely) invisible and incurable disease. In support of this argument, solid prognostication is increasingly recognized as both a priority and an unmet need by CLL patients themselves, who would benefit from accurate prognostic information in order to make educated life choices and, perhaps, also participate more actively in their care.

Several prognostic models have been developed for CLL haematologica | 2019; 104(2)

Refining prognosis in early-stage CLL

Figure 7. Prognostic algorithm regarding treatment probability for chronic lymphocytic leukemia (CLL). 5y-TP: treatment probability five years from diagnosis; 10yTP: treatment probability ten years from diagnosis; U-CLL: CLL with unmutated IGHV genes; M-CLL: CLL with mutated IGHV genes; TP53abn: deletion of chromosome 17p (del(17p)) and/or TP53 mutation; +12:trisomy 12, del(11q): deletion of chromosome 11q; SF3B1mut: SF3B1 mutation; #2: assignment to stereotyped subset #2. The pie chart refers to the entire cohort with each slice indicating the proportion of patients according to Binet clinical staging.

based on combinations of biomarkers and host-derived features. However, none has been adopted in everyday clinical practice,13 e.g. to help decide about the follow-up strategy amongst untreated patients. These indices were mainly based on Cox regression models, where each respective training/validation cohort was considered as a single group. Herein, we followed a novel approach evaluating prognosis separately within M-CLL and U-CLL focusing on TTFT in the largest ever series studied to this purpose. We report that within early stage M-CLL, TP53abn, +12 and assignment to stereotyped subset #2 identified a subgroup of patients with uniformly shorter TTFT compared to the remaining early Binet A M-CLL. Similarly, within U-CLL the presence of TP53abn, del(11q) and SF3B1 mutations was found to be associated with the shortest TTFT, whereas the remaining U-CLL female patients had a significantly longer TTFT compared to males.

Classification of CLL patients based on the SHM status of the clonotypic BcR IG into M-CLL and U-CLL categories offers robust prognostic information, differing from other prognostic markers (e.g. genetic aberrations) that may evolve over time.15,17,18,43 Of note, studies by us and others have documented an asymmetric distribution of certain genetic aberrations in patients with distinct immunogenetic features extending from the M-CLL or UCLL categorization to different stereotyped subsets.32,44,45 This has prompted speculation that particular modes of immune signaling initiated by specific BcR IG may trigger different pathways of clonal evolution leading to the emergence of distinct disease variants.

For these reasons, the BcR IG appears an obvious starting point for developing a biologically-orientated prognoshaematologica | 2019; 104(2)

tication scheme for CLL, as in our present study, offering the possibility to dissect the precise impact of a given biomarker within a particular immunogenetic category (e.g.

M-CLL or U-CLL). For example, within M-CLL, +12 was found to define a subgroup of patients with a TTFT similar to that of patients harboring TP53abn, whereas, in contrast, +12 was associated with favorable outcome within U-CLL (Figure 6A and B). Of note, mutations within the NOTCH1 gene had no impact on survival among U-CLL cases with +12 (Online Supplementary Figure S6). These findings may explain: i) why +12 is considered as an intermediate-risk aberration in prognostic indices where CLL is evaluated as one group regardless of the SHM status;38 and, ii) the contradictory results reported in different cohorts with different relative proportion of M-CLL and U-CLL patients, regarding the significance of a given indicator that can show an asymmetric distribution within each SHM group.

Our initial results based on Cox regression analysis were validated internally, being highly reproducible in an independent validation cohort. In particular, the median TTFT for the subgroups of patients with the shortest TTFT in both SHM categories, namely TP53abn/+12/#2 and TP53abn/SF3B1mut/del(11q) for M-CLL and U-CLL respectively, was almost identical between the validation and the training cohort. Interestingly, the latter exhibited significant differences in terms of the biological background compared to the training cohort (Online Supplementary Table S2); this may be taken as further evidence for the robustness of our approach, since similar results were obtained across cohorts with differing patient composition. Cox regression results were further confirmed through the application of an alternative statistical 367

P. Baliakas et al.

approach, namely binary recursive partitioning, which offers a different framework, thereby conveying a hierarchical order of importance and classification for the evaluated prognostic factors.

Admittedly, despite providing a robust risk stratification scheme, the prognostic indices proposed here will not solve the problem of outliers, while they also overlook the potential effect of other variables with proven prognostic significance, e.g. cytogenetic complexity or methylation signatures. Thus, they cannot be considered the last word in biomarker-orientated risk stratification for TTFT. They do, however, highlight the need for further studies, while providing the conceptual frame of compartimentalization.

It should be further emphasized that our approach is built on the pivotal role of IGHV SHM status in the prognostic setting, which appears to be less important in the era of novel agents, namely BTK and BCL2 inhibitors, where response rates seem to be similar between M-CLL and UCLL.46,47 Recently, the International Prognostic Index for patients with CLL (CLL-IPI) was developed for assessing overall survival (OS).7 This has provided a robust prognostic classification scheme as it includes well-characterized patients followed in the context of clinical trials. A caveat of CLL-IPI concerns the fact that the evaluated patients had been treated with various regimens in the context of different clinical trials. Moreover, CLL-IPI does not allow the identification of distinct groups within each SHM category. For example, following the CLL-IPI score, a young (<65 years), early-stage patient, negative for TP53abn and belonging to M-CLL would never be characterized as very high risk. It is important to note, therefore, that the clinically aggressive stereotyped subset #2 would be overlooked by the CLL-IPI as it largely falls into the category of M-CLL cases lacking TP53abn, with 60% of patients also under 65 years of age. This is not a trivial issue, considering that subset #2 accounts for up to 5% of all CLL requiring treatment and, therefore, is equal in size to the CLL-IPI very high risk group with very limited if any overlap.30 A final, more general concern is that CLL-IPI was developed based on the analysis of cases treated prior to the introduction of novel therapies, which are likely to change the treatment expectations and OS in CLL, thus eventually creating the need for new predictive schemes.48 Our study identified subset #2 membership and SF3B1 mutations as prognostically important biomarkers for early-stage M-CLL and U-CLL, respectively. In contrast, other recurrent gene mutations such as NOTCH1 or BIRC3 failed to reach significance even in univariable analysis. Interestingly, SF3B1 mutations are remarkably enriched within subset #2 (approx. 50% of cases harbor a mutation), however, their impact within this very aggressive subset remains equivocal.30,32 Overall, these results

References 1. Swerdlow SH, Campo E, Pileri SA, et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390.

368

emphasize the value of investigating IG sequences for stereotyped subset #2 membership (easily determined through the use of a free online tool available at: http://bat.infspire.org/arrest/assignsubsets/49) and searching for SF3B1 mutations in routine clinical practice as this would enable a more accurate assessment of prognosis (TTFT) at diagnosis of CLL. In conclusion, we propose a novel approach to prognostic assessment in CLL grounded on the fact that not all CLL are equal, but instead that M-CLL and U-CLL categories are fundamentally different regarding their ontogeny and molecular landscape, at least for early-stage patients (Figure 7). Our results support that compartmentalizing CLL with the BcR IG as the starting point allows accurate prognostication in early-stage CLL. This further shows that the relative weight of well established prognostic indicators differs based on the immunogenetic features of each individual case. From a broader perspective, such compartmentalized approaches might prove relevant in other B-cell lymphomas as well e.g. diffuse large B-cell lymphoma where different biomarkers are emerging as prognostically relevant for the activated B-cell or the germinal center subtype, respectively,50 that display distinct immunogenetic features and signaling signatures.50 Funding This work was supported in part by the Swedish Cancer Society, the Swedish Research Council, the Knut and Alice Wallenberg Foundation, Karolinska Institutet, Stockholm, the Lion’s Cancer Research Foundation, Uppsala, the Marcus Borgström Foundation and Selander’s Foundation, Uppsala; H2020 “AEGLE, An analytics framework for integrated and personalized healthcare services in Europe” by the EU; H2020 “MEDGENET, Medical Genomics and Epigenomics Network” (No.692298) by the EU; H2020 “CLLassify, Innovative risk assessment for individualizing treatment in chronic lymphocytic leukemia” (No.702714) by the EU; Associazione Italiana per la Ricerca sul Cancro AIRC Investigator grants #20246, and Special Program Molecular Clinical Oncology AIRC 5 per mille #9965; Progetti di Rilevante Interesse Nazionale (PRIN) #2015ZMRFEA, MIUR, Rome ,Italy; TRANSCAN-179 NOVEL JTC 2016; project CEITEC 2020 (LQ1601) by MEYS-CZ, project AZV-MH-CZ 15-30015A-4/2015; JCS was funded by Bloodwise (11052, 12036), the Kay Kendall Leukaemia Fund (873), Cancer Research UK (C34999/A18087, ECMC C24563/A15581), Wessex Medical Research and the Bournemouth Leukaemia Fund; Special Program Molecular Clinical Oncology 5 x 1000 No. 10007, Associazione Italiana per la Ricerca sul Cancro Foundation Milan, Italy; Progetto Ricerca Finalizzata RF-2011-02349712, Ministero della Salute, Rome, Italy.

TM is recipient of a Marie Sklodowska-Curie individual fellowship (grant agreement No. 702714), funded by the EU H2020 research and innovation programme.

2. Binet JL, Auquier A, Dighiero G, et al. A new prognostic classification of chronic lymphocytic leukemia derived from a multivariate survival analysis. Cancer.

1981;48(1):198-206.

3. Rai KR, Sawitsky A, Cronkite EP, Chanana AD, Levy RN, Pasternack BS. Clinical stag-

ing of chronic lymphocytic leukemia.

Blood. 1975;46(2):219-234.

4. Cramer P, Hallek M. Prognostic factors in chronic lymphocytic leukemia-what do we need to know? Nat Rev Clin Oncol.

2011;8(1):38-47.

5. Gentile M, Mauro FR, Rossi D, et al.

haematologica | 2019; 104(2)

Refining prognosis in early-stage CLL

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

Italian external and multicentric validation of the MD Anderson Cancer Center nomogram and prognostic index for chronic lymphocytic leukaemia patients: analysis of 1502 cases. Br J Haematol.

2014;167(2):224-232.

Haferlach C, Dicker F, Weiss T, et al.

Toward a comprehensive prognostic scoring system in chronic lymphocytic leukemia based on a combination of genetic parameters. Genes Chromosomes Cancer. 2010;49(9):851-859.

International CLLIPIwg. An international prognostic index for patients with chronic lymphocytic leukaemia (CLL-IPI): a metaanalysis of individual patient data. Lancet Oncol. 2016;17(6):779-790.

Morabito F, Cutrona G, Gentile M, et al.

Definition of progression risk based on combinations of cellular and molecular markers in patients with Binet stage A chronic lymphocytic leukaemia. Br J Haematol. 2009;146(1):44-53.

Oscier DG, Gardiner AC, Mould SJ, et al.

Multivariate analysis of prognostic factors in CLL: clinical stage, IGVH gene mutational status, and loss or mutation of the p53 gene are independent prognostic factors.

Blood. 2002;100(4):1177-1184.

Rossi D, Rasi S, Spina V, et al. Integrated mutational and cytogenetic analysis identifies new prognostic subgroups in chronic lymphocytic leukemia. Blood. 2013; 121(8):1403-1412.

Shanafelt TD, Jenkins G, Call TG, et al.

Validation of a new prognostic index for patients with chronic lymphocytic leukemia. Cancer. 2009;115(2):363-372.

Wierda WG, O'Brien S, Wang X, et al.

Prognostic nomogram and index for overall survival in previously untreated patients with chronic lymphocytic leukemia. Blood.

2007;109(11):4679-4685.

Baliakas P, Mattsson M, Stamatopoulos K, Rosenquist R. Prognostic indices in chronic lymphocytic leukaemia: where do we stand how do we proceed? J Intern Med. 2016;279(4):347-357.

Fabbri G, Dalla-Favera R. The molecular pathogenesis of chronic lymphocytic leukaemia. Nat Rev Cancer. 2016; 16(3): 145-162.

Sutton LA, Rosenquist R. Deciphering the molecular landscape in chronic lymphocytic leukemia: time frame of disease evolution. Haematologica. 2015;100(1):7-16.

Sutton LA, Rosenquist R. The complex interplay between cell-intrinsic and cellextrinsic factors driving the evolution of chronic lymphocytic leukemia. Semin Cancer Biol. 2015;34:22-35.

Landau DA, Carter SL, Stojanov P, et al.

Evolution and impact of subclonal mutations in chronic lymphocytic leukemia.

Cell. 2013;152(4):714-726.

Landau DA, Tausch E, Taylor-Weiner AN, et al. Mutations driving CLL and their evolution in progression and relapse. Nature.

2015;526(7574):525-530.

Stevenson FK, Caligaris-Cappio F. Chronic lymphocytic leukemia: revelations from the B-cell receptor. Blood. 2004; 103(12):4389-4395.

Damle RN, Wasil T, Fais F, et al. Ig V gene mutation status and CD38 expression as novel prognostic indicators in chronic lymphocytic leukemia. Blood. 1999;94(6):18401847.

Hamblin TJ, Davis Z, Gardiner A, Oscier

haematologica | 2019; 104(2)

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

DG, Stevenson FK. Unmutated Ig V(H) genes are associated with a more aggressive form of chronic lymphocytic leukemia.

Blood. 1999;94(6):1848-1854.

Fischer K, Bahlo J, Fink AM, et al. Longterm remissions after FCR chemoimmunotherapy in previously untreated patients with CLL: updated results of the CLL8 trial. Blood. 2016;127(2):208-215.

Rossi D, Terzi-di-Bergamo L, De Paoli L, et al. Molecular prediction of durable remission after first-line fludarabine-cyclophosphamide-rituximab in chronic lymphocytic leukemia. Blood. 2015;126(16):19211924.

Thompson PA, Tam CS, O'Brien SM, et al.

Fludarabine, cyclophosphamide, and rituximab treatment achieves long-term diseasefree survival in IGHV-mutated chronic lymphocytic leukemia. Blood. 2016;127(3):303309.

Kulis M, Heath S, Bibikova M, et al.

Epigenomic analysis detects widespread gene-body DNA hypomethylation in chronic lymphocytic leukemia. Nat Genet.

2012;44(11):1236-1242.

Marincevic M, Cahill N, Gunnarsson R, et al. High-density screening reveals a different spectrum of genomic aberrations in chronic lymphocytic leukemia patients with 'stereotyped' IGHV3-21 and IGHV434 B-cell receptors. Haematologica. 2010; 95(9):1519-1525.

Ntoufa S, Vardi A, Papakonstantinou N, et al. Distinct innate immunity pathways to activation and tolerance in subgroups of chronic lymphocytic leukemia with distinct immunoglobulin receptors. Mol Med. 2012;18:1281-1291.

Baliakas P, Hadzidimitriou A, Sutton LA, et al. Clinical effect of stereotyped B-cell receptor immunoglobulins in chronic lymphocytic leukaemia: a retrospective multicentre study. Lancet Haematology. 2014; 1(2):74-84.

Agathangelidis A, Vardi A, Baliakas P, Stamatopoulos K. Stereotyped B-cell receptors in chronic lymphocytic leukemia. Leuk Lymphoma. 2014;55(10):2252-2261.

B aliakas P, Agathangelidis A, Hadzidimitriou A, et al. Not all IGHV3-21 chronic lymphocytic leukemias are equal: prognostic considerations. Blood. 2015; 125(5):856-859.

Xochelli A, Baliakas P, Kavakiotis I, et al.

Chronic Lymphocytic Leukemia with Mutated IGHV4-34 Receptors: Shared and Distinct Immunogenetic Features and Clinical Outcomes. Clin Cancer Res. 2017; 23(17):5292-5301.

Jeromin S, Haferlach C, Dicker F, Alpermann T, Haferlach T, Kern W. Differences in prognosis of stereotyped IGHV3-21 chronic lymphocytic leukaemia according to additional molecular and cytogenetic aberrations.

Leukemia.

2016;30(11):2251-2253.

Baliakas P, Mattsson M, Hadzidimitriou A, et al. No improvement in long-term survival over time for chronic lymphocytic leukemia patients in stereotyped subsets #1 and #2 treated with chemo(immuno)therapy. Haematologica. 2018;103(4):e158-e161.

Baliakas P, Hadzidimitriou A, Sutton LA, et al. Recurrent mutations refine prognosis in chronic lymphocytic leukemia. Leukemia.

2015;29(2):329-336.

Jeromin S, Weissmann S, Haferlach C, et al.

SF3B1 mutations correlated to cytogenetics

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

and mutations in NOTCH1, FBXW7, MYD88, XPO1 and TP53 in 1160 untreated CLL patients. Leukemia. 2014;28(1):108-117.

Baliakas P, Hadzidimitriou A, Agathangelidis A, et al. Prognostic relevance of MYD88 mutations in CLL: the jury is still out. Blood. 2015;126(8):10431044.

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.

Dohner H, Stilgenbauer S, Benner A, et al.

Genomic aberrations and survival in chronic lymphocytic leukemia. N Engl J Med. 2000;343(26):1910-1916.

Agathangelidis A, Darzentas N, Hadzidimitriou A, et al. Stereotyped B-cell receptors in one-third of chronic lymphocytic leukemia: a molecular classification with implications for targeted therapies.

Blood. 2012;119(19):4467-4475.

Harrell FE, Jr., Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15(4):361-387.

Hothorn T, Hornik K, Zeileis A. Unbaised Recursive Partitioning: A Conditional Inference Framework.

Journal of Computational and Graphical Statistics.

2006;15(3):651-674.

Hallek M. Chronic lymphocytic leukemia: 2015 Update on diagnosis, risk stratification, and treatment. Am J Hematol.

2015;90(5):446-460.

Ljungstrom V, Cortese D, Young E, et al.

Whole-exome sequencing in relapsing chronic lymphocytic leukemia: clinical impact of recurrent RPS15 mutations.

Blood. 2016;127(8):1007-1016.

Rossi D, Spina V, Bomben R, et al.

Association between molecular lesions and specific B-cell receptor subsets in chronic lymphocytic leukemia. Blood. 2013; 121(24):4902-4905.

Sutton LA, Young E, Baliakas P, et al.

Different spectra of recurrent gene mutations in subsets of chronic lymphocytic leukemia harboring stereotyped B-cell receptors.

Haematologica. 2016;101(8):959-967.

Byrd JC, Furman RR, Coutre SE, et al.

Three-year follow-up of treatment-naive and previously treated patients with CLL and SLL receiving single-agent ibrutinib.

Blood. 2015;125(16):2497-2506.

Roberts AW, Davids MS, Pagel JM, et al.

Targeting BCL2 with Venetoclax in Relapsed Chronic Lymphocytic Leukemia.

N Engl J Med. 2016;374(4):311-322.

Lamanna N, O'Brien S. Novel agents in chronic lymphocytic leukemia.

Hematology Am Soc Hematol Educ Program. 2016;2016(1):137-145.

Bystry V, Agathangelidis A, Bikos V, et al.

ARResT/AssignSubsets: a novel application for robust subclassification of chronic lymphocytic leukemia based on B cell receptor IG stereotypy. Bioinformatics. 2015; 31(23):3844-3846.

Karube K, Enjuanes A, Dlouhy I, et al.

Integrating genomic alterations in diffuse large B-cell lymphoma identifies new relevant pathways and potential therapeutic targets. Leukemia. 2018;32(3):675-684.

369

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(2):370-379

Plasma Cell Disorders

Allogeneic transplantation of multiple myeloma patients may allow long-term survival in carefully selected patients with acceptable toxicity and preserved quality of life Christine Greil,1 Monika Engelhardt,1 Gabriele Ihorst,2 Katja Schoeller,1 Hartmut Bertz,1 Reinhard Marks,1 Robert Zeiser,1 Justus Duyster,1 Hermann Einsele,3 Jürgen Finke1 and Ralph Wäsch1

Department of Hematology, Oncology and Stem Cell Transplantation, Medical Center University of Freiburg, Faculty of Medicine; 2Clinical Trials Unit, Faculty of Medicine, University of Freiburg and 3Department of Internal Medicine II, University Hospital of Würzburg, Germany

1

ABSTRACT

D

Correspondence: ralph.waesch@uniklinik-freiburg.de

Received: June 29, 2018.

Accepted: September 14, 2018.

Pre-published: September 20, 2018. doi:10.3324/haematol.2018.200881 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/370 ©2019 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.

370

espite significantly improved survival and response rates in patients diagnosed with multiple myeloma, it still remains an incurable disease with a poor outcome, especially in high-risk groups. Allogeneic stem cell transplantation offers a potentially curative option but remains controversial due to considerable treatment-related toxicity. We analyzed 109 consecutive myeloma patients who had received reduced-intensity conditioning allogeneic transplantation at the Freiburg University Medical Center between 2000 and 2016. Although most patients were heavily pre-treated in high-risk constellations, the overall response rate was high with 70%, the median overall survival (OS) 39.2%, and the median progression-free survival (PFS) 14.2 months, with a median follow up of 71.5 months. Survival was significantly better in patients with response to previous therapies than in those with progressive disease (median OS 65 vs. 11.5 months, P=0.003; median PFS 18.4 vs. 5.1 months, P=0.001). Moreover, survival of patients transplanted in first-line was significantly prolonged compared to relapsed/refractory disease (median OS not reached vs. 21.6 months, P<0.001; median PFS 47.7 vs. 9.6 months, P<0.001). The non-relapse mortality was relatively low with a cumulative incidence of 12.4% at ten years. Acute graft-versus-host disease (GvHD) grade II-IV was observed in 25%, and moderate or severe chronic GvHD in 24%. Quality of life (QoL) assessed with the revised Myeloma Comorbidity Index before and after transplantation remained unchanged. Our data suggest that allogeneic transplantation in the context of novel immunotherapeutic approaches may enable long-term survival and even a potential cure in a carefully selected subgroup of high-risk multiple myeloma patients with acceptable toxicity and preserved QoL.

Introduction Despite a remarkable increase in effective treatment options, multiple myeloma (MM) still remains mostly incurable. Nevertheless, survival of patients diagnosed with MM has significantly improved over the last few years, although outcome may be poor with a median overall survival (OS) of only 2-3 years in subgroups of patients with higher stage and high-risk cytogenetics.1,2 Allogeneic stem cell transplantation (allo-SCT) may help to achieve long-term progression-free survival (PFS) and offers a potentially curative option due to a graft-versus-myeloma (GvM) effect.3 However, allo-SCT remains controversial because of considerable toxicity, especially due to immunosuppression and subsequent infections, the risk of graft-versus-host disease (GvHD), and thus a potentially high nonrelapse mortality (NRM).4 Although in general allo-SCT is not routinely conducted haematologica | 2019; 104(2)

Outcome in MM after allo-SCT Table 1. Patients’ characteristics.

All n=109 Median age [years] (range) 56 (30 - 70) Median body weight [kg] (range) 75 (40 - 103) Sex male /female n (%) 56 (51) / 53 (49) MM /plasma cell leukemia n (%) 106 (97) / 3 (3) Treatment within DSMM-trial 40 (37) / 69 (63) yes / no n (%) First-line treatment/ 46 (42) / 63 (58) salvage situation n (%) Stage according to ISS n (%) I / II 62 (57) III β2-MG ≥ 5.5mg/L 40 (37) n.e.

7 (6) Cytogenetics n (%) Summary of all subgroups 57 (52) Standard risk 20 (35*) High risk 25 (44*) Poor risk 8 (14*) Ultra high risk 4 (7*) Del13-analysis only ** 33 (30) n.e.

19 (18)

Distribution according to therapeutical concept First-line Salvage treatment situation n=46 n=63

Distribution according to time point of allo-SCT 2000 - 2004 2005 - 2009 2010 - 2016 n=19 n=37 n=53

53 (30 - 67) 72 (40 - 99) 24 (52) / 22 (48) 44 (96) / 2 (4) 34 (74) / 12 (26)

58 (42 - 70) 78 (42 - 103) 32 (51) / 31 (49) 62 (98) / 1 (2) 7 (11) / 56 (89)

55 (39 - 67) 57 (43 - 70) 71 (40 - 95) 73 (51 - 98) 9 (47) / 10 (53) 20 (54) / 17 (46) 18 (95) / 1 (5) 36 (97) / 1 (3) 5 (26) / 14 (74) 19 (51) / 18 (49)

55 (30 - 69) 78 (42 - 103) 27 (51) / 26 (49) 52 (98) / 1 (2) 16 (30) / 37 (70)

46 (100) / 0 (0)

0 (0) / 63 (100)

10 (53) / 9 (47) 17 (46) / 20 (54)

18 (34) / 35 (66)

26 (57) 20 (43) 0 (0)

36 (57) 20 (32) 7 (11)

7 (37) 12 (63) 0 (0)

26 (70) 8 (22) 3 (8)

29 (55) 20 (38) 4 (7)

20 (44) 8 (40*) 9 (45*) 1 (5*) 2 (10*) 18 (39) 8 (17)

37 (59) 12 (33*) 16 (43*) 7 (19*) | 2 (5*) 15 (24) 11 (17)

2 (11) 1 (50*) 1 (50*) 0 (0*) 0 (0*) 6 (31) 11 (58)

10 (27) 2 (20*) 6 (60*) 0 (0* 2 (20*) 24 (65) 3 (8)

45 (85) 17 (38*) 18 (40*) 8 (18*) 2 (4*) 3 (6) 5 (9)

MM: multiple myeloma; DSMM: German Myeloma Study Group; ISS: International Staging System; n.e.: not evaluated, β2-MG: beta-2 microglobulin. Cytogenetics: high risk: nonhyperdiploid karyotype, t(4;14), t(14;16), t(14;20), del 17p; poor risk: gain 1q, del 1p; ultra-high risk: ≥ 3 chromosomal aberrations. *% of patients with complete cytogenetic analysis. **Implemented into International Myeloma Working Group protocols.

in patients with MM, over the last decades the number of transplantations has increased.5 However, there are few clear treatment guidelines, as it is often performed on an individual basis in relapsed/refractory MM and not in the context of clinical studies.

Despite a high remission rate of up to 50% in retrospective analyses, early approaches in the 1980s and 1990s with high-dose myeloablative conditioning regimens were limited to younger patients with relapsed/refractory disease due to the high therapy-related toxicity, with NRM rates of 40-60%.6 NRM could be reduced through improved supportive care and a more rigorous patient selection, but long-term survival was only achieved in 1025% of patients.6 Consequently, a tandem approach was developed to separate myeloablation with maximal tumor cytoreduction achieved through high-dose chemotherapy followed by autologous (auto)-SCT, and allo-SCT with a less myelosuppressive but highly immunosuppressive regimen to reduce treatment-related organ toxicities but allow sufficient engraftment and GvM effect.7 In the era before the introduction of immunomodulatory drugs (IMID) and proteasome inhibitors (PI), several clinical trials were performed to analyze the combination of autoand subsequent reduced-intensity conditioning (RIC) alloSCT in the first-line treatment. Only two large trials with long-term follow up reached significance. In both studies, PFS and OS were superior with auto-allo-SCT compared to haematologica | 2019; 104(2)

tandem auto-SCT.8,9 The results suggested that high-risk cytogenetics may be overcome by allo-SCT. None of the trials showed inferiority of the auto-allo-SCT approach compared to single or double auto-SCT-study arms.10-14 Similarly, most retrospective studies comparing salvage allo- with a second auto-SCT in patients with relapsed MM after auto-SCT showed an improvement in PFS or a lower relapse rate after allo-SCT, but no benefit regarding OS, mostly due to high NRM.15-18 However, in one study, after a very long follow up, PFS and OS were both superior with the allo-SCT approach.19 There have been no prospective studies comparing allo- with auto-SCT in salvage situations.

Encouraging results were obtained with the combination of allo-SCT and immunomodulatory therapeutic approaches. The induction of a sustained anti-neoplastic effect by donor lymphocyte infusions (DLI) could be demonstrated in patients who had relapsed after allo-SCT,3,20 also in combination with IMID and PI.21 Several clinical trials are currently ongoing that may help to define the role of allo-SCT, particularly in the context of new immunomodulatory approaches, emphasizing the importance of this therapeutic concept. In the light of this background, here we analyzed a substantial number of MM patients who had received RIC allo-SCT at our University Medical Center between 2000 and 2016 with regard to treatment response, survival, adverse events, and quality of life (QoL).

371

C. Greil et al.

Table 2. Previous treatment and transplantation procedure.

Remission state at allo-SCT n (%) CR vgPR / PR SD / MR PD Median number of prior treatment lines n (range) Pre-treatment with PI / IMID n (%) Number of prior auto-SCT n (%) 0 1 2 Median time between auto- and allo-SCT [months] (range) Median time between ID and allo-SCT [months] (range) Interval between auto- and allo-SCT n < / > 8 months HLA compatibility n (%) HLA-identical Related / syngeneic Unrelated HLA-non-identical unrelated Donor sex male / female n (%) Stem cell source PB / BM n (%) CMV-status n (%) Donor positive /negative Patient positive / negative

13 (12) 15 (14) / 27 (25) 23 (21) / 1 (1) 30 (28) 3 (1 - 8) 54 (50) / 47 (43) 9 (8) 74 (68) 26 (24) 17.3 (1.1 - 104.2) 27.7 (4.8 - 137.4) 47 / 53 of 100 92 (84) 43 (39) / 2 (2) 47 (43) 17 (16) 71 (65) / 38 (35) 108 (99) / 1 (1) 49 (45) / 60 (55) 67 (61) / 42 (39)

CR: complete remission; (vg)PR: (very good) partial remission; SD: stable disease; MR: minimal response; PD: progressive disease; PI: proteasome inhibitor; IMID: immunomodulatory drug; auto-SCT: autologous stem cell transplantation; ID: initial diagnosis; allo-SCT: allogeneic stem cell transplantation; HLA: Human Leukocyte Antigen; PB: peripheral blood; BM: bone marow; CMV: cytomegalovirus.

Methods

mediate-fit (R-MCI 4-6), and frail patients (R-MCI 6-9) with strikingly different median OS rates of 10.1, 4.4 and 1.2 years, respectively.24

Patientsâ&#x20AC;&#x2122; description and data source We retrospectively analyzed 109 consecutive patients diagnosed with MM who had received RIC allo-SCT between 2000 and 2016 at the University Hospital of Freiburg. Data were analyzed as of January 2018. Patient data were retrieved from our institution's electronic medical records. Depending on the aggressiveness of their disease, patients received follow up on a regular basis. The analysis was carried out according to the Declaration of Helsinki and Good Clinical Practice guidelines. All patients gave their written informed consent for institutional-initiated research studies, approved by our institutional review board. Thirty-seven percent of our patients were treated in pre-emptive settings within different clinical trials of the German Myeloma Study Group (DSMM). Cytogenetic analysis was conducted in 82% of the patients. However, in 30%, FISH was only performed to detect the presence of deletion 13q14, an aberration that until recently was thought to be associated with shorter survival,22 and therefore had been implemented into the protocols of the DSMM and International Myeloma Working Group (IMWG) trials. Remission status was defined according to IMWG criteria.23 Relapse was defined as increase of serum paraprotein or occurrence of extramedullary disease. The Revised Myeloma Comorbidity Index (R-MCI) was determined as previously described.24,25 Karnofsky Performance Status (KPS), age, impairment of renal and lung function, frailty, and cytogenetic risk-group had been identified as significant determinants for OS and were combined in a weighted score, allowing identification of fit (R-MCI 0-3), inter372

Conditioning regimen and graft-versus-host disease prophylaxis Patients were treated with different fludarabine-containing RIC regimens (Online Supplementary Table S1).

In almost all patients, stem cells were harvested from the peripheral blood; only one patient received bone marrow. Fortyone percent of all transplants were conducted with a related donor, 43% with an HLA-matched unrelated donor, and 16% with an HLA-mismatched unrelated donor. Allo-SCT was performed from a male versus female donor in 65% versus 35%, respectively (Table 2). Cyclosporin A was used for GvHD prophylaxis, either in combination with alemtuzumab (37%) or mycophenolate mofetil (50%) or methotrexate (11%) with (41%) or without (59%) antithymocyte globulin (Online Supplementary Table S1).26

Statistical analysis Data were analyzed using SAS statistical software version 9.4 (SAS Institute Inc., Cary, NC, USA). OS and PFS were calculated as time from allo-SCT to death from any cause and first observation of relapse or death. NRM was defined as death without progressive disease. Patients without observation of the event of interest at the last follow up were treated as censored observations. OS and PFS rates were estimated and reported using the Kaplan-Meier method. Relapse and NRM were considered to be competing risks. Therefore, relapse and NRM rates were estimathaematologica | 2019; 104(2)

Outcome in MM after allo-SCT

A

B

C

D

Figure 1. Outcome analysis of the entire cohort. (A) Kaplan-Meier estimates for overall survival (OS). (B) Kaplan-Meier estimates for progression-free survival (PFS).

(C) Cumulative incidence of relapse rate (RR). (D) Cumulative incidence of non-relapse mortality (NRM). y: year; allo-SCT: allogeneic stem cell transplantation; CI: Confidence Interval.

ed as cumulative incidence rates using Aalen Johansen estimator27 and compared with Fine and Gray regression models for competing risks.28 P<0.05 was considered statistically significant.

Prognostic factors were investigated in bivariate regression models adjusting for the time point of allo-SCT.

Results Patients’ characteristics Median patients’ age was 56 years (range 30-70) with a balanced male:female ratio. Most patients had been diagnosed with MM; 3 had plasma cell leukemia. In 42%, alloSCT was planned as first-line treatment due to a high-risk constellation according to cytogenetic analysis, International Staging System (ISS) and/or lack of response to prior treatment,29 mostly following a prior auto-SCT and within the DSMM protocols (87%). The majority were treated in terms of individual salvage attempts due to relapsed/refractory disease after extensive pre-treatment. A complete cytogenetic analysis was performed in 52% (57 of 109); another 30% were examined for deletion 13q14 only.

Twenty-five out of those 57 patients (44%) analyzed according to the current IMWG-consensus30 showed highrisk cytogenetics with a non-hyperdiploid karyotype or detection of one of the following chromosomal aberrations: translocation (4;14), (14;16) or (14;20) or deletion 17p, respectively. Eight of 57 (14%) had poor risk cytogehaematologica | 2019; 104(2)

netics due to detection of gain 1q or deletion 1p. Four of 57 (7%) were classified in the ultra-high risk-group with three or more of those chromosomal aberrations.30 Forty of 109 (37% of the total cohort) had high-risk disease according to the ISS with a beta (β)2-microglobulin ≥5.5 mg/L (ISS III). The risk profile according to the ISS and cytogenetics of patients treated in first-line and those with relapsed/refractory disease was similar.

As the observation period was long (17 years), summing up a heterogenous cohort due to changes in therapeutical concepts over time, we additionally compared three patient subgroups to distinguish between those with allo-SCT performed between 2000-2004, 2005-2009 or 2010-1016. There was no substantial difference in patients’ characteristics between these groups; however, in the latest cohort, slightly more patients were transplanted due to relapsed/refractory disease. As expected, complete cytogenetic analyses were mainly performed in recent years. Patients’ characteristics are summarized in Table 1.

At the time point of allo-SCT, 28% of the patients showed evidence of progressive disease (PD). The majority proceeded to allo-SCT after sufficient response to prior treatment: 12% complete remissions (CR), 14% very good partial remissions (vgPR), 25% PR, 21% stable diseases (SD), and 1% minimal responses (MR). Patients underwent a median of three treatment lines prior to alloSCT (range 1-8). As we analyzed data of patients treated between 2000 and 2017, only 50% versus 43% received 373

C. Greil et al.

A

B

C

D

Figure 2. Subgroup analysis of disease activity prior to allogeneic stem cell transplantation (allo-SCT). Complete remission, (very good) partial remission, partial remission, stable disease or minimal response defined as inactive disease (n=79) versus those with progressive disease defined as active disease (n=30). (A) Kaplan-Meier estimates for overall survival (OS). (B) Kaplan-Meier estimates for progression-free survival (PFS). (C) Cumulative incidence of relapse rate (RR). (D) Cumulative incidence of non-relapse mortality (NRM); mo: months; y: year; CI: Confidence Interval.

regimens containing PI or IMID, respectively. Ninety-two percent of the cohort received prior auto-SCT, the majority of them as a single transplant; 24% had received prior auto-SCT as tandem transplant or with a second transplant in the case of relapse.31 In 47%, the period between auto- and allo-SCT was shorter than eight months; the median interval was 17.3 months (range 1.1-104.2).

Median time between initial diagnosis (ID) and allo-SCT was 27.7 months (range 4.8-137.4). Table 2 shows MMtreatment parameters before allo-SCT. Transplant data such as HLA-compatibility of the donor, stem cell source, and CMV-status are summarized in Table 2.

Graft-versus-host disease and engraftment Half of the cohort did not develop any sign of acute GvHD (aGvHD). In 25%, only mild symptoms occurred corresponding to aGvHD grade I, the remaining 25% were diagnosed with aGvHD grade II-IV, of whom only 10% had grade III or IV. Symptoms occurred at a median of 47 days after allo-SCT (range 3-150). In 58% of the patients, no symptoms of chronic GvHD (cGvHD) were detected; 24% suffered from moderate or severe cGvHD (Table 3).

Hematologic recovery with an absolute neutrophil count higher than 0.5x109/L and a platelet count higher than 20x109/L was reached at a median of 18 days (range 374

10-54) and 12 days (range 5-48), respectively (Table 3).

At the time point of analysis, 42% of the patients were still alive. NRM was relatively low (13%) (Table 3). Most patients died from PD with overlapping infection and/or GvHD; GvHD was the primary cause of death in only 2 patients.

Treatment response, survival after allo-stem cell transplantation and post-transplant therapy Overall response rate was high (70%) (Table 3). At the first response evaluation conducted on day 30 after alloSCT, 39% of the patients showed CR, 14% vgPR, and 18% PR. SD or MR was detected in 23% versus 4%, respectively, while only 3% had PD. In nearly all patients, best response to treatment had already been reached at this time point. In 4 patients, follow-up examinations revealed further improvement from vgPR to CR, leading to an overall CR rate of 42%.

Thirty-two percent received DLI after allo-SCT, nearly all due to serological PD. Only one patient was treated prophylactically because of a decreasing donor chimerism, consistent with the observation that chimerism analysis probably does not provide any further information for therapy management.32 PI and IMID were administered in the post-transplant setting in 43% and 36%, respectively, thereof only in haematologica | 2019; 104(2)

Outcome in MM after allo-SCT Table 3. Response and treatment-related toxicity.

Alive / dead (%) Primary cause of death (%), PD partly overlapping infection GvHD / GvHD only NRM (%) Death within 100 days after allo-SCT (%) Best response after allo-SCT (%) CR vgPR / PR SD / MR PD aGvHD grade (%) 0 I/ II 27 III/ IV aGvHD onset (range) [day after allo-SCT] cGvHD (%) no / mild moderate / severe n.e.

Median engraftment ANC > 0.5x109/L [day after allo-SCT] (range) PLT > 20x109/L CMV-reactivation yes / no (%)

ORR

42 (39) 15 (14) / 20 (18)

46 (42) / 63 (58) 49 (45) 19 (17) 7 (6) / 2 (2) 14 (13) 8 (7) 77 (70) 25 (23) / 4 (4) 3 (3) 54 (50) (25) / 17 (15) 9 (8) / 2 (2) 47 (3 - 150) 63 (58) / 13 (12) 18 (17) / 8 (7) 7 (6) 18 (10 - 54) 12 (5 - 48) 51 (47) / 58 (53)

allo SCT: allogeneic stem cell transplantation; PD: progressive disease; GvHD: graft-versus-host disease; NRM: non-relapse mortality; CR: complete remission; ORR: overall response rate; (vg)PR: (very good) partial remission; SD: stable disease; MR: minimal response; PD: progressive disease; aGvHD: acute GvHD; cGvHD: chronic GvHD; n.e.: not evaluated; ANC: absolute neutrophil count; PLT: platelet count; CMV: cytomegalovirus.

32% (15 of 47) and 23% (9 of 39) as maintenance therapy without evidence of relapse, mostly within different clinical DSMM trials.

With a substantial median follow up of 71.5 months, we observed a median OS of 39.2 months (95%CI: 23.4-73.7) (Figure 1A and Online Supplementary Table S2). Median PFS was 14.2 months (95%CI: 11.9-20.1) (Figure 1B and Online Supplementary Table S2). Interestingly, Kaplan-Meier curves reached a plateau after about ten years, suggesting long-term survival in selected patients with a 10-year OS of 28.4% and 10-year PFS of 24% (Figure 1B).

The median cumulative incidence of relapse within the first year was 33.6% (95%CI: 25.7-43.9) and 67.6% in ten years (95%CI: 58.6-77.9) (Figure 1C and Online Supplementary Table S3). Similar to the survival curves, there was only a slight further increase in the cumulative incidence of relapse from the second year (51.4%, 95%CI: 42.6-61.9) to ten years, again pointing towards a relapsefree long-term survival.

The median cumulative incidence of NRM within the first year was relatively low with 8.4% (95%CI: 4.5-15.7) and 12.4% in ten years (95%-CI: 7.4-20.6) (Figure 1D and Online Supplementary Table S3). Again, there was no substantial increase in this cumulative incidence from the second year (11.4%, 95%CI: 6.7-19.4) to ten years, emphasizing the low incidence of late complications.

Subgroup analyses To determine those patients who benefited particularly from allo-SCT, we conducted different subgroup analyses.

Disease activity prior to allo-SCT. First, we distinguished between patients with sufficient response to initial treatment reaching MR or better (inactive disease) and those haematologica | 2019; 104(2)

with active disease and evidence of PD at the time point of allo-SCT. We observed a statistically significantly longer OS for patients responding to previous therapies compared to those with PD with a median OS of 65.0 versus 11.5 months, respectively (P=0.003) (Figure 2A and Online Supplementary Table S2). Similarly, in patients showing MR or better, median PFS was statistically significantly prolonged with 18.4 months compared to only 5.1 months in those with active disease right before allo-SCT (P=0.001) (Figure 2B and Online Supplementary Table S2).

Similar to PFS data, the cumulative incidence of relapse differed between both groups; however, this difference did not reach statistical significance. In patients with sufficient response to prior treatment, the cumulative incidence of relapse within the first year after allo-SCT was 26.0% versus 53.3% in those with PD (P=0.087); similar results were obtained for longer observation periods (2, 5, 10 years) (Figure 2C and Online Supplementary Table S3).

No significant difference in NRM was observed between these two subgroups (13.3 vs. 6.5% in the first year; P=0.332) (Figure 2D and Online Supplementary Table S3).

Allo-SCT as first-line treatment versus in relapsed/refractory MM. In the second subgroup analysis, we distinguished between patients who were allo-transplanted in first-line due to a high-risk constellation, mostly following a prior auto-SCT, and those who received allo-SCT with relapsed/refractory disease after extensive pre-treatment.

Here, the differences between these two groups of patients were the most distinct. In patients allo-transplanted in first-line, the median OS was not reached, compared to 21.6 months in relapsed/refractory MM (P<0.001) (Figure 3A and Online Supplementary Table S2); the 5-year OS was 50.2% versus 5.4%, respectively. Similarly, the dif375

C. Greil et al.

A

B

C

D

Figure 3. Subgroup analysis of therapeutic concept for patients with allogeneic stem cell transplantation (allo-SCT) within their first-line therapy (tandem approach, n=46) versus those with relapsed/refractory (r/r) disease (n=63). (A) Kaplan-Meier estimates for overall survival (OS). (B) Kaplan-Meier estimates for progressionfree survival (PFS). (C) Cumulative incidence of relapse rate (RR). (D) Cumulative incidence of non-relapse mortality (NRM). mo: months; y: year; n.r.: not reached; CI: Confidence Interval.

ference in PFS was statistically significant with a median PFS of 47.7 months after allo-SCT in first-line versus 9.6 months in relapsed/refractory patients (P<0.001) (Figure 3B and Online Supplementary Table S2) and 5-year PFS was 49.8% versus 3.7%, respectively. The cumulative incidence of relapse was considerably lower in patients transplanted in first-line with 11.0% within the first year compared to 50.3% after allo-SCT in cases of relapsed/refractory MM (P<0.0001) (Figure 3C and Online Supplementary Table S3). Comparable to the other subgroup analyses, there was no difference in cumulative incidence of NRM (8.8 vs. 8.1% in the first year; P=0.793) (Figure 3D and Online Supplementary Table S3).

Disease activity after allo-SCT, HLA-compatibility and cytogenetic risk group. As expected, the remission status after allo-SCT had a substantial impact on survival, with a median OS of only a few weeks in the case of a lack of response, significantly shorter PFS, and an enhanced cumulative incidence of relapse. There was no difference in the cumulative incidence of NRM (data not shown).

In a subgroup analysis of HLA-compatibility, PFS and OS were impaired in patients who received allo-SCT from an HLA-non-identical donor compared to those after HLA-identical transplantation. However, the difference between the two groups did not reach statistical significance (Online Supplementary Table S2 and Online Supplementary Figure S1). Likewise, there was no statisti376

cally significant difference in the cumulative incidence of relapse and NRM (Online Supplementary Table S3 and Online Supplementary Figure S1).

When analyzing the 57 patients with complete cytogenetic analysis with respect to cytogenetic risk, a longer OS and PFS was observed in the standard risk group in comparison to patients with high risk, poor risk or ultra-high risk (Online Supplementary Table S2 and Online Supplementary Figure S2). Again, there was no difference in the cumulative incidence of relapse and NRM (Online Supplementary Table S3 and Online Supplementary Figure S2).

A subclassification according to cytogenetic risk of the 20 patients with available complete cytogenetic analysis transplanted in first-line showed no impact on survival. In the standard risk group, median OS was 77.2 months compared to 65.8 months in those patients with high risk, poor risk or ultra-high risk cytogenetics; 5-year OS was 100% versus 73.3%, respectively (P=0.256) (Online Supplementary Table S2). The median PFS was 47.7 versus 68.3 months, and 5-year PFS 43.8% versus 64.2%, respectively (P=0.874) (Online Supplementary Table S2).

Time point of allo-SCT. Over the last two decades, great improvements have been achieved in terms of anti-myeloma strategies, transplantation procedure, and supportive care. Therefore, survival may depend on the time point of allo-SCT during our long observation period of 17 years: haematologica | 2019; 104(2)

Outcome in MM after allo-SCT

Table 4. Revised-Myeloma Combordity Index (R-MCI) assessment of 46 patients alive at the time point of analysis.

Median R-MCI (range) Median KPS [%] (range) Moderate-severe frailty [number of patients] (%) Median age [years] (range) Median eGFR [ml/min/1.73 m²] (range) Moderate-severe lung function impairment [number of patients] (%)

ID

Prior allo-SCT

Current (last follow up)

Median change from ID to current P

Median change from prior allo-SCT to current P

4 (0 - 6)

3 (0 - 5)

3 (0 - 7)

80 (40 - 100)

90 (60 - 100)

90 (50 - 100)

15 (33)

9 (20)

14 (30)

0 (-5 - 5) 0.766 10 (-30 - 50) 0.008 -

0 (-3 - 5) 0.065 0 (-20 - 40) 0.411 -

51 (29 - 63)

53 (29 - 67)

60 (33 - 79)

86 (8 -121)

85 (32 - 126)

68 (16 -113)

4 (9)

0 (0)

2 (4)

8 (2 - 20) < 0.001 -11 (-77 - 66) 0.028 -

6 (1 - 20) < 0.001 -14 (-17 - 44) < 0.001 -

ID: initial diagnosis; allo-SCT: allogeneic stem cell transplantation; KPS: Karnofsky Performance Status; eGFR: estimated glomerular filtration rate.

Figure 4. Quality of life (QoL) assessed with the Revised-Myeloma Comorbidity Index (R-MCI). Single factors [Karnofsky Performance Status (KPS), frailty, age, impairment of renal and lung function] are shown at initial diagnosis, the time point right before allogeneic stem cell transplantation (allo-SCT) and at the last follow up. eGFR: estimated glomerular filtration rate; ID: initial diagnosis.

2000-2004, 2005-2009 versus 2010-2016. Indeed, the highest cumulative incidence of NRM was observed after alloSCT performed between 2000-2004 (Online Supplementary Figure S3). However, the improvement in patients transplanted more recently did not reach statistical significance, most probably due to the number of patients in the group (only 19 patients in the allo-SCT years 2000-2004).

Interestingly, the lowest relapse rate was documented in the earliest transplantation period, but this fact seems to have been overturned by treatment toxicity, as OS was improved in the cohort transplanted after the year 2010 (Online Supplementary Figure S3). Median OS was 20.1 months (95%CI: 7.5-112.5) for patients with allo-SCT between 2000-2004, 25.6 months (95%CI: 14.0-57.6) for patients transplanted between 2005-2009, and 73.7 months (95%CI: 25.2-not reached) in the most recent cohort, respectively. Again, the difference was not statistically significant (P=0.190).

Survival of patients with sufficient response to initial treatment remained significantly better than survival of haematologica | 2019; 104(2)

those with active disease when adjusting this factor for the time point of allo-SCT (P=0.004) (Online Supplementary Table S4). The same was true for the comparison of patients transplanted in first-line with those with relapsed/refractory disease (P<0.0001) (Online Supplementary Table S4).

Quality of life and comorbidity assessment Since allo-SCT may cause long-term or late onset side effects that can influence patients' QoL, and due to our prior comorbidity evaluations to assess treatment tolerance, we performed the R-MCI in the 46 patients still alive at the last follow up at three different time points: at ID, right before allo-SCT, and at last follow up.24 At ID, the median R-MCI was 4, corresponding to the intermediate-fit risk-group; this improved to 3 prior to allo-SCT, and remained at 3 at the last follow up, thus reflecting slightly fitter patients (Table 4 and Figure 4). A comparison of R-MCI assessed at ID and right before alloSCT to the current R-MCI assessed at the last follow up 377

C. Greil et al.

did not reveal any significant differences (P=0.766 and P=0.065, respectively) (Table 4), but suggested that QoL under allo-SCT improved rather than deteriorated.

The R-MCI worsened from the time point of allo-SCT to the last follow up in 48% (22 of 46) of the examined patients, whereas the score improved or was unaffected in 26% each (52%). However, a decrease was caused only by aging and age-related impaired renal function in 59% (13 of 22). Only 27% (6 of 22) of the patients with decreased R-MCI after allo-SCT showed signs of (mostly moderate) GvHD. In 14% (3 of 22), the general condition was worsening due to occurrence of an independent illness (stroke, second malignancy, LKM1-positive autoimmune hepatitis). Interestingly, only 45% (10 of 22) reached CR in this cohort, whereas 58% (7 of 12) of the group with improved R-MCI were in CR at the last follow up, indicating the influence of disease activity on the patientsâ&#x20AC;&#x2122; QoL. The cohort was too small to allow further statistical analyses.

We also assessed each of the 5 risk factors within the RMCI and their changes upon allo-SCT separately (see Table 4 and Figure 4).

Discussion Allo-SCT has been conducted in a large group of patients diagnosed with MM at our academic center, especially in relapsed/refractory situations in heavily pre-treated patients, mostly showing high-risk disease due to cytogenetic analysis and/or ISS. We observed a high ORR of 70%, with a median PFS of 14.2 months and a median OS of 39.2 months. Of note, only a moderate rate of high-grade GvHD occured. Survival was even better in patients with sufficient response to induction therapy (median OS, 65.0 months), and best in those treated within the first-line therapy (median OS not reached), independently of the time point of allo-SCT. As survival curves reached a plateau, and late relapses rarely occurred, a possible long-term survival or even cure may be presumed for a subgroup of MM patients. There was no statistically significant difference in survival when comparing patients transplanted in first-line with standard risk and those with high, poor or ultra-highrisk cytogenetics according to the actual IMWGconsensus.30 Similar findings have been obtained in previous trials mostly distinguishing patients with or without detection of deletion 13q14,22 both results suggesting that high-risk cytogenetics may be overcome by allo-SCT.8,9,32â&#x20AC;&#x201C;34 Available data from prior retrospective trials on allo-SCT in MM are inconsistent due to divergent therapeutic concepts and heterogeneous cohorts.4,35 Thus, comparison of survival and toxicity data is challenging.

For patients transplanted in first-line due to a high-risk constellation, mostly following a prior auto-SCT in terms of a tandem concept, the median OS reported in different studies ranges between 34 months and not reached; the best OS has been achieved with a long follow up, as in our analysis.9-12,36 For the cohort with relapsed/refractory dis-

References 1. Kumar SK, Dispenzieri A, Lacy MQ, et al.

Continued improvement in survival in multiple myeloma: changes in early mortality and outcomes in older patients.

378

ease, a median OS of 13-24 months was recorded in different trials,17,37 with our result of 21.6 months lying in the upper range of these values. Our PFS data were superior for patients with allo-SCT in first-line therapy with a median of 47.7 months compared to 19-35 months reported in the literature.9-12,36 The median PFS reached in our cohort of relapsed/refractory patients was also quite high with 9.6 months compared to 7-10 months in previous studies.17,37 The therapy-associated toxicity at our center, with a cumulative incidence of NRM of 13.4% in two years after allo-SCT performed as first-line treatment, was comparable to the rates of 11-16% found in published trials.10,12,36 We observed a relatively low NRM in relapsed/refractory MM with 10% in two years, contrary to prior analyses with rates of 11-43%.15â&#x20AC;&#x201C;18,38-40 Accordingly, the incidence of cGvHD was comparatively low, with detection of cGvHD of any grade in 36%, whereas cGvHD was observed in 3266% of patients in other cohorts.8,10,12,14,36,40 In order to make an objective assessment of comorbidity and therapy-associated restrictions, we analyzed the RMCI and its single risk factors over time. The median RMCI of our allo-SCT cohort was 4 at ID, but improved remarkably to 3 at the assessment right before allo-SCT and at the last follow up. Formally, this increase implicates a risk-group shift from intermediate-fit (R-MCI 4-6) to fitter patients (R-MCI 0-3),24 bearing in mind that patients had aged by almost a decade with a substantial age-related deterioration in renal function, which even under-estimated the improvement in patients' QoL. These findings suggest that QoL under allo-SCT can indeed improve or at least may not necessarily be impaired, most probably as a consequence of treatment response, as a reduction in illness-induced limitations may outweigh therapy-associated impairment.

Taken together, our data suggest that allo-SCT may enable long-term survival and a potential cure in a carefully selected subgroup of MM patients with tolerable toxicity under appropriate supportive therapy. RIC allo-SCT should preferentially be performed within clinical trials.

However, it may be considered individually in younger patients with good performance status and high-risk disease in the initial course of therapy. Such an approach has the potential of achieving a significantly better survival,35 especially in the context of novel agents and immunotherapy approaches,41 and could improve rather than impair QoL, as shown with our R-MCI subanalyses. MRD-guided immunotherapy with DLI, IMID, PI and monoclonal antibodies may significantly improve outcome even more.8,9,42,43 Our long-term retrospective single-center study may help to further evaluate the impact and redefine the role of allo-SCT in patients with MM in a rapidly changing treatment scenario. Future prospective trials should be designed with combinations of new drugs that allow profound cytoreduction before allo-SCT and that can enhance the efficacy of GvM through immunomodulatory effects after transplantation, thus leading to long-term disease control and survival even in high-risk MM patients.

Leukemia. 2014;28(5):1122-1128.

2. Fonseca R, Abouzaid S, Bonafede M, et al.

Trends in overall survival and costs of multiple myeloma, 2000-2014. Leukemia.

2017;31(9):1915-1921.

3. Tricot G, Vesole DH, Jagannath S, Hilton J,

Munshi N, Barlogie B. Graft-versus-myeloma effect: proof of principle. Blood. 1996; 87(3):1196-1198.

4. Dhakal B, Vesole DH, Hari PN. Allogeneic stem cell transplantation for multiple myeloma: is there a future? Bone Marrow

haematologica | 2019; 104(2)

Outcome in MM after allo-SCT

Transplant. 2016;51(4):492-500.

5. Sobh M, Michallet M, Gahrton G, et al.

Allogeneic hematopoietic cell transplantation for multiple myeloma in Europe: trends and outcomes over 25 years. A study by the EBMT Chronic Malignancies Working Party.

Leukemia.

2016; 30(10):2047-2054.

6. Gahrton G, Svensson H, Cavo M, et al.

Progress in allogeneic bone marrow and peripheral blood stem cell transplantation for multiple myeloma: a comparison between transplants performed 1983-93 and 1994-98 at European Group for Blood and Marrow Transplantation centres. Br J Haematol. 2001;113(1):209-216.

7. Maloney DG. Allografting with nonmyeloablative conditioning following cytoreductive autografts for the treatment of patients with multiple myeloma. Blood.

2003;102(9):3447-3454.

8. Gahrton G, Iacobelli S, Bjorkstrand B, et al.

Autologous/reduced-intensity allogeneic stem cell transplantation vs autologous transplantation in multiple myeloma: longterm results of the EBMT-NMAM2000 study. Blood. 2013;121(25):5055-5063.

9. Giaccone L, Storer B, Patriarca F, et al. Longterm follow-up of a comparison of nonmyeloablative allografting with autografting for newly diagnosed myeloma. Blood.

2011;117(24):6721-6727.

10. Garban F. Prospective comparison of autologous stem cell transplantation followed by dose-reduced allograft (IFM99-03 trial) with tandem autologous stem cell transplantation (IFM99-04 trial) in high-risk de novo multiple myeloma. Blood. 2006; 107(9):3474-3480.

11. Moreau P, Garban F, Attal M, et al. Longterm follow-up results of IFM99-03 and IFM99-04 trials comparing nonmyeloablative allotransplantation with autologous transplantation in high-risk de novo multiple myeloma. Blood. 2008; 112(9):39143915.

12. Rosinol L, Perez-Simon JA, Sureda A, et al.

A prospective PETHEMA study of tandem autologous transplantation versus autograft followed by reduced-intensity conditioning allogeneic transplantation in newly diagnosed multiple myeloma. Blood. 2008; 112(9):3591-3593.

13. Lokhorst HM, van der Holt B, Cornelissen JJ, et al. Donor versus no-donor comparison of newly diagnosed myeloma patients included in the HOVON-50 multiple myeloma study. Blood. 2012; 119(26):62196225.

14. Krishnan A, Pasquini MC, Logan B, et al.

Autologous haemopoietic stem-cell transplantation followed by allogeneic or autologous haemopoietic stem-cell transplantation in patients with multiple myeloma (BMT CTN 0102): a phase 3 biological assignment trial. Lancet Oncol. 2011; 12(13):1195-1203.

15. de Lavallade H, El-Cheikh J, Faucher C, et al. Reduced-intensity conditioning allogeneic SCT as salvage treatment for relapsed multiple myeloma. Bone Marrow Transplant. 2008;41(11):953-960.

16. Mehta J, Tricot G, Jagannath S, et al. Salvage autologous or allogeneic transplantation for multiple myeloma refractory to or relapsing after a first-line autograft? Bone Marrow Transplant. 1998;21(9):887-892.

haematologica | 2019; 104(2)

17. Qazilbash MH, Saliba R, De Lima M, et al.

Second autologous or allogeneic transplantation after the failure of first autograft in patients with multiple myeloma. Cancer.

2006;106(5):1084-1089.

18. Patriarca F, Einsele H, Spina F, et al.

Allogeneic stem cell transplantation in nultiple myeloma relapsed after Autograft: a multicenter retrospective study based on donor availability. Biol Blood Marrow Transplant. 2012;18(4):617-626.

19. Patriarca F, Bruno B, Einsele H, et al. LongTerm Follow-Up of a Donor versus NoDonor Comparison in Patients with Multiple Myeloma in First Relapse after Failing Autologous Transplantation. Biol Blood Marrow Transplant. 2018;24(2):406409.

20. Lokhorst HM. The occurrence of graft-versus-host disease is the major predictive factor for response to donor lymphocyte infusions in multiple myeloma. Blood. 2004; 103(11):4362-4364.

21. Kröger N, Badbaran A, Lioznov M, et al.

Post-transplant immunotherapy with donor-lymphocyte infusion and novel agents to upgrade partial into complete and molecular remission in allografted patients with multiple myeloma. Exp Hematol.

2009;37(7):791-798.

22. Zojer N, Königsberg R, Ackermann J, et al.

Deletion of 13q14 remains an independent adverse prognostic variable in multiple myeloma despite its frequent detection by interphase fluorescence in situ hybridization. Blood. 2000;95(6):1925-1930.

23. Rajkumar SV, Dimopoulos MA, Palumbo A, et al. International Myeloma Working Group updated criteria for the diagnosis of multiple myeloma. Lancet Oncol. 2014; 15(12):e538-e548.

24. Engelhardt M, Domm A-S, Dold SM, et al.

A concise revised Myeloma Comorbidity Index as a valid prognostic instrument in a large cohort of 801 multiple myeloma patients. Haematologica. 2017;102(5):910921.

25. 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.

26. Finke J, Bethge WA, Schmoor C, et al.

Standard graft-versus-host disease prophylaxis with or without anti-T-cell globulin in haematopoietic cell transplantation from matched unrelated donors: a randomised, open-label, multicentre phase 3 trial. Lancet Oncol. 2009;10(9):855-864.

27. Aalen OO, Johansen S. An empirical transition matrix for non-homogeneous Markov chains based on censored observations.

Scand J Stat. 1978;5(3):141-150.

28. Fine JP, Gray RJ. A Proportional Hazards Model for the Subdistribution of a Competing Risk. J Am Stat Assoc. 1999; 94(446):496-509.

29. Chng WJ, Dispenzieri A, Chim C-S, et al.

IMWG consensus on risk stratification in multiple myeloma. Leukemia. 2014; 28(2):269-277.

30. Sonneveld P, Avet-Loiseau H, Lonial S, et al.

Treatment of multiple myeloma with highrisk cytogenetics: a consensus of the International Myeloma Working Group.

Blood. 2016;127(24):2955-2962.

31. Garderet L, Iacobelli S, Koster L, et al.

Outcome of a salvage third autologous stem cell transplantation in multiple myeloma. Biol Blood Marrow Transplant.

2018;24(7):1372-1378.

32. Rasche L, Röllig C, Stuhler G, et al.

Allogeneic hematopoietic cell transplantation in multiple myeloma: focus on longitudinal assessment of donor chimerism, extramedullary disease, and high-risk cytogenetic features. Biol Blood Marrow Transplant. 2016;22(11):1988-1996.

33. Björkstrand B, Iacobelli S, Hegenbart U, et al. Tandem autologous/Reduced-Intensity Conditioning allogeneic stem-cell transplantation versus autologous transplantation in myeloma: long-term follow-up. J Clin Oncol. 2011; 29(22):3016-3022.

34. Kröger N, Badbaran A, Zabelina T, et al.

Impact of high-risk cytogenetics and achievement of molecular remission on long-term freedom from disease after autologous-allogeneic tandem transplantation in patients with multiple myeloma.

Biol Blood Marrow Transplant. 2013; 19(3):398-404.

35. Gay F, Engelhardt M, Terpos E, et al. From transplant to novel cellular therapies in multiple myeloma: European Myeloma Network guidelines and future perspectives. Haematologica. 2018;103(2):197-211.

36. Bruno B, Rotta M, Patriarca F, et al. A comparison of allografting with autografting for newly diagnosed myeloma. N Engl J Med. 2007;356(11):1110-1120.

37. Auner HW, Szydlo R, van Biezen A, et al.

Reduced intensity-conditioned allogeneic stem cell transplantation for multiple myeloma relapsing or progressing after autologous transplantation: a study by the European Group for Blood and Marrow Transplantation. Bone Marrow Transplant.

2013;48(11):1395-1400.

38. Wirk B, Byrne M, Dai Y, Moreb JS. Outcomes of salvage autologous versus allogeneic hematopoietic cell transplantation for relapsed multiple myeloma after initial autologous hematopoietic cell transplantation. J Clin Med Res. 2013; 5(3):174-184.

39. Freytes CO, Vesole DH, LeRademacher J, et al. Second transplants for multiple myeloma relapsing after a previous autotransplant-reduced-intensity allogeneic vs autologous transplantation. Bone Marrow Transplant. 2014;49(3):416-421.

40. Einsele H, Schäfer H-J, Hebart H, et al.

Follow-up of patients with progressive multiple myeloma undergoing allografts after reduced-intensity conditioning. Br J Haematol. 2003;121(3):411-418.

41. Köhler M, Greil C, Hudecek M, et al.

Current developments in immunotherapy in the treatment of multiple myeloma.

Cancer. 2018;124(10):2075-2085.

42. Giaccone L, Evangelista A, Patriarca F, et al.

Impact of new drugs on the long-term follow-up of upfront tandem autograft-allograft in multiple myeloma. Biol Blood Marrow Transplant. 2018;24(1):189-193.

43. Htut M, D’Souza A, Krishnan A, et al.

Autologous/allogeneic hematopoietic cell transplantation versus tandem autologous transplantation for multiple myeloma: comparison of long-term postrelapse survival. Biol Blood Marrow Transplant. 2018; 24(3):478-485.

379

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(2):380-391

Plasma Cell Disorders

Long-term follow up of tandem autologous-allogeneic hematopoietic cell transplantation for multiple myeloma

Enrico Maffini,1 Barry E. Storer,1,2 Brenda M. Sandmaier,1,3 Benedetto Bruno,4 Firoozeh Sahebi,5 Judith A. Shizuru,6 Thomas R. Chauncey,1,3,7 Parameswaran Hari,8 Thoralf Lange,9 Michael A.Pulsipher,10 Peter A. McSweeney,11 Leona Holmberg,1,2 Pamela S. Becker,1,3 Damian J. Green,1,3 Marco Mielcarek,1,3 David G. Maloney,1,3* and Rainer Storb1,3*

Fred Hutchinson Cancer Research Center, Clinical Research Division, Seattle, WA, USA; University of Washington School of Public Health, Seattle, WA, USA and 3Department of Medicine, Seattle, WA, USA; 4University of Turin, Department of Molecular Biotechnology and Health Sciences, Turin, Italy; 5City of Hope National Medical Center/Southern California Kaiser Permanente Medical Group, Duarte, CA, USA; 6Stanford University, CA, USA; 7VA Puget Sound Medical Health Care System, Seattle, WA, USA; 8Medical College of Wisconsin, Milwaukee, USA; 9University of Leipzig, Germany; 10Childrenâ&#x20AC;&#x2122;s Hospital of Los Angeles, CA, USA and 11Colorado Blood Cancer Institute, Denver, CO, USA 1 2

*Co-senior authors.

ABSTRACT

W

Correspondence: rstorb@fredhutch.org

Received: June 21, 2018.

Accepted: September 24, 2018.

Pre-published: September 27, 2018. doi:10.3324/haematol.2018.200253 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/380 Š2019 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.

380

e previously reported initial results in 102 multiple myeloma (MM) patients treated with sequential high-dose melphalan and autologous hematopoietic cell transplantation followed by 200 cGy total body irradiation with or without fludarabine 90 mg/m2 and allogeneic hematopoietic cell transplantation. Here we present longterm clinical outcomes among the 102 initial patients and among 142 additional patients, with a median follow up of 8.3 (range 1.0-18.1) years. Donors included human leukocyte antigen identical siblings (n=179) and HLA-matched unrelated donors (n=65). A total of 209 patients (86%) received tandem autologous-allogeneic upfront, while thirty-five patients (14%) had failed a previous autologous hematopoietic cell transplantation before the planned autologous-allogeneic transplantation. Thirty-one patients received maintenance treatment at a median of 86 days (range, 61-150) after allogeneic transplantation. Fiveyear rates of overall survival (OS) and progression-free survival (PFS) were 54% and 31%, respectively. Ten-year OS and PFS were 41% and 19%, respectively. Overall non-relapse mortality was 2% at 100 days and 14% at five years. Patients with induction-refractory disease and those with high-risk biological features experienced shorter OS and PFS. A total of 152 patients experienced disease relapse and 117 of those received salvage treatment. Eighty-three of the 117 patients achieved a clinical response, and for those, the median duration of survival after relapse was 7.8 years. Moreover, a subset of patients who became negative for minimal residual disease (MRD) by flow cytometry experienced a significantly lower relapse rate as compared with MRD-positive patients (P=0.03). Our study showed that the graft-versus-myeloma effect after non-myeloablative allografting allowed long-term disease control in standard and high-risk patient subsets. Ultra-high-risk patients did not appear to benefit from tandem autologous/allogeneic hematopoietic cell transplantation because of early disease relapse.

Incorporation of newer anti-MM agents into the initial induction treatments before tandem hematopoietic cell transplantation and during maintenance might improve outcomes of ultra-high-risk patients.

Clinical trials included in this study are registered at: clinicaltrials.gov identifiers: 00075478, 00005799, 01251575, 00078858, 00105001, 00027820, 00089011, 00003196, 00006251, 00793572, 00054353, 00014235, 00003954. haematologica | 2019; 104(2)

Tandem transplantation for multiple myeloma

Introduction Allogeneic hematopoietic cell transplantation (HCT) is a potentially curative treatment for multiple myeloma (MM) but its role is controversial. The first clinical experience with myeloablative regimens proved to be curative for a small proportion of patients but was accompanied by unacceptably high non-relapse mortality (NRM) rates.1,2 The introduction of less intensive conditioning regimens for allogeneic HCT, which relied on graft-versus-tumor (GvT) effects for tumor eradication, lowered NRM but at the expense of higher disease relapse rates.3,4 In the late 1990s, combining cytoreductive high-dose chemotherapy before autologous HCT with subsequent minimal intensity conditioning allogeneic HCT, an approach aimed at inducing GvT effects, proved to be less toxic than myeloablative allogeneic HCT and was well tolerated.5,6 Seven prospective trials compared clinical outcomes of autologous HCT versus tandem autologous/minimal intensity allogeneic HCT in newly diagnosed MM patients and yielded discordant results regarding depth of response, overall survival (OS), and progression-free survival (PFS).

Differences in conditioning regimens, as well as graft-versus-host disease (GvHD) prophylaxis, including ATG use, patient selection, definition of MM risk profiles, and duration of follow up, made meaningful comparisons between trials difficult.7-15 We previously reported initial results in 102 MM patients given tandem high-dose melphalan and autologous HCT followed by 200 cGy total body irradiation (TBI) with or without fludarabine 90mg/m2 and HLA-matched HCT from related or unrelated donors.16 Here we update the early observations and add results from 142 additional patients treated with the same approach for a total of 244 patients with a median follow up of 8.3 years (range, 1.0-18.1).

Methods Patients From August 1998 to January 2016, 244 MM patients completed sequential treatment with high-dose melphalan and autologous HCT followed by 200 cGy TBI ± fludarabine and allogeneic, granulocyte colony stimulating factor (G-CSF)-mobilized peripheral blood mononuclear cell (PBMC) infusion. One hundred and sixty-four (67%) patients were transplanted at the Fred Hutchinson Cancer Research Center (Fred Hutch), Seattle, WA, USA, and 80 (33%) patients received their transplant at eight other institutions. All patients included in the analysis were treated under eighteen clinical trials that were co-ordinated by Fred Hutch, approved by each institution’s review board, and registered at “clinicaltrials.gov”. All patients and donors signed written informed consent in accordance with the Declaration of Helsinki.

The nature of the analysis is retrospective, and we present clinical data for those 244 patients who received both autologous and allogeneic HCT. Patients' characteristics are detailed in Table 1.

Median age at diagnosis was 51 years (range, 25-67). Ninetyseven (42%) patients had high-risk cytogenetics. Fifty-seven (25%) had high-risk disease according to International Staging System (ISS) stage III and 36 (16%) according to Revised ISS (RISS) stage. Ninety-one (37%) had received more than one induction therapy line for unresponsive disease. A total of 209 patients (86%) received tandem autologous-allogeneic upfront while 35 patients (14%) had failed a previous autologous HCT before the planned autologous-allogeneic HCT. haematologica | 2019; 104(2)

Definitions and risk assessment Beta-2 (β2)-microglobulin and serum albumin values at diagnosis were available for 225 (92%) patients and were used to calculate risk according to the ISS.17 The R-ISS was introduced in 2015,18 and we calculated it retrospectively for 217 (89%) patients.

Lactate dehydrogenase (LDH) serum levels19 at diagnosis were available in 216 patients. Conventional cytogenetics and/or fluorescence in situ hybridization (FISH) studies at diagnosis and at any time before allogeneic HCT were available for 232 patients. Highrisk cytogenetics were defined as follows: t(4;14);20 t(14;16);21 t(14;20)22 by FISH; del (17/17p),23 1q21 amplifications24 both by FISH and conventional karyotyping; and non-hyperdiploid karyotype25 by conventional cytogenetics. Plasma cell leukemia included circulating plasma cells ≥ 20% of complete blood count or ≥2000 plasma cells per microliter.26 Extramedullary disease at diagnosis was defined as extramedullary plasmacytomas.27 Patients were considered high risk if they had one of the following: ISS stage III, high-risk genetic lesions, extramedullary disease presentation, plasma cell leukemia, LDH levels ≥ 2 upper normal limits or failed previous autologous HCT. Ultra-high-risk was defined as having ≥ 2 adverse factors.23,28 All patients not meeting previous criteria were considered standard risk.

HLA-typing Patients and donors were matched for HLA-A, HLA-B and HLA-C by at least intermediate resolution DNA typing and for HLA-DRB1 and DQB1 by high-resolution techniques, as previously described.29 Donors were HLA-identical siblings in 179 cases and HLA-matched unrelated in 65 cases; 11 unrelated donors were mismatched with their recipients for a single HLA allele (n=7) or antigen (n=4).

Autologous hematopoietic cell transplantation After induction treatment, patients proceeded to mobilization and collection of PBMC. Mobilization regimens included: cyclophosphamide plus dexamethasone (35% of patients), cyclophosphamide plus etoposide and dexamethasone (CED) (24%), cyclophosphamide plus paclitaxel (16%), VTD-PACE (bortezomib-thalidomide-dexamethasone-cisplatin-doxorubicincyclophosphamide-etoposide) (8%), VRD-PACE (bortezomiblenalidomide-dexamethasone-cisplatin-doxorubicin-cyclophosphamide-etoposide) (5%), carfilzomib plus RD-PACE (lenalidomide-dexamethasone-cisplatin-doxorubicin-cyclophosphamideetoposide) (1%), cyclophosphamide plus etoposide and carboplatinum (CEP) (2%), bendamustine plus etoposide and dexamethasone (BED) (1%), Hyper-CVAD (cyclophosphamide-vincristinedoxorubicine-dexamethasone-adenosine arabinoside-mesnamethotrexate) (1%), or G-CSF (10 mg/kg) alone in 7% of the patients. After PBMC collection, patients received melphalan at 200 mg/m2 intravenously (N.B. 3 patients received melphalan 140 mg/m2 because of impaired renal function) before autologous PBMC infusion, with a median of 7.8 (range, 2.1-30.4) × 106 CD34+ cells/kg actual body weight.

Allogeneic hematopoietic cell transplantation After complete recovery from autologous HCT, patients proceeded to allogeneic HCT at a median of 75 days (range, 40-281).

No further therapy was given between autologous and allogeneic HCT. The conditioning regimen for allogeneic HCT consisted of 200 cGy TBI at 7 cGy/minute from a linear accelerator (n=163) or two opposing Cobalt-60 sources (n=81). Recipients of unrelated grafts (n=65) received in addition three daily doses of fludarabine for a total of 90 mg/m2. PBMC grafts contained a median of 9.0 (range, 1.7-24.0) x 106 CD34+ cells/kg actual body weight. Postgrafting immunosuppression included mycophenolate mofetil 381

E. Maffini et al.

Table 1. Patients’ characteristics.

Characteristics Baseline characteristics at diagnosis Median age, years (range) Male: female Isotype IgG IgA Light chains only Non-secretory IgD Plasma cell leukemia Renal failure (serum creatinine > 2 mg/dL) LDH values ≥ 2 ULN ISS Stage I Stage II Stage III R-ISS Stage I Stage II Stage III Cytogenetics, high-risk ≥ 2 high-risk cytogenetic abnormalities del(17p) t (4;14) amp1q Others [hypoploidy; t (14;16); t (14;20)] Extramedullary disease (plasmacytomas) Disease risk Standard risk High risk Ultra-high risk Characteristics at autologous HCT Median time from diagnosis to autologous HCT, years (range) Failed previous autograft Induction regimens VAD-based IMiDs-based PIs-based IMiDs + PIs Other (MP; HD-Dex; Dex-Cy) Median induction lines of therapy, n (range) Characteristics at allogeneic HCT Median age, years (range) Patients > 60 years old Median time from autograft to allogeneic HCT, days (range) Patients with induction therapy-refractory disease Sibling, unrelated donor Median CD34+/kg infused, n (range) Median CD3+/kg infused, n (range)

Total (n)

%

244 51 (25–67) 143–101

59/41

151 51 30 3 3 6 39 60/216 225 72 96 57 217 38 143 36 97/232 31 28 22 9 7 50 214 62 73 79 244 0.8 (0.2–18.1) 35

62 22 12 1 1 2 18 28

125 30 15 56 18 1 (1–5) 244 53 (25–71) 50 75 (40–281) 42 179 – 65 9.00 x 106 (1.7–24.0) 3.28 x 108 (0.4–11.7)

51 12 6 24 7

32 43 25 89 18 66 16 42 32 29 23 9 7 20 88 28 35 37

14

20 17 73/27

continued on the next page

382

haematologica | 2019; 104(2)

Tandem transplantation for multiple myeloma

continued from the previous page

CMV serostatus Recipients positive Recipients negative with Donor positive Recipients and Donor negative HCT-CI score 0 1 or 2 ≥3 KPS scale < 80% Allogeneic HCT conditioning TBI 200 cGy TBI 200 cGy + fludarabine 90mg/m2 GvHD prophylaxis MMF + CSP MMF + TAC MMF + CSP/TAC + SRL Median follow up for surviving patients years(range)

212 113 30 69 206/244 59 79 68 42/218

87 53 14 33 84 29 38 33 19

164 80

67 33

176 56 12 8.3 (1.0–18.1)

72 23 5

CMV: cytomegalovirus; CSP: cyclosporine; Dex-Cy: dexamethasone + cyclophosphamide; GvHD: graft-versus-host disease; HCT: hematopoietic cell transplantation; HCT-CI: hematopoietic cell transplantation-comorbidity index; HD-Dex: high-dose dexamethasone; IMiDs: immunomodulatory drugs; KPS: Karnofsky Performance Status; LDH: lactate dehydrogenase; MMF: mycophenolate mofetil; MP: melphalan + prednisone; n: number; PIs: proteasome inhibitors; R-ISS: Revised-International Staging System; SRL: sirolimus; TAC: tacrolimus; TBI: total body irradiation; ULN: upper limit of normal; VAD: vincristine + doxorubicin + dexamethasone.

Table 2. Disease response.

Disease status CR VGPR PR RD/PD

Status at autologous HCT n= 244 % 12 35 106 91

5 14 42 39

Status at allogeneic HCT n= 244 % 62 47 93 42

26 19 38 17

Best response after allogeneic HCT n= 244 % 111 42 49 42

46 17 20 17

CR: complete remission; HCT: hematopoietic cell transplantation; n: number; PD: progressive disease; PR: partial response; RD: refractory disease; VGPR: very good partial response.

(MMF) (from a minimum of 28 days for sibling recipients to a maximum of 180 days for unrelated donors) and a calcineurin inhibitor (CNI) of either cyclosporine (n=176) or tacrolimus (n=56) for a minimum of 80 days with a subsequent taper to 180 days, as previously described.5 Twelve patients received sirolimus in addition to MMF and CNI at the dose of 2 mg orally once daily from day -3 to day +80 (n=4), day +180 (n=6), and day +365 (n=2).30 Thirty-one patients included in the analysis also received bortezomib (n=21; either at 1.6 mg/m2 intravenously or 2.6 mg/m2 subcutaneously every 14 days for up to 9 months) or lenalidomide (n=10; starting dose of 10 mg per day, range: 5-25 mg per day, on days 1-21 of each 28-day cycle, for 12 cycles of planned treatment) as maintenance treatment after allogeneic HCT, per protocol, as specified in the Results section.

Chimerism evaluation Donor chimerism was assessed at days 28, 56, 84, 180 and 365 after allogeneic HCT on peripheral blood CD3+ T lymphocytes and CD33+ myeloid cells, while unfractionated marrow was analyzed only on day +84. This involved FISH analyses in sex-mismatched pairs and polymerase chain reaction-based studies of polymorphic microsatellite regions in all other patients.31 haematologica | 2019; 104(2)

Disease response assessment Disease responses were based on the 2016 Uniform Response Criteria developed by the International Multiple Myeloma Working Group32 with some minor modifications. Complete response (CR) required negative immunofixation (IFIX) on the serum and urine, disappearance of any soft tissue plasmacytomas and/or osteolytic bone lesions, <5% plasma cells in bone marrow aspirates, and no evidence of clonal disease on flow cytometry analysis; very good partial response (VGPR) was defined as serum and urine M-protein detectable by IFIX but not on electrophoresis (SPEP) or ≥90% reduction in serum M-protein plus urine M-protein level <100 mg/24 hours (h). Partial response (PR) required ≥75% reduction of serum M-protein and reduction in 24 h urinary M-protein by ≥90% or to <200 mg/24 h and no increases in sizes or numbers of soft tissue plasmacytomas and/or lytic bone lesions; stable or progressive disease (PD) before autologous HCT was defined as chemo-refractory disease, while the achievement of at least a PR as chemotherapy-sensitivity disease. Patients were evaluated both before autologous HCT and before allogeneic HCT in order to estimate the baseline levels of disease activity before each transplantation, again on days 28, 56, 84 and 180 after allogeneic HCT, and thereafter on a clinical basis. Disease evaluation includ383

E. Maffini et al.

ed serum and urine SPEP and IFIX for M-protein detection and quantification, plasma cell quantification, cytogenetics and FISH studies in the marrow, and radiological imaging to assess for osteolytic lesions/plasmacytomas whenever appropriate. Six-color multi-parameter flow cytometry analysis of marrow cells for detection of minimal residual disease (MRD) was carried out for a subset of patients who achieved IFIX-negative CR after tandem autologous-allogeneic HCT and were treated at Fred Hutch (n=28). Samples were analyzed at the University of Washington Hematopathology Laboratory. The sensitivity of the flow cytometry assay for plasma cell neoplasms ranged from 0.01 to 0.001%.

MRD negativity (MRDNEG) status was defined as no evidence of quantifiably detectable disease.

sion, respectively. Probabilities of OS and PFS were estimated using the Kaplan-Meier method; cumulative incidences of relapse, NRM, and GvHD were estimated taking competing risks into account. Cox and Fine & Gray regression models were used to

Graft-versus-host disease evaluation Grading of acute and chronic GvHD was performed according to previously described methods.33,34 Information regarding the administration of systemic immunosuppressive treatment for GvHD was collected prospectively.

End points and statistical methods Primary objectives of this study were OS and PFS. Secondary end points included: cumulative incidences of acute GvHD, chronic GvHD, NRM, disease response, and disease relapse. We also examined response to treatment and survival among those patients who experienced disease relapse after allogeneic HCT. OS, PFS, and NRM were defined as the times from allogeneic HCT to death, death or progression, and death without progres-

Figure 1. Prevalence curve of patients alive requiring immunosuppressive treatments (IST) for chronic graft-versus-host-disease (GvHD). Black line represents the standard survival curve (OS) for the entire patient cohort (n=244). Patients alive and on IST for chronic GvHD treatment are represented by the blue curve.

The graphical difference between the two curves is the fraction of patients alive and off IST, during different time points.

Table 3. Causes of death after allogeneic hematopoietic cell transplantation.

Events Disease progression NRM Acute GvHD Gut GvHD Cerebral aspergillosis Cerebral ischemia r/to septic emboli Aspergillosis + TTP/HUS CMV disease Sepsis Chronic GvHD Respiratory syncytial virus pneumonia Sepsis Bronchiolitis obliterans Pneumonia Invasive aspergillosis Second cancers Lung cancer Esophageal cancer Pancreatic cancer Other Traumatic head injury Severe grand mal seizures ARDS and alveolar hemorrhage Congestive heart failure

Number

Time (months) after allografting

104 40 10 4 1 1 1 2 1 20 1 7 6 5 1 5 2 1 2 5 1 1 2 1

Median time: 36.6 (range: 1â&#x20AC;&#x201C;179) Median time: 10.9 (range: 2â&#x20AC;&#x201C;183) Median: 3.5 (3, 3, 3, 4) (5) (3) (6) (2, 4) (5) Median: 12 (7) (6, 8, 18, 23, 11, 58, 15) (9, 10, 14, 16, 37, 112) (7, 7, 11, 13, 142) (11) Median: 121 (4, 13) (130) (121, 182) Median: 45 (45) (64) (3, 3) (136)

ARDS: acute respiratory distress syndrome; CMV: cytomegalovirus; GvHD: graft-versus-host disease; NRM: non-relapse mortality; TTP/HUS: thrombotic thrombocytopenic purpura and hemolytic-uremic syndrome.

384

haematologica | 2019; 104(2)

Tandem transplantation for multiple myeloma

determine factors influencing HCT outcomes in univariate and multivariate analyses. Multivariate risk factors analysis included only variables significant at 0.05 level in univariate analysis for any OS, PFS, and relapse incidence. Acute GvHD, chronic GvHD, and post-transplant MM maintenance treatment (administered to 31 patients) were considered as time-dependent co-variates. Since no single variable, other than grade II-IV acute GvHD, had a significant correlation with NRM by univariate analysis, NRM was not incorporated into multivariate analysis.

Results Disease response before and after autologous hematopoietic cell transplantation Disease responses are summarized in Table 2. At the time of autologous HCT, there were no statistical significant differences among 125 patients who received vincristine â&#x20AC;&#x201C; doxorubicin â&#x20AC;&#x201C; dexamethasone (VAD)-based induction, mainly between 1998 and 2006, and those treated with immunomodulatory/proteasome inhibitor (n=101) triplet regimens (2006-2016) in terms of response rate: 74 (59%) versus 68 (68%) achieved at least a PR, respectively, and only 12 patients were in CR before autologous HCT. Of the 18 who received other induction therapies (melphalan plus prednisone, n=6; high-dose dexamethasone only, n=11; high-dose dexamethasone plus cyclophosphamide, n=1), 11 achieved PR and 7 PD. After

high-dose melphalan and autologous HCT, 62 patients (26%) were in CR, 47 (19%) VGPR, 93 (38%) PR, and 42 (17%) had PD.

Allogeneic hematopoietic cell transplantation Sixty-seven percent of the patients had allogeneic HCT within the first 3 months after autologous HCT, 29% between 3 and 6 months and 4% beyond 6 months.

Reasons for allogeneic HCT delay included: delayed hemopoietic recovery (59%), abnormal hepatic/renal function (15%), active infection requiring intravenous antibiotics (13%), persisting mucositis (4%), cytomegalovirus reactivation requiring intravenous therapy (6%), and patient choice (3%). Results after allogeneic HCT are presented with a median follow up of 8.3 (range, 1.0â&#x20AC;&#x201C;18.1) years among surviving patients.

Engraftment and GvHD - All 244 patients achieved sustained engraftment after allogeneic HCT. Median values of donor chimerism on CD3+ T cells in peripheral blood on days 28, 84, and 180 were 89%, 95%, and 100%, respectively, while median values of donor chimerism of CD33+ myeloid cells were 96%, 100%, and 100%, respectively. The cumulative incidence of grade II-IV acute GvHD was 44% with a median onset of 39 days (range, 6124), of which 33% was grade II, 7% grade III, and 4% grade IV. Sibling recipients (n=176) experienced less grade II-IV acute GvHD than unrelated (n=65) recipients (38% vs. 64%; P<0.001). Of the 228 patients who survived

A

B

C

D

Figure 2. Long-term clinical outcomes after tandem autologous-allogeneic hematopoietic cell transplantation (HCT) for the entire population and stratified for disease-risk groups. Non-relapse mortality (NRM) and relapse incidence (A). Overall survival (OS) and progression-free survival (PFS) for the entire cohort (n=244) (B).

OS (C) and PFS (D) for disease-risk stratification groups with standard-risk (n=62), high-risk (n=73), and ultra-high-risk (n=79).

haematologica | 2019; 104(2)

385

E. Maffini et al.

longer than 100 days post-transplant, 122 developed chronic GvHD requiring treatment. The cumulative incidence of extensive chronic GvHD requiring systemic immunosuppression was 46% [95% Confidence Interval (CI): 39.5-52.4] at one year and 55% (95%CI: 47.7-60.7) at five years. No differences in chronic GvHD incidence were observed between related and unrelated recipients.

Of all the surviving patients, 24% required immunosuppressive therapy for chronic GvHD at five years, 12% at ten years and 4% at 15 years, respectively (Figure 1).

Non-relapse mortality - NRM was 2% at day 100, 14% at five years and 15% at ten years after allogeneic HCT, respectively (Figure 2A). GvHD and treatment-related complications accounted for 30 of the 40 non-relapserelated deaths, while 5 patients died of secondary malignancies (Table 3) between 0.3 and 15.2 years after allografting and the remaining 5 died from other causes. Of note, no secondary hematologic cancers were observed.

By univariate analysis, only the finding of grade II-IV acute GvHD [Hazard Ratio (HR) 3.12; 95%CI: 1.6-6.2; P=0.001] and ISS stage III at diagnosis (HR: 3.92; 95%CI: 2.0-7.6; P<0.001) significantly impacted NRM. Chronic GvHD, poor performance status, and comorbidities were not associated with higher risk of NRM. Disease response and relapse - After allogeneic HCT, 111 (46%) patients achieved CR, 42 (17%) achieved VGPR, 49 achieved (20%) PR, and 42 (17%) failed to achieve a response (Table 2). Median time to best response after allografting was 6.68 months (range, 0.7-73.4). Among the 111 patients who achieved a CR as their best response, 46 remained in CR while 65 relapsed. Accordingly, the 10year relapse incidence was 66% (95%CI: 59-72) (Figure 2A). Nineteen (12%) patients had late relapses beyond five years after allogeneic HCT (range, 5.2-9.8 years).

Among patients who did not have extramedullary disease at diagnosis and who relapsed after allogeneic HCT (n=112), 28 (25%) showed extramedullary relapse (23 without evidence of marrow involvement). By multivariate analysis, ultra-high-risk patients (HR: 4.99; 95%CI: 2.9-8.7) and those with induction therapy-refractory disease before allografting (HR: 5.35; 95%CI: 3.4-8.6) had significantly higher relapse risks. The development of chronic GvHD did not protect against disease relapse (HR, 0.92; 95%CI: 0.6-1.3; P=0.66) (Table 4). Among patients with available marrow samples who achieved CR after allogeneic HCT (n=28), those with positive MRD (MRDPOS ) detected by flow cytometry (n=15) experienced a higher disease relapse rate than MRNEG patients (n=13) (HR: 10.4; 95%CI: 1.3-82.2; P=0.03).

Overall and progression-free survival - With a median follow up of 8.3 years (range, 1-18.1), median OS and PFS were 6.4 (95%CI: 3.9-9.2) years and 1.9 (95%CI: 1.4-2.6) years, respectively. Five-year OS and PFS were 54% (95%CI: 48-60) and 31% (95%CI: 25-36), respectively.

Ten-year OS was 41% (95%CI: 34-48) and PFS was 19% (95%CI:13-24) (Figure 2B). By univariate analyses, ISS and R-ISS stage III, LDH >2 upper normal limits, high-risk cytogenetics, grade II-IV acute GvHD, extramedullary disease, induction-refractory disease, and a prior failed autologous HCT were all strongly associated with inferior OS and PFS. By multivariate analysis, only high and ultra-high disease risk and induction-refractory disease remained strongly associated with worsened rates of OS and shorter PFS (Table 4). Among patients with standard-risk disease (n=62), the median OS was not reached, whereas the 386

median PFS was 6.5 (95%CI: 4.2-9.6) years. High-risk patients (n=73) experienced a median OS of 8.4 (95%CI: 3.9-10.2) years with a PFS of 2.5 (95%CI: 1.4-3.7) years, while ultra-high-risk patients (n=79) had a 2.3 (95%CI: 1.2-3.3) years median OS and 0.7 (95%CI: 0.6-0.9) year PFS (Figure 2C and D). Patients who proceeded to tandem transplantation after a previously failed autologous HCT (n=35) had poor outcomes, with a median OS of 1.2 years (95%CI: 0.6-2.0) years and a median PFS of 0.6 years (95%CI: 0.2-0.7). Similarly, patients who progressed after melphalan and autologous HCT (n=42) did poorly, having a median OS of 1.2 years (95%CI: 0.5-3.0) and median PFS of only 0.4 years (95%CI: 0.2-0.6) (Figure 3A and B); median time to relapse was 4.5 (0.1-61.9) months. Patients with extramedullary disease at diagnosis (n=50) showed median OS and PFS of 2.03 (95%CI: 1.0-3.5) and 0.95 (95%CI: 0.46-1.1) years, respectively. Patients achieving CR after allogeneic HCT displayed superior OS (median 14.9 vs. 3.2 years; HR: 3.2; 95%CI: 2.2-4.6), and PFS (median 6.9 vs.

0.9 years; HR: 4.6; 95%CI: 3.4-6.6) as compared with those who did not enter CR. Maintenance treatments - Between May 2009 and February 2016, 31 patients received post-transplant maintenance treatment with bortezomib (n=21) or lenalidomide (n=10) starting between 61 and 150 days (median

A

B

Figure 3. Kaplan-Meier estimates of overall survival (OS) and progression-free survival (PFS) stratified for disease-status after autologous hematopoietic cell transplantation (HCT). Overall survival (OS) (A) and progression-free survival (PFS) (B) among patients with progressive disease (n=42; blue line) and responders (n=202; black line).

haematologica | 2019; 104(2)

Tandem transplantation for multiple myeloma

Figure 4. Survival from disease relapse/progression after allogeneic hematopoietic cell transplantation for an overall 152 relapsed patients.

86) after allogeneic HCT. Seventeen patients completed the planned treatment (lenalidomide, n=7; bortezomib, n=10). Disease progression was the reason for early treatment discontinuation in 9 of the treated patients. One patient on lenalidomide stopped the treatment due to an acute GvHD flare which was successfully treated. Other causes included: patient choice (n=1), diarrhea not GvHDrelated (n=1), severe headache (n=1), and liver function abnormalities (n=1). By univariate analysis, maintenance therapy was not associated with any clinical outcome.

Median OS was 6.3 (95%CI: 3.6-not reached). There was no difference in terms of median PFS between those patients who received maintenance treatment (n=31) after allogeneic HCT (2.56 years; 95%CI: 0.88-4.22) and those (n=175) who achieved a response after allogeneic HCT but did not receive maintenance (2.59 years; 95%CI: 1.863.50).

Survival after disease progression - One hundred and fiftytwo patients (62%) experienced relapse or progression.

Median survival after the first relapse/progression was 2.9 years (95%CI: 1.9-3.7) for the entire cohort (n=152) (Figure 4). Twenty-eight of the 152 received palliative best supportive care and died after a median of 2.1 months.

One patient died during conditioning for a planned subsequent allogeneic HCT. Data on salvage therapy were not available for 5 patients (3%). One hundred and nineteen patients received a total of 228 lines of treatment, with a median of two lines (range, 1-9) of therapy each.

Treatments included lenalidomide (n=67), bortezomib

(n=54), thalidomide (n=36), alkylators/anthracyclines (n=27), pomalidomide (n=10), carfilzomib (n=9), daratumumab (n=1), and others (n=6). Twenty-seven (23%) of the 119 treated patients were unresponsive to salvage treatments and died after a median of 9.2 months (95%CI: 5.5-12.0), while the 83 patients who achieved at least a PR had a median survival of 7.8 (95%CI: 5.9-10.6) years.

Disease responses for those 83 patients included: 29 CR (15 of these patients are still in CR, 4 are alive with PD, and 10 have since died), 13 VGPR (3 are still in VGPR, 2 are alive with PD, and 8 have since died), and 41 PR (9 currently with stable disease, 10 alive with PD, while 22 have since died). Data on disease response after salvage treatments were not available for 9 patients. Eighteen patients received a median of 2 donor lymphocyte infusions (range, 1-4) (preceded by chemotherapy in 9 patients) after a median of 1.4 years post-allografting (range, 0.9-8.3).

Four of these achieved a partial response, while the remaining 14 did not show a response. Patients who relapsed during the first 18 months after allogeneic HCT (n=82) had a worsened prognosis (median survival after relapse/progression: 1.1 years; 95%CI: 0.6-1.9) compared to those (n=70) who relapsed beyond 18 months after allogeneic HCT (median survival after relapse/progression: 7.2 years; 95%CI: 4.3-10.6; P<0.0001).

Discussion Advances in the understanding of MM biology have led to novel treatments that have dramatically prolonged PFS and OS. First-line autologous HCT has remained standard of care for eligible patients. Three randomized trials reported significantly superior median PFS, ranging between 25 and 32 months, as compared to conventional chemotherapy.35-37 PFS was further improved up to 43-50 months after “new drugs” were employed both in the induction and consolidation/maintenance phases.38-40 Whether double autologous transplants are superior to a single autograft remains to be determined.41-43 In our series, an overall median PFS of 1.9 years may appear modest.

However, when applying retrospective risk stratification, median PFS was 6.5 years in standard-risk patients, 2.5 year in high-risk patients, and 0.7 years in ultra-high-risk patients. Extramedullary relapse without marrow relapse has been frequent after allografting.44-46 Sanctuary sites may be less accessible to graft-versus-myeloma effects than marrow. Of note, extra-medullary relapse occurred in 25% of current patients who did not have extramedullary involvement at diagnosis. Overall NRM was low (2% at

Table 4. Multivariate† risk factors in 211 patients.

Relapse High risk Ultra-high-risk Chemorefractory Age ≥ 60 years at allo HCT Unrelated donor

HR (95% CI)

P

1.74 (1.1–2.9) 4.99 (2.9–8.7) 5.35 (3.4–8.6) 1.16 (0.8–1.8) 1.54 (1.0–2.3)

0.03 <0.0001 <0.0001 0.48 0.04

Progression-free survival HR (95% CI) P

Overall survival HR (95% CI) P

1.62 (1.1–2.5) 3.47 (2.1–5.7) 4.61 (3.0–7.1) 1.22 (0.8–1.8) 1.47 (1.0–2.1)

2.48 (1.4–4.3) 3.87 (2.1–7.2) 3.28 (2.1–5.2) 1.33 (0.9–2.0) 1.32 (0.9–2.0)

0.03 <0.0001 <0.0001 0.29 0.04

0.001 <0.0001 <0.0001 0.18 0.20

HR: Hazard Ratio; CI: Confidence Interval; allo HCT: allogeneic hematopoietic cell transplantation. †Includes variables significant at 0.05 level in univariate analysis for any end point.

haematologica | 2019; 104(2)

387

E. Maffini et al.

100 days and 14% at 5 years) and, with a median follow up of 8.3 years, median OS was 6.4 years. By risk stratification, median OS was not reached in standard-risk patients, while high-risk and ultra-high-risk patients had worse outcomes with median OS of 8.4 and 2.3 years, respectively. Of note, only a minority of our patients achieved a complete remission after induction treatments, and a subset of them received tandem autologous-allogeneic HCT beyond first line. Moreover, none of our patients received recently Food and Drug Administrationapproved monoclonal antibodies, such as daratumumab and elotuzumab, which have been associated with remarkable response rates.

Although restricted to a small group of patients, we demonstrated that the achievement of MRDNEG predicted long-term CR among patients with IFIX-negative CR after autologous HCT. Whether long-term persistence of MRDNEG indicates disease eradication is unclear. Multiparameter flow cytometry and PCR-based methods are two sensitive techniques currently used to evaluate MRD in MM. Evaluation of MRD through immuno-phenotyping is more broadly available than PCR-based methods. In the present series, patients who achieved MRDNEG by flow cytometry experienced a significantly lower relapse rate as compared with MRDPOS patients (P=0.03). These findings confirm previous observations by Giaccone et al. who reported on the clinical impact of immuno-phenotypic remission after allografting in 66 MM patients.47 Conditioning was 2 Gy TBI-based in 55 of the 66 patients.

After a median follow up of 7.1 years, patients who achieved conventional CR and MRDNEG disease status had better clinical outcomes in terms of OS (median not reached) and PFS (median 59 months). Moreover, Ladetto et al. reported a PCR-based molecular analysis of MRD after minimal-intensity TBI-based conditioning in newly diagnosed patients who had not been exposed to new drugs.48 After a median follow up of 12.1 years, the median OS and PFS were not reached in patients who achieved PCR MRD negativity. Overall, MRD studies support the hypothesis that potentially curative graft-versus-myeloma effects after minimal intensity conditionings allowed longterm disease control and persistent yet non-progressive MRD in a subset of MM patients.

Whether graft-versus-myeloma effects are associated with chronic GvHD is still a subject of debate. Ringdén et al. evaluated the impact of acute and chronic GvHD on relapse and survival in 177 patients transplanted from HLA-identical siblings after non-myeloablative or reduced-intensity conditioning.49 Acute GvHD was associated with a significantly higher risk of TRM, while limited chronic GvHD significantly reduced the risk of relapse.

However, the reduced relapse risk did not translate into better OS. Crawley et al. reported that chronic GvHD was associated with better PFS and OS after reduced-intensity conditioning.3 In the present study, we report an incidence of chronic GvHD of 55%, which, as in other comparative prospective trials,7,9 was not associated with better disease control. A trend towards higher chronic GvHD rates after the introduction of minimal/reduced-intensity conditioning was shown in a Center for International Blood and Marrow Transplant Research (CIBMTR) analysis on 1207 MM recipients between 1989 and 2005. Overall, 50% of the patients who survived at least five years in the 20012005 cohort developed chronic GvHD, 65% of whom had extensive involvement.4 Discontinuation of all immuno388

suppressive agents is a surrogate for achieving immunotolerance, and is associated with improved quality of life.

Importantly, only a minority of current survivors remained on immunosuppressive drugs long-term: 24%, 12%, and 4% at five, ten, and 15 years, respectively.

Indications for allografting in MM have greatly changed over the years due to remarkable advances in the understanding of disease biology and new treatment modalities that increased median survival rates up to 8-10 years in standard-risk patients. However, relapse has remained a major issue, and poor outcomes have been observed in patients with high-risk/ultra-high-risk disease.50 Interestingly, Sobh et al. described trends and clinical outcomes of allogeneic HCT for MM in Europe between 1990 and 2012.51 The study included 7333 patients who were divided into 3 groups: 1) allogeneic HCT upfront (n=1924); 2) tandem autologous-allogeneic HCT (n=2004); and 3) allogeneic HCT as a second-line treatment or beyond (n=3405). A steady increase in numbers of allogeneic HCT over the years was observed. The use of upfront allogeneic HCT increased up to the year 2000, followed by a decline thereafter, representing 12% of allogeneic HCT performed in 2012. Tandem autologous-allogeneic HCT peaked around the year 2004 and represented 19% of allogeneic HCTs in 2012. Allogeneic HCT as salvage after at least one autograft has steadily increased over recent years and represented 69% of allogeneic HCTs in 2012. Unfortunately, only a minority of these patients were enrolled in controlled trials and remarkable heterogeneity in using allogeneic HCT was observed among different European countries.

The potential role combining “new drugs” with graft-versus-myeloma effects has not yet fully been explored. In a Phase II study, the feasibility of using bortezomib within a reduced-intensity conditioning regimen and as maintenance therapy post allograft was evaluated.52 Conditioning consisted of fludarabine, melphalan, and bortezomib, while maintenance treatment consisted of 7 cycles of bortezomib. Sixteen high-risk patients who had relapsed after an autograft were prospectively enrolled.

Nine of 16 patients (56%) achieved CR and 5 of 16 (31%) achieved PR after allogeneic HCT. In this heavily pretreated high-risk population, 3-year cumulative incidences of NRM, relapse, and OS were 25%, 54% and 41%, respectively. For the first time, this trial showed safety and efficacy of an intensified conditioning with a “new drug” in poor prognosis patients. Moreover, the concept of maintenance treatment after an allograft was also introduced. Our group recently published the results of a prospective Phase II single-center trial evaluating bortezomib as maintenance treatment after tandem autologous/allogeneic HCT for high-risk MM. At a median follow up of 51 months, a net benefit in terms of OS and PFS was shown among newly diagnosed patients over those with relapsed/persistent disease, suggesting that bortezomib maintenance may add a survival benefit among untreated patients. Treatment-related toxicity was limited, without any GvHD exacerbations.53 Different observations were reported with immunomodulatory drugs.

Somewhat compromised by an unacceptably high dropout rate, the Phase III BMT CTN 0102 trial did not show a benefit of thalidomide maintenance after tandem autologous/allogeneic HCT.13 Lenalidomide maintenance was evaluated in a study by the HOVON Group54 where the unexpectedly high toxicity profile, mainly exacerbation of haematologica | 2019; 104(2)

Tandem transplantation for multiple myeloma

acute GvHD, led to early discontinuation in 87% of the patients. In a Phase II CIBMTR trial on 30 patients,55 the use of lenalidomide was feasible if given at lower doses. A lower toxicity profile of lenalidomide maintenance was also reported by Kroger et al. in relapsed patients after an autograft and rescued with a myeloablative allograft.56 Importantly, a synergy between new drugs and graftversus-myeloma effects with a far safer toxicity profile has been described in the relapsed setting, suggesting that allogeneic-HCT and new drugs may be complementary. In our study, the median duration of survival of 7.8 years from the first relapse/progression among patients who achieved a response to salvage treatments supports this concept. These findings have been confirmed by two other recent reports.57,58 An update of an Italian study focused on the role of “new drugs” in long-term clinical outcomes.57 Median OS from first relapse was 7.5 years in the autologous/allogeneic group and two years in the tandem autologous group (P=0.01). Htut et al. compared the post-relapse OS after autologous/allogeneic HCT versus tandem autografts in patients reported to CIBMTR between 2000 and 2010.58 Six-year post-relapse OS was significantly better in the autologous/allogeneic group as compared with the tandem autografts group, 44% versus 35%, respectively (P=005). Taken together, these findings suggest a synergy between new agents and the donorderived immunological milieu.

The current role of allografting in multiple myeloma is controversial though the procedure is still used at many centers. Prospective studies were designed before agents with potent anti-myeloma activity became readily available. However, despite the recent introduction of very effective pharmacological therapies, there remains a subset of high-risk patients accounting for about 10-15% of new diagnoses, whose dismal prognosis is further compounded when early relapse (within 18 months from firstline treatment) is observed.59,60 The negative impact of adverse cytogenetics on clinical outcomes was not overcome by allogeneic HCT in our series. More aggressive plasma cell clones may have escaped graft-versus-myeloma effects after non-myeloablative 2 Gy TBI. Instead, the impact of certain high-risk cytogenetics was partly neutralized by graft-versus-myeloma effects in a study by Kröger et al.61 on 73 patients treated with autologous HCT followed by reduced-intensity melphalan 140 mg/m2 plus fludarabine, where no significant differences in PFS between patients with del(17p13) and/or t(4;14) and those without these abnormalities were observed after a median follow up of six years. A French

References 1. Gahrton G, Tura S, Ljungman P, et al.

Allogeneic bone marrow transplantation in multiple myeloma. European Group for Bone Marrow Transplantation. N Engl J Med. 1991;325(18):1267-1273.

2. Bensinger WI, Buckner CD, Anasetti C, et al. Allogeneic marrow transplantation for multiple myeloma: An analysis of risk factors on outcome. Blood. 1996;88(7):27872793.

3. Crawley C, Lalancette M, Szydlo R, et al.

haematologica | 2019; 104(2)

trial also showed no differences in clinical outcomes between t(4;14) and non-t(4;14) patients.62 Another study

by Rasche et al.63 showed no differences in survival outcomes in patients carrying del(17p), t(4;14) or amp(1q21) as compared to those with normal cytogenetics. We speculated that the incorporation of melphalan 140 mg/m2, as employed in the studies by Kroger and Rasche, in the conditioning regimen before allogeneic HCT added more cytotoxicity and might have resulted in superior tumor cell-kill and, therefore, better survival among those with high-risk cytogenetics. Moreover, almost 50% of our patients with adverse cytogenetics did not receive “new agents” as part of induction treatment that, in part, are able to overcome certain high-risk genetic features.64 In summary, our study showed that tandem autologous/minimal intensity allogeneic HCT for MM was safe and characterized by low acute and long-term toxicities.

Patients among standard and high-risk categories were able to achieve long-term sustained remissions, while patients with ultra-high-risk disease did not benefit from the tandem HCT. Similarly, patients with progressive disease after autologous HCT failed to respond to allogeneic HCT and succumbed to the disease, suggesting that graft-versus-myeloma effects alone are inadequate to control refractory and high-tumor burden disease states.

Allogeneic HCT may be employed as a platform for posttransplant immune-based strategies such as novel immunomodulatory drugs or proteasome inhibitors, CAR T- and N-cell infusions, and bispecific T-cell engagers in selected high-risk populations where prognosis remains poor even in the era of new drugs.65-67 Funding Research reported in this article was supported by the National Cancer Institute of the National Institutes of Health under award numbers CA078902, CA018029 and CA015704. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health, which had no involvement in the in study design; the collection, analysis and interpretation of data; the writing of the report; nor in the decision to submit the article for publication.

Acknowledgments We wish to thank Michelle Bouvier for study management; Joshua J. Latos, Ethan A. Melville for assistance with data retrieval and Helen Crawford for assistance with manuscript preparation and figure layout/formatting. We also thank the patients who participated on the clinical protocols and the physicians, nurses and staff who cared for them.

Outcomes for reduced-intensity allogeneic transplantation for multiple myeloma: an analysis of prognostic factors from the Chronic Leukemia Working Party of the EBMT. Blood. 2005;105(11):4532-4539.

4. Kumar S, Zhang MJ, Li P, et al. Trends in allogeneic stem cell transplantation for multiple myeloma: a CIBMTR analysis.

Blood. 2011;118(7):1979-1988.

5. Maloney DG, Molina AJ, Sahebi F, et al.

Allografting with nonmyeloablative conditioning following cytoreductive autografts for the treatment of patients with multiple myeloma. Blood. 2003;102(9):3447-3454.

6. Kroger N, Sayer HG, Schwerdtfeger R, et al. Unrelated stem cell transplantation in multiple myeloma after a reduced-intensity conditioning with pretransplantation antithymocyte globulin is highly effective with low transplantation-related mortality.

Blood. 2002;100(12):3919-3924.

7. Bjorkstrand B, Iacobelli S, Hegenbart U, et al. Tandem autologous/reduced-intensity conditioning allogeneic stem-cell transplantation versus autologous transplantation in myeloma: long-term follow-up. J Clin Oncol. 2011;29(22):3016-3022.

8. Bruno B, Rotta M, Patriarca F, et al. A com-

389

E. Maffini et al.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

390

parison of allografting with autografting for newly diagnosed myeloma. N Engl J Med. 2007;356(11):1110-1120.

Giaccone L, Storer B, Patriarca F, et al. Longterm follow-up of a comparison of nonmyeloablative allografting with autografting for newly diagnosed myeloma. Blood.

2011;117(24):6721-6727.

Garban F, Attal M, Michallet M, et al.

Prospective comparison of autologous stem cell transplantation followed by dosereduced allograft (IFM99-03 trial) with tandem autologous stem cell transplantation (IFM99-04 trial) in high-risk de novo multiple myeloma. Blood. 2006;107(9):34743480.

Moreau P, Garban F, Attal M, et al. Longterm follow-up results of IFM99-03 and IFM99-04 trials comparing nonmyeloablative allotransplantation with autologous transplantation in high-risk de novo multiple myeloma. Blood. 2008;112(9):39143915.

Rosinol L, Perez-Simon JA, Sureda A, et al.

A prospective PETHEMA study of tandem autologous transplantation versus autograft followed by reduced-intensity conditioning allogeneic transplantation in newly diagnosed multiple myeloma. Blood. 2008; 112(9):3591-3593.

Krishnan A, Pasquini MC, Logan B, et al.

Autologous haemopoietic stem-cell transplantation followed by allogeneic or autologous haemopoietic stem-cell transplantation in patients with multiple myeloma (BMT CTN 0102): a phase 3 biological assignment trial.

Lancet Oncol.

2011;12(13):1195-1203.

Lokhorst HM, van der Holt B, Cornelissen JJ, et al. Donor versus no-donor comparison of newly diagnosed myeloma patients included in the HOVON-50 multiple myeloma study. Blood. 2012;119(26):62196225.

Gahrton G, Iacobelli S, Bjorkstrand B, et al.

Autologous/reduced-intensity allogeneic stem cell transplantation vs autologous transplantation in multiple myeloma: longterm results of the EBMT-NMAM2000 study. Blood. 2013;121(25):5055-5063.

Rotta M, Storer BE, Sahebi F, et al. Longterm outcome of patients with multiple myeloma after autologous hematopoietic cell transplantation and nonmyeloablative allografting. Blood. 2009;113(14):33833391.

Greipp PR, San Miguel J, Durie BG, et al.

International staging system for multiple myeloma. J Clin Oncol. 2005;23(15):34123420.

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.

Patel K, Orlowski RZ, Shah N, et al. Role of serum lactate dehydrogenase (LDH) as a prognostic marker for autologous hematopoietic stem cell transplantation for multiple myeloma (abstract). Blood.

2012;120(21):3115-3115.

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.

Neben K, Jauch A, Bertsch U, et al.

Combining information regarding chromosomal aberrations t(4;14) and del(17p13) with the International Staging System classification allows stratification of myeloma patients undergoing autologous stem cell

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

transplantation. Haematologica. 2010; 95(7):1150-1157.

Teoh PJ, Chung TH, Sebastian S, et al. p53 haploinsufficiency and functional abnormalities in multiple myeloma. Leukemia.

2014;28(10):2066-2074.

Boyd KD, Ross FM, Chiecchio L, et al. A novel prognostic model in myeloma based on co-segregating adverse FISH lesions and the ISS: analysis of patients treated in the MRC Myeloma IX trial. Leukemia. 2012; 26(2):349-355.

Hanamura I, Stewart JP, Huang Y, et al.

Frequent gain of chromosome band 1q21 in plasma-cell dyscrasias detected by fluorescence in situ hybridization: incidence increases from MGUS to relapsed myeloma and is related to prognosis and disease progression following tandem stem-cell transplantation. Blood. 2006;108(5):17241732.

Van Wier S, Braggio E, Baker A, et al.

Hypodiploid multiple myeloma is characterized by more aggressive molecular markers than non-hyperdiploid multiple myeloma.

Haematologica.

2013; 98(10):1586-1592.

Nowakowski GS, Witzig TE, Dingli D, et al. Circulating plasma cells detected by flow cytometry as a predictor of survival in 302 patients with newly diagnosed multiple myeloma. Blood. 2005;106(7):22762279.

Usmani SZ, Heuck C, Mitchell A, et al.

Extramedullary disease portends poor prognosis in multiple myeloma and is overrepresented in high-risk disease even in the era of novel agents. Haematologica.

2012;97(11):1761-1767.

Sonneveld P, Avet-Loiseau H, Lonial S, et al.

Treatment of multiple myeloma with highrisk cytogenetics: a consensus of the International Myeloma Working Group.

Blood. 2016;127(24):2955-2962.

Petersdorf EW, Gooley TA, Anasetti C, et al. Optimizing outcome after unrelated marrow transplantation by comprehensive matching of HLA class I and II alleles in the donor and recipient. Blood. 1998;92(10): 3515-3520.

Sandmaier BM, Maloney DG, Storer BE, et al. Sirolimus combined with mycophenolate mofetil (MMF) and cyclosporine (CSP) significantly improves prevention of acute graft-versus-host-disease (GVHD) after unrelated hematopoietic cell transplantation (HCT): Results from a phase III randomized multi-center trial. Blood. 2016; 128(22):#506 (abstract); http://www.bloodjournal.org/content/128/522/506.

Martin PJ. Documentation of engraftment and characterization of chimerism after hematopoietic cell transplantation. In: Forman SJ, Negrin RS, Antin JH, Appelbaum FR, eds.

Thomas' Hematopoietic Cell Transplantation, 5th Edition. Chichester, UK: John Wiley & Sons, Ltd, 2016:272-280.

Kumar S, Paiva B, Anderson KC, et al.

International Myeloma Working Group consensus criteria for response and minimal residual disease assessment in multiple myeloma. Lancet Oncol. 2016;17(8):e328e346.

Glucksberg H, Storb R, Fefer A, et al.

Clinical manifestations of graft-versus-host disease in human recipients of marrow from HL-A-matched sibling donors.

Transplantation. 1974;18(4):295-304.

Shulman HM, Sullivan KM, Weiden PL, et al. Chronic graft-versus-host syndrome in man. A long-term clinicopathologic study

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

of 20 Seattle patients. Am J Med. 1980;69: 204-217.

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.

Fermand JP, Katsahian S, Divine M, et al.

High-dose therapy and autologous blood stem-cell transplantation compared with conventional treatment in myeloma patients aged 55 to 65 years: long-term results of a randomized control trial from the Group Myelome-Autogreffe. J Clin Oncol. 2005;23(36):9227-9233.

Child JA, Morgan GJ, Davies FE, et al.

High-dose chemotherapy with hematopoietic stem-cell rescue for multiple myeloma.

N Engl J Med. 2003;348(19):1875-1883.

McCarthy PL, Owzar K, Hofmeister CC, et al. Lenalidomide after stem-cell transplantation for multiple myeloma. N Engl J Med. 2012;366(19):1770-1781.

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.

Attal M, Lauwers-Cances V, Hulin C, et al.

Lenalidomide, bortezomib, and dexamethasone with transplantation for myeloma. N Engl J Med. 2017;376(14):1311-1320.

Cavo M, Tosi P, Zamagni E, et al.

Prospective, randomized study of single compared with double autologous stemcell transplantation for multiple myeloma: Bologna 96 clinical study. J Clin Oncol.

2007;25(17):2434-2441.

Sonneveld P, van der Holt B, Segeren CM, et al. Intermediate-dose melphalan compared with myeloablative treatment in multiple myeloma: long-term follow-up of the Dutch Cooperative Group HOVON 24 trial. Haematologica.

2007;92(7): 928-935.

Mai EK, Benner A, Bertsch U, et al. Single versus tandem high-dose melphalan followed by autologous blood stem cell transplantation in multiple myeloma: long-term results from the phase III GMMG-HD2 trial. Br J Haematol. 2016;173(5):731-741.

Perez-Simon JA, Sureda A, FernandezAviles F, et al. Reduced-intensity conditioning allogeneic transplantation is associated with a high incidence of extramedullary relapses in multiple myeloma patients.

Leukemia. 2006;20(3):542-545.

Byrne JL, Fairbairn J, Davy B, Carter IG, Bessell EM, Russell NH. Allogeneic transplantation for multiple myeloma: late relapse may occur as localised lytic lesion/plasmacytoma despite ongoing molecular remission. Bone Marrow Transplant. 2003;31(3):157-161.

Vincent L, Ceballos P, Plassot C, et al.

Factors influencing extramedullary relapse after allogeneic transplantation for multiple myeloma. Blood Cancer J. 2015;5:e341.

Giaccone L, Brunello L, Festuccia M, et al.

Clinical impact of immunophenotypic remission after allogeneic hematopoietic cell transplantation in multiple myeloma.

Bone Marrow Transplant. 2015;50(4):511516.

Ladetto M, Ferrero S, Drandi D, et al.

Prospective molecular monitoring of minimal residual disease after non-myeloablative allografting in newly diagnosed multiple myeloma. Leukemia. 2016;30(5):12111214.

RingdĂŠn O, Shrestha S, da Silva GT, et al.

Effect of acute and chronic GVHD on

haematologica | 2019; 104(2)

Tandem transplantation for multiple myeloma

50.

51.

52.

53.

54.

55.

relapse and survival after reduced-intensity conditioning allogeneic transplantation for myeloma. Bone Marrow Transplant.

2012;47(6):831-837.

Lonial S, Boise LH, Kaufman J. How I treat high-risk myeloma. Blood. 2015;126(13): 1536-1543.

Sobh M, Michallet M, Gahrton G, et al.

Allogeneic hematopoietic cell transplantation for multiple myeloma in Europe: trends and outcomes over 25 years. A study by the EBMT Chronic Malignancies Working Party.

Leukemia.

2016;30(10):2047-2054.

Caballero-Velazquez T, Lopez-Corral L, Encinas C, et al. Phase II clinical trial for the evaluation of bortezomib within the reduced intensity conditioning regimen (RIC) and post-allogeneic transplantation for high-risk myeloma patients. Br J Haematol. 2013;162(4):474-482.

Green DJ, Maloney DG, Storer BE, et al.

Tandem autologous/allogeneic hematopoietic cell transplantation with bortezomib maintenance therapy for high-risk myeloma. Blood Adv. 2017;1(24):2247-2256.

Kneppers E, van der HB, Kersten MJ, et al.

Lenalidomide maintenance after nonmyeloablative allogeneic stem cell transplantation in multiple myeloma is not feasible: results of the HOVON 76 Trial. Blood.

2011;118(9):2413-2419.

Alsina M, Becker PS, Zhong X, et al.

Lenalidomide maintenance for high-risk multiple myeloma after allogeneic hematopoietic cell transplantation. Biol

haematologica | 2019; 104(2)

56.

57.

58.

59.

60.

61.

Blood Marrow Transplant. 2014;20(8): 1183-1189.

Kroger N, Zabelina T, Klyuchnikov E, et al.

Toxicity-reduced, myeloablative allograft followed by lenalidomide maintenance as salvage therapy for refractory/relapsed myeloma patients.

Bone Marrow Transplant. 2013;48(3):403-407.

Giaccone L, Evangelista A, Patriarca F, et al.

Impact of new drugs on the long-term follow-up of upfront tandem autograft-allograft in multiple myeloma. Biol Blood Marrow Transplant. 2018;24(1):189-193.

Htut M, D'Souza A, Krishnan A, et al.

Autologous/allogeneic hematopoietic cell transplantation versus tandem autologous transplantation for multiple myeloma: Comparison of long-term postrelapse survival. Biol Blood Marrow Transplant.

2018;24(3):478-485.

Walker BA, Boyle EM, Wardell CP, et al.

Mutational spectrum, copy number changes, and outcome: Results of a sequencing study of patients with newly diagnosed myeloma. J Clin Oncol.

2015;33(33):3911-3920.

Thanendrarajan S, Tian E, Qu P, et al. The level of deletion 17p and bi-allelic inactivation of TP53 has a significant impact on clinical outcome in multiple myeloma.

Haematologica. 2017;102(9):e364-e367.

Kroger N, Badbaran A, Zabelina T, et al.

Impact of high-risk cytogenetics and achievement of molecular remission on long-term freedom from disease after autologous-allogeneic tandem transplanta-

62.

63.

64.

65.

66.

67.

tion in patients with multiple myeloma.

Biol Blood Marrow Transplant. 2013;19(3): 398-404.

Roos-Weil D, Moreau P, Avet-Loiseau H, et al. Impact of genetic abnormalities after allogeneic stem cell transplantation in multiple myeloma: a report of the Societe Francaise de Greffe de Moelle et de Therapie Cellulaire. Haematologica.

2011;96(10):1504-1511.

Rasche L, Rollig C, Stuhler G, et al.

Allogeneic hematopoietic cell transplantation in multiple myeloma: Focus on longitudinal assessment of donor chimerism, extramedullary disease, and high-risk cytogenetic features. Biol Blood Marrow Transplant. 2016;22(11):1988-1996.

Avet-Loiseau H, Leleu X, Roussel M, et al.

Bortezomib plus dexamethasone induction improves outcome of patients with t(4;14) myeloma but not outcome of patients with del(17p). J Clin Oncol. 2010;28(30):46304634.

Gay F, Engelhardt M, Terpos E, et al. From transplant to novel cellular therapies in multiple myeloma: European Myeloma Network guidelines and future perspectives. Haematologica. 2018;103(2):197-211.

Garfall AL, Maus MV, Hwang WT, et al.

Chimeric antigen receptor T cells against CD19 for multiple myeloma. N Engl J Med. 2015;373(11):1040-1047.

Mikkilineni L, Kochenderfer JN. Chimeric antigen receptor T-cell therapies for multiple myeloma. Blood. 2017;130(24):25942602.

391

ARTICLE Ferrata Storti Foundation

Stem Cell Transplantation

The TLR7 ligand R848 prevents mouse graft-versus-host disease and cooperates with anti-interleukin-27 antibody for maximal protection and regulatory T-cell upregulation

Mélanie Gaignage,1 Reece G. Marillier,1 Perrine M. Cochez,1 Laure Dumoutier,1 Catherine Uyttenhove,1,2 Jean-Paul Coutelier1 and Jacques Van Snick1,2

Haematologica 2018 Volume 104(2):392-402

de Duve Institute, Université Catholique de Louvain and 2Ludwig Cancer Research, Brussels, Belgium

1

ABSTRACT

I

Correspondence: jacques.vansnick@bru.licr.org

Received: April 12, 2018.

Accepted: September 7, 2018.

Pre-published: September 13, 2018. doi:10.3324/haematol.2018.195628 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/392 ©2019 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.

392

n spite of considerable therapeutic progress, acute graft-versus-host disease still limits allogeneic hematopoietic cell transplantation. We recently reported that mouse infection with nidovirus lactate dehydrogenase elevating virus impairs disease in non-conditioned B6D2F1 recipients of parental B6 spleen cells. As this virus activates TLR7, we tested a pharmacological TLR7 ligand, R848, in this model and observed complete survival if donor and recipients were treated before transplantation. Mixed lymphocyte culture performed 48 h after R848-treatment of normal mice demonstrated that both T-cell allo-responsiveness and antigen presentation by CD11b+ and CD8α+ dendritic cells were inhibited. These inhibitions were dependent on IFNAR-1 signaling. In the B6 to B6D2F1 transplantation model, R848 decelerated, but did not abrogate, donor T-cell implantation and activation. However, it decreased interferon-gamma, tumor necrosis factor-alpha and interleukin-27 while upregulating active transforming growth factor-beta 1 plasma levels. In addition, donor and recipient Foxp3+ regulatory T-cell numbers were increased in recipient mice and their elimination compromised disease prevention. R848 also strongly improved survival of lethally irradiated BALB/c recipients of B6 hematopoietic cells and this also correlated with an upregulation of CD4 and CD8 Foxp3+ regulatory T cells that could be further increased by inhibition of interleukin-27. The combination of anti-interleukin-27p28 monoclonal antibody and R848 showed strong synergy in preventing disease in the B6 to B6D2F1 transplantation model when recipients were sublethally irradiated and this also correlated with upregulation of regulatory T cells.

We conclude that R848 modulates multiple aspects of graft-versus-host disease and offers potential for safe allogeneic bone marrow transplantation that can be further optimized by inhibition of interleukin-27.

Introduction Allogeneic hematopoietic cell transplantation (HCT) is an important therapeutic option for a wide range of malignant and non-malignant disorders, including acquired and genetic anomalies.1-3 However, it remains plagued by bone marrow failure or graft-versus-host disease (GvHD), which develop in approximately 50% of allogeneic HCT recipients.4 With a death rate in the range of 15%,5 acute GvHD thus remains a major medical challenge, especially in older patients and in steroid-resistant individuals in whom mortality reaches 90%.6 A hallmark of acute GvHD is the release of inflammatory cytokines and a pathogenic contribution has been demonstrated for tumor necrosis factor-alpha (TNFα),7 interleukin(IL)-6,8 IL-239 and more recently IL-27.10,11 This last, a member of the IL-6 and IL12 families, is a cytokine with multiple activities12 that both promotes Th1 type immune reactions13 and restrains excessive inflammation.14 IL-27 is formed by noncovalently linked p28 (also called IL-30)15 and Epstein-Barr virus-induced gene 3 product (EBI3), which are structurally related to IL-12p35 and IL-12p40, respectively.13 IL-27 haematologica | 2019; 104(2)

A TLR7 ligand prevents mouse GvHD

binds to a heterodimeric receptor, comprising WSX-1/IL27Rα and gp130. This receptor signals through STAT1 and STAT3 and is expressed on hematopoietic stem cells, all lymphoid and myeloid cells, vascular endothelium and keratinocytes.16 IL-27 is produced by macrophages, monocytes and dendritic cells (DC) and its receptor is expressed on memory CD4 and CD8 T cells.13,17,18 In the context of GvHD, IL-27 inhibition confers protection by downregulating interferon-gamma (IFNγ) and upregulating Foxp3-expressing regulatory T cells (Treg) that represent a major control point of GvHD severity, as reported previously.19-21 We recently made the surprising observation that lactate dehydrogenase elevating virus (LDV), a single-stranded positive-sense RNA enveloped mouse nidovirus, enhanced survival in the lethal form of acute GvHD induced by B6 spleen cell injection into non-conditioned B6D2F1 recipients.22 In this model, which is mainly characterized by severe bone marrow failure23 and liver destruction, LDV protection correlated with a transient impairment of antigen presentation by DC and with alterations in T-cell allo-responsiveness, both dependent on IFNAR-1 signaling, in agreement with the reported GvHD inhibition by type I interferons.24 As LDV activates Toll-like receptor 7 (TLR7),25 we considered the possibility that the TLR7 agonist resiquimod (R848),26 an imidazoquinoline originally developed as an antiviral agent27 and an immune adjuvant,28 could replicate the protective effects of the virus. This idea is somewhat counterintuitive since R848 has adjuvant activity and promotes Th1 immune reactions.29 However, prolonged exposure of mice to the compound was reported to induce a disruption of the lymphoid system resembling that induced by human immunodeficiency virus.30 We here report that a transient treatment of mice with R848 just before allogeneic HCT conferred strong protection against GvHD which correlated with inhibition of inflammatory cytokine production, upregulation of transforming growth factor-beta (TGF-β) and Treg and could be further enhanced by anti-IL-27 monoclonal antibody (antiIL-27) therapy.

Methods Mice Mice were bred under specific pathogen-free conditions at the animal facility of the Ludwig Cancer Research Brussels Branch under the direction of Guy Warnier (DVM). Experimental protocols and animal handling were approved by the ethical committee of the Medical Faculty of the Université de Louvain (accreditation n.: 2016/UCL/MD/010). IFN-α/βR-/- 129/Sv mice (IFNAR-1-/-) were a gift from Dr. M. Aguet.31

Reagents Donor and/or recipient mice were injected intraperitoneally 48 h and 24 h or 0 h before donor cell implantation with 25 mg/mouse resiquimod (R848) (Cat. ALX-420-038-M005, EnzoLifeSciences, NY, USA). Anti-IL-27p28 monoclonal antibody (aIL-27) (clone: MM27.7B1) is a previously described mouse IgG2a antibody.32 Recipient animals were given 0.5 mg intraperitoneally on days 0 and 6 after transplantation. Anti-CD25 antibody (clone: PC61) was generated in-house from hybridomas obtained from the American Type Culture Collection (Manassas, VA, USA).33 Donor and recipient mice were treated with two intraperitoneal injections of antiCD25 antibody (400 mg/mouse) 6 days before and 1 day after donor cell transfer. haematologica | 2019; 104(2)

Other detailed methods All other methods are described in the Online Supplementary Methods.

Results R848 prevents lethal parent to F1 non-conditioned graftversus-host disease The effect of R848 was first tested in non-conditioned (nc) GvHD to avoid the inflammatory cytokine burst induced by host irradiation. The agent was administered to recipient B6D2F1 and/or donor B6 mice 24 and 48 h before donor spleen cell injection. Data pooled from five experiments showed 100% mortality in control mice by day 25.

R848 treatment of either recipient or donor resulted in survival rates of 40% and 60%, respectively, while combined treatment of both resulted in 100% long-term survival (Figure 1A). Morbidity evaluated by weight loss was also totally suppressed after donor and recipient treatments (Figure 1B). Timing of R848 administration was important since treatments given too early (6-5 days before transplantation) or too late (5-6 days after transplantation) were not protective (Figure 1C). R848 treatment also minimized hepatocyte destruction but mononuclear cell infiltration was still present (Figure 1D). This tissue protection was confirmed by suppression of serum amyloid A (SAA)1/2 and SAA3 mRNA expression in liver cells (Figure 1E).

Finally, after 14 days, host spleen cell numbers had dropped from ±50x106 to ±2x106 in the control GvHD group but remained unchanged or even slightly increased after R848 treatment, demonstrating host spleen cell protection by R848. Spleen cell implantation by R848-treated B6 donors was low after 14 days, approximately 4x106 cells, but increased with time, resulting in a permanent chimerism reaching 20x106 B6 cells per spleen after 100 days (Figure 1F).

R848 treatment inhibits production of IFNγ, TNFα and IL-27 but stimulates active TGF-β1 secretion in non-conditioned graft-versus-host disease B6→ncB6D2F1 GvHD was characterized by high concentrations of plasma IFNγ and IL-27 that reached maxima at day 10 and returned to nearly undetectable levels by day 14. R848 treatment essentially abolished these cytokine peaks (Figure 2A). RNA analysis confirmed upregulation of Ifng expression in spleen and liver of untreated transplanted mice and its complete silencing by R848 (Figure 2B). A similar inhibition was seen for TNFα (Figure 2A), a cytokine that also contributes to GvHD pathology.34 Given the potential implication of TGF-β in the control of GvHD,35,36 we also measured TGF-β1 and TGF-β3 by enzyme-linked immunosorbent assays selectively detecting the active forms of these cytokines and observed a strong upregulation of the former (Figure 2A), but not the latter (data not shown) after R848 treatment. As shown in Figure 2A, active TGF-β1 was upregulated from day 6 to day 14, but was no longer detectable at day 50 (data not shown).

R848 treatment before hematopoietic cell transplantation inhibits T-cell allo-responsiveness and major histocompatibility complex presentation by conventional dendritic cells through type I interferon signaling Optimal prevention of B6→ncB6D2F1 GvHD required treatment of both donor and recipient with R848. Since 393

M. Gaignage et al.

A

B

C D

E

F

Figure 1. R848 treatment of recipient B6D2F1 and donor B6 mice before spleen cell transfer completely blocks non-conditioned graft-versus-host disease.

B6D2F1 mice, treated with R848 or not (25 mg at 48 and 24 h before transplantation), were injected intraperitoneally with 60x106 spleen cells from B6 mice nontreated (NT) or treated with R848, 48 h before cell transfer. Mice were monitored for (A) mortality and (B) weight loss. (C) R848 treatment kinetics: B6D2F1 and B6 mice were treated with R848 (25 mg) twice (6 and 5 days before; 2 and 1 days before and 5 and 6 days after B6 cells transfer) and recipients were monitored for mortality. (D) Liver sections were prepared from NT B6D2F1, control or R848 14 day-GvHD mice. Hematoxylin & eosin–stained slides were analyzed for histopathological damage and representative sections illustrate mononuclear inflammatory cells present in portal tracts (green arrow) and large-dilated sinusoidal spaces (red arrow) are indicated. Scale bars in the upper and lower panels represent 200 and 50 μm, respectively. (E) RNA was extracted from total liver and analyzed by realtime quantitative polymerase chain reaction for SAA1/2 and SAA3 expression. (F) After 14 days of GvHD induction, spleen cells from NT B6 → NT B6D2F1 (control) and R848 B6 → R848 B6D2F1 (R848) GvHD mice were recovered and stained with anti-H-2Dd and H-2Db antibodies to assess the presence of host and donor cells by flow cytometry. On the left, representative mouse dot plots illustrating percentages (gated on total living cells) and on the right, absolute numbers of host (H-2Db) and donor (H-2Dd/b) cells are shown. Overall survivals are depicted (***P<0.001 by the log-rank test). Data are from three to five experiments (*P<0.05, ***P<0.001 by the Kruskal-Wallis test with Dunn multiple comparison test or ANOVA–Bonferroni post-test).

host T cells are not reactive against the implanted parental cells, the action of R848 on the F1 partner is likely on antigen presentation. To assess the influence of R848 on this process and on T-cell responsiveness, normal B6D2F1 and B6 mice were injected with R848 or phosphate-buffered saline for two consecutive days before spleen cell collection.

B6 spleen cells were then incubated with B6D2F1 irradiated adherent spleen cells as a source of antigen-presenting cells (APC). In control mice, this combination induced strong proliferation and IFNγ production. R848 treatment of either B6D2F1 stimulating or B6 responder mice inhibited proliferation and IFNγ production (Figure 3A), indicating that 48 h of treatment with R848 of otherwise non-manipulated mice impaired both antigen presenting and responder cells.

TLR7 activation leads to production of type I interferons.26 We, therefore, tested the implication of IFN in R848mediated T-cell suppression in allogeneic mixed lymphocyte cultures using spleen cells from 129/Sv (H-2Db) or 394

129/Sv-IFNAR-1-/- mice as responder cells and B6D2F1 irradiated adherent spleen cells as APC. This experiment confirmed the inhibitory effect of R848 on the allogeneic responder cells in a different mouse strain and showed that, in 129/Sv-IFNAR-1-/- responder cells, proliferation and IFNγ production were not inhibited (Figure 3B). R848-induced inhibition was not correlated with an increase in Foxp3+ Treg (Figure 3C) and depletion of Treg with anti-CD25 PC61 antibody before R848 treatment did not prevent the inhibition of IFNγ production during the allogeneic mixed lymphocyte culture (Figure 3D).

We previously showed that TLR7 viral stimulation transiently inhibits conventional DC (cDC), the only splenic cells able to activate an allogeneic T-cell response in vitro.22 To evaluate the effect of R848 on allogeneic antigen presentation in the allogeneic mixed lymphocyte culture, CD11b+ cDC, CD8a+ cDC and pDC were purified from 129/Sv mice 48 h after R848 treatment and co-cultured with FVB (H-2q) haematologica | 2019; 104(2)

A TLR7 ligand prevents mouse GvHD Figure 2. R848 treatment lowers inflammatory cytokines but increases active transforming growth factor beta-1 in nonconditioned graft-versus-host disease.

GvHD was induced by the transfer of 60x106 B6 spleen cells into B6D2F1 recipients. One group was left non-treated (Control GvHD) and in one group, both recipients and donors were treated with R848 48 and 24 h before transplant (R848 GvHD). (A) IFNγ, IL-27p28, active TGFβ-1 and TNFα were measured in plasma by enzyme-linked immunosorbent assay. (B) Spleen and liver cells were analyzed by real-time quantitative polymerase chain reaction on days 6, 10 and 14. IFNγ gene expression, normalized against β-actin, is shown. RNA was extracted from total spleen and liver. Data are pooled from two independent experiments and representative of three (*P<0.05, **P<0.01, ***P< 0.001 by the Kruskal-Wallis test with Dunn multiple comparison test).

A

B

CD4 T cells. CD11b+ and CD8α+ cDC were the main stimulating cells in the mixed lymphocyte culture and this capacity was impaired in mice treated with R848 (Figure 3E). This suppression of APC was also dependent on IFNAR-1 (Figure 3F).

These data indicate that at the time of donor cell transfer, R848 administration inhibited both antigen stimulation by cDC and T-cell responsiveness in a type 1 interferon-dependent process. This impairment of T-cell allo-responsiveness induced by R848 after 48 h did not involve Treg.

R848 treatment impairs effector donor T cells and increases regulatory T cells in a non-conditioned graft-versus-host disease model To evaluate the effect of R848 on responder T cells during GvHD, we first used CFSE-labeled CD5 B6 cells. Six days after transplantation, staining dropped considerably in B6 cells transplanted into B6D2F1 but not into B6 recipients.

This loss of CFSE staining, although diminished, still occurred in R848-treated mice with GvHD, indicating that donor cell division was not completely suppressed by R848 (Figure 4A). The slower expansion of donor T cells after R848 treatment was confirmed by the decrease in implanted donor CD4 and CD8 T cells 14 days after transplantation (Figure 4B). However, ultimately R848 did not abrogate B6 donor cell engraftment as chimerism was still detected up to 100 days after transplantation (see Figure 1G). Donor haematologica | 2019; 104(2)

CD8 T cells were also five times less abundant in R848-protected GvHD than in control ncGvHD. These effects were not the consequence of R848 toxicity for T cells since host CD4 T cells remained unchanged (Figure 4B).

The persistence of donor T cells in the R848-protected mice raised the question of the activation state of these cells. We evaluated the proportion of naïve and memory cells with CD62L and CD44 labeling, respectively. The massive loss of CD62L expression by CD4 T cells observed in control ncGvHD was partially inhibited by R848 (67% in control mice, 3.9% in control GvHD and 26% in R848-protected GvHD) and the same trend was seen for CD8 T cells, indicating that donor T-cell activation was partially inhibited by R848 (Figure 4C). In contrast, the upregulation of CD44 on CD4 T cells was not significantly modified by R848 (2.50-fold upregulation in control and 2.45 fold with R848, Figure 4C). With regards to the activation marker CD69, its expression was unchanged by R848 on CD8 T cells and even somewhat enhanced on CD4 T cells (Figure 4D). In contrast to B6 donor cells, there was no significant change in CD44/CD62L T-cell proportions in control and R848 recipient groups. However, in control GvHD, as the B6D2F1 splenocytes were destroyed, their numbers dramatically dropped (Online Supplementary Figure S1A) whereas in the R848 group, the numbers of CD44+ and CD62L+ T cells were significantly more important than in control GvHD but remained in the same order as in naïve mice. In 395

M. Gaignage et al.

A

E

B F

C D

Figure 3. In vivo treatment of mice with R848 affects responder and presenting cells in mixed lymphocyte culture: role of IFNAR-1. (A) B6D2F1 and B6 mice were treated or not with R848 (25 mg) 48 and 18 h before mixed lymphocyte culture of B6 responder cells and irradiated B6D2F1 APC. After 48 h, (left) proliferation and (right) IFNγ production were determined by 3H-thymidine incorporation and enzyme-linked immunosorbent assay, respectively. (B) Spleen cells from 129/Sv and 129/Sv IFNAR-1-/- mice were collected 48 h after in vivo R848 treatment and incubated with B6D2F1 APC. Proliferations and IFNγ were measured. (C) 129/Sv spleen cells cells were stained for CD4, LIVE/DEAD® and Foxp3 to determine the percentage and absolute numbers of Treg. (D) Treg were depleted with PC61 antibody in B6 mice 4 days before R848 treatment. B6 spleen cells were collected 48 h after in vivo R848 treatment and incubated with B6D2F1 APC and IFNγ was measured after 72 h. (E) FVB (H2q) splenocytes were incubated without APC or with CD11b+ cDC, CD8a+ cDC or pDC purified by MACS beads and FACS sorting from normal and R848-treated 129/Sv mice. After 48 h, proliferation was recorded. (F) Spleen cells from 129/Sv and 129/Sv IFNAR-1-/- mice were collected 48 h after R848 treatment and co-cultured with FVB responder splenocytes. Proliferation and IFNγ were measured. Data are from two to four experiments in all panels (*P<0.05, ***P<0.001 by the Kruskal-Wallis test with Dunn multiple comparison test).

addition, an increase was observed for CD69+ CD4 T cells in R848-protected mice as compared to control B6D2F1 mice (Online Supplementary Figure S1B).

We next examined the influence of R848 treatment on Treg in the same experiments. First, we observed a 10-fold decrease in the proportion of donor and recipient Foxp3+ CD4 T cells 14 days after B6 spleen cell transplantation. In contrast, in R848-treated GvHD mice, recipient Foxp3+ CD4 T cells increased 4-fold while donor Treg returned to normal B6 levels (Figure 4E).

Together, these data indicate that during ncGvHD R848 affected donor CD4 and CD8 T-cell implantation and activation but the inhibition was only partial and did not prevent the establishment of permanent chimerism. On the other hand, R848 prevented GvHD-induced loss of donor and recipient Treg and even increased the latter above normal levels.

396

Regulatory T cells contribute to R848-mediated prevention of non-conditioned graft-versus-host disease The contribution of Treg to the protective effect of R848 was tested by depletion of Treg using anti-CD25 PC61 antibody treatment of B6 donors and B6D2F1 recipients. PC61 antibody was injected 4 days before R848 treatment and a second injection was administered 1 day after B6 spleen cell transplantation. As compared to donor CD4 T cells from R848-treated mice with GvHD, B6 CD4 T cells from the PC61-R848 GvHD group engrafted host spleen faster and their numbers were significantly higher at 14 and 20 days after B6 cell transfer and continued to expand up to day 50.

PC61 antibody completely depleted their Foxp3+ Treg population which remained totally absent during the course of GvHD in contrast to that in the R848-treated mice, in which they expanded more than 6-fold from day 14 to day 50 (Figure 5A). Regarding host Treg, in PC61-R848-treated haematologica | 2019; 104(2)

A TLR7 ligand prevents mouse GvHD

A

B

C

E

D

Figure 4. R848 treatment lowers effector T cells but increases regulatory T cells during non-conditioned graft-versus-host disease. (A) B6D2F1 and B6 mice, treated with R848 or not, received 15x106 B6 CD5+ cells labeled with CFSE from B6 mice not treated (NT) or treated with R848 48 h before cell transfer. After 6 days, antiH-2Dd, -H-2Db and -TCRβ antibodies were used to analyze the proliferation level of B6 T cells in host spleen by flow cytometry. (B-E) 14 days after total B6 spleen cell transfer, spleen cell subsets were enumerated by FACS, using anti-H2Dd, -H2Db, -CD4, -CD8, -CD44, -CD62L and -CD69 antibodies. (B) Plots show absolute numbers of CD4 and CD8 T cells. (C) CD44, CD62L and (D) CD69 expression by T cells is represented with density plots or histograms for percentages and plots for absolute numbers. (E) H-2Dd, H-2Db, LIVE/DEAD®, CD4, TCRβ and Foxp3 staining was used to determine by flow cytometry the percentage (left) and absolute number (right) of Treg in host spleen 14 days after induction of GvHD. Data are from two to three experiments in all panels (*P<0.05, **P<0.01, ***P<0.001 by the Kruskal-Wallis test with Dunn multiple comparison test or Mann–Whitney unpaired t-test).

haematologica | 2019; 104(2)

397

M. Gaignage et al.

A

B

C

E

D

Figure 5. Foxp3+ regulatory T cells are involved in graft-versus-host disease prevention by R848. Treg were depleted with PC61 antibody in donor and recipient mice 4 days before R848 treatment and 1 day after B6 cell transfer in the non-conditioned GvHD model. (A) After 14, 20 and 50 days of B6 cell implantation, H-2Dd, H2Db, LIVE/DEAD®, CD4, TCRβ and Foxp3 staining was used to evaluate Treg populations by flow cytometry. Mice were monitored for (B) weight loss and (C) mortality.

(D) Anti-CD44 and -CD69 antibodies were used to analyze T-cell activation after 14 and 20 days. (E) IFNγ, IL-27p28 and active TGFβ-1 were measured by enzymelinked immunosorbent assay in plasma prepared from mice bled on days 6, 14 and 20. Data are from two to three experiments in all panels (*P<0.05, ** P<0.01, ***P<0.001 by the Kruskal-Wallis test with Dunn multiple comparison test, Mann–Whitney unpaired t-test and ANOVA–Bonferroni post-test).

mice, half of the initial Treg population recovered 14 days after GvHD induction. In contrast, in R848-treated GvHD mice, B6D2F1 Treg numbers almost doubled compared to normal B6D2F1 mice and increased four times versus PC61R848 ncGVHD mice and their levels remained unchanged up to day 50 after transplantation (Online Supplementary Figure S2A,B).

PC61 treatment of the R848 GvHD mice resulted in weight loss starting from day 17, a few days later than in control GvHD mice. The percentage weight loss was finally the same in both groups, suggesting a significant contribution of Treg in the prevention of ncGVHD morbidity by R848 (Figure 5B). However, PC61 treatment only partially decreased the survival of R848-treated mice (70% versus 90%) (Figure 5C). This trend was observed in two additional experiments.

In order to test whether Treg depletion affected the level of donor T-cell activation, we evaluated CD44 and CD69 expression levels 14 and 20 days after ncGVHD induction.

When Treg were depleted in R848-treated mice, CD44+ and CD69+ B6 CD4 and CD8 T cells were significantly 398

increased and CD69 levels even exceeded those of the control ncGVHD group. Compared to day 14 levels, the B6 CD69+ T-cell population tripled at day 20, indicating that an absence of Treg increased expansion of memory and activated donor T cells (Figure 5D). However, Treg depletion by PC61 did not seem to influence early cytokine production since no significant differences in IFNγ, IL-27p28 and active TGF-β1 plasma concentrations were observed between R848- and PC61-R848-treated mice (Figure 5E).

Together, the data suggest that Treg from donors and recipients contributed to R848-mediated GvHD prevention. However, despite the depleting treatment, a small population of host Treg remained present, which could explain why R848 protection was not completely abrogated and resulted in death of only 30% of PC61-R848-treated mice. As shown previously, R848 GvHD protection correlates with a strong drop in pro-inflammatory cytokines and this was still observed after Treg depletion, which could also explain why the protective effect of R848 was not completely suppressed by Treg depletion. haematologica | 2019; 104(2)

A TLR7 ligand prevents mouse GvHD

Figure 6. R848 and anti-interleukin27p28 monoclonal antibody administration to recipient mice prevents acute graft-versus-host disease in irradiated models. (A) BALB/c recipient mice were irradiated with 8 Gy and treated or not with R848 (25 mg/mouse) 48 h and 0 h before transplantation of 2x106 B6 CD5+ splenocytes and 107 B6 T-cell depleted bone marrow cells (TCD-BM). In addition, both recipient groups were treated or not (NT) with 0.5 mg antiIL-27 (aIL-27) on days 0 and 6. Mice were monitored for survival. (B) Six days after B6-cell transfer, IFNγ and IL-27p28 were measured by enzymelinked immunosorbent assay in plasma from irradiated-GvHD BALB/c mice. (C) B6D2F1 recipient mice were irradiated with 5 Gy and treated or not with R848 (25 mg/mouse) 48 and 24 h before transplantation of 40x106 splenocytes and 107 TCD-BM cells from B6 mice similarly treated with R848 or not. In addition, both recipient groups were treated or not with aIL-27 (0.5 mg) on days 0 and 6.

Mice were monitored for survival.

Data are representative of two to three experiments. (*P<0.05, **P<0.01, ***P<0.001 by the log rank test and Kruskal-Wallis test with Dunn multiple comparison test).

A

B

C

R848 cooperates with anti-interleukin-27p28 monoclonal antibody in regulatory T-cell upregulation and graft-versus-host disease prevention in conditioned models B6 spleen cell transfer into ncB6D2F1 recipients is an optimal strategy for dissecting the mechanisms involved in immune-mediated hematopoietic cell destruction and can be used as a model of induced bone marrow failure23 but does not fully replicate allogeneic HCT. To test R848 in full allogeneic HCT, BALB/c mice were injected with R848 24 h before irradiation (8 Gy) and immediately after B6 spleen and bone marrow cell transfer (B6→8Gy-BALB/c). All untreated mice rapidly developed acute GvHD and started to die from day 7 whereas 90% of the mice injected with R848 survived more than 40 days (Figure 6A).

However, R848 did not completely prevent the GvHD reaction since IL-27p28 and IFNγ were still upregulated in R848-protected mice (Figure 6B). This incomplete suppression of a GvHD reaction was even more striking in sublethally irradiated (5 Gy) B6D2F1 recipients of B6 spleen cells, in which R848 pre-treatment of host and donor before transplantation, as in the experiments described above in non-conditioned mice, solely delayed disease and no longer increased survival significantly (Figure 6C), contrary to the total protection seen in non-irradiated recipients. This also correlated with the persistence of pro-inflammatory cytokines IL-27p28 and IFNγ, which R848 treatment failed to decrease (data not shown). As IL-27, which has been haematologica | 2019; 104(2)

shown to contribute to GvHD pathology,10,11 was still present in the irradiated models, we tested the combination of R848 and anti-IL-27 and observed that this resulted in 100% survival of the 5 Gy-irradiated B6D2F1 recipients (Figure 6C).

The anti-IL-27-R848 combination also operated in B6→8Gy-BALB/c GvHD as it significantly decreased the concentration of plasma IFNγ, which was not the case when either agent was used separately (Figure 6B).

However, it did not prevent T-cell engraftment since we observed the same numbers of CD4 and CD8 T cells (Figure 7A) and total B6 donor cell implantation (Figure 7B). Already after 6 days, the number of B6 CD4 T cells recovered from the BALB/c spleen was equivalent to the total number injected and for CD8 T cells the number of cells had increased ±10-fold. These numbers were not modified by R848, anti-IL-27 or the anti-IL-27-R848 combination (Figure 7A). Similarly, upregulation of the memory marker CD44 and the activation marker CD69 in B6 CD4 and CD8 T cells recovered 6 days after transplantation were not modified by R848. The only difference was observed for the CD62L naïve cell marker that was completely lost in B6 CD4 and CD8 T cells recovered from control GvHD mice but remained significantly higher in the group treated with the anti-IL-27-R848 combination (Online Supplementary Figure S3A).

As we previously showed that Treg contribute to the R848 protection in B6→ncB6D2F1 GvHD, Treg popula399

M. Gaignage et al.

A

D

B

C

Figure 7. The combination of anti-interleukin-27p28 monoclonal antibody and R848 induces strong regulatory T-cell activation in acute graft-versus-host disease in irradiated recipients. BALB/c recipient mice were irradiated with 8 Gy and treated or not with R848 (25 mg/mouse) 48 and 0 h before transplantation of 2x106 CD5+ splenocytes from B6 mice. In addition, recipients were treated or not with anti-interleukin-27p28 (aIL-27) on days 0 and 6. After (A) 6 and (B) 40 days of GvHD induction, spleen cells were recovered and stained with anti-H-2Dd, -H-2Db, -CD4 and -CD8 for FACS spleen cell subset analysis. (C-D) H-2Dd, H-2Db, LIVE/DEAD®, CD4, CD8, LAP and Foxp3 staining was added on 6-day-GvHD spleen cells to evaluate Treg populations. Data are from two to three experiments (**P<0.01, ***P<0.001 by the Kruskal-Wallis test with Dunn multiple comparison test).

tions were also analyzed 6 days after B6→8 Gy-BALB/c GvHD induction. The numbers of Foxp3+ CD4 T cells were significantly higher in the R848-treated group than in the control group and adding anti-IL-27 to the R848 treatment further increased the Treg population 2-fold (Figure 7C). Sixty-five percent of these Foxp3+ CD4 cells were positive for latent TGF-β1-associated protein (LAP), attesting their state of activation (Figure 7D). The stimulating effect of the anti-IL-27-R848 combination on Foxp3 expression was even more striking for CD8 Foxp3+ T cells, which were barely detectable in the donor B6 spleen but reached levels equivalent to their CD4 counterparts in the anti-IL-27-R848 group. This was not seen in mice treated with either R848 or anti-IL-27 (Figure 7C).

Interestingly, following anti-IL-27 treatment, B6 Foxp3+ CD4 T cells increased 3-fold more than in the control GvHD group and 75% of these cells were activated. In contrast, R848 administration did not significantly amplify these cells, which correlated with the failure of the TLR7 ligand to increase survival in B6→5Gy-B6D2F1 mice. The anti-IL-27-R848 combination induced a strong increase of Foxp3+ CD4 T cells, which were 5-fold more numerous than in control GvHD mice and 85% were LAP+. In this irradiated model, the Foxp3+ CD8 population, although present, was substantially less abundant (Online Supplementary Figure S3B).

400

Thus, R848 treatment failed to provide complete protection against conditioned GvHD but full protection was restored when anti-IL-27 was added. The combination was necessary to abrogate IFNγ production and to induce maximal Foxp3+ CD4 and CD8 cell responses.

Discussion Innate pattern recognition receptors, including TLR, are implicated in the control of GvHD.37 Most often TLR stimulation aggravates disease, as reported for TLR438 and TLR9,39 but the consequences may differ depending on the time of TLR stimulation with respect to allogeneic HCT as observed for the TLR7/8 agonist 3M-001.40,41 The present study demonstrates the prevention of murine GvHD by the TLR7 ligand resiquimod/R848 and further improvement of protection by inhibition of IL-27.10,11 Our results indicate that R848 administration to recipients 48 h before and at the time of allogeneic HCT promoted survival in lethally irradiated recipients of fully allogeneic hematopoietic cells and in the lethal form of GvHD induced in ncB6D2F1 recipients of B6 parental spleen cells. In this model, R848 inhibited IFNγ, IL-27 and TNFα production while upregulating TGF-β1, thus altering the cytokine balance. However, this suppression of Th1 cytokine produchaematologica | 2019; 104(2)

A TLR7 ligand prevents mouse GvHD

tion was not observed in mice with conditioned GvHD which were also protected by R848, suggesting that inhibition of IFNγ, IL-27 and TNFα was not the only underlying mode of action. This conclusion was indirectly confirmed in the sublethally irradiated B6→B6D2F1 model in which R848 was not protective on its own but still enhanced protection by anti-IL-27. Together, these results indicate that the protective activity of R848 involves but is not limited to IL-27 inhibition as its effect could still be improved by antibody-mediated IL-27 inhibition.

In the B6→ncB6D2F1 GvHD model, complete protection required treatment of both the donor and recipient. As the contribution of F1 recipients to GvHD is limited to antigen presentation, these results suggest that R848 impaired both APC and T-cell responders. Analyses performed just before GvHD induction confirmed this hypothesis since DC antigen presentation and T-cell allo-responsiveness were inhibited in spleen cells collected from R848-treated mice. In agreement with the literature,42 we observed that R848 induced a transient loss and functional inhibition of splenic cDC (both CD11b+ and CD8α+), the main cells able to induce allogeneic responses in vitro and that are known to play an important role in GvHD induction in vivo.43,44 This drop in splenic DC was very transient and antigen presentation returned to normal after 7 days (unpublished observation), suggesting that a short inactivation of host DC is sufficient to alter the initiation of GvHD. R848 induced a similar inhibition of the capacity of T cells to respond to allogeneic in vitro stimulation. Type I interferons seem to be critical in the suppression of DC and T-cell allo-responsiveness

by R848 as both remained unaltered in R848-treated IFNAR-1-/- mice. This observation is in line with reported inhibition of DC and CD4 T cells by type I interferons.24 Importantly, the inhibition of T-cell allo-responsiveness

by R848 in vivo treatment, demonstrated by ex vivo mixed lymphocyte cultures, did not prevent their implantation as chimerism was maintained for months. Moreover, the implanted T cells completely lost naïve T-cell marker CD62L and showed only partial inhibition of CD44 and CD69 memory and activation marker upregulation. This implies the existence of other regulatory mechanisms permitting the persistence of donor T cells in the host with reduced GvHD manifestations. A likely explanation was the effect of R848 on donor and recipient Foxp3+ Treg. The number of these cells dropped dramatically during GvHD but not in R848-treated mice in which their numbers even increased compared to basal control levels. This was particularly striking for Foxp3+ CD8 T cells. Moreover, the presence of LAP on their surface demonstrated that these cells were in an activated state. These results are in agreement with the upregulation of Treg by R848 reported in an asthma model.45 Given the implication of Treg in the control of GvHD,19,20 this Treg stimulation probably contributed to the protective effect of R848. This conclusion was substantiated by the observation that depletion of Treg by anti-CD25 antibody treatment restored morbidity to levels equivalent to those of control GvHD mice although survival was only partly impaired. The importance of Treg upregulation for GvHD prevention by R848 was further demonstrated in the B6→5Gy-B6D2F1 model in which R848 treatment failed to induce a robust Treg response which correlated with poor survival. Interestingly, adding anti-IL-27p28 monoclonal antibody to R848 restored the Treg response and resulted in complete protection. These results are in line with reported haematologica | 2019; 104(2)

GvHD impairment by in vitro R848 stimulation of Treg46 and reveal the R848-anti-IL-27 combination as a new option for maximal Treg upregulation and prevention of GvHD without inhibiting donor cell implantation.

R848 upregulated plasma levels of active TGF-β1 during GvHD, which probably contributed to the observed increase in Treg and to the inhibition of allogeneic T-cell responsiveness that characterized T cells recovered from R848-treated mice. These results are in agreement with the reported severity of murine GvHD induced by donors lacking SMAD335 as well as the predictive impact of TGF-β expression in human allogeneic HCT donors on GvHD occurrence.47 The beneficial effect of high TGF-β1 levels could be related not only to Foxp3+ Treg cell expansion but also to the capacity of TGF-β to inhibit cytotoxic T-cell development.48 The complete and long-lasting donor Treg elimination by anti-CD25 antibody only partly suppressed GvHD prevention by R848. This phenomenon could be explained by the presence of a small host Treg population, which recovered after 14 days and remained stable during the course of the GvHD. This partial suppression of R848 protection could also be explained by the other suppressive mechanisms induced by the drug, such as type I interferon-mediated inhibition of DC and T cells and the downregulation of Th1 cytokines mentioned earlier.

In summary, our results demonstrate that, in the context of allogeneic HCT, R848 alters both alloantigen presentation by cDC and Th1-cell responsiveness in a process dependent on type I interferons and correlating with increased TGF-β1 as well as Treg expansion and diminished IFNγ, TNFα and IL-27 production. When not sufficient on its own, R848 protection can be further enhanced by antiIL-27p28 monoclonal antibody leading to maximal expansion of Foxp3+ Treg cells. A remarkable feature of this novel GvHD preventive procedure is that it needs to be applied only just before and at the time of allogeneic HCT and results in permanent coexistence of the host and allogeneic T cells with a significant reduction in GvHD symptoms.

Transposing these data to human conditions raises a number of issues. One is that R848 activation could be more complex in humans as it activates both TLR7 and TLR8, the latter being inactive in mice. Should this raise problems one could, however, use other agents that activate only TLR7 in humans, such as CL264, a 9-benzyl-8 hydroxyadenine derivative. Another issue is the expression of TLR7 by human not murine CD4 T cells.49,50 This could in fact further improve the efficacy of TLR7-based prevention of GvHD since engaging TLR7 in human CD4 T cells induces a NFATc2-dependent anergy.50 Acknowledgments MG is a FNRS-FRIA PhD Fellow at the Universite Catholique de Louvain. RGM is a Haas-Teichen Fellow of the de Duve Institute. JPC is FNRS research director. The technical help of Pamela Cheou and Emilie Hendrickx and the editorial assistance of Suzanne Depelchin are gratefully acknowledged.

Funding This work was supported by the Belgian National Fund for Scientific Research (FNRS), the “Fondation Contre le Cancer” (FCC)(2010-165), the Joseph Maisin Fund, the Interuniversity Attraction Pole of the Belgian Federal Science Policy, and Actions de Recherche Concertée from Fédération Wallonie-Bruxelles.

401

M. Gaignage et al.

References 1. Appelbaum FR. Hematopoietic cell transplantation from unrelated donors for treatment of patients with acute myeloid leukemia in first complete remission. Best Pract Res Clin Haematol. 2007;20(1):67-75.

2. Mahmoud HK, Elhaddad AM, Fahmy OA, et al. Allogeneic hematopoietic stem cell transplantation for non-malignant hematological disorders. J Adv Res. 2015;6(3):449458.

3. Dvorak CC, Cowan MJ. Hematopoietic stem cell transplantation for primary immunodeficiency disease. Bone Marrow Transplant. 2008;41(2):119-126.

4. Ferrara JL, Levine JE, Reddy P, Holler E. Graft-versus-host disease.

Lancet.

2009;373(9674):1550-1561.

5. Choi SW, Reddy P. Current and emerging strategies for the prevention of graft-versushost disease. Nat Rev Clin Oncol.

2014;11(9):536-547.

6. Markey KA, MacDonald KP, Hill GR. The biology of graft-versus-host disease: experimental systems instructing clinical practice.

Blood. 2014;124(3):354-362.

7. Couriel DR, Saliba R, de Lima M, et al. A phase III study of infliximab and corticosteroids for the initial treatment of acute graft-versus-host disease. Biol Blood Marrow Transplant. 2009;15(12):1555-1562.

8. Chen X, Das R, Komorowski R, et al.

Blockade of interleukin-6 signaling augments regulatory T-cell reconstitution and attenuates the severity of graft-versus-host disease. Blood. 2009;114(4):891-900.

9. Das R, Chen X, Komorowski R, Hessner MJ, Drobyski WR. Interleukin-23 secretion by donor antigen-presenting cells is critical for organ-specific pathology in graft-versushost disease. Blood. 2009;113(10):23522362.

10. Marillier RG, Uyttenhove C, Goriely S, Marbaix E, Van Snick J. IL-27p28 is essential for parent-to-F1 acute graft-versus-host disease. Eur J Immunol. 2014;44(7):2064-2073.

11. Belle L, Agle K, Zhou V, et al. Blockade of interleukin 27 signaling reduces GVHD in mice by augmenting Treg reconstitution and stabilizing FOXP3 expression. Blood.

2016;128(16):2068-2082.

12. Yoshida H, Hunter CA. The immunobiology of interleukin-27. Annu Rev Immunol.

2015;33:417-443.

13. Pflanz S, Timans JC, Cheung J, et al. IL-27, a heterodimeric cytokine composed of EBI3 and p28 protein, induces proliferation of naive CD4(+) T cells. Immunity.

2002;16(6):779-790.

14. Yoshida H, Nakaya M, Miyazaki Y. Interleukin 27: a double-edged sword for offense and defense. J Leukoc Biol. 2009;86 (6):1295-1303.

15. Dibra D, Cutrera JJ, Li S. Coordination between TLR9 signaling in macrophages and CD3 signaling in T cells induces robust expression of IL-30. J Immunol. 2012;188 (8):3709-3715.

16. Pflanz S, Hibbert L, Mattson J, et al. WSX-1 and glycoprotein 130 constitute a signaltransducing receptor for IL-27. J Immunol.

2004;172(4):2225-2231.

17. Villarino AV, Larkin J, 3rd, Saris CJ, et al.

Positive and negative regulation of the IL-27 receptor during lymphoid cell activation. J Immunol. 2005;174(12):7684-7691.

18. Kastelein RA, Hunter CA, Cua DJ. Discovery and biology of IL-23 and IL-27: related but functionally distinct regulators of inflammation. Annu Rev Immunol.

402

2007;25:221-242.

19. Cohen JL, Trenado A, Vasey D, Klatzmann D, Salomon BL. CD4(+)CD25(+) immunoregulatory T cells: new therapeutics for graft-versus-host disease. J Exp Med. 2002;196(3):401-406.

20. Hoffmann P, Ermann J, Edinger M, Fathman CG, Strober S. Donor-type CD4(+)CD25(+) regulatory T cells suppress lethal acute graftversus-host disease after allogeneic bone marrow transplantation. J Exp Med. 2002;196(3):389-399.

21. Taylor PA, Lees CJ, Blazar BR. The infusion of ex vivo activated and expanded CD4(+)CD25(+) immune regulatory cells inhibits graft-versus-host disease lethality.

Blood. 2002;99(10):3493-3499.

22. Gaignage M, Marillier RG, Uyttenhove C, et al. Mouse nidovirus LDV infection alleviates graft versus host disease and induces type I IFN-dependent inhibition of dendritic cells and allo-responsive T cells. Immun Inflamm Dis. 2017;5(2):200-213.

23. Bloom ML, Wolk AG, Simon-Stoos KL, Bard JS, Chen J, Young NS. A mouse model of lymphocyte infusion-induced bone marrow failure. Exp Hematol. 2004;32(12):11631172.

24. Robb RJ, Kreijveld E, Kuns RD, et al. Type IIFNs control GVHD and GVL responses after transplantation. Blood. 2011;118(12): 3399-3409.

25. Ammann CG, Messer RJ, Peterson KE, Hasenkrug KJ. Lactate dehydrogenase-elevating virus induces systemic lymphocyte activation via TLR7-dependent IFNalpha responses by plasmacytoid dendritic cells.

PLoS One. 2009;4(7):e6105.

26. Hemmi H, Kaisho T, Takeuchi O, et al.

Small anti-viral compounds activate immune cells via the TLR7 MyD88-dependent signaling pathway. Nat Immunol.

2002;3(2):196-200.

27. Miller RL, Meng TC, Tomai MA. The antiviral activity of Toll-like receptor 7 and 7/8 agonists. Drug News Perspect. 2008;21(2): 69-87.

28. Holbrook BC, Kim JR, Blevins LK, et al. A Novel R848-conjugated inactivated influenza virus vaccine is efficacious and safe in a neonate nonhuman primate model. J Immunol. 2016;15(2):555-564.

29. Wagner TL, Ahonen CL, Couture AM, et al.

Modulation of TH1 and TH2 cytokine production with the immune response modifiers, R-848 and imiquimod. Cell Immunol.

1999;191(1):10-19.

30. Baenziger S, Heikenwalder M, Johansen P, et al. Triggering TLR7 in mice induces immune activation and lymphoid system disruption, resembling HIV-mediated pathology. Blood.

2009;113(2):377-388.

31. Muller U, Steinhoff U, Reis LF, et al.

Functional role of type I and type II interferons in antiviral defense. Science.

1994;264(5167):1918-1921.

32. Uyttenhove C, Marillier RG, TacchiniCottier F, et al. Amine-reactive OVA multimers for auto-vaccination against cytokines and other mediators: perspectives illustrated for GCP-2 in L. major infection. J Leukoc Biol. 2011;89(6):1001-1007.

33. Lowenthal JW, Corthesy P, Tougne C, Lees R, MacDonald HR, Nabholz M. High and low affinity IL 2 receptors: analysis by IL 2 dissociation rate and reactivity with monoclonal anti-receptor antibody PC61. J Immunol. 1985;135(6):3988-3994.

34. Hattori K, Hirano T, Miyajima H, et al.

Differential effects of anti-Fas ligand and anti-tumor necrosis factor alpha antibodies

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

on acute graft-versus-host disease pathologies. Blood. 1998;91(11):4051-4055.

Giroux M, Delisle JS, Gauthier SD, et al.

SMAD3 prevents graft-versus-host disease by restraining Th1 differentiation and granulocyte-mediated tissue damage. Blood.

2011;117(5):1734-1744.

Hattori H, Matsuzaki A, Suminoe A, et al.

Polymorphisms of transforming growth factor-beta1 and transforming growth factorbeta1 type II receptor genes are associated with acute graft-versus-host disease in children with HLA-matched sibling bone marrow transplantation. Bone Marrow Transplant. 2002;30(10):665-671.

Penack O, Holler E, van den Brink MR. Graft-versus-host disease: regulation by microbe-associated molecules and innate immune receptors. Blood. 2010;115(10): 1865-1872.

Cooke KR, Gerbitz A, Crawford JM, et al.

LPS antagonism reduces graft-versus-host disease and preserves graft-versus-leukemia activity after experimental bone marrow transplantation. J Clin Invest. 2001;107(12): 1581-1589.

Calcaterra C, Sfondrini L, Rossini A, et al.

Critical role of TLR9 in acute graft-versushost disease. J Immunol. 2008;181(9):61326139.

Taylor PA, Ehrhardt MJ, Lees CJ, et al. TLR agonists regulate alloresponses and uncover a critical role for donor APCs in allogeneic bone marrow rejection. Blood. 2008;112(8): 3508-3516.

Jasperson LK, Bucher C, PanoskaltsisMortari A, et al. Indoleamine 2,3-dioxygenase is a critical regulator of acute graft-versus-host disease lethality.

Blood.

2008;111(6):3257-3265.

Assier E, Marin-Esteban V, Haziot A, Maggi E, Charron D, Mooney N. TLR7/8 agonists impair monocyte-derived dendritic cell differentiation and maturation. J Leukoc Biol.

2007;81(1):221-228.

Shlomchik WD, Couzens MS, Tang CB, et al. Prevention of graft versus host disease by inactivation of host antigen-presenting cells.

Science. 1999;285(5426):412-415.

Markey KA, Banovic T, Kuns RD, et al.

Conventional dendritic cells are the critical donor APC presenting alloantigen after experimental bone marrow transplantation.

Blood. 2009;113(22):5644-5649.

Van LP, Bardel E, Gregoire S, et al. Treatment with the TLR7 agonist R848 induces regulatory T-cell-mediated suppression of established asthma symptoms. Eur J Immunol.

2011;41(7):1992-1999.

Zogas N, Karponi G, Iordanidis F, et al. The ex vivo toll-like receptor 7 tolerance induction in donor lymphocytes prevents murine acute graft-versus-host disease.

Cytotherapy. 2018;20(1):149-164.

Baron C, Somogyi R, Greller LD, et al.

Prediction of graft-versus-host disease in humans by donor gene-expression profiling.

PLoS Med. 2007;4(1):e23.

Ranges GE, Figari IS, Espevik T, Palladino MA Jr. Inhibition of cytotoxic T cell development by transforming growth factor beta and reversal by recombinant tumor necrosis factor alpha. J Exp Med. 1987;166(4):991998.

Rahman AH, Taylor DK, Turka LA. The contribution of direct TLR signaling to T cell responses. Immunol Res. 2009;45(1):25-36.

Dominguez-Villar M, Gautron AS, de Marcken M, Keller MJ, Hafler DA. TLR7 induces anergy in human CD4(+) T cells.

Nat Immunol. 2015;16(1):118-128.

haematologica | 2019; 104(2)

ARTICLE

Blood Transfusion

Anti-HLA antibodies with complementary and synergistic interaction geometries promote classical complement activation on platelets

Ferrata Storti Foundation

Maaike Rijkers,1 David Schmidt,2 Nina Lu,1 Cynthia S.M. Kramer,3 Sebastiaan Heidt,3 Arend Mulder,3 Leendert Porcelijn,4 Frans H.J. Claas,3 Frank W.G. Leebeek,5 A.J. Gerard Jansen,1,5 Ilse Jongerius,6 Sacha S. Zeerleder,6 Gestur Vidarsson,2 Jan Voorberg1,7 and Masja de Haas3,4,8

Department of Cellular and Molecular Hemostasis, Sanquin Research and Landsteiner Laboratory Amsterdam UMC, University of Amsterdam; 2Department of Experimental Immunohaematology, Sanquin Research and Landsteiner Laboratory Amsterdam UMC, University of Amsterdam; 3Department of Immunohaematology and Blood Transfusion, Leiden University Medical Center; 4Department of Immunohaematology Diagnostics, Sanquin Diagnostic Services, Amsterdam; 5Department of Hematology, Erasmus University Medical Center, Rotterdam; 6Department of Immunopathology, Sanquin Research and Landsteiner Laboratory Amsterdam UMC, University of Amsterdam; 7 Department of Vascular Medicine, Amsterdam UMC, University of Amsterdam and 8 Center for Clinical Transfusion Research, Sanquin, Leiden, the Netherlands 1

Haematologica 2019 Volume 104(2):403-416

ABSTRACT

H

igh titers of HLA antibodies are associated with platelet refractoriness, causing poor platelet increments after transfusions in a subset of patients with HLA antibodies. Currently, we do not know the biological mechanisms that explain the variability in clinical responses in HLA alloimmunized patients receiving platelet transfusions.

Previously we showed that a subset of anti-HLA IgG-antibodies induces FcγRIIa-dependent platelet activation and enhanced phagocytosis. Here, we investigated whether anti-HLA IgG can induce complement activation on platelets. We found that a subset of anti-HLA IgG induced complement activation via the classical pathway, causing C4b and C3b deposition and formation of the membrane-attack complex. This resulted in permeabilization of platelet membranes and increased calcium influx.

Complement activation also caused enhanced α-granule release, as measured by CD62P surface exposure. Blocking studies revealed that platelet activation was caused by FcγRIIa-dependent signaling as well as HLA antibody induced complement activation. Synergistic complement activation employing combinations of monoclonal IgGs suggested that assembly of oligomeric IgG complexes strongly promoted complement activation through binding of IgGs to different antigenic determinants on HLA. In agreement with this, we observed that preventing anti-HLA-IgG hexamer formation using an IgG-Fc:Fc blocking peptide, completely inhibited C3b and C4b deposition. Our results show that HLA antibodies can induce complement activation on platelets including membrane attack complex formation, pore formation and calcium influx. We propose that these events can contribute to fast platelet clearance in vivo in patients refractory to platelet transfusions with HLA alloantibodies, who may benefit from functional-platelet matching and treatment with complement inhibitors.

Introduction HLA alloantibodies can develop upon transfusion,1 transplantation2 and during pregnancy.3,4 Leukoreduction of platelet transfusion products reduced HLA immunization by more than 50 percent,5 however, 20-30% of patients receiving multiple platelet transfusions still develop HLA alloantibodies.1,3,6 It is known that high titers of HLA antibodies are associated with platelet refractoriness.7 About 12-15% of patients, in need of chronic platelet transfusion support, become refractory to haematologica | 2019; 104(2)

Correspondence: m.dehaas@sanquin.nl

Received: July 11, 2018.

Accepted: September 19, 2018.

Pre-published: September 27, 2018. doi:10.3324/haematol.2018.201665 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/2/403 ©2019 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.

403

M. Rijkers et al.

platelet transfusions and repeatedly show poor increments of platelet counts caused by rapid clearance of the transfused platelets.3,6 HLA-matched platelet transfusions are commonly used for treatment of HLA alloimmunized patients. However, treatment with HLA-matched platelet concentrates is challenging due to the fact that it is often difficult to find a sufficiently high number of compatible donors for refractory patients. Current transfusion approaches for HLA alloimmunized patients are exclusively based on binding specificity of HLA antibodies but do not take into account functional properties of circulating HLA antibodies. Here, we have further characterized the pathogenic properties of different types of HLA-antibodies.

Previously, we showed that a subset of human monoclonal HLA antibodies and patient sera containing HLA antibodies induce FcγRIIa-dependent platelet activation and enhanced phagocytosis by macrophages.8 However, it remains unclear to which extent this HLA antibody-mediated activation of platelets contributes to platelet clearance and which other mechanisms contribute to platelet clearance in refractory patients. In the current study we have focused on the role of complement activation by HLA antibodies.

Platelets have been shown to promote complement activation via several mechanisms. It has been reported that activation of platelets, which leads to α-granule release and subsequent CD62P surface exposure, triggers deposition of complement C3b. C3b can bind directly to CD62P exposed on platelet surfaces, suggesting that platelet activation promotes complement deposition on platelets.9,10 In this case, the alternative pathway of the complement cascade is initiated, where binding of IgG and subsequent C1q deposition is bypassed. Subsequent binding of C3b facilitates further complement activation, finally leading to the formation of a membrane attack complex (MAC), also called the C5b-9 complex.9 Peerschke et al. also showed that C1q can bind to agonist-activated platelets, indicating a possible role for platelets in complement activation via the classical complement pathway.11 Platelet activation can also induce complement activation in the fluid phase, where the release of chondroitin sulfate by activated platelets is the trigger.12 Also, binding of C3 to activated platelets has been suggested to stimulate formation of platelet-leukocyte interactions.13 In addition, IgGcomplexes can induce platelet aggregation, which is strongly enhanced by addition of C1q.14 Mouse monoclonal antibodies (mAbs) directed to beta-2 microglobulin (β2M) and a pan HLA mAb have been shown to induce C3b binding and complement dependent cytotoxicity (CDC) on platelets when added at high concentrations.15,16 Platelet transfusion-related adverse events might be (partly) explained by complement activation in platelet products as standard storage conditions have been shown to induce complement activation with increasing C3a and C4d levels found in platelet concentrates upon prolonged storage.17 Here, we studied complement activation on platelets induced by HLA antibodies. Human HLA mAbs and sera from patients with refractory thrombocytopenia containing HLA antibodies were used to study the effect of complement deposition, formation of a MAC, platelet activation and permeabilization. Our results show that a subset of anti-HLA antibodies can induce complement activation 404

on platelets. We also showed that blocking pathways leading to complement deposition on platelets, prevented complement activation induced pathogenicity of HLA antibodies. Based on our findings, we propose that functional matching of platelet concentrates may be used to further improve treatment of refractory patients with HLA antibodies. Our results also suggest that complementdirected therapeutic interventions may be utilized to increase donor platelet survival in HLA-immunized refractory patients.

Methods Materials Detailed information on materials used can be found in the Online Supplementary Data.

Patient sera Blood samples of patients refractory to platelet transfusion were used following informed consent according to the Dutch established codes of conduct for responsible use of patient material and as approved by our institute.18 HLA antibody specificities in patient sera were determined by single antigen bead assay on Luminex platform (Labscreen SA, One Lambda, Inc.). Twelve sera positive for HLA antibodies and negative for other platelet specific antibodies were used.

Platelet isolation Platelets were isolated from citrated whole blood from healthy volunteers with known HLA type. All donors gave written informed consent and blood was drawn in accordance with Dutch regulations and after approval from the Sanquin Ethical Advisory Board in accordance with the Declaration of Helsinki. Platelets were isolated and washed as described before.19 Platelets were resuspended in platelet assay buffer (10 mM HEPES, 140 mM NaCl, 3 mM KCl, 0.5 mM MgCl2, 10 mM glucose and 0.5 mM NaHCO3, pH 7.4).

Complement deposition and platelet activation Platelets were used at a final concentration of 0.08*108 platelets/ml and mixed with indicated inhibitors, antibodies/sera (heat inactivated for 30 min at 56°C, 25% of total sample volume) and complement source (normal human serum) (25% of total volume) in platelet assay buffer. Mixtures were incubated for 30 min at 37°C while shaking (300 rpm), and then fixed by adding formaldehyde (final concentration of 1%). Platelets were washed with platelet assay buffer and stained for flow cytometry. AntiCD42a-FITC or CD41-APC-Cy7 antibodies were used to gate for platelets. Platelets were stained with mouse anti-human CD62PPE and mouse anti-human C3b-APC or mouse anti-human C4bAPC. For measuring the formation of a MAC, platelets were stained with rabbit anti-human C5b-9 antibody followed by the secondary antibody chicken anti-rabbit Alexa 647. Mean fluorescent intensities and/or percentage positive platelets were measured using flow cytometer FACSCanto II (Becton Dickinson, Franklin Lakes, NJ, USA).

Pore formation Complement assay was initiated as described above. After 30 min incubation at 37°C, platelets were washed once and resuspended in assay buffer containing live/dead marker for 30 min at room temperature (RT). Platelets were washed and stained with anti-CD42a (for gating purposes), anti-C3b and CD62P and analyzed using flow cytometry. haematologica | 2019; 104(2)

Anti-HLA antibodies with complementary and synergistic interaction

Calcium influx

Data and statistical analysis

Platelets were loaded with calcein violet and fluo-4 for 30 min at RT, washed and resuspended in platelet assay buffer.

Complement activation assay was initiated as described above.

After 20 min incubation at 37°C, platelets were diluted in assay buffer and calcium influx was measured immediately using flow cytometry by measuring fluo-4 signal. Platelets were gated upon their calcein violet fluorescence.

Flow cytometry data was analyzed using FlowJo version 10 (Ashland, OR, USA). Data are represented as either mean Âą standard deviation (SD) or all data points are shown. Statistical analyses were performed using GraphPad Prism 7 Version 7.02 (La Jolla, CA, USA); the analyses used are specified in the respective figure legends.

A

B

C

D

E

F

R2 0.979 P <0.005

haematologica | 2019; 104(2)

Figure 1. HLA mAb WIM8E5 induces C3b and C4b deposition on platelets. (A) Platelets were incubated with increasing concentrations of WIM8E5 and C3b deposition was measured employing flow cytometry.

Representative data from 1 donor. (B) C3b deposition on platelets incubated with 20 mg/ml WIM8E5 were compared to control platelets (no HLA antibody added). (n=7 different donors). (C) Representative flow cytometry graph of C4b deposition on platelets incubated with WIM8E5 (20 mg/ml). (D) C4b deposition on platelets from 7 different donors upon incubation with 20 mg/ml WIM8E5. (E) Comparison of C3b deposition on platelets from 3 donors with different amounts of WIM8E5 reactive antigens. Donor HLA types are indicated in the figure, strong WIM8E5 binding antigens in green, weak binding antigens in blue and non-binding antigens in red. (F) Five donors heterozygous for HLA-A2 (and no A3 or A32) with different HLA expression levels (as measured with anti-human IgG in the absence of serum). Level of WIM8E5 binding correlates with percentage C3b positive platelets. Strong binding antigens are indicated in green, weak binding antigens in blue and non-binding antigens in red.

MFI: mean fluorescence intensity. **P<0.01.

405

M. Rijkers et al.

Results HLA antibody WIM8E5 induces complement C3b and C4b deposition on platelets Human HLA mAbs recognizing different epitopes encoded by different HLA class I subtypes (Online Supplementary Table S1) were incubated with platelets from donors with an HLA type matching the specificity of the mAbs, in the presence of normal human serum.

Complement deposition of C3b on platelet surfaces increased upon incubation with increasing concentrations of HLA mAb WIM8E5 (specific for an epitope present on all HLA-A antigens except for HLA-A3 and-A32 and with reduced binding to HLA-A2) (Figure 1A). Statistically significant increased C3b deposition was observed with 20 mg/ml WIM8E5 (Figure 1B). Also, C4b deposition on platelet surfaces was significantly increased when platelets were incubated with WIM8E5 (Figure 1C-D). A

A

B

C

D

Figure 2. Combinations of HLA mAbs induce enhanced complement activation. (A) Four different HLA mAbs were incubated with HLA matched platelets both separately and in combination (20 mg/ml per antibody) and C3b deposition was measured. (B-C) Examples of C3b deposition induced by HLA mAbs separately and in combination (20 mg/ml per HLA mAb). (D) C3b deposition on platelets of 2 donors by a combination of anti-HLA-A2 antibodies SN607D8 and SN230G6 at 0.0125 - 10 mg/ml. HLA donor 1: A2 A31 B60 HLA donor 2: A2 A68 B51 B57.

406

haematologica | 2019; 104(2)

Anti-HLA antibodies with complementary and synergistic interaction

strong variation in the level of complement deposition of C3b and C4b was observed between donors, whose various HLA type matched in all cases with the specificity of WIM8E5 (Figure 1B and D). A gene dosage effect was observed; donors with two HLA-A antigen specificities compatible with WIM8E5 binding induced more C3b deposition compared to donors expressing one compatible HLA-A antigen and HLA-A2 (with reduced binding or

affinity) (Figure 1E). Only minor C3b deposition was observed when platelets from donors expressing only one compatible HLA-A antigen (e.g., A11 and A32) were used (Figure 1E right panel). Variable levels of C3b deposition on platelets of 5 donors expressing HLA-A2 in combination with another WIM8E5 HLA-A-compatible antigen were observed that correlated with WIM8E5 binding (Figure 1F) This suggests that the level of IgG opsoniza-

A

B

C

D

E

F

G

Figure 3. Complement deposition on platelets by HLA mAbs occurs via the classical complement pathway. (A) Pre-incubation of platelets with 50 mg/ml anti-C1q inhibits C3b deposition (n=7), (B) and C4b deposition (n=7). (C) Dose response of increasing concentrations anti-C1q to inhibit C3b deposition induced by 20 mg/ml WIM8E5 (n=4) (D) Effect of increasing concentrations of Fc:Fc blocking peptide DCAWHLGELVWCT on complement deposition induced by WIM8E5 (20 mg/ml) was measured by C3b deposition (n=3), (E) and C4b deposition (n=4). (F) Effect of 200 mg/ml Fc:Fc blocking peptide on C3b deposition by WIM8E5 and SN607D8/SN230G6 was compared to the irrelevant control peptide GWTVFQKRLDGSV (n=4). *P<0.05, **P<0.01.

haematologica | 2019; 104(2)

407

M. Rijkers et al.

tion, probably largely affected by the number of available epitopes for WIM8E5, is correlated with the level of platelet complement activation.

Combinations of anti-HLA mAbs induce complement deposition on platelets Next, less broadly reacting anti-HLA human IgG1 mAbs were tested for their ability to induce C3b deposition on platelet surfaces. No C3b deposition was observed when

SN607D8, SN230G6, HDG8D9 and BRO11F6 were incubated separately (Figure 2A; upper panel). However, by combining these HLA mAbs, thereby mimicking the polyclonal nature of HLA alloimmunized patient sera, several combinations promoted C3b deposition on platelets (Figure 2A; lower panel and Figure 2B). This effect was very strong when a combination of anti- HLA-A2 mAbs, SN607D8 and SN230G6, were used (Figure 2B). Also, combinations of monoclonal antibodies anti-HLA-A2

A

B

C

Figure 4. Platelet activation occurs via complement activation and FcγRIIadependent activation. (A) C3b deposition and CD62P exposure were measured in the presence and absence of serum upon incubation with WIM8E5, with or without pre-incubation of the platelets with Syk inhibitor (5 mM) or FcγRIIa blocking antibody IV.3 (10 mg/ml). (B) In the presence of serum, platelets were pre-incubated with anti-C1q (50 mg/ml), Syk inhibitor IV (5 μM) or both and CD62P and C3b deposition were measured upon addition of WIM8E5 (20 mg/ml). Representative flow cytometry graph, bar graph is mean ± SD (n=8). (C) Platelets were pre-incubated with IV.3 (10 mg/ml), Syk inhibitor IV (5 mM) and/or the C5 inhibitor Eculizumab (10 mg/ml) in the presence of serum. C3b and CD62P exposure were measured upon incubation with WIM8E5 (20 mg/ml). Representative flow cytometry graphs, bar graph represents mean ± SD (n=8). Syk inh: Syk inhibitor; *P<0.05; **P<0.01; ***: P<0.005; ****P<0.001; ns: not significant.

408

haematologica | 2019; 104(2)

Anti-HLA antibodies with complementary and synergistic interaction

(SN230G6) and anti-HLA-A11 (BRO11F6) (Figure 2B), anti-HLA-A11 (BRO11F6) and anti-HLA-B52 (OK8F11), or anti-HLA-B44 (DK7C11) and anti-HLA-B62 (OK8F11) induced C3b deposition on platelets. Only when incubated together, the two anti-HLA-A2 antibodies SN607D8 and SN230G6, which bind to opposite sides of the peptide binding groove in HLA-A2,20,21 were very effective in inducing complement activation. Even at concentrations as low as 0.2 Âľg/ml, these antibodies induced C3b deposition (Figure 2D). The levels of complement activation are

A

however dependent on the donor. Together, these results suggest that the geometry of binding of anti-HLA antibodies determines whether or not complement deposition can occur.

Complement deposition induced by anti-HLA antibodies occurs via the classical pathway Both classical- and alternative complement pathway activation have been described to occur on platelets.9,11,22 To discriminate between these pathways, platelets were

B

C

D

E

F

G

H

Figure 5. HLA antibodies induce the formation of a MAC, pore formation and calcium influx. (A) Complement activation was induced by SN607D8/SN230G6 and was inhibited by Eculizumab. Formation of a MAC was measured employing an anti-C5b-9 antibody. (n=4) (B) Representative flow cytometry plot showing SN607D8/SN230G6 induced formation of the C5b-9 complex (solid line). In the presence of Eculizumab the formation of the C5b-9 complex is inhibited (dashed line). (C) Pore formation upon incubation with increasing concentrations of SN607D8 and SN230G6 was measured using an impermeable dye, binding to free amines: referred to as live/dead marker (L/D). (n=4) (D) Pore formation was blocked with anti-C1q. (n=4) (E) Representative flow cytometry plot of pore formation measured by L/D marker upon complement activation by 10 mg/ml SN607D8/SN230G6 in presence and absence of 50 mg/ml anti-C1q. (F) Platelets were loaded with fluo-4 to measure a calcium influx upon incubation with increasing concentrations of SN607D8 and SN230G6. On the y-axis, the percentage of Fluo-4 positive platelets is depicted (n=4). (G) Pre-incubation of platelets with 50 mg/ml anti-C1q inhibited calcium influx induced by 10 mg/ml SN607D8/SN230G6. (n=4) (H) Representative flow cytometry plot of fluo-4 signal induced by 10 mg/ml SN607D8/SN230G6 in the presence and absence of 50 mg/ml anti-C1q. Data are represented as mean Âą SD. MFI: mean fluoresence intensity, *P<0.05; **P<0.01; ***P<0.005.

haematologica | 2019; 104(2)

409

M. Rijkers et al.

pre-incubated with a blocking antibody directed to the tail of C1q (anti-C1q), preventing binding of C1q to Fc-tails of immunoglobulins. Complete blockage of WIM8E5 induced C3b and C4b deposition on platelet surfaces was achieved by anti-C1q at a concentration of 50 mg/ml (Figure 3A-B). The blocking effect of anti-C1q was dose dependent (Figure 3C). These results indicate a crucial role for C1q binding in WIM8E5 induced C3b and C4b deposition, suggesting that complement activation occurs via the classical pathway.

It has been described that IgG molecules can form hexamers to which C1q can efficiently bind.23 Employing a synthetic peptide shown to interfere with Fc-dependent hexamer formation and subsequent complement deposition,23,24 we studied whether anti-HLA antibody-induced complement deposition was blocked by this peptide. IgGFc:Fc blocking peptide inhibited C3b and C4b deposition

by WIM8E5 or by the combination of SN607D8 and SN230G6 in a dose dependent manner (Figure 3D-E). No effects were observed with an irrelevant control peptide (Figure 3F-G). Together, these results suggest that antiHLA antibody-induced complement deposition on platelets involves Fc tail-mediated assembly of IgG hexamers, creating a suitable binding platform for C1q.

Anti-HLA antibodies induce complement- and FcγRIIa-dependent platelet activation We have recently shown that a subset of anti-HLA mAbs (including WIM8E5) can activate platelets directly through FcγRIIa in the absence of a complement source.8 Under these conditions, WIM8E5-induced CD62P exposure was completely blocked with either anti FcγRIIablocking antibody IV.3 or through Syk inhibitor IV (which blocks downstream signaling of FcγRIIa25) (Figure 4A).

Incubation of platelets with WIM8E5 in the presence of

A

B

C

D

serum as a complement source resulted in a marked increase in CD62P exposure on platelets (Figure 4A).

Under these conditions, CD62P exposure was only slightly inhibited by blocking FcγRIIa or with Syk inhibitor IV (Figure 4A). These results suggest that in the presence of complement activation, platelet activation is only partially FcγRIIa-dependent. To study the effect of complement activation on α-granule release, complement activation was blocked with anti-C1q. This significantly inhibited C3b deposition, but also only partly blocked α-granule release as measured by CD62P exposure (Figure 4B).

However, a combination of Syk inhibitor and anti-C1q blocked WIM8E5-induced CD62P exposure as well as C3b deposition completely (Figure 4B).

Previously, Wiedmer et al. described a mechanism where the formation of a MAC on platelet surfaces induces platelet activation and α-granule release caused by calcium influx through the MAC.26 We used a C5 blocking antibody (Eculizumab) to prevent cleavage of C5 and thereby the formation of a MAC. Like anti-C1q, Eculizumab partly inhibited CD62P surface exposure (Figure 4C). WIM8E5-induced CD62P exposure was completely inhibited when Eculizumab was combined with either Syk inhibitor IV or FcγRIIa blocking antibody IV.3 (Figure 4C). Similar results were obtained employing combinations of HLA mAbs (Online Supplementary Figure S1).

Together, these results suggest that in presence of a complement source, platelet activation by anti-HLA antibodies is dependent on formation of a MAC as well as FcγRIIadependent signaling.

MAC formation leads to calcium influx To confirm MAC formation, platelets activated with a combination of SN607D8 and SN230G6 were stained with an anti-C5b-9 antibody. As expected,

E

Figure 6. IVIg and C1 esterase inhibitor inhibit complement deposition on platelets. (A-B) Platelets were incubated with increasing concentrations of IVIg (0-10 mg/ml), after which WIM8E5 (20 mg/ml) (A) or SN607D8 and SN230G6 (0.2 mg/ml) (B) were added. C3b deposition was measured on the platelet surfaces. (n=3).

(C-E) Platelets were pre-incubated with 0-600 μg/ml C1 esterase inhibitor. C3b deposition upon incubation with (C) 20 mg/ml WIM8E5 or (D) 0.2 mg/ml SN607D8 and SN230G6 was measured. (E) Inhibitory effect of C1 esterase inhibitor on C4b deposition was measured upon incubation with WIM8E5 (20 mg/ml) or SN607D8 and SN230G6 (0.2 mg/ml) (n=3).

410

haematologica | 2019; 104(2)

Anti-HLA antibodies with complementary and synergistic interaction

SN607D8/SN230G6 induced enhanced C5b-9 on the platelet surface, which was blocked with anti-C1q (Figure 5A-B and Online Supplementary Figure S2).

Formation of a MAC is associated with cellular lysis through its end product which is pore formation.27 To

A

B

measure the permeability of platelet membranes, a live/dead marker (binding free amines, fluorescent signal increases in case of damaged membranes due to accessibility of intracellular amines) was added to platelets after incubation with SN607D8/SN230G6. Incubation with

D

E

C F

G

H

Figure 7. HLA antibody containing sera can induce complement activation on platelets. Twelve heat inactivated sera containing HLA antibodies were used to test complement activation and Syk-dependent activation on platelets from 3 different donors. (A-C) Complement activation (in presence of serum) was induced in presence and absence of 25 mg/ml anti-C1q. (D-F) Syk-dependent activation (as measured by CD62P exposure) was induced in presence and absence of 5 mM Syk inhibitor IV in the absence of a complement source. HLA typing of donor 1: A2 A3 B7, donor 2: A1 A2 B15:01 B35, donor 3: A11 A24 B35 B38. (G) IgG binding measured in absence of complement source employing an anti-human IgG antibody. (H) In green the antibodies present in sera matching the HLA typing of the platelet donors, and in red the expressed HLA-antigens for which no matching antibodies are present are indicated. Dotted lines in panel E-F correspond to background values of heat inactivated serum controls (no anti-platelet antibodies present). MFI: mean fluorescence intensity.

haematologica | 2019; 104(2)

411

M. Rijkers et al.

Figure 8. Proposed mechanism of anti-HLA antibody induced complement deposition. Upon binding of anti-HLA antibodies to HLA molecules on platelets, C1q can bind to HLA-bound IgGs. This leads to initiation of the classical complement pathway which results in C4b and C3b deposition on the platelet surface. Eventually this leads to the formation of a MAC which promotes the influx of Ca2+. Elevated intra-platelet Ca2+ levels induce platelet activation as measured by CD62P exposure. AntiHLA antibody-induced complement activation can be inhibited using anti-C1q blocking antibody or Eculizumab at the C1q of C5 level, respectively. Via a separate mechanism binding of HLA antibodies to HLA molecules can cross-link with FcγRIIa and induce platelet activation independent of the complement pathway.

these HLA mAbs induced a large increase in membrane permeability providing evidence for pore formation in platelet membranes (Figure 5C). Anti-HLA antibodies induced pore formation in a concentration-dependent manner (Online Supplementary Figure S2) and pore formation could be inhibited by anti-C1q (Figure 5D-E).

Comparison of C3b deposition, C5b-9 complex formation and pore formation revealed a clear correlation between these three parameters upon incubation of platelets derived of two different donors with increasing concentrations of SN607D8/SN230G6 (Online Supplementary Figure S2). Under the same conditions, an increase in calcium influx in platelets loaded with fluo-4 was measured (Figure 5F-H). Together, these results suggest that complement activation induced by HLA antibodies leads to formation of a MAC, with subsequent pore formation in platelet membranes resulting in Ca2+ influx.

Immunoglobulins (IVIg) and C1 esterase inhibitor inhibit HLA antibody-induced complement activation IVIg can inhibit FcγRIIa-dependent platelet activation by anti-HLA antibodies.8 Similar effects have been observed in the context of heparin-induced thrombocytopenia.28,29 Here, we also tested if IVIg affects complement activation and observed a dose-dependent inhibition of C3b deposition by IVIg (Figure 6A-B). Similarly, by inhibiting C1 employing the C1 esterase inhibitor, C3b and C4b deposition induced by HLA mAbs was inhibited (Figure C-E and Online Supplementary Figure S1). These results show that complement activation on platelets induced in vitro by HLA mAbs can be inhibited by IVIg and C1 esterase inhibitor in a dose dependent manner.

Sera containing HLA antibodies can induce complement activation on platelets, which is not correlated to Syk-mediated activation In order to confirm our data obtained with human monoclonal HLA antibodies, sera containing HLA antibodies 412

from 12 patients refractory to platelet transfusions were tested for their ability to induce C3b deposition on platelets (Online Supplementary Table S2). Depending on the HLA type of the donor platelets, C3b deposition was observed upon incubation with sera containing HLA antibodies (Figure 7A-C). In agreement with the results obtained for human mAbs, incubation with anti-C1q blocked C3b deposition on platelet membranes (Figure 7A-C). In case of strong complement activation, 25 mg/ml anti-C1q only partly inhibited C3b deposition. C3b deposition, however, was completely blocked by anti-C1q when the anti-HLA sera were diluted (Online Supplementary Figure S4). This confirms that HLA antibodies, as present in patient sera, induce complement activation via the classical pathway.

FcγRIIa-dependent α-granule release was also investigated with the same donors and platelet donors, in absence of complement source (Figure 7D-F). Some sera induced both C3b deposition and FcγRIIa-dependent platelet activation, however, similar to the results obtained with HLA mAbs, there were also sera which activated only one of these two pathways. These results suggest that HLA antibodies may induce complement activation and FcγRIIa-dependent platelet activation via distinct mechanisms. To confirm that complement activation induced by HLA antibodies present in patients sera correspond to that of HLA mAbs, platelet complement activation and Syk-dependent activation were blocked.

Similar to results obtained with HLA mAbs (Figure 4 and Online Supplementary Figure S1), both Syk-dependent and complement-dependent platelet activation occurs when platelets are incubated with HLA antibody containing sera (Online Supplementary Figure S5). Furthermore, C3b deposition induced by these sera could be inhibited employing the IgG-Fc:Fc blocking peptide (Online Supplementary Figure S6).

The level of IgG binding differed among sera and also depended on the donor platelets used (Figure 7G and haematologica | 2019; 104(2)

Anti-HLA antibodies with complementary and synergistic interaction

Online Supplementary Figure S3). Although some association between IgG binding and complement deposition and FcγRIIa-dependent platelet activation was observed, some sera displaying only limited IgG binding were still capable of inducing complement activation or Syk-dependent activation. This suggests that only a subset of antiHLA antibodies present in patient sera can induce C3b deposition on platelets.

Discussion Development of HLA alloantibodies remains a major cause of refractoriness to platelet transfusions. As yet, the pathogenic properties of HLA alloantibodies have not been sufficiently characterized. Here, we have shown that HLA antibodies are capable of inducing complement activation on platelets, leading to C4b and C3b deposition on platelet surfaces and the formation of a functional MAC. It was previously shown that platelet activation can lead to the binding of C3b,9 inducing complement activation via the alternative pathway. We have previously shown that HLA antibodies can induce FcγRIIa-dependent platelet activation, leading to CD62P surface exposure.8 Therefore, we tested whether the complement activation

by HLA antibodies occurs via the classical pathway, or whether CD62P surface exposure directly leads to C3b binding followed by activation of the alternative complement pathway. Since a blocking antibody directed to C1q completely blocked C3b and C4b deposition, complement activation by HLA antibodies appears to be fully dependent on the classical pathway. In line with this observation, blocking FcγRIIa-dependent activation had no effect on C3b deposition on platelets. Similar results were obtained in patient sera containing HLA antibodies. Also, the ability of HLA antibody-containing sera to induce complement activation was not directly correlated to FcγRIIa-dependent platelet activation. Some sera were capable of inducing both complement- and FcγRIIa-dependent activation of platelets.

We observed enhanced CD62P exposure in the samples in which complement was activated. In the presence of a complement source, CD62P exposure was only partly blocked upon inhibition of FcγRIIa-signaling.

Simultaneous blocking of complement activation and FcγRIIa-signaling resulted in complete inhibition of platelet activation as measured by CD62P exposure. This suggests that HLA antibodies can induce platelet activation by promoting complement deposition as well as FcγRIIa-dependent signaling. We speculate that both mechanisms may contribute to the rapid clearance of platelets in refractory patients.

In our study, we demonstrated that not all HLA mAbs and patient sera induce complement activation to a similar extent. Specific combinations of mAbs showed significantly enhanced C3b deposition; strong, synergistic complement deposition induced by incubation of platelets with low concentrations of two distinct HLA mAbs (SN607D8 and SN230G6) was observed. This suggests that structural organization of platelet bound IgG is crucial for initiation of the complement cascade. It has been shown that C1q binds most efficiently to IgG hexamers, for which Fc:Fc interactions are needed.23,30,31 In line with these observations, we could completely block complement deposition induced by HLA mAbs by adding an IgGhaematologica | 2019; 104(2)

Fc blocking peptide24 described to prevent these Fc:Fc interactions. The same blocking peptide inhibited C3b deposition on platelets induced by HLA antibody-containing patient sera. These results suggest that the propensity of polyclonal (patient)-derived HLA antibodies to form IgG hexamers on the platelet surface provides a binding platform for C1q which finally promotes the formation of C5b-9 complexes on platelet membranes. Additional properties of HLA antibodies such as specific subclass or composition of the glycan in the Fc part of IgGs are expected to potentially modulate complement activation of HLA antibodies on platelets.

In this respect, it is interesting to note that glycosylation of IgG1 has been described to affect the binding of IgG1 to C1q which also might affect the ability of individual antibodies to induce complement activation.32 Altogether, this might explain why only a subset of HLA mAbs and patient sera containing HLA antibodies induce complement activation. Interestingly, sera containing HLA antibodies matching the HLA type of the platelet donor did not always induce complement activation. This may partly be explained by differences in concentration and binding specificity of the different antibodies in these polyclonal sera, and varying levels of HLA expression on the donor platelets used in this study.7 Also, the presence of specific IgG subclasses might play a role, as different subtypes bind with different affinity to C1q.33 In this study, we show that low concentrations of specific HLA antibodies can induce complement activation in a synergistic manner. This observation suggests that depending on the HLA profile and HLA density on donor platelets, specific combinations of patient-derived HLA antibodies can potently induce complement activation. Our results indicate that the propensity of HLA antibodies to induce complement activation is not per se related to the titer and binding specificity of HLA antibodies in serum. Rather, our findings suggest that the ability of HLA antibodies to form hexameric IgG complexes on platelets determines their ability to induce complement activation.

Ideally, the ability of HLA antibodies of individual patients to induce platelet activation8 and /or complement deposition on donor platelets should be assessed before transfusion. Previous studies have shown that platelet cross match does increase platelet recovery in patients, but data on occurrence of hemorrhage and mortality are lacking.34 Implementation of functional tests focusing on HLA antibody-induced platelet activation and complement deposition may have the potential to functionally stratify platelet concentrates for specific HLA alloimmunized patients. This would further increase the pool of suitable donor platelets for treatment of HLA alloimmunized patients who are currently dependent on treatment with HLA-matched platelet concentrates. This would increase transfusion safety, lower the costs and avoid delays in provision of donor platelets, which is important for transfusion support in chronically transfused patients.

There are several treatment options for refractory thrombocytopenia, which from an immunological perspective could target both FcγR- and complementdependent clearance of donor platelets. Immunoglobulins (IVIg) could inhibit C3b deposition on platelets induced by HLA mAbs. Inhibition of complement deposition by IVIg has previously been shown in a number of autoimmune disorders.35,36 A randomized trial in HLA alloimmunized patients treated with IVIg revealed increased platelet-cor413

M. Rijkers et al.

rected count increments one hour after transfusion compared to placebo; however, no beneficial effects were observed 24 hours after transfusion.37 Thus, IVIg might be an effective way to prevent rapid platelet clearance induced by HLA antibodies. Similarly, Eculizumab could block formation of a MAC.38,39 Preliminary data of a clinical trial showed that Eculizumab could overcome platelet transfusion refractoriness in patients with HLA-alloantibodies.40 The C3 inhibitor Compstatin was not tested here, but is also used in clinical practice already41 and might have a beneficial effect for patients refractory for platelet transfusion caused by HLA alloantibodies. Also, C1 esterase inhibitor, which blocks complement activation at the beginning of the complement cascade, inhibited C3b and C4b deposition induced by HLA mAbs. Blocking complement activation at this point might be even more beneficial than blocking after formation of C4b and C3b. The potential beneficial effect of complement inhibitors in refractory patients receiving platelet transfusions needs to be confirmed in clinical studies.

In conclusion, in this study we have shown that HLA antibodies can induce complement activation on platelets

References 1. Pavenski K, Freedman J, Semple JW. HLA alloimmunization against platelet transfusions: pathophysiology, significance, prevention and management. Tissue Antigens.

2012;79(4):237-245.

2. Hickey MJ, Valenzuela NM, Reed EF, Reed EF. Alloantibody generation and effector function following sensitization to human leukocyte antigen. Front Immunol. 2016; 7:30.

3. Stanworth SJ, Navarrete C, Estcourt L, Marsh J. Platelet refractoriness – practical approaches and ongoing dilemmas in patient management. Br J Haematol. 2015; 171(3):297-305.

4. De Clippel D, Baeten M, Torfs A, et al.

Screening for HLA antibodies in plateletpheresis donors with a history of transfusion or pregnancy.

Transfusion.

2014;54(12)3036-3042.

5. The Trial to Reduce Alloimmunization to Platelets Study Group *. Leukocyte reduction and ultraviolet b irradiation of platelets to prevent alloimmunization and refractoriness to platelet transfusions. N Engl J Med. 1997;337(26):1861-1869.

6. Rebulla P. A mini-review on platelet refractoriness. Haematologica. 2005; 90(2):247253.

7. Jackman RP, Deng X, Bolgiano D, et al.

Low-level HLA antibodies do not predict platelet transfusion failure in TRAP study participants. Blood. 2013; 121(16):32613267.

8. Rijkers M, Saris A, Heidt S, et al. A subset of anti-HLA antibodies induces FcγRIIa dependent platelet activation.

Haematologica. 2018;103(10):1741-1752 9. del Conde I, Crúz MA, Zhang H, López JA, Afshar-kharghan V. Platelet activation leads to activation and propagation of the complement system. J Exp Med. 2005; 201(6):871-879.

10. Hamad OA, Nilsson PH, Wouters D, John

414

11.

12.

13.

14.

15.

16.

17.

18.

through the classical complement pathway, as is schematically represented in the visual abstract. We propose that C3b complement deposition induced by HLA alloantibodies in refractory patients might contribute to platelet clearance through their binding to C11b/CD18 on myeloid cells. Ongoing complement activation on HLA-sensitized platelets will lead to formation of a MAC allowing for Ca2+-influx and subsequent platelet activation.

Importantly, these effects of HLA antibodies may not only be prevented via HLA-matching between donors and recipient antibodies, but also by functional matching, since not all HLA antibodies are capable of inducing complement activation. Furthermore, inhibition of complement activation, as currently studied in clinical trials, may be of benefit to further optimize platelet support in HLA immunized patients.

Funding Supported by grant PPOC-2013-019 and PPOC-2015024P (Netherlands Ministry of Health). AJGJ is supported by a Clinical Fellowship of the European Hematology Association. We thank Diana Wouters for critical feedback on the experiments.

D, Ekdahl KN, Nilsson B. Complement component C3 binds to activated normal platelets without preceding proteolytic activation and promotes binding to complement receptor 1. J Immunol. 2010; 184(5):2686-2692.

Peerschke EIB, Yin W, Grigg S, Ghebrehiwet B. Blood platelets activate the classical pathway of human complement. J Thromb Haemost. 2006;4(9):2035-2042.

Hamad OA, Ekdahl KN, Nilsson PH, et al.

Complement activation triggered by chondroitin sulfate released by thrombin receptor-activated platelets. J Thromb Haemost.

2008;6(8):1413-1421.

Hamad OA, Mitroulis I, Fromell K, et al.

Contact activation of C3 enables tethering between activated platelets and polymorphonuclear leukocytes via CD11b/CD18.

Thromb Haemost. 2015;114(6):1207–1217.

Peerschke EIB, Ghebrehiwet B. C1q augments platelet activation in response to aggregated Ig. J Immunol. 1997; 159(11):5594-5598.

Pokrass MJ, Liu MF, Lindorfer MA, Taylor RP. Activation of complement by monoclonal antibodies that target cell-associated β2 -microglobulin : Implications for cancer immunotherapy. Mol Immunol. 2013; 56(4):549-560.

Meinke S, Sandgren P, Mörtberg A, et al.

Platelets made HLA deficient by acid treatment aggregate normally and escape destruction by complement and phagocytes in the presence of HLA antibodies.

Transfusion. 2016;56(2):370-382.

Chen J, Losos M, Yang S, Li J, Wu H, Cataland S. Increased complement activation during platelet storage. Transfusion 2017;57(9):2182–2188.

Nightingale MJ, Ceulemans J, Ágoston S, et al. The value to blood establishments of supplier quality audit and of adopting a European Blood Alliance collaborative approach. Blood Transfus. 2014;12(1):9198.

19. Rijkers M, van der Meer PF, Bontekoe IJ, et al. Evaluation of the role of the GPIb-IX-V receptor complex in development of the platelet storage lesion. Vox Sang.

2016;111(3):247-56.

20. Mulder A, Eijsink C, Kester MGD, et al.

Impact of peptides on the recognition of HLA class I molecules by human HLA antibodies. J Immunol. 2005;175(9):5950-5957.

21. Duquesnoy RJ, Marrari M, Jelenik L, Zeevi A, Claas FHJ, Mulder A. Structural aspects of HLA class I epitopes reacting with human monoclonal antibodies in Ig-binding, C1qbinding and lymphocytotoxicity assays.

Hum Immunol. 2013;74(10):1271-1279.

22. Speth C, Rambach G, Würzner R, et al.

Complement and platelets: Mutual interference in the immune network. Mol Immunol. 2015;67(1):108-118.

23. Diebolder CA, Diebolder CA, Beurskens FJ, et al. Complement is activated by IgG hexamers assembled at the cell surface.

Science. 2014;343(6176):1260-1263.

24. Delano WL, Ultsch MH, Vos AM De, Wells JA. Convergent solutions to binding at a protein-protein interface. Science. 2000; 287(5456):1279-1283.

25. Yamamoto N, Takeshita K, Shichijo M, et al. The orally available spleen tyrosine kinase inhibitor 2-[7-(3,4Dimethoxyphenyl)-imidazo[1,2-c]pyrimidin-5-ylamino]-nicotinamide dihydrochloride (BAY 61-3606) blocks antigen- induced airway inflammation in rodents. J Pharmacol Exp Ther. 2003; 306(3):11741181.

26. Wiedmer T, Ando B, Sims PJ. Complement C5b-9-stimulated platelet secretion is associated with a Ca2+ -initiated activation of cellular protein kinases. J Biol Chem.

1987;262(28):13674-13681.

27. Tegla CA, Cudrici C, Patel S, et al.

Membrane attack by complement: the assembly and biology of terminal complement complexes.

Immunol Res. 2011;51(1):45-60.

haematologica | 2019; 104(2)

Anti-HLA antibodies with complementary and synergistic interaction 28. Jones CG, Pechauer SM, Curtis BR, Bougie DW, Aster RH, Padmanabhan A. IgG in normal plasma inhibits HIT antibodymediated platelet activation: Implications for plasma exchange in HIT. Blood.

2018;131(6):703â&#x20AC;&#x201C;706.

29. Padmanabhan A, Jones CG, Pechauer SM, et al. IVIg for treatment of severe refractory heparin-induced thrombocytopenia. Chest.

2017;152(3):478-485.

30. Ugurlar D, Howes SC, Kreuk B De, et al.

Structures of C1-IgG1 provide insights into howdanger pattern recognition activates complement. Science. 2018;359(6377):794797.

31. Wang G, de Jong RN, van den Bremer ETJ, et al. Molecular basis of assembly and activation of complement component C1 in complex with immunoglobulin G1 and antigen. Mol Cell. 2016;63(1):135-145.

32. Dekkers G, Plomp R, Koeleman CAM, et al. Multi-level glyco-engineering techniques to generate IgG with defined Fc-gly-

haematologica | 2019; 104(2)

cans. Sci Rep. 2016;6(36964):1-12.

33. Vidarsson G, Dekkers G, Rispens T. IgG subclasses and allotypes: from structure to effector functions. Front Immunol. 2014; 5:520.

34. Vassallo RR, Fung M, Rebulla P, et al. Utility of cross-matched platelet transfusions in patients with hypoproliferative thrombocytopenia: A systematic review.

Transfusion 2014;54(4):1180-1191.

35. Basta M, Dalakas MC. High-dose intravenous immunoglobulin exerts its beneficial effect in patients with dermatomyositis by blocking endomysial deposition of activated complement fragments. J Clin Invest.

1994;94(5):1729-1735.

36. Kazatchkine M, Kaveri S. Immunomodulation of autoimmune and inflammatory diseases with intravenous immune globulin. N Engl J Med. 2001; 345(10):747-755.

37. Kickler T, Braine HG, Piantadosi S, Ness PM, Herman JH, Rothko K. A randomized,

38.

39.

40.

41.

placebo-controlled trial of intravenous gammaglobulin in alloimmunized thrombocytopenic patients. Blood. 1990; 75(1):313-316.

Thajudeen B, Sussman A, Bracamonte E. A case of atypical hemolytic uremic syndrome successfully treated with eculizumab. Case Rep Nephrol Urol. 2013;3(2):139146.

Ariceta G, Arrizabalaga B, Aguirre M, Morteruel E, Lopez-Trascasa M. Eculizumab in the treatment of atypical hemolytic uremic syndrome in infants. Am J kidney Dis. 2012;59(5):707-710.

Vo PT, Purev E, West KA, et al.

Complement inhibition using eculizumab overcomes platelet transfusion refractoriness in allo-immunized patients receiving HLA mismatched platelets. Blood.

2016;128(22):3840.

Huang Y. Evolution of compstatin family as therapeutic complement inhibitors. Expert Opin Drug Discov. 2018;13(5):435-444.

415