Haematologica, Volume 104, Issue 8 by Haematologica

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 8: August 2019 Cover Figure Bone marrow smear showing a giant plasmablast with two macronucleoli and deeply basophilic cytoplasm with Immunoglobulin accumulation in the form of amorphous substance or crystals in a patient with plasma cell myeloma. Courtesy of Prof. Rosangela Invernizzi.

Editorials 1503

Does iron let boys grow faster?! Günter Weiss

1505

The wolf of hypomethylating agent failure: what comes next? Anne Sophie Kubasch and Uwe Platzbecker

1508

Discontinuation of tyrosine kinase inhibitors in patients with chronic myelogeneous leukemia: it may be easier than you think Charles A. Schiffer

1511

The only thing that stops a bad microbiome, is a good microbiome Jessica R. Galloway-Peña and Robert R. Jenq

1513

Predicting risk for recurrence of arterial ischemic stroke in children: thrombophilia as another piece of the puzzle Ghada Aborkhees and Lesley Gayle Mitchell

Perspective Article 1515

“Somatic” and “pathogenic” - is the classification strategy applicable in times of large-scale sequencing? Constance Baer et al.

Review Articles 1521

Which are the most promising targets for minimal residual disease-directed therapy in acute myeloid leukemia prior to allogeneic stem cell transplant? Brian Ball and Eytan M. Stein

1532

How close are we to incorporating measurable residual disease into clinical practice for acute myeloid leukemia? Nicholas J. Short and Farhad Ravandi

Articles Iron Metabolism & its Disorders

1542

Rapid growth is a dominant predictor of hepcidin suppression and declining ferritin in Gambian infants Andrew E. Armitage et al.

Red Cell & its Disorders

1554

Clinical and biological features in PIEZO1-hereditary xerocytosis and Gardos channelopathy: a retrospective series of 126 patients Véronique Picard et al.

Myelodysplastic Syndromes

1565

A phase II study of guadecitabine in higher-risk myelodysplastic syndrome and low blast count acute myeloid leukemia after azacitidine failure Marie Sébert et al.

Myeloproliferative Neoplasms

1572

Epigenomic profiling of myelofibrosis reveals widespread DNA methylation changes in enhancer elements and ZFP36L1 as a potential tumor suppressor gene that is epigenetically regulated Nicolás Martínez-Calle et al.

Haematologica 2019; vol. 104 no. 8 - August 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

Myeloproliferative Neoplasms

1580

Characteristics and outcomes of patients with essential thrombocythemia or polycythemia vera diagnosed before 20 years of age: a systematic review Jean-Christophe Ianotto et al.

Chronic Myeloid Leukemia

1589

Observational study of chronic myeloid leukemia Italian patients who discontinued tyrosine kinase inhibitors in clinical practice Carmen Fava et al.

Acute Myeloid Leukemia

1597

RUNX1 inhibits proliferation and induces apoptosis of t(8;21) leukemia cells via KLF4-mediated transactivation of P57 Shuang Liu et al.

Acute Lymphoblastic Leukemia

1608

ZEB2 and LMO2 drive immature T-cell lymphoblastic leukemia via distinct oncogenic mechanisms Steven Goossens et al.

1617

DNMT3A mutation is associated with increased age and adverse outcome in adult T-cell acute lymphoblastic leukemia Jonathan Bond et al.

Non-Hodgkin Lymphoma

1626

Human leukocyte antigen class II expression is a good prognostic factor in adult T-cell leukemia/lymphoma Mai Takeuchi et al.

1633

The novel CD19-targeting antibody-drug conjugate huB4-DGN462 shows improved anti-tumor activity compared to SAR3419 in CD19-positive lymphoma and leukemia models Stuart W. Hicks et al.

Plasma Cell Disorders

1640

Once-weekly versus twice-weekly carfilzomib in patients with newly diagnosed multiple myeloma: a pooled analysis of two phase I/II studies Sara Bringhen et al.

Platelet Biology & its Disorders

1648

Platelet glycoprotein VI and C-type lectin-like receptor 2 deficiency accelerates wound healing by impairing vascular integrity in mice Surasak Wichaiyo et al.

1661

All-trans retinoic acid protects mesenchymal stem cells from immune thrombocytopenia by regulating the complementâ&#x20AC;&#x201C;interleukin-1Î˛ loop Xiaolu Zhu et al.

Hemostasis

1676

Recurrent stroke: the role of thrombophilia in a large international pediatric stroke population Gabrielle deVeber et al.

Stem Cell Transplantation

1682

Fecal microbiota transplantation before or after allogeneic hematopoietic transplantation in patients with hematologic malignancies carrying multidrug-resistance bacteria Giorgia Battipaglia et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/104/8.toc

e331

Diet and gender influence survival of transgenic Berkley sickle cell mice Om B. Jahagirdar et al http://www.haematologica.org/content/104/8/e331

e335

Stem cell transplantation for congenital dyserythropoietic anemia: an analysis from the European Society for Blood and Marrow Transplantation Maurizio Miano et al. http://www.haematologica.org/content/104/8/e335

Haematologica 2019; vol. 104 no. 8 - August 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation e340

Differential effects of therapeutic complement inhibitors on serum bactericidal activity against non-groupable meningococcal isolates recovered from patients treated with eculizumab Dan M. Granoff et al. http://www.haematologica.org/content/104/8/e340

e345

TP53 immunohistochemistry correlates with TP53 mutation status and clearance in decitabine-treated patients with myeloid malignancies Marianna B. Ruzinova et al. http://www.haematologica.org/content/104/8/e345

e349

Tracking myeloid malignancies by targeted analysis of successive DNA methylation at neighboring CG dinucleotides Monika Eipel et al. http://www.haematologica.org/content/104/8/e349

e352

CD371 cell surface expression: a unique feature of DUX4-rearranged acute lymphoblastic leukemia Dagmar Schinnerl et al. http://www.haematologica.org/content/104/8/e352

e356

Long-term follow up after third-party viral-specific cytotoxic lymphocytes for immunosuppression- and Epstein-Barr virus-associated lymphoproliferative disease Sajida Kazi et al. http://www.haematologica.org/content/104/8/e356

e360

Post-transplant outcome of ovarian tissue cryopreserved after chemotherapy in hematologic malignancies Catherine Poirot et al. http://www.haematologica.org/content/104/8/e360

e364

Ectopic ATP synthase Î˛ subunit proteins on human leukemia cell surface interact with platelets by binding glycoprotein IIb Ting Wang et al. http://www.haematologica.org/content/104/8/e364

e369

Role of factor VIII-binding capacity of endogenous von Willebrand factor in the development of factor VIII inhibitors in patients with severe hemophilia A Yohann RepessĂŠ et al. http://www.haematologica.org/content/104/8/e369

Case Reports Case Reports are available online only at www.haematologica.org/content/104/8.toc

e373

Shared cell of origin in a patient with Erdheim-Chester disease and acute myeloid leukemia Armin Ghobadi et al. http://www.haematologica.org/content/104/8/e373

e376

Epstein-Barr virus-associated lymphoproliferative disease during imatinib mesylate treatment for chronic myeloid leukemia Junko Yamaguchi et al. http://www.haematologica.org/content/104/8/e376

e380

Rescue factor VIII replacement to secure hemostasis in a patient with hemophilia A and inhibitors on emicizumab prophylaxis undergoing hip replacement Elena Santagostino et al. http://www.haematologica.org/content/104/8/e380

Comment Comment is available online only at www.haematologica.org/content/104/8.toc

e383

Tamoxifen erythroid toxicity revealed by studying the role of nuclear receptor co-activator 4 in erythropoiesis Antonella Nai et al. http://www.haematologica.org/content/104/8/e383

Haematologica 2019; vol. 104 no. 8 - August 2019 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

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

Ancient Greek

aÂma [haima] = blood a·matow [haimatos] = of blood lÒgow [logos]= reasoning haematologicus (adjective) = related to blood

Scientific Latin

Modern English

haematologica (adjective, plural and neuter, used as a noun) = hematological subjects The oldest hematology journal, publishing the newest research results. 2018 JCR impact factor = 7.570

EDITORIALS Does iron let boys grow faster?! GĂźnter Weiss1,2 1

Department of Internal Medicine II, Innsbruck Medical University and 2Christian Doppler Laboratory for Iron Metabolism and Anemia Research, Innsbruck, Austria E-mail: GĂ&#x153;NTER WEISS - guenter.weiss@i-med.ac.at doi:10.3324/haematol.2019.222018

ron is an essential nutrient for the body as it plays a part in multiple enzymatic processes, including DNA synthesis, mitochondrial respiration, oxygen transport, hormone formation, and cellular metabolism.1 Iron deficiency and iron deficiency anemia (the latter arising from limited availability of the metal for heme biosynthesis) are global health problems that affect around two billion people. These are particularly important in infants because they have a negative impact on children's growth and mental development.2 Such a situation is highly prevalent in developing countries. Thus, efforts have been made to substitute iron to avoid such developmental defects in children. However, the unbiased administration of iron supplements to childrenâ&#x20AC;&#x2122;s diets in tropical regions resulted in a significant increase in morbidity and mortality from infectious diseases.3 These can be attributed to the fact that iron is an essential nutrient also for most pathogens but also impacts on the efficacy of anti-microbial immune effector pathways.4,5 Subsequent studies have shown that mild iron deficiency in infants even offers protection from specific infections such as severe malaria.6 This has left physicians with the dilemma as to how to identify children who may benefit from iron supplementation while avoiding the risk of an adverse outcome from infection. Thus, several diagnostic approaches have been adopted to identify those children who may respond to iron supplementation therapy. In this context, the determination of the iron hormone hepcidin has attracted great interest. Hepcidin is a liverderived peptide which controls body iron homeostasis upon binding to the only known cellular iron export protein ferroportin, resulting in its internalization and degradation.1 Hepcidin expression is induced by body iron loading or inflammatory signals, including those arising from systemic infections, whereas iron deficiency (as well as, among others, hypoxia and anemia) reduce hepcidin expression.7 Accordingly, low hepcidin levels enable dietary or orally supplemented iron to be absorbed from the duodenum, whereas high-circulating hepcidin levels impair iron transfer from duodenal enterocytes to the circulation.8 In other words, subjects with true iron deficiency efficiently absorb iron from the duodenum, whereas persistent inflammation impairs iron uptake from the gut with iron remaining in the intestine.8 This not only results in a blunted response to oral iron therapy, but also increases the availability of iron for the intestinal microbiome. This leads to subtle alterations of the composition of the microbiota with an increase in pathogenic bacteria and promotion of intestinal inflammation.9 Thus, hepcidin determination in children has been seen to be a reliable diagnostic test to predict the response to oral iron therapy.10 This is also of interest as infection inducible inflammatory signals impact on cytokine formation and stimulate hepcidin production, resulting in the development of functional iron deficiency, particularly in countries with a high endemic burden of infectious diseases. This functional iron deficiency is characterized by iron retention in reticuloendothelial cells and the emergence of anemia of inflammation haematologica | 2019; 104(8)

or anemia of chronic disease which poorly responds to oral iron.11 However, in tropical countries, due to nutritional iron deficiency and/or chronic blood loss on the basis of intestinal infestation with hookworms, counter-regulatory factors can impact on hepcidin levels. Studies in animal models have shown that the inhibitory signals exerted by iron deficiency dominate over hepcidin induction by inflammation.12 This has also been confirmed in clinical trials in young women and in patients with inflammatory bowel disease and low-grade inflammation showing good absorption of oral iron.13,14 This would suggest that low hepcidin levels, even in an inflammatory setting, would predict sufficient oral iron absorption. To gain greater insight into how hepcidin levels are regulated and affected by different factors in a primary care setting, and how these change in early infancy over time, Armitage and co-workers analyzed data from two birth cohorts in The Gambia, Western Africa, adopting a longitudinal approach to the analysis.15 They took repeat measurements of serum concentrations of hepcidin, iron, the iron storage protein ferritin, and soluble transferrin receptor (sTfR) (which is a marker for the needs of iron for erythropoiesis) and studied the results for associations of these markers with birth weight, growth, seasonality, infection, anemia, and nutrition. Children were investigated from birth until one year of age. First, the authors observed that low iron and hepcidin levels at birth were associated with a lower birthweight, pointing to the importance of sufficient maternal iron supplementation during pregnancy. Second, they also found a decrease in hepcidin, iron and ferritin levels over time which is indicative for incorporation of the metal into the growing body. Of note, a greater weight gain was associated with more severe iron deficiency as reflected by low ferritin and hepcidin levels. This also indicated that the faster growth of children is paralleled by or even a consequence of more efficient incorporation of iron in the body where it is used for erythropoiesis and enzymatic complexes including myoglobin in muscle cells. However, such faster growing children are more likely to become iron deficient because dietary iron availability cannot compensate for the increased incorporation of iron in the body. Thus, such children need specific attention in order to avoid unwanted negative effects of iron deficiency on their development from one year of age onwards; based on the data presented by Armitage et al.,15 these infants can be identified by low hepcidin levels at the age of 12 months, but this also predicts that they will respond to oral iron therapy. Most surprisingly, the significant association between growth promotion and iron deficiency was most pronounced in boys. Even at five months of age, a higher prevalence of both iron deficiency and anemia became evident in males as compared to female subjects. Of note, at this early stage, there was a negative association between higher hepcidin levels and gain of weight and length in both sexes, confirming that infection-driven elevation of hepcidin negatively impacts on iron absorption.8 1503

Editorials

Figure 1. Factors impacting on iron availability and absorption for infants. Dietary iron is absorbed in the duodenum. The bioavailability and quantitative absorption of iron are dependent on the one hand by the molecular/heme iron content of the diet and on the other hand on the concentration of hepcidin. The latter blocks the transfer of iron from the duodenal enterocyte to the circulation, which is a prerequisite for iron availability for cells and tissues. Among other factors, hepcidin expression is stimulated by infection and inflammation, inhibited by iron deficiency, anemia and hypoxia, and can be influenced by genetic polymorphisms of iron metabolism genes in either direction. It is anticipated that efficient incorporation of the metal by infants is positively associated with promotion of growth and mental development.

Nonetheless, this leads to questions about the mechanisms underlying gender-specific differences in developmental growth and iron handling in infancy. Iron absorption and iron utilization for erythropoiesis are known to be affected by genetic polymorphisms in different iron metabolism and erythropoietic genes.2,16 Apart from the description of sex-linked anemia in mice and males,1 no specific genetic defects with higher prevalence in females have been described. One might also speculate that cultural differences in feeding procedures between boys and girls or dietary additives in addition to breast feeding which impact on iron bioavailability, may play a role in this setting. It could also be that there is a higher driving force of iron to be incorporated into muscle tissue in boys than in girls, although this would be surprising at this early stage of development. The latter is believed to be rather driven by sex-specific effects of hormones which would not be evident in infancy. Later on in life, this may become more relevant, because testosterone promotes muscular development whereas estrogens have a positive effect on inflammatory pathways which may negatively impact on dietary iron absorption.7 The same also holds true for hormonal effects on hepcidin expression, which is reduced by testosterone but likewise only becomes important in adolescence. Differences in the prevalence of infections with associated impairment of 1504

dietary iron absorption also do not appear to account for this because fewer females than males were affected by infections. Another issue could arise from sex-specific differences in intestinal infestation with hookworms which aggravates iron losses by duodenal bleeding. Nonetheless, it is also plausible that more sustained growth is independent of iron absorption, meaning that iron deficiency is the consequence, and not the cause, of growth that is actually driven by other factors. Thus, the issue of sex-specific differences in iron handling and putative iron-mediated growth promotion remains a matter of speculation which should be addressed in future prospective trials. Not surprisingly, the authors15 also found that hepcidin levels are much affected by markers of inflammation, namely C-reactive protein (CRP), but also by seasonality, both of which point to a role for infections in their impact on hepcidin levels. This observation generates new knowledge which can help predict the optimal time frame for iron substitutions; this could include recommending those months with the lowest seasonal burden of infections as this would increase efficacy or iron absorption and reduce the risk of an increased incidence or unfavorable course of infections. Moreover, iron administration has been shown to be quite safe when preventive measures for reducing the burden of infectious diseases are undertaken. A recent haematologica | 2019; 104(8)

Editorials

report demonstrated that the use of insecticide to impregnate bed nets and screening for parasites in blood reduced the malaria risk in children on iron supplementation.17 Moreover, recent data demonstrate that iron deficiency negatively impacts on immunological responses to diphtheria vaccine leaving children insufficiently protected against such infections (N Stoffel, Zurich, oral presentation, Bioiron Meeting 2019). Thus, this study by Armitage15 and co-workers is an important step forward to gain more insights into the relative contribution of different regulatory mechanisms on circulating biomarker concentrations such as hepcidin and how this impacts on predicting therapeutic efficacy and the risk:benefit ratio of iron supplementation in a primary care setting. Future studies will have to clarify the optimal timing and dose of iron supplementation to children, whether or not a continuous administration via dietary iron fortification or a once daily or once every other day application is preferable.18 It will also be necessary to identify those children who might be at risk of unwanted effects of iron supplementation mainly arising from an increased morbidity and mortality from infections. Finally, we await further information on the impact of iron supplementation on growth and mental development, functionality of the immune system, efficacy of preventive measures such as vaccination, and the consequences of iron-mediated alterations of the intestinal microbiota on childrenâ&#x20AC;&#x2122;s health.

References 1. Muckenthaler MU, Rivella S, Hentze MW, Galy B. A Red Carpet for Iron Metabolism. Cell. 2017;168(3):344-361. 2. Camaschella C. Iron deficiency. Blood. 2019;133(1):30-39. 3. Weiss G, Carver PL. Role of divalent metals in infectious disease susceptibility and outcome. Clin Microbiol Infect. 2018;24(1):16-23.

4. Armitage AE, Drakesmith H. Genetics. The battle for iron. Science. 2014;346(6215):1299-1300. 5. Soares MP, Weiss G. The Iron age of host-microbe interactions. EMBO Rep. 2015;16(11):1482-1500. 6. Gwamaka M, Kurtis JD, Sorensen BE, et al. Iron deficiency protects against severe Plasmodium falciparum malaria and death in young children. Clin Infect Dis. 2012;54(8):1137-1144. 7. Girelli D, Nemeth E, Swinkels DW. Hepcidin in the diagnosis of iron disorders. Blood. 2016;127(23):2809-2813. 8. Theurl I, Aigner E, Theurl M, et al. Regulation of iron homeostasis in anemia of chronic disease and iron deficiency anemia: diagnostic and therapeutic implications. Blood. 2009;113(21):5277-5286. 9. Paganini D, Uyoga MA, Zimmermann MB. Iron Fortification of Foods for Infants and Children in Low-Income Countries: Effects on the Gut Microbiome, Gut Inflammation, and Diarrhea. Nutrients. 2016;8(8). 10. Prentice AM, Doherty CP, Abrams SA, et al. Hepcidin is the major predictor of erythrocyte iron incorporation in anemic African children. Blood. 2012;119(8):1922-1928. 11. Weiss G, Ganz T, Goodnough LT. Anemia of inflammation. Blood. 2019;133(1):40-50. 12. Theurl I, Schroll A, Nairz M, et al. Pathways for the regulation of hepcidin expression in anemia of chronic disease and iron deficiency anemia in vivo. Haematologica. 2011;96(12):1761-1769. 13. Stoffel NU, Lazrak M, Bellitir S, et al. The opposing effects of acute inflammation and iron deficiency anemia on serum hepcidin and iron absorption in young women. Haematologica. 2019 Jan 10. [Epub ahead of print] 14. Reinisch W, Staun M, Tandon RK, et al. A randomized, open-label, non-inferiority study of intravenous iron isomaltoside 1,000 (Monofer) compared with oral iron for treatment of anemia in IBD (PROCEED). Am J Gastroenterol. 2013;108(12):1877-1888. 15. Armitage AE, Agbla SC, Betts M, et al. Rapid growth is a dominant predictor of hepcidin suppression and declining ferritin in Gambian infants. Haematologica. 2019;104(8):1542-1553. 16. Andrews NC. Genes determining blood cell traits. Nat Genet. 2009;41(11):1161-1162. 17. Aimone AM, Brown P, Owusu-Agyei S, Zlotkin SH, Cole DC. Impact of iron fortification on the geospatial patterns of malaria and nonmalaria infection risk among young children: a secondary spatial analysis of clinical trial data from Ghana. BMJ Open. 2017;7(5):e013192. 18. Stoffel NU, Cercamondi CI, Brittenham G, et al. Iron absorption from oral iron supplements given on consecutive versus alternate days and as single morning doses versus twice-daily split dosing in iron-depleted women: two open-label, randomised controlled trials. Lancet Haematol. 2017;4(11):e524-e533.

The wolf of hypomethylating agent failure: what comes next? Anne Sophie Kubasch1,2,3 and Uwe Platzbecker1,2,3 1

Medical Clinic and Policlinic 1, Hematology and Cellular Therapy, Leipzig University Hospital, Germany; 2German MDS Study Group (G-MDS) and 3European Myelodysplastic Syndromes Cooperative Group (EMSCO group, www.emsco.eu)

E-mail: UWE PLATZBECKER - uwe.platzbecker@medizin.uni-leipzig.de doi:10.3324/haematol.2019.222794

yelodysplastic syndromes (MDS) and acute myeloid leukemia (AML) are clonal hematopoietic stem/progenitor cell (HSPC) disorders mainly affecting the elderly population.1 Hypomethylating agents (HMA) like azacitidine and decitabine have become the standard of care in elderly patients with highrisk (HR) MDS or AML unfit for intensive treatment approaches. Until today, responses to HMA have occured in less than 50% of patients and are not durable, with only a few patients achieving long-lasting remissions.2,3 Prognostic clinical markers, such as presence of peripheral blasts, high transfusion burden, and poor performance status, have been identified as indicators of a worse outcome of HMA-based therapy.1,4 Moreover, responses to HMA are especially short-lived in patients with adverse haematologica | 2019; 104(8)

risk cytogenetic abnormalities compared to those with normal karyotype.1 Craddock et al. evaluated the impact of mutational profile on clinical response to azacitidine by analyzing 250 patients with newly diagnosed, relapsed, or refractory AML or HR-MDS. Lower complete response (CR) rates occurred in patients with an IDH2 and STAG2 mutation, higher CR rates in patients with NPM1 mutation. Mutations in CDKN2A, IDH1, TP53, NPM1, and FLT3ITD were associated with a worse overall survival (OS) in univariate analysis, while multivariate analysis showed a decrease in OS in patients with CDKN2A, IDH1, or TP53 mutations. Moreover, ASXL1 and ETV6 were associated with short response duration after azacitidine treatment.5 Despite all efforts to try to select patients based on 1505

Editorials

their cytogenetic and molecular characteristics, failing HMA therapy is still associated with a dismal prognosis reflected by a median survival of six months.6 Until now, almost nothing has been known about the mechanisms underlying HMA-resistance. Thus, as frequently as possible, patients experiencing HMA failure should be evaluated for clinical trial options, given the current absence of any available standard treatment in that setting. In clinically fit patients with HR-MDS or secondary AML (sAML) and normal karyotype, intensive chemotherapy with a subsequent allogeneic stem cell transplantation may also be considered. In this issue of the Journal, SĂŠbert et al. report results of a phase II study of the Groupe Francophone des MyĂŠlodysplasies (GFM) investigating the novel HMA guadecitabine (SGI-110) as a salvage treatment in HRMDS and low blast count AML (<30% bone marrow blasts) patients after azacitidine failure.7 Guadecitabine is a dinucleotide of decitabine and deoxyguanosine with similar potency but longer half life due to resistance to cytidine deaminase degradation. This results in an extended exposure of blasts to its active metabolite decitabine. The study included fifty-six patients (median age 75 years) who failed or relapsed after at least six previous azacitidine cycles. Patients' characteristics indicated a study population with advanced disease, including 87.5% of patients carrying high-risk somatic mutations such as ASXL1 (25%), RUNX1 (21%), TP53 (20%) and U2AF1 (20%).7 Patients achieving hematologic response after 3, 6 or 9 cycles of guadecitabine (60 mg/m2/day subcutaneously days 1-5 of 28-day treatment cycles) were considered to be responders and were allowed to continue treatment until loss of response. SĂŠbert et al. identified eight (14.3%) responding patients, including two CR, one partial response (PR), three hematologic improvements

(HI), and two marrow CR (mCR). Median response duration was 11.5 months and median OS 7.1 months; responders had a prolonged median OS of 17.9 months.7 Therefore, even after failure to an HMA, HMA-based treatment with guadecitabine can be an effective alternative treatment option prolonging survival in a small proportion of HR-MDS and AML patients, with a toxicity profile similar to that of standard HMA. The authors present an analysis of prognostic factors for response and prolonged OS after guadecitabine treatment. Especially patients with primary azacitidine failure, absence or limited number of somatic mutations and lower methylation level in blood during the first cycle of treatment benefited from guadecitibaine treatment.7 A phase III trial comparing guadecitabine with treatment of choice in MDS patients after HMA failure is currently ongoing (clinicaltrials.gov identifier: 02907359). One other promising approach to improving efficacy of hypomethylation in HR-MDS and AML patients is the addition of the orally selective B-cell lymphoma 2 (BCL-2) inhibitor venetoclax to HMA. BCL-2 protein, a key regulator of leukemic blast survival, has been reported to play an important role in regulating apoptosis via the intrinsic mitochondrial cell death. Overexpression of the BCL-2 protein has been shown to be associated with poor outcomes, conferring chemotherapeutic resistance in AML.8 Recent data suggest that 400 mg of venetoclax has an optimal benefit-risk profile when used in combination with azacitidine.8 This combination has already demonstrated impressive rates of CR both in the frontline and relapse settings: AML patients treated with first-line venetoclax and HMA showed favorable overall response rates (ORR) (CR: 71% and CRi: 74%).9,10 Median duration of response after achieving CR was 21.2 months and median overall survival was 16.9 months.9 Comparing these results with

Figure 1. New options after hypomethylating agent failure.

1506

haematologica | 2019; 104(8)

Editorials

historically poor outcome data of single agent azacitidine treatment (CR rates approx. 20%; OS not exceeding 12 months in AML patients3), it becomes clear that this novel targeted combined strategy will potentially dominate the future treatment landscape in HR-MDS and AML patients not eligible for intensive induction therapy. Nevertheless, previous clinical trials in AML demonstrated a toxicity profile that represents cause for concern. Over 50% of included patients developed grade ≥3 neutropenia, leading to a high incidence of treatment interruption and subsequent study discontinuation due to progressive disease (PD).9,11 Recruitment is currently underway for a clinical study evaluating the combination of venetoclax with azacitidine in patients with HR-MDS after HMA failure (clinicaltrials.gov identifier: 02966782). It is known that HMA can reduce immune response by upregulation of inhibitory immune checkpoint molecule expression. Therefore, preventing resistance to HMA by combining HMA and checkpoint inhibition is another possible new treatment strategy which is currently under investigation in several clinical trials. We reported on a patient with sAML undergoing single agent pembrolizumab (anti-PD-1) treatment.12 After two months of therapy, platelet count increased in line with a response according to International Working Group (IWG) 2018 criteria, together with clearance of IDH1 mutation.12 Recently Daver et al. reported on a phase II study evaluating response to azacitidine and nivolumab in relapsed/refractory AML patients. In HMA pretreated patients, ORR was 22% and median OS for the 70 included patients was 6.3 months.13 Gemtuzumab ozogamicin (GO), a humanized antiCD33 antibody conjugate, is currently licensed by both the US Food and Drug Administration and the European Medicines Agency in combination with daunorubicin and cytarabine for the treatment of de novo CD33-positive AML patients. Moreover, available data suggest activity of GO in combination with HMA. The maturation of AML blasts increases CD33 expression after HMA therapy, resulting in an enhanced uptake of GO by blast cells.14 A phase II clinical trial in older AML patients evaluated the combination of hydroxyurea followed by azacitidine for seven days and GO on day eight. Results demonstrated CR in 44% of patients in the good risk group (age 6069 years or performance status 0-1) and 35% (19 of 59 patients) CR rate in the poor risk group (age ≥70 years and performance status 2 or 3).14 In a phase II study in newly diagnosed or relapsed HR-MDS and AML patients, the combination of decitabine with GO achieved CR/CRi in 35% of patients (39 of 110 patients).15 Rigosertib (ON-01910), a multikinase inhibitor, is currently undergoing evaluation in a randomized phase III trial (clinicaltrials.gov identifier: 02562443) in HR-MDS patients after HMA failure. Results of a previous phase III study demonstrated that patients treated with rigosertib had longer (8.6 vs. 5.3 months) median OS compared to patients receiving best supportive care after HMA failure.16 The combination of rigosertib with azacitidine after HMA failure was recently evaluated in a phase II trial, showing an ORR of 54%, including 8% CR in this patient population; the safety profile was similar to those described for azacitidine alone.17 haematologica | 2019; 104(8)

One interesting new therapeutic target in patients failing HMA is the selective inhibition of AXL, a surface membrane protein kinase receptor on blast cells. Signaling through AXL seems to stimulate a number of pro-survival pathways and enables malignant cells to develop resistance to conventional chemotherapies.18 Preclinical studies with bemcentinib, an orally selective small molecule AXL inhibitor, demonstrated in vitro and in mouse models that leukemic proliferation was blocked by interference with AXL signaling.18 Thus, AXL represents a promising target and bemcentinib a possible new treatment option for HR-MDS or AML patients.18 The efficacy and safety of bemcentinib is currently being evaluated in a phase II study (BERGAMO trial; clinicaltrials.gov identifier: 03824080) within the European Myelodysplastic Syndromes Cooperative Group (EMSCO) in patients with HR-MDS or AML after HMA failure. Other potentially available therapeutic approaches after failing HMA include the use of targeted molecular therapies, e.g. with IDH or FLT3-inhibitors. IDH mutations are quite common in MDS (10-15% of patients) and data in relapsed AML have so far proved promising.19 FLT3-inhibitors have already been approved in the US for second-line treatment of patients with AML and may, therefore, offer a therapeutic option in rare FLT3 mutated cases with disease progression.20 In conclusion, patients with HR-MDS or AML failing HMA remain a population with a dismal outcome and limited therapeutic options. In the future, a personalized targeted treatment strategy on the basis of the patient’s molecular profile, cytogenetics, and previous therapies may be the best approach. Until then, translational studies based on a variety of prospective clinical trials are urgently required to overcome the enormous unmet medical need for additional treatment options.

References 1. Platzbecker U. Treatment of MDS. Blood. 2019;133(10):1096-1107. 2. Fenaux P, Mufti GJ, Hellstrom-Lindberg E, et al. Efficacy of azacitidine compared with that of conventional care regimens in the treatment of higher-risk myelodysplastic syndromes: a randomised, open-label, phase III study. Lancet Oncol. 2009;10(3):223-232. 3. Dombret H, Seymour JF, Butrym A, et al. International phase 3 study of azacitidine vs conventional care regimens in older patients with newly diagnosed AML with >30% blasts. Blood. 2015;126(3):291299. 4. Itzykson R, Thepot S, Quesnel B, et al. Prognostic factors for response and overall survival in 282 patients with higher-risk myelodysplastic syndromes treated with azacitidine. Blood. 2011;117(2):403-411. 5. Craddock CF, Houlton AE, Quek LS, et al. Outcome of Azacitidine Therapy in Acute Myeloid Leukemia Is not Improved by Concurrent Vorinostat Therapy but Is Predicted by a Diagnostic Molecular Signature. Clin Cancer Res. 2017;23(21):6430-6440. 6. Komrokji RS. Treatment of Higher-Risk Myelodysplastic Syndromes After Failure of Hypomethylating Agents. Clin Lymphoma Myeloma Leuk. 2015;15 Suppl:S56-59. 7. Sebert M, Renneville A, Bally C, et al. A phase II study of guadecitabine in higher-risk myelodysplastic syndrome and low blast count acute myeloid leukemia after azacitidine failure. Haematologica. 2019, 104(8):1565-1571. 8. Dinardo CD, Pratz KW, Potluri J, et al. Durable response with venetoclax in combination with decitabine or azacitadine in elderly patients with acute myeloid leukemia (AML). J Clin Oncol. 2018;36(15_suppl): 7010-7010. 9. DiNardo CD, Pratz K, Pullarkat V, et al. Venetoclax combined with decitabine or azacitidine in treatment-naive, elderly patients with acute myeloid leukemia. Blood. 2019;133(1):7-17.

1507

Editorials 10. Pollyea DA, Pratz KW, Jonas BA, et al. Venetoclax in Combination with Hypomethylating Agents Induces Rapid, Deep, and Durable Responses in Patients with AML Ineligible for Intensive Therapy. Blood. 2018;132(Suppl 1):285-285. 11. DiNardo CD, Pratz KW, Letai A, et al. Safety and preliminary efficacy of venetoclax with decitabine or azacitidine in elderly patients with previously untreated acute myeloid leukaemia: a non-randomised, open-label, phase 1b study. Lancet Oncol. 2018;19(2):216228. 12. Kubasch AS, Wehner R, Bazzurri S, et al. Clinical, molecular, and immunological responses to pembrolizumab treatment of synchronous melanoma and acute myeloid leukemia. Blood Adv. 2018;2(11):1187-1190. 13. Daver N, Garcia-Manero G, Basu S, et al. Efficacy, Safety, and Biomarkers of Response to Azacitidine and Nivolumab in Relapsed/Refractory Acute Myeloid Leukemia: A Nonrandomized, Open-Label, Phase II Study. Cancer Discov. 2019;9(3):370-383. 14. Nand S, Othus M, Godwin JE, et al. A phase 2 trial of azacitidine and gemtuzumab ozogamicin therapy in older patients with acute myeloid leukemia. Blood, 2013;122(20):3432-3439. 15. Daver N, Kantarjian H, Ravandi F, et al. A phase II study of decitabine and gemtuzumab ozogamicin in newly diagnosed and

16.

17.

18. 19. 20.

relapsed acute myeloid leukemia and high-risk myelodysplastic syndrome. Leukemia. 2016;30(2):268-273. Garcia-Manero G, Fenaux P, Al-Kali A, et al; ONTIME study investigators. Rigosertib versus best supportive care for patients with high- risk myelodysplastic syndromes after failure of hypomethylating drugs (ONTIME): a randomised, controlled, phase 3 trial. Lancet Oncol. 2016;17(4):496-508. Navada SC, Garcia-Manero G, Atallah EL, et al. Phase 2 Expansion Study of Oral Rigosertib Combined with Azacitidine (AZA) in Patients (Pts) with Higher-Risk (HR) Myelodysplastic Syndromes (MDS): Efficacy and Safety Results in HMA Treatment Naïve & Relapsed (Rel)/Refractory (Ref) Patients. Blood. 2018;132(Suppl 1):230. Medyouf H. The microenvironment in human myeloid malignancies: emerging concepts and therapeutic implications. Blood. 2017;129(12): 1617-1626. Stein EM, DiNardo CD, Pollyea DA, et al. Enasidenib in mutant IDH2 relapsed or refractory acute myeloid leukemia. Blood. 2017;130(6):722-731. Cortes J, Perl AE, Döhner H, et al. Quizartinib, an FLT3 inhibitor, as monotherapy in patients with relapsed or refractory acute myeloid leukaemia: an open-label, multicentre, single-arm, phase 2 trial. Lancet Oncol. 2018;19(7):889-903.

Discontinuation of tyrosine kinase inhibitors in patients with chronic myelogeneous leukemia – You can do this at home if you read the instructions Charles A. Schiffer Joseph Dresner Chair for Hematologic Malignancies, Departments of Oncology and Medicine, Wayne State University School of Medicine, Karmanos Cancer Institute, Detroit, MI, USA E-mail: CHARLES A. SCHIFFER - Schiffer@karmanos.org doi:10.3324/haematol.2019.222216

he exciting story of the clinical use of imatinib mesylate for the treatment of leukemias driven by the bcr/abl mutation began in the late 1990s and dramatic effectiveness was immediately apparent in all stages of the diseases. Although there was concern that these benefits might not persist, we now know, after almost twenty years of follow up, that a high proportion of chronic phase patients attain deep molecular responses and enjoy an overall survival comparable to that of agematched controls.1 It was originally expected that lifelong treatment would be needed, but in recent years, trials from around the world have shown that tyrosine kinase inhibitors (TKI) can be successfully discontinued in some patients who have achieved sustained deep molecular responses.2,3 These were conducted as part of clinical trials at CML research institutions by experienced CML clinicians. In this issue of the Journal, Italian clinicians from a wide range of institutions of the Gruppo Italiano Malattie Ematologiche dell'Adulto (GIMEMA) describe a large group of chronic phase patients who had therapy discontinued, many presumably as a consequence of patients’ requests to doctors, who were now comfortable with the accumulated results.4 With a median follow up of 34 months, 60% of patients remained in what has been termed “treatment-free remission” (TFR),5 a result consistent with or perhaps slightly superior to those from earlier trials. As in other trials, the relapse rate was somewhat lower in patients with longer exposures to TKI and all 1508

patients who had molecular relapse were successfully retreated with either their original TKI or were switched to another TKI if their motivation for discontinuation was toxicity; these retreated patients usually achieved the level of their original response. Most CML patients in the US (and to some extent elsewhere) are not followed in specialty hematology centers. This means that the next question in the TKI saga is whether discontinuation can be managed safely by nonspecialist oncologists. The process is not very difficult to understand and there are few risks if patients are selected and followed appropriately. The criteria for study entry and monitoring differed somewhat amongst the published trials, but a consensus approximation would include: - TKI treatment for a minimum of three years; - continuous deep molecular response [minimum MR4 (BCR-ABL1 ≤0.01% using the International Scale, IS)] on multiple testing for at least two years. Some studies required reduction to < MR4.5, although outcomes seem comparable (including the results from this GIMEMA experience) using either molecular cutoff; - use of a quantitative polymerase chain reaction (qPCR) test sensitive to a level of at least MR4.5 in a laboratory with a rapid turn-around time; - monitoring of peripheral blood transcripts every 4-6 weeks for 6-8 months, then bimonthly for approximately one year followed by every three months thereafter for a minimal follow up of three years; haematologica | 2019; 104(8)

Editorials

“Relapse” is defined by loss of MR3 [major molecular response (MMR)], and it is essential to be aware that values can sometimes fluctuate between MR3 and MR4, in part because of the variability of the assay; therefore, at least two values with loss of MMR should be documented before therapy is resumed. This is perhaps the part of the process with the most subtleties, and consultation with CML specialists is sometimes advisable. The pattern of relapse raises interesting issues about CML biology. Remarkably, despite the inclusion of patients with continuously undetectable transcripts for many years using very sensitive techniques, molecular relapses can be detected within the first 1-2 months of discontinuation. Virtually all relapses occur within the first 6-8 months of cessation, with very few emerging with long-term follow up which is now, in many studies, over five years. The rapidity of relapse in patients attests to the resilience of dormant CML progenitors which are capable of re-emerging almost immediately after the suppressive pressure of the TKI is released; it is a humbling reminder of the difficulties to be faced in eliminating stem cells in other leukemias. Perhaps even more fascinating is the observation of prolonged remissions lasting for many years. While it is theoretically possible that CML “stem” cells have been eliminated (Figure 1), this would seem unlikely. I am not aware of any studies of bone marrows from patients in

long-term TFR evaluating whether bcr/abl positive colonies can be grown in vitro. Interest has been shown in the possibility of immune suppression of remaining bcr/abl precursors, with some focus on the role of T-natural killer (NK) cells,6,7 perhaps stimulated by observations of a possible salutary effect of proliferation of NK cells after dasatinib treatment.8,9 Results of these studies have been, at best, inconclusive. Nonetheless, this remains an interesting hypothesis. I have seen two patients in apparent TFR who experienced molecular relapse after 1-2 years of follow up: the first after chemotherapy for another cancer and the other after prolonged use of corticosteroids for treatment of the TKI “withdrawal” syndrome. Could the relapses have been related to “immunosuppression” from these other treatments? This is speculative at best, but it would be interesting to see if other such patients are identified. Changes in the marrow microenvironment might also play a role in either the continued containment of growth or, alternatively, promote rapid recurrence. It is also not clear whether late relapses will develop with longer follow up. As illustrated in Figure 1, it is possible that successful TKI therapy reduces the number of CML precursors to the levels found in ‘preclinical’ CML. Little is known about the duration of the ‘incubation’ period after the initial mutagenic event or how long it takes for CML to become clinically identifiable. Perhaps

Figure 1. Chronic myeloid leukemia (CML) has a preclinical phase of unknown duration and usually is diagnosed with obvious “bulk” myeloid proliferation. After tyrosine kinase inhibitor (TKI) treatment, the cellular burden is markedly reduced to below the level of molecular detection (horizontal line). It is possible that the CML precursors are entirely eliminated (X) or tiny amounts of residual disease persist which have the potential to become clinically apparent after a long period of time, analogous to what occurs in the original presentation of the disease.

haematologica | 2019; 104(8)

1509

Editorials

the most relevant data come from observations after the atomic bomb events in Japan where the incidence of CML peaked at a median of ten years but continued at an increased rate for years thereafter.10 These findings have implications about the frequency or even the necessity of long-term PCR monitoring of patients in long-term remission. I have adopted a non-data-driven approach and continue testing approximately every six months after three years of undetectable transcripts. However, we urgently need further information about this. It is also important to appreciate that only a minority of chronic phase patients can achieve long-term TFR. Most treatment trials describe the rates of molecular response using cumulative incidence analyses, meaning that a patient achieved that level of response at least once. It is more difficult, however, to identify the rates of sustained response, a requirement for considering stopping. In the imatinib-based German CML IV and IRIS trials (the two largest studies with long-term follow up), the rate of MR 4.5 was approximately 50% at five years and >60% at ten years, respectively, although these estimates were not based on the “intent to treat” population11,12 and only included patients for whom data were available at these time points. Using a somewhat generous estimate of 40% sustained MR4.5 in newly diagnosed chronic phase patients, and a relapse rate of 50%, only approximately 20% of patients will successfully achieve TFR. Calculations may differ somewhat using continuous MR4 as the eligibility cutoff or if patients were treated initially with second-generation TKI which produce higher response rates. Nonetheless, the reality is that the large majority of patients will require life-long treatment, and even those who stop successfully would have required many years of treatment prior to a trial of cessation. Therefore, optimal CML treatment will continue to depend on the skills of physicians familiar with ameliorating the side effects of therapy and health systems that deal more effectively with the costs of this chronic treatment. And this raises the question of whether all patients require the “standard” dose to maintain response or if many of the benefits of stopping, such as reduced side effects and costs, can be achieved with lower doses. Again, data are fragmentary, but many clinical trials and observational studies report that a significant proportion of patients are maintained long term on lower than the initial “standard” doses of TKI.13,14 A recent pilot trial in newly diagnosed patients demonstrated what appear to be identical response rates with less toxicity, using 50 mg of dasatinib versus the standard 100 mg dose.15 Indeed, based on these data, I have been decreasing the dasatinib dose to 50 mg in patients with stable high-grade responses on 100 mg with no evidence of loss of benefit in approximately a dozen patients.16 Perhaps the most systematic data come from the UK DESTINY trial in which patients eligible for consideration for a TFR trial had their doses reduced by 50% for one year before drug discontinuation.17 Two per cent of patients who entered the trial with levels of MR4 lost MMR during the first year, while only 18% of those who began with sustained MR3 lost MMR within the year. It is not known how patients would have fared long term 1510

on the lower dose since they went on to the TFR portion of the study. Those who “relapsed” within the first year had therapy restarted, but it is possible that that may not have always been necessary. In fact, a recent modeling exercise using data from large clinical trials suggests that the rise in transcript numbers after dose reduction can be transient in many patients, and that MMR response might have recovered without increasing the dose.18 It is, therefore, clear that a substantial number of patients can do well with lower doses of TKI, but prospective trials addressing this question would be welcomed. To conclude, the report from the GIMEMA group confirms that TFR can be achieved in routine clinical practice and indicates that discontinuation be considered in appropriately selected patients outside the clinical trial setting.4 The relapse rate has been consistently in the 50% range in all trials, and future research should focus on the mechanisms by which recurrence is suppressed in the hope that new approaches, possibly immunomodulatory, can improve these results. In addition, patients should continue to be monitored to assess whether very late relapses develop.

References 1. Bower H, Björkholm M, Dickman PW, et al. Life expectancy of patients with chronic myeloid leukemia approaches the life expectancy of the general population. J Clin Oncol. 2016;34(24):2851-2857. 2. Etienne G, Guilhot J, Rea D, Rigal-Huguet F, et al. Long-Term Follow-Up of the French Stop Imatinib (STIM1) Study in Patients With Chronic Myeloid Leukemia. J Clin Oncol. 2017;35(3):298-305. 3. Saussele S, Richter J, Guilhot J, et al. Discontinuation of tyrosine kinase inhibitor therapy in chronic myeloid leukaemia (EURO-SKI): a prespecified interim analysis of a prospective, multicentre, nonrandomised, trial. Lancet Oncol. 2018;19(6):747-757. 4. Fava C, Rege-Cambrin G, Dogliotti I, et al. Observational Study of CML Chronic Myeloid Leukemia Italian patients who discontinued Tyrosine Kinase Inhibitors in clinical practice. Haematologica. 2019; 104(8):1589-1596. 5. Hughes TP, Ross DM. Moving treatment-free remission into mainstream clinical practice in CML. Blood. 2016;128(1):17-23. 6. Ilander M, Olsson-Strömberg U, Schlums H, et al. Increased proportion of mature NK cells is associated with successful imatinib discontinuation in chronic myeloid leukemia. Leukemia. 2017;31(5):11081116. 7. Réa D, Henry G, Khaznadar Z, et al. Natural killer-cell counts are associated with molecular relapse-free survival after imatinib discontinuation in chronic myeloid leukemia: the IMMUNOSTIM study. Haematologica. 2017;102(8):1368-1377. 8. Mustjoki S1, Ekblom M, Arstila TP, et al. Clonal expansion of T/NKcells during tyrosine kinase inhibitor dasatinib therapy. Leukemia. 2009;23(8):1398-1405. 9. Schiffer CA, Cortes J, Hochhaus A, et al. Lymphocytosis following treatment with dasatinib in chronic myeloid leukemia: effects on response and toxicity. Cancer. 2016;122(9):1398-1407. 10. Hsu WL, Preston DL, Soda M, et al. The Incidence of Leukemia, Lymphoma and Multiple Myeloma among Atomic Bomb Survivors: 1950–2001. Radiat Res. 2013;179(3):361-382. 11. Hehlmann R, Lauseker M, Saußele S, et al. Assessment of imatinib as first-line treatment of chronic myeloid leukemia: 10-year survival results of the randomized CML study IV and impact of non-CML determinants. Leukemia. 2017;31(11):2398-2406. 12. Hochhaus A, Larson RA, Guilhot F, et al. Long-Term Outcomes of Imatinib Treatment for Chronic Myeloid Leukemia. N Engl J Med. 2017;376(10):917-927. 13. Faber E, Divoká M, Skoumalová I, et al. A lower dosage of imatinib is sufficient to maintain undetectable disease in patients with chronic myeloid leukemia with long-term low-grade toxicity of the treatment. Leuk Lymphoma. 2016;57(2):370-375. 14. Visani G, Breccia M, Gozzini A, et al. Dasatinib, even at low doses, is an effective second-line therapy for chronic myeloid leukemia

haematologica | 2019; 104(8)

Editorials

patients resistant or intolerant to imatinib. Results from a real lifebased Italian multicenter retrospective study on 114 patients. Am J Hematol. 2010;85(12):960-963. 15. Clark RE, Polydoros F, Apperley JF, et al. De-escalation of tyrosine kinase inhibitor dose in patients with chronic myeloid leukaemia with stable major molecular response (DESTINY): an interim analysis of a non-randomised, phase 2 trial. Lancet Haematol. 2017;4(7):e310-e316. 16. Naqvi K, Jabbour E, Skinner J, et al. Early results of lower dose dasa-

tinib (50 mg daily) as frontline therapy for newly diagnosed chronicphase chronic myeloid leukemia. Cancer. 2018;124(13):2740-2747. 17. Schiffer CA. The evolution of dasatinib dosage over the years and its relevance to other anticancer medications. Cancer. 2018;124(13): 2687-2689. 18. Fassoni AC, Baldow C, Roeder I, Glauche I. Reduced tyrosine kinase inhibitor dose is predicted to be as effective as standard dose in chronic myeloid leukemia: a simulation study based on phase III trial data. Haematologica. 2018;103(11):1825-1834.

The only thing that stops a bad microbiome, is a good microbiome Jessica R. Galloway-Peña1,2 and Robert R. Jenq1,3 1

Department of Genomic Medicine; 2Department of Infectious Diseases, Infection Control and Employee Health and 3Department of Stem Cell Transplantation, The University of Texas MD Anderson Cancer Center, Houston, TX, USA. E-mail: jrgalloway@mdanderson.org/rrjenq@mdanderson.org doi:10.3324/haematol.2019.222430

ultidrug resistant (MDR) bacterial colonization in the gut is frequently induced by excessive use of antibiotics.1 Fecal microbiota transplantation (FMT) has been shown to be quite successful in treating refractory and recurrent Clostridium difficile infection.2 Thus, current research is focusing on how FMT may also help in decolonizing MDR organisms (MDRO) and in preventing recurrent MDR infections.3 Decolonization of MDRO via FMT may be particularly useful in patients with hematologic malignancies, such as those undergoing hematopoietic stem cell transplantation (HSCT),4 as use of chemotherapeutic agents and frequent administration of antibiotics can favor the selection of resistant pathogens.5,6 In spite of the increasing evidence that the feasibility and safety of FMT in immunocompromised cohorts is comparable to that of immunocompetent patients, administering FMT in the setting of hematologic malignancy remains a cause for concern due to perceived risks of translocation and sepsis.7,8 Given the growing body of literature associating a dysbiotic microbiome with adverse HSCT outcomes and treatment-related toxicities, including infection, delivering a diverse microbiome via FMT to immunocompromised patients may provide a variety of benefits, such as promoting colonization resistance and reducing the risk of bacterial translocation.9 Thus, attempts to better characterize the safety and efficacy of FMT in these patients are merited. In this issue of the Journal, Battipaglia et al.10 describe a retrospective case series of 10 patients with hematologic malignancies undergoing FMT for MDRO colonization before or after allogeneic HSCT. In this study, the authors show both the safety and efficacy of using FMT for decolonization of carbapenem-resistant Enterobactericeae (CRE), carbapenem-resistant Pseudomonas (CRP), and vancomycin-resistant Enterococcus (VRE). Notably, the study reports FMT both pre- and post transplant. The majority of patients who received FMT prior to transplant did not have recurrent MDRO even after HSCT, indicating the prophylactic use of FMT. Interestingly, the procedure remained effective for longhaematologica | 2019; 104(8)

term MDRO decolonization in the majority of patients despite the use of broad-spectrum antibiotics in some of the patients after FMT. This implies that FMT can potentially achieve full decolonization of MDRO rather than merely reducing the levels of MDRO below the limit of detection. While FMT was shown to be successful in decolonizing the MDRO studied, the FMT did not always prevent additional post-transplant infections from other bacteria susceptible to antibiotics. Interestingly, only 50% of patients concomitantly colonized with extended spectrum β-lactamase (ESBL)-producing Enterobacteriaceae obtained decolonization. This is reminiscent of a case report by Stalenhoef et al. where FMT successfully eradicated a Pseudomonas aeruginosa urinary tract infection, while stool cultures remained positive for extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae three months after FMT.11 Thus, the higher frequency of failure of FMT to eradicate the ESBL-producing Enterobacteriaceae in these two studies may suggest potential limitations to this therapy. Although the specific mechanisms underlying the success of FMT for MDRO colonization remain unclear, Figure 1 depicts an overview of the general concepts regarding the use of FMT for MDRO in patients with hematologic malignancies. Given that this study looked specifically at CRE, CRP, and VRE, it remains unclear if other MDRO may be equally responsive to FMT. Furthermore, given the seemingly discrepant results for CRE, CRP, and VRE compared to ESBL-producing organisms, one might consider that distinct mechanisms of action underlie how FMT mediates response for different MDRO. Due to the retrospective nature of the study, in contrast to a controlled study, it is unclear how physicians decided to treat each patient with FMT case by case. Moreover, there was a large variation between cases in the time of FMT relative to HSCT and the MDRO colonization/infection in both pre- and post-HSCT groups. Thus, it remains to be determined what the ideal timeframe for FMT is in both scenarios. The use of relat1511

Editorials

Figure 1. Fecal microbiota transplant (FMT) as a potential decolonization strategy for multi-drug resistant organisms (MDRO) in hematologic malignancy patients. After cytotoxic chemotherapy, hematopoietic stem cell transplantation (HSCT), and antibiotic administration, the likelihood that resistant bacteria replace diverse microbiota increases. FMT can potentially restore susceptibility to antimicrobials by replacing resistant bacteria with a diverse microflora when used as a decolonization strategy in response to infection or colonization with MDRO. When administered prior to cancer treatment, FMT may potentially mitigate dysbiosis and selection of resistance.

ed donors was preferred as it was perceived that common environmental exposures would reduce additional risk of transferring infectious agents between the donor and recipient. Intriguingly, neither of the two cases using an unrelated donor was successfully decolonized; given so few numbers, we cannot determine if this result is significant. Consequently, the best choice of donor remains to be explored. If it were to be found that related donors are in fact preferable to universal donors, appropriate screening and regulations will be an important consideration in the future. One critical piece missing from this study is the understanding of the microbiome in this process. Although theoretically FMT decolonization works via restoration of microbial diversity leading to colonization resistance and displacement of the MDRO,12,13 experimentally showing what micro-organisms were important for decolonization in each case, which organisms presented robust and durable colonization, as well as if resistance genes were completely displaced after FMT would strengthen these types of studies and vastly improve the understanding of the mechanism by which FMT decolonizes MDRO. This represents an important future opportunity for investigators. With MDR infections set to be the worldâ&#x20AC;&#x2122;s leading cause of morbidity and mortality by the year 2050, set to 1512

surpass even cancer, and with few new antimicrobials in the pipeline, the need for novel and different approaches to treat MDR infections is critical.14 Moreover, we need to improve our clinical understanding of the antibiotic resistome, particularly in immunocompromised patients who experience repeated exposures to antimicrobials. Although no large randomized controlled trials have been performed to study the efficacy and safety of FMT for MDR organisms in the immunocompromised patient, this study and others have provided some promising evidence, and suggest that a non-antibiotic therapy for MDR colonization and infections may become common practice in the future for patients with hematologic malignancies.15-17 The use of FMT as a decolonization agent both as a prophylactic and treatment measure may prove effective in preventing MDR outbreaks and transmission, prolonged in-hospital care, recurrent infection, and improve the overall outcomes of HSCT patients. Given the numerous potential benefits, and demonstrated safety and efficacy, the fear and negative perception of FMT in the cancer setting is unjustified.15 Acknowledgments JGP is supported by the NIH (K01 AI143881). RRJ is supported by the NIH (R01 HL124112) and the Cancer Prevention and Research Institute of Texas (RR160089). haematologica | 2019; 104(8)

Editorials

References 1. Modi SR, Collins JJ, Relman DA. Antibiotics and the gut microbiota. J Clin Invest. 2014; 124(10):4212-4218. 2. Quraishi MN, Widlak M, Bhala N, et al. Systematic review with meta-analysis: the efficacy of faecal microbiota transplantation for the treatment of recurrent and refractory Clostridium difficile infection. Aliment Pharmacol Ther. 2017;46(5):479-493. 3. Saha S, Tariq R, Tosh PK, et al. Faecal microbiota transplantation for eradicating carriage of multidrug-resistant organisms: a systematic review. Clin Microbiol Infect. 2019;25(8):958-963. 4. Bilinski J, Grzesiowski P, Sorensen N, et al. Fecal microbiota transplantation in patients with blood disorders inhibits gut colonization with antibiotic-resistant bacteria: results of a prospective, single-center study. Clin Infect Dis. 2017;65(3):364-370. 5. Galloway-Pena J, Brumlow C, Shelburne S. Impact of the microbiota on bacterial infections during cancer treatment. Trends Microbiol. 2017;25(12):992-1004. 6. Montassier E, Gastinne T, Vangay P, et al. Chemotherapy-driven dysbiosis in the intestinal microbiome. Aliment Pharmacol Ther. 2015;42(5):515-528. 7. Shogbesan O, Poudel DR, Victor S, et al. A systematic review of the efficacy and safety of fecal microbiota transplant for Clostridium difficile infection in immunocompromised patients. Can J Gastroenterol Hepatol. 2018;2018:1394379. 8. Wang S, Xu M, Wang W, et al. Systematic review: adverse events of fecal microbiota transplantation. PLoS One. 2016;11(8):e0161174. 9. Shono Y, van den Brink MRM. Gut microbiota injury in allogeneic haematopoietic stem cell transplantation. Nat Rev Cancer.

2018;18(5):283-295. 10. Battipaglia G, Malard F, Rubio MT, et al. Fecal microbiota transplantation before or after allogeneic hematopoietic transplantation in patients with hematologic malignancies carrying multidrug-resistance bacteria. Haematologica 2019;104(8):1682-1688. 11. Stalenhoef JE, Terveer EM, Knetsch CW, et al. Fecal microbiota transfer for multidrug-resistant gram-negatives: a clinical success combined with microbiological failure. Open Forum Infect Dis. 2017;4(2):ofx047. 12. Kelly CR, Kahn S, Kashyap P, et al. Update on fecal microbiota transplantation 2015: indications, methodologies, mechanisms, and outlook. Gastroenterology. 2015;149(1):223-237. 13. Khoruts A, Sadowsky MJ. Understanding the mechanisms of faecal microbiota transplantation. Nat Rev Gastroenterol Hepatol. 2016;13(9):508-516. 14. O'Neill J. Review on antimicrobial resistance antimicrobial resistance: tackling a crisis for the health and wealth of nations. Review on Antimicrobial Resistance. London, 2014. Available from: http://amr-review.org 15. Abu-Sbeih H, Ali FS, Wang Y. Clinical review on the utility of fecal microbiota transplantation in immunocompromised patients. Curr Gastroenterol Rep. 2019;21(4):8. 16. Wardill HR, Secombe KR, Bryant RV, et al. Adjunctive fecal microbiota transplantation in supportive oncology: Emerging indications and considerations in immunocompromised patients. EBioMedicine. 2019;44: 730-740. 17. DeFilipp Z, Hohmann E, Jenq RR, et al. Fecal microbiota transplantation: restoring the injured microbiome after allogeneic hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2019;25(1): e17-e22.

Predicting risk for recurrence of arterial ischemic stroke in children: thrombophilia as another piece of the puzzle Ghada Aborkhees and Lesley Gayle Mitchell Department of Pediatrics, University of Alberta, Edmonton, AB, Canada E-mail: LESLEY GAYLE MITCHELL - Lesley.Mitchell@albertahealthservices.ca doi:10.3324/haematol.2019.222695

ecurrent arterial ischemic stroke (AIS) is increasingly recognized as a significant cause of mortality and morbidity in the pediatric population. Identifying risk factors for recurrent AIS is essential for developing strategies for secondary stroke prevention. While multiple risk factors have been identified for AIS events, the only confirmed risk factor for initial AIS recurrence is the presence of vasculopathy, particularly moyamoya disease.1-3 In a meta-analysis, prothrombotic risk factors were found to be associated with AIS in pediatric patients.4 However, the role of thrombophilia as an independent risk factor for recurrent AIS has not been established due to a paucity of research in the area and the lack of statistical power in the published studies. In the current edition of Haematologica, deVeber et al. report on an international prospective cohort study which recruited 894 pediatric patients from centers in Germany, Canada and the UK.5 The primary objective of the study was to determine the association of prothrombotic risk factors and/or underlying stroke subtypes with risk for recurrent stroke. The authors excluded asymptomatic strokes and transient ischemic attacks due to the difference in underlying disease as well as the differing outcomes from symptomatic strokes. Sickle cell disease and moyamoya vasculopathies were also excluded as their haematologica | 2019; 104(8)

recurrence rates and risk factors differ from those of other subtypes of pediatric AIS. The authors report an overall AIS recurrence rate of 17.9% in the cohort studied. The study confirmed the association of vasculopathy as a risk factor for AIS recurrence. The novel approach in the current study was the examination of the role of thrombophilia as an independent risk factor for AIS recurrence. Study patients were excluded if they had thrombophilic markers with established pathophysiological relevance such as homozygous protein C and homozygous antithrombin deficiency. Analysis of the study data showed that the following were independent risk factors for recurrence: antithrombin deficiency (hazard ratio 3.9; 95% confidence interval: 1.4-10.9), increased lipoprotein(a) (hazard ratio 2.3; 95% confidence interval: 1.3-4.1) and more than one prothrombotic marker (hazard ratio 1.9; 95% confidence interval: 1.1-3.2). The results obtained from this study highlight the importance of screening AIS cases for thrombophilia in order to identify the children at risk of AIS recurrence. The reported study is a valuable addition to the previous efforts to identify the risk factors for AIS recurrence. There are significant strengths of the study design. The first was the relatively large sample size, which provided adequate power to determine the association of pro1513

Editorials

thrombotic markers with recurrence. Second, recruiting patients by collaboration of investigators from three countries and including multiple sites supports the generalizability of the results. As the study sites were large tertiary or quaternary centers, another key element of the design was the exclusion of referrals from outside the catchment area, which minimizes referral bias. One study showed that there is a higher rate of AIS recurrence among referrals, providing evidence of differences in the populations of patients which would affect generalizability of the results.3 Furthermore, prothrombotic marker testing was performed at each site, which demonstrates the ability to determine these markers in different clinical laboratories. Finally, there was a clear clinical and radiologically combined definition of initial AIS and recurrent stroke events. There are a few limitations to the study. The first is combining all three clinical entities for determination of association of prothrombotic markers with recurrence risk. While this could be considered a strength, as the results are generalizable to the pediatric population with AIS, there is a missed opportunity to determine markers specific to each clinical population. As the mechanisms for AIS and AIS recurrence likely differ based on the underlying disorder, it is reasonable to predict that the markers will vary by diagnosis. Another important limitation is the extended study period from 1990 to 2016. While this allowed the enrollment of a large number of patients with incident AIS, over time there are variations in clinical practice including index case diagnosis and interventions. These factors could influence patientsâ&#x20AC;&#x2122; outcomes as well as characterization of the study populations, which affects the generalizability of the study results. However, even with the limitations, the results of the study are valid and are a valuable contribution to this area of research. In pediatric AIS studies international collaboration is essential to assemble adequately powered cohorts for determination of predictive markers for AIS recurrence. In futures studies, with recent publications in the area including the current paper, there are now avenues for determin-

1514

ing predictive models which include clinical variables and biomarkers. For example, the Childhood AIS Standardized Classification and Diagnostic Evaluation (CASCADE) classification is a consensus-based standardized tool for classifying arteriopathic and non-arteriopathic AIS. Patients with arteriopathy classified as CASCADE 2 (unilateral focal cerebral arteriopathy) and 3 (bilateral cerebral arteriopathy) have an increased risk of AIS recurrence as do those classified as CASCADE 5 (patients with cardio-embolism).6 A recent publication by Fullerton et al. identified inflammatory markers associated with risk of recurrence in arteriopathic patients.7 Therefore, a model could be based on a combination of clinical variables and both thrombophilia and inflammatory markers. The current publication by deVeber et al. will help to shape future studies determining predictive models. These predictive models will allow secondary prophylaxis interventions to be targeted to only those children at risk of recurrence, ultimately improving the care of children who have had an AIS.

References 1. Strater R, Becker S, von EA, et al. Prospective assessment of risk factors for recurrent stroke during childhood--a 5-year follow-up study. Lancet. 2002;360(9345):1540-1545. 2. Ganesan V, Prengler M, Wade A, Kirkham FJ. Clinical and radiological recurrence after childhood arterial ischemic stroke. Circulation. 2006;114(20):2170-2177. 3. Stacey A, Toolis C, Ganesan V. Rates and risk fctors for arterial ischemic stroke recurrence in children. Stroke. 2018;49(4):842-847. 4. Kenet G, Lutkhoff LK, Albisetti M, et al. Impact of thrombophilia on risk of arterial ischemic stroke or cerebral sinovenous thrombosis in neonates and children: a systematic review and meta-analysis of observational studies. Circulation. 2010;121(16):1838-1847. 5. deVeber G, Kirkham F, Shannon K, et al. Recurrent stroke: the role of thrombophilia in a large international pediatric stroke population. Haematologica. 2019;104(8):1676-1681. 6. Bohmer M, Niederstadt T, Heindel W, et al. Impact of childhood arterial ischemic stroke standardized classification and diagnostic evaluation classification on further course of arteriopathy and recurrence of childhood stroke. Stroke. 2018 Dec 7. [Epub ahead of print] 7. Fullerton HJ, deVeber GA, Hills NK, et al. Inflammatory biomarkers in childhood arterial ischemic stroke: correlates of stroke cause and recurrence. Stroke. 2016;47(9):2221-2228.

haematologica | 2019; 104(8)

PERSPECTIVE ARTICLE

“Somatic” and “pathogenic” - is the classification strategy applicable in times of large-scale sequencing?

Ferrata Storti Foundation

Constance Baer, Wencke Walter, Stephan Hutter, Sven Twardziok, Manja Meggendorfer, Wolfgang Kern, Torsten Haferlach and Claudia Haferlach MLL Munich Leukemia Laboratory, Munich, Germany

Introduction In the early days of sequencing only a small number of bases was evaluated because of the labor-intensive nature of the procedure. Genes were identified to play a role in the pathogenesis of neoplasms in animal models and cell lines. Subsequently, these mutations were analyzed in samples from patients and their impact on prognosis was evaluated. The list of examples is long: e.g. TP53 was found to be universally mutated across cancers,1 and NPM1 is now among the most frequently analyzed genes in acute myeloid leukemia2 and the mutation defines its own acute myeloid leukemia subtype in the current World Health Organization classification.3 High-throughput sequencing has changed the landscape. It is now possible to sequence a huge number of genes up to exomes and even whole genomes in a comparably short time at affordable cost. The challenge is no longer the sequencing, but rather the evaluation of the results and the interpretation of their impact on diagnosis, prognosis or therapeutic decisions. This has led to some major changes in the way we view sequencing data. In 2015, the American College of Medical Genetics and Genomics (ACMG) and the Association for Molecular Pathology (AMP) recommended changing the terms mutation and polymorphism to “variant”. Variants are then further subdivided into five categories depending on the likelihood of their association with the disease.4 The definition was designed for hereditary diseases and therefore addresses germline variants. The vast majority of genetic events in cancer are somatic.5 Acquired variants represent potential drug targets or biomarkers. Testing a sample, e.g. of colon cancer, for mutations is frequently performed by comparing variants from a biopsy to those in leukocytes as reference to identify only somatic variants (tumor-normal comparison).6,7 The classical tumor-normal workflow is challenging in studies with large or historic cohorts because of additional sequencing costs, or limited availability of reference material. Leukemia represents another challenge, since blood cannot be used as the reference material. In addition, the growing knowledge of genetic complexity and tumor heterogeneity challenges the historic binary variant classifications (mutation, polymorphism) in the somatic field as well (Figure 1).

Today Ideally, the results of tumor sequencing are compared to those of reference material with an unaltered germline sequence. Clonal hematopoiesis of indeterminate potential has made us familiar with the idea that mutations are acquired as part of the aging process.8 Blood cells are strongly affected by the continuous accumulation of somatic changes as a consequence of lifelong proliferation,9 but the phenomenon could apply to all types of reference material.10,11 Tissue formed of cells that divide less quickly (e.g. cerebral tissue)12 would be preferred as a reference; however, this is not a practical approach for routine analysis. Easily accessible sources of reference material are hair follicles, nails, urine, T cells, fibroblast cultures, buccal swabs, saliva, and skin biopsies, but poor DNA yield and the presence of leukemic cell contamination (e.g. DNA from blood in nails) are potential challenges to the use of such material.13-15 In the absence of available reference material, the variant allele frequency (VAF) can be used to distinguish germline from somatic variants. A germline variant is present with a 50% (heterozygous) or 100% (homozygous) VAF. An acquired variant is usually present with a lower VAF, because it is not present in all cells. The caveat is that other factors can also contribute to VAF. Firstly, technical issues haematologica | 2019; 104(8)

Haematologica 2019 Volume 104(8):1515-1520

Correspondence: TORSTEN HAFERLACH torsten.haferlach@mll.com Received: February 10, 2019. Accepted: May 15, 2019. Pre-published: July 4, 2019. doi:10.3324/haematol.2019.218917 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1515 ©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.

1515

C. Baer et al.

(polymerase chain reaction/amplification bias) can contribute to skewed VAF. Secondly, somatic mutations can also occur with a VAF of 50% if the number of malignant cells in the analyzed sample is high. Lastly, genetic features influence the VAF. A deletion of 17p, would cause all germline variants in the deleted locus (foremost TP53) to appear with VAF not around 50%. Copy number gains and copy neutral loss of heterozygosity also influence VAF. Databases are frequently used for variant interpretation. Curated databases for variants meet high quality standards, but contain only a small number of variants.16 Databases with large numbers of contributors, such as dbSNP, contain more variants. Initially, many variants were classified as benign germline variants if they were listed in the dbSNP.17 All databases, which allow individual submitters to add data, are error prone. The diversity of the exome was impressively demonstrated by Lek et al., who analyzed 60,706 individuals of different ethnicities and found on average one variant for every eight bases of the exome. Data were collected by the Exome Aggregation Consortium (ExAC), which included a col-

lection of whole exome sequencing data from a broad range of studies. ExAC has recently been extended to genomes (gnomAD).18 As of today, gnomAD contains data from 141,456 individuals. This allows for a representative overview of the population and ethnic groups. As such, it can be used to exclude frequent variants in the population from candidate disease-associated aberrations. However, it must be kept in mind that the metadata from such a collection includes many blood samples and even individuals with an (undiagnosed) hematologic disease. The V617F mutation in the JAK2 gene, which is found in the majority of patients with myeloproliferative neoplasms, is also found in 0.04% of individuals in the gnomAD dataset, so variants in gnomAD cannot be classified per se as benign. Considering the difficulties, it can be questioned whether the strict separation of somatic and germline is necessary. There are a number of reasons why they should be distinguished: (i) only somatic changes prove clonal outgrowth in the case of clonal hematopoiesis of indeterminate potential and clonal cytopenia of undetermined significance8 and they are used for the calculation

Figure 1. Clinical questions for current variant classifications. The complexity of variant classification challenges the terms “mutation” and “polymorphism”, which were clearly associated with clinical relevance in the past. The classification, interpretation and consequences of each variant depend on the context. The identification of somatic changes proves clonality, and somatic aberrations qualify as markers of minimal/measurable residual disease. Other clinical issues are the availability of targeted therapies for mutations and the World Health Organization classification, which genetically defines certain entities. Finally, inherited variants are not only discussed in the context of familial predisposition, but are also relevant if family members are considered as stem cell donors. AlloSCT: allogeneic stem cell transplantation; MRD: minimal/measurable residual disease; WHO: World Health Organization.

1516

haematologica | 2019; 104(8)

Variant classification in the sequencing era

of tumor mutation burden;19 (ii) the presence of a genetic aberration is frequently used as a marker for monitoring minimal (measurable) residual disease (MRD).20 A germline variant would not be eradicated by treatment (with the exception of allogeneic bone marrow transplantation), and is therefore not informative as a MRD marker; (iii) germline variants can predispose to or cause cancer and other diseases. Tarailo-Graovac et al. found that 2.8% of individuals from the ExAC database have variants which are implicated in a wide variety of Mendelian disorders.21 The 2017 World Health Organization classification now also recognizes neoplasms with germline predisposition and mutations in ETV6, RUNX1 and other genes.3 Knowledge of pathogenic germline variants is of importance for family counseling and to determine whether family members can be considered as stem cell donors. Germline variants are important modulators of a patient’s outcome and treatment response. An aberration in the DNA damage repair pathway could increase the risk of therapy-related myeloid neoplasms.22 An impressive example that the variant itself is of greater importance than its origin is the response to treatment with olaparib. Patients respond if they have either a germline or acquired variant of BRCA1/2.23 It is therefore essential to know whether the mutation is pathogenic or actionable. The definition of pathogenic or actionable is not trivial and is highly context-dependent. For inherited diseases, a pathogenic variant is usually understood as being causal. A clear variant-phenotype relationship has

been recognized in a few cases (e.g. cystic fibrosis), but for many other disease types such a relationship is more elusive. The five-tier system for the classification of hereditary variants currently recommended by the ACMG and AMP also recognizes the categories of “likely pathogenic” and “likely benign”.4 Here, we focus on variants in cancer. When translating genetic findings into the clinic, a variant might have a different value depending on the immediate question at hand. Sukhai et al. proposed the term “actionable” to describe variants which affect patients’ management.24 The 2017 guidelines for variants in cancer from the AMP, American Society of Clinical Oncology (ASCO), and the College of American Pathologists (CAP) suggest a fourtier system: tier I, strong clinical significance; tier II, potential clinical significance; tier III, unknown clinical significance; and tier IV, a benign or likely benign variant (Figure 2).25 The interpretation should be further subdivided into the categories “diagnostic”, “prognostic” and “predictive”. It is suggested that the four-tier system is applied to each category. The highest level of evidence is given to biomarkers that predict response or resistance to Food and Drug Administration-approved therapies and to variants from professional diagnostic and prognostic guidelines.25 The BRAF V600E mutation is such an example, as its presence allows vemurafenib treatment.26 At first glance, a tiered system seems to be a well-standardized approach, but each case presents its own challenges. One example would be a patient with cytopenia

Figure 2. Decision tree for variant classification and clinical-decision making. Sequencing data are used to answer different clinical questions. Here we separate the issue of germline versus somatic and biological functions (e.g. pathogenic vs. benign). The questions are closely related in everyday life, however the sources and evidence supporting decisions are different (examples are outlined here). The definitions of the clinical significance of the four-tier classification system are from the guidelines by Li et al.25 WES: whole-exome sequencing; WGS: whole-genome sequencing; VAF: variant allele frequency; FDA: Food and Drug Administration; SCT: stem cell transplantation; MRD: minimal/measurable residual disease.

haematologica | 2019; 104(8)

1517

C. Baer et al.

and an acquired variant in the TET2 gene. The variant could certainly prove clonality, which is essential for diagnosing clonal cytopenia of undetermined significance. The variant would therefore be tier I or tier II in the diagnostic classification. However, if the same variant is found in an patient with acute myeloid leukemia, for whom the therapeutic procedure remains to be defined, the classification becomes more complicated. Currently, no specific therapy is available for TET2-positive malignancies. The variant could therefore end up in tier III if a strict interpretation of the guidelines is used. Hematologic diseases are closely related and mutations are generally typical, but not exclusive to one disease. For example, the L265P variant in MYD88 is found in 90% of all patients with Waldenström macroglobulinemia, but is also present, at a lower frequency, in patients with other B-cell neoplasms.27 Consequently, no variant has the sensitivity or specificity to qualify as “diagnostic” when the rules are applied strictly. For a patient, all three of the categories (diagnostic, prognostic, predictive) are important. The presence of a MYD88 mutation suggests that ibrutinib therapy is an option.28 Therefore, L265P would probably be classified as tier I in most interpretations. Unlike MYD88 L265P, many genes do not have welldescribed mutation hotspots. A large variety of variants are observed in genes such as RUNX1, CEBPA and DNMT3A. Databases can be helpful when the variant has been described before. An example is the Catalog of Somatic Mutations in Cancer (COSMIC),29 which has collected information on mutations from peer-reviewed journals since 2004.30 The UMD database contains 6,870 variants for TP53 alone.31 COSMIC is manually curated and the UMD-TP53 has developed its own data-driven curation strategy.32 For genes which are less well understood than TP53, databases are less helpful. Alternatively, algorithms that predict the effect of variants on protein structure could be used. In silico analyses are immediately available and can be performed without expert knowledge. Major influencing factors include the type of amino acid exchange and the location of the variant in conserved or functional domains. The dbNSFP database33 contains pre-calculated values for all possible single nucleotide variants which result in amino-acid or splice-site changes in the human genome from 18 different algorithms. However, the read-out is not necessarily “yes” or “no”. Different algorithms almost never come to exactly the same result. Some well-known pathogenic variants are not rated high enough by common algorithms. For example, the W515L mutation in the MPL gene is known to be typical in myeloproliferative neoplasms, but is rated as damaging/pathogenic by only seven of the 18 algorithms. Currently, most of the algorithms are trained on single nucleotide variants and cannot be applied to indels. Finally, it should be emphasized that a high pathogenicity score is not always a synonym for causality or actionability.

Tomorrow There is no one-step solution for variant classification. Searching software for genetic variant interpretation on Google gives millions of hits. In a cross-laboratory comparison of variant classification, there was only 34% concordance.34 The data provided by whole exome sequenc1518

ing or whole genome sequencing will inevitably make the analysis more complex, since the number of identified variants is many times greater than with panel sequencing. Variants will be found in all the approximately 20,000 human genes, but not all of them will be either relevant or redundantly mutated in a disease. A comparison of two large study sets (TCGA and BeatAML)35 confirmed that 33 genes are frequently mutated in acute myeloid leukemia, but there was diversity regarding genes with a mutation frequency of 2% or less. Over 2,000 genes were found to be mutated in only one of the datasets or even one patient. Current guidelines, such as those from the AMP,36 are based on characterized genes. Therefore, Kaur et al.7 argue that panels are preferred for breast cancer.37 They can achieve deeper coverage, which is synonymous with greater sensitivity. Sensitivity in the 1-3% range is increasingly required because subclones should be detected,38 or because clonal hematopoiesis of indeterminate potential is already diagnosed if a mutation is found with a VAF of 2%.39 The sensitivity of whole genome sequencing is currently in the range of 15-20%, but the technique allows simultaneous identification of structural and copy number variations and the cost of sequencing is decreasing.40 We therefore suggest short-term strategies for current diagnostic use of large-scale datasets and long-term approaches to advance our understanding of the malignant process and therapeutic options. Overstating the importance of a variant is clearly dangerous, but a report with variants of unknown clinical significance can be difficult to translate into clinical consequences. Here, we outline major aspects for today’s usage. First, collaboration between different laboratory branches is needed to compare genetic and other biomarkers and between laboratories and physicians to tailor personalized answers. For example, by integration of different laboratory results, a patient in remission according to morphology, but still with a VAF of 50% could be identified to have a rare and possibly less relevant germline variant. Another example derives from the growing awareness of germline predisposition e.g. with SAMD9/SAMD9L mutations in myelodysplastic syndromes.41 If the family background and reference material are provided, testing can be adjusted. Second, databases are the cornerstones of variant interpretation. An impressive example of the success of combined forces is gnomAD, which is now the worldwide reference for germline variants. Third, in the context of monitoring, serial testing can reveal the outgrowth of a clone with a specific variant and highlight clinical relevance, as demonstrated by retrospective studies.38,42 Well-documented information from multiple time points is a resource for variant classification, also for following patients with the same variant, and ideally should be included in databases and classification algorithms in the future. Finally, filtering for known and well-studied changes is always a valid first step. The first wholegenome sequencing studies in hematology demonstrated respectable sensitivity and specificity when filtering for known copy number variations, structural variations and genes.43,44 Mutations outside coding regions are difficult to associate with functions. They influence gene expression by altering transcription factor binding, alternative splicing, and certain genomic variants are likely to be causal for the acquisition of chromosomal aberrations.45-47 haematologica | 2019; 104(8)

Variant classification in the sequencing era

Furthermore, they can influence pharmacogenetics,48 and the effect of the same somatic mutation may differ between patients depending on other acquired or inherited genetic factors. Artificial intelligence is a logical choice to exploit the full potential of the data and leave the binary mutation/polymorphism classification behind. The use of artificial intelligence in clinical oncology, genome interpretation, and especially in variant reporting has gained momentum.49,50 Data available from manually classified variants can be used to train deep neural networks.49 The advantage of this approach is that the algorithm is able to autonomously extract relevant features for classification

References 15. 1. Koshland DE Jr. Molecule of the year. Science. 1993;262(5142):1953. 2. Heath EM, Chan SM, Minden MD, et al. Biological and clinical consequences of NPM1 mutations in AML. Leukemia. 2017;31(4):798-807. 3. 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. 4. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 2015;17(5):405-424. 5. Vogelstein B, Papadopoulos N, Velculescu VE, et al. Cancer genome landscapes. Science. 2013;339(6127):1546-1558. 6. Mwenifumbo JC, Marra MA. Cancer genome-sequencing study design. Nat Rev Genet. 2013;14(5):321-332. 7. Doroshow JH, Kummar S. Translational research in oncology--10 years of progress and future prospects. Nat Rev Clin Oncol. 2014;11(11):649-662. 8. Bejar R. CHIP, ICUS, CCUS and other fourletter words. Leukemia. 2017;31(9):18691871. 9. Lee-Six H, Obro NF, Shepherd MS, et al. Population dynamics of normal human blood inferred from somatic mutations. Nature. 2018;561(7724):473-478. 10. Fernandez LC, Torres M, Real FX. Somatic mosaicism: on the road to cancer. Nat Rev Cancer. 2016;16(1):43-55. 11. Yokoyama A, Kakiuchi N, Yoshizato T, et al. Age-related remodelling of oesophageal epithelia by mutated cancer drivers. Nature. 2019;565(7739):312-317. 12. Lynch M. Rate, molecular spectrum, and consequences of human mutation. Proc Natl Acad Sci U S A. 2010;107(3):961-968. 13. Padron E, Ball MC, Teer JK, et al. Germ line tissues for optimal detection of somatic variants in myelodysplastic syndromes. Blood. 2018;131(21):2402-2405. 14. Thiede C, Prange-Krex G, Freiberg-Richter J, Bornhauser M, Ehninger G. Buccal swabs but not mouthwash samples can be used to obtain pretransplant DNA fingerprints from recipients of allogeneic bone marrow trans-

haematologica | 2019; 104(8)

16.

17.

18.

19.

20.

21.

22.

23. 24.

25.

26.

and identify important combinations not only for genetic information but for all types of biomarker. There is no need for any manually defined set of rules. This is especially useful for variant interpretation because, as described above, it is basically impossible to capture the entire complexity using a simple set of rules. The output of such algorithms could indicate clinically relevant likelihoods. However, in order to aid clinical decision-making, a future report should not just be a list of variants with their individual classifications but rather a personalized summary of all genetic information (including structural and copy number variations) and other biomarkers and their combined meaning.

plants. Bone Marrow Transplant. 2000;25 (5):575-577. Imanishi D, Miyazaki Y, Yamasaki R, et al. Donor-derived DNA in fingernails among recipients of allogeneic hematopoietic stemcell transplants. Blood. 2007;110(7):22312234. Ritter DI, Roychowdhury S, Roy A, et al. Somatic cancer variant curation and harmonization through consensus minimum variant level data. Genome Med. 2016;8(1):117. Kitts A, Phan L, Ward M, Holmes B. The Database of Short Genetic Variation (dbSNP). The NCBI Handbook, 2nd ed. National Center for Biotechnology Information (US), Bethesda (MD). 2013:259298. Lek M, Karczewski KJ, Minikel EV, et al. Analysis of protein-coding genetic variation in 60,706 humans. Nature. 2016;536(7616): 285-291. Yarchoan M, Hopkins A, Jaffee EM. Tumor mutational burden and response rate to PD1 inhibition. N Engl J Med. 2017;377 (25):2500-2501. 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. Tarailo-Graovac M, Zhu JYA, Matthews A, van Karnebeek CDM, Wasserman WW. Assessment of the ExAC data set for the presence of individuals with pathogenic genotypes implicated in severe Mendelian pediatric disorders. Genet Med. 2017;19(12):1300-1308. McNerney ME, Godley LA, Le Beau MM. Therapy-related myeloid neoplasms: when genetics and environment collide. Nat Rev Cancer. 2017;17(9):513-527. George A, Banerjee S, Kaye S. Olaparib and somatic BRCA mutations. Oncotarget. 2017;8(27):43598-43599. Sukhai MA, Craddock KJ, Thomas M, et al. A classification system for clinical relevance of somatic variants identified in molecular profiling of cancer. Genet Med. 2016;18 (2):128-136. Li MM, Datto M, Duncavage EJ, et al. Standards and guidelines for the interpretation and reporting of sequence variants in cancer: a joint consensus recommendation of the Association for Molecular Pathology, American Society of Clinical Oncology, and College of American Pathologists. J Mol Diagn. 2017;19(1):4-23. Dietrich S, Glimm H, Andrulis M, et al.

27. 28.

29.

30.

31. 32.

33.

34.

35.

36.

37.

38.

39.

BRAF inhibition in refractory hairy-cell leukemia. N Engl J Med. 2012;366(21):20382040. Yu X, Li W, Deng Q, et al. MYD88 L265P mutation in lymphoid malignancies. Cancer Res. 2018;78(10):2457-2462. Treon SP, Xu L, Hunter Z. MYD88 Mutations and response to ibrutinib in Waldenstrom's macroglobulinemia. N Engl J Med. 2015;373(6):584-586. Forbes SA, Beare D, Boutselakis H, et al. COSMIC: somatic cancer genetics at highresolution. Nucleic Acids Res. 2017;45(D1): D777-D783. Bamford S, Dawson E, Forbes S, et al. The COSMIC (Catalogue of Somatic Mutations in Cancer) database and website. Br J Cancer. 2004;91(2):355-358. Soussi T, Wiman KG. TP53: an oncogene in disguise. Cell Death Differ. 2015;22(8):12391249. Edlund K, Larsson O, Ameur A, et al. Datadriven unbiased curation of the TP53 tumor suppressor gene mutation database and validation by ultradeep sequencing of human tumors. Proc Natl Acad Sci U S A. 2012;109(24):9551-9556. Liu X, Wu C, Li C, Boerwinkle E. dbNSFP v3.0: A one-stop database of functional predictions and annotations for human nonsynonymous and splice-site SNVs. Hum Mutat. 2016;37(3):235-241. Amendola LM, Jarvik GP, Leo MC, et al. Performance of ACMG-AMP variant-interpretation guidelines among Nine Laboratories in the Clinical Sequencing Exploratory Research Consortium. Am J Hum Genet. 2016;98(6):1067-1076. Tyner JW, Tognon CE, Bottomly D, et al. Functional genomic landscape of acute myeloid leukaemia. Nature. 2018;562 (7728):526-531. McClure RF, Ewalt MD, Crow J, et al. Clinical significance of DNA variants in chronic myeloid neoplasms: a report of the Association for Molecular Pathology. J Mol Diagn. 2018;20(6):717-737. Kaur P, Porras TB, Ring A, Carpten JD, Lang JE. Comparison of TCGA and GENIE genomic datasets for the detection of clinically actionable alterations in breast cancer. Sci Rep. 2019;9(1):1482. Chen J, Kao YR, Sun D, et al. Myelodysplastic syndrome progression to acute myeloid leukemia at the stem cell level. Nat Med. 2019;25(1):103-110. Valent P, Kern W, Hoermann G, et al. Clonal

1519

C. Baer et al.

40.

41. 42. 43.

1520

hematopoiesis with oncogenic potential (CHOP): separation from CHIP and roads to AML. Int J Mol Sci. 2019;20(3). Wrzeszczynski KO, Felice V, Abhyankar A, et al. Analytical validation of clinical wholegenome and transcriptome sequencing of patient-derived tumors for reporting targetable variants in cancer. J Mol Diagn. 2018;20(6):822-835. Kennedy AL, Shimamura A. Genetic predisposition to MDS: clinical features and clonal evolution. Blood. 2019;133(10):1071-1085. da Silva-Coelho P, Kroeze LI, Yoshida K, et al. Clonal evolution in myelodysplastic syndromes. Nat Commun. 2017;8:15099. Klintman J, Barmpouti K, Knight SJL, et al.

Clinical-grade validation of whole genome sequencing reveals robust detection of lowfrequency variants and copy number alterations in CLL. Br J Haematol. 2018;182(3): 412-417. 44. Hรถllein A, Twardziok SO, Walter W, et al. WGS and RNA-seq is superior to conventional diagnostic tests in multiple myeloma: ready for prime time? Blood. 2018;132 (Suppl. 1):4442. 45. Khurana E, Fu Y, Chakravarty D, et al. Role of non-coding sequence variants in cancer. Nat Rev Genet. 2016;17(2):93-108. 46. Pellagatti A, Armstrong RN, Steeples V, et al. Impact of spliceosome mutations on RNA splicing in myelodysplasia: dysregulated

47.

48. 49.

50.

genes/pathways and clinical associations. Blood. 2018;132(12):1225-1240. Loh PR, Genovese G, Handsaker RE, et al. Insights into clonal haematopoiesis from 8,342 mosaic chromosomal alterations. Nature. 2018;559(7714):350-355. Relling MV, Evans WE. Pharmacogenomics in the clinic. Nature. 2015;526(7573):343350. Sundaram L, Gao H, Padigepati SR, et al. Predicting the clinical impact of human mutation with deep neural networks. Nat Genet. 2018;50(8):1161-1170. Zomnir MG, Lipkin L, Pacula M, et al. Artificial intelligence approach for variant reporting. JCO Clin Cancer Inform. 2018;2.

haematologica | 2019; 104(8)

REVIEW ARTICLE

Which are the most promising targets for minimal residual disease-directed therapy in acute myeloid leukemia prior to allogeneic stem cell transplant?

Ferrata Storti Foundation

Brian Ball and Eytan M. Stein

Memorial Sloan Kettering Cancer Center, New York, NY, USA

ABSTRACT

Haematologica 2019 Volume 104(8):1521-1531

inimal residual disease has emerged as an important prognostic factor for relapse and survival in acute myeloid leukemia. Eradication of minimal residual disease may increase the number of patients with long-term survival; however, to date, strategies that specifically target minimal residual disease are limited. Consensus guidelines on minimal residual disease detection by immunophenotypic and molecular methods are an essential initial step for clinical trials evaluating minimal residual disease. Here, we review promising targets of minimal residual disease prior to allogeneic stem cell transplantation. Specifically, the focus of this review is on the rationale and clinical development of therapies targeting: oncogenic driver mutations, apoptosis, methylation, and leukemic immune targets. We review the progress made in the clinical development of therapies against each target and the challenges that lie ahead.

Introduction For over 45 years, standard therapy for fit patients with newly diagnosed acute myeloid leukemia (AML) has been induction chemotherapy with cytarabine and an anthracycline.1 Despite most patients achieving morphological remission with intensive chemotherapy, the prognosis for long-term survival in AML remains poor. Advances in multiparameter flow cytometry and molecular testing, including real-time quantitative polymerase chain reaction, digital polymerase chain reaction and next-generation sequencing, have enabled detection of minimal or measurable residual disease (MRD) far below a threshold of 5% blasts required for morphological remission.2 Among patients receiving induction chemotherapy, complete remission (CR) with persistent MRD occurs in a substantial 40% of patients.3 Mounting evidence has shown that the presence of MRD detectable prior to myeloablative allogeneic stem cell transplantation (SCT) is associated with shorter survival and increased risk of relapse that is similar to the risk in patients with active disease.4-7 Eradication of MRD prior to allogeneic SCT has the potential to increase longterm survival in AML. However, few studies have reported on the outcomes of patients converting from MRD-positive to MRD-negative disease after treatment with consolidation therapies. In the HOVON/SAKK AML 42A study, post-remission treatment with either chemotherapy, autologous or allogeneic SCT led to a change from MRD-positive to MRD-negative status in 7/21 (33%) patients.8 In the GIMEMA study, late MRD clearance (induction positive, consolidation negative MRD status) was observed in 15/134 (11%) patients and was associated with similar rates of 5-year overall survival and relapse-free survival as those of patients with early MRD clearance (induction negative, consolidation negative MRD status). MRD status after consolidation was the only factor independently associated with both a shorter duration of relapse-free survival and overall survival in multivariate analaysis, suggesting a more favorable outcome from MRD conversion after post-remission chemotherapy.9 Given the modest rates of MRD conversion with consolidation chemotherapy, more effective therapies capable of eradicating MRD prior to transplantation are urgently needed. As a reservoir for relapse, MRD would ideally be targeted by therapies that haematologica | 2019; 104(8)

Correspondence: EYTAN M. STEIN steine@mskcc.org

Received: April 15, 2019. Accepted: July 1, 2019. doi:10.3324/haematol.2018.208587 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1521 ÂŠ2019 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.

1521

B. Ball and E.M. Stein et al.

reduce the potential for recurrence by eliminating leukemia regenerating cells. AML is a heterogeneous disease that includes populations of bulk leukemic blasts and leukemic stem cells that are thought to be more refractory to treatment than others.10 Leukemic stem cells were initially defined phenotypically by specific cell surface markers CD34+ CD38- and functionally by an ability to initiate leukemia in animal transplant models.11 Cellular tracking of leukemic cell populations demonstrated the persistence of either leukemic stem cell subclones or more committed leukemia cells that retained stemness transcriptional programs from disease initiation to relapse.12 Therefore, central to the development of MRD targeting is the ability of the novel therapies to eradicate leukemic stem cells. In this review, we discuss MRD targets of therapeutic potential. We focus on the therapies that have been developed for each target and, if available, evidence of efficacy in reducing MRD prior to allogeneic SCT.

Targeting oncogenic driver mutations

NCT02668653 for quizartinib. In a single-arm, phase II study (NCT02283177) of crenolanib in combination with standard induction and consolidation chemotherapy followed by crenolanib maintenance for 1 year, 24 out of 29 (83%) patients achieved CR and 20 out of 25 evaluable patients (80%) achieved MRD-negative disease, as determined by multiparameter flow cytometry.26,27 Similarly, in a phase I study in patients with newly diagnosed FLT3mutated AML, gilteritinib in combination with induction and consolidation led to a high CR rate of 77% (n=23/30).28 A phase I study of quizartinib in combination with induction and consolidation in newly diagnosed AML led to CR in six of nine (67%) patients and a morphological leukemia-free state in two of nine (22%) patients with FLT3-ITD mutations.29 The high response rates of next-generation FLT3 inhibitors in combination with chemotherapy in early phase studies led to the development of randomized studies comparing gilteritinib (NCT03836209) and crenolanib (NCT02283177) to midostaurin in combination with induction and consolidation chemotherapy.

Fms-like tyrosine kinase 3 (FLT3)

Isocitrate dehydrogenases (IDH1 and IDH2)

Fms-like tyrosine kinase 3 (FLT3) is the most commonly mutated gene in AML with FLT3 internal tandem duplications (ITD) and FLT3 tyrosine kinase domain (TKD) mutations occurring in 22-32% and 8% of newly diagnosed cases, respectively.13,14 In a large population-based study the incidence of FLT3-ITD mutations was lower at 18.9% and decreased with age.15 FLT3-ITD mutations are associated with worse prognosis and increased risk of relapse with allogeneic transplantation.13,14,16-18 As monotherapy, FLT3 inhibitors are capable of inducing molecular remissions and gilteritinib (Xospata) is approved for relapsed or refractory FLT3-mutated AML.19,20 Quizartinib has also demonstrated efficacy as monotherapy in patients with relapsed or refractory FLT3-ITD-mutated AML.21 The combination of FLT3 inhibitors with chemotherapy has the potential to induce deeper remissions than induction chemotherapy alone. Midostaurin (Rydapt) is a first-generation FLT3 inhibitor that was originally developed as a protein kinase C inhibitor and found to have inhibitory activity against multiple tyrosine kinases including FLT3.22 The phase III RATIFY trial randomized younger patients with newly diagnosed FLT3-TKD or FLT3-ITD mutated AML to midostaurin in combination with induction and consolidation chemotherapy or placebo with standard chemotherapy. Patients in the midostaurin arm had a significantly longer median overall survival (74.7 vs. 25.6 months, P=0.009) leading to approval of the regimen. In this study, MRD was not assessed; however, among patients undergoing allogeneic SCT, midostaurin in combination with chemotherapy led to a near significant increase in overall survival (P=0.07) and a significant decrease in cumulative incidence of relapse [hazard ratio (HR) 0.47, P=0.02].23,24 Next-generation FLT3 inhibitors have greater specificity and higher potency. Type I inhibitors such as gilteritinib and crenolanib are active against FLT3-TKD or FLT3-ITD mutations. In contrast, FLT3-TKD mutations in the activation loop and gatekeeper domain confer resistance to type 2 inhibitors such as quizartinib.25 Active clinical trials evaluating next-generation FLT3 inhibitors in combination with induction and consolidation include NCT02283177 for crenolanib, NCT02236013 for gilteritinib, and

Mutations involving the isocitrate dehydrogenase-1 (IDH1) and -2 (IDH2) genes occur in about 6-10% and 913% of newly diagnosed cases of AML, respectively.30-35 Mutant IDH has neomorphic enzyme activity leading to aberrant production of the oncometabolite 2-hydroxyglutarate.33,36 Accumulation of 2-hydroxyglutarate competitively inhibits Îą-ketoglutarate-dependent enzymes including TET2, a DNA hydroxymethylase resulting in global hypermethylation, a block in cellular differentiation, an increase in self-renewal and enhancement of leukemic transformation.36-38 Ivosidenib (Tibsovo) and enasidenib (Idhifa) are oral inhibitors of mutant IDH1 and IDH2, respectively and are approved for relapsed or refractory IDH1- and IDH2-mutant AML.39,40 In relapsed or refractory AML, ivosidenib led to clearance of IDH1 mutations in seven out of 25 (28%) patients who achieved either CR or CR with incomplete count recovery (CRi).39 Similarly, treatment with enasidenib in relapsed or refractory AML led to IDH2 mutation clearance in nine out of 29 (31%) patients achieving a CR.41 Preliminary results from a phase I study of ivosidenib or enasidenib in combination with standard induction and consolidation chemotherapy in patients with newly diagnosed IDH-mutated AML demonstrated that the combination was well tolerated. Among patients treated with ivosidinib, responses [CR, CRi or CR with imcomplete platelet recovery (CRp)] occurred in 26 out of 28 (93%) and 33 out of 45 (73%) patients with de novo and secondary IDH1-mutated AML, respectively. In the enasidenib group responses occurred in 33 out of 45 (73%) and 20 out of 32 (63%) patients with de novo and secondary IDH2-mutated AML, respectively. Furthermore, IDH-mutation clearance was observed in nine out of 22 (41%) of the patients with IDH1 mutations and in 11 out of 31 (30%) of those with IDH2 mutations. MRD negativity by multiparameter flow cytometry was observed in eight out of nine (89%) patients with IDH1 mutations and seven out of 12 (58%) of those with IDH2 mutations.42 Although IDH inhibitors and chemotherapy may increase MRD-negative rates, further studies are needed to determine the impact of the combination on survival after allogeneic SCT. A phase III, randomized study of ivosidenib or enasidenib in combination with

1522

haematologica | 2019; 104(8)

Promising targets for MRD therapies in AML

induction and consolidation chemotherapy followed by maintenance therapy in newly diagnosed AML or myelodysplastic syndrome (MDS) with excess blasts-2 with an IDH1 or IDH2 mutation (NCT03839771) will soon begin enrollment. The observation that cancer stem cells are resistant to therapies targeting BCR-ABL in chronic myeloid leukemia and JAK2 V617F in myeloproliferative neoplasms raises concern regarding the ability of targeted therapies to eradicate leukemic stem cells.21,22 If indeed FLT3 and IDH1/2 inhibitors are unable to eradicate leukemic stem cells, then targeted therapy may reduce or maintain low levels of bulk disease but will likely not be curative unless combined with allogeneic SCT or other therapies targeting leukemic stem cell. A leukemic stem cell population that is refractory to targeted therapy may also contribute to clonal evolution and the acquisition of secondary resistance mutations. Clinical studies evaluating FLT3 and IDH inhibitors as maintenance therapy after induction and consolidation and allogeneic SCT are also essential to determine the optimal duration of treatment. In the phase II AMLSG 16-10 trial, treatment with midostaurin in combination with induction and consolidation chemotherapy followed by maintenance midostaurin for 1 year after allogeneic SCT was associated with improved 1-year eventfree survival when compared to that of historical controls with FLT3-ITD-mutated AML [HR 0.58; 95% confidence interval (95% CI): 0.48-0.7; P<0.001].43

Targets of apoptosis evasion B-cell lymphoma 2 (BCL2) Evasion of apoptosis is a hallmark of malignant tumor progression, allowing for tumor survival and resistance to cancer treatments.37 The anti-apoptotic protein B-cell lymphoma 2 (BCL2) is overexpressed in AML and associated with resistance to chemotherapy and poor outcomes.44 The prosurvival BCL2 family of proteins such as BCL2 and MCL1 sequester the apoptosis initiator protein BIM to prevent initiation of apoptosis.45 Aberrant BCL2 expression is also essential for maintaining oxidative phosphorylation in quiescent leukemic stem cells. BCL2 inhibition reduces oxidative phosphorylation and preferentially induces cell death in leukemic stem cells.46,47 Venetoclax is an oral, BH3 mimetic that selectively binds BCL2, displacing pro-apoptotic proteins leading to apoptosis.48 Monotherapy with venetoclax demonstrated clinical activity in early phase studies but was associated with modest response rates and a short duration of response.49 Combinations of venetoclax with both lowdose cytarabine and hypomethylating agents in previously untreated, newly diagnosed elderly patients not eligible for chemotherapy resulted in high response rates and durable remissions leading, to Food and Drug Administration (FDA) approval of these regimens.50,51 Venetoclax and hypomethylating agents led to a CR or CRi with MRD-negative disease by multiparameter flow cytometry in 45% of patients.52 Similarly, treatment with venetoclax and low-dose cytarabine led to MRD-negative disease in 32% of patients in CR or CRi.53 This spurred the development of trials evaluating venetoclax in combination with 7+3 (NCT03709758), CPX-351 (NCT03629171), or FLAG-IDA (NCT03214562) based induction regimens in newly diagnosed patients eligible for chemotherapy. In haematologica | 2019; 104(8)

a phase I study of venetoclax in combination with FLAGIDA in relapsed or refractory AML, treatment was well tolerated and eight of 11 patients achieved a CR or CRi.54 The high MRD-negative rates associated with venetoclax combinations are encouraging; however, additional phase III studies are needed to determine if there is a survival benefit, in particular among patients who undergo allogeneic SCT.

Tumor protein 53 (TP53) p53 is a transcription factor that is activated by cellular stress and promotes cell cycle arrest, senescence and apoptosis.55 Loss of p53 induces oncogenic self-renewal in mouse hematopoietic progenitor cells.56 In AML, inactivating mutations in the TP53 gene occur in 7-18% of patients with newly diagnosed AML and are enriched in patients with other poor prognostic features including complex karyotype and therapy-related disease.57,58 The co-occurrence of TP53 mutations and a complex karyotype is associated with an especially dismal prognosis and a high rate of relapse after allogeneic SCT.59 In AML, p53 inactivation more commonly results from overexpression of negative regulators.60,61 MDMX and MDM2 inhibit p53 transactivation and induce its ubiquitination with subsequent degradation.62 Idasanutlin is an oral selective MDM2 inhibitor capable of activating apoptosis in a p53-dependent manner.63 Current trials evaluating the combination of this MDM2 inhibitor with chemotherapy include a phase I/II study (NCT03850535) of idasanutlin in combination with standard induction chemotherapy in newly diagnosed AML and a phase III study (NCT02545283) of idasanutlin with or without cytarabine in relapsed or refractory AML. Despite many patients achieving deep and durable remissions with apoptosis inhibitors, primary and secondary resistance is known to occur. In particular, RAS pathway mutations and TP53 mutations are associated with decreased responses to venetoclax.47,51,64,65 MCL-1 also serves as a redundant pro-survival pathway that mediates resistance to venetoclax.48,49 In cell lines resistant to BCL2 inhibition, idasanutlin led to induction of apoptosis through p53 activation and MCL1 degradation.52 MCL1 mimetics currently in active trials as monotherapy and in combination with venetoclax include S64315 (Servier) (NCT02979366, NCT03672695), AMG 176 (Amgen) (NCT02675452, NCT03797261), and AMG 397 (Amgen) (NCT03465540). Additionally, TP53-mutant AML are resistant to MDM2 inhibitors and prolonged exposure to idasanutlin in cancer cell lines has been associated with the development of TP53 mutations. APR-246 is a prodrug that is converted to the Michael acceptor methylene quinuclidinone, which covalently binds mutated p53 cysteine residues 124 and 277, leading to refolding and restoration of p53 function.66,67 In a phase Ib study of APR-246 in combination with azacitidine in patients with TP53-mutant MDS and AML, all 11 evaluable patients responded with nine patients achieving CR (82%) and eight having clearance of p53 mutations (73%).68

Methylation The hypomethylating agents 2â&#x20AC;&#x2122;deoxy-5-azacitidine (decitabine) and 5-azacitidine (azacitidine) are approved for the treatment of MDS and newly diagnosed AML patients unfit for chemotherapy.69-71 Azacitidine and 1523

B. Ball and E.M. Stein et al.

decitabine are nucleoside analogs that irreversibly bind the methylase DNMT1 leading to global hypomethylation, resulting in altered expression and cell death.72,73 Low doses of hypomethylating agents disrupt immune evasion by inducing expression of tumor-associated antigens such as cancer/testis antigens in AML cell lines and antigen presentation molecules such as human leukocyte antigen class I antigens.74-77 Hypomethylating agents also upregulate expression of endogenous retroviruses that activate viral recognition and interferon response pathways.78,79 In contrast, treatment with hypomethylating agents induced expression of programmed cell death protein 1 (PD1), programmed death-ligand 1 and 2 (PD-L1 and PD-L2) and cytotoxic T-cell ligand antigen 4 (CTLA-4) in patients with MDS, AML and chronic myelomonocytic leukemia and was associated with resistance to treatment with hypomethylating agents.80 In the RELAZA2 trial, patients with advanced MDS or AML who achieved a CR after conventional chemotherapy or allogeneic SCT but had MRD, detected by either quantitative polymerase chain reaction for mutant NPM1 or other leukemia-specific fusion genes or by flow cytometry, were treated with azacitidine.81 The study met its primary endpoint with 31 out of 53 (58%) patients being relapse-free at 6 months. Reassessment of MRD status revealed that 19 out of 53 patients achieved MRD negativity and 12 out of 19 MRD-negative patients maintained MRD negativity without hematologic relapse during the median follow-up time of 23 months. Post-hoc analysis demonstrated a difference in relapse-free survival (HR 0.2, P<0.0001), but not overall survival (HR 0.4, P=0.112), between responders and non-responders to azacitidine.81

Immunotherapy targets Immunotherapy is an approach that uses the potency of the immune system as a therapeutic modality against cancer.82,83 The rationale for immunotherapy in AML lies in the curative potential of allogeneic SCT as post-remission therapy mediated by a graft-versus-leukemia effect. Similarly, immunotherapy leverages the adaptive immune system, specifically antibodies from B cells and the T-cell receptor on T cells to recognize antigens expressed on the cancer cells. In AML, immunotherapy has the potential to target unique leukemic stem cell surface antigens, thereby selectively eradicating these cells.

Immune checkpoints: PD1, PD-L1, CTLA-4 Immune modulating antibodies against negative regulators of T-lymphocyte activation, including anti-CTLA-4 and anti-PD1/PD-L1 have produced unprecedented rates of durable responses in a variety of malignancies.83 In AML, responses to checkpoint inhibitors as monotherapy have been modest. A phase I study of patients treated with the anti-PD1 antibody pidilizumab revealed a response in only one out of eight patients with AML with a reduction in blast percentages from 50% to 5%.84 A phase I/Ib study of 28 patients with relapsed hematologic malignancies after allogeneic transplantation, including 12 patients with AML, evaluated the anti-CTLA-4 antibody ipilimumab, given at doses of 3 mg/kg and 10 mg/kg. Responses were only observed with the ipilimumab 10 mg/kg dose in seven out of 22 (32%) patients and included CR in four patients with extramedullary AML and one 1524

patient with MDS that progressed to AML. Dose-limiting chronic graft-versus-host disease of the liver or gut occurred in four patients but resolved when the treatment was withheld and steroids were administered.85 Active phase II studies evaluating anti-PD1 therapy as postremission treatment include NCT02532231 with nivolumab and NCT02708641 with pembrolizumab. In order to enhance responses to checkpoint inhibition in AML, combinations with chemotherapy, hypomethylating agents, and other checkpoint inhibitors are under investigation. In a phase II study (NCT02464657) patients with newly diagnosed AML received induction chemotherapy with idarubicin and cytarabine followed by nivolumab 3 mg/kg starting on day 24 and continued every 2 weeks for up to 1 year; 34 out of 44 patients (77%) achieved a CR or CRi and 18 out of 43 (53%) had undetectable MRD by multiparameter flow cytometry. Responses were durable and the median overall survival was 18.5 months, which compared favorably to that of a contemporary cohort of patients treated with idarubicin and cytarabine induction alone. Among 18 patients who underwent allogeneic SCT, 13 (72%) developed graft-versus-host disease and eight responded to treatment.86 Increased expression of PD1, PD-L1, and CTLA-4 is associated with resistance to treatment with hypomethylating agents but has the potential to sensitize leukemia cells to checkpoint-blocking monoclonal antibodies.74,80 In a phase II study of azacytidine and nivolumab 3 mg/kg on days 1 and 14 in relapsed/refractory AML, responses occurred in 23 patients (overall response rate, 33%) including 15 patients (22%) with CR or CRi. The median overall survival for all patients enrolled was 6.3 months, while that of the patients who achieved any type of response (CR, CRi, partial response or hematologic improvement) or had stable disease was 16.2 months. When compared to controls from historical hypomethylating agent-based clinical trials, patients receiving nivolumab and hypomethylating agents had an increased response rate (33% vs. 20%) and significantly longer median overall survival (6.3 vs. 4.6 months).87 The phase II PEMAZA study is evaluating azacitidine in combination with pembrolizumab in patients achieving CR after induction chemotherapy but with detectable MRD (NCT03769532).

Dendritic cells Dendritic cells are the most potent antigen-presenting cells capable of priming new responses or enhancing existing antigen-specific immune responses.88,89 Mature dendritic cells facilitate cytotoxic T-lymphocyte activation through antigen presentation on major histocompatibility complex class 1 molecules, termed cross-presentation and by upregulating co-stimulatory molecules, such as CD80 and CD86.89,90 Dendritic cell vaccination approaches differ in the source of dendritic precursors, maturation methods, target antigen, antigen loading, and in the administration of the vaccine.89 A phase II study of patients with AML in first CR after induction chemotherapy at high risk for relapse and without a matched sibling donor for allogeneic hematopoietic SCT revealed that treatment with WT1 mRNA-electroporated dendritic cell vaccine led to a clinical response in 13 out of 30 patients (30%) with nine patients achieving molecular remission by WT1 transcript levels.91 The 5-year overall survival rate was 40% among vaccine recipients and compared favorably to a 5-year overall survival rate of 24.7% observed in historical conhaematologica | 2019; 104(8)

Promising targets for MRD therapies in AML

trols.91 Additionally, the dendritic cell vaccine elicited WT1-specific CD8+ T-cell responses resulting in expression that correlated with long-term survival.91 Another prospective study of a vaccine composed of patientderived AML cells fused with autologous dendritic cells in patients in CR after induction chemotherapy not eligible for allogeneic SCT led to sustained remission in 12 of 17 patients receiving at least one dose of vaccine and a 4-year progression-free survival rate of 71%: the median progression-free and overall survival had not been reached.92 The vaccine was well tolerated with the most common adverse events being erythema, pruritis and/or induration at the vaccine site.92 The dendritic cell/AML fusion also induced CD8+ T-cell specific responses and an increased circulating leukemia-reactive T-cell population that persisted for more than 6 months.92

Antibody drug conjugates and bispecific T-cell engaging therapy Cluster of differentiation 33 (CD33) The development of an antibody-based therapy targeting antigens expressed on leukemic blasts to eradicate MRD is supported by the efficacy of the CD19/CD3 bispecific antibody, blinatumomab in B-cell acute lymphoblastic leukemia.73 CD33 is a transmembrane sialic acid-binding immunoglobulin-like lectin (SIGLEC) family protein that is expressed by cells of the myeloid lineage but not hematopoietic stem cells.93-95 CD33 is expressed on leukemic blasts as well as CD34+/CD38- leukemic stem cells.96 CD33 levels are highest in acute promyelocytic leukemia and AML with NPM1, FLT3-ITD and KMT2A mutations and lower in those with core-binding factor translocations or complex cytogenetics.97 Gemtuzumab ozogamicin (GO) is a human antibody conjugated to a

DNA-damaging calicheamicin derivative by an acid-labile linker.98 Based on promising results from three single-arm phase II studies at a dose of 9 mg/m2 given every 2 weeks, GO was initially granted FDA approval for patients >60 years of age with CD33+ AML who were not candidates for aggressive chemotherapy.99 However, GO was later withdrawn from the commercial market in October 2010 after the confirmatory phase III SWOG S0106 study showed no survival benefit and increased treatment-related mortality in patients treated with GO compared to those given standard induction.100 Subsequent studies have evaluated reduced and fractionated dosing of GO to decrease treatment-related toxicity.100-103 A large meta-analysis from five randomized controlled trials of patients with newly diagnosed AML receiving GO with induction chemotherapy revealed that the addition of GO was associated with a reduced risk of relapse (odds ratio 0.81, P=0.0001) and improved overall survival at 5 years (odds ratio 0.9, P=0.01), especially in patients with favorable and intermediate-risk cytogenetics.104 Additionally, the NCRI AML17 trial demonstrated a lower rate of veno-occlusive disease and early mortality but no difference in relapse or survival at 4 years between patients given GO at a dose of 3 mg/m2 or a dose of 6 mg/m2.105 As a result GO received FDA approval for adults with newly diagnosed AML, whose tumor expresses the CD33 antigen. Retrospective analysis of adult patients with NPM1-mutated AML enrolled in the ALFA-0701 trial revealed that GO in combination with induction chemotherapy increased the proportion of patients with MRD-negative disease at the end of treatment, as determined by NPM1 gene transcript levels, when compared to those treated with chemotherapy alone (91% vs. 61%, P=0.028).106 This has led to a phase II trial of fractionated GO on days 1, 4, and 7 in patients with MRD after at least one cycle of induction chemotherapy. (NCT03737955)

Figure 1. Active clinical trials evaluating minimal residual disease-directed therapies arranged by trial design. Trials with induction and consolidation-based combinations are shown on the left, non-chemotherapy post-remission therapies are shown on the right. Studies evaluating post-allogeneic transplant minimal residual disease therapies are not included.

haematologica | 2019; 104(8)

1525

B. Ball and E.M. Stein et al. Table 1. Outcomes of clinical trials targeting minimal residual disease with induction chemotherapy or as post-remission therapy.

MRD target

Drug name

Trial

Combination

Population

Clinical phase

Efficacy

MRD negative rate

Ref.

FLT3 TKD FLT3 ITD

Gilteritinib (ASP 2215)

NCT02236013

Gilteritinib with induction and consolidation chemotherapy

Newly diagnosed AML

Not reported

(28)

FLT3 TKD FLT3 ITD

Crenolanib

NCT02283177

20/25 (80%) by MPFC

(26, 27)

Quizartinib

NCT01892371

Newly diagnosed FLT3-mutated AML Newly diagnosed AML

FLT3 ITD

Not reported

(29)

IDH1 IDH2

Ivosidenib Enasidenib

NCT02632708

Crenolanib with induction and consolidation chemotherapy Quizartinib with induction and consolidation chemotherapy Ivosidenib or enasidenib in combination with induction and consolidation chemotherapy

Newly diagnosed AML with an IDH1 and/or IDH2 mutation

Among FLT3 mutated patients: CR 23/30 (77%) CRc (CR/CRp/ CRi) 27/30 (90%) CRc 100% at 120 mg dose CR 24/ 29 (83%) 2 patients relapsed with a median follow-up of 14 months Among FLT3-ITD mutated patients CR 6/9 (67%) MLFS 2/9 (22%) Ivosidenib De novo AML CRc (CR, CRi, CRp) 26/28 (93%) Secondary AML CRc 6/13 (46%)

IDH1 MC: 9/22 (41%) of responding patients by NGS

(42)

BCL-2

Venetoclax

NCT03214562

DNMT1

Azacitidine

NCT01462578

PD1

Nivolumab

NCT02464657

Venetoclax in combination with FLAG-IDA None

Nivolumab in combination with standard induction and consolidation chemotherapy

Relapsed or refractory AML

Advanced MDS or AML in CR, MRD+ after induction or allo-SCT High-MDS or AML, chemotherapy naĂŻve

I/II

AMG 330 is a bispecific T-cell engager (BiTE) antibody construct that binds CD33 on leukemic blasts and CD3 on T cells.107 Preliminary results from a phase I study (NCT02520427) of AMG330, revealed serious adverse events in 23 out of 35 patients (66%) including cytokine release syndrome in 11 patients. The cytokine release syndrome was mitigated with step-up dosing, corticosteroid pretreatment, intravenous fluids, tocilizumab, and drug interruption. Two patients had a CR and two had a CRi during dose escalation.108

Cluster of differentiation 123 (CD123) CD123 is the alpha chain of the interleukin-3 receptor heterodimer and is expressed at higher levels in leukemic stem cells than on normal hematopoietic bone marrow stem cells.109,110 CD123+CD34+CD38- leukemic cells are capable of initiating and maintaining leukemia in NOD/SCID mice.110 IMGN632 is a CD123-targeting antibody-drug conjugate consisting of a CD123 antibody linked to a DNA alkylating indolino-benzodiazepine dimer (IGN) via a protease cleavable linker.111 In a phase I 1526

Enasidenib De novo AML CRc 33/45 (73%) Secondary AML CRc 20/32 (63%) CR+CRi 8/11 (73%)

Primary endpoint: relapse-free at 6 months post- treatment 31/53 (58%) CR+CRi 34/44 (77%)

IDH2 MC: 11/37 (30%) of responding patients by NGS

Not reported

(54)

19/53 (36%) by NPM1 or fusion gene transcript levels

(81)

18/34 (53%) by MPFC after induction

(86)

continued on the next page

trial of IMGN632 (NCT03386513) in patients with relapsed or refractory CD123+ hematologic malignancies, four out of 12 (33%) patients with AML achieved a CR or CRi.112 Elzonris (tagraxofusp or SL-401) is a recombinant fusion protein consisting of human interleukin-3 fused via a Met-His linker to a truncated diptheria toxin that is currently FDA-approved for the treatment of blastic plasmacytoid dendritic-cell neoplasm.113,114 The interleukin-3 domain binds to the interleukin-3 receptor leading to translocation of the diphtheria A fragment and thus to inactivation of protein synthesis and cell death. A phase I/II study of SL-401 as consolidation therapy for patients in first or second CR is ongoing. (NCT02270463)

C-type lectin-like molecule-1 (CLL1 or CLEC12A) C-type lectin-like molecule-1 (CLL1 or CLEC12A) is a transmembrane glycoprotein that functions as an inhibitory receptor. CLL-1 is expressed on leukemic blasts in the majority of cases and selectively expressed in leukemic CD34+CD38- cells but not normal hematopoietic stem cells. Moreover, CLL1+ CD34+ cells are serially transhaematologica | 2019; 104(8)

Promising targets for MRD therapies in AML

continued from the previous page

MRD target

Drug name

Trial

Combination

Population

Clinical phase

Dendritic cells

WT1- mRNA dendritic cells

NCT00965224

None

Dendritic cells

hTERTdendritic cells

NCT00510133

None

Dendritic cells

AML/ dendritic cell fusion

NCT01096602

None

Dendritic cells

DCPrime (DCP-001)

NCT00965224

None

CD33

Gemtuzumab ozoganicin

NCT00927498

GO in combination with standard induction and consolidation

AML or II MDS RAEB1/2 in PR or CR or smoldering course with high risk of relapse AML in first or II second CR after induction or consolidation Newly diagnosed II or first relapsed AML in CR ineligible for allo-SCT AML in CR after II â&#x2030;Ľ 1 course of chemotherapy and age >60 years or < 60 years without allo-SCT donor Untreated III de novo AML

Efficacy

MRD negative rate

Ref.

Clinical response rate 13/30 (43%) 5-year OS 40% vs. 24.7% historical control

9/30 (30%) by WT1 transcript levels

(91)

Recurrence free at a median 52 months follow up 11/19 (74%) 4-year progressionfree survival 71%

Not reported

(125)

Not reported

(92)

Clinical response rate 12/20 (43%) 5-years OS 40%

9/13 by (126) normalization of WT1 transcript levels

GO vs. control Median as: 27.5 vs. 21.8 months (P=0.16) Median event-free survival: 17.3 vs. 9.5 months (P<0.01)

Post- induction (106, 127) GO vs. control 39% vs. 7%, P<0.01 Post-treatment GO vs. control 91% vs. 61%,P=0.03 by NPM1 mutation transcript levels

MRD: minimal residual disease; Ref.: references; FLT3 TKD: fms-like tyrosine kinase 3 (FLT3) gene tyrosine kinase domain mutations; FLT3 ITD: FLT3 internal tandem duplications; AML: acute myeloid leukemia; CR: complete remission; CRi: complete remission with incomplete count recovery; CRp: complete remission with incomplete platelet recovery; CRc: complete remission - composite; MPFC: multiparameter flow cytometry; MLFS: morphological leukemia-free state; IDH1/ IDH2: isocitrate dehydrogenase-1 and -2; MC: mutation clearance; NGS: nextgeneration sequencing; BCL-2: B-cell lymphoma-2; FLAG-IDA: fludarabine, cytarabine, idarubicin, and granulocyte colony-stimulating factor; DNMT1: DNA methyltransferase 1; allo-SCT: allogeneic stem cell transplant; NPM1: nucleophosmin 1; PD1: programmed death protein 1; MDS: myelodysplastic syndrome; WT1: Wilms tumor 1; RAEB 1/2: refractory anemia with excess blasts; PR: partial response; OS: overall survival; hTERT: human telomerase reverse transcriptase; GO: gemtuzumab ozogamicin

plantable in NOD/SCID mice suggesting a self-renewal ability.115 MCLA-117 is a potent bispecific T-cell engager that directs CD3+ T cells to leukemia cells expressing CLL1.116 A phase I clinical trial of MCLA-117 in patients with relapsed or refractory AML or in elderly patients not eligible for chemotherapy is currently recruiting patients (NCT03038230).

Chimeric antigen receptor therapy Chimeric antigen receptors (CAR) are engineered extracellular receptors joined to intracellular signaling domains that reprogram immune cells for therapeutic purposes.117 The development of second-generation CAR with an additional CD28 or 41BB co-stimulatory domain has allowed for effective responses.117 CAR-T cells kill tumor cells and promote immune surveillance directly by persisting and indirectly by cross-priming tumor-infiltrating lymphocytes through antigen release.10-12 CAR therapy targeting CD19 is extremely effective in B-cell malignancies, resulting in the approval of tisagenlecleucel (Kymriah) for the treatment of pediatric B-cell acute lymphoblastic leukemia that is refractory or in second relapse and axicabtagene ciloleucel (Yescarta) in large B-cell lymphomas after two or more lines of systemic therapy. A phase I study of autologous CAR-T cells with specihaematologica | 2019; 104(8)

ficity for a difucosylated carbohydrate antigen Lewis (Le)Y coupled to the cytoplasmic domains of CD28 and TCRÎś chain produced a transient cytogenetic remission in one out of three patients with MRD at the time of infusion. Another patient with MRD prior to the infusion of CART cells had persistent cytogenetic MRD but sustained MRD negativity by multiparameter flow cytometry for 23 months. Although LeY CAR-T cells persisted up to 10 months after infusion, most patients relapsed within the first 5 months suggesting possible antigen escape. None of the patients developed grade 3 or 4 toxicity.118 In AML, the ideal CAR target that is highly expressed in myeloid blasts and spares normal myeloid progenitor cells and vital tissues has not yet been identified. In preclinical studies anti-CD33 CAR-T cells resulted in a reduction of normal myeloid progenitors.119,120 Similarly, anti-CD123 CAR-T cells have demonstrated myeloablation in a xenograft mouse model.121 CLL1 CAR-T cells are cytotoxic to normal mature myeloid cells but not to normal myeloid progenitor cells or hematopoietic stem cells.122 An extensive proteomic and transcriptomic analysis revealed four potential CAR targets, ADGRE2, CCR1, CD70, and LILRB2, with high expression in AML, AML leukemic stem cells, and low expression in normal tissues, normal hematopoietic stem and progenitor cells and resting/acti1527

B. Ball and E.M. Stein et al.

vated T cells. However, none of the targets showed a profile comparable with that of CD19 in B-cell malignancies.123 This suggests that combinatorial strategies may be necessary for targeting AML with CAR-T cells. An approach for combination includes bispecific T cells that co-express two CAR or a dual-specific CAR (CAR/CAR T cells) allowing T-cell recognition of target cells that express any of two given antigens.123 Alternatively, the combination of a CAR that alone is insufficient to activate a T cell and a chimeric co-stimulatory receptor (CAR/CCR T cells) restricts T-cell recognition to dual antigen-expressing target cells. The latter approach requires pan-expression of CAR targets on AML cells, which was not seen by Perna and colleagues.117 Persistent CAR-T-cell mediated myelotoxicity may necessitate incorporation of CAR-T cells with conditioning regimens prior to allogeneic SCT. An alternative approach currently in development is the use of genetically modified donor allografts that lack expression of CAR-T-cell targets, such as CD33, followed by administration anti-CD33 CAR-T cells after transplantation.124

Conclusion Advancements in flow cytometry, quantitative polymerase chain reaction analysis and more recently nextgeneration sequencing continue to push the limits of detection of residual disease and open the door to therapies aimed at eradicating it. As MRD is a significant negative prognostic factor for relapse and survival in AML following allogeneic SCT, therapies capable of eliminating MRD are urgently needed to increase the number of patients cured of their disease. Here, we have reviewed the most promising MRD targets with therapeutic potential based on efficacy in reducing MRD and potential for targeting leukemia repopulating cells mediating relapse. The targets discussed are by no means an exhaustive list and will continue to be refined as single-cell sequencing and xenograft studies better characterize leukemia populations in MRD that mediate relapse. Ultimately, incorporation of MRD into clinical practice will require pivotal trials that demonstrate an improvement in survival with MRD-directed approaches. Moving forward with MRD-

References 1. Yates JW, Wallace HJ Jr, Ellison RR, Holland JF. Cytosine arabinoside (NSC-63878) and daunorubicin (NSC-83142) therapy in acute nonlymphocytic leukemia. Cancer Chemother Rep. 1973;57(4):485-488. 2. 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. 3. Freeman SD, Hills RK, Virgo P, et al. Measurable residual disease at iduction redefines partial response in acute myeloid leukemia and stratifies outcomes in patients at standard risk without NPM1 mutations. J Clin Oncol. 2018;36(15):14861497. 4. Buckley SA, Appelbaum FR, Walter RB. Prognostic and therapeutic implications of

1528

targeted therapies will require a standardized method for detecting MRD and rigorous assessment of the safety and efficacy of these therapies. The European LeukemiaNet MRD working group has recently provided recommendations for assessment of MRD by multiparameter flow cytometry and molecular testing.2 These consensus recommendations aid the standardization of MRD testing should be incorporated into all AML clinical trials. Additional issues that will need to be addressed include the optimal timing of MRD assessments. MRD after induction, second induction and consolidation may have varying prognostic impact. Differences in time to initial response and the duration of response among MRD therapies may also affect the interval of MRD assessments. In particular, IDH inhibitors typically take a longer time to produce an initial response and may warrant later MRD assessments at later timepoints than MRD therapies with a faster onset of effect. The use of MRD as a surrogate endpoint for survival for clinical trials in AML has the potential to accelerate drug development. Although MRD has a significant impact on prognosis, the mortality associated with treating MRD also needs to be considered. The experience with CD33targeted therapies demonstrates that toxicities associated with treatment may outweigh the potential benefit associated with eradicating MRD. In addition, MRD as a surrogate endpoint would not capture the impact of MRD therapies on transplant outcomes. For example, vadastuximab and GO were associated with an increased risk of veno-occlusive disease after transplantation. T-cell-activating therapies such as checkpoint inhibitors, dendritic cell vaccines and CAR-T cells have the potential to increase the risk of graft-versus-host disease after transplantation. Therefore, initial studies evaluating the safety of MRD-directed therapies should include post-transplant outcomes to identify late toxicities. The development of MRD-directed therapies may be facilitated in other ways. Similar to clinical trials in acute lymphoblastic leukemia and pediatric AML, current and future clinical trials in patients with AML who are fit for allogeneic SCT should include an intensification arm with MRD-directed therapies. This has the potential to increase the number of trials evaluating MRD therapies.

minimal residual disease at the time of transplantation in acute leukemia. Bone Marrow Transplant. 2013;48(5):630-641. 5. Walter RB, Gyurkocza B, Storer BE, et al. Comparison of minimal residual disease as outcome predictor for AML patients in first complete remission undergoing myeloablative or nonmyeloablative allogeneic hematopoietic cell transplantation. Leukemia. 2015;29(1):137-144. 6. 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. 7. Kayser S, Benner A, Thiede C, et al. Pretransplant NPM1 MRD levels predict outcome after allogeneic hematopoietic stem cell transplantation in patients with acute myeloid leukemia. Blood Cancer J. 2016;6:e449.

8. Terwijn M, Putten WLJv, 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. 9. Maurillo L, Buccisano F, Del Principe MI, et al. Toward optimization of postremission therapy for residual disease-positive patients with acute myeloid leukemia. J Clin Oncol. 2008;26(30):4944-4951. 10. Kreso A, Dick John E. Evolution of the cancer stem cell model. Cell Stem Cell. 2014;14(3):275-291. 11. Lapidot T, Sirard C, Vormoor J, et al. A cell initiating human acute myeloid leukaemia after transplantation into SCID mice. Nature. 1994;367(6464):645-648. 12. Shlush LI, Mitchell A, Heisler L, et al. Tracing the origins of relapse in acute myeloid leukaemia to stem cells. Nature. 2017;547(7661):104-108.

haematologica | 2019; 104(8)

Promising targets for MRD therapies in AML 13. 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. 14. Frohling S, Schlenk RF, Breitruck J, et al. Prognostic significance of activating FLT3 mutations in younger adults (16 to 60 years) with acute myeloid leukemia and normal cytogenetics: a study of the AML Study Group Ulm. Blood. 2002;100(13):4372-4380. 15. Nagel G, Weber D, Fromm E, et al. Epidemiological, genetic, and clinical characterization by age of newly diagnosed acute myeloid leukemia based on an academic population-based registry study (AMLSG BiO). Ann Hematol. 2017;96(12):1993-2003. 16. Thiede C, Steudel C, Mohr B, et al. Analysis of FLT3-activating mutations in 979 patients with acute myelogenous leukemia: association with FAB subtypes and identification of subgroups with poor prognosis. Blood. 2002;99(12):4326-4335. 17. Kottaridis PD, Gale RE, Frew ME, et al. The presence of a FLT3 internal tandem duplication in patients with acute myeloid leukemia (AML) adds important prognostic information to cytogenetic risk group and response to the first cycle of chemotherapy: analysis of 854 patients from the United Kingdom Medical Research Council AML 10 and 12 trials. Blood. 2001;98(6):17521759. 18. Sengsayadeth SM, Jagasia M, Engelhardt BG, et al. Allo-SCT for high-risk AML-CR1 in the molecular era: impact of FLT3/ITD outweighs the conventional markers. Bone Marrow Transplant. 2012;47(12):1535-1537. 19. Perl A, Martinelli, G, Cortes, JE, et al. Gilteritinib significantly prolongs overall survival in patients with FLT3-mutated (FLT3mut+) relapsed/refractory (R/R) acute myeloid leukemia (AML): results from the phase III ADMIRAL trial Abstract #CCT184. Presented at the American Association for Cancer Research Annual Meeting 2019. 20. Levis M, Perl, AE, Altman, JK, et al. Impact of minimal residual disease and achievement of complete remission/complete remission with partial hematologic recovery (CR/CRh) on overall survival following treatment with gilteritinib in patients with relapsed/refractory (R/R) acute myeloid leukemia (AML) with FLT3 mutations. Blood. 2018;132. 21. Cortes J, Perl, AE, Dohner, H, et al. Quizartinib, an FLT3 inhibitor, as monotherapy in patients with relapsed or refractory acute myeloid leukaemia: an open-label, multicentre, single-arm, phase 2 trial. Lancet Oncol. 2018;19(7):889-903. 22. Propper DJ, McDonald AC, Man A, et al. Phase I and pharmacokinetic study of PKC412, an inhibitor of protein kinase C. J Clin Oncol. 2001;19(5):1485-1492. 23. 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. 24. Stone RM, Mandrekar SJ, Sanford BL, et al. The addition of midostaurin to standard chemotherapy decreases cumulative incidence of relapse (CIR) in the international prospective randomized, placebo-controlled, double-blind trial (CALGB 10603 / RATIFY [Alliance]) for newly diagnosed acute myeloid leukemia (AML) patients with FLT3 mutations. Blood. 2017;130 (Suppl 1):2580. 25. Smith CC, Lin K, Stecula A, Sali A, Shah NP. FLT3 D835 mutations confer differential

haematologica | 2019; 104(8)

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

resistance to type II FLT3 inhibitors. Leukemia. 2015;29(12):2390-2392. Wang E, Tallman M, Stone R, et al. Low relapse rate in younger patients â&#x2030;¤ 60 years old with newly diagnosed FLT3-mutated acute myeloid leukemia (AML) treated with crenolanib and cytarabine/anthracycline chemotherapy. ASH Annual Meeting 2017. Blood. 2017;130(Suppl 1):566-566. Stone R, Collins, R, Tallman, MS, et al. . Effect of cytarabine/anthracycline/ crenolanib induction on minimal residual disease (MRD) in newly diagnosed FLT3 mutant AML. ASCO Annual Meeting 2017. J Clin Oncol. 2017;35(15_suppl): 7016. Pratz KW, Cherry M, Altman JK, et al. updated results from a phase 1 study of gilteritinib in combination with induction and consolidation chemotherapy in subjects with newly diagnosed acute myeloid leukemia (AML). ASH Annual Meeting 2018. 2018;132(Suppl 1):564. Altman JK, Foran JM, Pratz KW, Trone D, Cortes JE, Tallman MS. Phase 1 study of quizartinib in combination with induction and consolidation chemotherapy in patients with newly diagnosed acute myeloid leukemia. Am J Hematol. 2018;93(2):213221. Medeiros BC, Fathi AT, DiNardo CD, Pollyea DA, Chan SM, Swords R. Isocitrate dehydrogenase mutations in myeloid malignancies. Leukemia. 2017;31(2):272-281. 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. 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. Ward PS, Patel J, Wise DR, et al. The common feature of leukemia-associated IDH1 and IDH2 mutations is a neomorphic enzyme activity converting alpha-ketoglutarate to 2-hydroxyglutarate. Cancer Cell. 2010;17(3):225-234. 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. 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. Losman JA, Looper RE, Koivunen P, et al. (R)-2-hydroxyglutarate is sufficient to promote leukemogenesis and its effects are reversible. Science. 2013;339(6127):16211625. Kats LM, Reschke M, Taulli R, et al. Protooncogenic role of mutant IDH2 in leukemia initiation and maintenance. Cell Stem Cell. 2014;14(3):329-341. 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. DiNardo CD, Stein EM, de Botton S, et al. Durable remissions with ivosidenib in IDH1-mutated relapsed or refractory AML. N Engl J Med. 2018;378(25):2386-2398. Stein EM, DiNardo CD, Pollyea DA, et al. Enasidenib in mutant IDH2 relapsed or

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

refractory acute myeloid leukemia. Blood. 2017;130(6):722-731. Amatangelo MD, Quek L, Shih A, et al. Enasidenib induces acute myeloid leukemia cell differentiation to promote clinical response. Blood. 2017;130(6):732-741. Stein E, DiNardo, CD, Fathi, AT, et al. Ivosidenib or enasidenib combined with induction and consolidation chemotherapy in patients with newly diagnosed AML with an IDH1 or IDH2 mutation is safe, effective, and leads to MRD-negative complete remissions. ASH Annual Meeting. Blood. 2018:132(Suppl 1). Schlenk RF, Weber D, Fiedler W, et al. Midostaurin added to chemotherapy and continued single-agent maintenance therapy in acute myeloid leukemia with FLT3-ITD. Blood. 2019;133(8):840. Mehta SV, Shukla SN, Vora HH. Overexpression of Bcl2 protein predicts chemoresistance in acute myeloid leukemia: its correlation with FLT3. Neoplasma. 2013;60(6):666-675. Luedtke DA, Niu X, Pan Y, et al. Inhibition of Mcl-1 enhances cell death induced by the Bcl-2-selective inhibitor ABT-199 in acute myeloid leukemia cells. Signal Transduct Target Ther. 2017;2:17012. Lagadinou ED, Sach A, Callahan K, et al. BCL-2 inhibition targets oxidative phosphorylation and selectively eradicates quiescent human leukemia stem cells. Cell Stem Cell. 2013;12(3):329-341. Pollyea DA, Stevens BM, Jones CL, et al. Venetoclax with azacitidine disrupts energy metabolism and targets leukemia stem cells in patients with acute myeloid leukemia. Nat Med. 2018;24(12):1859-1866. Souers AJ, Leverson JD, Boghaert ER, et al. ABT-199, a potent and selective BCL-2 inhibitor, achieves antitumor activity while sparing platelets. Nat Med. 2013;19(2):202208. Konopleva M, Pollyea DA, Potluri J, et al. Efficacy and biological correlates of response in a phase II study of venetoclax monotherapy in patients with acute myelogenous leukemia. Cancer Discov. 2016;6(10):11061117. Wei A, Strickland, SA, Roboz, GJ, et al. Phase 1/2 study of venetoclax with lowdose cytarabine in treatment-naive, elderly patients with acute myeloid leukemia unfit for intensive chemotherapy: 1-year outcomes. ASH Annual Meeting 2017. Blood. 2017;130(Suppl 1):890. DiNardo CD, Pratz KW, Letai A, et al. Safety and preliminary efficacy of venetoclax with decitabine or azacitidine in elderly patients with previously untreated acute myeloid leukaemia: a non-randomised, open-label, phase 1b study. Lancet Oncol. 2018;19(2):216-228. Pollyea D, Pratz, KW, Jonas, BA, et al. Venetoclax in combination with hypomethylating agents induces rapid, deep, and durable responses in patients with AML ineligible for intensive therapy ASH Annual Meeting 2018. Wei A, Strickland, SA, Hou, J, et al. Venetoclax with low-dose cytarabine induces rapid, deep, and durable responses in previously untreated older adults with AML ineligible for intensive chemotherapy ASH Annual Meeting 2018. DiNardo C, Albitar, M, Kadia, TM, et al. Venetoclax in combination with FLAG-IDA chemotherapy (FLAG-V-I) for fit, relapsed/refractory AML patients: interim results of a phase 1b/2 dose escalation and

1529

B. Ball and E.M. Stein et al.

55. 56.

57.

58.

59.

60.

61.

62. 63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

1530

expansion study. ASH Annual Meeting 2018. Blood. 2018;132(Suppl 1):4048. Kastenhuber ER, Lowe SW. Putting p53 in context. Cell. 2017;170(6):1062-1078. Zhao Z, Zuber J, Diaz-Flores E, et al. p53 loss promotes acute myeloid leukemia by enabling aberrant self-renewal. Genes Dev. 2010;24(13):1389-1402. Kadia TM, Jain P, Ravandi F, et al. TP53 mutations in newly diagnosed acute myeloid leukemia: clinicomolecular characteristics, response to therapy, and outcomes. Cancer. 2016;122(22):3484-3491. Hou HA, Chou WC, Kuo YY, et al. TP53 mutations in de novo acute myeloid leukemia patients: longitudinal follow-ups show the mutation is stable during disease evolution. Blood Cancer J. 2015;5:e331. Rucker FG, Schlenk RF, Bullinger L, et al. TP53 alterations in acute myeloid leukemia with complex karyotype correlate with specific copy number alterations, monosomal karyotype, and dismal outcome. Blood. 2012;119(9):2114-2121. Bueso-Ramos CE, Yang Y, deLeon E, McCown P, Stass SA, Albitar M. The human MDM-2 oncogene is overexpressed in leukemias. Blood. 1993;82(9):2617-2623. Li L, Tan Y, Chen X, et al. MDM4 overexpressed in acute myeloid leukemia patients with complex karyotype and wild-type TP53. PLoS One. 2014;9(11):e113088. Karni-Schmidt O, Lokshin M, Prives C. The roles of MDM2 and MDMX in cancer. Annu Rev Pathol. 2016;11:617-644. Ding Q, Zhang Z, Liu JJ, et al. Discovery of RG7388, a potent and selective p53-MDM2 inhibitor in clinical development. J Med Chem. 2013;56(14):5979-5983. Goldberg A, Horvat, TZ, Hsu, M, et al. Venetoclax combined with either a hypomethylating agent or low-dose cytarabine shows activity in relapsed and refractory myeloid malignancies. ASH Annual Meeting 2017. Blood. 2017;130(Suppl 1):1353. Aldoss I, Yang D, Aribi A, et al. Efficacy of the combination of venetoclax and hypomethylating agents in relapsed/refractory acute myeloid leukemia. Haematologica. 2018;103(9):e404-e407. Lambert JM, Gorzov P, Veprintsev DB, et al. PRIMA-1 reactivates mutant p53 by covalent binding to the core domain. Cancer Cell. 2009;15(5):376-388. Zhang Q, Bykov VJN, Wiman KG, Zawacka-Pankau J. APR-246 reactivates mutant p53 by targeting cysteines 124 and 277. Cell Death Dis. 2018;9(5):439. Sallman D, DeZern, AE, Steensma, DP, et al. Phase 1b/2 combination study of APR-246 and azacitidine (AZA) in patients with TP53 mutant myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML). ASH Annual Meeting 2018. Fenaux P, Mufti GJ, Hellstrom-Lindberg E, et al. Efficacy of azacitidine compared with that of conventional care regimens in the treatment of higher-risk myelodysplastic syndromes: a randomised, open-label, phase III study. Lancet Oncol. 2009;10(3):223-232. Kantarjian H, Issa JP, Rosenfeld CS, et al. Decitabine improves patient outcomes in myelodysplastic syndromes: results of a phase III randomized study. Cancer. 2006;106(8):1794-1803. Dombret H, Seymour JF, Butrym A, et al. International phase 3 study of azacitidine vs conventional care regimens in older patients with newly diagnosed AML with >30% blasts. Blood. 2015;126(3):291-299. Hollenbach PW, Nguyen AN, Brady H, et al.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83. 84.

85.

86.

A comparison of azacitidine and decitabine activities in acute myeloid leukemia cell lines. PLoS One. 2010;5(2):e9001. Ball B, Zeidan A, Gore SD, Prebet T. Hypomethylating agent combination strategies in myelodysplastic syndromes: hopes and shortcomings. Leuk Lymphoma. 2017;58(5):1022-1036. Wolff F, Leisch M, Greil R, Risch A, Pleyer L. The double-edged sword of (re)expression of genes by hypomethylating agents: from viral mimicry to exploitation as priming agents for targeted immune checkpoint modulation. Cell Commun Signal. 2017;15(1):13. Almstedt M, Blagitko-Dorfs N, DuqueAfonso J, et al. The DNA demethylating agent 5-aza-2'-deoxycytidine induces expression of NY-ESO-1 and other cancer/testis antigens in myeloid leukemia cells. Leuk Res. 2010;34(7):899-905. Atanackovic D, Luetkens T, Kloth B, et al. Cancer-testis antigen expression and its epigenetic modulation in acute myeloid leukemia. Am J Hematol. 2011;86(11):918922. Fonsatti E, Nicolay HJ, Sigalotti L, et al. Functional up-regulation of human leukocyte antigen class I antigens expression by 5aza-2'-deoxycytidine in cutaneous melanoma: immunotherapeutic implications. Clin Cancer Res. 2007;13(11):33333338. Chiappinelli KB, Strissel PL, Desrichard A, et al. Inhibiting DNA methylation causes an interferon response in cancer via dsRNA including endogenous retroviruses. Cell. 2015;162(5):974-986. Roulois D, Loo Yau H, Singhania R, et al. DNA-demethylating agents target colorectal cancer cells by inducing viral mimicry by endogenous transcripts. Cell. 2015;162(5): 961-973. Yang H, Bueso-Ramos C, DiNardo C, et al. Expression of PD-L1, PD-L2, PD-1 and CTLA4 in myelodysplastic syndromes is enhanced by treatment with hypomethylating agents. Leukemia. 2014;28(6):1280-1288. Platzbecker U, Middeke JM, Sockel K, et al. Measurable residual disease-guided treatment with azacitidine to prevent haematological relapse in patients with myelodysplastic syndrome and acute myeloid leukaemia (RELAZA2): an open-label, multicentre, phase 2 trial. Lancet Oncol. 2018;19(12):1668-1679. Khalil DN, Smith EL, Brentjens RJ, Wolchok JD. The future of cancer treatment: immunomodulation, CARs and combination immunotherapy. Nat Rev Clin Oncol. 2016;13(5):273-290. Ribas A, Wolchok JD. Cancer immunotherapy using checkpoint blockade. Science. 2018;359(6382):1350-1355. Berger R, Rotem-Yehudar R, Slama G, et al. Phase I safety and pharmacokinetic study of CT-011, a humanized antibody interacting with PD-1, in patients with advanced hematologic malignancies. Clin Cancer Res. 2008;14(10):3044-3051. Davids MS, Kim HT, Bachireddy P, et al. Ipilimumab for patients with relapse after allogeneic transplantation. N Engl J Med. 2016;375(2):143-153. Assi R, Kantarjian, HM, Daver, NG, et al. Results of a phase 2, open-label study of idarubicin (I), cytarabine (A) and nivolumab (Nivo) in patients with newly diagnosed acute myeloid leukemia (AML) and highrisk myelodysplastic syndrome (MDS). ASH Annual Meeting 2018. Blood. 2018;132 (Suppl 1):905.

87. Daver N, Garcia-Manero G, Basu S, et al. Efficacy, safety, and biomarkers of response to azacitidine and nivolumab in relapsed/refractory acute myeloid leukemia: a nonrandomized, open-label, phase II study. Cancer Discov. 2019;9(3):370-383. 88. Weinstock M, Rosenblatt J, Avigan D. Dendritic cell therapies for hematologic malignancies. Mol Ther Methods Clin Dev. 2017;5:66-75. 89. Sabado RL, Balan S, Bhardwaj N. Dendritic cell-based immunotherapy. Cell Res. 2017;27(1):74-95. 90. Caux C, Vanbervliet B, Massacrier C, et al. B70/B7-2 is identical to CD86 and is the major functional ligand for CD28 expressed on human dendritic cells. J Exp Med. 1994;180(5):1841-1847. 91. Anguille S, Van de Velde AL, Smits EL, et al. Dendritic cell vaccination as postremission treatment to prevent or delay relapse in acute myeloid leukemia. Blood. 2017;130( 15):1713-1721. 92. Rosenblatt J, Stone RM, Uhl L, et al. Individualized vaccination of AML patients in remission is associated with induction of antileukemia immunity and prolonged remissions. Sci Transl Med. 2016;8(368): 368ra171. 93. Andrews RG, Takahashi M, Segal GM, Powell JS, Bernstein ID, Singer JW. The L4F3 antigen is expressed by unipotent and multipotent colony-forming cells but not by their precursors. Blood. 1986;68(5):1030-1035. 94. Appelbaum FR, Bernstein ID. Gemtuzumab ozogamicin for acute myeloid leukemia. Blood. 2017;130(22):2373. 95. Walter RB, Appelbaum FR, Estey EH, Bernstein ID. Acute myeloid leukemia stem cells and CD33-targeted immunotherapy. Blood. 2012;119(26):6198-6208. 96. Krupka C, Kufer P, Kischel R, et al. CD33 target validation and sustained depletion of AML blasts in long-term cultures by the bispecific T-cell-engaging antibody AMG 330. Blood. 2014;123(3):356-365. 97. Khan N, Hills RK, Virgo P, et al. Expression of CD33 is a predictive factor for effect of gemtuzumab ozogamicin at different doses in adult acute myeloid leukaemia. Leukemia. 2017;31(5):1059-1068. 98. Hamann PR, Hinman LM, Beyer CF, et al. An anti-CD33 antibody-calicheamicin conjugate for treatment of acute myeloid leukemia. Choice of linker. Bioconjug Chem. 2002;13(1):40-46. 99. Bross PF, Beitz J, Chen G, et al. Approval summary: gemtuzumab ozogamicin in relapsed acute myeloid leukemia. Clin Cancer Res. 2001;7(6):1490-1496. 100. Petersdorf SH, Kopecky KJ, Slovak M, et al. A phase 3 study of gemtuzumab ozogamicin during induction and postconsolidation therapy in younger patients with acute myeloid leukemia. Blood. 2013;121(24): 4854-4860. 101. Castaigne S, Pautas C, Terre C, et al. Effect of gemtuzumab ozogamicin on survival of adult patients with de-novo acute myeloid leukaemia (ALFA-0701): a randomised, open-label, phase 3 study. Lancet. 2012;379(9825):1508-1516. 102. Burnett AK, Hills RK, Milligan D, et al. Identification of patients with acute myeloblastic leukemia who benefit from the addition of gemtuzumab ozogamicin: results of the MRC AML15 trial. J Clin Oncol. 2011;29(4):369-377. 103. Burnett AK, Russell NH, Hills RK, et al. Addition of gemtuzumab ozogamicin to induction chemotherapy improves survival in older patients with acute myeloid leukemia. J

haematologica | 2019; 104(8)

Promising targets for MRD therapies in AML Clin Oncol. 2012;30(32):3924-3931. 104. Hills RK, Castaigne S, Appelbaum FR, et al. Addition of gemtuzumab ozogamicin to induction chemotherapy in adult patients with acute myeloid leukaemia: a metaanalysis of individual patient data from randomised controlled trials. Lancet Oncol. 2014;15(9):986-996. 105. Burnett A, Cavenagh J, Russell N, et al. Defining the dose of gemtuzumab ozogamicin in combination with induction chemotherapy in acute myeloid leukemia: a comparison of 3 mg/m2 with 6 mg/m2 in the NCRI AML17 Trial. Haematologica. 2016;101(6):724-731. 106. Lambert J, Lambert J, Nibourel O, et al. MRD assessed by WT1 and NPM1 transcript levels identifies distinct outcomes in AML patients and is influenced by gemtuzumab ozogamicin. Oncotarget. 2014;5 (15):6280-6288. 107. Laszlo GS, Gudgeon CJ, Harrington KH, et al. Cellular determinants for preclinical activity of a novel CD33/CD3 bispecific Tcell engager (BiTE) antibody, AMG 330, against human AML. Blood. 2014;123(4): 554-561. 108. Ravandi F, Stein, AS, Kantarjian, HM, et al. A Phase 1 first-in-human study of AMG 330, an anti-CD33 bispecific T-cell engager (BiTE®) antibody construct, in relapsed/refractory acute myeloid leukemia (R/R AML) ASH Annual Meeting 2018. 109. Munoz L, Nomdedeu JF, Lopez O, et al. Interleukin-3 receptor alpha chain (CD123) is widely expressed in hematologic malignancies. Haematologica. 2001;86(12):12611269. 110. Jordan CT, Upchurch D, Szilvassy SJ, et al. The interleukin-3 receptor alpha chain is a unique marker for human acute myelogenous leukemia stem cells. Leukemia. 2000;14(10):1777-1784. 111. Kovtun Y, Jones GE, Adams S, et al. A

haematologica | 2019; 104(8)

CD123-targeting antibody-drug conjugate, IMGN632, designed to eradicate AML while sparing normal bone marrow cells. Blood advances. 2018;2(8):848-858. 112. Daver NG, Erba HP, Papadantonakis N, et al. A phase I, first-in-human study evaluating the safety and preliminary antileukemia activity of IMGN632, a novel CD123-targeting antibody-drug conjugate, in patients with relapsed/refractory acute myeloid leukemia and other CD123-positive hematologic malignancies. ASH Annual Meeting 2018. Blood. 2018;132(Suppl 1):27. 113. Frankel AE, Woo JH, Ahn C, et al. Activity of SL-401, a targeted therapy directed to interleukin-3 receptor, in blastic plasmacytoid dendritic cell neoplasm patients. Blood. 2014;124(3):385-392. 114. Pemmaraju N, Lane AA, Sweet KL, et al. Tagraxofusp in blastic plasmacytoid dendritic-cell neoplasm. 2019;380(17):1628-1637. 115. van Rhenen A, van Dongen GAMS, Kelder A, et al. The novel AML stem cell–associated antigen CLL-1 aids in discrimination between normal and leukemic stem cells. Blood. 2007;110(7):2659. 116. Van Loo PF, Doornbos R, Dolstra H, Shamsili S, Bakker L. Preclinical evaluation of MCLA117, a CLEC12AxCD3 bispecific antibody efficiently targeting a novel leukemic stem cell associated antigen in AML. Blood. 2015;126(23):325. 117. June CH, Sadelain M. Chimeric antigen receptor therapy. N Engl J Med. 2018;379 (1):64-73. 118. Ritchie DS, Neeson PJ, Khot A, et al. Persistence and efficacy of second generation CAR T cell against the LeY antigen in acute myeloid leukemia. Mol Ther. 2013;21(11):2122-2129. 119. Kenderian SS, Ruella M, Shestova O, et al. CD33-specific chimeric antigen receptor T cells exhibit potent preclinical activity against human acute myeloid leukemia.

Leukemia. 2015;29(8):1637-1647. 120. Pizzitola I, Anjos-Afonso F, Rouault-Pierre K, et al. Chimeric antigen receptors against CD33/CD123 antigens efficiently target primary acute myeloid leukemia cells in vivo. Leukemia. 2014;28(8):1596-1605. 121. Gill S, Tasian SK, Ruella M, et al. Preclinical targeting of human acute myeloid leukemia and myeloablation using chimeric antigen receptor-modified T cells. Blood. 2014;123(15):2343-2354. 122. Tashiro H, Sauer T, Shum T, et al. Treatment of acute myeloid leukemia with T cells expressing chimeric antigen receptors directed to C-type lectin-like molecule 1. Mol Ther. 2017;25(9):2202-2213. 123. Perna F, Berman SH, Soni RK, et al. Integrating proteomics and transcriptomics for systematic combinatorial chimeric antigen receptor therapy of AML. Cancer Cell. 2017;32(4):506-519 e505. 124. Kim MY, Yu K-R, Kenderian SS, et al. Genetic inactivation of CD33 in hematopoietic stem cells to enable CAR T cell immunotherapy for acute myeloid leukemia. Cell. 2018;173(6):1439-1453.e1419. 125. Khoury HJ, Collins RH, Jr., Blum W, et al. Immune responses and long-term disease recurrence status after telomerase-based dendritic cell immunotherapy in patients with acute myeloid leukemia. Cancer. 2017;123(16):3061-3072. 126. van de Loosdrecht AA, van Wetering S, Santegoets S, et al. A novel allogeneic offthe-shelf dendritic cell vaccine for postremission treatment of elderly patients with acute myeloid leukemia. Cancer Immunol Immunother. 2018;67(10):1505-1518. 127. Lambert J, Pautas C, Terré C, et al. Gemtuzumab ozogamicin for de novo acute myeloid leukemia: final efficacy and safety updates from the open-label, phase III ALFA-0701 trial. Haematologica. 2019;104 (1):113.

1531

REVIEW ARTICLE Ferrata Storti Foundation

How close are we to incorporating measurable residual disease into clinical practice for acute myeloid leukemia? Nicholas J. Short and Farhad Ravandi

Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA

Haematologica 2019 Volume 104(8):1532-1541

ABSTRACT

Correspondence: FARHAD RAVANDI fravandi@mdanderson.org Received: February 8, 2019. Accepted: June 5, 2019. Pre-published: July 4, 2019.

ssessment of measurable residual disease, also called “minimal residual disease,” in patients with acute myeloid leukemia in morphological remission provides powerful prognostic information and complements pretreatment factors such as cytogenetics and genomic alterations. Based on data that low levels of persistent or recurrent residual leukemia are consistently associated with an increased risk of relapse and worse longterm outcomes, its routine assessment has been recommended by some experts and consensus guidelines. In addition to providing important prognostic information, the detection of measurable residual disease may also theoretically help to determine the optimal post-remission strategy for an individual patient. However, the full therapeutic implications of measurable residual disease are uncertain and thus controversy exists as to whether it should be routinely incorporated into clinical practice. While some evidence supports the use of allogeneic stem cell transplantation or hypomethylating agents for some subgroups of patients in morphological remission but with detectable residual leukemia, the appropriate use of this information in making clinical decisions remains largely speculative at present. To resolve this pressing clinical issue, several ongoing studies are evaluating measurable residual disease-directed treatments in acute myeloid leukemia and may lead to new, effective strategies for patients in these circumstances. This review examines the common technologies used in clinical practice and in the research setting to detect residual leukemia, the major clinical studies establishing the prognostic impact of measurable residual disease in acute myeloid leukemia, and the potential ways, both now and in the future, that such testing may rationally guide therapeutic decision-making.

doi:10.3324/haematol.2018.208454

Introduction

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1532

Acute myeloid leukemia (AML) is a heterogeneous disease, with a widely variable likelihood of cure with conventional therapies that depends on both patientand disease-related characteristics.1 Historically, pretreatment characteristics such as the patients’ age, karyotype, and, more recently, genetic mutations have been the primary determinants of prognosis in patients with newly diagnosed AML. How well the leukemia responds to initial treatment also provides important information about chemosensitivity of an individual’s leukemia that cannot always be predicted from pretreatment characteristics. One measure of treatment response is the achievement of complete remission, defined as bone marrow with <5% blasts with recovery of peripheral blood elements and no evidence of extramedullary disease.2 However, among patients <60 years of age who achieve complete remission with induction and then receive consolidation chemotherapy, the cure rate is only approximately 50% (and significantly lower in older patients), suggesting that morphological assessment of the bone marrow alone is inadequate to discriminate relapse risk.3 Accurate assessment of the risk of relapse is imperative in the treatment of AML, as the decision to pursue consolidation chemotherapy rather than allogeneic hematopoietic stem cell transplantation (HSCT) in first remission is largely decided by the anticipated risk of relapse in the absence of HSCT balanced against the risk of relapse after HSCT as well as the expected transplant-related

©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.

1532

haematologica | 2019; 104(8)

MRD in AML

mortality.4 Although improvements in genomic classification of AML have refined our current risk stratification systems, relapses are unfortunately still common even with risk-adapted treatment. Assessment of measurable residual disease (MRD), also called “minimal residual disease,” allows for the detection and quantification of lower levels of residual leukemia that can be detected by morphological assessment alone.5,6 In both pediatric and adult acute lymphoblastic leukemia, MRD is routinely used to refine prognostic assessment and also allocate post-remission therapies, particularly in the pediatric population.7-11 Similarly, studies utilizing various methods of MRD determination have consistently shown that MRD is highly prognostic in AML and complements (and sometimes supersedes) historically relevant pretreatment characteristics.12-18 However, many questions remain as to the optimal MRD assay, appropriate timing of MRD assessment, and whether MRD should guide treatment. Herein, we review methods of MRD detection and the prognostic impact of MRD across AML subtypes. We also offer some practical considerations for MRD testing in the clinical setting and review the available data on how MRD can potentially guide post-remission treatment decisions in AML.

Methods of measurable residual disease assessment Several methods of MRD assessment are available in the clinical and research settings, each of which has its own advantages and disadvantages. The European

LeukemiaNet MRD Working Party has released comprehensive, consensus recommendations regarding their appropriate use and technical limitations.6 These various MRD methods vary in their level of sensitivity, applicability to the vast majority of patients with AML versus only certain subtypes, cost, and level of technical expertise needed to obtain an accurate result, all of which may influence the choice of assay for a particular patient.19 The major differences among MRD technologies are summarized in Table 1.

Karyotype analysis and fluorescence in situ hybridization In patients with an abnormal pretreatment karyotype, the persistence of an abnormal leukemia-associated karyotype at remission suggests the presence of residual disease. Persistent cytogenetic abnormalities have been associated with worse survival in several studies and may also identify patients who could benefit from HSCT in first remission.20,21 While cytogenetic analysis at remission adds prognostic information in patients who achieve complete remission, the sensitivity of this method is relatively poor, as it can only detect one abnormal metaphase out of 20 (i.e. sensitivity ≈5%). Furthermore, up to 50% of adult patients present with cytogenetically normal AML, further limiting karyotype analysis as a universal MRD marker.1 Fluorescence in situ hybridization may also be used to detect persistent cytogenetic abnormalities with slightly increased sensitivity compared to conventional cytogenetics. However, the maximum sensitivity achieved with fluorescence in situ hybridization is approximately 1%, which is not adequate to detect low levels of clinically relevant MRD.22

Table 1. Methods of measurable residual disease assessment in acute myeloid leukemia.

Method Conventional karyotyping

Sensitivity ~5%

Advantages

Disadvantages

• Common in routine clinical practice

• Poor sensitivity • Time-consuming and labor-intensive • Applicable only to patients with baseline abnormal karyotype (~50%) • Worse sensitivity than MFC or PCR • Applicable only to patients with baseline abnormal karyotype (~50%) • Potential for immunophenotypic shifts (mitigated by using a DfN-based approach) • Requires significant technical expertise to interpret • Limited standardization across laboratories • Appropriate molecular targets present in <50% of cases (<35% in older adults) • Many mutations are not suitable for MRD detection (e.g. FLT3) • Time-consuming and labor-intensive • Results may take several days • Low sensitivity with most commonly used platforms • May be confounded by persistence of preleukemic mutations (e.g. CHIP) • Results may take several days • Expensive • Not standardized • Requires complex bioinformatics

FISH

Up to 10-2

• Useful for numeric cytogenetic abnormalities (i.e. gains or deletions)

MFC for LAIP or DfN

10-3 to 10-5

• • • •

RT-qPCR

10-4 to 10-6

• Sensitive • Well standardized • Can be run in any laboratory with RT-qPCR capabilities

NGS

Highly variable (1% to 10-6)

Sensitive Fast (results usually available within 24 hours) Relatively inexpensive Applicable to >90% of AML cases

• Potential for very high sensitivity (depending on technology) • Can test multiple genes at once

FISH: fluorescence in situ hybridization; MFC: multiparameter flow cytometry; PCR: polymerase chain reaction; LAIP: leukemia-associated immunophenotypes; DfN: difference from normal; AML: acute myelod leukemia; RT-qPCR: real-time quantitative polymerase chain reaction; NGS: next-generation sequencing; MRD: measurable residual disease; CHIP: clonal hematopoiesis of indeterminate potential.

haematologica | 2019; 104(8)

1533

N.J. Short and F. Ravandi et al.

Multiparameter flow cytometry Multiparameter flow cytometry (MFC) uses a panel of fluorochrome-labeled monoclonal antibodies to identify aberrantly expressed antigens on leukemic blasts. This MFC-based MRD analysis may be accomplished through either the tracking of leukemia-associated immunophenotypes in the pretreatment and remission samples or the use of “difference from normal” analysis.22 Leukemia-associated immunophenotypes consist of the aberrant expression of antigens compared to that on normal myeloid precursors, cross-lineage antigen expression (e.g. expression of lymphoid antigens on myeloblasts), over- or underexpression of antigens normally expressed, and aberrant co-expression of antigens normally found in early or late hematopoietic differentiation.6 In contrast, the “difference from normal” approach is used to detect any differences in the remission immunophenotype compared to the highly stereotypical normal immunophenotype distribution.23 The advantage of “difference from normal” analysis is that it does not require knowledge of the diagnostic immunophenotype; furthermore, it may also be less susceptible to immunophenotypic shifts that can occur as a direct result of therapy or due to a shift in clonal architecture.24,25 Workflows that incorporate both of these methods (i.e., a leukemia-associated immunophenotype-based, “difference from normal” approach) may help to further optimize MRD assessment.6 Some studies suggest that addition of leukemia stem cell markers (e.g. CLL1, CD44, CD123, and CD184, among others) to flow MRD antibody panels may add additional prognostic information to standard MFC-based MRD, particularly by identifying those patients at very high risk of relapse (i.e. those who are positive for both MRD and leukemia stem cell markers).26-28 MFC-based MRD assessment can achieve a sensitivity of 10-3 to 10-5, which is dependent on the number of cells analyzed, gating method, and number of antibody colors used; in most cases, a sensitivity of 10-4 is achieved. Compared to real-time quantitative polymerase chain reaction (RT-qPCR), MFC is significantly faster and less labor-intensive. It also has the advantage of being applicable to more than 90% of patients with AML, unlike other methods that rely on specific genetic or molecular targets.22 Despite these advantages, the interpretation of MFC MRD is not standardized in most countries, including the USA, and requires significant technical expertise on the part of the interpreting pathologist, which can lead to inter-laboratory discordance. Because of the complexity of interpreting flow-based MRD, there is interest in the use of machine-based learning artificial intelligence to reduce the potential for bias or other subjective errors in MRD interpretation. Such artificial intelligence-based algorithms are promising and may result in more clinically significant MRD results, although further validation of this technology is needed.29

Polymerase chain reaction RT-qPCR can be used to monitor recurrent genomic alterations in certain subtypes of AML. To be a useful marker for PCR-based MRD assessment, the target gene fusion or mutation should be stable throughout the disease course and its presence should reflect true persistent disease (rather than preleukemic subclones). Suitable targets that have been evaluated in large studies include PML-RARA in acute promyelocytic leukemia, CBFB1534

MYH11 and RUNX1-RUNX1T1 in core-binding factor AML, and mutant NPM1.15,30-36 In contrast, other mutations may emerge or disappear at the time of relapse (e.g., mutant FLT337,38) and are therefore generally unreliable MRD markers for assessing meaningful “MRD negativity,” although their persistence likely represents residual disease in most cases. One disadvantage of PCR is that appropriate, validated targets are present in less than 50% of patients with AML, and the incidence of these AML subtypes declines substantially with increasing age at diagnosis.1 To overcome this limitation, attempts have been made to track residual disease using markers that are expressed at significantly higher levels in leukemic blasts than in normal hematopoietic cells. For example, several studies have evaluated monitoring levels of WT1 or EVI1 mRNA transcripts over the course of treatment as a marker of MRD.39-42 While these targets may provide some prognostic information, they are generally not specific enough for residual leukemia to be used in routine practice.6 PCR-based MRD assessment has the advantage of being highly sensitive (sensitivity ≈ 10-4 to 10-6, depending on the input of RNA/DNA) and is generally better standardized than MFC.19 To further refine relapse risk, ultrasensitive digital droplet PCR technologies have been developed in order to detect low levels of residual gene mutations.43 In contrast to standard PCR, digital droplet PCR does not require calibration standards, and is thus faster and more precise and reproducible.44 Due to its absolute quantification of DNA copy numbers, it can also provide information about clonality and subclonality. This technology may have greater sensitivity and be better able to quantify very low levels of MRD than standard PCR22; however, whether this added sensitivity will translate into more accurate prognostic discrimination is yet to be determined.

Next-generation sequencing Targeted next-generation sequencing (NGS) panels are commonly used at the time of diagnosis to identify prognostic gene mutations or mutations that may be therapeutically targeted (e.g. FLT3 or IDH1/2 mutations). Several studies have also evaluated re-using similar NGS panels at the time of remission to assess the relationship between decline in mutational burden and clinical outcomes.17,45,46 This methodology relies on a similar principal as PCR regarding the tracking of genomic or molecular targets, although NGS is able to target multiple genes at once, or even the entire genome, if desired. Although there is understandable excitement about the potential for high-throughput NGS-based MRD monitoring in AML, there are several practical considerations that limit its clinical use at present. NGS MRD assessment is relatively expensive, is not standardized, and requires complicated bioinformatics. The interpretation of NGS MRD results is further complicated by the presence of preleukemic clones that may not fully clear even in patients who achieve deep, long-term remissions with chemotherapy, such as mutations associated with clonal hematopoiesis of indeterminate potential (CHIP).47-49 CHIP mutations, particularly DNMT3A, TET2, and ASXL1, commonly persist in patients who do not relapse, suggesting that they should not be routinely used as MRD markers.17,46 Furthermore, at present, the sensitivity of NGS for MRD assessment is generally ~1% with most commonly haematologica | 2019; 104(8)

MRD in AML

used platforms, due largely to the intrinsic error rate of sequencing. This lack of sensitivity is often because the assays are designed for identification of mutations in baseline samples (for which high sensitivity is generally not needed) and then repurposed for assessment of MRD (for which adequate sensitivity is imperative). However, depending on the NGS platform used and the amount of input DNA, NGS can theoretically achieve a sensitivity of 10-6, making it an attractive potential option for very sensitive MRD detection. Advances in NGS technologies, including molecular barcoding and duplex sequencing, may improve the sensitivity of NGS and allow for the identification of very low levels of residual leukemia.50,51

Prognostic Impact of measurable residual disease in acute myeloid leukemia Achievement of MRD negativity has been shown to be a powerful prognostic factor in numerous studies of patients undergoing frontline AML therapy. Based on the consistent impact of MRD on long-term outcomes across multiple studies and AML subtypes, consensus guidelines from the European LeukemiaNet support “complete remission without MRD” as an official AML response criterion.2 There are also ongoing efforts in the USA and elsewhere to validate MRD as a surrogate endpoint for accelerated regulatory approval of novel drugs and combinations. Here we review some of the major studies supporting the clinical use of MRD to provide prognostic and predictive information in AML.

Multiparameter flow cytometry-based measurable residual disease studies MRD as assessed by MFC is highly prognostic when measured at various time points, including after induction, during and after consolidation, and peri-transplant.12,13,16,18,5256 While studies have used varied cutoffs to define MRD “negativity,” many have used levels of ≤0.1% to define negativity, both because this level of sensitivity can be reliably achieved with flow-based assays and because this level appears to provide the best discrimination for relapse risk at most time points.13,56 However, it is important to note that some studies have also suggested that even levels of residual leukemia below the 0.1% threshold may be associated with worse outcomes than lack of detectable MRD.12,53 Several studies in younger adults ≤60 years of age with newly diagnosed AML undergoing standard frontline chemotherapy have shown that achievement of MRD negativity is associated with a significantly lower risk of relapse and better survival.13,16,18 The impact of MRD status consistently adds additional prognostic information beyond that provided by pretreatment characteristics, such as cytogenetics and genetic mutations. Notably, most studies have evaluated MRD in younger adults and the data regarding the impact of MRD in patients >60 years are relatively scant. However, the available data suggest that achievement of MRD negativity in the older population may be predictive for lower relapse risk whether the patients are treated with intensive chemotherapy or hypomethylating agents.12,57 Because most older patients with AML will not be candidates for allogeneic HSCT, the potential therapeutic implications of persistent MRD in these patients is less clear than in their younger counterparts. haematologica | 2019; 104(8)

The National Cancer Research Institute AML17 trial is the largest study of MRD in AML to date and provides some key insights into the impact of MRD in AML.18 This study enrolled 1,874 adults <60 years of age with AML who received standard daunorubicin plus cytarabinebased induction, followed by risk-adapted chemotherapy consolidation, with or without allogeneic HSCT. Patients with MRD positivity (defined as ≥0.1% by MFC assay) after cycle 1 had similar 5-year overall survival as patients who only achieved a partial response (51% vs. 46%, respectively), highlighting the poor outcomes associated with persistent MRD. Given the large size of this population, subgroup analyses were possible, including those among patients with standard-risk AML without NPM1 mutations. Of note, this is a particularly important group to refine prognosis using MRD, since there is controversy as to whether standard-risk patients (as determined by pretreatment characteristics) should routinely undergo HSCT in first remission; it is also a population in which MFC-based MRD assessment plays an integral role given the absence of a reliable molecular MRD target (e.g. mutant NPM1).1,22 Among patients with standard-risk AML without NPM1 mutation, the 5-year overall survival rates for patients who achieved MRD negativity after two cycles and for patients who remained MRD-positive were 63% and 33%, respectively (P=0.003), with the difference being driven by a significantly higher relapse rate in the MRD-positive group. Thus, MRD status after induction or in early consolidation may be able to stratify standard-risk patients according to risk of relapse and, consequently, help to determine the potential benefit from consolidative allogeneic HSCT. Detectable MRD immediately prior to HSCT was also associated with increased risk of post-HSCT relapse and worse overall survival in several studies.52-55 These observations are further supported by a meta-analysis of 19 studies evaluating pre-HSCT MRD (the majority of which assessed MRD by MFC).58 In one study of patients with AML who underwent HSCT, those in complete remission but with detectable MRD had a similar post-HSCT relapse rate and overall survival as those transplanted when not in morphological remission.54 The 3-year overall survival rates for patients who were in complete remission but MRD-positive versus those with active AML at the time of HSCT were 26% and 23%, respectively, compared to 73% for patients who were MRD-negative. The impact of post-HSCT MRD status is, however, less clear.55,59 While patients with persistent or recurrent MRD after HSCT appear to have relatively poor outcomes, one analysis suggests that pre-HSCT MRD status carries relatively more prognostic information than post-HSCT MRD status.55 In this study, only pre-HSCT MRD was independently associated with long-term outcomes. The 3-year overall survival rate for patients with pre-HSCT MRD that persisted after the transplant was 19% and only slightly higher for those with pre-HSCT MRD that cleared with transplantation (3-year overall survival rate: 29%).

Polymerase chain reaction-based measurable residual disease studies The prognostic impact of MRD detected by PCR has been shown primarily in acute promyelocytic leukemia, core-binding factor leukemias, and NPM1-mutated AML, as the target alterations in these AML subtypes represent stable, foundational genomic lesions that become unde1535

N.J. Short and F. Ravandi et al.

tectable in long-term survivors (with rare exceptions60,61) and persist or re-emerge in the setting of relapsed disease.19 In acute promyelocytic leukemia, the persistence of PML-RARA transcripts strongly predicts for relapse, and patients with persistent transcripts invariably relapse without intervention.30,31,62 Among patients with acute promyelocytic leukemia treated with chemotherapy plus all-trans retinoic acid, initiation of arsenic trioxide for those with positive MRD prevents morphological relapse in the majority of cases.31 However, in this context, it is important to confirm a positive MRD assessment with repeat testing, given the potential for false positives. Consensus guidelines have therefore defined molecular remission as an important therapeutic milestone and recommend molecular assessment in routine clinical practice.63 The link between MRD status and prognosis, as well as an effective salvage therapy for MRD-positive disease, makes acute promyelocytic leukemia a model for MRD-guided therapies, although it is unclear whether routine, universal monitoring is still required in the era of regimens capable of curing >95% of non-high-risk patients.64,65 Core-binding factor AML accounts for 20-25% of cases of AML in younger patients, but <10% in older adults.2,66 PCR-based MRD assessment of CBFB-MYH11 and RUNX1-RUNX1T1 transcripts for patients with inv(16) and t(8;21), respectively, has been consistently associated with risk of relapse across several studies.32,33,67-70 In the largest study of core-binding factor AML and MRD, 278 patients [163 with t(8;21) and 115 with inv(16)] were treated.33 A >3 log reduction in RUNX1-RUNX1T1 in the bone marrow and CBFB-MYH11 copy number >10 in the peripheral blood after induction were the most prognostic measures of MRD for risk of relapse. Levels of MRD above these thresholds were associated with overall survival in the t(8;21) group but not in the inv(16) group, possibly due to better salvage options fpr patients with the latter subtype. Thresholds of MRD were identified that could predict relapse in 100% of patients in both groups and, notably, rising MRD levels on sequential samples predicted morphological relapse, suggesting a potential role for pre-emptive therapy in this population. Monitoring of mutant NPM1 throughout treatment may also provide important prognostic information in AML.15,34-36 Mutations in NPM1 are present in approximately 30% of younger patients with AML and in up to 60% of cases of AML with normal karyotype.71 Like PMLRARA, CBFB-MYH11 and RUNX1-RUNX1T1, these mutations are stable at diagnosis and relapse, making them ideal targets for molecular MRD assessment.72 In the largest study of NPM1-based MRD (the National Cancer Research Institute AML17 trial), 346 patients with NPM1mutated AML received intensive chemotherapy and were monitored for mutant NPM1 levels by PCR in the bone marrow and peripheral blood after each cycle of chemotherapy.15 With a median sensitivity of 10-5, MRD negativity in the peripheral blood after two cycles of chemotherapy was achieved by approximately 85% of patients and was the most prognostic MRD measure. Lack of achievement of this MRD milestone was the only independent prognostic factor for death in multivariate analysis [hazard ratio (HR) 4.84; 95% confidence interval (95% CI): 2.57-9.15; P<0.001). When MRD was assessed in the peripheral blood at this time point, the 3-year overall survival rate was 24% for those who were MRD-positive and 1536

75% for those who were MRD-negative (P<0.001). Similar to the experience with acute promyelocytic leukemia and core-binding factor AML, rising levels of mutant NPM1 transcripts on sequential testing reliably predicted relapse. In a separate study, patients with NPM1 transcripts detectable by PCR immediately prior to HSCT had significantly worse 5-year overall survival rates than those who were MRD-negative (40% vs. 89%: P=0.007); furthermore, the overall survival of those in complete remission prior to HSCT but with detectable NPM1 MRD was similar to that of patients transplanted with active disease.73 Together, these studies suggest that PCR-based MRD for mutant NPM1 is prognostic across clinical contexts, both after induction/early consolidation and before HSCT.

Next-generation sequencing-based measurable residual disease studies Compared to both MFC and PCR, the use of NGS as an instrument for detecting MRD is relatively new and consequently there are fewer studies to guide its use in clinical practice.17,45,46 In a study of 50 patients who underwent paired whole genome or exon sequencing at diagnosis and remission, 48% had persistent leukemia-associated mutations in at least 5% of bone marrow cells (i.e. variant allelic frequency ≥2.5%).45 These patients with persistent mutations had shorter event-free survival (median: 6.0 vs. 17.9 months, P<0.001) and overall survival (median: 10.5 vs. 42.2 months, P=0.004), compared to those of patients with lower post-remission mutation burden. In a study from the MD Anderson Cancer Center of 131 patients undergoing induction for newly diagnosed AML, complete molecular clearance (i.e. absence of detectable mutations by NGS) was independently associated with decreased risk of relapse and better event-free survival even when MFC-based MRD status was also available.46 These results were strengthened by removing mutations associated with CHIP (i.e. DNMT3A, TET2, and ASXL1, or “DTA” mutations) from the analysis, suggesting that persistence of these preleukemic mutations may not be associated with worse outcomes. In the largest study of NGS MRD in AML, 482 patients were evaluated with NGS at diagnosis and remission, 430 (89.2%) of whom had a leukemia-associated mutation detected at baseline.17 Mutations persisted in 51% of patients at variant allelic frequencies ranging from 0.02-47%. CHIP-associated DTA mutations were not associated with relapse. When DTA mutations were excluded from analysis, residual disease, as detected by NGS, was independently associated with risk of relapse, relapse-free survival and overall survival. Importantly, NGS MRD added prognostic information to MFC MRD status. Four-year relapse rates were similar among patients who were MRD-positive by NGS/MRD-negative by MFC and those who were MRDnegative by NGS/MRD-positive by MFC (52.3% vs. 49.8%, respectively), whereas patients who were MRDpositive or MRD-negative by both assays had highly divergent risks of relapse (73.3% vs. 26.7%, respectively). Together, these studies suggest that NGS-based MRD may complement MFC MRD assays, as these two approaches evaluate different targets (i.e. leukemia-associated mutations and aberrant surface protein expression), and also that reduction and/or clearance of mutations associated with CHIP may not be necessary for cure.17,46 Due to the background error rate of commonly used haematologica | 2019; 104(8)

MRD in AML

NGS-based technologies, the sensitivity of these assays is generally no better than ~1%, limiting their discrimination for small amounts of residual disease. In a study of patients with AML undergoing HSCT, an error-corrected, molecular barcoded NGS MRD assay was able to detect residual mutations prior to HSCT with a median variant allelic frequency of 0.33% (range, 0.016%-4.91%), suggesting that such an assay can detect lower levels of MRD than can standard NGS MRD assays.51 Among the 96 evaluable patients, 45% were MRD-positive prior to HSCT using this approach. The 5-year cumulative incidence of relapse was 66% for those who were MRD-positive versus 17% for those who were MRD-negative (HR 5.58; P<0.001). On multivariate analysis, MRD remained an independent predictor of relapse and also overall survival (HR 3.0; P=0.004). Further studies of these more sensitive, error-corrected NGS technologies in other contexts are warranted.

Practical considerations for the evaluation of measurable residual disease in clinical practice While MRD response has been reliably shown to be predictive for better outcomes, there is significant heterogeneity across studies depending on the specific population of patients, regimen used, AML subtype, MRD method and target, timing of MRD assessment, and threshold to define adequate response, among other factors. All of these variables must be considered when interpreting study results and attempting to generalize to a patient in clinical practice. Despite these challenges, we routinely assess MRD in our clinical practice, as this information helps to refine prognostic assessment and, in some cases, informs therapeutic decisions (e.g. identifies patients who may be candidates for MRD-directed clinical trials or HSCT). Consensus recommendations from the European LeukemiaNet support this approach of routine MRD assessment in AML.6 We also routinely discuss the findings with our patients, particularly when this information is used for clinical decision-making. However, despite the strong association of MRD with clinical outcomes, it is imperative to remember (and to convey to the patient) that “MRD-negativity” does not equate to “cure,” as cumulative incidences of relapse of approximately 2050% are still observed in “MRD-negative” patients, depending on the assay and clinical context. In contrast, while “MRD-positive” patients almost invariably relapse without subsequent therapy, this is also not an absolute certainty. Thus, an acknowledgment of the limitations of our current MRD technologies is required by the clinician to avoid “over-interpreting” MRD information. When using MRD in clinical practice, it is important to choose a complementary MRD assay tailored to the AML subtype of a particular patient. For patients with an appropriate target (e.g. those with acute promyelocytic leukemia, core-binding factor AML or NPM1-mutated AML), PCR is preferred to MFC since the clinical data in these populations are more robust for the former and it also has a higher sensitivity.2,6,22 When MFC is used, it is imperative that it is performed in a laboratory with extensive experience in interpretation to ensure accuracy of the assessment. Whether peripheral blood can replace bone marrow for MRD testing is an area of controversy, although it is generally agreed that an accurate and sensitive peripheral haematologica | 2019; 104(8)

blood-based MRD assay would be welcome to both clinicians and patients, given the less invasive nature of this approach. While some studies have suggested similar findings with these two sources of assay material,74,75 most of the available clinical data on the prognostic and predictive impact of MRD were derived from bone marrow specimens. Furthermore, bone marrow likely provides additional sensitivity, with detectable MRD levels approximately 1-log higher than those in the peripheral blood.6 For example, in core-binding factor AML, a “negative” peripheral blood MRD assay could miss positive bone marrow MRD in up to 40% of cases during therapy and 10-15% of cases during follow up.33 Conversely, in a study of PCR-based MRD in NPM1-mutated AML, discrimination for survival was better when peripheral blood was used than when bone marrow was used.15 Thus, the optimal source for MRD assessment is likely dependent on assay, regimen, and time. Future studies of peripheral blood MRD monitoring are needed across MRD technologies and treatment contexts, and should include investigation of the potential use of circulating cell-free DNA as a source for MRD assays similarly to what has been achieved for some solid tumors.76,77 Initial efforts are already underway in AML.78,79 While the optimal prognostic time point of MRD assessment varies slightly based on the specific regimen and other factors, most studies have shown that detectable MRD is associated with worse outcomes regardless of when it is measured. For patients treated with intensive chemotherapy in the frontline setting, we routinely assess for MRD, at a minimum, at the end of induction and consolidation. It is reasonable to assess MRD at the end of the first cycle of consolidation as well, particularly in patients who remained MRD-positive after induction. In several studies, this time point (i.e. after 2 cycles of chemotherapy) is the most discriminatory for relapse and long-term survival.15,18 MRD status before transplantation is also highly associated with risk of postHSCT relapse and may identify patients who may be candidates for investigational post-HSCT maintenance strategies.52-55

Therapeutic implications of measurable residual disease in acute myeloid leukemia While it is important to obtain MRD information for prognostic purposes, ultimately the goal of developing MRD assays that can more accurately determine risk of relapse in an individual patient with AML is so that postremission therapies can be tailored accordingly. Currently, risk stratification in AML is largely based on cytogenetic and molecular alterations present at the time of diagnosis; these factors are currently the primary disease-related considerations for deciding to pursue HSCT in first remission.2 In contrast, in acute lymphoblastic leukemia, the decision to pursue HSCT in first remission is determined in large part by MRD status, particularly in the pediatric population.7-9 Furthermore, the CD3-CD19 bispecific T-cell engaging antibody blinatumomab is highly effective in eradicating MRD in patients with acute lymphoblastic leukemia and is approved by the United States Food and Drug Administration for this use.80 While some data support the use of MRD to guide therapeutic decisions in AML, the role of MRD-guided 1537

N.J. Short and F. Ravandi et al. Table 2. Ongoing measurable residual disease-based clinical trials in adults with acute myeloid leukemia.

Agent(s)

Study Phase

Timing of MRD positivity

Autologous WT1-TCRc4 gene-transduced CD8+ T cells

I/II

First CR/CRi

Azacitidine + avelumab BL-8040 (CXCR4 antagonist) + atezolizumab DCP-001 (dendritic cell vaccine) FLYSYN (anti-FLT3 monoclonal antibody)

I/II Ib/II

Any time First CR/CRi

II I

First CR/CRi Any time

II II

Any time Before or after allogeneic HSCT

II I II II (randomized) I/II

Gemtuzumab ozogamicin Guadecitabine (SGI-110) + donor lymphocyte infusion Lenalidomide Lenalidomide Nivolumab Nivolumab (vs. observation) Tagraxofusp (SL-401)

Major inclusion/exclusion criteria

ClinicalTrials.gov identifier

• Must have HLA-A*02:01 expression and elevated baseline WT1 expression • Excludes patients with prior allogeneic HSCT --• Excludes patients with prior allogeneic HSCT

NCT02770820

• Excludes patients with prior allogeneic HSCT • Must have NPM1-mutated AML and FLT3 expression on leukemic cells • Excludes patients with prior allogeneic HSCT • Must be CD33+ ---

NCT03697707 NCT02789254

NCT03737955 NCT02684162

First or second CR/CRi After allogeneic HSCT First or second CR/CRi First CR/CRi

--• MRD detected by CD34+ chimerism assay -----

NCT02126553 NCT02370888 NCT02126553 NCT02275533

First or second CR/CRi

---

NCT02270463

NCT03699384 NCT03154827

MRD: measurable residual disease; CR: complete remission; CRi: complete remission with incomplete hematologic recovery; HSCT: hematopoietic stem cell transplantation; AML: acute myeloid leukemia.

therapies is less clear in AML than in acute lymphoblastic leukemia. With the possible exceptions of allogeneic HSCT and hypomethylating agents, effective treatments for MRD-positive disease have not been established in AML. It has yet to be conclusively shown that eradication of MRD that persists or recurs after initial treatment is associated with improved outcomes in AML; however, given the available evidence (as discussed in detail below), we do frequently use HSCT or hypomethylating agents for patients with MRD-positive disease. Enrollment of these patients in MRD-directed clinical trials with novel agents and combinations is also imperative (Table 2).

Measurable residual disease-guided allogeneic hematopoietic stem cell transplantation While it is common practice to refer patients who are MRD-positive to HSCT, robust data supporting this practice are lacking. Nevertheless, while the outcomes of MRD-positive patients are relatively poor regardless of whether HSCT is or is not performed, several studies suggest that outcomes may be improved with HSCT.14,18,70,81 In a multicenter study of 116 patients with AML harboring t(8;21), patients received either allogeneic HSCT or chemotherapy, with or without autologous HSCT, based on MRD response.70 High-risk patients were defined as those who did not achieve a major molecular response (i.e., a >3-log reduction in RUNX1/RUNX1T1 transcript levels in bone marrow, determined by PCR analysis) after the second consolidation, or who lost the major molecular response within 6 months of achieving it. Among patients in the high-risk group, those who underwent allogeneic HSCT had a lower cumulative incidence of relapse compared to that of the patients who did not 1538

undergo allogeneic HSCT (22.1% vs. 78.9%, respectively; P<0.001), which translated into an improvement in overall survival (71.6% vs. 26.7%, respectively; P=0.007). Conversely, the overall survival rate was lower in lowrisk patients who underwent allogeneic HSCT (75.7% vs. 100%; P=0.013). These data suggest that MRD could reasonably be used to assign patients with t(8;21) to postremission therapies. Data for similar approaches in corebinding factor AML with inv(16) are lacking. The role of allogeneic HSCT in patients with intermediate-risk AML is controversial, and it is possible that MRD may be able to guide post-remission therapeutic decisions in this group.2,4 In a study of younger patients (age 18-60 years) with newly diagnosed NPM1-mutated AML, HSCT was associated with better disease-free survival and overall survival in patients with non-favorable risk disease who had a <4-log reduction in mutant NPM1 in the peripheral blood at the time of remission (diseasefree survival: HR, 0.25; 95% CI: 0.06-0.98; P=0.047; overall survival: HR, 0.25; 95% CI: 0.06-0.98; P=0.047), whereas HSCT did not improve outcomes for patients with a >4-log reduction.14 Similarly in the National Cancer Research Institute AML17 study of standard-risk patients without NPM1 mutations, HSCT appeared to preferentially benefit those who remained MRD-positive by MFC after the second cycle of chemotherapy but not those who were MRD-negative, although the interaction between HSCT and MRD status did not reach statistical significance (P=0.16), possibly because of the small number of patients in the cohort who underwent HSCT in first remission.18 Although definitive evidence is lacking, collectively these studies provide some support for the use of MRD to guide post-remission therapies in patients with intermediate-risk AML. haematologica | 2019; 104(8)

MRD in AML

Hypomethylating agents for measurable residual disease-positive acute myeloid leukemia

Conclusions

Although HSCT has historically been the preferred consolidation option for patients with high-risk AML, there is also interest in the use of hypomethylating agents for these patients, including those with detectable MRD. Several studies have suggested that treatment with hypomethylating agents (azacitidine or decitabine) may be beneficial for patients with AML and persistent or recurrent MRD.82-85 The RELAZA2 trial was the largest prospective study to assess this issue.84 One hundred and ninety patients with AML or myelodysplastic syndrome who achieved complete remission after chemotherapy or HSCT were monitored for MRD for 24 months with either PCR for mutant NPM1, leukemia-specific gene fusions (DEK-NUP214, RUNX1-RUNX1T1, or CBFBMYH11) or donor chimerism in flow cytometry-sorted CD34+ cells (for patients who had undergone prior allogeneic HSCT). Patients who developed detectable MRD received a standard dose of azacitidine for up to 24 cycles. Among the 53 evaluable patients who became MRD-positive and received treatment with azacitidine, 19 (36%) achieved MRD-negativity after 6 months of treatment. Median relapse-free survival and overall survival times for these MRD-positive patients who received azacitidine was 10 months and 31 months, respectively. Importantly, MRD response was associated with improved relapse-free survival (HR, 0.2; 95% CI: 0.1-0.5; P<0.0001) and a trend towards improved overall survival (HR, 0.4; 95% CI: 0.11.3; P=0.112). These data suggest that hypomethylating agents may be a reasonable option for patients with MRD-positive AML, particularly those who are not suitable candidates for allogeneic HSCT. Other studies have, however, failed to demonstrate a benefit for the routine use of hypomethylating agents in the maintenance setting.86,87 A phase III randomized study of an oral formulation of azacitidine as maintenance therapy in patients with AML in remission will likely provide further important information in this area (NCT01757535).

Despite substantial heterogeneity across studies of MRD in AML (e.g. variations in AML subtype, MRD methodology and target, cutoff, regimen, and timing of assessment), achievement of MRD negativity has been reliably shown to predict risk of relapse and long-term outcomes. More sensitive assays are still needed, however, as up to 50% of patients who are â&#x20AC;&#x153;MRD-negativeâ&#x20AC;? according to commonly applied MRD technologies still relapse. Beyond the important prognostic information that MRD provides, the ultimate goal of assessing MRD is to use this information to appropriately guide decisions regarding post-remission therapies. This approach is, however, strongly dependent on the availability of therapies that are effective in eradicating small amounts of residual leukemia (as with blinatumomab in acute lymphoblastic leukemia). While some studies suggest that allogeneic HSCT or hypomethylating agents may provide benefit for some patients with persistent or recurrent MRD, long-term outcomes with these approaches are largely disappointing. Innovative treatments are therefore needed for these patients. There are several ongoing MRD-directed clinical trials that are evaluating novel immune-based strategies (e.g. checkpoint inhibitors, vaccines, and T-cell-based therapies such as bi-specific T-cell engaging antibodies and chimeric antigen receptor T cells) which may be able to overcome the chemoresistant phenotype associated with MRD-positive disease. Novel combinations with Bcl-2 inhibitors (e.g. venetoclax) may also be promising in this setting. Ultimately, enrollment of high-risk patients into these trials is imperative in order to advance our understanding of how to use MRD information to drive clinical decisions and to improve outcomes of patients with AML.

References 1. Short NJ, Rytting ME, Cortes JE. Acute myeloid leukaemia. Lancet. 2018;392 (10147):593-606. 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. Mayer RJ, Davis RB, Schiffer CA, et al. Intensive postremission chemotherapy in adults with acute myeloid leukemia. Cancer and Leukemia Group B. N Engl J Med. 1994;331(14):896-903. 4. Cornelissen JJ, Blaise D. Hematopoietic stem cell transplantation for patients with AML in first complete remission. Blood. 2016;127 (1):62-70. 5. Hourigan CS, Karp JE. Minimal residual disease in acute myeloid leukaemia. Nat Rev Clin Oncol. 2013;10(8):460-471. 6. Schuurhuis GJ, Heuser M, Freeman S, et al. Minimal/measurable residual disease in AML: a consensus document from the

haematologica | 2019; 104(8)

10.

11.

Funding Supported by an MD Anderson Cancer Center Support Grant (CA016672).

European LeukemiaNet MRD Working Party. Blood. 2018;131(12):1275-1291. Short NJ, Jabbour E, Albitar M, et al. Recommendations for the assessment and management of measurable residual disease in adults with acute lymphoblastic leukemia: a consensus of North American experts. Am J Hematol. 2019;94(2):257-265. Pui CH, Pei D, Coustan-Smith E, et al. Clinical utility of sequential minimal residual disease measurements in the context of risk-based therapy in childhood acute lymphoblastic leukaemia: a prospective study. Lancet Oncol. 2015;16(4):465-474. Vora A, Goulden N, Wade R, et al. Treatment reduction for children and young adults with low-risk acute lymphoblastic leukaemia defined by minimal residual disease (UKALL 2003): a randomised controlled trial. Lancet Oncol. 2013;14(3):199-209. Campana D, Pui CH. Minimal residual disease-guided therapy in childhood acute lymphoblastic leukemia. Blood. 2017;129(14): 1913-1918. Bruggemann M, Raff T, Kneba M. Has MRD

12.

13.

14.

15.

monitoring superseded other prognostic factors in adult ALL? Blood. 2012;120(23):44704481. 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. 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. 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. Ivey A, Hills RK, Simpson MA, et al. Assessment of minimal residual disease in standard-risk AML. N Engl J Med. 2016;374

1539

N.J. Short and F. Ravandi et al. (5):422-433. 16. Ravandi F, Jorgensen J, Borthakur G, et al. Persistence of minimal residual disease assessed by multiparameter flow cytometry is highly prognostic in younger patients with acute myeloid leukemia. Cancer. 2017;123 (3):426-435. 17. 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. 18. Freeman SD, Hills RK, Virgo P, et al. Measurable residual disease at induction redefines partial response in acute myeloid leukemia and stratifies outcomes in patients at standard risk without NPM1 mutations. J Clin Oncol. 2018;36(15):1486-1497. 19. Grimwade D, Freeman SD. Defining minimal residual disease in acute myeloid leukemia: which platforms are ready for "prime time"? Blood. 2014;124(23):33453355. 20. Marcucci G, Mrozek K, Ruppert AS, et al. Abnormal cytogenetics at date of morphologic complete remission predicts short overall and disease-free survival, and higher relapse rate in adult acute myeloid leukemia: results from Cancer and Leukemia Group B study 8461. J Clin Oncol. 2004;22(12):24102418. 21. Chen Y, Cortes J, Estrov Z, et al. Persistence of cytogenetic abnormalities at complete remission after induction in patients with acute myeloid leukemia: prognostic significance and the potential role of allogeneic stem-cell transplantation. J Clin Oncol. 2011;29(18):2507-2513. 22. Ravandi F, Walter RB, Freeman SD. Evaluating measurable residual disease in acute myeloid leukemia. Blood Adv. 2018;2(11):1356-1366. 23. Wood BL. Principles of minimal residual disease detection for hematopoietic neoplasms by flow cytometry. Cytometry B Clin Cytom. 2016;90(1):47-53. 24. Baer MR, Stewart CC, Dodge RK, et al. High frequency of immunophenotype changes in acute myeloid leukemia at relapse: implications for residual disease detection (Cancer and Leukemia Group B study 8361). Blood. 2001;97(11):3574-3580. 25. Langebrake C, Brinkmann I, Teigler-Schlegel A, et al. Immunophenotypic differences between diagnosis and relapse in childhood AML: implications for MRD monitoring. Cytometry B Clin Cytom. 2005;63(1):1-9. 26. Terwijn M, Zeijlemaker W, Kelder A, et al. Leukemic stem cell frequency: a strong biomarker for clinical outcome in acute myeloid leukemia. PLoS One. 2014;9(9): e107587. 27. Bradbury C, Houlton AE, Akiki S, et al. Prognostic value of monitoring a candidate immunophenotypic leukaemic stem/progenitor cell population in patients allografted for acute myeloid leukaemia. Leukemia. 2015;29(4):988-991. 28. Zeijlemaker W, Grob T, Meijer R, et al. CD34(+)CD38(-) leukemic stem cell frequency to predict outcome in acute myeloid leukemia. Leukemia. 2019;33(5):1102-1112. 29. Ko BS, Wang YF, Li JL, et al. Clinically validated machine learning algorithm for detecting residual diseases with multicolor flow cytometry analysis in acute myeloid leukemia and myelodysplastic syndrome. EBioMedicine. 2018;37:91-100. 30. Burnett AK, Grimwade D, Solomon E, Wheatley K, Goldstone AH. Presenting white blood cell count and kinetics of molecular remission predict prognosis in acute promyelocytic leukemia treated with

1540

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

all-trans retinoic acid: result of the randomized MRC trial. Blood. 1999;93(12):41314143. Grimwade D, Jovanovic JV, Hills RK, et al. Prospective minimal residual disease monitoring to predict relapse of acute promyelocytic leukemia and to direct pre-emptive arsenic trioxide therapy. J Clin Oncol. 2009;27(22):3650-3658. Krauter J, Gorlich K, Ottmann O, et al. Prognostic value of minimal residual disease quantification by real-time reverse transcriptase polymerase chain reaction in patients with core binding factor leukemias. J Clin Oncol. 2003;21(23):4413-4422. Yin JA, O'Brien MA, Hills RK, Daly SB, Wheatley K, Burnett AK. Minimal residual disease monitoring by quantitative RT-PCR in core binding factor AML allows risk stratification and predicts relapse: results of the United Kingdom MRC AML-15 trial. Blood. 2012;120(14):2826-2835. Schnittger S, Kern W, Tschulik C, et al. Minimal residual disease levels assessed by NPM1 mutation-specific RQ-PCR provide important prognostic information in AML. Blood. 2009;114(11):2220-2231. 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. Hubmann M, Kohnke T, Hoster E, et al. Molecular response assessment by quantitative real-time polymerase chain reaction after induction therapy in NPM1-mutated patients identifies those at high risk of relapse. Haematologica. 2014;99(8):13171325. Nazha A, Cortes J, Faderl S, et al. Activating internal tandem duplication mutations of the fms-like tyrosine kinase-3 (FLT3-ITD) at complete response and relapse in patients with acute myeloid leukemia. Haematologica. 2012;97(8):1242-1245. Cloos J, Goemans BF, Hess CJ, et al. Stability and prognostic influence of FLT3 mutations in paired initial and relapsed AML samples. Leukemia. 2006;20(7):1217-1220. Smol T, Nibourel O, Marceau-Renaut A, et al. Quantification of EVI1 transcript levels in acute myeloid leukemia by RT-qPCR analysis: a study by the ALFA group. Leuk Res. 2015;39(12):1443-1447. 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. Gianfaldoni G, Mannelli F, Ponziani V, et al. Early reduction of WT1 transcripts during induction chemotherapy predicts for longer disease free and overall survival in acute myeloid leukemia. Haematologica. 2010;95 (5):833-836. Nomdedeu JF, Hoyos M, Carricondo M, et al. Bone marrow WT1 levels at diagnosis, post-induction and post-intensification in adult de novo AML. Leukemia. 2013;27(11): 2157-2164. Parkin B, Londono-Joshi A, Kang Q, Tewari M, Rhim AD, Malek SN. Ultrasensitive mutation detection identifies rare residual cells causing acute myelogenous leukemia relapse. J Clin Invest. 2017;127(9):34843495. Kanagal-Shamanna R. Digital PCR: principles and applications. Methods Mol Biol. 2016;1392:43-50.

45. 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. 46. 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. 47. 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. 48. 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. 49. Jaiswal S, Fontanillas P, Flannick J, et al. Agerelated clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371(26):2488-2498. 50. Salk JJ, Schmitt MW, Loeb LA. Enhancing the accuracy of next-generation sequencing for detecting rare and subclonal mutations. Nat Rev Genet. 2018;19(5):269-285. 51. Thol F, Gabdoulline R, Liebich A, et al. Measurable residual disease monitoring by NGS before allogeneic hematopoietic cell transplantation in AML. Blood. 2018;132 (16):1703-1713. 52. Walter RB, Gooley TA, Wood BL, et al. Impact of pretransplantation minimal residual disease, as detected by multiparametric flow cytometry, on outcome of myeloablative hematopoietic cell transplantation for acute myeloid leukemia. J Clin Oncol. 2011;29(9):1190-1197. 53. Walter RB, Buckley SA, Pagel JM, et al. Significance of minimal residual disease before myeloablative allogeneic hematopoietic cell transplantation for AML in first and second complete remission. Blood. 2013;122(10):1813-1821. 54. 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. 55. Zhou Y, Othus M, Araki D, et al. Pre- and post-transplant quantification of measurable ('minimal') residual disease via multiparameter flow cytometry in adult acute myeloid leukemia. Leukemia. 2016;30(7): 1456-1464. 56. Feller N, van der Pol MA, van Stijn A, et al. MRD parameters using immunophenotypic detection methods are highly reliable in predicting survival in acute myeloid leukaemia. Leukemia. 2004;18(8):1380-1390. 57. Boddu P, Jorgensen J, Kantarjian H, et al. Achievement of a negative minimal residual disease state after hypomethylating agent therapy in older patients with AML reduces the risk of relapse. Leukemia. 2018;32(1): 241-244. 58. 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. 59. Shah MV, Jorgensen JL, Saliba RM, et al. Early post-transplant minimal residual disease assessment improves risk stratification in acute myeloid leukemia. Biol Blood Marrow Transplant. 2018;24(7):1514-1520. 60. Nucifora G, Larson RA, Rowley JD. Persistence of the 8;21 translocation in patients with acute myeloid leukemia type M2 in long-term remission. Blood. 1993;82(3):712-715.

haematologica | 2019; 104(8)

MRD in AML 61. Miyamoto T, Nagafuji K, Akashi K, et al. Persistence of multipotent progenitors expressing AML1/ETO transcripts in longterm remission patients with t(8;21) acute myelogenous leukemia. Blood. 1996;87(11): 4789-4796. 62. Diverio D, Rossi V, Avvisati G, et al. Early detection of relapse by prospective reverse transcriptase-polymerase chain reaction analysis of the PML/RARalpha fusion gene in patients with acute promyelocytic leukemia enrolled in the GIMEMA-AIEOP multicenter "AIDA" trial. Blood. 1998;92(3): 784-789. 63. Sanz MA, Grimwade D, Tallman MS, et al. Management of acute promyelocytic leukemia: recommendations from an expert panel on behalf of the European LeukemiaNet. Blood. 2009;113(9):18751891. 64. 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. 65. Abaza Y, Kantarjian H, Garcia-Manero G, et al. Long-term outcome of acute promyelocytic leukemia treated with all-trans-retinoic acid, arsenic trioxide, and gemtuzumab. Blood. 2017;129(10):1275-1283. 66. Cancer Genome Atlas Research Network, 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. 67. Boddu P, Gurguis C, Sanford D, et al. Response kinetics and factors predicting survival in core-binding factor leukemia. Leukemia. 2018;32(12):2698-2701. 68. Leroy H, de Botton S, Grardel-Duflos N, et al. Prognostic value of real-time quantitative PCR (RQ-PCR) in AML with t(8;21). Leukemia. 2005;19(3):367-372. 69. Weisser M, Haferlach C, Hiddemann W, Schnittger S. The quality of molecular response to chemotherapy is predictive for the outcome of AML1-ETO-positive AML and is independent of pretreatment risk factors. Leukemia. 2007;21(6):1177-1182. 70. Zhu HH, Zhang XH, Qin YZ, et al. MRDdirected risk stratification treatment may

haematologica | 2019; 104(8)

71.

72.

73.

74.

75.

76.

77.

78.

79.

improve outcomes of t(8;21) AML in the first complete remission: results from the AML05 multicenter trial. Blood. 2013;121 (20):4056-4062. Falini B, Martelli MP, Bolli N, et al. Acute myeloid leukemia with mutated nucleophosmin (NPM1): is it a distinct entity? Blood. 2011;117(4):1109-1120. Jain P, Kantarjian H, Patel K, et al. Mutated NPM1 in patients with acute myeloid leukemia in remission and relapse. Leuk Lymphoma. 2014;55(6):1337-1344. Kayser S, Benner A, Thiede C, et al. Pretransplant NPM1 MRD levels predict outcome after allogeneic hematopoietic stem cell transplantation in patients with acute myeloid leukemia. Blood Cancer J. 2016;6(7):e449. Maurillo L, Buccisano F, Spagnoli A, et al. Monitoring of minimal residual disease in adult acute myeloid leukemia using peripheral blood as an alternative source to bone marrow. Haematologica. 2007;92(5):605611. Zeijlemaker W, Kelder A, OussorenBrockhoff YJ, et al. Peripheral blood minimal residual disease may replace bone marrow minimal residual disease as an immunophenotypic biomarker for impending relapse in acute myeloid leukemia. Leukemia. 2016;30(3):708-715. Umetani N, Giuliano AE, Hiramatsu SH, et al. Prediction of breast tumor progression by integrity of free circulating DNA in serum. J Clin Oncol. 2006;24(26):4270-4276. Abbosh C, Birkbak NJ, Swanton C. Early stage NSCLC - challenges to implementing ctDNA-based screening and MRD detection. Nat Rev Clin Oncol. 2018;15(9):577586. Assi RE, Albitar M, Ma W, et al. Comparison of somatic mutations profiles from nextgeneration sequencing (NGS) of cell-free DNA (cfDNA) versus bone marrow (BM) in acute myeloid leukemia (AML). J Clin Oncol. 2018;36(15_suppl):7051-7051. Nakamura S, Yokoyama K, Shimizu E, et al. Prognostic impact of circulating tumor DNA status post-allogeneic hematopoietic stem cell transplantation in acute myeloid

80.

81.

82.

83.

84.

85.

86.

87.

leukemia and myelodysplastic syndrome. Blood. 2018;132(Suppl 1):247. Gokbuget N, Dombret H, Bonifacio M, et al. Blinatumomab for minimal residual disease in adults with B-cell precursor acute lymphoblastic leukemia. Blood. 2018;131(14): 1522-1531. Buccisano F, Maurillo L, Spagnoli A, et al. Cytogenetic and molecular diagnostic characterization combined to postconsolidation minimal residual disease assessment by flow cytometry improves risk stratification in adult acute myeloid leukemia. Blood. 2010;116(13):2295-2303. Sockel K, Wermke M, Radke J, et al. Minimal residual disease-directed preemptive treatment with azacitidine in patients with NPM1-mutant acute myeloid leukemia and molecular relapse. Haematologica. 2011;96(10):1568-1570. Platzbecker U, Wermke M, Radke J, et al. Azacitidine for treatment of imminent relapse in MDS or AML patients after allogeneic HSCT: results of the RELAZA trial. Leukemia. 2012;26(3):381-389. Platzbecker U, Middeke JM, Sockel K, et al. Measurable residual disease-guided treatment with azacitidine to prevent haematological relapse in patients with myelodysplastic syndrome and acute myeloid leukaemia (RELAZA2): an open-label, multicentre, phase 2 trial. Lancet Oncol. 2018;19(12):1668-1679. Ragon BK, Daver N, Garcia-Manero G, et al. Minimal residual disease eradication with epigenetic therapy in core binding factor acute myeloid leukemia. Am J Hematol. 2017;92(9):845-850. Boumber Y, Kantarjian H, Jorgensen J, et al. A randomized study of decitabine versus conventional care for maintenance therapy in patients with acute myeloid leukemia in complete remission. Leukemia. 2012;26(11): 2428-2431. Blum W, Sanford BL, Klisovic R, et al. Maintenance therapy with decitabine in younger adults with acute myeloid leukemia in first remission: a phase 2 Cancer and Leukemia Group B study (CALGB 10503). Leukemia. 2017;31(1):34-39.

1541

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(8):1542-1553

Iron metabolism & its Disorders

Rapid growth is a dominant predictor of hepcidin suppression and declining ferritin in Gambian infants

Andrew E. Armitage,1* Schadrac C. Agbla,2* Modupeh Betts,3 Ebrima A. Sise,3 Momodou W. Jallow,3 Ellen Sambou,4 Bakary Darboe,3 Archibald Worwui,3 George M. Weinstock,5 Martin Antonio,4 Sant-Rayn Pasricha,1,6,7 Andrew M. Prentice,3 Hal Drakesmith,1,8 Momodou K. Darboe3,** and Brenda Anna Kwambana-Adams4,9,**

MRC Human Immunology Unit, MRC Weatherall Institute of Molecular Medicine, University of Oxford, Oxford, UK; 2Faculty of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London, UK; 3MRC Unit The Gambia at the London School of Hygiene and Tropical Medicine, Banjul, The Gambia, Africa; 4WHO Collaborating Center for New Vaccines Surveillance, MRC Unit The Gambia at the London School of Hygiene and Tropical Medicine, Banjul, The Gambia, Africa; 5The Jackson Laboratory for Genomic Medicine, Farmington, CT, USA; 6Walter and Eliza Hall Institute for Medical Research, Melbourne, VIC, Australia; 7Department of Medical Biology, The University of Melbourne, VIC, Melbourne, Australia; 8Haematology Theme, Oxford Biomedical Research Centre, Oxford, UK and 9NIHR Global Health Research Unit on Mucosal Pathogens, Division of Infection and Immunity, University College London, London, UK 1

*Contributed equally to this work. **Contributed equally as senior co-authors.

ABSTRACT

Correspondence: ANDREW E. ARMITAGE/BRENDA ANNA KWAMBANA-ADAMS andrew.armitage@imm.ox.ac.uk/ rekgbak@ucl.ac.uk Received: October 25, 2018. Accepted: January 31, 2019. Pre-published: February 7, 2019. doi:10.3324/haematol.2018.210146 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1542 ÂŠ2019 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.

1542

ron deficiency and iron deficiency anemia are highly prevalent in low-income countries, especially among young children. Hepcidin is the major regulator of systemic iron homeostasis. It controls dietary iron absorption, dictates whether absorbed iron is made available in circulation for erythropoiesis and other iron-demanding processes, and predicts response to oral iron supplementation. Understanding how hepcidin is itself regulated is therefore important, especially in young children. We investigated how changes in iron-related parameters, inflammation and infection status, seasonality, and growth influenced plasma hepcidin and ferritin concentrations during infancy using longitudinal data from two birth cohorts of infants in rural Gambia (n=114 and n=193). This setting is characterized by extreme seasonality, prevalent childhood anemia, undernutrition, and frequent infection. Plasma was collected from infants at birth and at regular intervals, up to 12 months of age. Hepcidin, ferritin and plasma iron concentrations declined markedly during infancy, with reciprocal increases in soluble transferrin receptor and transferrin concentrations, indicating declining iron stores and increasing tissue iron demand. In cross-sectional analyses at 5 and 12 months of age, we identified expected relationships of hepcidin with iron and inflammatory markers, but also observed significant negative associations between hepcidin and antecedent weight gain. Correspondingly, longitudinal fixed effects modeling demonstrated weight gain to be the most notable dynamic predictor of decreasing hepcidin and ferritin through infancy across both cohorts. Infants who grow rapidly in this setting are at particular risk of depletion of iron stores, but since hepcidin concentrations decrease with weight gain, they may also be the most responsive to oral iron interventions. haematologica | 2019; 104(8)

Rapid growth and hepcidin suppression in infancy

Introduction Iron deficiency anemia (IDA) affected 1.24 billion people and was the most common cause of Years Lived with Disability (YLD) in low- and low-middle income countries in 2016.1 Iron deficiency (ID) and anemia disproportionately affect young children in such socio-economic settings.2 Beyond anemia, ID during infancy is linked to impaired cognitive and behavioral development, potentially yielding irreversible long-term individual and societal impacts.3,4 Neonates are born with a maternal iron endowment that is initially largely contained within hemoglobin and influenced by birthweight, gestational age, and the timing and method of cord clamping.5-7 In healthy term infants, this birth endowment meets the iron needs of blood volume expansion, brain development and tissue accretion for several months, compensating for the very low level of iron in breast milk;5 after this, complementary feeding must supply iron.6-9 Iron-poor diets and high infection burdens, when combined with the significant physiological iron demands of growth and development during infancy, render young LMIC children especially vulnerable to ID/IDA. The hepatic hormone hepcidin regulates systemic iron handling.10 Hepcidin inhibits the cellular iron exporter ferroportin, preventing iron recycling by erythrophagocytic macrophages and dietary iron uptake through enterocytes.11 Hepcidin induction by iron prevents iron overload, while its production during inflammation/infection generates an acute anti-infective hypoferremia;12,13 moreover, chronically raised hepcidin causes ID and IDA,14 and chronic ID and IDA can protect against malaria.15-17 In contrast, hepcidin suppression during iron demand, for example via erythroferrone activity in response to erythropoietic stress, increases serum iron availability.18 In African preschool children, hepcidin reflects iron status and effectively predicts utilization of orally administered iron.19,20 However, its regulation during the first year of life remains poorly characterized.6 The Gambia is an African country characterized by high burdens of infection, undernutrition and anemia, together with extreme seasonality, all of which have the potential to influence iron control through effects on hepcidin.21 In the clearly demarcated wet season, nutritional availability and quality deteriorate, infections are more prevalent, and growth faltering is more prominent. Here, we used two longitudinal birth cohort studies from rural communities in The Gambia to investigate how relative changes in iron, infection and inflammation, erythropoietic drive, seasonality, and growth rate predict changes in hepcidin and the iron stores marker ferritin during the first year of life.

Methods Details of cohort characteristics, methods and statistics are presented in the Online Supplementary Appendix and Online Supplementary Tables S1 and S2.

- VPM study: Vaccination and Paediatric Microbiome (VPM) study. Conducted in the Western Region; March 2013-September 2015; n=114; - VA study: a vitamin A supplementation randomized controlled trial.22 Conducted in the West Kiang Region; September 2001-October 2004; n=193.

Laboratory analysis

Ferritin, plasma iron, soluble transferrin receptor (sTfR), α(1)acid glycoprotein (AGP), C-reactive protein (CRP), transferrin (VPM only), and hemoglobin (VA only) concentrations were measured by automated analyzers. Hepcidin was measured by manual ELISA (Bachem/Peninsula Laboratories, San Carlos, CA, USA).21

Definitions Anthropometric Z-scores (World Health Organization child growth standards): generated using STATA package zscore06.23 Wet season: July-October. Weaning: first recorded occurrence of any feeding other than exclusive breastfeeding. Positive recent infection: occurrence within the previous four weeks of ear infection, chest infection, meningitis/sepsis, or other symptoms associated with infection (including fever, diarrhea, vomiting), or if antibiotics were administered within the previous two weeks (VPM); or of diarrhea, vomiting, fever, cough or clinic attendance (VA). Iron deficiency (ID): ferritin <12 μg/L or ferritin <30 μg/L if CRP >5 mg/L; anemia: Hb <11 g/dL; inflammation: CRP >5 mg/L; iron deficiency anemia (IDA): anemia in the presence of ID.21

Statistical analysis RStudio (v1.1), Stata14 (Statacorp-LP), R (www.r-project.org) and Prism7 (GraphPad Software) were used for statistical analysis and graphics. To address the potential bias that may arise when deriving parameter estimates in the presence of missing data,24,25 as discussed in detail in the Online Supplementary Appendix, we implemented multiple imputation by chained equations (MICE) to generate 100 independent datasets with missing data imputed;26 after analyzing imputed datasets, estimates were pooled using Rubin’s rules.27 Continuous variables were log10-transformed prior to multiple imputation. Subsequent analyses were performed on both imputed datasets and original pre-imputation datasets; the latter are shown in the Online Supplementary Appendix. Analyses comprised: - univariate correlations with hepcidin and ferritin, computing Pearson correlation coefficients; - multivariate cross-sectional analyses at five and 12 months of age incorporating parameters returning P<0.2 in univariate analysis; Seemingly Unrelated Regression analysis,28 enabled simultaneous modeling of the correlated outcomes, hepcidin and ferritin; - longitudinal panel fixed effects models [estimating DriscollKraay standard errors29 using Stata command xtscc (st0128)30] to enable investigation of how within-infant changes over time in time-variant explanatory variables predict changes in hepcidin and ferritin addressing any confounding induced by time-invariant characteristics (e.g. sex, genotype, ethnicity), whether measured or unmeasured; to visualize relative effect sizes, data were standardized, fixed effects models were refitted, and forest plots were generated.

Ethical considerations Cohorts We obtained biochemical, anthropometric and infection data at serial time points [birth (cord blood), 2, 5, 9 (VA cohort only) and 12 months of age] from two rural birth-cohort studies in The Gambia, West Africa: haematologica | 2019; 104(8)

The Gambia Government/Medical Research Council (MRC) Joint Ethics Committee approved both studies (VPM: SCC1315; VA: SCC844); VA was also approved by the Ethics Committee of the London School of Hygiene and Tropical Medicine (LSHTM), Banjul, The Gambia. All participants were recruited via approved 1543

A.E. Armitage et al.

protocols; parents and/or legal guardians gave written informed consent for infants. Studies were performed in accordance with the Declaration of Helsinki.

Results Decline in hepcidin and iron status through infancy High plasma concentrations of hepcidin and ferritin in

the absence of inflammation suggest that the majority of infants were iron replete at birth and up to two months of age (Figure 1A and B). Hepcidin concentrations declined from two to 12 months of age in both VPM and VA infant cohorts. In both cohorts, declines were most marked in males born in the lower birthweight group (Figure 1A). Similar patterns of decline were observed for ferritin, indicating rapid depletion of iron stores through infancy (Figure 1B). The decline in hepcidin and ferritin occurred

Figure 1. Changes in hepcidin and iron/inflammatory biomarkers across the first year of life in Gambian infants. Plots summarize changes in (A) hepcidin, (B) ferritin, (C) C-reactive protein (CRP), (D) Îą(1)-acid glycoprotein (AGP), (E) plasma iron, (F) soluble transferrin receptor (sTfR), and (G) transferrin (VPM only), and (H) hemoglobin (VA cohort) occurring across the first year of life in two cohorts of Gambian infants, VPM (left panels) and VA (right panels). Plots depict mean for cross-sectional data at each time point, stratifying by sex (male: blue; female: red) and birthweight group [World Health Organization weight-for-age z-score (WAZ) -0.5: above -0.5: solid line, or below -0.5: dashed line], summarizing 100 datasets in which any missing data was imputed by multiple imputation, combined using Rubinâ&#x20AC;&#x2122;s combination rules, as described in detail in the Methods section and the Online Supplementary Methods. 95% Confidence Intervals are omitted for clarity, but are given in Online Supplementary Table S4. Equivalent plots based on the original data prior to multiple imputation are shown in Online Supplementary Figure S1. VPM: Vaccination and Paediatric Microbiome study; VA: a vitamin A supplementation randomized controlled trial.22

1544

haematologica | 2019; 104(8)

Rapid growth and hepcidin suppression in infancy

despite higher inflammatory burden later in infancy, reflected by increased CRP and AGP concentrations (Figure 1C and D). Plasma iron showed the most striking decline of any iron biomarker, with concentrations falling from birth in all infant groups, irrespective of sex or birthweight, to concentrations well below normal expected values for iron replete young children (typically >10 μmol/L)31 by five months of age in both cohorts (Figure 1E). Concentrations of sTfR, commonly used as a surrogate for erythropoietic iron demand, were reduced at two months of age (until when erythropoietic output is likely to have been low32) in both cohorts. Erythropoietic activity typically increases from around this age, and is likely influenced by the declining plasma iron availability; consistent with this, sTfR and transferrin concentrations were both elevated by 12 months of age, again most notably in lower birthweight males (Figure 1F and G). Accordingly,

this group had the lowest hemoglobin concentration by 12 months of age (Figure 1H), and by this time the population had experienced a heavy burden of iron deficiency and anemia: 67.7% (95%CI: 56.2-79.2%) and 36.4% (95%CI: 25.6-47.3%) of females, and 77.8% (95%CI: 65.6-90.0%) and 45.3% (95%CI: 34.5-56.1%) of males in VPM and VA, respectively, were classified as iron deficient at 1 year of age; 86.5% (95%CI: 79.5-93.5%) of females and 93.1% (95%CI: 87.8-98.4%) of males were classified as anemic in VA (in which hemoglobin was measured) (Online Supplementary Table S3).

Associations with hepcidin and ferritin concentrations at five and 12 months of age: cross-sectional analyses To explore the relationships of hepcidin and ferritin with other variables during infancy, we next performed cross-sectional univariable analyses at month 5 (the time

Table 1. Cross-sectional correlations between hepcidin, ferritin, and explanatory variables in infants at five months of age.

Data summarize 100 datasets in which any missing data were imputed using multiple imputation as described in detail in the Methods section and the Online Supplementary Appendix. Infection was coded as “no infection=0” and “occurrence of infection=1’; Season was coded as “dry season (November-June)=0” and “wet season (July-October)=1”; Sex was coded as “female=0” and “male=1”; Head Circ.: head circumference.“r” denotes the Pearson correlation coefficient. Significance level is highlighted with blue shading. Dark blue: P<0.001; mid-blue: P<0.01; light blue: P<0.05.

haematologica | 2019; 104(8)

1545

A.E. Armitage et al.

point nearest typical weaning) (Table 1) and at month 12 (Table 2). There were significant correlations in both cohorts of hepcidin and ferritin with sTfR, transferrin (assessed in VPM only), CRP and AGP, suggesting hepcidin is regulated similarly in infants compared to older children/adults.21 Lower birthweight correlated significantly with lower hepcidin concentrations at five and 12 months in VPM, and at five months in VA, likely reflecting how individuals born with smaller iron endowments are at greater risk of

iron depletion during infancy.8 However, we also found strong evidence that the extent of antecedent weight gain (especially weight gain between 0 and 5 months of age) was negatively associated with the hepcidin and ferritin concentrations observed at both five and 12 months of age (Tables 1 and 2). In contrast, there were no significant associations of hepcidin and ferritin concentrations at any time point with subsequent weight gain in either cohort (data not shown). We then constructed multivariate models to investigate

Table 2. Cross-sectional correlations between hepcidin, ferritin, and explanatory variables in infants at 12 months of age.

1546

haematologica | 2019; 104(8)

Rapid growth and hepcidin suppression in infancy

the hierarchy of factors associating with hepcidin and ferritin concentrations at five and 12 months. In these models, higher weight gain since birth remained a significant independent predictor of reduced hepcidin at five months of age in both cohorts (Table 3), and at 12 months in VPM (Table 4), irrespective of birthweight; however, weight change did not associate with ferritin concentration in these analyses. At 12 months of age, hepcidin was positively associated with CRP and season and inversely associated with sTfR, while at both time points and in each cohort we found significant inverse associations of ferritin with sTfR (Tables 3 and 4). Together, these analyses not only highlight associations of hepcidin with markers of erythropoietic drive, inflammation and season that align with previous observations

in older African pre-school children,21 but also reveal consistent evidence that prior weight gain predicts hepcidin (although not ferritin) at key time points across these infant populations, irrespective of birthweight and especially in earlier infancy.

Factors predicting changes in hepcidin and ferritin over time within infants: longitudinal analysis We then utilized the longitudinal data, fitting fixed effects models to examine how changes over time in timevariant parameters (sTfR, CRP, plasma iron, infections, season, weaning status and growth) predicted changes over time in hepcidin and ferritin within infants. Fixed effects models eliminate confounding caused by any timeinvariant characteristic, measured or unmeasured, and

Figure 2. The relative influence of time-variant factors on changes in hepcidin and ferritin over time during the first year of life in Gambian infants: standardized Forest plots summarizing fixed effect models. Plots depict how a change of one standard deviation of an explanatory variable over time induces a change in standard deviation of outcome variables hepcidin (left panels) and ferritin (right panels) over time within a child in (A) the VPM cohort and (B) the VA cohort. Outcome variables are modeled simultaneously as â&#x20AC;&#x153;seemingly unrelated variablesâ&#x20AC;? to account for hepcidin/ferritin correlation, and data represent the pooled analysis of 100 datasets in which any missing data were imputed by multiple imputation, combined using Rubinâ&#x20AC;&#x2122;s combination rules, as described in detail in the Online Supplementary Methods. Weaning was defined as the first recorded occurrence of any type of feeding other than exclusive breastfeeding. sTfR: soluble transferrin receptor; CRP: Creactive Protein; Hb: hemoglobin. The wet season was classified as July-October. Plots depict standardized coefficients with 95% Confidence Intervals. *P<0.05; **P<0.01; ***P<0.001. Interactions of weight change with birthweight group [above or below World Health Organization weight-for-age Z-score (WAZ) of -0.5, i.e. close to the median observed across these two cohorts] and sex were modeled. #P<0.05; ##P<0.01; ###P<0.001; respectively, for differences in the coefficients for weight change with respect to higher birthweight females as the reference group. Unstandardized and standardized coefficients are given in Table 5; equivalent analysis based on the dataset prior to multiple imputation is given in Online Supplementary Table S5. MH: higher birthweight males; ML: lower birthweight males; FH: higher birthweight females; FL: lower birthweight females; VPM: Vaccination and Paediatric Microbiome study; VA: a vitamin A supplementation randomized controlled trial.22

haematologica | 2019; 104(8)

1547

A.E. Armitage et al.

enable estimation of within-child effects, contrasting the above cross-sectional models. Moreover, standardizing coefficients allows comparison of the relative effect sizes of explanatory variables on outcome variables (Tables 5 and 6 and Figure 2). Infection and increases in CRP over time associated with increasing hepcidin, while rising sTfR predicted hepcidin decline within infants, suggesting the models captured effects on hepcidin in expected directions. Increased plasma iron over time associated with decreased hepcidin in both cohorts. In the VPM dataset, the transition from dry to wet season, when infections are more prevalent, also associated with increased hepcidin (Table 5 and Figure 2A). However, consistent with the cross-sectional (i.e. between-infant) analyses, the dynamic predictor of hepcidin decline that had the largest effect size in fixed effects models (i.e. within-infant) in both cohorts was weight gain (Tables 5 and 6 and Figure 2). The effect of weight gain on hepcidin suppression was highly significant in all groups of infants but was most pronounced in males: that is, a given increase in weight had a stronger suppressive effect on hepcidin over time in males (especially those

born in the lower birthweight group in VPM) than in females. Weight gain also significantly predicted ferritin decline in both cohorts (in contrast to the multivariate cross-sectional analyses at 5 and 12 months). The effect of weight gain on ferritin decline, like hepcidin, was larger in males in the VPM cohort (Table 5 and Figure 2A) although no sex difference on ferritin was observed in the VA cohort.

Discussion Hepcidin is the key regulator of iron absorption and dictates whether absorbed iron is sequestered or made available in the circulation.10 Understanding how hepcidin is regulated is therefore vital in understanding the differential drivers of iron status, and how and whether iron interventions aimed at combating ID might be effectively targeted.6 Here, we have taken advantage of serial sampling of infants from two Gambian cohorts to describe changes in hepcidin and ferritin levels during the first year of life, and to examine how they are influenced by different factors using both cross-sectional and time series analysis.

Table 3. Cross-sectional multivariate associations between hepcidin, ferritin and explanatory variables in infants at five months of age.

A. VPM – 5 months – N=114 Outcome variable: HEPCIDIN Plasma iron, µmol/L sTfR, mg/L CRP, mg/L Infection in last 4 weeks Sex Weight change (0-5 months), kg Season Outcome variable: FERRITIN sTfR, mg/L CRP, mg/L Sex Weight change (0-5 months), kg

B. VA – 5 months – N=193 Outcome variable: HEPCIDIN sTfR, mg/L CRP, mg/L Infection in last 4 weeks Weight change (0-5 months), kg Outcome variable: FERRITIN Plasma iron, µmol/L sTfR, mg/L CRP, mg/L Hb, g/dL Sex Weight change (0-5 months), kg

Coefficient

95% CI

-0.127 -0.641 0.218 0.103 -0.050 -5.543 0.247

-0.252 – -0.002 -1.592 – 0.310 -0.076 – 0.513 -0.144 – 0.350 -0.238 – 0.139 -8.786 – -2.300 0.080 – 0.413

0.047 0.186 0.145 0.412 0.606 0.001 0.004

-1.076 0.265 -0.251 -1.828

-2.11 – -0.039 -0.043 – 0.574 -0.463 – -0.039 -0.463 – 1.607

0.042 0.092 0.020 0.297

Coefficient

95% CI

-0.700 0.148 -0.054 -3.307

-1.362 – -0.039 0.033 – 0.263 -0.283 – 0.175 -6.069 – -0.545

0.039 0.012 0.645 0.019

0.057 -1.127 -0.031 1.627 -0.066 -0.173

-0.129 – 0.244 -1.953 – -0.301 -0.175 – 0.113 -1.000 – 4.253 -0.347 – 0.216 -4.364 – 4.018

0.545 0.008 0.670 0.224 0.648 0.935

The outcome variables hepcidin and ferritin were handled using Seemingly Unrelated Regressions, with World Health Organization birth weight-for-age Z-score used as an instrument for weight change. Only variables with P<0.2 in univariate regressions were included in multivariable models, omitting correlated indices as described in the Methods section. Data summarise 100 datasets in which any missing data were imputed using multiple imputation as described in detail in the Methods section and the Online Supplementary Appendix. Infection was coded as “no infection=0” and “occurrence of infection=1’; Season was coded as “dry season (November-June)=0” and “wet season (July-October)=1”; Sex was coded as “female=0” and “male=1”.

1548

haematologica | 2019; 104(8)

Rapid growth and hepcidin suppression in infancy

We demonstrate that a marked decline in iron status through infancy is reflected by progressive reduction in hepcidin concentrations, and that while hepcidin appears to be regulated by similar factors in infants, older children and adults, weight gain is the most prominent dynamic predictor of the extent of hepcidin decline in infants. The data highlight post-natal growth as being strongly associated with the onset of iron depletion, and potentially associated with downstream sequelae, including anemia in vulnerable populations of infants. The decline of hepcidin during infancy observed in both cohorts is consistent with data from previous smaller studies in Zimbabwean and Kenyan infants,33-35 which likewise described more pronounced effects in males. The parallel declines in hepcidin and ferritin, at a time when inflammation (which would be expected to increase both markers) is rising, illustrate the rapidity and severity of the onset of ID. Contrasting the dynamics of hepcidin in African infants, hepcidin levels were reported as stable during the first year in Europeans,36,37 presumably reflecting higher iron provision in weaning diets and lower cumulative exposure to infection and consequent inflammation. Human breastmilk has a low iron content5 and

Gambian babies are predominantly breastfed for the first 6 months and beyond.38 Moreover, as our data demonstrate, inflammation, infection and the season when infections are more prevalent were all associated with relatively raised hepcidin concentrations, even in cases of low iron status, potentially contributing to iron depletion in African infants. One notable observation was the dramatic depletion of plasma iron between birth and 5 months of age to concentrations well below the previously described typical levels in infancy;31 these low levels persisted throughout the remainder of the first year. Availability of plasma iron is critical for supplying iron-demanding tissues. During infancy, iron is essential for erythropoiesis, but is also required by the developing brain which is a key consumer of iron,4 as well as by muscle and for tissue accretion.39 The proportional iron demands of these tissues are not constant through the first year of life: for example, erythropoietin levels and erythropoietic output are relatively low in neonates,32 despite early infancy being a period of rapid growth. Effective adaptive immunity, critical in LMIC infants, also requires iron; this is demonstrated by the recent discovery that a mutation in the Tfrc gene

Table 4. Cross-sectional multivariate associations between hepcidin, ferritin and explanatory variables in infants at five months of age.

A. VPM – 12 months – N=114 Outcome variable: HEPCIDIN sTfR, mg/L CRP, mg/L Infection in last 4 weeks Weight change (0-12 months), kg Season FERRITIN sTfR, mg/L CRP, mg/L Infection in last 4 weeks Sex

B. VA – 12 months – N=193 Outcome variable: HEPCIDIN sTfR, mg/L CRP, mg/L Hb, g/dL Sex Weight change (0-12 months), kg Season Outcome variable: FERRITIN sTfR, mg/L CRP, mg/L Hb, g/dL Sex Weight change (0-12 months), kg Season

Coefficient

95% CI

-1.708 0.221 0.167 -2.833 0.343

-2.568 – -0.849 0.100 – 0.342 -0.068 – 0.401 -5.583 – 0.082 0.104 – 0.582

<0.001 <0.001 0.163 0.044 0.005

-1.200 0.170 0.223 -0.247

-2.225 – -0.175 0.001 – 0.339 -0.089 – 0.537 -0.535 – 0.040

0.022 0.049 0.161 0.091

Coefficient

95% CI

-1.191 0.163 0.330 -0.287 -1.362 0.267

-1.870 – -0.511 0.040 – 0.286 -1.784 – 2.443 -0.485 – -0.089 -4.298 – 1.574 0.081 – 0.453

0.001 0.009 0.760 0.005 0.363 0.005

-1.450 0.109 0.275 -0.055 -1.325 0.112

-2.095 – -0.805 -0.016 – 0.234 -1.844 – 2.395 -0.243 – 0.132 -4.168 – 1.518 -0.077 – 0.301

<0.001 0.087 0.799 0.562 0.361 0.247

The outcome variables hepcidin and ferritin were handled using Seemingly Unrelated Regressions, with World Health Organization birth weight-for-age Z-score used as an instrument for weight change. Only variables with P<0.2 in univariate regressions were included in multivariable models, omitting correlated indices as described in the Methods section. Data summarize 100 datasets in which any missing data were imputed using multiple imputation as described in detail in the Methods section and the Online Supplementary Appendix. Infection was coded as “no infection=0” and “occurrence of infection=1’; Season was coded as “dry season (November-June)=0” and “wet season (July-October)=1”; Sex was coded as “female=0” and “male=1”.

haematologica | 2019; 104(8)

1549

A.E. Armitage et al.

(encoding transferrin receptor 1, TfR1), which abrogated lymphocyte iron acquisition, was the cause of fatal combined immunodeficiency due to an inability to protect against infections.40 Thus, it is conceivable that low plasma iron availability may similarly impair adaptive immunity. The longitudinal analyses paradoxically suggested that changes in plasma iron inversely predicted changes in hepcidin; the reason for this is unclear, but could be driven by events in the first two months, when serum iron and hepcidin appear to differ most in their kinetics. Soluble transferrin receptor is frequently used as a biomarker of tissue iron demand, and is typically associated with erythropoiesis.41 Concentrations of sTfR fell between birth and two months of age in both cohorts, presumably reflecting low erythropoietic demand at this point consequent upon the decline of the high birth hemoglobin concentration and recycling of its iron; however, sTfR concentrations were raised by 12 months of age, especially in males. By this time, elevated sTfR strongly associated with suppressed hepcidin, and fixed effects modeling, demonstrated how increases in sTfR within an infant over time predicted hepcidin reduction (overcoming the hepcidin-stimulatory effect of inflammation). Erythropoietic drive can suppress liver hepcidin production through activity of the recently characterized erythroblast-derived

hormone erythroferrone, liberating iron to meet erythropoietic demands. 18Erythroferrone assays were not available at the time of this analysis, but it will be important to investigate its role during infancy, especially in LMIC settings. In addition, our cross-sectional and longitudinal analyses both also revealed weight gain to be strongly associated with declining hepcidin during infancy. Tissue accretion and erythropoietic expansion are widely acknowledged as consumers of iron in infancy, yet variability in growth rate within and between infants is less frequently considered to be a risk factor for earlier onset of iron depletion and deficiency in infancy than other risk factors (such as premature birth, early cord clamping, low birthweight or early consumption of cow’s milk).6,9 In univariable analysis, lower birthweight did correlate as expected with reduced hepcidin at five months in both cohorts, yet there was more consistent evidence of a relationship between antecedent growth and hepcidin (noting that there was no such relationship of hepcidin or ferritin with subsequent weight gain); moreover, in multivariate cross-sectional models, prior weight gain remained significantly inversely associated with hepcidin, irrespective of birthweight. Our longitudinal fixed effects models added strength to these observations, with more rapid weight gain predicting a more profound decline in hepcidin over

Table 5. The relative influence of time-variant factors on changes in hepcidin and ferritin over time during the first year of life in Gambian infants: fixed effect models – VPM cohort.

Variable Outcome variable: HEPCIDIN Season Infection in last 4 weeks Weaning Plasma iron, µmol/L sTfR, mg/L CRP, mg/L Weight, kg lower birthweight females lower birthweight males higher birthweight males Outcome variable: FERRITIN Season Infection in last 4 weeks Weaning Plasma iron, µmol/L sTfR, mg/L CRP, mg/L Weight, kg lower birthweight females lower birthweight males higher birthweight males

Coefficient (unstandardized)

95% CI (unstandardized)

Coefficient (standardized)

95% CI (standardized)

0.106 0.207 -0.060 -0.080 -0.974 0.105 -1.402 -0.281 -1.064 -0.313

0.013 – 0.200 0.119 – 0.294 -0.165 – 0.046 -0.150 – -0.010 -1.206 – -0.741 0.042 – 0.168 -1.786 – -1.018 -0.687 – 0.125 -1.781 – -0.348 -0.628 – 0.002

0.092 0.141 -0.047 -0.108 -0.328 0.145 -0.411 -0.083 -0.312 -0.092

0.011 – 0.172 0.081 – 0.202 -0.131 – 0.037 -0.202 – -0.015 -0.407 – -0.248 0.059 – 0.231 -0.522 – -0.301 -0.202 – 0.037 -0.520 – -0.104 -0.184 – 0.000

0.026 <0.001 0.265 0.025 <0.001 0.001 <0.001 0.172 0.005 0.052

0.104 0.085 -0.438 -0.048 -1.176 0.069 -1.760 -0.099 -0.626 -0.966

0.033 – 0.176 -0.000 – 0.171 -0.586 – -0.291 -0.124 – 0.028 -1.439 – -0.914 0.017 – 0.120 -2.144 – -1.377 -0.559 – 0.361 -1.080 – -0.172 -1.560 – -0.371

0.065 0.042 -0.251 -0.047 -0.286 0.068 -0.374 -0.021 -0.133 -0.205

0.020 – 0.110 -0.000 – 0.085 -0.337 – -0.166 -0.121 – 0.028 -0.349 – -0.223 0.017 – 0.120 -0.455 – -0.293 -0.119 – 0.077 -0.229 – -0.037 -0.331 – -0.079

0.005 0.050 <0.001 0.215 <0.001 0.010 <0.001 0.672 0.007 0.002

Data represent the pooled analysis of 100 datasets in which any missing data were imputed by multiple imputation, combined using Rubin’s combination rules, as described in detail in Online Supplementary Methods. Unstandardized coefficients indicate the effect of a unit change over time of an explanatory variable on changes over time within a child in the outcome variables hepcidin and ferritin, which were handled using seemingly unrelated regression to account for strong correlations between outcome variables. Standardized coefficients indicate the effect of a change of one standard deviation of an explanatory variable over time on standard deviation of outcome variables over time within a child. Infection was coded as “no infection=0” and “occurrence of infection=1”; Season was coded as “dry season (November-June)=0” and “wet season (JulyOctober)=1”; Sex was coded as “female=0” and “male=1”. Interactions of weight change with birthweight group (groups indicated in italics: above or below WHO weight-for-age z-score of -0.5) and sex were modeled with coefficients/P-values representing differences for weight change relative to higher birthweight females as reference group.

1550

haematologica | 2019; 104(8)

Rapid growth and hepcidin suppression in infancy

time. In the within-subject analyses, there was also a highly significant effect of weight gain on ferritin decline, apparently contrasting the cross-sectional models in which no association of ferritin with weight gain was found; this discrepancy likely reflects the better efficiency of analyzing data longitudinally. The hepcidin-suppressive effect of rapid growth therefore warrants mechanistic evaluation. It may relate to known hepcidin-regulatory signals, such as decreased transferrin saturation and hepatic iron. Growth factors, especially insulin-like growth factor 1 (IGF-1), play important roles in infant growth and development;42 hepatocyte growth factor (HGF), epidermal growth factor (EGF), and platelet-derived growth factor BB (PDGF-BB) are able to down-regulate hepcidin transcription.43,44 These or related growth factors should, therefore, be considered as candidate contributors to hepcidin suppression during infancy. Furthermore, erythroferrone provides a precedent for an iron-demanding process, erythropoiesis, evolving to produce a hepcidin suppressive hormone to facilitate iron supply. Further consideration should be given to the extent to which decreasing transferrin saturation due to, or erythroferrone derived from, an expanding erythron accounts for the observed relationship between rapid growth and hepcidin suppression and iron depletion at

different stages of infancy, compared to growth and development of other tissues such as muscle and brain. Whether non-erythroid iron-demanding tissues that grow rapidly during infancy similarly up-regulate erythroferrone itself or other, as yet undefined, signals should be investigated. The proportional effect of weight gain on hepcidin was also more pronounced in males. This sex difference might reflect differences in hormonal regulation of hepcidin exacerbating effects of weight gain: for example, testosterone suppresses hepcidin in mice45 and humans,46 and is elevated in males during the 'mini-puberty' of early infancy.47 The availability of two well-sized cohorts providing longitudinal data provides a key strength of the analyses presented here, allowing investigation of factors affecting changes in hepcidin and ferritin within infants through the first year. Causal inference using longitudinal data faces two sources of confounding: time-variant confounding and time-invariant confounding. Unlike other regression models, fixed effects panel models control for any timeinvariant confounding, measured or unmeasured. Despite this benefit, our models are still unlikely to fully capture all potential time-variant explanatory variables, limiting inference of causation. Since our analyses were retrospec-

Table 6. The relative influence of time-variant factors on changes in hepcidin and ferritin over time during the first year of life in Gambian infants: fixed effect models – VA cohort.

Variable Outcome variable:HEPCIDIN Season Infection in last 4 weeks Plasma iron, µmol/L sTfR, mg/L CRP, mg/L Hb, g/dL Weight, kg lower birthweight females ower birthweight males higher birthweight males Outcome variable: FERRITIN Season Infection in last 4 weeks Plasma iron, µmol/L sTfR, mg/L CRP, mg/L Hb, g/dL Weight, kg lower birthweight females lower birthweight males higher birthweight males

Coefficient (unstandardized)

95% CI (unstandardized)

Coefficient (standardized)

95% CI (standardized)

0.063 0.116 -0.194 -1.161 0.133 0.115 -1.581 -0.484 -0.745 -1.226

-0.040 – 0.166 0.009 – 0.224 -0.342 – -0.466 -1.513 – -0.809 0.060 – 0.205 -0.610 – 0.840 -2.342 – -0.819 -1.083 – 0.116 -1.229 – -0.260 -1.819 – -0.633

0.038 0.072 -0.204 -0.316 0.160 0.012 -0.347 -0.106 -0.163 -0.269

-0.024 – 0.100 0.005 – 0.140 -0.356 – -0.052 -0.406 – -0.225 0.073 – 0.246 -0.061 – 0.084 -0.516 – -0.179 -0.238 – 0.026 -0.270 – -0.057 -0.398 – -0.140

0.231 0.035 0.009 <0.001 <0.001 0.751 <0.001 0.112 0.003 <0.001

-0.044 0.073 -0.008 -1.762 0.014 -0.158 -2.278 0.273 0.254 -0.164

-0.154 – 0.065 -0.078 – 0.225 -0.137 – 0.120 -2.205 – -1.319 -0.048 – 0.076 -1.354 – 1.037 -3.165 – -1.391 -0.494 – 1.039 -0.606 – 1.115 -1.011 – 0.683

-0.024 0.042 -0.008 -0.442 0.016 -0.015 -0.461 0.055 0.051 -0.033

-0.085 – 0.036 -0.045 – 0.129 -0.132 – 0.115 -0.552 – -0.332 -0.053 – 0.084 -0.126 – 0.096 -0.641 – -0.281 -0.100 – 0.211 -0.122 – 0.225 -0.205 – 0.139

0.425 0.338 0.895 <0.001 0.651 0.792 <0.001 0.481 0.558 0.700

Data represent the pooled analysis of 100 datasets in which any missing data were imputed by multiple imputation, combined using Rubin’s combination rules, as described in detail in Online Supplementary Methods. Unstandardized coefficients indicate the effect of a unit change over time of an explanatory variable on changes over time within a child in the outcome variables hepcidin and ferritin, which were handled using seemingly unrelated regression to account for strong correlations between outcome variables. Standardized coefficients indicate the effect of a change of one standard deviation of an explanatory variable over time on standard deviation of outcome variables over time within a child. Infection was coded as “no infection=0” and “occurrence of infection=1’; Season was coded as “dry season (November-June)=0” and “wet season (JulyOctober)=1”; Sex was coded as “female=0” and “male=1”. Interactions of weight change with birthweight group (groups indicated in italics: above or below WHO weight-for-age z-score of -0.5) and sex were modeled with coefficients/P-values representing differences for weight change relative to higher birthweight females as reference group.

haematologica | 2019; 104(8)

1551

A.E. Armitage et al.

tive investigations of cohorts that had been enrolled for other purposes, certain desirable variables were unavailable: for example, hemoglobin data were not collected in VPM, and for both cohorts, we cannot exclude the possibility that participants took iron-containing supplements, breast-milk substitutes or fortified complementary diets during the study period since this information was not systematically collected. Furthermore, given the known impact of season on iron status in this region,21 it would have been desirable for enrollment to span at least one full year in each cohort. This was not the case in VPM, meaning the majority of VPM 12-month samples were taken in the dry season (Online Supplementary Table S2A), possibly explaining in part the apparent higher prevalence of ID in VPM at 12 months (Online Supplementary Table S3). Nevertheless, we believe our models are still of great value in characterizing predictors of hepcidin and ferritin decline in infancy. Other limitations should also be considered. Self-reporting of infectious episodes may have poor sensitivity and specificity for actual infection. Plasma biomarkers of inflammation may only capture current or recent inflammatory insults; since previous inflammation may contribute to reduced iron status through chronically raised hepcidin, higher resolution tracking of inflammatory events between key time points would be desirable. Finally, whether the effects we describe are maintained in distinct LMIC infant populations, most notably regions with a high malarial burden, should be further investigated. Universal iron interventions for LMIC infants have proven to be relatively ineffective in reducing the burden of anemia, partly because only a proportion of anemia is

References 1. GBD-2016-Disease-and-Injury-Incidenceand-Prevalence-Collaborators. Global, regional, and national incidence, prevalence, and years lived with disability for 328 diseases and injuries for 195 countries, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet. 2017;390(10100):1211-1259. 2. WHO. The global prevalence of anaemia in 2011. Geneva; 2015. 3. Georgieff MK. Long-term brain and behavioral consequences of early iron deficiency. Nutr Rev. 2011;69(Suppl 1):S43-S48. 4. Cusick SE, Georgieff MK. The Role of Nutrition in Brain Development: The Golden Opportunity of the "First 1000 Days". J Pediatr. 2016;175:16-21. 5. Lonnerdal B. Development of iron homeostasis in infants and young children. Am J Clin Nutr. 2017;106(Suppl 6):1575S-1580S. 6. Domellof M, Braegger C, Campoy C, et al. Iron requirements of infants and toddlers. J Pediatr Gastroenterol Nutr. 2014;58(1):119129. 7. Dewey KG, Chaparro CM. Session 4: Mineral metabolism and body composition iron status of breast-fed infants. Proc Nutr Soc. 2007;66(3):412-422. 8. Ziegler EE, Nelson SE, Jeter JM. Iron stores of breastfed infants during the first year of life. Nutrients. 2014;6(5):2023-2034.

1552

thought to be iron responsive.48 Iron absorption is facilitated in the presence of low hepcidin concentrations, but inhibited when hepcidin is raised (such as during infection/inflammation), potentially explaining why many infants respond poorly to iron interventions.20,49,50 Growth rate has previously been linked to reduced ferritin and iron deficiency,51-54 and has been shown to be predictive of hemoglobin response to iron supplementation;55 hepcidin suppression during rapid growth likely explains the latter effect. Since rapid growth predicts greater hepcidin suppression, our data suggest that simple weight monitoring could enable identification of rapidly growing infants to whom oral iron might be advantageously targeted. Acknowledgments The authors would like to thank the study participants and their families, the field teams and laboratory staff. We thank Rita WegmĂźller for overseeing laboratory analyses at the MRC Keneba field station and for helpful discussions. The original VA study was funded by the UK Medical Research Council (MRC) (see below). Funding The original VPM study was funded by the Bill & Melinda Gates Foundation (BMGF). The current analyses were supported by BMGF (OPP 1055865) awarded to the MRC International Nutrition Group which is core-funded by MCA760-5QX00 from the MRC and the UK Department for International Development (DFID) under the MRC/DFID Concordat agreement. AEA and HD receive core-funding through the MRC Human Immunology Unit, MRC Weatherall Institute of Molecular Medicine, University of Oxford (MC_UU_12010/3).

9. Chaparro CM. Setting the stage for child health and development: prevention of iron deficiency in early infancy. J Nutr. 2008;138(12):2529-2533. 10. Ganz T. Systemic iron homeostasis. Physiol Rev. 2013;93(4):1721-1741. 11. Nemeth E, Tuttle MS, Powelson J, et al. Hepcidin regulates cellular iron efflux by binding to ferroportin and inducing its internalization. Science. 2004;306(5704):2090-2093. 12. Drakesmith H, Prentice AM. Hepcidin and the iron-infection axis. Science. 2012;338(6108):768-772. 13. Arezes J, Jung G, Gabayan V, et al. Hepcidin-induced hypoferremia is a critical host defense mechanism against the siderophilic bacterium Vibrio vulnificus. Cell Host Microbe. 2015;17(1):47-57. 14. De Falco L, Sanchez M, Silvestri L, et al. Iron refractory iron deficiency anemia. Haematologica. 2013;98(6):845-853. 15. Gwamaka M, Kurtis JD, Sorensen BE, et al. Iron deficiency protects against severe Plasmodium falciparum malaria and death in young children. Clin Infect Dis. 2012;54(8):1137-1144. 16. Jonker FA, Calis JC, van Hensbroek MB, et al. Iron status predicts malaria risk in Malawian preschool children. PLoS One. 2012;7(8):e42670. 17. Muriuki JM, Mentzer AJ, Kimita W, et al. Iron Status and Associated Malaria Risk Among African Children. Clin Infect Dis.

2018 Sep 14. [Epub ahead of print] 18. Kautz L, Jung G, Valore EV, Rivella S, Nemeth E, Ganz T. Identification of erythroferrone as an erythroid regulator of iron metabolism. Nat Genet. 2014;46(7): 678-684. 19. Pasricha SR, Atkinson SH, Armitage AE, et al. Expression of the iron hormone hepcidin distinguishes different types of anemia in African children. Sci Transl Med. 2014;6(235):235re233. 20. Prentice AM, Doherty CP, Abrams SA, et al. Hepcidin is the major predictor of erythrocyte iron incorporation in anemic African children. Blood. 2012;119(8):19221928. 21. Atkinson SH, Armitage AE, Khandwala S, et al. Combinatorial effects of malaria season, iron deficiency, and inflammation determine plasma hepcidin concentration in African children. Blood. 2014; 123(21):3221-3229. 22. Darboe MK, Thurnham DI, Morgan G, et al. Effectiveness of an early supplementation scheme of high-dose vitamin A versus standard WHO protocol in Gambian mothers and infants: a randomised controlled trial. Lancet. 2007;369(9579):2088-2096. 23. Leroy JF. zscore06: Stata command for the calculation of anthropometric z-scores using the 2006 WHO child growth standards. 2011. Available from:https://econpapers.repec.org/software/bocbocode/ s457279.htm

haematologica | 2019; 104(8)

Rapid growth and hepcidin suppression in infancy

24. Graham JW. Missing data analysis: making it work in the real world. Annu Rev Psychol. 2009;60:549-576. 25. Sterne JA, White IR, Carlin JB, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ. 2009;338:b2393. 26. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate Imputation by Chained Equations in R. J Stat Softw. 2011;45(3):1-67. 27. Rubin DB. Multiple imputation for nonresponse in surveys John Wiley & Sons., 2004. 28. Zellner A. An Efficient Method of Estimating Seemingly Unrelated Regressions and Tests for Aggregation Bias. J Am Stat Assoc. 1962;57(298):348-368. 29. Driscoll JC, Kraay AC. Consistent covariance matrix estimation with spatially dependent panel data. The Review of Economics and Statistics. 1998;80(4):849860. 30. Hoechle D. Robust standard errors for panel regressions with cross-sectional dependence. Stata Journal. 2007;7(3):281. 31. Ritchie RF, Palomaki GE, Neveux LM, Navolotskaia O, Ledue TB, Craig WY. Reference distributions for serum iron and transferrin saturation: a comparison of a large cohort to the world's literature. J Clin Lab Anal. 2002;16(5):246-252. 32. Finne PH, Halvorsen S. Regulation of erythropoiesis in the fetus and newborn. Arch Dis Child. 1972;47(255):683-687. 33. Mupfudze TG, Stoltzfus RJ, Rukobo S, et al. Hepcidin decreases over the first year of life in healthy African infants. Br J Haematol. 2014;164(1):150-153. 34. Jaeggi T, Moretti D, Kvalsvig J, et al. Iron status and systemic inflammation, but not gut inflammation, strongly predict genderspecific concentrations of serum hepcidin in infants in rural Kenya. PLoS One. 2013; 8(2):e57513. 35. Atkinson SH, Uyoga SM, Armitage AE, et al. Malaria and Age Variably but Critically Control Hepcidin Throughout Childhood in Kenya. EBioMedicine. 2015;2(10):14781486. 36. Aranda N, Bedmar C, Arija V, et al. Serum

haematologica | 2019; 104(8)

37.

38.

39. 40.

41.

42.

43.

44.

45.

46.

hepcidin levels, iron status, and HFE gene alterations during the first year of life in healthy Spanish infants. Ann Hematol. 2018;97(6):1071-1080. Berglund S, Lonnerdal B, Westrup B, Domellof M. Effects of iron supplementation on serum hepcidin and serum erythropoietin in low-birth-weight infants. Am J Clin Nutr. 2011;94(6):1553-1561. Eriksen KG, Johnson W, Sonko B, Prentice AM, Darboe MK, Moore SE. Following the World Health Organization's Recommendation of Exclusive Breastfeeding to 6 Months of Age Does Not Impact the Growth of Rural Gambian Infants. J Nutr. 2017;147(2):248-255. Beard JL. Iron biology in immune function, muscle metabolism and neuronal functioning. J Nutr. 2001;131(2S-2):568S-579S. Jabara HH, Boyden SE, Chou J, et al. A missense mutation in TFRC, encoding transferrin receptor 1, causes combined immunodeficiency. Nat Genet. 2016;48(1):74-78. Pfeiffer CM, Looker AC. Laboratory methodologies for indicators of iron status: strengths, limitations, and analytical challenges. Am J Clin Nutr. 2017;106(Suppl 6):1606S-1614S. Netchine I, Azzi S, Le Bouc Y, Savage MO. IGF1 molecular anomalies demonstrate its critical role in fetal, postnatal growth and brain development. Best Pract Res Clin Endocrinol Metab. 2011;25(1):181-190. Goodnough JB, Ramos E, Nemeth E, Ganz T. Inhibition of hepcidin transcription by growth factors. Hepatology. 2012; 56(1):291-299. Sonnweber T, Nachbaur D, Schroll A, et al. Hypoxia induced downregulation of hepcidin is mediated by platelet derived growth factor BB. Gut. 2014;63(12):19511959. Latour C, Kautz L, Besson-Fournier C, et al. Testosterone perturbs systemic iron balance through activation of epidermal growth factor receptor signaling in the liver and repression of hepcidin. Hepatology. 2014;59(2):683-694. Bachman E, Feng R, Travison T, et al. Testosterone suppresses hepcidin in men: a

47.

48.

49.

50.

51.

52.

53.

54.

55.

potential mechanism for testosteroneinduced erythrocytosis. J Clin Endocrinol Metab. 2010;95(10):4743-4747. Forest MG, Sizonenko PC, Cathiard AM, Bertrand J. Hypophyso-gonadal function in humans during the first year of life. 1. Evidence for testicular activity in early infancy. J Clin Invest. 1974;53(3):819-828. Pasricha SR, Armitage AE, Prentice AM, Drakesmith H. Reducing anaemia in low income countries: control of infection is essential. BMJ. 2018;362:k3165. Cercamondi CI, Egli IM, Ahouandjinou E, et al. Afebrile Plasmodium falciparum parasitemia decreases absorption of fortification iron but does not affect systemic iron utilization: a double stable-isotope study in young Beninese women. Am J Clin Nutr. 2010;92(6):1385-1392. Moretti D, Goede JS, Zeder C, et al. Oral iron supplements increase hepcidin and decrease iron absorption from daily or twice-daily doses in iron-depleted young women. Blood. 2015;126(17):1981-1989. Michaelsen KF, Milman N, Samuelson G. A longitudinal study of iron status in healthy Danish infants: effects of early iron status, growth velocity and dietary factors. Acta Paediatr. 1995;84(9):1035-1044. Thorsdottir I, Gunnarsson BS, Atladottir H, Michaelsen KF, Palsson G. Iron status at 12 months of age -- effects of body size, growth and diet in a population with high birth weight. Eur J Clin Nutr. 2003;57(4):505-513. Sherriff A, Emond A, Hawkins N, Golding J. Haemoglobin and ferritin concentrations in children aged 12 and 18 months. ALSPAC Children in Focus Study Team. Arch Dis Child. 1999;80(2):153-157. Yang Z, Lonnerdal B, Adu-Afarwuah S, et al. Prevalence and predictors of iron deficiency in fully breastfed infants at 6 mo of age: comparison of data from 6 studies. Am J Clin Nutr. 2009;89(5):1433-1440. Domellof M, Dewey KG, Lonnerdal B, Cohen RJ, Hernell O. The diagnostic criteria for iron deficiency in infants should be reevaluated. J Nutr. 2002;132(12):36803686.

1553

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(8):1554-1564

Red Cell & its Disorders

Clinical and biological features in PIEZO1hereditary xerocytosis and Gardos channelopathy: a retrospective series of 126 patients

Véronique Picard,1,2 Corinne Guitton,3 Isabelle Thuret,4 Christian Rose,5 Laurence Bendelac,1 Kaldoun Ghazal,6 Patricia Aguilar-Martinez,7 Catherine Badens,8 Claire Barro,9 Claire Bénéteau,10 Claire Berger,11 Pascal Cathébras,12 Eric Deconinck,13 Jacques Delaunay,14 Jean-Marc Durand,15 Nadia Firah,16 Frédéric Galactéros,17 Bertrand Godeau,18 Xavier Jaïs,19 Jean-Pierre de Jaureguiberry,20 Camille Le Stradic,21 François Lifermann,22 Robert Maffre,1 Gilles Morin,23 Julien Perrin,24 Valérie Proulle,1 Marc Ruivard,25 Fabienne Toutain,26 Agnès Lahary27 and Loïc Garçon1,28

Laboratoire d'Hématologie, Center Hospitalier Universitaire (CHU) Bicêtre, Assistance publique – Hôpitaux de Paris (AP-HP), Le Kremlin-Bicêtre; 2Université Paris Sud Paris Saclay, Faculté de Pharmacie, Chatenay Malabry; 3Service de Pédiatrie Générale, CHU Bicêtre et Filière MCGRE, AP-HP, Le Kremlin-Bicêtre; 4Service de Pédiatrie, Hôpital La Timone, Aix Marseille University, Marseille; 5Service d’Oncologie et d’Hématologie, Hôpital Saint Vincent de Paul, Lille; 6Laboratoire de Biochimie, CHU Bicêtre, AP-HP, Le KremlinBicêtre; 7Laboratoire d’Hématologie Biologique, CHU Saint-Eloi, Montpellier; 8Service de Génétique Médicale, Hôpital La Timone, Marseille; 9Laboratoire d’Hématologie Biologique, CHU Grenoble, Grenoble; 10Génétique Médicale, CHU Nantes, Nantes; 11Service d’Hématologie-Oncologie Pédiatrique, CHU, Saint-Etienne; 12Service de Médecine Interne, CHU Saint-Etienne; 13Service d’Hématologie, CHU Jean Minioz, Besançon; 14Centre Catherine de Sienne, Nantes; 15Service de Médecine Interne, Hôpital La Timone, Marseille; 16Service de Pédiatrie, Centre Hospitaliere (CH) Pau; 17Centre de Référence des Syndromes Drépanocytaires Majeurs, Hôpital Henri-Mondor, AP-HP, Créteil; 18Service de Médecine Interne, CHU Henri Mondor, AP-HP, Créteil; 19Service de Pneumologie, CHU Bicêtre, AP-HP, Le Kremlin-Bicêtre; 20Service de Médecine Interne, Sainte Anne, Toulon; 21 Service de Pédiatrie –Néonatologie, CH de Bretagne Sud, Lorient; 22Service de Médecine interne, CH Dax; 23Génétique Médicale, CHU Amiens; 24Laboratoire d'Hématologie, CHRU Nancy; 25Service de Médecine Interne, CHU Estaing, Clermont-Ferrand; 26Service d’Hémato-Oncologie Pédiatrique, CHU Sud, Rennes; 27 Laboratoire d'Hématologie, CHU Rouen and 28Equipe d’Accueil 4666 HEMATIM Université de Picardie Jules Verne and Service d’Hématologie Biologique, CHU Amiens, France 1

Correspondence: LOÏC GARÇON Garcon.Loic@chu-amiens.fr Received: September 4, 2018. Accepted: January 15, 2019. Pre-published: January 24, 2019. doi:10.3324/haematol.2018.205328 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1554 ©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.

1554

ABSTRACT

e describe the clinical, hematologic and genetic characteristics of a retrospective series of 126 subjects from 64 families with hereditary xerocytosis. Twelve patients from six families carried a KCNN4 mutation, five had the recurrent p.Arg352His mutation and one had a new deletion at the exon 7-intron 7 junction. Forty-nine families carried a PIEZO1 mutation, which was a known recurrent mutation in only one-third of the cases and private sequence variation in others; 12 new probably pathogenic missense mutations were identified. The two dominant features leading to diagnosis were hemolysis that persisted after splenectomy and hyperferritinemia, with an inconstant correlation with liver iron content assessed by magnetic resonance imaging. PIEZO1-hereditary xerocytosis was characterized by compensated hemolysis in most cases, perinatal edema of heterogeneous severity in more than 20% of families and a major risk of post-splenectomy thrombotic events, including a high frequency of portal thrombosis. In KCNN4-related disease, the main symptoms were more severe anemia, hemolysis and iron overload, with no clear sign of red cell dehydration; therefore, this disorder would be better described as a ‘Gardos channelopathy’. These data on the largest series to date indicate that PIEZO1-hereditary xerocytosis and Gardos channelopathy are not the same disease although they share hemolysis, a high rate of haematologica | 2019; 104(8)

A retrospective series of 126 HX cases

iron overload and inefficient splenectomy. They demonstrate the high variability in clinical expression as well as genetic bases of PIEZO1-hereditary xerocytosis. These results will help to improve the diagnosis of hereditary xerocytosis and to provide recommendations on the clinical management in terms of splenectomy, iron overload and pregnancy follow-up.

Introduction Hereditary stomatocytoses are dominant red cell membrane disorders characterized by an increased cation leak through the membrane and a subsequent alteration in cell hydration.1,2 Also called hereditary xerocytosis (HX) (MIM #194380), dehydrated stomatocytosis is the most frequent of these disorders, with an estimated incidence of 1:50000 births.3,4 It is characterized by chronic hemolysis5 as well as non-hematologic features including transient perinatal edema, pseudohyperkalemia,6 iron overload7 and postsplenectomy thromboembolic complications8 that often dominate the phenotype and delay the diagnosis. In addition, diagnosis is complex because of the lack of an easily available biological test. Osmotic gradient ektacytometry is the most efficient diagnostic tool so far,9 showing a typical left-shifted curve with increased osmotic resistance. However, it is only performed in a few specialized laboratories. HX is most often caused by gain-of-function mutations in PIEZO1 which encodes a mechanosensitive ion channel that translates a mechanic stimulus into an increased intracellular calcium concentration.10–12 Recent reports have shown that PIEZO1 mutants induce an inappropriate increase in intracellular Ca2+ which in turn activates the Gardos channel inducing potassium leak, water loss and consequently erythrocyte dehydration.13 Of note, erythrocyte dehydration related to PIEZO1 gain-of-function mutations has also been associated with protection against malaria.14 In HX, it can be assumed that PIEZO1 mutations lead to a stronger dehydration and subsequently to hemolysis. Besides PIEZO1 mutations, gain-of-function mutations in KCNN4, encoding the Gardos channel, directly cause another subset of HX.15,16 Up to now, with the exception of a recent report,17 the clinical and biological features of HX have been mostly described in small series. Large-scale clinical and biological descriptions at diagnosis and comparisons with the underlying genotypes are required. We report here a retrospective study describing the clinical and biological features in the largest series to date including 126 patients from 64 families with HX diagnosed based on ektacytometry and/or genetics between 1993 and 2017.

Methods Patients and data collection This retrospective cohort included 126 patients from 64 apparently unrelated families who were diagnosed with HX when tested for red cell membrane diseases in a context of chronic hemolysis after elimination of other causes such as hemoglobinopathies except otherwise mentioned, at the hematology laboratory of Bicêtre hospital, between 1993 and 2017. The study was conducted according to French ethics regulations regarding retrospective, non-interventional studies. Data were collected directly from clinihaematologica | 2019; 104(8)

cal records and through the French Cohort of Hereditary Stomatocytosis, declared to the French National Commission on Informatics and Liberty. Twenty families have been previously reported and are presented here with complementary phenotypic or genotypic data.10,11,13,15,18,19

Phenotypic red cell analysis Red blood cell analyses were performed in the same laboratory and were processed within 24 h after blood drawing and transport at 4°C. Cell counts, red cell and reticulocyte indices were measured using an ADVIA2120 analyzer (Siemens®). Blood smears were observed after May-Grünwald-Giemsa staining using standard methods. Screening for red cell membrane disorders was performed by osmotic gradient ektacytometry, as previously described.20–22

Gene analysis Genetic analysis was performed with informed consent according to the Declaration of Helsinki. Genomic DNA was extracted from blood samples collected on EDTA using the QIAamp or QIASymphony DSP Midi blood DNA extraction kit (Quiagen®). The PIEZO1 and KCNN4 coding exons and exon-intron boundaries were sequenced as described in the Online Supplementary Data. Each sequence variation with an allele frequency under 5% in the GnomAD v2.0.1 database was recorded. Family studies were performed when possible and pathogenicity scores were calculated using various bioinformatic software (SIFT 6.2.0, Mutation Taster 2013, Polyphen-2 2.2.6). The predicted pathogenicity is described in Table 1.

Statistical analysis Statistical analyses were performed using parametric tests with two-tailed P values. Statistical significance used was α=0.05. We used the Student t-test for quantitative variables. All numerical values are expressed as mean values ± standard error of mean.

Results Patients and clinical features at diagnosis We identified 126 patients with HX, including 64 probands and 62 family members, between 1993 and 2017. The diagnosis of HX was based on clinical phenotype, red cell parameters and morphology, a normal eosin5-maleimide (EMA) binding test and a typical osmotic gradient ektacytometry curve for 103 patients, and genetic testing with non-typical or normal ektacytometry for 12 patients. Eleven patients presenting typical clinical or biological features and belonging to families with genetically proven HX were also included although they had no ektacytometry or genetic analysis. For probands, the mean age at diagnosis was 32±18 years (n=64), with the range extending from the perinatal period to more than 88 years old (Figure 1A). Clinical and biological features at diagnosis were comparable between probands and family members including mainly non1555

V. Picard et al.

spherocytic chronic hemolysis in 84% of all cases, hyperferritinemia (36%), thrombosis after splenectomy (11%), and perinatal edema in 17% of all cases (Figure 1B). Cholelithiasis was noted in 52% of cases (n=67) with the incidence increasing with age: 22% before 20 years old (n=23), 62% between 20 and 40 years (n=29) and 80% after 40 years (n=15). Splenomegaly was noted at diagnosis in 56% of cases (n=47): 67% before the age of 20 years vs. 51% after the age of 40 years . The mean bilirubin and haptoglobin levels were, respectively, 2.5±1.4 mg/dL (n=36) and 0.28±0.24 g/L (n=24), among whom only nine had a level <0.1 g/L). All other notable medical issues are listed in Online Supplementary Table S1.

Biological diagnosis and genetic testing in hereditary xerocytosis Ektacytometry was performed for 115 patients (90%), including 62/64 index cases and 53/62 family members. Genetic analysis was available for 103 subjects (81%) including 55 index cases (86%). Overall, 56 probands (87%) were diagnosed thanks to a typical dehydrated osmotic gradient ektacytometry profile (Online Supplementary Figure S1) and six (6/56, 11%) following genetic testing. Two index cases (one fetus and one neonate) died and could not be tested; the diagnosis was made after family study. Out of the 56 patients with a typical osmotic gradient ektacytometry profile, 49 subjects could be genotyped. Rare PIEZO1 coding sequence variations were recorded for all these families whereas no variation was detected in KCNN4 (Table 1 and Online Supplementary Table S2). These PIEZO1 sequence variations co-segregated with the disease in 30/49 informative families. In 19/49 families (39%), a known PIEZO1 recurrent mutation in exon 51 was identified: (i) p.Leu2495_Glu2496dup (10 probands, 20 subjects; in 1 subject, it was associated with a second substitution p.Val2474Met), (ii) p.Arg2456His (6 probands, 15 subjects), p.Arg2488Gln (3 probands, 6 subjects, associated with the polymorphisms p.Gly718Ser in cis in 1 family and (iii) p.Arg1925Trp in another family). The other 30 families (61%) carried rare private PIEZO1 missense sequence variations (n=34). Six families were described or carried known mutations considered pathogenic (in 1 case, 2 mutations in cis: p.Gly782Ser and p.Arg808Gln); 15 families carried new unique mutations that we scored as probably pathogenic (n=11) or of unknown significance (n=4) as defined in Table 1. A probably pathogenic sequence variation in exon 16 (p.Asp669Tyr) was present in two apparently unrelated families. In seven families, we identified two associated missense sequence variations; in two of them, one mutation was scored probably pathogenic (p.Val598Leu and p.Gly2433Arg) and was associated with the quite frequent p.Pro2510Leu substitution, the other mutations were scored as being of ‘unknown significance’ and their role needs to be further established. Overall, ten families (20%) carried two PIEZO1 rare sequence variations, in cis in three families, and whose transmission could not be defined in seven (Online Supplementary Table S2). Finally, one subject carried an apparently homozygous mutation, possibly due to hemizygosity since from exon 5 to 51, all PIEZO1 variations were homozygous (data from the parents were not available). This patient had chronic compensated hemolysis with a hemoglobin concentration of 175 g/L, reticulocyte count of 500x109/L, 1556

a moderate increase in ferritin level and a typical osmotic gradient ektacytometry profile. In one case we only found the PIEZO1 c.1013C>A (p.Ser 338Tyr) substitution scored as a polymorphism. Overlapping features with lymphedema were identified in two families: one patient with a mild HX phenotype presented two sequence variations (p.Leu939Met, p.Phe2458Leu) previously described in cis in the parent of a patient with PIEZO1-related lymphedema;23 three out of four related patients carrying two substitutions (p.Gly782Ser and p.Arg808Gln) in cis in exon 18 co-segregating with the disease through three generations presented with severe perinatal edema, including one with persistent lymphedema during adulthood. In six families (12 patients), osmotic gradient ektacytometry did not lead to the diagnosis of HX, since the profiles were either atypical (n=1) or normal (n=5) (Online Supplementary Figure S1). All index cases had chronic hemolytic anemia/hyperferritinemia, four of them had a family history of undiagnosed dominant hemolysis. Gene analysis identified a mutation in KCNN4 in all of them (with no PIEZO1 nucleotide variation). Five families, two

Figure 1. Main clinical and biological features at diagnosis of the probands and their family members. (A) Repartition of age at diagnosis for the 64 index cases. The mean age was 32±18 years (range, 0-88). (B) Biological and clinical features at the time of diagnosis of hereditary xerocytosis (HX) in the 64 index cases and 62 family members; only post-splenectomy thromboses are shown; other: B19 parvovirus infection (n=1), persistent isolated jaundice after birth (n=1), systematic exploration in a context of familial porphyria leading to an “incidental” diagnosis of PIEZO1-related HX (n=1). The figure does not show one case of extramedullar hematopoiesis. NSCH: non-spherocytic chronic hemolysis; PNE: perinatal edema; NA: data not available (4 patients).

haematologica | 2019; 104(8)

A retrospective series of 126 HX cases Table 1. Heterozygous rare PIEZO1 coding sequence variations identified in families with hereditary xerocytosis. (overall allele frequency <5% as reported in the GnomAD v2.0.1 database, http://www.gnomad.broadinstitute.org).

Mutation in Variant dbSNP NM_001142864.2 Exon

Amino acid

Frequency GnomAD %

Cons nt AA

rs112081600

c.1013C>A

p.Ser338Tyr

3.86

None

c.1792G>A

p.Val598Met

None

p.Val598Leu

p.Met605Val

0.0016

None

c.1792G>C (1) c.1813A>G (2) c.1815G>A

p.Met605Ile

None

c.2005G>T

p.Asp669Tyr

None

c.2042T>C

p.Phe681Ser

rs755885744

p. Gly718Ser

0.004

p.Gly782Ser

0.41

p.Arg808Gln

0.41

p.Leu939Met

0.11

None

c.2152G>A (3) c.2344G>A (4) c.2423G>A (4) c.2815C>A* (5) c.4069A>G

p. Ile1357Val

None

c.4071C>G

p.Ile1357Met

rs587776990

c.4073G>C

p.Arg1358Pro

rs767365106

c.4556A>C (6) c.5728G>A (6) c.5773C>T (7) c.5981C>T

p.Gln1519Pro

0.038

p.Glu1910Lys

0.19

p.Arg1925Trp

0.054

p.Ser1994Phe

p.Ala2003Thr

None

c.6007G>A (8) c.6008C>A

p.Ala2003Asp

None

c.6016G>A

p.Val2006Ile

None

c.6019A>C

p.Met2007Leu

rs587776989

c.6058G>A**

p.Ala2020Thr

rs556794769

p.Arg2110Gln

0.0026

rs587776991

c.6329G>A Hmz c.6380C>T

p.Thr2127Met

0.0037

+/+/+/+/+ +/+/+/+ +/+ + -+/+ +/+/+/+ + ++ -+/-+ + + +/+ +/+/+/+ + + + + +/+ +

0.001 B 0.998 Pr 0.993 Pr 0.952 Pr 0.968 Pr 0.956 Pr 0. 973 Pr 0.001 B 0.921 Pr 0.638 Po 0.999 Pr 0.985 Pr 0.998 Pr 0.998 Pr 0.827 Po 0.001 B 0.332 B 0.024 B 0.124 B 0.305 B 0.005 B 0.068 B 0.579 Po 0.649 Po 0.998 Pr

0.52 T 0.1 T 0.32 T 0.05 T 0.08 T 0.05 D 0 D 0.69 T 0.21 T 0.1 T 0.05 D 0 D 0 D 0 D 0.31 T 0.86 T 0.04 D 0.01 D 0.64 T 0.03 D 0.26 T 0 D 0 D 0.24 T 0.02 D

1 PM 0.998 DC 0.997 DC 0.993 DC 0.983 DC 1 DC 1 DC 1 PM 1 DC 0.999 DC 0.996 DC 1 DC 0.995 DC 1 DC 0,76 DC 1 PM 1 PM 1 DC 1 DC 1 DC 0.994 DC 1 DC 1 DC 1 DC 1 DC

None

c.6451T>C

p.Cys2151Arg

+ +

0.997 Pr

0 D

1 DC

None

rs200970763 rs202103485 rs201226914

rs200929552 rs201442593 None None

Predicted mutation effect scores PolyPhen-2 SIFT MT Predicted HumVar score value pathogenicity

n family (subject)ยง

Reported family (subject) ref

143 (2)

121 (1)

1(1)13

132 (1)

155 (3)

19 (1)

227,52 (4)

114 (1)

1(1) 13

111 (2)

1 (2) 11, 18

147 (4)

1 (2) 11,18

147 (4)

1 (2) 11, 18

134 (1)

*23

138 (2)

158 (1)

137 (1)

1 (1)10

18 (1)

157 (1)

139 (1)

124 (1)

113 (2)

1 (4) 11, 18

142 (1)

154 (3)

12 (3)

**10

141 (1)

16 (1)

1 (1)10

117 (3)

continued on the next page

haematologica | 2019; 104(8)

1557

V. Picard et al. continued from the previous page

Mutation in Variant dbSNP NM_001142864.2 Exon

Amino acid

Frequency GnomAD %

Cons nt AA + + + + + + +/+/+ +/+ +/+ /+ +/+/+ +/+ +/+ + +/+ +/nd

None

c.6479C>T

p.Pro2160Leu

None

c.6574C>A

p. Leu2192Ile

None

c.6601G>T

p.Val2201Phe

rs563555492

c.6829C>A (2) c.6922C>G

p.Leu2277Met

0.53

p.Gln2308Glu

c.7297G>C (9) c.7367G>A

p.Gly2433Arg

p.Arg2456His

c.7374C>G* (5) c.7391A>C

p.Phe2458Leu

p.His2464Pro

c.7420G>A (10) c.7463G>A (3, 7) c.7467G>C

p.Val2474Met

0.016

p.Arg2488Gln

0.00066

p.Glu2489Asp

c.7471C>T (8) c.7479_7484dup (10) c.7529C>T (1, 9)

p.Arg2491Trp

0.029

p.L2495_E2496 dup p.Pro2510Leu

None None rs587776988 rs202127176 None rs200243384 rs749288233 None rs201746476 rs587776992 rs61745086

Predicted mutation effect scores PolyPhen-2 SIFT MT Predicted HumVar score value pathogenicity

0.983 Pr 0.754 Po 0.826 Po 0.873 Po 0.437 B 0.811 Po 0.998 Pr 0.557 Po 0.394 B 0.973 Pr 0.978 Pr 0.995 Pr 0.892 Pr nd

0 D 0.06 T 0.01 D 0.07 T 0.21 T 0.01 D 0 D 0.25 T 0.07 T 0.04 D 0 D 0 D 0.02 D nd

1 DC 1 DC 1 DC 0.676 DC 0.994 DC 1 DC 1 DC 1 DC 1 DC 1 DC 1 DC 1 DC 1 PM nd

n family (subject)§

Reported family (subject) ref

145 (1)

153 (1)

128 (2)

155 (3)

149 (1)

17 (1)

61,4,12,33,40,56 (15)

2 (8) 19

134 (1)

*23

126 (2)

122 (1)

311.31.57 (6)

1(1) 11, 18

146(3)

124 (1)

105.15.16.22.25 (20) 27.32 (2)

8(14) 10

29.36.44.50.51

0.68

+ +

0.991 Pr

0 D

1 DC

Cons: phylogenetic conservation of nucleotide (nt) and amino acids (AA) according to Alamut software v2.10.0 (-- not conserved, - weakly conserved, +/- moderately conserved, + highly conserved). Predicted mutation effect scores from Polyphen-2 v2.2.6 (B: benign, Po: possibly damaging, Pr: probably damaging), SIFT v6.2.0 (T: tolerated, D: deleterious when score <0.05) and Mutation Taster v2013 (PM: polymorphism, DC: disease causing) algorithms. Pathogenicity: Y pathogenic as described in previous reports; PP scored probably pathogenic: mutation identified in a family with typical dehydrated ektacytometry, absent in population databases, affecting a moderately or highly conserved amino acid and predicted as pathogenic by at least two algorithms, PM polymorphism: sequence variation known in population databases and amino acid weakly or not conserved and predicted as tolerated by at least two algorithms, US: unknown significance if not pathogenic, probably pathogenic or polymorphism. §: the data shown for each mutation are: total number of families, identification of each family according to Online Supplementary Table S2 (in superscript), and the total number of patients (in brackets). NB: c.6329G>A was homozygous (hmz). *Both sequence variations present on the same allele were described in a parent of a child with PIEZO1-linked lymphedema23 **This mutation was described in an apparently unrelated family10

already reported and three new families, carried the recurrent c.1055G>A KCNN4 mutation leading to the p.Arg352His Gardos channel substitution.15 The last family included three members with marked anemia and splenomegaly; all three carried a novel KCNN4 28 bp deletion (c.1109_1119+17del). This deletion encompassed the exon-intron 7 junction and its consequences are not known so far, although the use of an alternate splice site can be hypothesized.

Hematologic features at diagnosis The mean hemoglobin level of the whole cohort was 131±20 g/L (n=115). Most subjects with HX (68%) were not anemic at diagnosis and presented with compensated hemolysis. No patient was transfusion-dependent on a regular basis; occasional transfusions were necessary for 1558

eight patients. Analysis of red cell indices indicated a trend towards macrocytosis (MCV 97±8 fL) and a high reticulocyte count (294±102x109/L). The mean corpuscular hemoglobin concentration was 351±13 g/L, with 38% of the patients having a level above 360 g/L (Figure 2A). After exclusion of neonates, 29 patients out of 97 from 20 families (30%) had a hemoglobin level below 120 g/L, including 8/12 KCNN4-mutation carriers. Interestingly, seven adults presented at diagnosis with a hemoglobin level above 160 g/L, including two who were referred for “polycythemia” (hemoglobin level of 181 g/L and 175 g/L), with hemolytic features. All these patients carried a PIEZO1 mutation (Figure 2B). Regarding neonatal and childhood hematologic features, 19 cases were diagnosed before the age of 1 year (including 11 probands). Focusing on 15 cases with suffihaematologica | 2019; 104(8)

A retrospective series of 126 HX cases

cient data, seven had neonatal jaundice, requiring ex sanguino transfusion in one case and phototherapy in the other six. During the first month of life, anemia was more marked, 48% having a hemoglobin level below 120 g/L (Figure 2B). Five infants were transfused during their first year of life. After the age of 1 year, no patient was regularly transfused except one KCNN4-HX infant with severe anemia requiring in utero then post-natal transfusions, who underwent splenectomy with a partial improvement thereafter.13 These data demonstrate the more severe hematologic phenotype after birth, and the progressive improvement as the patients become older. Focusing on hematologic parameters for each geno-

type, KCNN4-HX cases were characterized by a lower hemoglobin level (102±13 vs. 134±19 g/L, P<0.001), reticulocyte count (178.1±63.1 vs. 307.5±106 x109/L, P<0.001) and mean corpuscular hemoglobin concentration (332±12.9 vs. 354±24.1 g/L, P<0.05) than PIEZO1-HX cases (Figure 2C-E). The normal mean corpuscular hemoglobin concentration as well as the non-typical ektacytometry profile reflected the absence of clear erythrocyte dehydration in KCNN4-HX. In terms of red cell morphology, stomatocytes were noted on the blood smear in the majority of PIEZO1-HX (65%) samples and were considered as “few” or “rare” on semi-quantitative evaluation in 75% of them.

Figure 2. Main hematologic data and red cell indices in PIEZO1 - and KCNN4-hereditary xerocytosis. (A) Hematologic features in the whole population of patients (index cases + family members positively tested): results are shown as mean hemoglobin (Hb, g/L), mean corpuscular volume (MCV, fL), reticulocyte count (Ret, x109/L) and mean corpuscular hemoglobin concentration (MCHC, g/L). (B) Percentages of patients for each hemoglobin subgroup in the neonatal period (age <1 month, n=14, black boxes) in comparison with adults and infants after the age of 1 month (n=97, gray boxes). Forty-eight percent of neonates had a hemoglobin value under 100 g/L, while in 70% of adults and infants older than 1 month, hemoglobin level was in the normal range for the age. (C) Hemoglobin level was higher in PIEZO1-hereditary xerocytosis (HX) (n=80) than in KCNN4-HX (n=12): mean hemoglobin 134±19 g/L vs. 102±13, P<0.001. (D) Reticulocyte count in PIEZO1-HX vs. KCNN4-HX: 307.5±106 (x109/L) vs. 178.1±63.1 (x109/L), P<0.001. (E) MCHC in PIEZO1-HX vs. KCNN4-HX: 354±24.1 vs. 332 ±12.9 g/L, P<0.05. The MCHC value was obtained on an ADVIA2120 blood analyzer. (F) Frequency of typical stomatocytes on peripheral blood smear examination between PIEZO1-HX (n=63) and KCNN4-HX (n=10): +: rare, ++: few, +++: many.

haematologica | 2019; 104(8)

1559

V. Picard et al.

Stomatocytes were absent in ten of the 12 cases of KCNN4-HX and rare in the other two cases (Figure 2F). Finally, with regards to PIEZO1-HX, we compared carriers of p.Arg2456His (n=15) and p.Leu2495_Glu2496dup (n=20) to those with all other mutations. Hemoglobin level was lower for both recurrent mutations, although only statistically significantly for p.Leu2495_Glu2496dup [mean hemoglobin: 121.3±11.1 g/L (p.Leu2495_Glu2496dup) vs. 130.6±14.5 g/L (p.Arg2456His) vs. 143.7±16.7 g/L (all other mutations); P<0.05]. There was no difference in terms of reticulocyte count (290±74 x109/L vs. 344±76 x109/L vs. 325±131 x109/L, P=NS).

1560

Splenectomy and thrombotic events We recorded thrombotic events in four non-splenectomised patients: a 51-year old man with transient cerebral stroke, a 31-year old female HbS carrier with splenic infarction who was then splenectomized and presented severe thrombotic events after splenectomy, an 18-year old female with painful splenic infarcts and a 52-year old man with hepatic artery thrombosis after liver transplantation for hepatocarcinoma. In contrast, thrombotic events were frequent in splenectomized patients. Overall, 16 patients underwent splenectomy, including eight PIEZO1-HX patients from eight families and four KCNN4-subjects from two families. Genetic diagnosis

Figure 3. Hyperferritinemia and iron overload. (A, B) Ferritin level at diagnosis was correlated to the age of patients: the later in life the diagnosis was made, the higher the ferritin level was: the mean ferritin level was 273 ±141 ng/mL in patients < 20 years vs. 717±441 ng/mL in patients between 20-40 years (P<0.05) vs. 1409±653 ng/mL in those older than 40 years (P<0.05). (C) Percentages of patients receiving chelation therapy, depending on the age at diagnosis. (D) Efficiency of iron chelation on liver iron content (LIC), evaluated by magnetic resonance imaging (MRI), between patients at diagnosis (n=20) and patients at last follow-up (n=14): mean LIC: 200±103 μmol/g vs. 88±42 μmol/g, P<0.001. (E, F) Correlation between LIC assessed by liver MRI and ferritin level (< or > 1000 ng/mL) for patients for whom these two data were available simultaneously: LIC: 318±31 μmol/g for patients with ferritin >1000 ng/mL (n=7) vs. 113±68 μmol/g for patients with ferritin <1000 ng/mL (n=25), P<0.001. (E) Eight patients with a ferritin level below 1000 ng/mL already had a LIC >150 μmol/g, and (F) no clear correlation was found between LIC and ferritin below 1000 ng/mL.

haematologica | 2019; 104(8)

A retrospective series of 126 HX cases

was not available for four patients. Splenectomy was performed for persistent hemolysis and/or symptomatic splenomegaly before the diagnosis of HX was made, with a mean delay of 15 years between splenectomy and diagnosis (range, 1-32 years). The mean age at splenectomy was 24 years (range, 4-41 years). Splenectomy did not abrogate hemolysis: the mean hemoglobin level and reticulocyte count were 112±19 g/L and 280±134 x109/L, respectively, in splenectomized patients. Twelve (75%) splenectomized patients (8 patients with PIEZO1 mutation and 4 not genotyped) experienced 19 thrombotic events, including five portal thromboses (1 previously reported13), five cerebral strokes, six episodes of venous thromboembolism and three cases of chronic thromboembolic pulmonary hypertension (Online Supplementary Figure S2). Thrombotic events occured with a mean delay of 13±8.6 years after splenectomy. All splenectomized PIEZO1-HX subjects (8/8) developed thrombotic events and received long-term anticoagulation therapy (Online Supplementary Figure S2). None of the four KCNN4-HX splenectomized patients experienced thrombosis, with a mean follow-up of 26.5 years (range, 2-44 years). One non-genotyped patient with severe thrombotic events after splenectomy (patient #5, Online Supplementary Figure S2) received anagrelide therapy for a JAK2V617F-negative essential thrombocythemia. Two patients presented heterozygous globin gene mutations in association with HX and experienced severe thrombosis after splenectomy: a woman with AS trait (patient #10, Online Supplementary Figure S2),24 and a male with PIEZO1HX, β-thalassemia trait and α-globin gene triplication (patient #11, Online Supplementary Figure S2). Both experienced cerebral stroke and chronic thromboembolic pulmonary hypertension.

Occurrence of perinatal edema in hereditary xerocytosis Twenty-two patients from 13 families had a known history of perinatal edema. We focused on 19 patients, from 11 families, with well-documented records (Online Supplementary Table S3). In 13 cases, no family history of HX was known when the perinatal edema occurred. The diagnosis of HX was made at birth or in the first month in 12 cases, in utero in two cases, or retrospectively, later in life, in a context of non-spherocytic chronic hemolysis and/or recurrence of perinatal edema during a second pregnancy in five cases. Genetic data were available for ten families (18 patients) and revealed PIEZO1 mutations in all of them. Recurrent mutations were found in five families and private sequence variations in the other five. Five families had histories of recurrent perinatal edema. In family 1, the grandmother had a history of fetal loss due to edema. In families 2, 10 and 11, perinatal edema was noted in two siblings; no data were available for the other affected members. In family 4, all affected members had a history of perinatal edema. In the other families, recurrence of perinatal edema could not be evaluated because of the absence of other HX cases and/or of other pregnancies. In terms of severity, we observed one death in utero at 27 weeks of gestation, one medical termination of pregnancy due to severe hydrops at 28 weeks of gestation, and one death 15 days after birth with refractory effusions and cerebral edema. Five cases required in utero punctures, and five required post-natal punctures. The forms of perinatal edema were hydrops (n=6), pleural effusion (n=5), ascites (n=11), hydramnios (n=2), jugular cyst (n=1), subcutahaematologica | 2019; 104(8)

neous edema (n=3), pericardial effusion (n=2) and hygroma (n=1). Anemia in a context of perinatal edema was documented in three cases, two of whom received in utero transfusions; the last one underwent ex sanguino transfusion at 23 weeks of gestation. The evolution was favorable in 14 cases with complete resorption of the perinatal edema, except for one patient who still presented moderate lymphedema of the lower limbs at adult age.

Hyperferritinemia Ferritin level at diagnosis was available for 49 patients. The mean ferritin level was 764±480 ng/mL and correlated with age (Figure 3A,B). The mean ferritin level at diagnosis was 1702±1048 ng/mL in adult KCNN4-HX (n=5) and 656±428 ng/mL in PIEZO1-HX (n=40). This hyperferritinemia was not related to transfusions since no patient was transfused on a regular basis. Among these patients, HFE genotyping was available for 45 patients: 14 carried p.His63Asp heterozygous mutations, three were heterozygous for the p.Cys282Tyr mutation and one was a composite heterozygous p.Cys282Tyr/ p.His63Asp. We did not find any difference in ferritin levels between wildtype and mutated patients (742±549 ng/mL vs. 874±534 ng/mL, respectively; P=NS). Diabetes mellitus was noted in three cases; one patient developed a hepatocarcinoma and underwent liver transplantation; hypogonadism was recorded in one case, hypothyroidism in one case and osteoporosis in three cases. The percentage of patients treated for iron overload increased with age (25% before the age of 20 years, 41% between 20 and 40 years, and 73% after the age of 40 years) (Figure 3C). Treatments included phlebotomy (n=15), deferasirox (n=9), deferoxamine (n=3) and deferiprone (n=1). Mean liver iron content, evaluated using magnetic resonance imaging, was 200±103 μmol/g at diagnosis (n=20) vs. 88±42 (n=14) at the last follow-up (P<0.001), showing that iron chelation was efficient in decreasing iron content in the liver (Figure 3D). There was a strong difference in liver iron content between patients with a ferritin level above 1000 ng/mL (n=7) and under 1000 ng/mL (n=25): 318±31 μmol/g vs. 113±68 μmol/g, respectively (Figure 3E). However, focusing on patients with ferritin levels below 1000 ng/mL, the correlation was poor (Figure 3F). Therefore, even patients with a moderate ferritin increase should undergo an evaluation of tissue iron content at diagnosis.

Pseudohyperkalemia Among the 35 patients for whom data on potassium concentration at the same time as ektacytometry were available, the potassium level was above the upper range in 18 cases (51%). However, no specific potassium release test at room temperature was performed.

Discussion Because of its heterogeneous presentation, HX is an underestimated condition, as suggested recently.25 The aim of our report is to describe precisely the mean phenotypic and genotypic HX features, in order to facilitate the diagnosis. Our first conclusion is that most HX subjects have a mild hematologic phenotype, with two-thirds of patients having a fully compensated hemolysis; non-erythroid features, including iron overload, perinatal edema and thrombotic events after splenectomy often deter1561

V. Picard et al.

mined the disease severity. Considering this heterogeneity, we investigated whether genetics could discriminate patients who would require a more intensive management. We identified 49 families with PIEZO1-HX, and six with KCNN4-HX, this ratio being in agreement with those in other reports.5 One recurrent mutation (p.Arg352His) accounted for most cases of KCNN4-HX (5/6 families). We also identified a new 18-base pair KCNN4 deletion, while only missense mutations have been reported so far, indicating a new genetic basis for this disorder. In the Gardos channelopathies reported so far, the single residue substitution increased the activity of the Gardos channel through an altered calmodulin binding.15,26 In this family, we observed the 18-base pair deletion overlapped the exon-intron 7 junction. We hypothesized that the use of a potential upstream splice site at c.G1104(TG/GT) might lead to an in-frame deletion of five residues (p.369_373VDISKdel), creating a Gardos channel missing five amino acids in the calmodulin-binding domain. The consequences on Gardos channel sequence and function remain however unclear and need to be tested. In contrast to Gardos channelopathy, recurrent mutations accounted for only one-third of PIEZO1-HX cases. The three recurrent mutations are located in exon 51, the main mutation hotspot that encodes the canal pore and the C-terminal region. Regions including exons 14 to 18 and exons 42 to 45 appeared as secondary ‘hotspots’, the former region affects peripheral helices forming the extracellular ‘blade’, the latter region encodes distinct α helices close to the canal pore. Indeed, we identified a new recurrent mutation, p.Asp669Tyr in exon 16, present in two unrelated families and recently reported in one other patient.17 In other families, one or two rare or undescribed private sequence variations were present, indicating very heterogeneous genetic backgrounds in PIEZO1-HX. Indeed, a PIEZO1 sequence variation, not scored as a polymorphism, was identified in all tested families except one. For the large majority of patients, the diagnosis was made based on phenotypic data before the genetic test was available. So far, no clear genotype-phenotype correlation could be drawn. Given the high number of PIEZO1 polymorphisms described in databases, the effect of these private, newly described mutations cannot be ascertained. From a practical point of view, these data underline the requirement for phenotype-based methods and functional experiments in addition to genetics to confirm PIEZO1-HX diagnosis, when unreported mutations are identified. If osmotic gradient ektacytometry can be seen as an indirect phenotypic test reflecting red cell dehydration, functional electrophysiological tests would represent a major advance in the characterization of new mutations. Some of the mutations involved here have been tested functionally, but not in a systematic manner because of the lack of available tools.10,13,26 The recent characterization of a PIEZO1 gain-offunction mutation through high-throughput patch clamping on red cells is promising in this respect.27 Recent in vitro studies showed that both PIEZO1-HX and KCNN4-HX share a common pathophysiology leading to red cell dehydration.13,26 Indeed, both disorders share hemolytic features and frequent hyperferritinemia. Hyperferritinemia was not related to transfusions and was frequently at the front line of the diagnosis. Hyperferritinemia is well described in chronic hemolytic diseases,28 but is notably much more frequent in HX than 1562

in hereditary spherocytosis.29 Although associated HFE mutations may worsen iron overload in HX,7 they could not be associated with more severe iron overload in our series. Alternative mechanisms of increased iron uptake may be involved, including chronic hypoxia, increased erythroferrone secretion, and erythroblast proliferation possibly associated with some inefficient erythropoiesis. Alternatively, expression of a mutated PIEZO1 or Gardos protein at the cell surface could directly deregulate hepcidin expression in liver cells or drive iron entry through the gut. From a practical point of view, we observed a weak correlation between ferritin and liver iron content, particularly for ferritin levels below 1000 ng/mL. In terms of clinical management, these data highlight the requirement for an annual iron status evaluation and for measurement of liver iron content by magnetic resonance imaging when ferritin increases above normal values. On the other hand, PIEZO1-HX and KCNN4-HX differed in several ways. First, in terms of severity, patients with PIEZO1-HX had a milder hematologic phenotype: 27% had a hemoglobin level below 120 g/L vs. 75% of patients with KCNN4-HX. It is worth noting that some patients with PIEZO1-HX had a hemoglobin level in the upper normal range or above it. Therefore, PIEZO1 gainof-function mutations may stimulate erythropoiesis by itself, explaining the ‘compensated hemolysis’ phenotype as a balance between hemolysis and increased erythropoiesis. Of note, red cell dehydration was not the main cause of hemolysis since it was predominant in PIEZO1HX, but discreet or absent in KCNN4-HX despite a more severe anemia. This difference has a practical consequence: ektacytometry, which responds to red cell hydration, identified PIEZO1-HX but not KCNN4-HX. Therefore, genetic testing should be performed to rule out this subset of HX in the case of undiagnosed hemolysis, even when ektacytometry is normal. It has been recently suggested to use the term "Gardos channelopathy”, instead of xerocytosis,30 for this variant and we agree with this proposal. Another difference between PIEZO-HX and Gardos channelopathy was the rate of post-splenectomy thrombosis. Thombosis occurred in 100% of PIEZO1-HX cases (8/8), in agreement with other reports,8,24,30–34 but in 0/4 Gardos-mutated patients. However, these data must be interpreted carefully because: (i) the number of Gardossplenectomized patients was low; (ii) four patients with thrombotic complications were not genotyped; and (iii) the persistence in all cases of hemolysis after splenectomy represents a risk by itself. Indeed, thrombosis is a welldescribed complication after splenectomy, particularly when removal of the spleen does not abrogate hemolysis. Several factors may be involved, including platelet aggregation, decrease in nitric oxide level, high level of circulating microparticles, high rate of phosphatidylserineexpressing red cells and increased reticulocyte adherence.35–37 However, PIEZO1 seems to have a specific role that may involve endothelial dysfunction.38,39 Thus, a mutated PIEZO1 at the cell surface could alter interactions between endothelial cells and dehydrated erythrocytes and favor thrombosis. Considering the type of thrombosis, we made two interesting findings. There was a high rate of portal thrombosis (40% of patients), which may even be underestimated since asymptomatic portal thrombosis was not systematically evaluated after splenectomy as recently suggested.40 Secondly, we haematologica | 2019; 104(8)

A retrospective series of 126 HX cases

observed severe thrombotic complications, including pulmonary hypertension in two patients carrying heterozygous globin gene mutations, one with A/S trait and the other with β-thalassemia trait + α-triplication, suggesting a synergistic deleterious effect of these conditions. Gene analysis was not available for the former, and the latter had two PIEZO1 missense sequence variations (p.Gln1519Pro and p.Glu1910Lys, scored as unknown significance and as a polymorphism, respectively), the role of these associated mutations on PIEZO1 function remains to be established. Finally, perinatal edema was observed in PIEZO1-HX and not in KCNN4-mutated patients in this series. However, the low number of KCNN4-mutated families does not allow definitive conclusion to be drawn on this issue and we recommend that pregnancies should be monitored closely in both genotypes. The severity of perinatal edema was heterogeneous, ranging from nuchal clarity to fetal loss or death after birth. This highlights the requirement for a careful pregnancy follow-up with at least monthly ultrasonography monitoring when one parent is affected. Whether severe perinatal edema is restricted to a subset of PIEZO1 mutations is unclear, but we observed recurrent perinatal edema in all informative families. Perinatal edema apparently did not involve the effect of PIEZO1 on red cells, since anemia was not present in most cases. The occurrence of chylous ascites and hygroma cysts evoked a primary defect in the lymphatic system.6,18,41,42 PIEZO1 is expressed in human embryo lymphatic vessels11 and its bi-allelic loss of function is associated with congenital lymphatic dysplasia.23,43 It is puzzling that gain-of-function mutations induced a similar phenotype in the perinatal period. Morpholino PIEZO1-knockdown in zebrafish induced anemia;31 hemolysis with stomatocytes on the blood smear was seen in some patients with congenital lymphatic dysplasia.23 Moreover, one patient in our series carried a mutation associated with congenital lymphatic dysplasia, while another patient with a history of perinatal edema had persistent lymphedema in adulthood. Edema and hemolysis may

References 1. Gallagher PG. Disorders of erythrocyte hydration. Blood. 2017;130(25):2699–2708. 2. Caulier A, Rapetti-Mauss R, Guizouarn H, Picard V, Garçon L, Badens C. Primary red cell hydration disorders: pathogenesis and diagnosis. Int J Lab Hematol. 2018;40 Suppl 1:68–73. 3. Delaunay J. The hereditary stomatocytoses: genetic disorders of the red cell membrane permeability to monovalent cations. Semin Hematol. 2004;41(2):165–172. 4. Andolfo I, Russo R, Gambale A, Iolascon A. New insights on hereditary erythrocyte membrane defects. Haematologica. 2016;101(11):1284–1294. 5. Andolfo I, Russo R, Gambale A, Iolascon A. Hereditary stomatocytosis: an underdiagnosed condition. Am J Hematol. 2018;93(1): 107–121. 6. Grootenboer S, Barro C, Cynober T, et al. Dehydrated hereditary stomatocytosis: a cause of prenatal ascites. Prenat Diagn. 2001;21(13):1114–1118. 7. Syfuss P-Y, Ciupea A, Brahimi S, et al. Mild dehydrated hereditary stomatocytosis revealed by marked hepatosiderosis. Clin

haematologica | 2019; 104(8)

represent two features in the spectrum of PIEZO1-related diseases which may sometimes overlap. Again, functional studies are needed to decipher the pathophysiology of both entities. In summary, we report here the clinical, biological and genetic features of the largest series of HX patients to date. We have uncovered the heterogeneous genetic bases of PIEZO1-HX reporting 12 novel mutations, and its varied clinical expression, characterized by a normal to high hemoglobin level, iron overload, occurrence of perinatal edema, and the high thrombosis rate after splenectomy. Gardos channelopathy was characterized by more severe hemolysis, and less erythrocyte dehydration. Taken together, our results provide a better picture of this disorder and will help the diagnosis and clinical management of patients, particularly in terms of contraindications of splenectomy, iron overload and pregnancy follow-up. Acknowledgments We particularly want to thank all the families, the patients and the French Registry of Hereditary Stomatocytosis. We thank all technicians and biologists of the Hematology Laboratory of Hôpital Bicêtre for their contribution and Mme Hélène Ponsin, for excellent resource management. We also thank the ARFH (Association Recherche et Formation en Hématopathologie) for its constant support. Finally, we thank all investigators and contributors: Dr. Madeleine Fénéant-Thibault, Dr. Gérard Tertian, Pr. Gilles Tchernia, Dr. Philippe Agape, Dr. Nathalie Aladjidi, Dr. Yannick Bacq, Dr. Fiorenza Barraco, Dr. Sophie Bayart, Dr. Jean-Sébastien Blade, Pr. Photis Beris, Dr. Jean-Yves Boucher, Dr. Marie Pierre Castex, Dr. Denis Cléau, Dr. Luc Darnige, Dr. Xavier Delbrel, Dr. Valérie Mathieu, Dr.Vincent Di Martino, Dr. Valérie Doireau, Dr. Bernard Drenou, Pr. Cécile Goujard, Dr. Isabelle Grulois, Pr. Xavier Jeunemaitre, Dr. MarieFrançoise Le Coz, Dr. Nicolas Limal, Dr. Jean-Philippe Miguet, Dr. Laetitia Morel, Dr. Philippe Nicoud, Dr. Emmanuel Plouvier, Pr. Jacques Pouchot, Dr. Michel Renoux, Dr. Anne-Laure Suc, and Dr. Hacene Zerazhi. We thank the RFH association for its support.

Lab Haematol. 2006;28(4):270–274. 8. Stewart GW, Amess JA, Eber SW, et al. Thrombo-embolic disease after splenectomy for hereditary stomatocytosis. Br J Haematol. 1996;93(2):303–310. 9. Delaunay J. The molecular basis of hereditary red cell membrane disorders. Blood Rev. 2007;21(1):1–20. 10. Albuisson J, Murthy SE, Bandell M, et al. Dehydrated hereditary stomatocytosis linked to gain-of-function mutations in mechanically activated PIEZO1 ion channels. Nat Commun. 2013;4:1884. 11. Andolfo I, Alper SL, De Franceschi L, et al. Multiple clinical forms of dehydrated hereditary stomatocytosis arise from mutations in PIEZO1. Blood. 2013;121(19):3925-3935. 12. Zarychanski R, Schulz VP, Houston BL, et al. Mutations in the mechanotransduction protein PIEZO1 are associated with hereditary xerocytosis. Blood. 2012;120(9):1908-1915. 13. Rapetti-Mauss R, Picard V, Guitton C, et al. Red blood cell Gardos channel (KCNN4): the essential determinant of erythrocyte dehydration in hereditary xerocytosis. Haematologica. 2017;102(10):e415–e418. 14. Ma S, Cahalan S, LaMonte G, et al. Common PIEZO1 allele in african populations causes RBC dehydration and attenu-

15.

16.

17.

18.

19.

20.

ates plasmodium infection. Cell. 2018;173 (2):443-455.e12. Rapetti-Mauss R, Lacoste C, Picard V, et al. A mutation in the Gardos channel is associated with hereditary xerocytosis. Blood. 2015;126(11):1273–1280. Glogowska E, Lezon-Geyda K, Maksimova Y, Schulz VP, Gallagher PG. Mutations in the Gardos channel (KCNN4) are associated with hereditary xerocytosis. Blood. 2015;126(11):1281–1284. Andolfo I, Russo R, Rosato BE, et al. Genotype-phenotype correlation and risk stratification in a cohort of 123 hereditary stomatocytosis patients. Am J Hematol. 2018;93(12):1509-1517. Grootenboer S, Schischmanoff PO, Laurendeau I, et al. Pleiotropic syndrome of dehydrated hereditary stomatocytosis, pseudohyperkalemia, and perinatal edema maps to 16q23-q24. Blood. 2000;96(7):2599– 2605. Beneteau C, Thierry G, Blesson S, et al. Recurrent mutation in the PIEZO1 gene in two families of hereditary xerocytosis with fetal hydrops. Clin Genet. 2014;85(3):293– 295. Cynober T, Mohandas N, Tchernia G. Red cell abnormalities in hereditary spherocyto-

1563

V. Picard et al.

21.

22.

23.

24.

25.

26.

27.

1564

sis: relevance to diagnosis and understanding of the variable expression of clinical severity. J Lab Clin Med. 1996;128(3):259–269. Llaudet-Planas E, Vives-Corrons JL, Rizzuto V, et al. Osmotic gradient ektacytometry: a valuable screening test for hereditary spherocytosis and other red blood cell membrane disorders. Int J Lab Hematol. 2018;40(1):94– 102. Lazarova E, Gulbis B, Oirschot B van, van Wijk R. Next-generation osmotic gradient ektacytometry for the diagnosis of hereditary spherocytosis: interlaboratory method validation and experience. Clin Chem Lab Med. 2017;55(3):394-402. Fotiou E, Martin-Almedina S, Simpson MA, et al. Novel mutations in PIEZO1 cause an autosomal recessive generalized lymphatic dysplasia with non-immune hydrops fetalis. Nat Commun. 2015;6:8085. Jaïs X, Till SJ, Cynober T, et al. An extreme consequence of splenectomy in dehydrated hereditary stomatocytosis: gradual thrombo-embolic pulmonary hypertension and lung-heart transplantation. Hemoglobin. 2003;27(3):139–147. Kaufman HW, Niles JK, Gallagher DR, et al. Revised prevalence estimate of possible hereditary xerocytosis as derived from a large U.S. laboratory database. Am J Hematol. 2018;93(1):E9–E12. Rapetti-Mauss R, Soriani O, Vinti H, Badens C, Guizouarn H. Senicapoc: a potent candidate for the treatment of a subset of hereditary xerocytosis caused by mutations in the Gardos channel. Haematologica. 2016;101 (11):e431–e435. Rotordam MG, Fermo E, Becker N, et al. A

28.

29.

30.

31.

32.

33.

34.

novel gain-of-function mutation of Piezo1 is functionally affirmed in red blood cells by high-throughput patch clamp. Haematologica. 2019;104(5):e179-e183. Barcellini W, Fattizzo B. Clinical applications of hemolytic markers in the differential diagnosis and management of hemolytic anemia. Dis Markers. 2015;2015:635670. Mariani M, Barcellini W, Vercellati C, et al. Clinical and hematologic features of 300 patients affected by hereditary spherocytosis grouped according to the type of the membrane protein defect. Haematologica. 2008;93(9):1310–1317. Fermo E, Bogdanova A, Petkova-Kirova P, et al. “Gardos Channelopathy”: a variant of hereditary Stomatocytosis with complex molecular regulation. Sci Rep. 2017;7(1): 1744. Perel Y, Dhermy D, Carrere A, et al. Portal vein thrombosis after splenectomy for hereditary stomatocytosis in childhood. Eur J Pediatr. 1999;158(8):628–630. Bergheim J, Ernst P, Brinch L, Gore DM, Chetty MC, Stewart GW. Allogeneic bone marrow transplantation for severe postsplenectomy thrombophilic state in leaky red cell membrane haemolytic anaemia of the stomatocytosis class. Br J Haematol. 2003;121(1):119–122. Murali B, Drain A, Seller D, Dunning J, Vuylsteke A. Pulmonary thromboendarterectomy in a case of hereditary stomatocytosis. Br J Anaesth. 2003;91(5):739–741. Carli P, Graffin B, Gisserot O, Landais C, De Jaureguiberry J-P. [Recurrence of thromboembolic disease after splenectomy for hereditary xerocytosis]. Rev Med Interne.

2007;28(12):879–881. 35. Gallagher PG, Chang SH, Rettig MP, et al. Altered erythrocyte endothelial adherence and membrane phospholipid asymmetry in hereditary hydrocytosis. Blood. 2003;101 (11):4625–4627. 36. Smith BD, Segel GB. Abnormal erythrocyte endothelial adherence in hereditary stomatocytosis. Blood. 1997;89(9):3451–3456. 37. Cappellini MD, Grespi E, Cassinerio E, Bignamini D, Fiorelli G. Coagulation and splenectomy: an overview. Ann N Y Acad Sci. 2005;1054:317–324. 38. Li J, Hou B, Tumova S, et al. PIEZO1 integration of vascular architecture with physiological force. Nature. 2014;515(7526):279–282. 39. Wang S, Chennupati R, Kaur H, Iring A, Wettschureck N, Offermanns S. Endothelial cation channel PIEZO1 controls blood pressure by mediating flow-induced ATP release. J Clin Invest. 2016;126(12):4527–4536. 40. Iolascon A, Andolfo I, Barcellini W, et al. Recommendations regarding splenectomy in hereditary hemolytic anemias. Haematologica 2017;102(8):1304–1313. 41. Entezami M, Becker R, Menssen HD, Marcinkowski M, Versmold HT. Xerocytosis with concomitant intrauterine ascites: first description and therapeutic approach. Blood. 1996;87(12):5392–5393. 42. Ami O, Picone O, Garçon L, et al. Firsttrimester nuchal abnormalities secondary to dehydrated hereditary stomatocytosis. Prenat Diagn. 2009;29(11):1071–1074. 43. Lukacs V, Mathur J, Mao R, et al. Impaired PIEZO1 function in patients with a novel autosomal recessive congenital lymphatic dysplasia. Nat Commun. 2015;6:8329.

haematologica | 2019; 104(8)

ARTICLE

Myelodysplastic Syndromes

A phase II study of guadecitabine in higher-risk myelodysplastic syndrome and low blast count acute myeloid leukemia after azacitidine failure Marie Sébert,1,2 Aline Renneville,3 Cécile Bally,2 Pierre Peterlin,1,4 Odile Beyne-Rauzy,1,5 Laurence Legros,1,6 Marie-Pierre Gourin,1,7 Laurence Sanhes,1,8 Eric Wattel, 1,9 Emmanuel Gyan,1,10 Sophie Park,1,11 Aspasia Stamatoullas,1,12 Anne Banos,1,13 Kamel Laribi,1,14 Simone Jueliger,15 Luke Bevan,15 Fatiha Chermat,1 Rosa Sapena,1 Olivier Nibourel,3 Cendrine Chaffaut,16 Sylvie Chevret,16 Claude Preudhomme,3 Lionel Adès,1,2 and Pierre Fenaux,1,2 on behalf of the Groupe Francophone des Myélodysplasies (GFM)

Ferrata Storti Foundation

Haematologica 2019 Volume 104(8):1565-1571

Groupe Francophone des Myélodysplasies, Paris, France; 2Hématologie Clinique, Hôpital Saint Louis, Paris, France; 3Laboratoire d’Hématologie, CHU de Lille, France; 4CHU de Nantes, France; 5IUCT ONCOPOLE Toulouse, France; 6CHU de Nice, France; 7CHRU de Limoges, France; 8CHU de Perpignan, France; 9CHU Lyon Sud, Lyon, France; 10CHRU de Tours, France; 11CHU de Grenoble, France; 12Centre Henri Becquerel, Rouen, France; 13CH de la Côte Basque, France; 14CHU Côte de Nacre, Caen, France; 15Astex Pharmaceuticals Inc., Cambridge, UK; 16Service de Biostatistiques, Hôpital Saint-Louis, APHP, Paris, France 1

ABSTRACT

igh-risk myelodysplastic syndrome/acute myeloid leukemia patients have a very poor survival after azacitidine failure. Guadecitabine (SGI-110) is a novel subcutaneous hypomethylating agent which results in extended decitabine exposure. This multicenter phase II study evaluated the efficacy and safety of guadecitabine in highrisk myelodysplastic syndrome and low blast count acute myeloid leukemia patients refractory or relapsing after azacitidine. We included 56 patients with a median age of 75 years [Interquartile Range (IQR) 69-76]. Fifty-five patients received at least one cycle of guadecitabine (60 mg/m2/d subcutaneously days 1-5 per 28-day treatment cycles), with a median of 3 cycles (range, 0-27). Eight (14.3%) patients responded, including two complete responses; median response duration was 11.5 months. Having no or few identified somatic mutations was the only factor predicting response (P=0.035). None of the 11 patients with TP53 mutation responded. Median overall survival was 7.1 months, and 17.9 months in responders (3 of whom had overall survival >2 years). In multivariate analysis, IPSS-R (revised International Prognostic Scoring System) score other than very high (P=0.03) primary versus secondary azacitidine failure (P=0.01) and a high rate of demethylation in blood during the first cycle of treatment (P=0.03) were associated with longer survival. Thus, guadecitabine can be effective, sometimes yielding relatively prolonged survival, in a small proportion of high-risk myelodysplastic syndrome/low blast count acute myeloid leukemia patients who failed azacitidine. (Trial registered at clinicaltrials.gov identifier: 02197676)

Introduction The first generation hypomethylating agents (HMA) azacitidine (AZA) or decitabine are considered to be the reference treatment for high-risk myelodysplastic syndromes (MDS) and low blast count acute myeloid leukemia (AML) (<30% marrow blasts) in elderly patients who are not candidates for allogeneic stem cell transplantation (allo-SCT). However, responses are seen in only 50-60% of patients, haematologica | 2019; 104(8)

Correspondence: PIERRE FENAUX pierre.fenaux@aphp.fr Received: October 5, 2018. Accepted: February 7, 2019. Pre-published: February 7, 2019. doi:10.3324/haematol.2018.207118 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1565 ©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.

1565

M. Sebert et al.

and the median overall survival (OS) of 18-24 months obtained with azacitidine remains modest.1 Moreover, median survival after HMA failure is only approximately six months.2 The hypomethylating activity of AZA and decitabine depends on their incorporation during the S phase of the cell cycle into RNA or DNA, respectively.3 This suggests a relationship between duration of drug exposure and effectiveness of the HMA. Half-life of first generation HMA is approximately 30 minutes, which might limit their activity in slowly dividing MDS cells.4 Furthermore, a recent study found a high response rate of 67% in unfavorable risk MDS/AML (specifically TP53 mutated patients) after serial 10-day cycles of decitabine, a regimen with a longer exposure to decitabine than the classical 5-day schedule.5 Guadecitabine (SGI-110) is a novel, second-generation hypomethylating drug. It is a dinucleotide of decitabine (the active metabolite) and deoxyguanosine, resistant to cytidine deaminase, the main enzyme responsible for decitabine degradation. Cleavage of the phosphodiester bond between the two parts of the dinucleotide results in a slow release of decitabine, prolonging the HMA exposure in cells.6 A phase I/II study found a 52% (55 of 107) response rate to guadecitabine in treatment naïve AML with tolerable toxicity.7 Guadecitabine was also studied in 19 patients with relapsed or refractory MDS after HMA, with a 32% response rate.8 These results prompted the GFM group to propose guadecitabine as a salvage treatment in a larger series of higher-risk MDS and low blast count AML patients after AZA failure, not candidates for intensive chemotherapy or allo-SCT.

Non-responders were eligible only in the absence of overt progression, i.e. at least doubling of marrow blast percentage between start of HMA and protocol screening. Patients eligible for intensive chemotherapy or allo-SCT were excluded. Other eligibility criteria included an Eastern Co-operative Oncology Group (ECOG) performance status of 0-2, and adequate liver and hepato-renal functions (creatinine <1.5 times the upper limit of normal, and creatinine clearance ≥+50 mL/min, total bilirubin and transaminase <1.5 times the upper normal limit). The protocol was approved by the Comité de Protection des Personnes Paris-Ile de France, the ethical committee whose approval is valid for all participating French institutions. All patients provided written informed consent.

Treatment Patients received 60 mg/m2 subcutaneous guadecitabine on days 1-5 of 28-day treatment cycles (the recommended drug regimen in previous studies).7,8 Treatment was continued until progression, death, unacceptable toxicity, or no response after six cycles (extended to 9 cycles after the first 20 patients). Dose reductions to 45 and even 30 mg/m2/d were allowed to manage toxicity.

Biological studies Somatic mutations were screened on bone marrow cells by a next-generation sequencing (NGS) assay for a selected panel of 36 genes (Online Supplementary Appendix and Online Supplementary Table S1) at inclusion for all patients and on sequential bone marrow samples in some responders. Global DNA methylation was measured in 53 patients by the long interspersed nuclear element (LINE1) methylation assay, and changes in methylation from baseline were assessed as described in the Online Supplementary Appendix.

End points Methods Trial design This was a national, GFM-sponsored multicenter phase II clinical trial (clinicaltrials.gov identifier: 02197676) evaluating the efficacy of guadecitabine in higher-risk MDS and low-blast count AML patients, refractory or relapsing after AZA treatment. A first cohort of 21 patients was planned with the objective to stop the study if four patients or less would respond after six cycles of guadecitabine or experience a high toxicity. After review by an independent Data Safety Monitoring Board (DSMB), toxicity was considered acceptable and five patients were responders. The extended cohort included 36 patients. Because late responders had been reported in previous studies (Issa et al., 2019, personal communication), response was also evaluated after nine cycles of guadecitabine in the extended cohort (Online Supplementary Figure S1).

Patients Inclusion criteria were: i) age >18 years; i) diagnosis of MDS or chronic myelomonocytic leukemia (CMML), with white blood cell (WBC) count <13x109/L according to the World Health Organization (WHO) 2008 criteria9 with international prognostic scoring system (IPSS) intermediate-2 or high-risk MDS10 or AML with 20-30% marrow basts [AML/refractory anemia with excess blasts in transformation (RAEB-t) according to the FrenchAmerican-British (FAB) classification11]; iii) refractoriness to azacitidine, i.e. at least six cycles without response [complete response (CR), partial response (PR), marrow CR or stable disease with hematologic improvement (HI), according to International Working Group (IWG) 2006 criteria]12 or relapse after a response. 1566

The primary end point was hematologic response (CR, PR, marrow CR or stable disease with HI according to IWG 2006 criteria)12 Secondary end points were duration of response, rate of progression to AML, overall survival, toxicity profile of guadecitabine. All patients who achieved a CR, PR, marrow CR or HI after 3, 6 or 9 (for the 36 patients of the extended cohort) cycles of guadecitabine were considered responders and were allowed to continue treatment until loss of response, progression, or death.

Statistical analysis Based on the results of phase III trials of decitabine for the treatment of MDS13 and AML,14 it was expected that guadecitabine would achieve at least 30% hematologic responses. A Bryant and Day 2-stage phase II design was used. Assuming a 20% response rate (CR+PR+ marrow CR + HI according to IWG 2006 criteria) under the null, controlling for type I and II error rates at α=0.05 and β=0.2 respectively, 19 patients had to be accrued for stage 1 of the trial to demonstrate a benefit of 20% (i.e. a response rate of >40%). At the end of Stage 1, the study would stop if there were four responders or less. If there were at least five responders, 35 additional patients had to be included in the study. We assumed that approximately 10% of the population may not be evaluable for response and that 56 patients should, therefore, be included. The response rate was estimated in the intention-to-treat sample, with 95% exact confidence interval. (95%CI:) The Kaplan and Meier method was used for analysis of progression-free survival and overall survival. Median and 25% and 75% quartiles, were estimated. Analysis was performed on SAS (SAS Cary, NC, USA) and R v.3.3.2 (https://www.R-project.org/) softwares. Two-sided P<0.05 was considered statistically significant. haematologica | 2019; 104(8)

Guadecitabine in selected MDS and AML after azacitidine failure

Results Patients’ baseline characteristics Between August 2014 and January 2016, we enrolled 56 patients in 13 French centers; one patient died from infection before receiving guadecitabine (Table 1 and Figure 1). Of the 56 patients (intent-to-treat population), 66% were males, with a median age of 75 years (range, 70-79). At inclusion, according to WHO classification, 11 (20%) patients had RAEB-1, 31 (55%) had RAEB-2, and 11 (20%) had AML. Thirty-four (61%) patients had very high-risk IPSS-R and 43 (77%) patients were red blood cell transfusion-dependent (TD), while ECOG status was >1 in five (9%) patients. The median number of prior AZA cycles was 13 (range 6-23). Forty-one (73%) patients had relapsed after a response to AZA, while 15 (27%) had primary resistance. The median IQR time interval between the HMA failure and study initiation was 50 days (33; 76) (range, 0-251 days). Forty-nine (87.5%) of the 56 patients had at least one somatic mutation, with a median number of two mutations (range 0-7), the most common being ASXL1 (n=14, 25%), RUNX1 (n=12, 21%), TP53 (n=11, 20%), U2AF1 Table 1. Patients’ baseline characteristics.

N=56 (%) Age (years) Sex Male Female ECOG performance status 0 1 2 WHO MDS-MLD CMML MDS-EB-1 MDS-EB-2 AML IPSS Int-1 Int-2 High NA IPSS-R Low/Int High Very High NA AZA first response Primary resistance (≥ 6 cycles) Relapse after response Median n. of AZA cycles Transfusion dependence

75 [70-79]

Treatment outcomes Fifty-five patients received at least one cycle of guadecitabine. The median number of treatment cycles received was three (range, 0-27), eight patients having received only one cycle. Most patients received the planned dose in all treatment cycles, but 18 patients had a dose reduction. Eight of 56 (14.3%) patients responded, including two CR, one PR, three hematologic improvements, and two marrow CR (Table 2). Seven patients had responded by three cycles and one additional patient by six cycles, but we observed no later response (after 9 cycles) in this study. The median duration of response was 11.5 months (95%CI: 9; not available) (Figure 3B), and response was longer than 12 months in four patients (13, 19, 21, 31 months, respectively). Median overall survival was 7.1 months [95%CI: (5.6;11.8)] with a one-year survival of 33% (95%CI: 22.9;48.4) (Figure 4A). Responders to guadecitabine had a median OS of 17.9 months, and three patients had >2-year OS (24, 34, and 42 months, respectively). Forty-nine patients had died, because of progressing disease in 28 (57.1%), infection in 13 (26.5%), bleeding in two, heart

37 (66%) 19 (34%) 20 (36%) 31 (55%) 5 (9%) 2 (4%) 1 (2%) 11 (20%) 11 (20%) 4 (7%) 27 (48%) 23 (41%) 2 (4%) 4 (7%) 13 (23%) 34 (61%) 5 (9%) 15 (27%) 41 (73%) 13 [9-23] 43 (77%)

Data are median [range] or number (n) / (%); ECOG: Eastern Cooperative Oncology Group; WHO: World Health Organization; RCMD: refractory cytopenia with multilineage dysplasia; CMML: chronic myelomonocytic leukemia; RAEB: refractory anemia with excess blasts; RAEB-t: RAEB in transformation; AML: acute myeloid leukemia; IPSS: International Prognostic Scoring System; Int: intermediate; NA: not available; IPSSR: revised IPSS; AZA: azacitidine.

haematologica | 2019; 104(8)

(n=11, 20%), and DNMT3A (n=11, 20%) (Figure 2A). Median variant allele frequency (VAF) of those mutations was 29.5% (Online Supplementary Figure S2). Baseline methylation levels of LINE-1 were similar in blood and bone marrow samples with an average of 73% and 71%, respectively.

Figure 1. Flow chart of the study.

1567

M. Sebert et al.

failure in one, and from unknown cause in five. None of the patients had received allogeneic SCT.

Prognostic factors of response and overall survival Response was seen in four (26%) of the 15 patients with primary failure, and four (9%) of the 41 relapsing patients (P=0.12). The median number of mutations was one [range, 0-3 in responders, compared with 2 (range, 06) mutations in non-responders (P=0.035)], and the response rate was significantly higher in patients with no

detectable somatic mutations compared to patients with at least one somatic mutation (P=0.036). None of the 11 patients with TP53 mutation achieved response. Clonal architecture was followed during treatment in five responders with mutations at baseline. There was no significant decrease in VAF of the mutated clone(s) at hematologic response (Figure 2B). Treatment with guadecitabine resulted in a maximum LINE-1 demethylation relative to baseline (D0) of 12.3% on day 8 of cycle 1 in peripheral blood samples and of 3.3% on day 28 of

Figure 2. Molecular characteristics of the patients after azacitidine (AZA) failure and during guadecitabine treatment. (A) Spectrum of mutations in the 56 high-risk myelodypslastic syndrome (MDS) patients included (refractory to or relapsing after AZA therapy) in 36 selected genes. Each column represents an individual patient sample, and each colored cell represents a mutation of the gene or gene group listed to the left of that row. The number of mutations in each row is indicated in the column on the right. Darker cells of patient numbers indicate responders to guadecitabine. (B) Evolution of different clones, according to variant allele frequency (VAF), in five patients responding to guadecitabine after AZA failure.

1568

haematologica | 2019; 104(8)

Guadecitabine in selected MDS and AML after azacitidine failure

Table 2. Baseline characteristics and outcome of the responders.

Patient n.

Sex

WHO

#2 #4 #9 #10 #21 #24 #29 #53

F M F M M M M M

AML RAEB-1 RAEB-1 AML RAEB-2 RAEB-2 RCMD RAEB-2

Karyotype

Somatic mutations

+8, del20q No Del1p,del5q, del11q No +1, +21 PHF6, RUNX1 Normal No -Y, +8 SETBP1, RIT1 Normal BCOR, STAG2 Del20q, +21 RUNX1, U2AF1, ASXL1 Normal EZH2, SETBP1

AZA first response

Response

Response duration (months)

Survival (months)

Relapse Relapse Primary failure Relapse Relapse Primary failure Relapse Primary failure

CR Marrow CR Marrow CR HI + Marrow CR PR CR HI HI + Marrow CR

13* 19* 21* 10 9 31* 7 8

42** 19 24** 10 14 34** 10 12

*Relatively long-term responders. **Relatively long-term survivors. F: female; M: male; WHO: World Health Organization; AZA: azacitidine; RCMD: refractory cytopenia with multilineage dysplasia; CMML: chronic myelomonocytic leukemia; RAEB: refractory anemia with excess blasts; RAEB-t: RAEB in transformation; AML: acute myeloid leukemia.

Table 3. Grade III-IV non-hematologic toxicities during first nine cycles of guadecitabine treatment.

Patients N=23 Pulmonary Cardiovascular Musculary/denutrition Transaminase Renal failure Gastro-intestinal Neurological Uro-genital Endocrinological

N (%) 7 (12.5%) 4 (7.1%) 4 (7.1%) 4 (7.1%) 3 (5.3%) 3 (5.3%) 2 (3.6%) 2 (3.6%) 2 (3.6%)

N: number.

cycle 1 in bone marrow samples (Online Supplementary Figure S3). Except for somatic mutations, no other baseline parameter had significant prognostic value for response, including age, sex, ECOG status, transfusion dependency, baseline hemoglobin, platelet, absolute neutrophil count, bone marrow blast percentage, cytogenetics, IPSS, IPSS-R, type of AZA failure (primary or secondary), LINE1 baseline methylation, or demethylation rate with treatment. Overall survival was significantly shorter in patients with high IPSS (HR=1.81, 95%CI: 1.1-2.97; P=0.02), with very high IPSS-R (HR=1.5, 95%CI: 1.2-1.87; P=0.0004), and TP53 mutation (HR=2.23, 95%CI: 1.09-4.57; P=0.028), and longer in patients with high demethylation rate in blood on day 8 of the first cycle (P=0.02) (Online Supplementary Figure S4). There was a trend towards shorter survival in patients with a higher number of somatic mutations (HR=1.18, 95%CI: 0.97-1.44; P=0.099) and prolonged OS in patients with primary AZA failure (HR=0.51, 95%CI: 0.25-1.01; P=0.054), and low baseline level of methylation in blood on day 8 of the first cycle (P=0.066) and in bone marrow on day 28 of the first cycle (P=0.083). In multivariate analysis, IPSS-R (P=0.03), demethylation rate in blood (P=0.03) and the type of AZA failure (primary vs. secondary; P=0.01) remained predictive of OS. Using the recent prognostic model for MDS patients having failed hypomethylating agents15 that includes ECOG >1, very poor cytogenetics, age, bone marrow blasts >20%, transfusion dependency, platelets <30 haematologica | 2019; 104(8)

x109/L, 21 of 49 patients were classified as low-risk and 28 as high-risk, with a median OS of 9.2 vs. 5.7 months, respectively (HR=1.7, 95%CI: 0.8-3.8; P=0.16) (Figure 4B). This compared with 11 and 4.5 months, respectively, for the patients included in the prognostic model of Nazha et al. who had received various treatments after HMA failure.

Side effects Ninety-nine serious adverse events (SAE) occurred in 44 patients, and they were mostly hematologic, with myelosuppression in 88 of 99 (88%) of events. Thirteen patients were hospitalized for febrile neutropenia with a median duration of hospitalization of 14 days. Grade III-IV nonhematologic toxicities occurring in at least 3% of patients are shown in Table 3. Regarding toxicity at injection points, patients reported less pain and less secondary lesions with subcutaneous guadecitabine injections than with previous AZA injections.

Discussion In this phase II study, treatment with guadecitabine after AZA failure was generally safe in this elderly population, with limited dose reductions. It yielded a modest ORR of 14.3% and median OS of 7.1 months, but a few longer-term responders were seen, and some biological prognostic factors of response could be identified. Responders to guadecitabine in our study had a median OS of 17.9 months, compared with six months in nonresponders, the reported median survival of high-risk MDS patients after AZA failure in the literature.2 In previous smaller series of decitabine salvage after AZA failure, median OS ranged between 5.9 and 11.8 months.2,15,16 Compared with those series, and also our experience using decitabine in high-risk MDS/CMML patients after AZA failure (that reported no CR and a median response duration of only 3 months),17 the current two CR and median duration of response of 11.5 months achieved with guadecitabine, with 4 of 8 responses exceeding one year, may appear slightly better. A recent study comparing a 5-day regimen of guadecitabine (60-90 mg/m2/d) to a 10day regimen (60 mg/m2/d) in relapsed or refractory AML reported a response rate of 16% and 30.2% (P=0.1), and an OS of 5 and 7.1 months, respectively, with no significant difference between the two regimens.18 However, 1569

M. Sebert et al. A

Figure 3. Response to guadecitabine of patients treated after azacitidine failure. (A) Cumulative incidence of response. (B) Duration of response.

Figure 4. Overall survival of patients treated with guadecitabine after azacitidine failure. (A) Overall survival. (B) Survival according to Nazha score.

only four patients in this study had received first-line treatment with HMA (more than 1 cycle), and predictive factors of response and survival were not analyzed. In the present series, we observed significant demethylation in blood and bone marrow samples after one cycle of treatment, even if there was no significant correlation between the demethylation rate and the hematologic response, possibly because of the small number of responders. Similar results were reported using decitabine after AZA failure.16 However, a higher demethylation rate in blood on day 8 of cycle 1 was significantly associated with longer OS and remained a significant prognostic factor in multivariate analysis, suggesting that demethylation achieved with guadecitabine after AZA failure is a mechanism implicated in response. This could possibly also explain the relatively good response rate of patients with 1570

primary AZA failure, who might have experienced AZA resistance due to low level and duration of HMA exposure, but subsequently responded to guadecitabine, which induced greater demethylation. To our knowledge, this is the first study analyzing the prognostic factors of response and OS in high-risk MDS/AML receiving a HMA after failure of a first HMA. Previous studies reported that TET2 mutations,19-21 particularly at a VAF >10% and in the absence of co-occurring ASXL1 mutations,20 were associated with higher response rates, whereas a higher number of detectable mutations predicted for lower likelihood of response and complete response, as well as shorter response duration to HMA; however, those studies involved HMA naĂŻve patients.21 In the present study, the only predictive factor of response to guadecitabine was the absence or small number of somathaematologica | 2019; 104(8)

Guadecitabine in selected MDS and AML after azacitidine failure

ic mutations, associated with a better response to treatment, while no response was observed in 11 TP53 mutated patients. Thus, TP53 mutated MDS/AML patients may have high response rates (although of short duration) with early use of HMA (especially a 10-day regimen of decitabine),5 but after HMA failure, those patients may be particularly resistant to further HMA therapy, even with a different agent. Regarding OS, our multivariate analysis showed that primary AZA failure (vs. secondary failure), low to high R-IPSS and higher demethylation in blood (on day 8 of cycle1) were associated with better OS, factors that could help select patients more likely to benefit from second-line treatment with guadecitabine. The survival impact of guadecitabine did not seem to differ between “low-risk”

References 9. 1. Fenaux P, Mufti GJ, Hellstrom-Lindberg E, et al. Efficacy of azacitidine compared with that of conventional care regimens in the treatment of higher-risk myelodysplastic syndromes: a randomised, open-label, phase III study. Lancet Oncol. 2009; 10(3):223-232. 2. Prébet T, Gore SD, Esterni B, et al. Outcome of high-risk myelodysplastic syndrome after azacitidine treatment failure. J Clin Oncol. 2011;29(24):3322-3327. 3. Karahoca M, Momparler RL. Pharmacokinetic and pharmacodynamic analysis of 5-aza-2’-deoxycytidine (decitabine) in the design of its dose-schedule for cancer therapy. Clin Epigenetics. 2013;5(1):3. 4. Issa J-PJ, Kantarjian HM. Targeting DNA methylation. Clin Cancer Res. 2009; 15(12):3938-3946. 5. Welch JS, Petti AA, Miller CA, et al. TP53 and Decitabine in Acute Myeloid Leukemia and Myelodysplastic Syndromes. N Engl J Med. 2016; 375(21):2023-2036. 6. Chuang JC, Warner SL, Vollmer D, et al. S110, a 5-Aza-2’-deoxycytidine-containing dinucleotide, is an effective DNA methylation inhibitor in vivo and can reduce tumor growth. Mol Cancer Ther. 2010;9(5):14431450. 7. Kantarjian HM, Roboz GJ, Kropf PL, et al. Guadecitabine (SGI-110) in treatmentnaive patients with acute myeloid leukaemia: phase 2 results from a multicentre, randomised, phase 1/2 trial. Lancet Oncol. 2017;18(10):1317-1326. 8. Issa J-PJ, Roboz G, Rizzieri D, et al. Safety and tolerability of guadecitabine (SGI-110) in patients with myelodysplastic syndrome and acute myeloid leukaemia: a multicen-

haematologica | 2019; 104(8)

10.

11.

12.

13.

14.

15.

and “high-risk” patients, based on Nazha et al.’s scoring system for HMA failure patients. Altogether, our study suggests that some selected patients [primary AZA failure and low to high IPSS-R, patients with no or few somatic mutations (especially no TP53 mutations), patients with higher demethylation rate in blood during the first cycle of treatment] may benefit from guadecitabine treatment after AZA failure. Our results also suggest that primary AZA failures may be better candidates than secondary AZA failures to receive guadecitabine. An international phase III study (clinicaltrials.gov identifier: 02907359) is underway to compare guadecitabine treatment with best investigator’s choice in high-risk MDS patients relapsing or failing after first-line AZA or decitabine.

tre, randomised, dose-escalation phase 1 study. Lancet Oncol. 2015;16(9):1099-1110. Vardiman JW, Thiele J, Arber DA, et al. The 2008 revision of the World Health Organization (WHO) classification of myeloid neoplasms and acute leukemia: rationale and important changes. Blood. 2009;114(5):937-951. Greenberg P, Cox C, LeBeau MM, et al. International scoring system for evaluating prognosis in myelodysplastic syndromes. Blood. 1997;89(6):2079-2088. Bennett JM, Catovsky D, Daniel MT, et al. Criteria for the diagnosis of acute leukemia of megakaryocyte lineage (M7). A report of the French-American-British Cooperative Group. Ann Intern Med. 1985;103(3):460462. Cheson BD, Greenberg PL, Bennett JM, et al. Clinical application and proposal for modification of the International Working Group (IWG) response criteria in myelodysplasia. Blood. 2006;108(2):419425. Lübbert M, Suciu S, Baila L, et al. Low-dose decitabine versus best supportive care in elderly patients with intermediate- or highrisk myelodysplastic syndrome (MDS) ineligible for intensive chemotherapy: final results of the randomized phase III study of the European Organisation for Research and Treatment of Cancer Leukemia Group and the German MDS Study Group. J Clin Oncol. 2011;29(15):1987-1996. Kantarjian HM, Thomas XG, Dmoszynska A, et al. Multicenter, randomized, openlabel, phase III trial of decitabine versus patient choice, with physician advice, of either supportive care or low-dose cytarabine for the treatment of older patients with newly diagnosed acute myeloid leukemia. J Clin Oncol. 2012;30(21):2670-2677. Nazha A, Komrokji RS, Garcia-Manero G, et al. The efficacy of current prognostic

16.

17.

18.

19.

20.

21.

22.

models in predicting outcome of patients with myelodysplastic syndromes at the time of hypomethylating agent failure. Haematologica. 2016;101(6):e224-227. Borthakur G, Ahdab SE, Ravandi F, et al. Activity of decitabine in patients with myelodysplastic syndrome previously treated with azacitidine. Leuk Lymphoma. 2008;49(4):690-695. Duong VH, Bhatnagar B, Zandberg DP, et al. Lack of objective response of myelodysplastic syndromes and acute myeloid leukemia to decitabine after failure of azacitidine. Leuk Lymphoma. 2015;56(6):17181722. Harel S, Cherait A, Berthon C, et al. Outcome of patients with high risk Myelodysplastic Syndrome (MDS) and advanced Chronic Myelomonocytic Leukemia (CMML) treated with decitabine after azacitidine failure. Leuk Res. 2015;39(5):501-504. Roboz GJ, Kantarjian HM, Yee KWL, et al. Dose, schedule, safety, and efficacy of guadecitabine in relapsed or refractory acute myeloid leukemia. Cancer. 2018; 124(2):325-334. Itzykson R, Kosmider O, Cluzeau T, et al. Impact of TET2 mutations on response rate to azacitidine in myelodysplastic syndromes and low blast count acute myeloid leukemias. Leukemia. 2011;25(7):11471152. Bejar R, Lord A, Stevenson K, et al. TET2 mutations predict response to hypomethylating agents in myelodysplastic syndrome patients. Blood. 2014;124(17):2705-2712. Montalban-Bravo G, Takahashi K, Patel K, et al. Impact of the number of mutations in survival and response outcomes to hypomethylating agents in patients with myelodysplastic syndromes or myelodysplastic/myeloproliferative neoplasms. Oncotarget. 2018;9(11):9714-9727.

1571

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(8):1572-1579

Myeloproliferative Neoplasms

Epigenomic profiling of myelofibrosis reveals widespread DNA methylation changes in enhancer elements and ZFP36L1 as a potential tumor suppressor gene that is epigenetically regulated

Nicolás Martínez-Calle,1,2# Marien Pascual,1,2# Raquel Ordoñez,1,2# Edurne San José Enériz,1,2 Marta Kulis,3 Estíbaliz Miranda,1,2 Elisabeth Guruceaga,4 Víctor Segura,4 María José Larráyoz,5 Beatriz Bellosillo,6 María José Calasanz,2,5 Carles Besses,7 José Rifón,2,8 José I. Martín-Subero,2,9,10 Xabier Agirre1,2* and Felipe Prosper1,2,8*

Área de Hemato-Oncología, Centro de Investigación Médica Aplicada, IDISNA, Universidad de Navarra, Pamplona; 2Centro de Investigación Biomédica en Red de Cáncer (CIBERONC), Madrid; 3Fundació Clínic per a la Recerca Biomèdica, Barcelona; 4Unidad de Bioinformática, Centro de Investigación Médica Aplicada, Universidad de Navarra, Pamplona; 5CIMA Laboratory of Diagnostics, Universidad de Navarra, Pamplona; 6 Departmento de Patología, Hospital del Mar, Barcelona; 7Departmento de Hematología, Hospital del Mar, Barcelona; 8Departamento de Hematología, Clínica Universidad de Navarra, Universidad de Navarra, Pamplona; 9Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Barcelona and 10Departament de Fonaments Clinics, Facultat de Medicina, Universitat de Barcelona, Barcelona, Spain. 1

#These authors share first authorship

*These authors share senior authorship

ABSTRACT

Correspondence: XABIER AGIRRE xaguirre@unav.es FELIPE PROSPER fprosper@unav.es Received: August 23, 2018. Accepted: January 15, 2019. Pre-published: January 17, 2019. doi:10.3324/haematol.2018.204917 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1572 ©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.

1572

n this study we interrogated the DNA methylome of myelofibrosis patients using high-density DNA methylation arrays. We detected 35,215 differentially methylated CpG, corresponding to 10,253 genes, between myelofibrosis patients and healthy controls. These changes were present both in primary and secondary myelofibrosis, which showed no differences between them. Remarkably, most differentially methylated CpG were located outside gene promoter regions and showed significant association with enhancer regions. This aberrant enhancer hypermethylation was negatively correlated with the expression of 27 genes in the myelofibrosis cohort. Of these, we focused on the ZFP36L1 gene and validated its decreased expression and enhancer DNA hypermethylation in an independent cohort of patients and myeloid cell-lines. In vitro reporter assay and 5’-azacitidine treatment confirmed the functional relevance of hypermethylation of ZFP36L1 enhancer. Furthermore, in vitro rescue of ZFP36L1 expression had an impact on cell proliferation and induced apoptosis in SET-2 cell line indicating a possible role of ZFP36L1 as a tumor suppressor gene in myelofibrosis. Collectively, we describe the DNA methylation profile of myelofibrosis, identifying extensive changes in enhancer elements and revealing ZFP36L1 as a novel candidate tumor suppressor gene.

Introduction Philadelphia chromosome-negative myeloproliferative neoplasms, namely polycythemia vera (PV), essential thrombocythemia (ET) and primary myelofibrosis (MF), are characterized by a clonal transformation of hematopoietic progenitors leading to expansion of fully differentiated myeloid cells.1 Primary MF carries the worst prognosis of all Philadelphia chromosome-negative myeloproliferative neoplasms, with progressive marrow fibrosis, extramedullary hematopoiesis, mild to severe splenomegaly and an increased risk of transformation into leukemia.2 Secondary MF can also arise from PV and ET (hereafter referred to as post-PV and haematologica | 2019; 104(8)

ZFP36L1 enhancer DNA methylation in MF

post-ET MF, respectively) by mechanisms that are still poorly understood and are clinically and morphologically indistinguishable from primary MF.3 MF has been intensively studied from the genetic perspective;4,5 in fact, the modified World Health Organization (WHO) diagnostic criteria for Philadelphia chromosome-negative myeloproliferative neoplasms require the demonstration of a genetic marker of clonal hematopoiesis (JAK2V617F, CALR or MPL mutations).6 The frequency of mutations on relevant epigenetic genes (i.e., DNMT3A, EZH2 and ASXL1) suggests that MF might have an epigenetic component that, to our knowledge, remains poorly characterized.5 So far, epigenetic changes such as DNA methylation have been scarcely addressed in MF7 partly due to the limited changes in promoter DNA methylation compared to those in other hematologic malignancies, as previously published by our group.8 DNA methylation of CpG islands (CGI) (mostly on putative promoter regions) has been traditionally studied in both normal and neoplastic hematopoiesis.9,10 However, highthroughput platforms offer a wider coverage of the genome, allowing a better understanding of DNA methylation dynamics in regions distant from CGI.11 In this regard, enhancer regions have been characterized as potentially relevant sites of DNA methylation outside CGI.12-14 Chromatin immunoprecipitation-sequencing studies have enabled reliable mapping of genome-wide active enhancer regions based on histone modifications (e.g., H3K4me1 and H3K27Ac),15,16 allowing the identification of enhancers playing a role in dynamic transcriptional regulation during hematopoiesis.17 The present work describes a comprehensive genomewide analysis of DNA methylation in MF patients, coupled with a gene expression analysis and information on functional chromatin states, compared with those of samples from healthy donors.16 Focusing on potential epigenetic alterations in enhancer regions, we identified ZFP36L1 as a potential tumor suppressor gene with relevance for the pathogenesis of MF.

Methods Patients’ samples and clinical data Samples from MF patients (n=39) were bone marrow, granulocytes or total peripheral blood cells. The MF cohort comprised cases of primary MF (n=22), post-ET MF (n=7) and post-PV MF (n=10). Peripheral blood cells from healthy donors (n=6) were used as control samples in this study. All patients were diagnosed using the 2008 version of the WHO classification system of hematologic malignancies.18 Data on JAK2V617F mutation status were retrospectively available for all patients, whereas no data on CALR and MPL mutations were available. The patients’ data are accessible from the Gene Expression Omnibus (GSE118241). Samples and patients’ data were provided by the Biobank of the University of Navarra and were processed following standard operating procedures approved by the local Ethics & Scientific Committee. Prior to the collection of samples, all patients consented to the use of their data and to the use of stored material for research purposes.

DNA methylation profiling DNA methylation was assessed using a Human-Methylation 450K Bead-Chip kit (Illumina, Inc., San Diego, CA, USA) and the data were analyzed by Bioconductor open source software. The haematologica | 2019; 104(8)

analytical pipeline implemented several filters to exclude technical and biological biases and take into account the performance characteristics of Infinium I and Infinium II assays.19 Differentially methylated CpG were defined as previously described.13,19 Details on the experimental procedures, annotation of CpG sites, detection of differentially methylated regions, and Gene Ontology analysis20 are described in the Online Supplementary Methods.

Identification of candidate genes targeted by aberrant DNA methylation in enhancers Data on gene expression profiling from primary MF and healthy peripheral blood samples were obtained from the publicly available Gene Expression Omnibus accession bank number GSE26049.21 Data were further processed using R and the open source Limma package.22 Further details are described in the Online Supplementary Methods.

Luciferase reporter assays The CpG-free vector (pCPG-L), kindly provided by Dr. Michael Rehli,23 was used to clone the ZFP36L1 enhancer region. Luciferase experiments were performed in triplicate and the details are described in the Online Supplementary Methods. Primer sequences are available in Online Supplementary Table S1.

ZFP36L1 binding motif search To further validate the potential relevance of the ZFP36L1 gene in MF, the DREME motif discovery algorithm24 was used to assess enrichment of genes with the ZFP36L1 consensus binding sequence among those genes differentially expressed in MF [false discovery rate (FDR)≤ 0.05].

Overexpression of ZFP36L1 A vector containing the ZFP36L1 open reading frame was kindly provided by Dr. Murphy and subcloned into a PL-SIN-GK vector.25 Further details are described in the Online Supplementary Methods.

Statistical analysis For parametric group comparisons one-way analysis of variance (ANOVA) with the Dunnet correction was used, whereas for nonparametric group comparisons the Kruskall-Wallis test with the Dunn correction was employed. Paired data were analyzed with a Friedman non-parametric test with the Dunn correction for multiple comparisons, for the data with single measurements. Twoway ANOVA with the Tukey correction was used for data with multiple paired measurements. All tests were performed using Prism 7TM software (GraphPad, La Jolla, CA, USA). Details of other experimental procedures are given in the Online Supplementary Methods.

Results Myelofibrosis is characterized by a specific DNA methylation pattern enriched in enhancer regions In order to provide an exhaustive analysis of the DNA methylation profile in patients with MF, we analyzed the DNA methylome of patients with primary MF, secondary MF (including post-ET/post-PV MF) and healthy donors as controls, using the Human- Methylation 450K array. The first result worth highlighting was the epigenetic similarity between primary and post-ET/post-PV MF. Interestingly, with a FDR<0.05, no differentially methylated CpGs were found between primary and secondary MF. Furthermore, we did not identify any differentially methy1573

N. MartĂnez-Calle et al.

lated CpG between post-ET and post-PV MF. However, both unsupervised principal component analysis (PCA) (Figure 1A) and hierarchical clustering studies (Online Supplementary Figure S1A) using all CpG analyzed confirmed an explicit segregation and a clear epigenetic differ-

ence between samples from patients with MF and those from healthy controls. These results allowed us hereafter to consider all MF samples as a single sample cohort. Next, we sought to interrogate differences in DNA methylation between MF samples and healthy controls. In

Figure 1. Patients with myelofibrosis have a different DNA methylation profile from controls, with changes located primarily in enhancer regions. (A) Unsupervised principal component analysis (PCA) showing a differential DNA methylation profile of myelofibrosis (MF) patients and healthy controls with no differences between primary and secondary MF. (B) Distribution of differentially methylated CpG according to CpG island mapping (left graph) or functional chromatin analysis (right graph) grouped by DNA methylation status of the probes (legend). *Pâ&#x2030;¤0.05. (C) Heatmap of DNA methylation levels of differentially methylated CpG sites located in enhancer regions in MF patients and healthy controls. (D) GO-PANTHER analysis of genes adjacent to differenatially methylated CpG located in enhancer regions. Analysis of hypermethylated and hypomethylated genes is shown in the left and right panels, respectively. PC1: principal component 1; PC2: principal component 2; PMF: primary myelofibrosis; PV: polycythemia vera; MF: myelofibrosis; ET: essential thrombocythemia; CGI: CpG islands; DMC: differentially methylated CpG.

1574

haematologica | 2019; 104(8)

ZFP36L1 enhancer DNA methylation in MF

this supervised analysis, we detected 35,215 differentially methylated CpG (FDR≤0.05) corresponding to 10,253 coding genes. Among all of these differentially methylated CpG, 65.3% were hypomethylated (corresponding to 22,998 CpG) and the remaining 34.7% were hypermethylated (a total of 12,217 CpG), suggesting that loss of DNA methylation is the predominant alteration in MF. Global DNA hypomethylation has also been a common finding in other hematologic malignancies such as chronic lymphocytic leukemia, multiple myeloma and acute myeloid leukemia.13,26,27 Analysis of the genomic location of differentially methylated CpG showed that both hyper- and hypomethylated CpG were underrepresented in classical CGI and significantly enriched outside CpG islands (Figure 1B). This is an interesting finding, because traditionally, neoplasms acquire hypomethylation outside CGI and hypermethylation inside the islands,13,27 and suggests that patterns of methylation gain in MF might differ from those of other neoplasms. To shed light onto the specific function of the differentially methylated CpG, the chromatin state of each CpG was categorized adapting a publicly available annotation of chromatin immunoprecipitation-sequencing data from CD34+ hematopoietic progenitor cells, in which four distinct states were defined: promoter (with H3K4me3), active enhancer (with H3K4me1 and H3K27ac), transcribed regions (showing H3K36me3) and heterochromatin (including H3K9me3 and H3K27me3).16 Both hyper- and hypo-methylated CpG showed significant enrichment in enhancer regions, together with a striking underrepresentation in promoter regions (Figure 1B). Unsupervised clustering of differentially methylated CpG located exclusively in enhancer regions (Online Supplementary Table S2) displayed a clear segregation of the majority of MF patients from healthy controls (Figure 1C) identifying 4,182 hypermethylated and 10,935 hypomethylated probes. These results suggest that patients with MF show an intrinsic aberrant pattern of DNA methylation preferentially located in enhancer regions of the genome. To further characterize the aberrant DNA methylation of enhancer regions in MF, GO-PANTHER enrichment analysis was performed separately in differentially methylated genes. GO terms with an adjusted FDR<0.05 were selected, showing in the case of hypermethylated enhancers relevant cellular processes such as cellular defense response or induction of apoptosis (Figure 1D).

DNA methylation of enhancer regions is associated with gene expression profile in myelofibrosis DNA methylation levels of enhancer regions were correlated with the expression of host and adjacent coding genes using publicly available gene expression data of an independent cohort of MF patients and healthy donors (GSE26049).21 Fold increases in gene expression values were grouped according to the hypermethylated (Δβ>0.4) or hypomethylated (Δβ<-0.4) enhancer status in MF versus controls. This analysis showed that enhancer DNA hypermethylation was associated with decreased gene expression of host/adjacent coding genes. In contrast, hypomethylated enhancer regions were not associated with increased gene expression (Figure 2A). Next, we designed a more stringent approach to identify the set of genes underlying the most significant and substantial changes in enhancer DNA methylation (FDR haematologica | 2019; 104(8)

<0.01, Δβ>0.4), coupled with downregulation of their expression (logFC<0) (Figure 2B). After identifying a number of potential candidates (27 genes), we focused on ZFP36L1, which codes for a RNA-binding protein that mediates the decay of unstable mRNA with AU rich elements in the 3’ untranslated region.28,29 Interestingly, the enhancer region associated with this candidate gene was located in its intragenic region, presumably acting as a cisregulatory element of ZFP36L1 transcription. It is worth noting that this regulatory element was consistently hypermethylated in the cohort of MF patients and showed the largest number of hypermethylated enhancer-related CpG probes among the final 27-gene list. ZFP36L1 enhancer hypermethylation correlated with downregulation of expression in MF as compared to controls (Figure 2B and Online Supplementary Figure S1B), which was further confirmed in an independent cohort of MF patients and myeloid cell lines (Figure 2C). Bisulfite sequencing confirmed that DNA methylation of the enhancer region of ZFP36L1 was consistently higher in all MF samples and myeloid cell lines than in control samples, whereas the promoter region remained unmethylated (Figure 2D,E and Online Supplementary Figure S1C). Results obtained from luciferase-reporting assays demonstrated that the exogenous DNA methylation significantly reduced ZFP36L1 enhancer activity (Figure 2F). Moreover, 5’́-azacytidine hypomethylating treatment was able to reverse the DNA methylation levels of the enhancer region in vitro, partially restoring the gene expression levels of ZFP36L1 in the SET-2 cell line (Figure 2G,H).

ZFP36L1 acts as a tumor suppressor gene and potentially affects the myelofibrosis transcription profile We hypothesized that ZFP36L1 downregulation could lead to upregulation of its putative targets in MF. We used DREME, a motif discovery algorithm specifically designed to find short, core DNA-binding motifs enriched in the 3’ untranslated region of genes. We found that the GTATTTDT motif (E-value=4.5x10-15) was in fact overrepresented in transcripts upregulated in MF patients (Figure 3A). Subsequently, an analysis of motif enrichment was performed, revealing a significant enrichment of upregulated genes in MF patients among the group containing the mentioned motif (P=7.69x10-20; logFC >1; P<0.05). To complement DREME analysis, we searched the AREsite30 database for AU-rich elements to determine whether we could detect, among the genes differentially expressed (B-value>10) between MF and controls, an enrichment of these sequences in the upregulated subset. Of all the possible AU motifs, we focused on the most restricted 9, 11 and 13-mer motifs. Interestingly, we were able to identify an enrichment of a 9-mer sequence WTATTTATW (P=0.01) and a 13-mer sequence WWWTATTTATWWW (P=0.03) exclusively among the upregulated genes in MF patients (Figure 3A). Remarkably, both AU motifs strongly resemble the ZFP36L1 core-binding motif predicted by the DREME algorithm. Re-expression of ZPF36L1 was achieved through lentiviral infection of the SET-2 cell line. Seventy-two hours after infection, the levels of EGFP-positive cells used as the positive control confirmed successful infection, and the level of expression of ZFP36L1 confirmed 1575

N. Martínez-Calle et al.

satisfactory overexpression of the gene (Figure 3B-D). Rescue of ZFP36L1 expression resulted in a decrease of more than 50% in cell proliferation, alongside an increase of annexin V-positive cells as measured by flow cell cytometry (Figure 3E,F).

Discussion In the present study, we have extended previous knowledge regarding the DNA methylome in both primary and secondary MF, focusing specially on those CpG sites locat-

Figure 2. Aberrant enhancer DNA methylation regulates gene expression in myelofibrosis. (A) Violin density plots of expression of genes with differentially methylated CpG located in enhancer regions. The vertical axis represents log fold change in gene expression. The horizontal width of the plot represents density of data along the y axis. (B) Candidate genes with substantial changes in DNA methylation (FDR <0.01 and Δβ>0.4) and differential gene expression (logFC<0). Red bars represent the average DNA methylation of all enhancer-mapped probes, the black bars represent the average expression of all probes, and the error bars represent the standard deviation (SD) (C) ZFP36L1 downregulation validation by real-time quantitative polymerase chain reaction analysis of myelofibrosis (MF) patients and three myeloid cell lines (including SET-2) compared to healthy controls (n=3). (D,E) Bisulfite sequencing of the ZFP36L1 enhancer region (D) and promoter region (E) in healthy controls, cell lines and primary MF samples. For each sample, the graph shows the mean ± SD of ten CpG dinucleotides for enhancer regions and 15 CpG dinucleotides for promoter regions. (F) pCpG-L luciferase reporter assay showing the inhibition of luciferase activity after treatment of the ZFP36L1 enhancer region with Sss-I methyltransferase. (G) DNA methylation levels of the enhancer region – the same ten CpG dinucleotides as in (D) after 5-azacytidine treatment of SET-2. (H) ZFP36L1 expression levels after 5-azacytidine treatment of SET-2. Plots/bars indicate mean ± SD. FC: fold change; DMC: differentially methylated CpG; FDR: false discovery rate; CONTROL: healthy controls. MF; myelofibrosis; AZA: 5-azacytidine.

1576

haematologica | 2019; 104(8)

ZFP36L1 enhancer DNA methylation in MF

ed in enhancer regions of the genome. A preliminary analysis of the global DNA methylome revealed the absence of DNA methylation differences between primary and secondary MF. This constitutes the first key finding of the present study and allowed us to use all the MF samples in a single cohort for further analysis. Primary and secondary MF are known to have very similar biological features, presenting symptoms and clinical course and in fact, both entities are treated indistinctively according to most published guidelines3, 31 Nevertheless, some recent evidence from large retrospective trials has suggested that traditional prognostic factors may not be applicable to secondary MF as patients with post-ET MF seem to survive longer than those with post-PV MF or primary MF.3,31-33

The remarkably homogenous epigenetic profile of all our MF samples supports a common biological origin of primary and secondary MF.34 The DNA methylomes of the novel MF subtypes defined by the new 2016 WHO classification (prefibrotic and overt MF) remain to be characterized and it will be interesting to establish whether these subtypes have different methylation profiles. This aspect exceeded the possibilities of our cohort (retrospective availability of histology samples) but warrants further investigation. Although previous studies have already interrogated the DNA methylation landscape of MF,7 their findings are limited to small numbers of epigenetic abnormalities mainly focused on the study of promoter regions. Our genome-

Figure 3. ZFP36L1 rescue decreases cell viability in myelofibrosis. (A) Consensus binding motif for ZFP36L1 obtained by DREME motif discovery among transcripts with putative AU-rich motifs upregulated in myelofibrosis samples. (B) Efficiency of infection measured by the percentage of EGFP-positive cells after lentiviral infection. (C) Quantitative polymerase chain reaction validation of ZFP36L1 restoration in the SET-2 cell line after lentiviral infection. (D) ZFP36L1 protein restoration measured by western blot in the SET-2 cell line after lentiviral infection. (E,F) ZFP36L1 rescue with lentiviral vector infection in the SET-2 cell line decreased cell proliferation rate (E) and increased annexin V-positive cells (F).

haematologica | 2019; 104(8)

1577

N. Martínez-Calle et al.

wide approach of DNA methylation analysis using the 450k array allowed us to interrogate regulatory regions outside traditional promoters and obtain a deeper insight into the aberrant DNA methylome of MF. Changes in DNA methylation levels are known to cooperate with the deposition of chromatin marks, particularly H3K4 methylation, to render the enhancers/promoters accessible/inaccessible to the transcription machinery.35-37 Hence, the changes in DNA methylation observed in MF are expected to have an impact on the transcriptional profile of MF and potentially contribute to the MF malignant phenotype. Enhancer DNA methylation changes have been described to play a more prominent role in transcriptional regulation than promoter DNA methylation, governing processes such as hematopoietic differentiation and neoplastic transformation through the regulation of key transcription factors and genes.12,13,27,37,38 Translated into the context of Philadelphia chromosome-negative myeloproliferative neoplasms, this evidence might support the involvement of aberrant enhancer DNA methylation in the abnormal pattern of differentiation leading to MF. Enhancer hypermethylation has been reported in neutrophils,12 B cells,39 AML cells26 and myeloma13 adding evidence to dynamic enhancer DNA hypermethylation as a relevant regulatory mechanism of gene expression both in normal and neoplastic hematopoietic cells. Although the potential involvement of ZFP36L1 in myeloid differentiation has been described previously,40 our results suggest that epigenetic downregulation of ZFP36L1 might be a prominent event in the pathobiology of MF; more importantly, hypermethylation of an enhancer regulatory element represents a novel mechanism of disrupted gene expression in the context of MF and ZFP36L1. ZFP36L1 has been previously implicated in normal hematopoiesis41 and specifically associated with erythroid and myeloid differentiation,40 suggesting a possible role of this gene in MF onset and progression. Moreover, ZFP36L1 is also known to mediate mRNA decay of genes relevant to cell proliferation, survival and differentiation such as CDK6, TNFα, BCL2, NOTCH1 and STAT5B.42,43 Interestingly, the enhancer region associated with this candidate gene was consistently hypermethylated in the cohort of MF patients and was located in its intragenic region, presumably acting as a cis-regulatory element of ZFP36L1 transcription. The motif discovery experiments support our hypothesis of epigenetic deregulation of ZFP36L1, suggesting that MF samples with ZFP36L1 loss of expression experience upregulation of the gene’s putative targets. Consequently, when ZFP36L1 expression levels are restored with the lentiviral model,

References 1. Spivak JL. Myeloproliferative neoplasms. N Engl J Med. 2017;376(22):2168-2181. 2. Kim J, Haddad RY, Atallah E. Myeloproliferative neoplasms. Dis Mon. 2012;58(4):177-194. 3. Passamonti F, Rumi E, Caramella M, et al. A dynamic prognostic model to predict survival in post-polycythemia vera myelofibrosis. Blood. 2008;111(7):3383-3387. 4. Kim SY, Im K, Park SN, Kwon J, Kim J-A, Lee

1578

SET-2 cells lose their malignant proliferative phenotype, strengthening the tumor suppressor role of this gene in MF. Taken together, these results link ZFP36L1 to the pathobiology of MF, ultimately resulting in transcriptome deregulation of genes relevant to cell proliferation, survival and differentiation, as previously described.40,42,44,45

Conclusion The DNA methylation landscape of patients with primary MF or post-ET/post-PV MF is consistently different from that of healthy individuals. The absence of differences between primary MF and post-ET/post-PV MF suggests that the changes seen in MF are founding epigenetic alterations occurring at the level of stem cells of this myeloproliferative neoplasm and maintained in differentiated myeloid cells. Aberrant DNA methylation in MF is predominantly located in enhancer regions and has a significant impact on the expression of their target genes. Combining DNA methylation and gene expression data, we identified ZFP36L1 as an attractive new possible therapeutic target that shows a decrease of gene expression mediated by enhancer hypermethylation. Our results also suggest a direct effect of ZFP36L1 downregulation on the gene expression profile of MF, through upregulation of mRNA harboring canonical sites with AU-rich elements. In vitro rescue of ZFP36L1 expression had an impact on cell proliferation and induced apoptosis in the SET-2 cell line, indicating a possible role of ZFP36L1 as a tumor suppressor gene in MF. Moreover, treatment with 5’-azacytidine further evidenced the plausibility of ZFP36L1 pharmacological manipulation. Taken together, these results provide evidence of an unexplored therapeutic target in MF patients, which remains to be properly evaluated in the pre-clinical setting. Acknowledgments We particularly acknowledge the patients for their participation and the Biobank of the University of Navarra for its collaboration. We thank John J Murphy and Amor Alcaraz for providing ZFP36L1 expression constructs. This research was funded by grants from Instituto de Salud Carlos III (ISCIII) PI14/01867, PI16/02024 and PI17/00701, TRASCAN (EPICA), CIBERONC (CB16/12/00489; co-financed with FEDER funds), RTICC (RD12/0036/0068) and the Departamento de Salud del Gobierno de Navarra 40/2016. NM is supported by a FEHH-Celgene research grant, MP was supported by a Sara Borrell fellowship CD12/00540 and RO was supported by the Ministerio de Ciencia, Innovación y Universidades of Spain, Subprograma de Formación de Profesorado Universitario (FPU) award number FPU14/04331.

DS. CALR, JAK2, and MPL mutation profiles in patients with four different subtypes of myeloproliferative neoplasms: primary myelofibrosis, essential thrombocythemia, polycythemia vera, and myeloproliferative neoplasm, unclassifiable. Am J Clin Pathol. 2015;143(5):635-644. 5. Vainchenker W, Kralovics R. Genetic basis and molecular pathophysiology of classical myeloproliferative neoplasms. Blood. 2017;129(6):667-679. 6. 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. 7. Myrtue Nielsen H, Lykkegaard Andersen C, Westman M, et al. Epigenetic changes in myelofibrosis: distinct methylation changes in the myeloid compartments and in cases with ASXL1 mutations. Sci Rep. 2017;7 (1):6774. 8. Pérez C, Pascual M, Martín-Subero JI, et al. Aberrant DNA methylation profile of chron-

haematologica | 2019; 104(8)

ZFP36L1 enhancer DNA methylation in MF

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

ic and transformed classic Philadelphia-negative myeloproliferative neoplasms. Haematologica. 2013;98(9):1414-1420. Mendizabal I, Yi SV. Whole-genome bisulfite sequencing maps from multiple human tissues reveal novel CpG islands associated with tissue-specific regulation. Hum Mol Genet. 2016;25(1):69-82. Deaton AM, Webb S, Kerr ARW, et al. Cell type-specific DNA methylation at intragenic CpG islands in the immune system. Genome Res. 2011;21(7):1074-1086. Sandoval J, Heyn H, Moran S, et al. Validation of a DNA methylation microarray for 450,000 CpG sites in the human genome. Epigenetics. 2011;6(6):692-702. Rรถnnerblad M, Andersson R, Olofsson T, et al. Analysis of the DNA methylome and transcriptome in granulopoiesis reveals timed changes and dynamic enhancer methylation. Blood. 2014;123(17):e79-e89. Agirre X, Castellano G, Pascual M, et al. Whole-epigenome analysis in multiple myeloma reveals DNA hypermethylation of B cell-specific enhancers. Genome Res. 2015;25(4):478-487. Bell RE, Golan T, Sheinboim D, et al. Enhancer methylation dynamics contribute to cancer plasticity and patient mortality. Genome Res. 2016;26(5):601-611. Ernst J, Kheradpour P, Mikkelsen TS, et al. Mapping and analysis of chromatin state dynamics in nine human cell types. Nature. 2011;473(7345):43-49. Wijetunga NA, Delahaye F, Zhao YM, et al. The meta-epigenomic structure of purified human stem cell populations is defined at cis-regulatory sequences. Nat Commun. 2014;5:5195. Lara-Astiaso D, Weiner A, Lorenzo-Vivas E, et al. Immunogenetics. Chromatin state dynamics during blood formation. Science. 2014;345(6199):943-949. Tefferi A, Thiele J, Vardiman JW. The 2008 World Health Organization classification system for myeloproliferative neoplasms. Cancer. 2009;115(17):3842-3847. Kulis M, Merkel A, Heath S, et al. Wholegenome fingerprint of the DNA methylome during human B cell differentiation. Nat Genet. 2015;47(7):746-756. Mi H, Huang X, Muruganujan A, et al. PANTHER version 11: expanded annotation data from Gene Ontology and Reactome pathways, and data analysis tool enhancements. Nucleic Acids Res. 2017;45(D1):D183-D189. Skov V, Larsen TS, Thomassen M, et al. Whole-blood transcriptional profiling of interferon-inducible genes identifies highly

haematologica | 2019; 104(8)

22.

23.

24. 25.

26.

27.

28.

29.

30.

31.

32.

33.

upregulated IFI27 in primary myelofibrosis. Eur J Haematol. 2011;87(1):54-60. Ritchie ME, Phipson B, Wu D, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47-e47. Klug M, Rehli M. Functional analysis of promoter CpG methylation using a CpG-free luciferase reporter vector. Epigenetics. 2006;1(3):127-130. Bailey TL. DREME: motif discovery in transcription factor ChIP-seq data. Bioinformatics. 2011;27(12):1653-1659. Hotta A, Cheung AYL, Farra N, et al. Isolation of human iPS cells using EOS lentiviral vectors to select for pluripotency. Nat Methods. 2009;6(5):370-376. Qu Y, Siggens L, Cordeddu L, et al. Cancerspecific changes in DNA methylation reveal aberrant silencing and activation of enhancers in leukemia. Blood. 2017;129(7): e13-e25. 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. Hitti E, Bakheet T, Al-Souhibani N, et al. Systematic analysis of AU-rich element expression in cancer reveals common functional clusters regulated by key RNA-binding proteins. Cancer Res. May 2016. Baou M, Norton JD, Murphy JJ. AU-rich RNA binding proteins in hematopoiesis and leukemogenesis. Blood. 2011;118(22):57325740. Gruber AR, Fallmann J, Kratochvill F, Kovarik P, Hofacker IL. AREsite: a database for the comprehensive investigation of AUrich elements. Nucleic Acids Res. 2011;39(Database issue):D66-D69. Reilly JT, McMullin MF, Beer PA, et al. Use of JAK inhibitors in the management of myelofibrosis: a revision of the British Committee for Standards in Haematology Guidelines for Investigation and Management of Myelofibrosis 2012. Br J Haematol. 2014;167(3):418-420. Hernรกndez-Boluda J-C, Pereira A, Gรณmez M, et al. The International Prognostic Scoring System does not accurately discriminate different risk categories in patients with post-essential thrombocythemia and post-polycythemia vera myelofibrosis. Haematologica. 2014;99(4):e55-e57. Passamonti F, Giorgino T, Mora B, et al. A clinical-molecular prognostic model to predict survival in patients with post polycythemia vera and post essential thrombo-

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

cythemia myelofibrosis. Leukemia. 2017;31 (12):2726-2731. Brecqueville M, Rey J, Devillier R, et al. Array comparative genomic hybridization and sequencing of 23 genes in 80 patients with myelofibrosis at chronic or acute phase. Haematologica. 2014;99(1):37-45. Sharifi-Zarchi A, Gerovska D, Adachi K, et al. DNA methylation regulates discrimination of enhancers from promoters through a H3K4me1-H3K4me3 seesaw mechanism. BMC Genomics. 2017;18(1):964. King AD, Huang K, Rubbi L, et al. Reversible regulation of promoter and enhancer histone landscape by DNA methylation in mouse embryonic stem cells. Cell Rep. 2016;17(1):289-302. Blattler A, Yao L, Witt H, et al. Global loss of DNA methylation uncovers intronic enhancers in genes showing expression changes. Genome Biol. 2014;15(9):469. Hon GC, Rajagopal N, Shen Y, et al. Epigenetic memory at embryonic enhancers identified in DNA methylation maps from adult mouse tissues. Nat Genet. 2013;45(10):1198-1206. Bergmann AK, Castellano G, Alten J, et al. DNA methylation profiling of pediatric Bcell lymphoblastic leukemia with KMT2A rearrangement identifies hypomethylation at enhancer sites. Pediatr Blood Cancer. 2017;64(3):e26251. Vignudelli T, Selmi T, Martello A, et al. ZFP36L1 negatively regulates erythroid differentiation of CD34+ hematopoietic stem cells by interfering with the Stat5b pathway. Mol Biol Cell. 2010;21(19):3340-3351. Galloway A, Saveliev A, ล ukasiak S, et al. RNA-binding proteins ZFP36L1 and ZFP36L2 promote cell quiescence. Science. 2016;352(6284):453-459. Zekavati A, Nasir A, Alcaraz A, et al. Posttranscriptional regulation of BCL2 mRNA by the RNA-binding protein ZFP36L1 in malignant B cells. PLoS ONE. 2014;9(7): e102625. Liu J, Lu W, Liu S, et al. ZFP36L2, a novel AML1 target gene, induces AML cells apoptosis and inhibits cell proliferation. Leuk Res. 2018;68:15-21. Chen M-T, Dong L, Zhang X-H, et al. ZFP36L1 promotes monocyte/macrophage differentiation by repressing CDK6. Sci Rep. 2015;5(1):16229. Hodson DJ, Janas ML, Galloway A, et al. Deletion of the RNA-binding proteins ZFP36L1 and ZFP36L2 leads to perturbed thymic development and T lymphoblastic leukemia. Nat Immunol. 2010;11(8):717-724.

1579

ARTICLE Ferrata Storti Foundation

Myeloproliferative Neoplasms

Characteristics and outcomes of patients with essential thrombocythemia or polycythemia vera diagnosed before 20 years of age: a systematic review

Jean-Christophe Ianotto,1,2 Natalia Curto-Garcia,1 Marie Lauermanova,1,3 Deepti Radia,1 Jean-Jacques Kiladjian4 and Claire N. Harrison1

Department of Haematology, Guy’s and St Thomas’ NHS Trust, London, UK; 2Service d’Hématologie Clinique, Institut de Cancéro-Hématologie, Centre Hospitalier Régional et Universitaire de Brest, Brest, France; 3Institute of Hematology and Blood Transfusion, Prague, Czech Republic and 4Centre d’Investigation Clinique, Hôpital St Louis, Paris, France 1

Haematologica 2019 Volume 104(8):1580-1588

ABSTRACT

Correspondence: CLAIRE N HARRISON claire.harrison@gstt.nhs.uk Received: June 29, 2018. Accepted: January 21, 2019. Pre-published: January 24, 2019. doi:10.3324/haematol.2018.200832 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1580

lthough it is well known that myeloproliferative neoplasms occur in younger patients, few large cohorts of such patients have been reported. Thus, our knowledge about circumstances of diagnosis, outcome and treatment is limited, especially for children and young adults. We therefore performed a systematic review of cases, published since 2005, concerning patients aged below 20 years at the time of diagnosis of essential thrombocythemia or polycythemia vera. We identified 396 cases of essential thrombocythemia and 75 of polycythemia vera. The median age at diagnosis was 9.3 and 12 years, respectively, and females constituted 57.6% and 45% of the groups, respectively. Half of the patients were asymptomatic at diagnosis. The proportion of so-called triple negativity was high: 57% in essential thrombocythemia and 73% in polycythemia vera. The incidence of thrombosis during the follow-up was 9.3% in patients with polycythemia vera and less, 3.8%, in those with essential thrombocythemia. Venous events were predominant (84.2%), with hemorrhagic episodes being rarer (<5%). The risk of evolution also seemed low (2% to myelofibrosis and no reports of acute leukemia), but the median follow-up was only 50 months. Survival curves were not available. Half of the patients received an antithrombotic drug and 40.5% received a cytoreductive drug. All data should be analyzed with care because of the proportion of missing data (10.7% to 74.7%). This review highlights interesting points concerning this population of young patients with myeloproliferative neoplasms, including that such patients were identified as negative for all common driver mutations, but also shows the need for larger contemporary cohorts with longer follow-up to assess the true prognosis of these patients.

Introduction ©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.

1580

Essential thrombocythemia (ET) and polycythemia vera (PV) are the most prevalent myeloproliferative neoplasms (MPN). However, the median age at diagnosis of both conditions is over 60 years. Patients with these diseases are particularly exposed to risks of thromboembolic events and evolution into more aggressive disorders (myelofibrosis, myelodysplastic syndromes and acute myeloid leukemia), with a consequent heavy burden of morbidity and mortality.1 Current clinical guidelines concerning the diagnosis and management of ET and PV are generally written for older patients and emphasize that testing for a driver mutation and a bone marrow biopsy are fundamental to diagnose an MPN and that treatment should be adapted according to a classification into low risk or high risk based on the patients’ age and history of thrombosis or hemorrhage (prescription of aspirin and cytoreductive drugs) in order to reduce the occurrence of thrombosis.2-4 haematologica | 2019; 104(8)

Very young patients with ET and PV

Details of some large cohorts of young patients with MPN, defined sometimes as below 60 and at other times as below 40 years old, have already been published.5-8 Notwithstanding these publications, there are only sparse data concerning very young patients with MPN (aged below 20 years at diagnosis), particularly regarding details such as initial characteristics (reason for consultation, clinical features, bone marrow biopsy features) and outcomes (thrombosis, pregnancy, disease evolution, incidence of second cancer and survival). Furthermore, the utilization of therapeutic modalities in this population is largely unknown. For example, the proportions of very young patients treated with antiplatelet and/or cytoreductive therapies and the therapeutic goals are very poorly defined and, furthermore, there is no information about the potential, long-term sequelae of the treatments. Here we present a review of published cases of ET and PV patients below the age of 20 years at the time of diagnosis. The earliest data collection point we chose was 2005, coincident with important discoveries concerning the molecular pathogenesis of these conditions, so that we would have more information about the mutational status of the patients and to improve the likelihood of ruling out reactive conditions confounding the diagnosis.9-12 We describe the biological and clinical characteristics at the time of diagnosis and the incidence of vascular and longterm complications during the follow-up.

For this review, we analyzed papers referring only to patients with ET or PV who were under 20 years old at the time of the diagnosis. The keywords used during the research were: polycythemia vera, primary or essential thrombocythemia/thrombocytosis, myelofibrosis, myeloproliferative neoplasms or diseases, young patients/adults, children/childhood and pediatric cases. To avoid the bias of recording misdiagnosed PV cases, only papers published since 2005 were eligible, coincident with the date of the discovery of the JAK2V617F mutation. Furthermore, we directly excluded familial MPN cases, if this was clearly specified in the title or the text. All types of articles were collected: general reviews, cohort papers (including more than 5 patients) and case reports (including fewer than 5 patients).

Methods

Collected characteristics

PubMed research The purpose of this review was to learn more about young patients diagnosed with MPN: their characteristics at diagnosis and the incidence of complications. To our knowledge there has been no systematic review of the published cases. We used PubMed (https://www.ncbi.nlm.nih.gov /pubmed) to identify articles related to our topic.

Selection of the articles We identified 87 articles concerning MPN and young patients and we finally analyzed 46 articles after exclusion of redundant papers (same authors, same numbers of patients), uninformative cases (no information about diagnosis or outcomes), alternative myeloid diseases (primary myelofibrosis, acute leukemia), inadequate age identification (cohorts of young patients, but below 40 years old) as delineated in the PRISMA flowchart (Online Supplementary Figure S1) and the checklist (Online Supplementary Table S1). In total, we identified 46 informative articles: 19 cohort papers (16 on ET and 3 on PV) and 27 case reports (23 on ET and 11 on PV, with some concerning both conditions).9-56 We also added seven papers with useful information concerning epidemiology.57-63

All informative articles were printed (published articles and supplementary data) and searched data were extracted and reported in an Excel file. The following data collected at the time of diagnosis were recorded: age, sex, circumstances of diagnosis (e.g. thrombosis), symptoms (hyperviscosity, pain, fatigue, pruritus, microvascular events), full blood count (leukocyte, hemoglobin and platelet levels), previous cardiovascular events (thrombosis and hemorrhage),

Figure 1. Driver mutations among very young patients with (A) essential thrombocythemia (n=206) or (B) polycythemia vera (n=55). ET: essential thrombocythemia; PV: polycythemia vera; 3NEG: triple negative for the JAKV617F, CALR and MPL driver mutations. 2NEG: double negative for JAK2V617F and JAK2 exon 12.

haematologica | 2019; 104(8)

1581

J-C. Ianotto et al.

molecular status (JAK2V617F or JAK2 exon 12, positivity for CALR or MPL, JAK2 allele burden) and detailed molecular analysis. The duration of the follow-up was assessed as precisely as possible, the prescribed drugs and the incidence of complications were then recorded: rate and type of antithrombotic drugs (lowdose aspirin, clopidogrel, vitamin K antagonists, heparins), rate and type of cytoreductive drugs (hydroxycarbamide, anagrelide, interferon, ruxolitinib, others), together with information on venesection requirement, incidence of cardiovascular events (thromboses and hemorrhages), evolution (ET into PV, ET or PV into secondary myelofibrosis or acute leukemia), and death. With regards to the thrombotic events, the type of vessel and the localization were recorded.

Results Epidemiology Only a few papers considered epidemiological data specifically for the MPN population and, it was sometimes difficult to assess the presence or absence of chronic myelogenous leukemia among the cases described. Furthermore, there was great variability in the incidence of MPN between countries (United Kingdom, Denmark, Europe, Japan), in the timing of the observation (from 1980 to 2010) and in the age of the patients (below 14 to below 25 years old). Overall, we found that the global incidence of MPN, in children and young adults, can be estimated to be around 0.82/100,000 patients/year (range, 0.1 to 2.25): the incidence of ET is around 0.6/100,000 patients/year (range, 0.004 to 0.9), against 0.18 for PV and 0.53 for primary myelofibrosis (range, 0.003 to 1.5).

Clinical and biological data at diagnosis Clinical characteristics On analyzing the published literature, we were able to collect data on 471 patients, of whom 396 (84%) had ET and 75 (16%) had PV. These patientsâ&#x20AC;&#x2122; clinical and biological characteristics are summarized in Table 1. For each described parameter, we also give the number and the percentage of data available in the published population. The median age at diagnosis was 9.3 years for ET patients and 12 years for PV patients. The percentage of female cases was also different between the two groups (57.6% in ET and 45% in PV). The reason for the original consultation was unclear or unknown in most cases. At the time of the diagnosis, 49.6% of the ET patients and 47.5% of PV were declared to be asymptomatic. For the other patients, the two most frequent symptoms experienced were headaches (27.5% in ET and 30.5% in PV) followed by abdominal or bone pain (5.5% and 3.4%, respectively). As a potential bias, a group of 30 patients with ET were declared to suffer from microvascular disturbances but without a more precise description. Interestingly, 13.6% of PV patients and 4.7% of ET patients were diagnosed following a thrombotic or hemorrhagic event. Splenomegaly was the most frequent abnormal sign described in the papers: 54.7% of ET and 15.3% of PV patients had a palpable spleen. Surprisingly, its presence did not seem to have induced so many abdominal symptoms as there were discrepancies between frequencies of splenomegaly and reported abdominal pain. It is also hard to understand the much higher frequency of splenomegaly in ET and the fact that this does not seem to have correlated with abdominal vein thrombosis, for example.

Table 1. Clinical and biological characteristics at diagnosis of very young patients with essential thrombocythemia or polycythemia vera. Number of cases (%) Median age (years) Range ( years) Male (%) Reasons for consultation or symptoms, n (%) Asymptomatic Thrombosis Hemorrhage Splenomegaly Headaches Abdominal/bone pain Paresthesia/erythromelalgia Syncope Fatigue Pruritus Full blood counts at presentation Leukocytes, x 109/L Hemoglobin, g/L Platelets, x 109/L Driver mutations JAK2 exon14, n Allele burden, % JAK2 exon12, n CALR, n MPL, n

Essential thrombocythemia

Polycythemia vera

396 (84) 9.3 0.2-20 42.4 *236 (59.6) 117 (49.6) 7 (3) 4 (1.7) 129 (54.7) 65 (27.5) 13 (5.5) 11 (4.7) 3 (1.3) 2 (0.8) 0 *229 (57.8) 10.6 131 1192 *388 (98.2) 130 24.1 23 (type1, n=9; type2, n=6) 4 (L, n=2; K, n=1)

75 (16) 12 0.6-19 55 *59 (78.7) 28 (47.5) 5 (8.5) 3 (5.1) 9 (15.3) 18 (30.5) 2 (3.4) 1 (1.7) 3 (5.1) 4 (6.8) 3 (5.1) *67 (89.3) 13.2 157 799 *75 (100) 30 43.5 2 -

*indicates the number and percent of available data for each category of parameters.

1582

haematologica | 2019; 104(8)

Very young patients with ET and PV

Biological characteristics The results of the full blood count at diagnosis are presented in Table 1. For ET patients, the median leukocyte count was 10.6 109/L, the hemoglobin concentration was 131 g/L and the platelet count was 1192x109/L (maximum 4500x109/L). For PV patients, the median leukocyte count was 13.2x109/L, the hemoglobin concentration was 180 g/L (maximum level, 189 g/L), the maximum hematocrit was 72.5% and the platelet count was 799x109/L. It is difficult to understand how hereditary thrombocytosis and erythrocytosis were excluded for these patients. To assess the diagnosis of MPN, many authors wrote in the “Patients and methods” section that their patients fulfilled the diagnostic criteria according to the 2001, 2008 or 2016 World Health Organization classification. However, bone marrow results were described for less than 52% of the ET and 44% of the PV cases. Generally, the descriptions were short with a conclusion expressed as “compatible with MPN”.

Molecular analyses Considering the entire cohort of PV patients, we noted that only 37% and 2.5% were positive for JAK2 exon 14 and exon 12 mutations, respectively. Three studies comprehensively assessed the presence of both types of JAK2 mutations: the percentage of JAK2 exon 14 cases decreased to 24% whereas, the rate of JAK2 exon 12 mutations remained stable at 3% (Figure 1A).9,13,24 According to these results, the percentage of patients positive for the V617F mutation in exon 14 is far less than in adults, whereas that of exon 12 is identical. Consequently, the percentage of patients who do not harbor one of these two mutations is also high: 73%. Information on JAK2V617F allele burden was available in a small number of studies and the mean value was 43.5%. This finding of much lower rates of JAK2 mutation in young patients with PV is also unexpected and requires a prospective evaluation, including, for example, the role of red cell isotopic studies to confirm the diagnosis. The analysis in ET patients is more complex because of

the number of mutations to test. Notwithstanding, in a global analysis, the percentages of positivity were 31.7% for JAK2V617F, 5.6% for CALR and 1% for MPL mutations Analyzing the eight cohorts of ET patients in whom all driver mutations were tested, the proportions of positivity became 31% for JAK2V617F, 10% for CALR, and 2% for MPL and so 57% of the cases were triple-negative (Figure 1B).12,14,16,17,19,20,22,33 These percentages are quite different from those in adults, among whom there is a much higher frequency of triple-negative cases and lower percentages of JAK2- or CALR-positive cases.64 As for PV patients, the number of studies reporting JAK2V617F allele burden was small and the mean value was 24.1%. As discussed earlier the exclusion of hereditary cases is critical here. Interestingly, two groups have reported the results of next-generation sequencing analyses in this population.12,16 Among 68 patients tested the authors found that 35% did not carry any of the tested non-driver mutations. Most of the patients carried only one additional mutation. The description and the proportions of non-driver mutations are shown in Figure 2. In one study, patients with ET seemed to have more mutations than PV patients.16 The presence of mutations belonging to the high molecular risk group and inducing worse prognosis in primary myelofibrosis is uncommon in this population (ASXL1 in 4 cases and IDH1/2 in 1 case).65,66 As far as concerns young patients with ET or PV, the real clinical significance of these nondriver mutations cannot be assessed from the studies.

Outcomes and survival Thrombotic events As noted above, some very young patients were diagnosed with MPN because of the occurrence of a thrombotic event. The exact incidence of thrombosis at diagnosis was 14.7% and 4% in PV and ET patients, respectively. So, these events seemed quite infrequent compared to their frequency in adults.67-68 Importantly, the incidence of thrombosis after the diagnosis of MPN decreased in the PV cohort (9.3%), but remained stable in the ET patients (3.8%) (Table 2).

Figure 2. Description and numbers of non-driver mutations identified by next-generation sequencing in two studies (68 patients).12,16 mut: mutations.

haematologica | 2019; 104(8)

1583

J-C. Ianotto et al.

Global analyses of recurrence of thrombotic events are not available here since most of the data were published for entire groups and not for individuals. However, considering only the case reports (n=25), we recorded ten patients with ET who experienced a thrombotic event before or at diagnosis and only one had a recurrence during the follow up (10%). Among PV cases (n=10), five patients were in the same situation and two of them had a new thrombotic event (40%). Since this only represents a small group there is a risk of bias in these data. Interestingly, the overall ratio of arterial/venous events (r=0.2) demonstrates a clear predominance of venous events (84.2%). This situation was identical both before and after the diagnosis. Concerning the sites of the events, thromboses of the splanchnic territories were most frequent (75% of the venous events), with a large predominance of Budd-Chiari syndrome (62.5% of all venous events). Again, this is very different from the situation in the adult population (Online Supplementary Table S2).

Hemorrhagic events The number of hemorrhagic episodes seemed very low (1% before and 4.8% after the diagnosis in ET patients and 4% in both situations in PV patients). Importantly, the use of an antithrombotic drug did not seem to increase the risk of hemorrhage. The episodes described in the literature were mostly minor events, but their localization was usually unknown.

Transformation As one of the most significant complications of MPN, transformation into secondary myelofibrosis and/or acute myeloid leukemia is a major concern. Such transformation frequently provokes a deterioration of the Performance Status and the necessity to change treatment strategy (i.e., to use ruxolitinib, allogeneic transplantation or intensive chemotherapy). Reassuringly, transformation seemed unusual in very young patients with MPN, with only 2% of cases evolving into myelofibrosis and none transforming into acute myeloid leukemia. However, this and other information about complications during the follow-up should be interpreted carefully because of the relatively short median follow-up of the cohorts and the case reports (54 and 51.3 months for ET and PV patients, respectively).

Another malignancy and long-term sequelae of therapy Interestingly, Cario and colleagues observed that three among 36 PV patients (8.3%) were diagnosed with their MPN after having been cured from acute leukemia (2 cases) or lymphoma (1 case).9 There are only two cohorts of patients for which the occurrence of solid cancers in the young ET and PV patients was reported, with only one case of kidney cancer observed (1/97 patients, 1%). There is no information or evidence about the potential implications of previous chemotherapy or cytoreductive drugs on the occurrence of MPN or cancer, despite the possible risk

Table 2. Treatments and outcomes of very young patients with essential thrombocythemia or polycythemia vera. Antithrombotic drugs, n (%) Not treated Low dose aspirin Vitamin-K antagonists Subcutaneous heparin Cytoreductive drugs, n (%) Not-treated Hydroxycarbamide Anagrelide Interferon Venesections Ruxolitinib Busulfan/Melphalan/32P Allogeneic SCT Thrombo/Erythropheresis Complications before diagnosis, n (%) Thrombosis Hemorrhage Complications after diagnosis, n (%) Thrombosis Hemorrhage Transformation, n (%) Total Polycythemia vera Myelofibrosis Acute leukemia Death (n/%) Follow-up (months)

Essential thrombocythemia

Polycythemia vera

*203 (51.3) 104 (51.2) 88 (43.3) 11 (5.4) 7 (3.4) *239 (60.3) 112 (46.9) 31 (13) 50 (20.9) 11 (4.6) 0 0 0 0 1 (0.4) * 307 (77) 16 (4) 4 (1) * 307 (77) 15 (3.8) 19 (4.8) *264 (68.7) 7 (1.8) 0 7 (1.8) 0 0 54

*59 (78.7) 37 (62.7) 19 (32) 7 (11.9) 6 (10.2) * 62 (82.7) 6 (9.7) 16 (25.8) 0 6 (9.7) 28 (45.2) 1 (1.6) 7 (11.3) 3 (4.8) 4 (6.5) *69 (92) 11 (14.7) 3 (4) *69 (92) 7 (9.3) 3 (4) *19 (25.3) 2 (2.7) 2 (2.7) 0 3 (4) 51.3

*indicates the number and percent of available data for each category of parameters. 32P: radioactive phosphorus; SCT: stem cell transplantation

1584

haematologica | 2019; 104(8)

Very young patients with ET and PV

of these treatments modifying the genetic environment. Similarly, there are no data about the occurrence of other diseases which might be significant during the MPN, such as autoimmune, cardiac or inflammatory diseases. Surprisingly, only one study reported data on pregnancy: six young women experienced 15 pregnancies which resulted in nine healthy babies and six miscarriages (40%). This latter percentage is much higher than in the latest cohorts of young adult women with MPN, but should be interpreted with caution.69

Survival and death The mortality rate seemed low (3 cases, 0.65%): two patients died after the occurrence of Budd-Chiari syndrome despite adequate management, highlighting the fact that this event has a very high risk of morbidity and mortality, and one patient died of pneumonia that developed following a stroke. It should be noted that information about death was probably subject to reporting bias, and the median follow-up was short.

Treatments It is not clear in the published literature whether adult or pediatric staff made decisions on these patients nor what age cut-off, if any, might have been used to decide therapy. Also, it is not clear what strategy for venesection was used for these young patients. In our review, we observed a relatively high frequency of prescription of â&#x20AC;&#x153;antithromboticâ&#x20AC;? drugs (51.2% in ET and 62.7% in PV patients). The description of the treatments is provided in Table 2. The rate of prescription of aspirin was higher in ET patients than in PV patients (43.3% vs. 32%) whereas, PV patients were more frequently treated with vitamin K antagonists or low molecular weight heparin (22.1% vs. 8.8% in ET patients, respectively). These differences were probably due to the rate of thrombosis observed among PV patients (mostly Budd-Chiari syndrome). It is difficult to know whether these medications could have been responsible for the occurrence of the hemorrhages reported during the follow-up. As expected in this population, there was a very low number of high-risk patients (5.7%) based on a history of thrombosis (data on the number of patients with platelet counts over 1500x109/L were not available). Despite the low percentage of high-risk patients, most of the subjects were treated with a cytoreductive drug or related therapy (60.8%). The description of the treatments is available in Table 2. The reasons for prescribing these treatments were not explained. Most of the ET patients received what would be regarded as non-leukemogenic drugs, such as anagrelide (20.9%) or interferon (4.6%). Most of the PV patients were treated with phlebotomy (45.2% plus 6.5% erythropheresis), but a substantial proportion of them also received hydroxycarbamide (25.8%). The use of interferon seemed quite uncommon (9.7%), even for the pegylated formulation (one-third of the interferon-treated population). Interestingly, ruxolitinib (a JAK1 and JAK2 inhibitor) was prescribed in only one explicit case (three other cases were cited in a phase II trial, but without available data). Surprisingly, 16.1% of the PV patients were treated with melphalan, busulfan, radioactive phosphorus or allogeneic stem cell transplantation: most of these patients were treated in the 1980s and are represented by one cohort of patients.9 haematologica | 2019; 104(8)

Discussion We have described here the clinical and biological parameters of very young patients (aged <20 years at diagnosis of ET or PV) whose data have been published since 2005. Interestingly in this cohort of over 470 cases the clear majority (84%) had ET, which remains unexplained. Another interesting fact is that only a few cases were discovered because of the presence of symptoms (mostly headaches or migraines) or due to a thrombotic event. Importantly, the occurrence of thrombosis, hemorrhage or evolution of disease seemed quite rare. We found that a comprehensive, or total, description of all the published cases is almost impossible. As a possible bias, some cases could also have been published in large series of young patients (under 40 years old) and we were unable to extract data specifically pertinent to our age group of interest. Furthermore, almost all the information was reported in a global way, i.e. not for individual cases and, as shown, a lot of information concerning each specific data point was lacking (the amount of missing data varied from 10.7% to 74.7%), illustrating the difficulties of our approach. Concerning the diagnosis of MPN, bone marrow biopsy is generally regarded as essential, as illustrated recently by Putti and colleagues who demonstrated, in a descriptive study of biopsies from very young ET patients, that only 16 (76%) of the 21 bone marrow biopsies were compatible with a diagnosis of ET. Furthermore, they also confirmed the usefulness of this examination by identifying one case of PV, three cases of prefibrotic myelofibrosis and even one non-MPN case.19 There is no such study on bone marrow biopsies among PV patients. It should be noted that there is no evidence to tell us that the management or the outcome of very young patients with prefibrotic myelofibrosis needs to be different from that of ET patients, so this remains a matter for further evaluation. In fact, it is interesting to speculate that had more biopsies been performed, the large excess of ET diagnoses may have changed. Surprisingly, a large proportion of this cohort of young patients with MPN was found to be negative for all common driver mutations. For the patients with ET, this proportion was 57%, which is higher than that reported in adult series (between 10% to 20%).70,71 About three-quarters of the patients with PV were also negative for any JAK2 mutations, with the proportion varying between 63% and 75% depending on the cohorts.9,13,24 This raises the question of whether these were real cases of MPN. As many of them were old cases, were they misdiagnosed MPN or real MPN but with different mechanisms of proliferation than the JAK2 pathway? This suggests a potential place for next-generation sequencing in the population of patients negative for driver mutations. A new evaluation with recent cases will be useful to confirm or refute this observation. With regards to complications, we observed a low rate of both thrombosis and transformation (4.7% and 1.9%, respectively), which is in accordance with the numbers observed in the young population in general (people aged less than 40 years old) (Table 3) but, far less than in old people.5-8,67,68,72 Concerning the thrombotic risk, the classification of this population as a low- or very low-risk category seems appropriate (based on their age and history of thrombosis).2 Concerning the risk of transformation, reas1585

J-C. Ianotto et al.

Table 3. Comparison of the characteristics of the very young patients in this cohort versus those of other reported series of young patients less than 40 years old with polycythemia vera or essential thrombocythemia.

Number Median age, years Age range, years Male White blood count, 109/L Haemoglobin, g/L Platelets, 109/L JAK2V617F cases JAK2V617F allele burden Thrombosis before diagnosis Venous events Median follow-up, years Thrombosis after diagnosis Hemorrhage after diagnosis Transformation into PV Transformation into MF Transformation into AML Death

Our cohort

Essential thrombocythaemia Boddu Palandri Lussana 2018 2015 2014

Barbui 2012

Our cohort

Polycythaemia vera Boddu Lussana Passamonti 2018 2014 2003

396 9.3 0.2-20 42% 10.6 131 1192 33.5% 24.1% 4% 86% 4.5 3.8% 4.8% 0% 1.8% 0% 0%

105 25 16-39 27% 8.2 134 763 53% 15 6% ? 2.6 2% 2% ? 0% 2% 8%

178 33 13-40 38% 8.4 14 796 56% ? 13% 62% 7.6 8% 5% 0% 3% 0% 0%

75 12 0.6-19 55% 13.2 157 799 40% 43.5% 14.7% 78% 4.3 9.3% 4% na 2.7% 0% 4%

43 28 16-39 51% 9.4 149 547 92% 21% 14% ? 3.6 2% 5% na 0% 0% 0%

197 34 16-40 32% 8.6 142 850 63% 21% 8% 68% 10.3 10% ? 0% 5.5% 0.5% 2.7%

375 32,3 16-41 29% 8.6 142 708 100% ? 9% 55% 7.3 10% ? 0% 3% 0% 0.5%

97 35 18-40 44% 10.7 190 476 100% ? 12% 67% 7.9 18% ? na 9% 3% 4%

70 42 18-49 70% 12 210 544 ? ? 24% 29% 14 11% ? na 7% 7% 26%

PV: polycythemia vera; MF: myelofibrosis; AML: acute myeloid leukemia.

suringly, there were no cases of post-ET PV or acute myeloid leukemia, and the death rate was very low. However, we note a possible bias since most of the papers described particular data (e.g., thrombosis, bone marrow examinations, molecular analysis) as the main message, not reporting the parameters at diagnosis or detailed follow-up information; furthermore, we should also be cautious because of the short median follow-up of the patients included in this review (around 50 months). These factors could have led us to underestimate the rates of long-term complications. Thus, while the overall survival seems long, there are no published survival curves and the median follow-up, as discussed already, is quite short for this population of subjects who would normally be expected to live a further 60-70 years, if their life expectancy is similar to that of the general population. We noted a clear prevalence of venous thrombosis with a complete inversion of the ratio of arterial/venous thrombotic events (0.2), compared to that in cohorts of adults in whom this ratio is close to 0.67 (Online Supplementary Table S1).1 Interestingly, we observed a predominance of portal vein thrombosis and Budd-Chiari syndrome, mostly at diagnosis, a situation similar to that in the adult population in whom these thromboses are frequently associated with MPN and induce significant morbidity and mortality (2 of the 3 deaths reported in this review).9 On the other hand, 83% of the arterial events occurred during the follow-up and all these events were localized in the cerebral area. Given the age of the patients and the very low supposed rate of cardiovascular risk factors (not evaluated here as unavailable), and the different mutation profiles it will be interesting to understand the mechanism of these thromboses. Since the ECLAP study, the prescription of low-dose aspirin is highly recommended in older patients, especially with PV, to reduce the risk of thrombosis.73 The benefit of using this drug has not been proven for either ET patients 1586

or for very young patients. Furthermore, Alvarrez-Larran and colleagues have published a retrospective study on the use of antiplatelet drugs among young (defined as less than 60 years old) patients with ET, and found that CALRpositive patients experienced more hemorrhages when treated with aspirin.74 Effectively, given their metabolism, children are exposed to a higher risk of aspirin-related gastrointestinal and intracranial hemorrhages. Children below 12 years old seem particularly at risk of Reye syndrome because of the interaction between aspirin and coenzyme A reductase in mitochondria.75,76 In Reye syndrome, gastrointestinal symptoms (nausea and vomiting) are followed by progressive encephalopathy (i.e. somnolence) until coma and death due to multi-organ failure. Unfortunately, stopping the administration of aspirin does not automatically reverse the process. The incidence of Reye syndrome among very young patients with ET and PV is unknown and there have been no published cases in the past 12 years. According to international guidelines, in the adult ET and PV population, the prescription of a cytoreductive drug is limited to patients who are classified as being at high risk of thrombosis based on age (recommended if >60 years or younger with a cardiovascular risk factor) or platelet count >1500x109/L or a history of thrombosis.2 For patients belonging to the low-risk group, no other drug than aspirin is recommended. Thus, in this very young population with low rates of prior thrombosis, aspirin should often have been the only therapy. However, the proportion of patients receiving cytoreductive drugs was unexpectedly high in this population (60.8%), and the reasons were not explained. Given ongoing concerns about the safety of hydroxycarbamide it seems surprising perhaps that this drug was so frequently prescribed in this population of young patients. On the other hand, the relative innocuity of hydroxycarbamide has been proven in many cohorts of patients with haematologica | 2019; 104(8)

Very young patients with ET and PV

sickle cell disease.77 In contrast, the use of interferons was quite uncommon even though these drugs are non-leukemogenic and the first line of cytoreductive drugs according to the European LeukemiaNet recommendations. It is important to remember that all these medications, even hydroxycarbamide, are currently unlicensed in this population. Also, there are few data concerning the side effects and the long-term consequences of the use of these drugs. The European LeukemiaNet recommendations do not appear to have been followed closely in the population we analyzed, but at the same time, there are no adapted recommendations for the treatment of MPN in very young patients. This is potentially a substantial shortcoming

References 1. Tefferi A, Barbui T. Polycythemia vera and essential thrombocythemia: 2017 update on diagnosis, risk-stratification, and management. Am J Hematol. 2017;92(1): 94-108. 2. Barbui T, Barosi G, Birgegard G, et al; European LeukemiaNet. Philadelphia-negative classical myeloproliferative neoplasms: critical concepts and management recommendations from European LeukemiaNet. J Clin Oncol. 2011;29(6):761-770. 3. 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. 4. Tefferi A, Guglielmelli P, Larson DR, et al. Long-term survival and blast transformation in molecularly annotated essential thrombocythemia, polycythemia vera, and myelofibrosis. Blood. 2014;124(16):2507-2513. 5. Boddu P, Masarova L, Verstovsek S, et al. Patient characteristics and outcomes in adolescents and young adults with classical Philadelphia chromosome-negative myeloproliferative neoplasms. Ann Hematol. 2018;97(1):109-121. 6. Palandri F, Latagliata R, Polverelli N, et al. Mutations and long-term outcome of 217 young patients with essential thrombocythemia or early primary myelofibrosis. Leukemia. 2015;29(6):1344-1349. 7. Barbui T, Thiele J, Carobbio A, et al. Disease characteristics and clinical outcome in young adults with essential thrombocythemia versus early/prefibrotic primary myelofibrosis. Blood. 2012;120(3):569-571. 8. Passamonti F, Malabarba L, Orlandi E, et al. Polycythemia vera in young patients: a study on the long-term risk of thrombosis, myelofibrosis and leukemia. Haematologica. 2003;88(1):13-18. 9. Cario H, McMullin MF, Pahl HL. Clinical and hematological presentation of children and adolescents with polycythemia vera. Ann Hematol. 2009;88(8):713-719. 10. Costa G, Petit A, Ianotto JC, et al. ThrombocytĂŠmie essentielle (TE) chez lâ&#x20AC;&#x2122;enfant et le jeune adulte. Rev Oncol Hematol Ped. 2016;4(4):253-254. 11. El-Moneim AA, Kratz CP, BĂśll S, Rister M, Pahl HL, Niemeyer CM. Essential versus reactive thrombocythemia in children: retrospective analyses of 12 cases. Pediatr Blood Cancer. 2007;49(1):52-55. 12. Fu R, Liu D, Cao Z, et al. Distinct molecular

haematologica | 2019; 104(8)

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

since young people are not small adults and also because there is no real large cohort of patients in whom to assess the risks, outcomes and medications in a prospective manner. Thus, we propose a large pan-European study concerning very young patients. The aims of this study should be multiple: to gain a more accurate overview of the clinical and biological parameters concerning this population at diagnosis (symptoms and the way of making the diagnosis) and to obtain a clearer understanding of the incidence of thromboses, hemorrhages and disease progression. Such a study should also federate clinicians and biologists interested in the field in order to manage these patients better with the help of pediatricians.

abnormalities underlie unique clinical features of essential thrombocythemia in children. Leukemia. 2016;30(3):746-749. Giona F, Teofili L, Moleti ML, et al. Thrombocythemia and polycythemia in patients younger than 20 years at diagnosis: clinical and biologic features, treatment, and long-term outcome. Blood. 2012;119(10): 2219-2227. Giona F, Teofili L, Capodimonti S, et al. CALR mutations in patients with essential thrombocythemia diagnosed in childhood and adolescence. Blood. 2014;123(23):36773679. Ismael O, Shimada A, Hama A, et al. Mutations profile of polycythemia vera and essential thrombocythemia among Japanese children. Pediatr Blood Cancer. 2012;59(3): 530-535. Karow A, Nienhold R, Lundberg P, et al. Mutational profile of childhood myeloproliferative neoplasms. Leukemia. 2015;29(12): 2407-2409. Langabeer SE, Haslam K, McMahon C. CALR mutations are rare in childhood essential thrombocythemia. Pediatr Blood Cancer. 2014;61(8):1523. Nakatani T, Imamura T, Ishida H, et al. Frequency and clinical features of the JAK2 V617F mutation in pediatric patients with sporadic essential thrombocythemia. Pediatr Blood Cancer. 2008;51(6):802-805. Putti MC, Pizzi M, Bertozzi I, et al. Bone marrow histology for the diagnosis of essential thrombocythemia in children: a multicenter Italian study. Blood. 2017;129(22): 3040-3042. Randi ML, Geranio G, Bertozzi I, et al. Are all cases of paediatric essential thrombocythemia really myeloproliferative neoplasms? Analysis of a large cohort. Br J Haematol. 2015;169(4):584-589. Randi ML, Putti MC, Scapin M, et al. Pediatric patients with essential thrombocythemia are mostly polyclonal and V617FJAK2 negative. Blood. 2006;108(10): 3600-3602. Sekiya Y, Okuno Y, Muramatsu H, et al. JAK2, MPL, and CALR mutations in children with essential thrombocythemia. Int J Hematol. 2016;104(2):266-267. Teofili L, Cenci T, Martini M, et al. The mutant JAK2 allele burden in children with essential thrombocythemia. Br J Haematol. 2009;145(3):430-432. Teofili L, Giona F, Martini M, et al. The revised WHO diagnostic criteria for Ph-neg-

25.

26.

27. 28.

29.

30. 31.

32.

33.

34.

35.

ative myeloproliferative diseases are not appropriate for the diagnostic screening of childhood polycythemia vera and essential thrombocythemia. Blood. 2007;110(9):33843386. Veselovska J, Pospisilova D, Pekova S, et al. Most pediatric patients with essential thrombocythemia show hypersensitivity to erythropoietin in vitro, with rare JAK2 V617F-positive erythroid colonies. Leuk Res. 2008;32(3):369-377. Aladily TN, Mohammad RS, Al-Khader A, Awidi AS. Essential thrombocythemia in a two-year-old child, responsive to hydroxyurea but not aspirin. Oman Med J. 2017;32(3):243-246. Beatrice JM, Garanito MP. Essential thrombocythemia: a rare disease in childhood. Rev Bras Hematol Hemoter. 2013;35(4):287-289. Dua V, Yadav SP, Kumar V, Saxena R, Sachdeva A. Two cases of pediatric essential thrombocythemia managed effectively with hydroxyurea. Int J Hematol. 2012;96(6):810813. Hermeziu B, Franchi-Abella S, Plessier A, et al. Budd-Chiari syndrome and essential thrombocythemia in a child: favorable outcome after transjugular intrahepatic portosystemic shunt. J Pediatr Gastroenterol Nutr. 2008;46(3):334-337. Hwang J, Lee W. A case of essential thrombocythemia in an 8-year-old boy. Pediatr Hematol Oncol. 2008;25(4):325-330. Jaime-Perez JC, Gomez-Almaguer D. Platelet refractoriness to classical agonists in a child with essential thrombocythemia. Platelets. 2005;16(1):61-62. Jensen AW, Tefferi A, Arndt CA. Cerebral venous sinus thrombosis associated with essential thrombocytosis in a pediatric patient. J Pediatr Hematol Oncol. 2007;29(3):156-159. Kanai A, Morokawa H, Matsuda K, Shimazaki E, Nakazawa Y. Efficacy and safety of anagrelide in childhood essential thrombocythemia. Pediatr Int. 2017;59(9): 1017-1018. Khan AA, Kumar V, Anand I, Kumar M, Sharma P, Bhargava M. JAK2 mutation-negative essential thrombocythemia in a child presenting with cerebral venous thrombosis. Hematol Oncol Stem Cell Ther. 2012;5(1): 66-68. Kurosawa H, Okuya M, Matsushita T, et al. JAK2V617F mutation-positive childhood essential thrombocythemia associated with cerebral venous sinus thrombosis. J Pediatr

1587

J-C. Ianotto et al. Hematol Oncol. 2009;31(9):678-680. 36. Lackner H, Urban C, Benesch M, et al. Longterm use of anagrelide in the treatment of children with essential thrombocythemia. Eur J Haematol. 2006;77(4):358-359. 37. Langabeer SE, Haslam K, McMahon C. A prenatal origin of childhood essential thrombocythemia. Br J Haematol. 2013;163(5): 676-678. 38. Bertrand A, Heissat S, Caron N, et al. Deep vein thrombosis revealing myeloproliferative neoplasm in two adolescents. Arch Ped. 2014;21(5):497-500. 39. Pulini S, D'Amico E, Basilico R, et al. Portal venous thrombosis in a young patient with idiopathic myelofibrosis and intrahepatic extramedullary hematopoiesis: a difficult diagnosis, prognosis and management. Leukemia. 2007;21(11):2373-2375. 40. Robins EB, Niazi M. Essential thrombocythemia in a child with elevated thrombopoietin concentrations and skeletal anomalies. Pediatr Blood Cancer. 2008;50(4):859861. 41. Roy NB, Treacy M, Kench P. Childhood essential thrombocythemia. Br J Haematol. 2005;129(5):567. 42. Scherer S, Ferrari R, Rister M. Treatment of essential thrombocythemia in childhood. Pediatr Hematol Oncol. 2003;20(5):361-365. 43. Singh A, Nityanand S, Sonker A, Kumar S. Successful use of the cell separator Hemonetics multicomponent collection system+ for therapeutic thrombocytapheresis in a low body weight child of essential thrombocythemia. Asian J Transfus Sci. 2015;9(2):207-209. 44. Slone JS, Smith MC, Seegmiller AC, Sidonio RF, Yang E. Idiopathic myelofibrosis in children: primary myelofibrosis, essential thrombocythemia, or transient process? J Pediatr Hematol Oncol. 2013;35(7):559-565. 45. Szegedi I, Benko I, Mero G, Prinzinger A, Kappelmayer J, Kiss C. Long-lasting partial remission by interferon-alpha treatment in a child with essential thrombocythemia. Pediatr Blood Cancer. 2007;49(5):744-747. 46. Tokgoz H, Caliskan U, Yüksekkaya HA, Kucukkaya R. Essential thrombocythemia with Mpl W515K mutation in a child presenting with Budd-Chiari syndrome. Platelets. 2015;26(8):805-808. 47. Vafaie M, Jaseb K, Ghanavat M, Pedram M, Rahiminia T. Asymtomatic [sic] essential thrombocythemia in a child: a rare case report. Int J Hematol Oncol Stem Cell Res. 2013;7(2):35-37. 48. Wigton JC, Tersak JM. JAK2+ Essential Thrombocythemia in a young girl with Budd-Chiari syndrome: diagnostic and therapeutic considerations when adult disease strikes the young. J Pediatr Hematol Oncol.

1588

2016;38(1):70-73. 49. Benedik-Dolni ar M, Homan M, Brecelj J. JAK2V617F mutation in a 9-year-old girl with polycythemia vera and Budd-Chiari syndrome: a case report. J Pediatr Hematol Oncol. 2012;34(3):243-244. 50. Berman JN, Greer WL, Archambeault S, et al. JAK2 V617F positive polycythemia vera in a child with neurofibromatosis type I. Pediatr Blood Cancer. 2008;51(5):689-691. 51. Braconnier O, Monpoux F, Affanetti M, Bérard E, Sirvent N. [Polycythemia vera and JAK-2 mutation in childhood: a case report]. Arch Pediatr. 2007;14(10):1202-1205. 52. Coskun ME, Height S, Dhawan A, Hadzic N. Ruxolitinib treatment in an infant with JAK2+ polycythemia vera-associated BuddChiari syndrome. BMJ Case Rep. 2017;2017. 53. Goldstein G, Maor J, Kleinbaum Y, Palumbo M, Sidi Y, Salomon O. Budd-Chiari syndrome in very young adult patients with polycythemia vera: report of case series with good outcome with direct thrombin inhibitor treatment. Blood Coagul Fibrinolysis. 2013;24(8):848-853. 54. Guo HX, Chan GC, Chiang AK, Ho MH, Chan EY, Ha SY. Dural sinus thrombosis owing to polycythemia vera in a 12-year-old girl. Paediatr Int Child Health. 2012;32 (3):167-170. 56. Park MJ, Shimada A, Asada H, Koike K, Tsuchida M, Hayashi Y. JAK2 mutation in a boy with polycythemia vera, but not in other pediatric hematologic disorders. Leukemia. 2006;20(8):1453-1454. 57. Osgood EE. Polycythemia vera: age relationships and survival. Blood. 1965;26:243-256. 58. Reid MM, Saunders PW, Kernahan J. Myeloproliferative disease in children: a demographic study. J Clin Pathol. 1988;41 (8):883-885. 59. McNally RJ, Rowland D, Roman E, Cartwright RA. Age and sex distributions of hematological malignancies in the U.K. Hematol Oncol. 1997;15(4):173-189. 60. Hasle H. Incidence of essential thrombocythemia in children. Br J Haematol. 2000; 110(3):751. 61. Visser O, Trama A, Maynadié M, et al; RARECARE Working Group. Incidence, survival and prevalence of myeloid malignancies in Europe. Eur J Cancer. 2012;48(17): 3257-3266. 62. Horibe K, Saito AM, Takimoto T, et al. Incidence and survival rates of hematological malignancies in Japanese children and adolescents (2006-2010): based on registry data from the Japanese Society of Pediatric Hematology. Int J Hematol. 2013;98(1):74-88. 63. Titmarsh GJ, Duncombe AS, McMullin MF, et al. How common are myeloproliferative neoplasms? A systematic review and meta-

analysis. Am J Hematol. 2014;89(6):581-587. 64. Rumi E, Cazzola M. Diagnosis, risk stratification, and response evaluation in classical myeloproliferative neoplasms. Blood. 2017;129(6):680-692. 65. Vannucchi AM, Lasho TL, Guglielmelli P, et al. Mutations and prognosis in primary myelofibrosis. Leukemia. 2013;27(9):18611869. 66. Tefferi A, Lasho TL, Guglielmelli P, et al. Targeted deep sequencing in polycythemia vera and essential thrombocythemia. Blood Adv. 2016;1(1):21-30. 67. Tefferi A, Vannucchi AM, Barbui T. Polycythemia vera treatment algorithm 2018. Blood Cancer J. 2018;8(1):3. 68. Tefferi A, Vannucchi AM, Barbui T. Essential thrombocythemia treatment algorithm 2018. Blood Cancer J. 2018;8(1):2. 69. Alimam S, Bewley S, Chappell LC, et al. Pregnancy outcomes in myeloproliferative neoplasms: UK prospective cohort study. Br J Haematol. 2016;175(1):31-36. 70. Klampfl T, Gisslinger H, Harutyunyan AS, et al. Somatic mutations of calreticulin in myeloproliferative neoplasms. N Engl J Med. 2013;369(25):2379-2390. 71. Nangalia J, Massie CE, Baxter EJ, et al. Somatic CALR mutations in myeloproliferative neoplasms with nonmutated JAK2. N Engl J Med. 2013;369(25):2391-2405. 72. Lussana F, Carobbio A, Randi ML, et al. A lower intensity of treatment may underlie the increased risk of thrombosis in young patients with masked polycythemia vera. Br J Haematol. 2014;167(4):541-546. 73. Landolfi R, Marchioli R, Kutti J, et al; European Collaboration on Low-Dose Aspirin in Polycythemia Vera Investigators. Efficacy and safety of low-dose aspirin in polycythemia vera. N Engl J Med. 2004;350 (2):114-124. 74. Alvarez-Larrán A, Pereira A, Guglielmelli P, et al. Antiplatelet therapy versus observation in low-risk essential thrombocythemia with a CALR mutation. Haematologica. 2016;101 (8):926-931. 75. Belay ED, Bresee JS, Holman RC, Khan AS, Shahriari A, Schonberger LB. Reye's syndrome in the United States from 1981 through 1997. N Engl J Med. 1999;340 (18):1377-1382. 76. Bennett MJ, Russell LK, Tokunaga C, et al. Reye-like syndrome resulting from novel missense mutations in mitochondrial medium- and short-chain l-3-hydroxy-acyl-CoA dehydrogenase. Mol Genet Metab. 2006;89(1-2):74-79. 77. McGann PT, Ware RE. Hydroxyurea for sickle cell anemia: what have we learned and what questions still remain? Curr Opin Hematol. 2011;18(3):158-165.

haematologica | 2019; 104(8)

ARTICLE

Chronic Myeloid Leukemia

Observational study of chronic myeloid leukemia Italian patients who discontinued tyrosine kinase inhibitors in clinical practice

Carmen Fava,1 Giovanna Rege-Cambrin,1 Irene Dogliotti,1 Marco Cerrano,2 Paola Berchialla,1 Matteo Dragani,1 Gianantonio Rosti,3 Fausto Castagnetti,3 Gabriele Gugliotta,3 Bruno Martino,4 Carlo Gambacorti-Passerini,5 Elisabetta Abruzzese,6 Chiara Elena,7 Patrizia Pregno,8 Antonella Gozzini,9 Isabella Capodanno,10 Micaela Bergamaschi,11 Monica Crugnola,12 Monica Bocchia,13 Sara Galimberti,14 Davide Rapezzi,15 Alessandra Iurlo,16 Daniele Cattaneo,16 Roberto Latagliata,17 Massimo Breccia,17 Michele Cedrone,18 Marco Santoro,19 Mario Annunziata,20 Luciano Levato,21 Fabio Stagno,22 Francesco Cavazzini,23 Nicola Sgherza,24 Valentina Giai,25 Luigia Luciano,26 Sabina Russo,27 Pellegrino Musto,28 Giovanni Caocci,29 Federica Sorà,30 Francesco Iuliano,31 Francesca Lunghi,32 Giorgina Specchia,33 Fabrizio Pane,26 Dario Ferrero,2 Michele Baccarani3 and Giuseppe Saglio1

Department of Clinical and Biological Sciences, University of Turin, Orbassano; Hematology Division, Department of Molecular Biotechnologies and Health Sciences, University of Turin, Turin; 3Institute of Hematology "L. & A. Seràgnoli", St. Orsola University Hospital, Bologna; 4Azienda Ospedaliera "Bianchi Melacrino Morelli", Reggio Calabria; 5 University Milano Bicocca, San Gerardo Hospital, Monza; 6Haematology Unit, S. Eugenio Hospital, Rome; 7Hematology Hunit, Fondazione IRCCS Policlinico San Matteo, Pavia; 8 A.O. Città della Salute e della Scienza di Torino, Turin; 9SC Terapie Cellulari e Medicina Trasfusionale, AOU Careggi, Florence; 10Hematology, Azienda Unità Sanitaria Locale IRCCS, Reggio Emilia; 11Division of Hematology 1, IRCCS AOU San Martino-IST, Genoa; 12 Division of Hematology, University Hospital of Parma, Parma; 13Azienda Ospedaliera Universitaria, University of Siena, Siena; 14Hematology Department, University of Pisa, Pisa; 15S.C. Ematologia, ASO S. Croce e Carle, Cuneo; 16Haematology Division, Foundation IRCCS Ca' Granda Ospedale Maggiore Policlinico, Milan; 17Department of Cellular Biotechnologies and Hematology, University La Sapienza, Rome; 18UOC of Hematology, San Giovanni - Addolorata Hospital, Rome; 19Hematology Unit, University of Palermo, Palermo; 20Division of Hematology, Ospedale Cardarelli, Naples; 21Department Hematology-Oncology, Azienda Ospedaliera Pugliese-Ciaccio, Catanzaro; 22Chair and Hematology Section, Ferrarotto Hospital, Catania; 23Department of Medical Sciences Haematology and Physiopathology of Haemostasis Section, Ferrara; 24Division of Hematology, IRCCS Ospedale Casa Sollievo Sofferenza, San Giovanni Rotondo; 25Division of Haematology, SS Antonio e Biagio e Cesare Arrigo Hospital, Alessandria; 26Division of Hematology - Departments of Clinical Medicine and Surgery, University of Naples Federico II, Naples; 27Department of Internal Medicine, AOU Policlinico di Messina, Messina; 28 IRCCS, Centro Di Riferimento Oncologico Della Basilicata, Rionero in Vulture; 29 Department of Medical Sciences, University of Cagliari, Cagliari; 30Hematology Department, University Cattolica del Sacro Cuore - Policlinico A. Gemelli, Rome; 31Presidio Ospedaliero N. Giannetasio - Azienda ASL 3, Rossano; 32Division of Haematology and Bone Marrow Transplant, Ospedale San Raffaele IRCCS, Milan and 33Division of Haematology with Transplant – Outpatients, Azienda Ospedaliero-Universitaria Policlinico Consorziale di Bari, Bari, Italy

Ferrata Storti Foundation

Haematologica 2019 Volume 104(8):1589-1596

1 2

ABSTRACT

t is judged safe to discontinue treatment with tyrosine kinase inhibitors (TKI) for chronic myeloid leukemia (CML) in experimental trials on treatment-free remission (TFR). We collected a total of 293 Italian patients with chronic phase CML who discontinued TKI in deep molecular response. Seventy-two percent of patients were on treatment with imatinib, and 28% with second generation TKI at the time of discontinuation. Median duration of treatment with the last TKI was 77 months [Interquartile Range (IQR) 54;111], median duration of deep molecular response was 46 months (IQR 31;74). Duration of treatment with TKI and duration of deep molecular response were shorter with second generation TKI than with imatinib (P<0.001). Eighty-eight percent of patients discontinued as per clinical practice, and reasons for stopping treatment were: toxicity (20%), pregnancy (6%), and shared decision between treating physician and patient (62%). After a median follow up of 34 months (range, 12haematologica | 2019; 104(8)

Correspondence: CARMEN FAVA carmen.fava@unito.it Received: October 8, 2018. Accepted: February 27, 2019. Pre-published: February 28, 2019. doi:10.3324/haematol.2018.205054 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1589 ©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.

1589

C. Fava et al.

161) overall estimated TFR was 62% (95%CI: 56;68). At 12 months, TFR was 68% (95%CI: 62;74) for imatinib, 73% (95%CI: 64;83) for second generation TKI. Overall median time to restart treatment was six months (IQR 4;11). No progressions occurred. Although our study has the limitation of a retrospective study, our experience within the Italian population confirms that discontinuation of imatinib and second generation TKI is feasible and safe in clinical practice.

Introduction Chronic myeloid leukemia (CML) patients have reached a near-normal life expectancy thanks to tyrosine kinase inhibitors (TKI).1,2 These drugs, however, can cause several persistent low-grade side effects that affect quality of life, and can be associated in the long term with severe toxicities.3 For this lifelong disease, tolerance and adherence to treatment are an issue. Furthermore, thanks to the success of the therapies, patients grow older and accumulate comorbidities that require concomitant treatments that can possibly interfer with TKI. Younger patients have other problems because living with TKI interferes with family planning, availability for the job market, life insurance, and so on.4 Besides, as more and more patients are living with their disease, high treatment costs are becoming an important issue.5 Over recent years, several papers have reported on treatment discontinuation in CML patients in persistent deep molecular response (DMR).6-25 The majority of these studies have reported on patients who had achieved a DMR with imatinib. There are fewer data reported on the discontinuation of second generation TKI, and with a shorter follow up. The definitions of DMR and the criteria for treatment discontinuation, for molecular relapse, and for treatment resumption, varied among these studies. Therefore, the reported treatment-free remission (TFR) rate ranged widely (between 30% and 70%), with the first reports mostly showing a TFR rate of approximately 40%; more recent reports, which adopted less stringent criteria for treatment discontinuation and therapy resumption, showed a TFR rate of approximately 60%. Partly due to these different definitions, it is still difficult to identify the factors that may predict for the TFR rate, although some analyses have drawn attention to the predictive value of treatment duration, Sokal score, duration of molecular response (MR), and response to first-line TKI treatment. Prior studies were mostly academic or company-sponsored; these were mostly prospective in nature, with restricted and carefully selected inclusion criteria. Nowadays, doctors and patients are willing and ready to introduce TKI discontinuation in clinical practice. Very few data are available on the effects and the outcome of treatment discontinuation outside prospective studies and without a central control of MR. We report here on 293 adult patients who discontinued TKI outside studies, as per clinical practice.

Methods Study design and purpose We designed a retrospective observational study of Italian patients with Philadelphia positive (Ph+) CML in chronic phase who had discontinued TKI treatment in DMR, with a follow up after discontinuation over one year. All hematology centers 1590

belonging to the Italian Group for the Hematologic Diseases of the Adults (GIMEMA) were invited to participate; thirty-two centers contributed to this study. The primary end point of the study was the TFR rate after one year from TKI treatment discontinuation. Secondary end points included: longer-term TFR status, safety (including the outcome after treatment resumption and disease progression), identifying factors associated with MR. Data on the main disease characteristics were collected for each patient. These were: all treatments before and after discontinuation, duration of each treatment, response to each treatment, and the reasons for discontinuation. The cutoff date for this analysis was February 2017. The observational retrospective study protocol was approved by the ethics committees of all centers taking part.

Response definitions and statistical analysis Molecular response was assessed by quantitative polymerase chain reaction (qPCR) according to the standard methodology;26 all analyses were performed by the GIMEMA Laboratories Network (LabNet) for CML, expressed according to the International Scale. Major molecular response (MMR) was defined as a BCR-ABL1 ratio ≤0.1 with at least 10,000 ABL1 copies. Deep molecular response was defined as MR4 (BCR-ABL1 ratio ≤0.01% with at least 10,000 ABL1 copies), or MR4.5 (BCR-ABL1 ratio ≤0.0032% with at least 32,000 ABL1 copies), or MR5 (BCR-ABL1 ratio ≤0.001% with at least 100,000 ABL1 copies) confirmed at least three times before TKI discontinuation.26 In a few patients who discontinued TKI before the establishment of molecular standardization, DMR was defined as a level of BCR-ABL1 transcript undetectable by qPCR or by qualitative PCR, confirmed in at least two controls. The cytogenetic response was assessed according to European LeukemiaNet (ELN) criteria.27 Treatment-free response was assessed using the Kaplan-Meier method, from the date of TKI discontinuation to the date documenting the restart of therapy regardless of the reason. In fact, since this is a retrospective study, criteria for treatment resumption have not been pre-established. TFR was estimated using a KaplanMeier curve and 95% confidence interval (CI). Deaths were considered as censored events. For all the other patients, data were censored at the date of last qPCR. Continuous data were expressed as medians with interquartile ranges (IQR, i.e. 25th and 75th percentiles) as a measure of variability. A Mann-Whitney U test was used for comparison of quantitative variables and χ2 or Fisher exact test was used for categorical variables as appropriate. Clinical and biological variables at baseline were assessed as potential independent prognostic factors for MR by univariate analysis using Cox regression model. Variables were entered without any transformation or cut off. For the multivariate analysis, a stepwise backward selection procedure was carried out.28 The non-linear effect of continuous covariates was modeled using a restrictive cubic spline function, and its significance was assessed using the Wald test; similar methods were used to check interactions.29 The best fitting model was chosen according to the Akaike information criterion. P=0.05 was considered statistically significant. All analyses were carried out using R v.3.3.3 statistical software.30 haematologica | 2019; 104(8)

TKI discontinuation in CML clinical practice

Results Patients We collected data on 293 patients who discontinued TKI between June 2003 and February 2016. Overall, 34 of 293 patients (11.5%) suspended treatment because they were enrolled in the prospective interventional Imatinib Suspension and Validation (ISAV) study.13 All the other patients discontinued as per clinical practice, and the rea-

sons were: toxicity (20%, 58 of 293), pregnancy (6%, 17 of 293), and a shared decision between the treating physician and the patient (62%, 182 of 293). Finally, one patient discontinued the TKI because of chemotherapy for another neoplasia. Reason of discontinuation was not known for one patient. Patientsâ&#x20AC;&#x2122; characteristics are reported in Table 1. Median age was 49 years (IQR 38-60) at diagnosis and 59 years (IQR 48-70) at discontinuation. At the time of discontinu-

Table 1. Patientsâ&#x20AC;&#x2122; baseline characteristics. N Age at diagnosis (median [IQR]) Age at discontinuation (median [IQR]) Sex Males (%) Sokal Score n=263 (%) Low Intermediate High Type of transcript n=252 (%) b2a2 b2a3 b3a2 b3a3 e1a2 Last TKI (%) n=293 Imatinib Nilotinib Dasatinib Bosutinib Line of treatment at discontinuation (%) n=293 1st line 2nd line 3rd line 4th line Reasons for discontinuation (%) n=292 Shared decision Toxicity ISAV13 Pregnancy Chemotherapy for 2nd tumor MR at discontinuation (%) n=290 MR4 MR4.5 MR5 Transcript undetectable Duration of last TKI (median [IQR]) Duration of treatment with any TKI (median [IQR]) Duration of total treatment (median [IQR]) Time to DMR (median [IQR]) Duration of DMR (median [IQR])

Imatinib

2nd generation TKI

Overall

211 47 [36, 58] 58 [46,67]

82 55 [45, 67] 63 [51, 74]

293 49 [38, 60] 59 [48, 70]

117 (56)

44 (54)

161 (55)

114 (61) 52 (28) 20 (11)

40 (52) 28 (36) 9 (12)

154 (59) 80 (30) 29 (11)

42 (23) 0 (0) 141 (76.5) 1 (0.5) 0 (0)

20 (29) 1 (1.5) 46 (68) 0 (0) 1 (1.5)

62 (24.5) 1 (0.5) 187 (74) 1 (0.5) 1 (0.5)

211 (100) 0 (0) 0 (0) 0 (0)

0 (0) 58 (71) 23 (28) 1 (1)

211 (72) 58 (19.5) 23 (8) 1 (0.5)

129 (61) 81 (38.5) 1 (0.5) 0 (0)

33 (40) 36 (44) 12 (15) 1 (1)

162 (55) 117 (40) 13 (4.5) 1 (0.5)

135 (64) 28 (13.5) 34 (16) 12 (6) 1 (0.5)

47 (57) 30 (37) 0 (0) 5 (6) 0 (0)

182 (62) 58 (20) 34 (11.5) 17 (6) 1 (0.5)

70 (33) 61 (29) 41 (20) 37 (18) 96 [62, 120] 96 [62, 120] 104 [73, 142] 24 [12, 52] 53 [33, 82]

31 (38) 29 (36) 12 (15) 9 (11) 50 [32, 66] 73 [51, 98] 76 [52, 109] 13 [6, 26] 36 [25, 46]

101 (35) 90 (31) 53 (18) 46 (16) 77 [54, 111] 87 [59, 117] 98 [65, 133] 21 [10, 42] 46 [30, 73]

P 0.001 0.023 0.884 0.346

0.126

<0.001

0.315

<0.001 0.002 <0.001 <0.001 <0.001

IQR: interquartile ranges; TKI: tyrosine kinase inhibitor; MR: molecular response; DMR: deep molecular response.

haematologica | 2019; 104(8)

1591

C. Fava et al.

ation, 211 patients (72%) were on treatment with imatinib and 82 patients (28%) with either nilotinib (n=58), dasatinib (n=23), or bosutinib (n=1). There were no differences in age, sex, Sokal score and type of transcript between imatinib and second generation TKI. One hundred and sixty-two patients (55%) discontinued in first line, 117 patients (40%) in second line, 13 patients (4.5%) in third line, and one patient in fourth line. Among those who discontinued imatinib, 73 patients (35%) had been pre-treated with Îą-interferon (IFN) and seven patients had been submitted to allogeneic stem cell transplantation. Median duration of treatment with any TKI was 87 months (IQR 59-117) for all patients, 96 months (IQR 62120) for imatinib patients, and 73 (IQR 51-98) months for second generation TKI patients (P=0.002). Median duration of treatment with the last TKI was 77 months (IQR 54-111) for all patients, and 50 months (IQR 32; 66) for second generation TKI patients. Median duration of DMR was 46 months (IQR 30-73) for all patients, 53 months (IQR 33-82) for imatinib patients, and 36 months (IQR 2546) for second generation TKI patients (P<0.001). Overall, all patients but one had an optimal early response to last treatment. At three months of last TKI, 34% of patients were in MMR, 40% were in CCyR and/or had a transcript â&#x2030;¤1%, and 25% were in PCyR and/or had a transcript â&#x2030;¤10%. At treatment discontinuation the response was as follows: undetectable transcript in 16% of patients, MR4 in 35% of patients, MR4.5 in 31% of patients, and MR5 in 18%. There was no difference in the grade of molecular response at discontinuation between patients on imatinib and patients on second generation TKI (P=0.315).

Relapses and treatment-free remission At 12 months, the estimated TFR was 69% (95%CI: 6475) for all patients (Figure 1A), 68% (95%CI: 62-74) for imatinib patients (Figure 1B), 73% (95%CI: 64-83) for second generation TKI patients (Figure 1C). Median follow up was 34 months (IQR 24-53) for all patients, 42 months (IQR 26-56) for imatinib patients, and 26 months (IQR 21-34) for second generation TKI patients. At median follow up, TFR was 62% (95%CI: 5668) for all patients (at 34 months), 60% (95%CI: 54-67) for imatinib patients (at 42 months), 67% (95%CI: 57-78) for second generation TKI patients (at 26 months) (Figure 1). There was no significant difference in TFR between patients who had discontinued imatinib first-line versus imatinib after IFN versus further lines (P=0.35), and there was no difference in TFR between patients who discontinued second generation TKI frontline (n=33) versus secondline for intolerance (n=30) versus second-line for resistance (n=16) (P=0.16). Overall, 114 patients (39%) resumed treatment. Reasons for resuming were: loss of MR4 (19%), loss of MMR (70%), loss of CCyR (9%), other (2%). The reasons for restarting imatinib and second generation TKI were similar (P=0.13). Overall median time to restart treatment was six months (IQR 4-11). Although 75% of patients had restarted treatment by the end of the first year, the last treatment resumption was after 105 months of TFR. Median time to loss of MR4 was three months (IQR 2-7); median time to loss of MMR was four months (IQR 3-7), and median time to loss of CCyR was five months (IQR 4-6). Median time from loss of response to restarting treatment was one month (IQR 0-2).

Figure 1. Kaplan-Meier curves for Italian patients who discontinued tyrosine kinase inhibitor (TKI). (A) Overall population. (B) Patients who discontinued imatinib. (C) Patients who discontinued second generation TKI. Estimated treatment-free remission (TFR) is reported at 12 months for the overall population; at 12, 26 (median follow up for patients who discontinued second generation TKI), and 42 (median follow up for patients who discontinued imatinib) months for imatinib; at 12 and 26 months (median follow up for patients who discontinued second generation TKI) for second generation TKI. N: number.

1592

haematologica | 2019; 104(8)

TKI discontinuation in CML clinical practice

No progressions occurred. Nine deaths were reported but none of them was disease related. The patients who resumed therapy (Table 2) were given imatinib (n=77), nilotinib (n=22), dasatinib (n=9), bosutinib (n=3), ponatinib (n=1), or IFN (n=2). Most of the patients who stopped imatinib restarted imatinib after relapse, and patients who were on second generation TKI mainly stayed with second generation TKI. Ninety-four percent of the patients who were retreated achieved at least another MMR, and 82% of them achieved another DMR, fitting the criteria for a second attempt at discontinuation.31 In 20 patients who had lost MR4, and in four patients who had lost MMR, the treatment was not resumed fol-

lowing a shared decision with the doctor. Interestingly, they were still on the same response after a median time of 12 months (IQR 1-32).

Prognostic factors Univariate analysis – Univariate analysis was used to assess age (considered as continuous variable), sex (female vs. male), Sokal score (intermediate vs. low; high vs. low), type of therapy (second generation TKI vs. imatinib), line of therapy at stop (imatinib vs. imatinib post IFN; first-line second generation vs. second generation in second or further lines), type of transcript (b2a2 vs. others), duration of therapy with the last TKI and any TKI (continuous variables), duration of total treatment (continuous

Table 2. Type of retreatment after failure of discontinuation.

Type of retreatment (%) Imatinib Nilotinib Dasatinib Bosutinib Ponatinib IFNα

Overall (n=114)

2nd generation TKI (n=26)

Imatinib (n=88)

77 (67) 22 (19) 9 (8) 3 (3) 1 (1) 2 (2)

2 (8) 18 (69) 4 (15) 1 (4) 1 (4) 0 (0)

75 (85) 4 (5) 5 (6) 2 (2) 0 (0) 2 (2)

Table 3. Hazard Ratios (HRs) computed at univariate analysis.

HR Sex female vs. male Sokal score Intermediate vs. low high vs. low Type of therapy 2nd generation vs. imatinib Type of transcript b2a2 vs. others Age at discontinuation (older vs. younger; diff. of 22 ys) Duration of DMR* (diff. of 43 mos) Time to DMR before stop* (32 mos increase) Duration of therapy with last TKI* (57 mos increase) Duration of treatment with any TKIs* (58 mos increase) Duration of total treatment* (68 mos increase) Depth of MR at stop MR4.5 vs.MR4 MR5 vs. MR4 Undetectable vs. MR4 Line of therapy at stop 1st line vs. ≥ 2nd line Reason for discontinuation Pregnancy vs. shared decision with MD ISAV vs. shared decision with MD Toxicity vs.shared decision with MD

95%CI

1.17

0.81

1.69

0.41

0.74 1.66

0.47 0.98

1.17 2.81

0.19 0.06

0.8

0.52

1.23

0.31

0.93 0.76 1.01 0.97 1.04 0.85 0.79

0.58 0.58 0.77 0.75 0.80 0.64 0.62

1.49 0.98 1.31 1.27 1.37 1.13 1.02

0.77 0.04 0.97 0.84 0.73 0.27 0.07

0.67 0.68 0.7

0.42 0.4 0.4

1.07 1.14 1.19

0.1 0.14 0.18

1.53

1.04

2.24

0.03

1.57 1.40 0.73

0.81 0.82 0.43

3.05 2.38 1.21

0.18 0.22 0.22

*For each variable the difference of months between groups of patients considered for computing HR corresponds to the Interquartile Range (IQR); ys: years; DMR: deep molecular response; mos: months; MD: medical doctor.

haematologica | 2019; 104(8)

1593

C. Fava et al.

2nd generation

imatinib

Figure 2. Tyrosine kinase inhibitor (TKI)-treatmentfree remission (TFR) curves adjusted for age at discontinuation, Sokal score, line of therapy, and duration of disease.

Table 4. Median and Interquartile Range of duration of treatment in patients who discontinued treatment in first line or in second or further lines of therapy.

Lines of treatment at discontinuation 1st Line â&#x2030;Ľ2nd Line

Duration of total treatment [median (IQR)]

82 (60; 105) 128 (86;169)

<0.001

IQR: Interquartile Ranges.

Table 5. Multivariate Cox regression analysis for restarting therapy. Figures reported are Hazard Ratios and 95% confidence intervals.

HR Age at discontinuation (10 yrs difference) Sokal score Intermediate vs. low High vs. low Line of therapy: 2nd vs. 1st line 2nd generation TKIs vs. imatinib Duration of total therapy (one yr increase) in patients treated with imatinib* Duration of total therapy (one yr increase) in patients treated with 2nd generation TKIs**

95%CI

0.84

0.73

0.97

0.02

0.92 2.07 0.80 0.43 1.00 0.78

0.54 1.16 0.50 0.20 0.94 0.65

1.57 3.71 1.30 0.91 1.07 0.93

0.76 0.01 0.37 0.03 0.90 0.01

*HR =1 expresses no risk increase associated to the increase of 1 year of the duration of therapy in patients treated with imatinib. ** HR < 1 expresses the risk reduction associated to the increase of 1 year of the duration of therapy in patients treated with 2nd generation tyrosine kinase inhibitor (TKI); yr: years; HR: Hazard Ratios; CI: Confidence Intervals.

variable), time to DMR and DMR duration (continuous variables), depth of MR at stop (MR4.5 vs. MR4, MR5 vs. MR4 and Undetectable vs. MR4), reasons for discontinuation (pregnancy, ISAV study and toxicity vs. shared decision with medical doctor). The only statistically significant risk factors that affected TFR were age at discontinuation (with a higher risk for younger patients) and line of treatment (Table 3). When we assessed the duration of total treatment for patients who discontinued TKI in front line versus second line, we observed that patients who discontinued treatment front line had a significantly shorter duration of treatment (P<0.001) (Table 4). Multivariate analysis - The line of treatment lost statistical significance in a multivariate analysis including age at discontinuation, Sokal score, duration of total treatment, line of treatment, and type of TKI at discontinuation (Table 5). Patients treated with second generation TKI showed a better TFR (HR 0.43; 95%CI: 0.20-0.91) (Table 5 and Figure 2). Duration of total treatment was positively 1594

associated with TFR among patients treated with second generation TKI with a 22% risk reduction for one additional year of treatment (HR: 0.78; 95%CI: 0.65-0.93).

Discussion Although at present no guidelines explicitly recommend treatment discontinuation, this study showed that many physicians have already experienced TKI cessation in their clinical practice because of intolerance, toxicity, and patient desire to stop the treatment. This multi-center observational study has confirmed that treatment cessation was safe as no progression occurred and the overall TFR was 69% at 12 months, consistent with data reported in previous studies.6-25 After discontinuation, patients were monitored with the same frequency as in the EURO-SKI study: most of the patients had a molecular evaluation every month for the first six months, every six weeks for haematologica | 2019; 104(8)

TKI discontinuation in CML clinical practice

the subsequent six months, and then every three months.21 Although we may think that a stringent monitoring is protective, and indeed most of the relapses occurred during the first year, late relapses were not complicated by loss of complete hematologic remission or progression to advanced phases, even if monitoring was less frequent.32 Given this, we must mention that Italian centers rely on the Lab-net CML network, which ensures a standardized measurement of minimal residual disease, with a short turn-around time between sampling and reporting. The history of CML has been revolutionized by the introduction of imatinib, and while this has resulted in an extraordinary improvement in survival, second generation TKI have refined our concept of CML. The achievement of higher rates of DMR in shorter periods of time switched the goal of CML treatment from survival to cure, to the point that TFR was included in the data sheet of nilotinib.33 However, for the moment, a definitive treatment discontinuation is not yet an option for everybody. All the studies have tried to define prognostic factors for a successful TFR in order to increase the number of patients who can experience a successful discontinuation. In our study, having a high Sokal risk score at diagnosis was predictive for a worse outcome, in agreement with the STIM and the Korean studies.7,16 As in the ISAV trial,13 we showed that age might have a role in the maintenance of response, with an advantage for older patients. We retrospectively observed that our population was almost entirely characterized by an optimal early response at three months; this could explain why TFR was comparable when discontinuation occurred in a first-line setting or during subsequent lines of therapy. Duration of treatment was reported as a prognostic factor in many studies.7,15,16,21 In our analysis, the duration of total treatment for patients who discontinued TKI in second line was significantly longer compared to patients who discontinued TKI in front-line (128 vs. 82 months). This could possibly account for the lower risk of relapses in patients who discontinued TKI in second line as shown in the univariate analysis. In fact, in the multivariate analyses, the line of treatment lost significance. In our study, the total duration of treatment had a positive influence particularly on patients treated with second generation TKI: we observed a 22% reduction of the risk of resuming therapy per year of treatment.

References 1. Hochhaus A, Larson RA, Guilhot F, et al. Long-Term Outcomes of Imatinib Treatment for Chronic Myeloid Leukemia. N Engl J Med. 2017;376(10):917-927. 2. Sasaki K, Strom SS, Oâ&#x20AC;&#x2122;Brien S, et al. Relative survival in patients with chronicphase chronic myeloid leukaemia in the tyrosine-kinase inhibitor era: analysis of patient data from six prospective clinical trials. Lancet Haematol. 2015;2(5):e186193. 3. Steegmann JL, Baccarani M, Breccia M, et al. European LeukemiaNet recommendations for the management and avoidance of adverse events of treatment in chronic myeloid leukaemia. Leukemia. 2016; 30(8):1648-1671.

haematologica | 2019; 104(8)

In this study, we observed that patients who discontinued second generation TKI had a median duration of treatment with the last TKI of 50 months versus 96 months of treatment with imatinib (Table 1). The results are in line with those of several prospective studies, such as the ENEST Freedom, the ENEStop (median duration of treatment with nilotinib of 43 months and 53 months, respectively), and the EURO-SKI trials (median duration of treatment with imatinib of 91 months).20,21,25 Furthermore, the multivariate Cox proportional hazards regression model showed a better probability of TFR for patients treated with second generation TKI, with an estimated 57% relative risk reduction in favor of the second generation TKI. Even considering the quite large confidence interval, the minimum risk reduction is still 9%. These data are in keeping with the superiority of second generation TKI in deeply and rapidly reducing the level of disease. Importantly, almost all the patients who were retreated regained at least MMR, and 82% regained the DMR criteria for a second discontinuation attempt, which has been recently proven to be feasible.31 In fact, Legros et al. reported that 35% of patients who had a second discontinuation attempt (median total time of treatment of 103 months) remained free from relapse at three years.31 Those who have eventually restarted treatment had nonetheless taken advantage of a treatment 'holiday' without meaningful risks.

Conclusions This multicenter observational study included a substantial number of patients who were cared for in care institutes through clinical practice procedures, confirming that treatment discontinuation is safe and effective also outside controlled clinical trials. Taking into account all the evidence collected in the last ten years, we think that TKI discontinuation in patients in persistent DMR must be considered in routine clinical practice, as long as molecular monitoring is performed regularly in standardized laboratories, and in accordance with the criteria stated in the ESMO and NCCN recommendations.34,35 Acknowledgments We thank the Associazione Italiana Leucemie (AIL) for the continuous support to doctors and patients.

4. Narra RK, Flynn KE, Atallah E. Chronic Myeloid Leukemia-the Promise of Tyrosine Kynase Inhibitor Discontinuation. Curr Hematol Malig Rep. 2017;12(5):415-423. 5. Experts in Chronic Myeloid Leukemia. The price of drugs for chronic myeloid leukemia (CML) is a reflection of the unsustainable prices of cancer drugs: from the perspective of a large group of CML experts. Blood. 2013;121(22):4439-4442. 6. Rousselot P, Huguet F, Rea D, et al. Imatinib mesylate discontinuation in patients with chronic myelogenous leukemia in complete molecular remission for more than 2 years. Blood. 2007;109(1):58-60. 7. Etienne G, Guilhot J, Rea D, et al. LongTerm Follow-Up of the French Stop Imatinib (STIM1) Study in Patients With Chronic Myeloid Leukemia. J Clin Oncol. 2017;35(3):298-305.

8. Ross DM, Branford S, Seymour JF, et al. Safety and efficacy of imatinib cessation for CML patients with stable undetectable minimal residual disease: results from the TWISTER study. Blood. 2013;122(4):515522. 9. Rousselot P, Charbonnier A, ConyMakhoul P, et al. Loss of major molecular response as a trigger for restarting tyrosine kinase inhibitor therapy in patients with chronic-phase chronic myelogenous leukemia who have stopped imatinib after durable undetectable disease. J Clin Oncol. 2014;32(5):424-430. 10. Takahashi N, Kyo T, Maeda Y, et al. Discontinuation of imatinib in Japanese patients with chronic myeloid leukemia. Haematologica. 2012;97(6):903-906. 11. Thielen N, van der Holt B, Cornelissen JJ, et al. Imatinib discontinuation in chronic

1595

C. Fava et al.

12.

13.

14.

15.

16.

17.

18.

1596

phase myeloid leukaemia patients in sustained complete molecular response: a randomised trial of the Dutch-Belgian Cooperative Trial for Haemato-Oncology (HOVON). Eur J Cancer. 2013;49(15):32423246. Lee S-E, Choi SY, Bang J-H, et al. Predictive factors for successful imatinib cessation in chronic myeloid leukemia patients treated with imatinib. Am J Hematol. 2013; 88(6):449-454. Mori S, Vagge E, le Coutre P, et al. Age and dPCR can predict relapse in CML patients who discontinued imatinib: the ISAV study. Am J Hematol. 2015;90(10):910-914. Tsutsumi Y, Ito S, Ohigashi H, Shiratori S, Teshima T. Unplanned discontinuation of tyrosine kinase inhibitors in chronic myeloid leukemia. Mol Clin Oncol. 2016;4(1):89-92. Lee S-E, Choi SY, Song H-Y, et al. Imatinib withdrawal syndrome and longer duration of imatinib have a close association with a lower molecular relapse after treatment discontinuation: the KID study. Haematologica. 2016;101(6):717-723. Yhim H-Y, Lee N-R, Song E-K, et al. LongTerm Outcomes after Imatinib Mesylate Discontinuation in Chronic Myeloid Leukemia Patients with Undetectable Minimal Residual Disease. Acta Haematol. 2016;135(3):133-139. Ferrero D, Cerrano M, CrisĂ E, Aguzzi C, Giai V, Boccadoro M. How many patients can proceed from chronic myeloid leukaemia diagnosis to deep molecular response and long-lasting imatinib discontinuation? A real life experience. Br J Haematol. 2017;176(4):669-671. Imagawa J, Tanaka H, Okada M, et al. Discontinuation of dasatinib in patients with chronic myeloid leukaemia who have maintained deep molecular response for longer than 1 year (DADI trial): a multicentre phase 2 trial. Lancet Haematol.

2015;2(12):e528-535. 19. Rea D, Nicolini FE, Tulliez M, et al. Discontinuation of dasatinib or nilotinib in chronic myeloid leukemia: interim analysis of the STOP 2G-TKI study. Blood. 2017;129(7):846-854. 20. Hochhaus A, Masszi T, Giles FJ, et al. Treatment-free remission following frontline nilotinib in patients with chronic myeloid leukemia in chronic phase: results from the ENESTfreedom study. Leukemia. 2017;31(7):1525-1531 21. Saussele S, Richter J, Guilhot J, et al. Discontinuation of tyrosine kinase inhibitor therapy in chronic myeloid leukaemia (EURO-SKI): a prespecified interim analysis of a prospective, multicentre, non-randomised, trial. Lancet Oncol. 2018;19(6):747-757. 22. Mahon FX, Nicolini FE, Noel MP, et al. Preliminary report of the STIM2 study: a multicenter stop imatinib trial for chronic phase chronic myeloid leukemia de novo patients on imatinib [abstract]. Blood. 2013; 122(21):654. 23. Takahashi N, Tauchi T, Kitamura K, et al. Deeper molecular response is a predictive factor for treatment-free remission after imatinib discontinuation in patients with chronic phase chronic myeloid leukemia: the JALSG-STIM213 study. In J Hematol. 2018;107(2):185-193. 24. Shah NP, Paquette R, MĂźller MC, et al. Treatment-free remission (TFR) in patients with chronic phase chronic myeloid leukemia (CMLCP) and in stable deep molecular response (DMR) to dasatinib: the Dasfree Study. Blood. 2016; 128(22):1895. 25. Hughes TP, Boquimpani CM, Takahashi N, et al. Treatment-free remission in patients with chronic myeloid leukemia in chronic phase according to reasons for switching from imatinib to nilotinib: subgroup analysis from ENESTop. Blood. 2016; 128(22):792.

26. Cross NC, White HE, Colomer D, et al. Laboratory recommendations for scoring deep molecular responses following treatment for chronic myeloid leukemia. Leukemia. 2015;29(5):999-1003. 27. Baccarani M, Deininger MW, Rosti G, et al. LeukemiaNet recommendations for the management of chronic myeloid leukemia: 2013. Blood. 2013;122(6):872-884. 28. Sander Greenland, Rhian Daniel, Neil Pearce; Outcome modelling strategies in epidemiology: traditional methods and basic alternatives. Int J Epidemiol. 2016;45(2):565-575. 29. Vatcheva KP, Lee M, McCormick JB, Rahbar MH. Multicollinearity in Regression Analyses Conducted in Epidemiologic Studies. Epidemiology (Sunnyvale). 2016;6(2). 30. R Core Team (2017). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria 31. Legros L, Nicolini FE, Etienne G, et al. Second tyrosine kynase inhibitor discontinuation attempt in patients with chronic myeloid leukemia. Cancer. 2017; 123(22):4403-4410. 32. Kong JH, Winton EF, Heffner LT, et al. Does the frequency of molecular monitoring after tyrosine kynase inhibitor discontinuation affect outcomes of patients with chronic myeloid leukemia? Cancer. 2017;123(13):2482-2488. 33. http://www.ema.europa.eu/docs/en_GB/ d o c u m e n t _ l i b r a r y / EPAR_Product_Information/human/00079 8/WC500034394.pdf 34. Hochhaus A, Saussele S, Rosti G, et al. Chronic Myloid Leukemia: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up, Ann Oncol. 2017;28(suppl_4):iv41-iv51. 35. https://www.nccn.org/professionals/physician_gls/pdf/cml.pdf

haematologica | 2019; 104(8)

ARTICLE

Acute Myeloid Leukemia

RUNX1 inhibits proliferation and induces apoptosis of t(8;21) leukemia cells via KLF4mediated transactivation of P57

Ferrata Storti Foundation

Shuang Liu,1* Yanyan Xing,1* Wenting Lu,1 Shouyun Li,1 Zheng Tian,1 Haiyan Xing,1 Kejing Tang,1 Yingxi Xu,1 Qing Rao,1 Min Wang1 and Jianxiang Wang1,2 *These authors contributed equally to this work.

State Key Laboratory of Experimental Hematology and 2National Clinical Research Center for Blood Diseases, Institute of Hematology and Blood Diseases Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Tianjin, P.R. China

Haematologica 2019 Volume 104(8):1597-1607

ABSTRACT

UNX1 is a key transcription factor in hematopoiesis and its disruption is one of the most common aberrations in acute myeloid leukemia. RUNX1 alterations affect its DNA binding capacity and transcriptional activities, leading to the deregulation of transcriptional targets, and abnormal proliferation and differentiation of myeloid cells. Identification of RUNX1 target genes and clarification of their biological functions are of great importance in the search for new therapeutic strategies for RUNX1-altered leukemia. In this study, we identified and confirmed that KLF4, a known tumor suppressor gene, as a direct target of RUNX1, was down-regulated in RUNX1-ETO leukemia. RUNX1 bound to KLF4 promoter in chromatin to activate its transcription, while the leukemogenic RUNX1-ETO fusion protein had little effect on this transactivation. KLF4 was also identified as a novel binding partner of RUNX1. RUNX1 interacted with KLF4 through Runt domain and further co-activated its target genes. However, RUNX1-ETO competed with RUNX1 to bind KLF4 through Runt and ETO domains, and abrogated transcription of KLF4. Finally, overexpression experiments indicated that RUNX1 inhibited proliferation and induced apoptosis of t(8;21) leukemia cells via KLF4-mediated upregulation of P57. These data suggest KLF4 dysregulation mediated by RUNX1-ETO enhances proliferation and retards apoptosis, and provides a potential target for therapy of t(8;21) acute myeloid leukemia.

Introduction RUNX1, also known as AML1 and CBFα, is a critical transcription factor in hematopoiesis, which regulates various kinds of hematopoiesis-related genes, including cytokines, cytokine receptors, microRNA and other transcription factors.1 It is very versatile and also interacts with a number of other hematopoietic regulators, such as CBFβ, PU.1, GATA1, PAX5, and ETS1.2-6 These interaction partners provide RUNX1 with the potential to target genes primarily regulated by other transcription factors, and vice versa. RUNX1 is frequently involved in gene mutations and chromosomal translocations in leukemias, which indicates that the altered function of RUNX1 is closely related with leukemogenesis. Among them, t(8;21)(q22;q22) is one of the most common chromosomal translocations in acute myeloid leukemia (AML), which results in RUNX1-ETO fusion protein. RUNX1ETO fuses the N-terminus of RUNX1 including only runt domain (RHD) in-frame with the almost entire ETO protein. This leukemogenic fusion protein competes with wild-type RUNX1 in binding to its target genes and recruits a transcriptional co-repressor complex NCoR/SMRT/HDAC to further repress transcription of RUNX1 target genes.7-9 The dominant negative repressive effects of RUNX1-ETO on RUNX1 is considered to be the major pathogenic mechanism of t(8;21) AML, which causes blockage of normal hematopoietic differentiation and accumulation of immature myelocytes.10 haematologica | 2019; 104(8)

Correspondence: JIANXIANG WANG wangjx@ihcams.ac.cn MIN WANG wangjxm@ihcams.ac.cn Received: March 5, 2018. Accepted: February 20, 2019. Pre-published: February 21, 2019. doi:10.3324/haematol.2018.192773 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1597 ©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.

1597

S. Liu et al.

A large number of studies on RUNX1 and RUNX1-ETO have been performed to investigate their roles in normal hematopoiesis and leukemogenesis. However, the mechanisms by which RUNX1 and RUNX1-ETO regulate their target genes and interact with binding partners are far from clear. Identification of RUNX1 and RUNX1-ETO novel target genes and interacting proteins are still of great importance in order to develop new therapeutic strategies for t(8;21) leukemia. Our previous study reported that sodium phenylbutyrate (PB), one of the HDAC inhibitors, could induce t(8;21) leukemia cells to undergo differentiation and apoptosis.11 In another study, we identified PIG7 as a direct target gene of RUNX1 that was responsible for differentiation and apoptosis induction of t(8;21) leukemia cells.12 In addition to PIG7, a series of genes were also up-regulated in the process, including RUNX1, KLF4 and P57. Among them, KLF4, a transcription factor frequently deregulated in a variety of malignant tumors, including several examples in hematologic malignancies, was considered a tumor suppressor. In AML, KLF4 was negatively regulated by HDAC1 and overexpression of KLF4 could markedly repress proliferation of AML cells by blocking cell cycles and inducing the expression of P21 and P27.13 In T-cell acute lymphoblastic leukemia, KLF4 delayed disease progression through directly inhibiting T-cell associated genes, such as NOTCH1, BCL11B, GATA3 and TCF7, and activating the BCL2/BCLXL pathway.14 Another study by Morris et al. reported that KLF4-mediated anti-leukemic effects through regulation of microRNA networks, including MIR-150 and the MYC/MAX/MXD network, providing novel mechanistic support for AML treatment via increasing KLF4 expression.15 In this report, we identified KLF4 as both a novel target gene and a binding partner of RUNX1, which induced the expression of cell cycle inhibitor P57. We further confirmed that overexpression of either of RUNX1, KLF4 or P57 could inhibit proliferation and induce apoptosis of t(8;21) leukemia cell. Our results support the hypothesis that RUNX1, KLF4 and P57 might compose a transcriptional activation cascade in t(8;21) leukemia cells. RUNX1 inhibited proliferation and induced apoptosis of t(8;21) leukemia cells via KLF4-mediated upregulation of P57. We believe that reactivating the “RUNX1-KLF4-P57” signaling pathway might be a potentially potent therapeutic strategy for t(8;21) AML.

Methods

Lentiviral preparation and transduction of Kasumi-1 cells Details of lentiviral preparation and transduction of Kasumi-1 cells are available in the Online Supplementary Appendix.

Real-time polymerase chain reaction and western blot analyses Real-time quantitative polymerase chain reaction (PCR) and western blot analyses for gene expression are described in the Online Supplementary Appendix.

Co-immunoprecipitation assay Cells were collected and lysed in Cell lysis buffer for western blotting and immunoprecipitation (IP) (Beyotime, China) on ice for 30 minutes (min). Then the lysates were pre-cleared with protein A/G-Sepharose beads (Santa Cruz, CA, USA) at 4°C for 1h before IP with indicated primary antibodies or anti-IgG antibody (Beyotime) overnight. The protein-antibody complexes were incubated with protein A/G-Sepharose beads for another 4 hours (h). The beads were subsequently washed three times with cold lysis buffer and the bound proteins were separated by SDS-PAGE, followed by western blot assay.

Immunofluorescence analysis Cell preparation and immunofluorescence staining procedures have been described previously.16 Fluorescence images were taken on a spinning disk confocal microscope (PerkinElmer, USA) by a 100X oil-immersion objective.

Luciferase assay 5×104 CV-1 cells were transfected with 100 ng indicated pGL3 luciferase reporter plasmid, 50 ng control Renilla luciferase plasmid together with different combinations of transcription factor expression plasmid. At 48 h after transfection, cells were harvested and the luciferase activities were analyzed according to the Dual-Luciferase Reporter 1000 Assay System Technical Manual (Promega, USA) on a luminometer (Lumat LB 9507, Berthold Technologies GmbH & Co. KG, Baden-Württemberg, Germany).

Chromatin immunoprecipitation Chromatin immunoprecipitation (ChIP) assays were performed with a Pierce Agarose ChIP Kit according to the manufacturer’s instructions (Thermo Scientific, USA). Each ChIP reaction contained chromatin from 1×107 cells and 2 μg indicated antibody. Following IP, the binding sites of transcription factor to target gene promoter were analyzed by PCR. Anti-RUNX1 antibody (ChIP Grade, Abcam, ab23980) and Rabbit IgG antibody (Beyotime) were used in this study. Primers used in ChIP assays were listed in the Online Supplementary Table S1.

Cell culture All cell lines used in this study were purchased from ATCC (Manassas, VA, USA). Kasumi-1 and SKNO-1 cells were cultured in RPMI-1640 medium (Gibco-Life Technologies, Grand Island, NY, USA) supplemented with 20% fetal bovine serum (FBS) (Gibco). SKNO-1 was also supplemented with 10 ng/mL recombinant human granulocyte-macrophage colony-stimulating factor (Peprotech, USA). CV-1 and HEK-293T cells were maintained in DMEM (Gibco) supplemented with 10% FBS.

Plasmids construction Construction of the plasmids for luciferase assays, co-immunoprecipitation assays and overexpression experiments are described in the Online Supplementary Appendix.

1598

Cell proliferation, cell cycle, cell apoptosis and differentiation analyses Details are available in the Online Supplementary Appendix.

Statistical analysis All results were presented as mean with error bars indicating standard deviation of triplicate experiments. Statistical differences were determined by Student t-test using GraphPad Prism (v.5.0). P<0.05 was considered statistically significant; *P<0.05; **P<0.01; ***P<0.001.

haematologica | 2019; 104(8)

RUNX1 transactivated and interacted with KLF4

Results Phenylbutyrate up-regulated expressions of RUNX1, KLF4 and P57 in Kasumi-1 cells Gene expression profiles of PB-treated Kasumi-1 cells demonstrated that RUNX1, KLF4 and P57 were up-regulated during the process of cell differentiation and apoptosis induced by PB. To confirm the upregulation of these three genes, Kasumi-1 cells were treated with PB at different doses and incubation times. Then cells were harvested, and the mRNA and protein expressions of RUNX1, KLF4 and P57 were evaluated by qRT-PCR and western blot. The results clearly showed that PB up-regulated RUNX1, KLF4 and P57 expressions at both mRNA (Figure 1A) and protein (Figure 1B) level in a time- and dosedependent manner.

Identification of KLF4 as a novel target gene of RUNX1 and RUNX1-ETO To determine the mechanism of KLF4 upregulation by PB treatment, we first analyzed KLF4 promoter region (bp -2668 to -195) (Homo sapiens chromosome 9, GRCh37) for the transcription factors binding sites by Jaspar database. The result showed that there were seven putative RUNX1-binding sites (R1ď˝&#x17E;R7) within KLF4 promoter region (Figure 2A). We then constructed a pGL3-KLF4 promoter reporter plasmid (pGL3-KLF4) containing RUNX1 binding sites to analyze the transcriptional regulation effects of RUNX1 and RUNX1-ETO on KLF4. The luciferase activity of pGL3-KLF4 was activated by RUNX1 in a dose-dependent manner

with a nearly 12-fold increase at the dose of 200 ng (Figure 2B). However, there was a less than 3-fold increase of pGL3-KLF4 luciferase activity by RUNX1-ETO at the doses from 50 ng to 200 ng. To further determine the specific binding sites of RUNX1 and RUNX1-ETO on KLF4 promoter, ChIP assays were performed with a specific anti-RUNX1 antibody in Kasumi-1 cells. The results confirmed the direct binding of RUNX1 to KLF4 promoter region via four predicted binding sites at bp -2617 to -2607 (R1), bp -2312 to -2302 (R2), bp -1136 to -1126 (R4), and bp -764 to -754 (R5) (Figure 2C). Subsequently, western blot analyses were performed to verify the regulation relationship on protein level. 293T cells were transfected with pCMV5-vector, pCMV5-RUNX1 and pCMV5-RUNX1-ETO plasmids, respectively. After 48 h of transfection, cells were harvested for western blot. The results showed that KLF4 was clearly up-regulated by RUNX1 while RUNX1ETO had almost no effect (Figure 2D).

KLF4 is also a novel binding partner of RUNX1 and RUNX1-ETO De novo motif analysis of our previous RUNX1 Chip-Seq data demonstrated that KLF4 binding sites were significantly enriched within RUNX1 chip regions, which raised the hypothesis that KLF4 might co-localize and interact with RUNX1. To validate our hypothesis, immunofluorescence confocal imaging analysis and co-immunoprecipitation (Co-IP) assays were performed. Immunofluorescence analysis showed both exogenous and endogenous colocalization of RUNX1 and KLF4 in nuclei (Figure 3A). CoIP assay further confirmed the interaction between RUNX1 and KLF4 (Figure 3C and Online Supplementary Figure S1A). However, due to the low expression level of

Figure 1. Upregulation of RUNX1, KLF4 and P57 in t(8;21) leukemia cells by sodium phenylbutyrate (PB) treatment. (A) Relative expressions of RUNX1, KLF4 and P57 mRNA were measured by quantitative real-time polymerase chain reaction in Kasumi-1 cells after treatment with different doses of PB for the time (hours, h) indicated. The gene transcript levels were normalized to those of GAPDH and set to 1 in the control group. (B) Western blot analysis of RUNX1, KLF4 and P57 protein expressions in Kasumi-1 cells after PB treatment. Î˛-actin and H3 were used as internal loading controls.

haematologica | 2019; 104(8)

1599

S. Liu et al.

KLF4 in Kasumi-1 and SKNO-1 cells, the endogenous interaction between KLF4 and RUNX1 was less evident than that under conditions of exogenous overexpression in 293T cells. The co-localization and protein interaction of fusion protein RUNX1-ETO and KLF4 were also investigated. Similar to wild-type RUNX1, RUNX1-ETO colocalized and interacted with KLF4 in nuclei (Figure 3B and D and Online Supplementary Figure S1B).

Identification of the specific domains of RUNX1 and RUNX1-ETO mediating interaction with KLF4 A large number of studies have reported that the runt homology domain (RHD) was responsible for RUNX1

interacting with other transcription factors, such as PU.1, STAT5 and GATA1.3,4,17 To clarify whether RUNX1 interacted with KLF4 through RHD, full-length or truncated mutant (RHD deleted, ΔRHD) of RUNX1 and RHD domain were cloned into a pCMV5 MYC-tagged plasmid, respectively (Figure 4A). After co-transfecting 293T cells with each of the MYC-tagged RUNX1 constructs and FLAG-tagged KLF4 construct, Co-IP assays were performed with anti-FLAG antibody for IP and anti-MYC antibody for immunoblotting (IB). Both the full-length RUNX1 and RHD could directly interact with KLF4, while the truncated mutant RUNX1-ΔRHD failed to interact with KLF4 (Figure 4B). The specific domains of RUNX1-

1600

Figure 2. KLF4 is a novel target gene of RUNX1 and RUNX1-ETO. (A) Schematic representation of KLF4 promoter fragments fused to a pGL3-Basic Vector. The putative RUNX1 binding sites are indicated by orange boxes as follows: R1 (-2617 to -2607), R2 (-2312 to 2302), R3 (-2014 to -2004), R4 (-1136 to -1126), R5 (764 to -754), R6 (-399 to -394), and R7 (-263 to -258). Transcription start site (TSS) is indicated by an arrow. Numbers represent base pairs relative to TSS. (B) pGL3KLF4 promoter reporter plasmid (pGL3-KLF4) was cotransfected with increasing doses of pCMV5-RUNX1 or pCMV5-RUNX1-ETO into CV-1 cells. At 48 h after transfection, the luciferase transcriptional activity of pGL3KLF4 was measured and normalized to that of Renilla luciferase. (C) Chromatin immunoprecipitation analysis of RUNX1 binding sites to KLF4 promoter region in Kasumi-1 cells. The predicted binding sites R1, R2, R4 and R5 were successfully amplified, both from the input DNA and chromatin immunoprecipitation by an antiRUNX1 antibody, whereas no amplified product was obtained in the IgG control group. White stars indicate non-specific amplified bandings. (D) 293T cells were transfected with pCMV5-vector, pCMV5-RUNX1 and pCMV5-RUNX1-ETO, respectively. At 48 h after transfection, cells were harvested and the protein levels of KLF4 were assayed by western blot. β-actin and H3 were used as protein loading controls.

haematologica | 2019; 104(8)

RUNX1 transactivated and interacted with KLF4

ETO-mediated interaction with KLF4 were also investigated. The results showed that KLF4 not only interacted with RHD, but also with the ETO part of RUNX1-ETO (Figure 4A and C). Further study verified the physical interaction between KLF4 and ETO protein (Figure 4D).

dependent transactivation of target genes, a luciferase reporter plasmid containing KLF4 binding sites (KLF4Reporter) was constructed. Luciferase assay was then performed (Figure 5B). The results demonstrated that KLF4 was capable of transactivating its target genes reporter plasmid (KLF4-Reporter) in CV-1 cells. RUNX1 significantly enhanced the transcriptional activation effect of KLF4 in a dose-dependent manner, while RUNX1-ETO could only enhance the effect slightly at low doses (50-100 ng). The above results suggested that RUNX1-ETO might compete with RUNX1 for binding to KLF4 and abrogate transcription of the KLF4 target gene. To confirm this hypothesis, we performed co-transfection of RUNX1 and RUNX1ETO for luciferase assay using KLF4 target gene P57 promoter reporter. CV-1 cells were transfected with KLF4 or RUNX1 expression plasmid alone or in different combinations of KLF4, RUNX1 and RUNX1-ETO expression plasmids to investigate their regulating effect on P57 promoter reporter. The transactivation of P57 promoter mediated by

RUNX1-ETO competes with RUNX1 for binding to KLF4 We investigated the physiological interaction between RUNX1 and KLF4. We found that both RUNX1 and RUNX1-ETO interacted with KLF4 (Figure 3) and leukemic fusion protein RUNX1-ETO obstructed the physiological interaction between RUNX1 and KLF4. Fixed doses of pCMV5-FLAG-KLF4 and pCMV5-MYCRUNX1 were co-transfected with increasing doses of pCMV5-MYC-RUNX1-ETO into 293T cells. At 48 h after transfection, cell lysates were prepared for Co-IP assay. RUNX1-ETO competed with RUNX1 for binding to KLF4 in a dose-dependent manner (Figure 5A). To further investigate the effects of RUNX1 and RUNX1-ETO on KLF4-

Figure 3. Identification of KLF4 as a novel binding partner of RUNX1 and RUNX1-ETO. (A) KLF4 co-localized with RUNX1 in nuclei. (Top) Co-localization analysis of exogenous MYC-RUNX1 and FLAG-KLF4 in 293T cells at 48 h after transfection by immunofluorescence assay. Anti-MYC (green) and anti-FLAG (red) antibodies were used as primary antibodies; DAPI (blue) was used for nuclear staining. Scale bar represents 10 μm. (Middle and bottom) Co-localization analysis of endogenous RUNX1 and KLF4 in Kasumi-1 and SKNO-1 cells with anti-RUNX1 (green) and anti-KLF4 (red) antibodies; DAPI (blue) was used for nuclear staining. Scale bar represents 10 μm. (B) KLF4 co-localized with RUNX1-ETO in nuclei. (Top) Co-localization analysis of exogenous MYC-RUNX1-ETO and FLAG-KLF4 in 293T cells at 48 h after transfection. Antibodies were used as described in (A). Scale bar represents 10 μm. (Middle and bottom) Co-localization analysis of endogenous RUNX1-ETO and KLF4 in Kasumi-1 and SKNO-1 cells with anti-ETO (green) and anti-KLF4 (red) antibodies. DAPI (blue) was used for nuclear staining. Scale bar represents 10 μm. (C) KLF4 interacted with RUNX1. 293T cells were co-transfected with pCMV5-MYC-RUNX1 and pCMV5-FLAG-KLF4. At 48 h after transfection cells were harvested and cell lysates underwent immunoprecipitation (IP) with anti-FLAG or anti-MYC antibody. Immunoblotting (IB) analysis was performed with the other antibody. (D) KLF4 interacted with RUNX1-ETO. 293T cells were co-transfected with pCMV5-MYC-RUNX1-ETO and pCMV5-FLAG-KLF4. At 48 h after transfection, cells were harvested and cell lysates underwent IP with anti-FLAG or anti-MYC antibody. Immunoblotting analysis was performed with the other antibody. WB: western blotting.

haematologica | 2019; 104(8)

1601

S. Liu et al.

KLF4 and RUNX1 was abrogated by RUNX1-ETO in a dose-dependent manner (Figure 5C), suggesting that RUNX1-ETO competed with RUNX1 for binding to KLF4 and abrogated transcription of KLF4.

Biological roles of RUNX1 and KLF4 in t(8;21) leukemia cells The results seen in Figure 1 suggested that RUNX1 and KLF4 might contribute to the apoptosis and differentiation of Kasumi-1 cells induced by PB. Overexpression experiments were then performed to address their roles in t(8;21) leukemia cells. RUNX1 and KLF4 were overexpressed respectively in Kasumi-1 cells by a pCDH

lentivirus system, and the overexpression efficiencies were confirmed at both mRNA and protein levels by qRTPCR and western blot assay (Figure 6A). 3-(4,5dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4sulfophenyl)-2H-tetrazolium (MTS) assay was performed to evaluate their effects on cell proliferation. The results showed that both of them could markedly inhibit cell proliferation (Figure 6B). Cell cycle analysis was carried out with propidium iodide staining and examined by flow cytometry. KLF4 overexpression blocked cell cycle in G0/G1 phase and reduced S and G2/M cell proportion, while overexpression of RUNX1 had no significant effect on the cell cycle in Kasumi-1 cells (Figure 6C). Apoptosis

1602

Figure 4. Identification of the specific domains of RUNX1 and RUNX1-ETO mediating interaction with KLF4. (A) Schematic representation of full-length and truncation mutants of RUNX1 (top) and RUNX1-ETO (bottom). (B and C) FLAG-KLF4 and full-length or truncation mutants of RUNX1 (B) or RUNX1-ETO (C) with MYC tag were co-transfected into 293T cells. At 48 hours (h) after transfection, cell lysates were prepared and underwent immunoprecipitation (IP) with anti-FLAG antibody. Immunoblotting (IB) analyses were performed with anti-MYC antibody. Different combinations of cotransfection with KLF4 and full-length or truncation mutants of RUNX1/RUNX1ETO were labeled as KLF4/RUNX1, KLF4/RUNX1-Î&#x201D;RHD, KLF4/RHD, KLF4/RE, KLF4/RE-Î&#x201D;RHD and KLF4/RHD, respectively. (D) KLF4 and ETO expression plasmids were co-transfected into 293T cells. At 48 h after transfection, cells were harvested and cell lysates underwent IP with anti-KLF4 or anti-ETO antibody. IB analysis was performed with the other antibody. WB: western blotting.

haematologica | 2019; 104(8)

RUNX1 transactivated and interacted with KLF4

analysis demonstrated that both RUNX1 and KLF4 promoted apoptosis of Kasumi-1 cells (Figure 6D). Further evidence of cell apoptosis was provided by WrightGiemsa staining, which displayed typical apoptotic characteristics such as karyopyknosis, reduced ratio of nucleus to cytoplasm, nuclear fragmentation with intact cell membrane, and vacuolar degeneration in pCDH-RUNX1 and pCDH-KLF4 groups compared with that of the control group (Figure 6E). We next examined the effects of RUNX1 and KLF4 overexpression on Kasumi-1 cell differentiation. The expression levels of myeloid cell surface markers CD11b and CD15 were analyzed by flow cytometry at 48 h and 72 h after lentivirus infection. The results showed that there was a big increase in the proportion of after infection with CD11b+ and CD15+ cells pCDH-KLF4, whereas no parallel changes were found in pCDH-RUNX1 group (Figure 6F and Online Supplementary Figure S3).

Identification of P57 as a target gene of KLF4 As mentioned in Figure 1, the upregulation of P57 correlated with that of RUNX1 and KLF4 during PB-induced cell differentiation and apoptosis; however, the mechanism of this coherence was unknown. Previously published studies reported that KLF4 could up-regulate P57 in lymphatic endothelial cells and colon cancer cells.18,19 Based on this, we explored whether KLF4 up-regulated the expression of P57 in t(8;21) leukemia cells. Gene promoter analysis showed that there were four KLF4 putative binding sites (K1~K4) within P57 promoter region (bp -979 to +240) (Homo sapiens chromosome 11, GRCh38) (Figure

7A). We then cloned the promoter region of P57 containing KLF4 putative binding sites into a pGL3-Basic Vector to construct the pGL3-P57 plasmid. Next, the pGL3-P57 plasmid was co-transfected with different doses of KLF4 expression plasmid into CV-1 cells and the luciferase activity of pGL3-P57 was evaluated. KLF4 could transactivate pGL3-P57 in a dose-dependent manner with the maximal effect with an almost 8-fold increase at the dose of 100 ng (Figure 7B). Moreover, P57 protein level was found remarkably up-regulated in Kasumi-1 cells after KLF4 overexpression, which further substantiated the regulation relationship (Figure 7C). As discussed (see above), KLF4 was a downstream gene of RUNX1; we therefore concluded that RUNX1, KLF4 and P57 might make up a transcriptional activation cascade in t(8;21) leukemia cells. To further testify the “RUNX1-KLF4-P57” pathway, we over-expressed RUNX1 in Kasumi-1 cells and examined the expression level of putative downstream genes KLF4 and P57. The result showed that both KLF4 and P57 were remarkably up-regulated after RUNX1 overexpression (Figure 7D). Finally, we explored the biological functions of P57 in t(8;21) leukemia cells. Overexpression of P57 mediated inhibition in cell proliferation, blockage in cell cycle, and induction of cell apoptosis, similar to RUNX1 and KLF4 (Figure 7E). However, P57 overexpression had no effects on cell differentiation as assessed by myeloid markers CD11b and CD15 and morphology analysis (Figure 7E and Online Supplementary Figure S3). Taken together, the above results suggested that the “RUNX1KLF4-P57” pathway, activated by PB, was closely related to cell proliferation and apoptosis in t(8;21) leukemia cells.

Figure 5. RUNX1-ETO competes with RUNX1 for binding to KLF4. (A) Fixed dose of pCMV5-FLAG-KLF4 and pCMV5-MYC-RUNX1 were co-transfected with increasing dosages of pCMV5-MYC-RUNX1-ETO into 293T cells. At 48 hours (h) after transfection, cell lysates were prepared and underwent immunoprecipitation (IP) with antiFLAG antibody. Immunoblotting analyses were performed with anti-MYC antibody. β-actin and H3 served as loading controls for the whole cell lysate. (B) The effects of RUNX1 and RUNX1-ETO on KLF4-dependent transactivation of target genes. CV-1 cells were transfected with fixed dose of KLF4 target genes reporter plasmid (KLF4-Reporter), pCMV5-KLF4 and increasing doses of pCMV5-RUNX1 (top) or pCMV5-RUNX1-ETO (bottom). At 48 h after transfection, transfected cells were harvested for luciferase assay. The luciferase transcriptional activities of KLF4-Reporter were measured and normalized to that of Renilla luciferase. Cells transfected with only KLF4-Reporter were set as control. (C) The transcriptional regulatory effects of RUNX1 and RUNX1-ETO on KLF4 target gene P57 promoter reporter (pGL3P57). CV-1 cells were transfected with KLF4 or RUNX1 expression plasmid alone or different combinations of KLF4, RUNX1 and RUNX1-ETO expression plasmids together to investigate their regulating effect on pGL3-P57. At 48 h after transfection, transfected cells were harvested for luciferase assay and western blot analysis. The luciferase transcriptional activities of pGL3-P57 were measured and normalized to that of Renilla luciferase. Cells transfected with pGL3-P57 only were set as control.

haematologica | 2019; 104(8)

1603

S. Liu et al.

Discussion Dysfunction of RUNX1 through gene mutations and chromosome translocations occurs frequently in various myeloid malignancies. As a transcription factor, RUNX1 exerts its biological effects through transcriptional regulation of its target genes. Thus, identification of RUNX1 target genes and clarification of their roles would offer an intriguing insight into the mechanisms of how perturbed RUNX1 function may result in hematologic malignancies. For this purpose, we comprehensively analyzed RUNX1 Chip-Seq data with our previous gene expression profiles of PB-treated Kasumi-1 cells. The result revealed a list of putative RUNX1 target genes which had a high potential

of driving differentiation and apoptosis of t(8;21) leukemia cells. These targets also served as potential candidates for developing new therapies to treat t(8;21) AML. Among them, p53-induced gene 7 (PIG7), a gene transactivated by RUNX1 through a specific RUNX1-binding site located in the promoter region (bp -1511 to -1503) and promoted apoptosis and differentiation of leukemia cells, has been reported.12 In the present study, we focused our attention on KLF4. On the basis of previously published studies, KLF4 is likely to have similar cellular phenotypes as those observed with RUNX1 in regulating cell proliferation and differentiation. It is reported that RUNX1 mediated inhibition of AML cell survival through repression of VEGF expression, a major mediator of angiogenesis, prolifera-

Figure 6. Biological effects of RUNX1 and KLF4 overexpression on Kasumi-1 cell proliferation, apoptosis and differentiation. (A) Overexpression of RUNX1 and KLF4 in Kasumi-1 cells were mediated by a pCDH lentivirus system. At 72 hours (h) after infection, the infected cells were sorted by flow cytometry for GFP+ population and the expression levels of RUNX1 or KLF4 were measured by quantitative real-time polymerase chain reaction (left) and western blot (right), respectively. (B) MTS assay was performed to evaluate cell proliferation ability of GFP+ Kasumi-1 cells over-expressing RUNX1 or KLF4 at 72 h after lentivirus infection. (C) Cell cycle distribution of GFP+ Kasumi-1 cells over-expressing RUNX1 or KLF4 was analyzed by flow cytometry with PI staining at 48 h after cell sorting. (D) Cell apoptosis analysis of GFP+ Kasumi-1 cells over-expressing RUNX1 or KLF4 were performed by the flow cytometry at 48 h and 72 h after lentivirus infection. (E) Morphological assessment of GFP+ Kasumi-1 cells over-expressing RUNX1 or KLF4 at indicated days after infection. Cells were stained by Wright-Giemsa staining and observed by oil microscopy after flow sorting (magnification x100). (F) Flow cytometry analysis of cell surface markers CD11b and CD15 of GFP+ Kasumi-1 cells over-expressing RUNX1 or KLF4 at 48 h and 72 h after lentivirus infection.

1604

haematologica | 2019; 104(8)

RUNX1 transactivated and interacted with KLF4

tion, migration, and survival in AML.20 Loss of Runx1 down-regulated p19ARF to accelerate the development of MLL-ENL leukemia.21 Besides, RUNX1 co-operated with PU.1 to activate hematopoietic differentiation genes, such

as MCSF, and contributed to the downregulation of the erythroid gene expression program by repressing MIR451 transcription.22,23 Likewise, KLF4 overexpression inhibited cell proliferation of AML cell lines through regulation of

Figure 7. P57 is involved in â&#x20AC;&#x153;RUNX1-KLF4-P57â&#x20AC;? transcriptional activation cascade, which promotes cell proliferation inhibition and apoptosis induction of t(8;21) leukemia cells. (A) Schematic representation of P57 promoter fragments fused to a pGL3-Basic vector. The putative KLF4 binding sites are indicated by yellow boxes. Transcription start site (TSS) is indicated by an arrow. Numbers represent base pairs relative to TSS. (B) pGL3-P57 promoter reporter plasmid (pGL3-P57) was cotransfected with increasing doses of pCMV5-KLF4 into CV-1 cells. At 48 hours (h) after transfection, the luciferase transcriptional activity of pGL3-P57 was measured and normalized to that of Renilla luciferase. (C and D) Kasumi-1 cells were infected with pCDH lentivirus over-expressing KLF4 (C) and RUNX1 (D), respectively. At 72 h after infection, the protein levels of putative downstream genes P57 and KLF4 were evaluated by western blot assay. (E) Overexpression of P57 in Kasumi-1 cells was mediated by a pCDH lentivirus system. At 72 h after infection, the infected cells were sorted for GFP+ population and the overexpression efficiency was confirmed by quantitative real-time polymerase chain reaction and western blot assay, respectively. Then, the biological effects of P57 on cell proliferation, cell cycle distribution, cell apoptosis and differentiation were evaluated by MTS assay, flow cytometry, and cell morphological analysis (see Figure 6).

haematologica | 2019; 104(8)

1605

S. Liu et al.

microRNA networks and cell cycle inhibitors, and it also played important roles in regulating hematopoietic differentiation, especially myeloid and monocytic differentiation.15,24 Furthermore, a recent study described the role of RUNX1 in inducing intestinal goblet cell differentiation, acting through direct transcriptional activation of KLF4.25 However, the interactions between RUNX1 and KLF4 in AML and its association with leukemia development have remained largely unexplored. In this study, we demonstrated that KLF4 was directly regulated by RUNX1 through four of seven putative binding sites within the KLF4 promoter region. KLF4 expression could be markedly transactivated by RUNX1 in a dose-dependent manner while leukemogenic fusion protein RUNX1-ETO only had a slight effect. We expected KLF4 expression to be specifically lower in t(8;21) leukemia samples due to haploinsufficiency of RUNX1 in these cells. However, according to an analysis of a publicly available microarray gene expression profiling (GEP) dataset (containing 285 annotated AML cases and 10 healthy cases), KLF4 expression was consistently decreased in the majority of AML cases, and there was no significant difference between t(8;21) and non-t(8;21) groups.26,27 This suggested that KLF4 inactivation might be occurring in non-t(8;21) leukemia cells in distinctly different ways. One mechanism would be DNA hypermethylation, which has already been confirmed by two independent groups in chronic lymphocytic leukemia and adult Tcell leukemia.28,29 Other possibilities remain of great interest for further investigation. In addition to RUNX1, analysis of KLF4 promoter region also revealed putative binding sites correspond to STAT3 and NF-κB (date not shown), both of which are important transcription factors in the hematopoietic system, suggesting that RUNX1 might collaborate with these factors to regulate KLF4 expression. Clarification of the connections between RUNX1, STAT3 and NF-κB is the subject of our ongoing study. By analyzing RUNX1 Chip-Seq data, we identified several RUNX1 binding sites within the KLF4 promoter region. Furthermore, we found remarkable enrichment of KLF4 consensus motif within RUNX1 chip regions, which suggested the physical interaction between these two factors. Supporting this possibility, both exogenous and endogenous RUNX1 and KLF4 were confirmed to colocalize and interact with each other in nuclei by coimmunoprecipitation and immunofluorescence confocal imaging assay. To the best of our knowledge, this is the first report demonstrating KLF4 both as a direct target and a binding partner of RUNX1. The physical interaction between RUNX1 and KLF4 was mediated through RHD, a critical domain of RUNX1 responsible for DNA binding and commonly involved in protein-protein interaction. RUNX1-KLF4 interaction increased KLF4 transactivation capacity on a promoter reporter driven by KLF4-response elements only (KLF4-Reporter). RUNX1-ETO was also shown to interact with KLF4; however, it had almost no co-activation effect on KLF4-Reporter. The competitive protein-protein interaction experiments showed that RUNX1-ETO disrupted RUNX1-KLF4 interaction in a competitive manner, which might block RUNX1 from coactivating KLF4 target genes. Previous studies described the role of RUNX1 in mediating the interaction between

1606

PU.1 and NuAT and BAF families of co-activators. RUNX1-ETO displaced the co-activators from PU.1 and produced a striking switch to PU.1 interaction with corepressors, such as Dnmt1, Sin3A, Nurd, CoRest, and BWich.30 Whether interaction between RUNX1-ETO and KLF4 would abolish KLF4 binding capacity with co-activators or impact its DNA-binding ability are questions that need to be further addressed. RUNX1 and KLF4 are both up-regulated genes during HDAC inhibitor-induced differentiation and apoptosis of t(8;21) leukemia cells. In this study, we performed overexpression experiments to address their biological roles in t(8;21) leukemia cells. The results showed that both of them contributed to cell proliferation inhibition and cell apoptosis induction. Besides, KLF4 also promoted myeloid differentiation of t(8;21) leukemia cells by up-regulating myeloid markers CD11b and CD15. Furthermore, we reported the role of P57, a novel target gene of KLF4 in limiting proliferation and inducing apoptosis. Thus, RUNX1, KLF4 and P57 might make up a transcriptional activation cascade in regulating t(8;21) leukemia cells survival. To further investigate whether the anti-leukemic effects of the “RUNX1-KLF4-P57” pathway existed only in t(8;21) AML, we performed additional overexpression experiments of RUNX1, KLF4 and P57 in a non-RUNX1ETO expression cell line, the HL-60 cells. The result showed that overexpression of RUNX1 also up-regulated KLF4 and P57 expression in HL-60 cells (Online Supplementary Figure S2A and B), and overexpression of either RUNX1, KLF4 or P57 (Online Supplementary Figure S2C) could inhibit the proliferation and induce apoptosis in HL-60 cells (Online Supplementary Figure S2D and F), which was similar to the results observed in Kasumi-1 cells. The above results obtained from HL-60 cells suggested that the “RUNX1-KLF4-P57” pathway not only existed in RUNX1-ETO expressing cells but also in non-RUNX1ETO expressing cells. The effect of RUNX1-ETO on the “RUNX1-KLF4-P57” pathway needs to be further clarified and this is something that we will carry forward in our future studies. In summary, this study identified KLF4 as a target gene and a binding partner of RUNX1. KLF4 mediated proliferation inhibition and apoptosis induction of t(8;21) leukemia cells through transactivating P57. Restoration of the “RUNX1-KLF4-P57” signaling pathway might be an effective therapeutic strategy for t(8;21) AML. Acknowledgments The authors would like to thank the staff for their kindly assistance, especially Wanzhu Yang, Haoyue Liang and Weichao Fu in Core Facility of flow cytometry, State Key Laboratory of Experimental Hematology, Institute of Hematology and Blood Diseases Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College. Funding This work was supported by grants from the National Natural Science Foundation of China (81570147, 81430004 and 81800153), Foundation for Innovative Research Groups of the National Natural Science Foundation of China (81421002) and CAMS Initiative Fund for Medical Sciences (2016-I2M-1-001, 2016-I2M-1-007).

haematologica | 2019; 104(8)

RUNX1 transactivated and interacted with KLF4

References 1. Ichikawa M, Yoshimi A, Nakagawa M, Nishimoto N, Watanabe-Okochi N, Kurokawa M. A role for RUNX1 in hematopoiesis and myeloid leukemia. Int J Hematol. 2013;97(6):726-734. 2. Warren AJ, Bravo J, Williams RL, Rabbitts TH. Structural basis for the heterodimeric interaction between the acute leukaemiaassociated transcription factors AML1 and CBFbeta. Embo J. 2000;19(12):3004-3015. 3. Petrovick MS, Hiebert SW, Friedman AD, Hetherington CJ, Tenen DG, Zhang DE. Multiple functional domains of AML1: PU.1 and C/EBPalpha synergize with different regions of AML1. Mol Cell Biol. 1998;18(7):3915-3925. 4. Elagib KE, Racke FK, Mogass M, Khetawat R, Delehanty LL, Goldfarb AN. RUNX1 and GATA-1 coexpression and cooperation in megakaryocytic differentiation. Blood. 2003;101(11):4333-4341. 5. Libermann TA, Pan Z, Akbarali Y, et al. AML1 (CBFalpha2) cooperates with B cellspecific activating protein (BSAP/PAX5) in activation of the B cell-specific BLK gene promoter. J Biol Chem. 1999;274(35): 24671-24676. 6. Kim WY, Sieweke M, Ogawa E, et al. Mutual activation of Ets-1 and AML1 DNA binding by direct interaction of their autoinhibitory domains. Embo J. 1999; 18(6):1609-1620. 7. Laherty CD, Yang WM, Sun JM, Davie JR, Seto E, Eisenman RN. Histone deacetylases associated with the mSin3 corepressor mediate mad transcriptional repression. Cell. 1997;89(3):349-356. 8. Hassig CA, Fleischer TC, Billin AN, Schreiber SL, Ayer DE. Histone deacetylase activity is required for full transcriptional repression by mSin3A. Cell. 1997;89(3): 341-347. 9. Wang J, Hoshino T, Redner RL, Kajigaya S, Liu JM. ETO, fusion partner in t(8;21) acute myeloid leukemia, represses transcription by interaction with the human NCoR/mSin3/HDAC1 complex. Proc Natl Acad Sci U S A. 1998;95(18):10860-10865. 10. Lam K, Zhang DE. RUNX1 and RUNX1ETO: roles in hematopoiesis and leukemogenesis. Front Biosci (Landmark Ed).

haematologica | 2019; 104(8)

2012;17:1120-1139. 11. Wang J, Saunthararajah Y, Redner RL, Liu JM. Inhibitors of histone deacetylase relieve ETO-mediated repression and induce differentiation of AML1-ETO leukemia cells. Cancer Res. 1999;59(12):2766-2769. 12. Liu J, Xing H, Chen Y, et al. PIG7, transactivated by AML1, promotes apoptosis and differentiation of leukemia cells with AML1-ETO fusion gene. Leukemia. 2012;26(1):117-126. 13. Huang Y, Chen J, Lu C, et al. HDAC1 and Klf4 interplay critically regulates human myeloid leukemia cell proliferation. Cell Death Dis. 2014;5:e1491. 14. Li W, Jiang Z, Li T, et al. Genome-wide analyses identify KLF4 as an important negative regulator in T-cell acute lymphoblastic leukemia through directly inhibiting Tcell associated genes. Mol Cancer. 2015;14:26. 15. Morris VA, Cummings CL, Korb B, Boaglio S, Oehler VG. Deregulated KLF4 Expression in Myeloid Leukemias Alters Cell Proliferation and Differentiation through MicroRNA and Gene Targets. Mol Cell Biol. 2016;36(4):559-573. 16. Liu S, Lu W, Li S, et al. Identification of JL1037 as a novel, specific, reversible lysine-specific demethylase 1 inhibitor that induce apoptosis and autophagy of AML cells. Oncotarget. 2017;8(19):31901-31914. 17. Ogawa S, Satake M, Ikuta K. Physical and functional interactions between STAT5 and Runx transcription factors. J Biochem. 2008;143(5):695-709. 18. Choi D, Park E, Jung E, et al. ORAI1 Activates Proliferation of Lymphatic Endothelial Cells in Response to Laminar Flow Through Kruppel-Like Factors 2 and 4. Circ Res. 2017;120(9):1426-1439. 19. Ky N, Lim CB, Li J, Tam JP, Hamza MS, Zhao Y. KLF4 suppresses HDACi induced caspase activation and the SAPK pathway by targeting p57(Kip2). Apoptosis. 2009;14(9):1095-1107. 20. Ter Elst A, Ma B, Scherpen FJ, et al. Repression of vascular endothelial growth factor expression by the runt-related transcription factor 1 in acute myeloid leukemia. Cancer Res. 2011;71(7):27612771.

21. Nishimoto N, Arai S, Ichikawa M, et al. Loss of AML1/Runx1 accelerates the development of MLL-ENL leukemia through down-regulation of p19ARF. Blood. 2011;118(9):2541-2550. 22. Li X, Vradii D, Gutierrez S, et al. Subnuclear targeting of Runx1 is required for synergistic activation of the myeloid specific MCSF receptor promoter by PU.1. J Cell Biochem. 2005;96(4):795-809. 23. Kohrs N, Kolodziej S, Kuvardina ON, et al. MiR144/451 Expression Is Repressed by RUNX1 During Megakaryopoiesis and Disturbed by RUNX1/ETO. PLoS Genet. 2016;12(3):e1005946. 24. Feinberg MW, Wara AK, Cao Z, et al. The Kruppel-like factor KLF4 is a critical regulator of monocyte differentiation. Embo J. 2007;26(18):4138-4148. 25. Buchert M, Darido C, Lagerqvist E, et al. The symplekin/ZONAB complex inhibits intestinal cell differentiation by the repression of AML1/Runx1. Gastroenterology. 2009;137(1):156-164, 164.e1-3. 26. Valk PJ, Verhaak RG, Beijen MA, et al. Prognostically useful gene-expression profiles in acute myeloid leukemia. N Engl J Med. 2004;350(16):1617-1628. 27. Stirewalt DL, Meshinchi S, Kopecky KJ, et al. Identification of genes with abnormal expression changes in acute myeloid leukemia. Genes Chromosomes Cancer. 2008;47(1):8-20. 28. Filarsky K, Garding A, Becker N, et al. Kruppel-like factor 4 (KLF4) inactivation in chronic lymphocytic leukemia correlates with promoter DNA-methylation and can be reversed by inhibition of NOTCH signaling. Haematologica. 2016;101(6):e249253. 29. Yasunaga J, Taniguchi Y, Nosaka K, et al. Identification of aberrantly methylated genes in association with adult T-cell leukemia. Cancer Res. 2004;64(17):60026009. 30. Gu X, Hu Z, Ebrahem Q, et al. Runx1 regulation of Pu.1 corepressor/coactivator exchange identifies specific molecular targets for leukemia differentiation therapy. J Biol Chem. 2014;289(21):14881-14895.

1607

ARTICLE Ferrata Storti Foundation

Acute Lymphoblastic Leukemia

ZEB2 and LMO2 drive immature T-cell lymphoblastic leukemia via distinct oncogenic mechanisms

Steven Goossens,1,2,3 Jueqiong Wang,4 Cedric S. Tremblay,5 Jelle De Medts,6 Sara Tâ&#x20AC;&#x2122;Sas,1,2,3 Thao Nguyen,4 Jesslyn Saw,5 Katharina Haigh,4 David J. Curtis,5 Pieter Van Vlierberghe,1,3 Geert Berx,2,3 Tom Taghon,6 and Jody J. Haigh4,7,8

Department of Biomolecular Medicine, Ghent University, Ghent, Belgium; 2Department for Biomedical Molecular Biology, VIB-UGent Center for Inflammation Research (IRC), Ghent, Belgium; 3Cancer Research Institute Ghent (CRIG), Ghent, Belgium; 4Mammalian Functional Genetics Group, Australian Centre for Blood Diseases, Monash University, Melbourne, VIC, Australia; 5Stem Cell Research Group, Australian Centre for Blood Diseases, Monash University, Melbourne, VIC, Australia; 6Department of Diagnostic Sciences, Ghent University, Ghent, Belgium; 7Department of Pharmacology and Therapeutics, Rady Faulty of Health Sciences, University of Manitoba, Winnipeg, Manitoba, Canada and 8Research Institute in Oncology and Hematology (RIOH), Cancer Care Manitoba, Winnipeg, Manitoba, Canada 1

Haematologica 2019 Volume 104(8):1608-1616

*TT and JJH contributed equally to this work.

ABSTRACT

Correspondence: STEVEN GOOSSENS steven.goossens@ugent.be Received: September 30, 2018. Accepted: January 18, 2019. Pre-published: January 24, 2019. doi:10.3324/haematol.2018.207837 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1608

EB1 and ZEB2 are structurally related E-box binding homeobox transcription factors that induce epithelial to mesenchymal transitions during development and disease. As such, they regulate cancer cell invasion, dissemination and metastasis of solid tumors. In addition, their expression is associated with the gain of cancer stem cell properties and resistance to therapy. Using conditional loss-of-function mice, we previously demonstrated that Zeb2 also plays pivotal roles in hematopoiesis, controlling important cell fate decisions, lineage commitment and fidelity. In addition, upon Zeb2 overexpression, mice spontaneously develop immature T-cell lymphoblastic leukemia. Here we show that pre-leukemic Zeb2overexpressing thymocytes are characterized by a differentiation delay at beta-selection due to aberrant activation of the interleukin-7 receptor signaling pathway. Notably, and in contrast to Lmo2-overexpressing thymocytes, these pre-leukemic Zeb2-overexpressing T-cell progenitors display no acquired self-renewal properties. Finally, Zeb2 activation in more differentiated T-cell precursor cells can also drive malignant T-cell development, suggesting that the early T-cell differentiation delay is not essential for Zeb2mediated leukemic transformation. Altogether, our data suggest that Zeb2 and Lmo2 drive malignant transformation of immature T-cell progenitors via distinct molecular mechanisms.

Introduction ÂŠ2019 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.

1608

T-cell development starts with the migration of bone marrow-derived progenitor cells into the thymus. There, these newly arrived early T-cell progenitors rapidly lose their multipotent character and gradually reprogram into the T-cell lineage. Tcell commitment occurs through an orderly process that is tightly regulated by interplay between key signaling pathways and transcription factors.1 Once committed, immature T-cell progenitors undergo successive and dynamic stages of differentiation, including positive selection for the T-cell receptor complex in the cortex, as well as negative selection for removal of potential self-responsive cells in the medulla.2 Alterations in this process can lead to the development of T-cell acute lymphoblastic leukemia (T-ALL). During malignant transformation, a clonal expansion of immature T cells is selected for via the gradual accumulation of advantageous epigenetic changes and genetic mutations.3,4 E-proteins, E2A and HEB, play pivotal roles in early T-cell commitment, but also haematologica | 2019; 104(8)

Oncogenic mechanisms of ZEB1 and LM02 in T-ALL

at later stages of T-cell differentiation.5,6 These widely expressed basic helix-loop-helix (bHLH) transcription factors cooperate with more tissue/lineage-restricted bHLH proteins, such as LYL1 and SCL/TAL1, to form heterodimers that recognize a single 5’-CANNTG-3’ E-box motif in their target promoters and regulatory elements. As such, E-proteins regulate essential T-cell fate-determining factors including Rag1, Notch1, and the interleukin-7 receptor (Il7r). Only in the presence of the LIM-domainonly (LMO) proteins and LMO-binding protein 1 (LDB1) can bHLH heterodimers form larger multiprotein complexes with a second bHLH heterodimer or with other transcription factors, such as GATA proteins. LMO proteins do not have intrinsic DNA-binding capacity and act solely as essential scaffolding proteins for this multiprotein complex formation to bridge bipartite DNA motifs, e.g. two E-boxes.7,8 Activation of LMO1 and LMO2 genes, which results from chromosomal aberrations such as translocations, deletions or insertions in regulatory elements and promoters, has been recurrently observed in patients with TALL.9-11 Mouse models overexpressing Lmo1 and Lmo2 have demonstrated that both are potent oncogenic drivers within the T-cell lineage.12,13 From 5 months of age, CD2LMO2tg mice spontaneously develop T-ALL with an immature Lyl1+ expression profile.14 Unlike the thymus of normal mice, which is continously replenished by progenitors from the bone marrow, the pre-leukemic thymus of CD2-LMO2tg transgenic mice is self-sustaining.15,16 Similar to other long-lived pre-leukemic stem cells, Lmo2-overexpressing thymocyte precursors retain the ability to differentiate into the full spectrum of mature daughter cells, but in addition, their stem cell properties allow clonal expansion and subsequent acquisition of extra oncogenic driver mutations, eventually leading to the onset of a fully transformed leukemia. The pre-leukemic self-renewal capacity of CD2-LMO2tg thymocytes is restricted to the CD4-CD8double-negative (DN) precursor T cells, more specifically the CD4-CD8-CD44-CD25+ (DN3) subpopulation, and strictly depends on the expression of the bHLH protein LYL1.17 Zinc finger E-box binding homeobox transcription factors, ZEB1 and ZEB2, recognize a similar bipartite E-box motif in their target promoters and regulatory elements.18,19 and as such regulate epithelial-to-mesenchymal transition in the context of progression of solid tumors. In addition, ZEB expression has been correlated with the acquisition of cancer stem cell properties.20,21 Using loss-of-function mouse models, it was previously demonstrated that both ZEB proteins are also essential hematopoietic transcription factors that play pivotal roles at various cell fate decision check points during hematopoiesis,22-24 including the T-cell lineage.24-27 In addition, we recently showed that Zeb2 overexpression can result in spontaneous development of T-ALL with an immature Lyl1+ expression profile21 and a latency similar to that in CD2-LMO2tg/+ mice,14 suggesting a common oncogenic mechanism of action. In these mouse T-ALL, Zeb2 overexpression drives increased expression of Il7r and aberrant activation of the IL7RJAK/STAT signaling pathway.21 Activating IL7R mutations are also recurrently found in T-ALL patients.28 Interestingly, overexpression of gain-of-function mutant variants of IL7R in p19(Arf)-/- mouse hematopoietic progenitors resulted in a similar T-ALL formation with an immature Lyl1+ expression profile, and high levels of haematologica | 2019; 104(8)

Lmo2.29 Based on the high Lmo2 expression in the IL7R mutant tumors and phenotypic similarities with the CD2LMOtg mouse models, the authors suggested that T-ALL initiation in both models might act via converging downstream signaling pathways that result in aberrant preleukemic thymocyte self-renewal.29 In this study, we used transgenic mouse models to further analyze the effects of Zeb1 and Zeb2 overexpression on pre-leukemic T-cell differentiation. In contrast to Zeb1, Zeb2 overexpression resulted in a partial cell-autonomous differentiation delay and accumulation of a DN3 precursor T-cell population, similar to what has been described in the CD2-LMO2tg and IL7R mutant mouse models. However, Zeb2 overexpression was not associated with gain of pre-leukemic self-renewal capacity. Finally, using a late-acting Cre line, we demonstrated that the early T-cell differentiation defects are not essential for Zeb2-mediated T-ALL initiation. Collectively, our data indicate that Zeb2 and Lmo2 drive a similar immature T-ALL subtype, but via distinct oncogenic mechanisms.

Methods Animal experimentation and handling All experiments were performed according to the regulations and guidelines of the ethics committee for care and use of laboratory animals of Ghent University and Monash University. For thymocyte transplantation experiments, donor thymi were dissected under aseptic conditions. Single cells were prepared in cold phosphate-buffered saline using a 40 μM cell strainer. Cell concentrations were measured using a Burker cell counter chamber. Thymocytes (1x 107) were intravenously injected into 6- to 10week old syngeneic Ly5.1 recipients that were irradiated with a sublethal (550 Rad) dose 4 h before the transplant. One day before and 14 days after the irradiation, mice were kept on neomycin prophylaxis delivered at a dose of 1.7 mg/mL in acidified (pH 2.5) drinking water.

Histology Tissue samples were fixed in formalin, embedded in paraffin, sectioned at 6 μm and stained with hematoxylin and eosin (H&E) for histopathological examination as described in detail elsewhere.21

Flow cytometry Both lobes of thymi were carefully dissected in 1 mL of cold phosphate-buffered saline. Single-cell preparations were made using a 40 μm cell strainer and cell numbers were quantified using a Burker cell counting chamber. Cells were stained and analyzed either by LSRII (BD Biosciences) or FACSAria II (BD Biosciences), and FACSDiva or FlowJo software (BD Biosciences). Cell debris and cell aggregates were gated out and dead cells were discarded using the fixable Viability Dye eFluro506 (eBioscience). The antibodies used for flow cytometry are listed in Online Supplementary Table S1. Intracellular staining was done using a BD Cytofix/Cytoperm kit (BD Bioscience) according to the manufacturer’s guidelines.

In vitro differentiation of fetal hematopoietic progenitors towards the T-cell lineage The differentiation experiments were performed as described previously,30 plating 1 x 104 Lin-cKit+ hematopoietic progenitor cells per 24-well plates of OP9-DL1 feeders in OP9 culture medium supplemented with 5 ng/mL Flt3 ligand (R&D Systems) and 5 1609

S. Goossens et al.

ng/mL IL7 (or 1-0.2 ng/mL IL7 where indicated) (R&D Systems or Peprotech). OP9-DL1 bone marrow stromal cells30 were maintained in MEMalpha medium (Gibco), supplemented with 20% fetal bovine serum (FCS; Hyclone), penicillin (100 U/mL)-streptomycin (100 μg/mL), 2 mM L-glutamine (Gibco) and incubated at 37°C with 7% CO2 and 95% humidity. E13.5 fetal liver cells were stained with biotin-conjugated lineage cocktail antibodies [Gr-1 (Ly-6G & 6C), Ter119 (Ly-76), CD3ε, and B220 (CD45R); eBioscience], PE-conjugated streptavidin (BD Biosciences, 1:200) and CD117-APC (cKit, Immunosource, 1:200). Lin-cKit+ cells were sorted using a FACSAria II machine (BD Biosciences) and FACSDiva software. Dead cells were discarded via DAPI stain.

Real-time quantitative polymerase chain reactions Total RNA was isolated using a RNeasy Plus Mini Kit (Qiagen). cDNA was synthesized using the First Strand cDNA Synthesis Kit (Roche) with oligo(dT) primers, starting from equal amounts of RNA. Real-time quantitative polymerase chain reactions (qRTPCR) were performed using the LightCycler 480 SYBR Green I Master (Roche), monitored on a LichtCycler 480 system (Roche) and analyzed using qBase software from Biogazelle. Gene expression was standardized against the housekeeping genes β-actin, glyceraldehyde-3-phosphate dehydrogenase (Gapdh), RPL13 and TBP. All primers used are listed in Online Supplementary Table S2.

Statistical analysis Data are presented as the mean ± standard deviation. Comparisons between two data groups were performed using a two-sided Student t-test. *P<0.05, **P<0.01, ***P<0.001 (vs. control).

Results Overexpression of Zeb2 but not Zeb1 induces spontaneous T-cell acute lymphoblastic leukemia in the mouse We previously demonstrated that hematopoietic-specific overexpression of Zeb2 from the mouse ROSA26 locus resulted in spontaneous T-ALL development.21 To document the functional similarities and/or differences between the two ZEB family members, we recently generated a similar Zeb1 conditional overexpression mouse model using the same targeting strategies as previously described for Zeb231 (Online Supplementary Figure S1A). In these R26-Zeb1tg mice, an aminoterminal HA-tag Zeb1 cDNA, preceded by a floxed transcriptional stop cassette and followed by an IRES-eGFP, was targeted to the ROSA26 locus. The details on how these mice were generated have been submitted for publication elsewhere. To compare the oncogenic potential of Zeb1 and Zeb2, we crossed these newly generated R26-Zeb1tg mice to the same Cre line as we used for the R26-Zeb2tg, the Tie2-cre, which targets hematopoietic stem cells and their progeny.32 This resulted in a moderate 2- to 3-fold overexpression of the total Zeb1 mRNA levels in the thymus, similar to that in the R26-Zeb2tg mice (Online Supplementary Figure S1B). These Tie2cre, R26-Zeb1tg/tg mice (henceforth referred to as R26-Zeb1tg/tg) were born in normal Mendelian ratios and no obvious phenotypes were observed at a young age. While we previously reported that mice overexpressing Tie2-cre, R26-Zeb2tg/tg (henceforth referred to as R26-Zeb2tg/tg) spontaneously develop T-ALL starting from the age of 5 months with a penetrance of 53% at 15 1610

months of age, none of the R26-Zeb1tg/tg-overexpressing mice (n=15) showed any signs of leukemia development (Online Supplementary Figure S1C). These data indicate that, in contrast to Zeb2, Zeb1 is not an oncogenic driver of murine T-ALL.

Overexpression of Zeb2 results in thymic hypoplasia and delayed T-cell development Upon dissection, we noted that juvenile Zeb2-overexpressing mice, and not Zeb1-overexpressing mice, consistently had smaller thymi compared to their littermate controls. Except for a reduction of the medullary areas, no significant changes were observed in thymus architecture (Figure 1A). Next, detailed flow cytometric analysis was performed on thymi of 8-week old R26-Zeb2tg versus control Cre-negative littermates. Upon Zeb2 overexpression, a significant and dose-dependent decrease in total thymocyte numbers was observed, associated with a significant decrease in percentages and absolute numbers of mature CD3+ and CD4+CD8+ double-positive Thy1+ cell populations, combined with increased percentages of the immature CD4-CD8- double negative (DN) Thy1+ population (Figure 1B). These findings indicate that the thymic hypoplasia in R26-Zeb2tg animals results from a partial block or delay at an early stage during T-cell development. Within the R26-Zeb2tg DN-gated cells, an abnormal population of CD44- cells with intermediate levels of CD25 was observed, which was absent in the control thymus (Figure 1C). These cells are Thy1+, CD49b-, B220- and CD19-, confirming their T-cell lineage identity. Positive staining for CD28 and intracellular TCRβ indicated that these are early post-β-selected cells blocked or delayed at the DN3 to DN4 transition, more specifically at DN3C. In line with this notion, expression analysis of mouse and human T-cell populations (Online Supplementary Figure S2) revealed that endogenous ZEB2 levels are normally downregulated at this DN3-4 transition point. Notably, no T-cell differentiation block or delay was seen in the Zeb1-overexpressing mice (Online Supplementary Figure S1D), indicating that Zeb2 overexpression has effects on early T-cell differentiation that differ from those of Zeb1 overexpression.

The Zeb2-mediated differentiation delay is a hematopoietic cell-autonomous phenotype As the Tie2-cre line also targets endothelial cells next to hematopoietic cells, we wanted to exclude that paracrine changes in the thymic microenvironment or architecture cause the delay in T-cell differentiation in R26-Zeb2tg mice. To this end, we performed in vitro T-cell differentiation experiments with purified hematopoietic progenitor cells. Sorted fetal liver hematopoietic progenitors (Lin-, cKit+) from R26-Zeb2tg/tg or Cre-negative control E13.5 embryos were seeded on OP9-DL1 bone marrow stromal cells that express the Notch delta ligand-1,30 in the presence of 5 ng/mL Flt3 ligand and 5 ng/mL IL7. In vitro T-cell development was monitored for 4 weeks. A significant delay in T-cell development was already prominent within 1 week in the Zeb2-overexpressing co-cultures (Figure 2A), exemplified by an increased frequency of cKit+ DN cells. After 2 weeks, an increase of the DN3-like population (CD3-,CD8-,CD44-,CD25int) and a concomitant decrease in more mature cells (CD8+) were also observed, pointing to a hematopoietic cell-autonomous delay or block in early T-cell development upon Zeb2 overexpression (Figure 2A). haematologica | 2019; 104(8)

Oncogenic mechanisms of ZEB1 and LM02 in T-ALL

After 4 weeks of culture, a large proportion of the R26Zeb2tg/+ cells remained DN3 with intermediate levels of CD25, whereas all T cells had already differentiated beyond the DN stage in the control cultures (Figure 2B).

Zeb2-mediated increased interleukin-7 receptor signaling is involved in the T-cell differentiation delay In our previous study, we demonstrated that ZEB2 directly regulates the Il7r promoter.33 Consequently, Zeb2-overexpressing T-ALL tumors and derived cell lines have high levels of Il7r mRNA and are more responsive to exogenous IL7

levels. As downregulation of IL7 signaling, induced by the pre-TCR signal, is crucial for normal transition of DN3 to double-positive cells,34,35 we hypothesized that the Zeb2induced delay in T-cell differentiation is due to their inability to downregulate Il7r. We therefore tested whether lowering the amount of recombinant IL7 in the culture medium could overcome the observed delay in T-cell differentiation in vitro. Indeed, reduction of the concentration of recombinant IL7 from 5 ng/mL to 0.2 ng/mL rescued the delay in differentiation, resulting in similar developmental kinetics to that of the control co-cultures (Figure 2C).

Figure 1. Zeb2 overexpression results in a delay in T-cell development at the DN3 stage. (A) Hematoxylin & eosin-stained sections of paraffin-embedded thymi of mice with and without mono- or bi-allelic ROSA26-mediated overexpression. (B) Flow cytometric analysis of thymi of Zeb2-overexpressing mice versus control littermates. Absolute numbers of total thymocytes and percentages of CD3, CD4/8 double-positive (DP) and CD4/8 double-negative (DN) cell populations, DN1-4 and DN3A-C subpopulations are shown. (C) Representative dot plots of CD4/8 DN populations stained with CD25 and CD44. In R26-Zeb2tg/+ and R26-Zeb2tg/tg thymi, an abnormal DN3/4-like population is observed with intermediate CD25 levels (arrows). *P<0.05, **P<0.01, ***P<0.001 (vs control).

haematologica | 2019; 104(8)

1611

S. Goossens et al.

ZEB2 overexpression does not induce pre-leukemic thymocyte self-renewal For the following reasons, we subsequently hypothesized that thymocytes would also acquire pre-leukemic self-renewal capacity upon Zeb2 overexpression, similar to what has been observed in CD2-LMO2tg mice: (i) mice with Zeb2 overexpression phenocopy the IL7R gain-of-

function mutant and CD2-LMOtg immature T-ALL mouse models; (ii) ZEB2 and LMO2 bind similar bipartite E-boxcontaining regulatory elements; (iii) ZEB2 expression has been associated with acquisition of cancer stem cell properties in solid tumors; and (iv) Zeb2-overexpressing T-ALL tumors and their derived cell lines have increased leukemia-initiating potential. To test our hypothesis, we

Figure 2. Zeb2-mediated T-cell differentiation delay is independent of the thymic mirco-environment. (A,B) Flow cytometric analysis of 1- and 2-week in vitro differentiation cultures of E13.5 fetal liver hematopoietic progenitors on OP9-DL1 feeders. Percentages of DN1 (CD4/8- and CD44+, CD25-), DN2 (CD4/8- and CD44+, CD25+), DN3 (CD4/8-, CD44-, CD25+) and post-DN3 (sCD3 or CD8+) populations show a significant delay in differentiation upon Zeb2 overexpression (A) and the presence of a cell population with intermediate CD25 expression even after 4 weeks of co-culture, as exemplified by a representative dot plot (B). (C) Lowering the concentration of recombinant interleukin 7 (IL7) in OP9-DL1 co-cultures partially rescues the delay in T-cell differentiation of R26-Zeb2tg/tg fetal liver hematopoietic progenitors. Dot plots of representative cultures are shown as are the percentages of T-cell populations following addition of 5 ng/mL, 1 ng/mL and 0.2 ng/mL recombinant IL7. *P<0.05, **P<0.01, ***P<0.001 (vs. control).

1612

haematologica | 2019; 104(8)

Oncogenic mechanisms of ZEB1 and LM02 in T-ALL

backcrossed our R26-Zeb2tg model for ten generations to pure C57BL/6 background (CD45.2+) mice to enable thymocyte transplantation experiments. We injected 1x107 thymocytes from 8-week old R26-Zeb2tg/tg mice or littermate controls into the tail vein of sublethally irradiated syngeneic Ly5.1 (CD45.1+) recipients and used CD2LMO2tg thymocytes as the positive control. Via flow cytometric analysis of CD45.1 versus CD45.2, we analyzed how many of the donor versus recipient cells contributed to the repopulation of the thymus at 4 and 6 weeks after transplantation. Four weeks after transplantation, no CD45.2+ wildtype thymocytes could be detected, while varying low numbers of CD45.2+ Zeb2-overexpressing thymocytes were consistently seen at this time point. However, and in contrast to the CD2-LMO2tg thymocytes, none of the Zeb2-overexpressing CD45.2 thymocytes could be detected 6 weeks after transplantation (Figure 3). These data demonstrate that Zeb2-overexpressing thymocytes have prolonged survival capacity compared to that of wildtype thymocytes, but that they do not have longterm repopulation or self-renewal capacity.

The differentiation delay is not essential for the formation of Zeb2-mediated T-cell acute lymphoblastic leukemia Finally, we wondered whether the observed T-cell differentiation defect in Zeb2-overexpressing mice is essential for Zeb2-mediated T-ALL formation. To this end, we crossed our R26-Zeb2tg mice with the CD4-cre, a late-acting T-cell-specific Cre line that is only active after the above-described DN3C block/delay (Figure 4A). No thymocyte hypoplasia (Figure 4B), or early T-cell differentiation defects could be observed in 8-week old CD4-cre, R26-Zeb2tg/tg mice versus control littermates. Irrespective of the lack of this early T-cell differentiation block, these mice developed leukemia with approximately the same latency and same penetrance (Figure 4C) as observed with the early acting Tie2-cre line. Necropsy and histopathological examination was done on eight Tie2-cre,R26Zeb2tg/tg and five CD4-cre,R26-Zeb2tg/tg mice, as described in our previous study.21 All leukemic mice were diagnosed

with precursor T-cell lymphoblastic leukemia. In general, no immunophenotypic differences were observed between the leukemias developing in the early- versus lateacting Cre line. However, we noted that while all examined Tie2-cre,R26-Zeb2tg/tg mice had large mediastinal masses, two out of the five CD4-cre,R26-Zeb2tg/tg mice had no mediastinal mass. Furthermore, in these two leukemic mice, the neoplastic cells appeared smaller. Since the phenotype in both models is not fully penetrant and we observed large variation in leukemia onset and phenotype, only a limited number (8 vs. 5) leukemic mice could be analyzed fully and we cannot, therefore, draw strong conclusions from our observations.

Discussion In various solid cancer types, ZEB expression has been correlated with both poor prognosis and patientsâ&#x20AC;&#x2122; outcomes. Indeed, ZEB1 and ZEB2 have been demonstrated to increase cancer cell invasion and dissemination via the regulation of epithelial-to-mesenchymal transition. In addition, as for other regulators of epithelial-to-mesenchymal transition, their expression has been associated with the acquisition of cancer stem cell properties and therapy resistance.36 It has been debated whether these structurally very related family members compensate for each other or whether they also have unique functions. The differences in phenotypes between Zeb1 and Zeb2 knockout mice can be explained by their complementary expression patterns.37 Studies with compound Zeb1/Zeb2 double knockouts have shown that they have at least partly overlapping, compensatory functions, but could not rule out that they also have unique functions. Overexpression of ZEB1/ZEB2 in various epithelial cancer cell lines catalyzed similar phenotypes, with overlapping downstream targets.18,38,39 Mild differences between ZEB1/2-mediated epithelial-to-mesenchymal transition were often reported, but these could be attributed to differences in overexpression levels and/or the cellular context. Here, we provide

Figure 3. Zeb2 overexpression does not induce pre-leukemic thymocyte self-renewal. Percentages of CD45.2+ donor-derived thymocytes that contribute to the repopulation of sublethally irradiated syngeneic CD45.1+ recipients 4 or 6 weeks after transplantation.

haematologica | 2019; 104(8)

1613

S. Goossens et al.

compelling evidence, for the first time, that ZEB1 and ZEB2 can have unique functions. Overexpression in the same cellular context, using the same targeting strategy, expressed from the same endogenous promoter and in a relevant in vivo setting, resulted in clearly distinct phenotypes for ZEB1 and ZEB2. Using luciferase reporter constructs, others have previously shown that ZEB1 and ZEB2 also have opposing effects on TGFÎ˛1-mediated repression of the 3TP and p21 promoter elements.40 At that time, the authors hypothesized that these seemingly opposite effects were mediated by the putative differential recruitment of ZEB1/2-specific co-activators/repressors, such as p300 and P/CAF. This differential recruitment of co-factors may specifically switch ZEB1 from a repressor into an activator. However, others have counteracted this hypothesis by demonstrating that ZEB1 and ZEB2 are equally potent at binding p300 and P/CAF. Furthermore, although the extensive list of ZEB-interacting proteins is growing continuously, no unique interactors have been

documented thus far. In this study, we prove that the distinct phenotypes observed are not due to differences in cellular contexts or mRNA (over)expression levels. Unfortunately, no instruments are available to test whether these similar mRNA levels also translate into similar transgene protein levels. Cell-context-dependent differences in post-translational regulation of both ZEB family members may result in more or less of the ZEB1/2 transgene protein and may explain the phenotypic differences observed between the two models. It is important to mention here that although early T-cell development of R26-Zeb1tg mice is normal and these mice do not spontaneously develop leukemia, this does not exclude that other hematopoietic lineages are not affected by the Zeb1 overexpression. In the context of T-cell malignancies, ZEB1 and ZEB2 seem to act in an opposite manner. A tumor suppressor role for ZEB1 in T-cell leukemias/lymphomas has previously been suggested,41 based on its expression, the muta-

Figure 4. Zeb2-mediated T-cell differentiation delay is not essential for Zeb2-induced T-cell acute lymphoblastic leukemia formation. (A) Flow cytometric analysis of the percentages of eGFP+ cells in DN3 versus DP demonstrating differential Cre activity between the early-acting Tie2-cre line and the late-acting CD4-cre line in early T-cell precursors. (B) Flow cytometric analysis of thymi of Zeb2-overexpressing mice versus control littermates upon intercross with either the Tie2-cre or CD4cre line. Absolute numbers of total thymocytes and percentages of CD3, CD4/8 double-positive (DP) and CD4/8 double-negative (DN) cell populations demonstrate that there is no differentiation delay in the late-acting CD4-cre line. (C) Kaplan-Meier curve for leukemia-free survival of CD4-cre, R26-Zeb2tg/tg (n=18) versus Tie2cre, R26-Zeb2tg/tg (n=21) mice. *P<0.05, **P<0.01, ***P<0.001 (vs. control); NS, not significant.

1614

haematologica | 2019; 104(8)

Oncogenic mechanisms of ZEB1 and LM02 in T-ALL

tions found in patients and the observed spontaneous Tcell lymphoma development in a mouse model expressing a carboxyterminal truncated form of ZEB1. This is in contrast with the oncogenic role we have described for Zeb2 in T-ALL.21 In this study, we demonstrated that pre-leukemic T-cell differentiation is also affected by Zeb2 overexpression. Maintenance of Zeb2 expression results in a partial differentiation block or delay at an early DN3C stage, coinciding with the T-cell developmental stage at which Zeb2 expression is normally downregulated in mice, as well as in humans. This delay could be partially rescued by decreasing IL7R pathway activation, suggesting that the delay is caused by an inability of the Zeb2-overexpressing thymocytes to downregulate IL7R expression. Interestingly, other mouse models have shown a similar block in T-cell differentiation at this key transition point, including mice that overexpress the IL7r.35,42 CD2-LMO2tg animals accumulate a similar aberrant DN3 population at a young age. Notably, our detailed flow cytometric analysis revealed that LMO2overexpressing cells are delayed slightly earlier in T-cell differentiation compared to the Zeb2-overexpressing thymocytes, at the DN3A stage, and before T-cell rearrangement (data not shown). As ZEB have been previously associated with the acquisition of cancer stem cell properties, we hypothesized that the cause of spontaneous T-ALL formation in R26-Zeb2tg could be due to acquired self-renewal of pre-leukemic thymocytes, as previously demonstrated for LMO2. This was supported by the fact that both proteins bind similar regulatory elements, and that both factors can drive murine T-ALL development with a similar immature expression profile and increased stem cell properties. Nevertheless, using thymocyte transplantation experiments, we demonstrated that R26-Zeb2tg thymocytes are not able to reconstitute an irradiated thymus and therefore have no pre-

References 1. Rothenberg EV. T cell lineage commitment: identity and renunciation. J Immunol. 2011;186(12):6649-6655. 2. Morris GP, Allen PM. How the TCR balances sensitivity and specificity for the recognition of self and pathogens. Nat Immunol. 2012;13(2):121-128. 3. Meijerink JP. Genetic rearrangements in relation to immunophenotype and outcome in T-cell acute lymphoblastic leukaemia. Best Pract Res Clin Haematol. 2010;23(3):307318. 4. Van Vlierberghe P, Ferrando A. The molecular basis of T cell acute lymphoblastic leukemia. J Clin Invest. 2012;122(10):33983406. 5. Engel I, Murre C. The function of E- and Id proteins in lymphocyte development. Nat Rev Immunol. 2001;1(3):193-199. 6. Wang LH, Baker NE. E Proteins and ID proteins: helix-loop-helix partners in development and disease. Dev Cell. 2015;35(3):269280. 7. Wadman I, Li J, Bash RO, et al. Specific in vivo association between the bHLH and LIM proteins implicated in human T cell leukemia. EMBO J. 1994;13(20):4831-4839.

haematologica | 2019; 104(8)

leukemic self-renewal capacity. These data indicate different oncogenic mechanism for ZEB2 and LMO2. Upon transplantation, we observed that the survival of R26Zeb2tg thymocytes was longer than that of wildtype thymocytes, with this difference most probably being associated with increased IL7R signaling. Whether the increase in Il7r expression is involved in tumor initiation in R26Zeb2tg animals remains to be determined. Finally, we used a late-acting CD4-Cre line to show that Zeb2 could also transform thymic precursor cells at later stages during their development, suggesting that the delay at the DN3C stage is not essential for leukemic transformation. Interestingly, a forward genetic screening used the same CD4-Cre line to activate the â&#x20AC;&#x153;sleeping beauty transposaseâ&#x20AC;? and induce oncogenic transformation specifically at later stages of T-cell development, which resulted in TALL with an immature expression profile.43 These data indicate that cases of immature murine T-ALL can also originate in cells beyond the DN stage. Furthermore, our data suggest that a Zeb2-transformed T-ALL with an immature/stem cell expression profile can originate from a more differentiated cell without self-renewal capacity. In summary, we conclude that multiple oncogenes such as ZEB2 and LMO2 are able to induce subtypes of immature murine T-ALL via distinct oncogenic mechanisms of action. In addition, Zeb2 can also drive T-cell transformation in more differentiated T-cell precursor cells with similar kinetics. Acknowledgments This work was supported by the Australian NHMRC (grant 1047995 to JJH), the Worldwide Cancer Research Fund, the Swiss Bridge Foundation, Stand Up Against Cancer Fund, the Ghent University Special Research Fund and the Fund for Scientific Research-Flanders.

8. Wadman IA, Osada H, Grutz GG, et al. The LIM-only protein Lmo2 is a bridging molecule assembling an erythroid, DNA-binding complex which includes the TAL1, E47, GATA-1 and Ldb1/NLI proteins. EMBO J. 1997;16(11):3145-3157. 9. Boehm T, Foroni L, Kaneko Y, Perutz MF, Rabbitts TH. The rhombotin family of cysteine-rich LIM-domain oncogenes: distinct members are involved in T-cell translocations to human chromosomes 11p15 and 11p13. Proc Natl Acad Sci U S A. 1991;88(10):4367-4371. 10. Van Vlierberghe P, van Grotel M, Beverloo HB, et al. The cryptic chromosomal deletion del(11)(p12p13) as a new activation mechanism of LMO2 in pediatric T-cell acute lymphoblastic leukemia. Blood. 2006;108(10): 3520-3529. 11. Goossens S, Van Vlierberghe P. Novel oncogenic noncoding mutations in T-ALL. Blood. 2017;129(24):3140-3142. 12. Fisch P, Boehm T, Lavenir I, et al. T-cell acute lymphoblastic lymphoma induced in transgenic mice by the RBTN1 and RBTN2 LIMdomain genes. Oncogene. 1992;7(12):23892397. 13. Neale GA, Rehg JE, Goorha RM. Ectopic expression of rhombotin-2 causes selective expansion of CD4-CD8- lymphocytes in the

14.

15.

16.

17.

18.

19.

20.

thymus and T-cell tumors in transgenic mice. Blood. 1995;86(8):3060-3071. Smith S, Tripathi R, Goodings C, et al. LIM domain only-2 (LMO2) induces T-cell leukemia by two distinct pathways. PloS One. 2014;9(1):e85883. McCormack MP, Young LF, Vasudevan S, et al. The Lmo2 oncogene initiates leukemia in mice by inducing thymocyte self-renewal. Science. 2010;327(5967):879-883. Cleveland SM, Smith S, Tripathi R, et al. Lmo2 induces hematopoietic stem cell-like features in T-cell progenitor cells prior to leukemia. Stem Cells. 2013;31(5):882-894. McCormack MP, Shields BJ, Jackson JT, et al. Requirement for Lyl1 in a model of Lmo2driven early T-cell precursor ALL. Blood. 2013;122(12):2093-2103. Comijn J, Berx G, Vermassen P, et al. The two-handed E box binding zinc finger protein SIP1 downregulates E-cadherin and induces invasion. Mol Cell. 2001;7(6):12671278. Remacle JE, Kraft H, Lerchner W, et al. New mode of DNA binding of multi-zinc finger transcription factors: delta EF1 family members bind with two hands to two target sites. EMBO J. 1999;18(18):5073-5084. Wellner U, Schubert J, Burk UC, et al. The EMT-activator ZEB1 promotes tumorigenic-

1615

S. Goossens et al.

21.

22.

23.

24.

25.

26.

27.

1616

ity by repressing stemness-inhibiting microRNAs. Nat Cell Biol. 2009;11(12): 1487-1495. Goossens S, Radaelli E, Blanchet O, et al. ZEB2 drives immature T-cell lymphoblastic leukaemia development via enhanced tumour-initiating potential and IL-7 receptor signalling. Nat Commun. 2015;6:5794. Goossens S, Janzen V, Bartunkova S, et al. The EMT regulator Zeb2/Sip1 is essential for murine embryonic hematopoietic stem/progenitor cell differentiation and mobilization. Blood. 2011;117(21):56205630. Li J, Riedt T, Goossens S, et al. The EMT transcription factor Zeb2 controls adult murine hematopoietic differentiation by regulating cytokine signaling. Blood. 2017;129(4):460-472. Higashi Y, Moribe H, Takagi T, et al. Impairment of T cell development in delta EF1 mutant mice. J Exp Med. 1997;185(8): 1467-1479. Omilusik KD, Best JA, Yu BF, et al. Transcriptional repressor ZEB2 promotes terminal differentiation of CD8(+) effector and memory T cell populations during infection. J Exp Med. 2015;212(12):2027-2039. Dominguez CX, Amezquita RA, Guan TX, et al. The transcription factors ZEB2 and Tbet cooperate to program cytotoxic T cell terminal differentiation in response to LCMV viral infection. J Exp Med. 2015;212(12):2041-2056. Guan T, Dominguez CX, Amezquita RA, et al. ZEB1, ZEB2, and the miR-200 family form a counterregulatory network to regulate CD8(+) T cell fates. J Exp Med. 2018;215(4):1153-1168.

28. Zhang J, Ding L, Holmfeldt L, et al. The genetic basis of early T-cell precursor acute lymphoblastic leukaemia. Nature. 2012;481 (7380):157-163. 29. Treanor LM, Zhou S, Janke L, et al. Interleukin-7 receptor mutants initiate early T cell precursor leukemia in murine thymocyte progenitors with multipotent potential. J Exp Med. 2014;211(4):701-713. 30. Schmitt TM, de Pooter RF, Gronski MA, Cho SK, Ohashi PS, Zuniga-Pflucker JC. Induction of T cell development and establishment of T cell competence from embryonic stem cells differentiated in vitro. Nat Immunol. 2004;5(4):410-417. 31. Tatari MN, De Craene B, Soen B, et al. ZEB2-transgene expression in the epidermis compromises the integrity of the epidermal barrier through the repression of different tight junction proteins. Cell Mol Life Sci. 2014;71(18):3599-3609. 32. Kisanuki YY, Hammer RE, Miyazaki J, Williams SC, Richardson JA, Yanagisawa M. Tie2-Cre transgenic mice: a new model for endothelial cell-lineage analysis in vivo. Dev Biol. 2001;230(2):230-242. 33. Goossens S, Radaelli E, Blanchet O, et al. ZEB2 drives immature T-cell lymphoblastic leukaemia development via enhanced tumour-initiating potential and IL-7 receptor signalling. Nat Commun. 2015;6:5794. 34. El Kassar N, Lucas PJ, Klug DB, et al. A dose effect of IL-7 on thymocyte development. Blood. 2004;104(5):1419-1427. 35. Munitic I, Williams JA, Yang Y, et al. Dynamic regulation of IL-7 receptor expression is required for normal thymopoiesis. Blood. 2004;104(13):4165-4172. 36. Goossens S, Vandamme N, Van Vlierberghe

37.

38.

39.

40.

41.

42.

43.

P, Berx G. EMT transcription factors in cancer development re-evaluated: beyond EMT and MET. Biochim Biophys Acta Rev Cancer. 2017;1868(2):584-591. Miyoshi T, Maruhashi M, Van De Putte T, Kondoh H, Huylebroeck D, Higashi Y. Complementary expression pattern of Zfhx1 genes Sip1 and deltaEF1 in the mouse embryo and their genetic interaction revealed by compound mutants. Dev Dyn. 2006;235(7):1941-1952. Vandewalle C, Comijn J, De Craene B, et al. SIP1/ZEB2 induces EMT by repressing genes of different epithelial cell-cell junctions. Nucleic Acids Res. 2005;33(20):6566-6578. Eger A, Aigner K, Sonderegger S, et al. DeltaEF1 is a transcriptional repressor of Ecadherin and regulates epithelial plasticity in breast cancer cells. Oncogene. 2005;24(14): 2375-2385. Postigo AA. Opposing functions of ZEB proteins in the regulation of the TGF beta/BMP signaling pathway. EMBO J. 2003;22(10): 2443-2452. Hidaka T, Nakahata S, Hatakeyama K, et al. Down-regulation of TCF8 is involved in the leukemogenesis of adult T-cell leukemia/lymphoma. Blood. 2008;112(2): 383-393. Trigueros C, Hozumi K, Silva-Santos B, et al. Pre-TCR signaling regulates IL-7 receptor alpha expression promoting thymocyte survival at the transition from the double-negative to double-positive stage. Eur J Immunol. 2003;33(7):1968-1977. Berquam-Vrieze KE, Nannapaneni K, Brett BT, et al. Cell of origin strongly influences genetic selection in a mouse model of TALL. Blood, 2011;118(17):4646-4656.

haematologica | 2019; 104(8)

ARTICLE

Acute Lymphoblastic Leukemia

DNMT3A mutation is associated with increased age and adverse outcome in adult T-cell acute lymphoblastic leukemia

Ferrata Storti Foundation

Jonathan Bond,1,2,3 Aurore Touzart,1 Stéphane Leprêtre,4 Carlos Graux,5 Mario Bargetzi,6,7 Ludovic Lhermitte,1 Guillaume Hypolite,1 Thibaut Leguay,8 Yosr Hicheri,9 Gaëlle Guillerm,10 Karin Bilger,11 Véronique Lhéritier,12 Mathilde Hunault,13 Françoise Huguet,14 Yves Chalandon,6,15 Norbert Ifrah,13 Elizabeth Macintyre,1 Hervé Dombret,16 Vahid Asnafi1 and Nicolas Boissel16

Université Paris Descartes Sorbonne Cité, Institut Necker-Enfants Malades (INEM), Institut National de Recherche Médicale (INSERM) U1151, and Laboratory of OncoHematology, Assistance Publique-Hôpitaux de Paris (AP-HP), Hôpital Necker EnfantsMalades, Paris, France; 2Systems Biology Ireland, School of Medicine, University College Dublin, Ireland; 3National Children's Research Centre, Children's Health Ireland at Crumlin, Dublin, Ireland; 4INSERM U1245 and Department of Hematology, Centre Henri Becquerel and Normandie Université UNIROUEN, Rouen, France; 5Department of Hematology, Université Catholique de Louvain (UCL), Centre Hospitalier Universitaire (CHU) Namur - Godinne site, Yvoir, Belgium; 6University Medical Department, Division of Oncology, Hematology and Transfusion Medicine, Kantonsspital Aarau, Aarau, Switzerland; 7Swiss Group for Clinical Cancer Research (SAKK), Bern, Switerland; 8 Department of Hematology, CHU de Bordeaux, France; 9Hematology Service, Hôpital St Eloi, Montpellier, France; 10Hematology Service, CHU de Brest, Brest, France; 11 Hematology Service, CHU Hautepierre, Strasbourg, France; 12Group for Research on Adult Acute Lymphoblastic Leukemia, Coordination Office, Centre Hospitalier Lyon Sud, Lyon, France; 13PRES LUNAM, CHU Angers Service des Maladies du Sang and CRCINA INSERM, Angers, France; 14Department of Hematology, CHU de Toulouse, Institut Universitaire du Cancer de Toulouse Oncopole, Toulouse, France; 15Department of Oncology, Hematology Division, University Hospital, Geneva, Switzerland and 16Université Paris Diderot, Institut Universitaire d’Hématologie, EA-3518, AP-HP, University Hospital Saint-Louis, Paris, France 1

Haematologica 2019 Volume 104(8):1617-1625

Correspondence: ABSTRACT

he prognostic implications of DNMT3A genotype in T-cell acute lymphoblastic leukemia are incompletely understood. We performed comprehensive genetic and clinico-biological analyses of T-cell acute lymphoblastic leukemia patients with DNMT3A mutations treated during the GRAALL-2003 and -2005 studies. Eighteen of 198 cases (9.1%) had DNMT3A alterations. Two patients also had DNMT3A mutations in nonleukemic cell DNA, providing the first potential evidence of age-related clonal hematopoiesis in T-cell acute lymphoblastic leukemia. DNMT3A mutation was associated with older age (median 43.9 years vs. 29.4 years, P<0.001), immature T-cell receptor genotype (53.3% vs. 24.4%, P=0.016) and lower remission rates (72.2% mutated vs. 94.4% non-mutated, P=0.006). DNMT3A alterations were significantly associated with worse clinical outcome, with higher cumulative incidence of relapse (HR 2.33, 95% CI: 1.05-5.16, P=0.037) and markedly poorer event-free survival (HR 3.22, 95% CI: 1.81-5.72, P<0.001) and overall survival (HR 2.91, 95% CI: 1.56-5.43, P=0.001). Adjusting for age as a covariate, or restricting the analysis to patients over 40 years, who account for almost 90% of DNMT3Amutated cases, did not modify these observations. In multivariate analysis using the risk factors that were used to stratify treatment during the GRAALL studies, DNMT3A mutation was significantly associated with shorter event-free survival (HR 2.33, 95% CI: 1.06 – 4.04, P=0.02). Altogether, these results identify DNMT3A genotype as a predictor of aggressive T-cell acute lymphoblastic leukemia biology. The GRAALL-2003 and -2005 studies were registered at http://www.ClinicalTrials.gov as #NCT00222027 and #NCT00327678, respectively. haematologica | 2019; 104(8)

NICOLAS BOISSEL nicolas.boissel@aphp.fr Received: May 24, 2018. Accepted: January 10, 2019. Pre-published: January 17, 2019. doi:10.3324/haematol.2018.197848 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1617 ©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.

1617

J. Bond et al.

Introduction Mutations in the DNA methyltransferase 3 alpha gene (DNMT3A) have been reported in a range of hematologic malignancies, most frequently in myeloid neoplasia, including acute myeloid leukemia (AML),1-5 myelodysplastic syndromes,6 myeloproliferative neoplasms7 and myeloproliferative neoplasm/myelodysplastic overlap syndromes.8,9 DNMT3A alterations in lymphoid malignancies are less common, and reports to date are confined to T-lineage disease.10-16 In all cases, DNMT3A mutations increase in frequency with age, and are extremely rare in children and adolescents.17-19 Multiple studies have reported that DNMT3A alterations correlate with poor outcome in AML.1,2,4,20-22 In comparison, the prognostic influence of DNMT3A mutation in T-cell acute lymphoblastic leukemia (T-ALL) is poorly characterized. Patients with DNMT3A alterations were reported to have shorter survival in three moderately sized (55 to 93 patients) T-ALL cohorts.9,11,13 DNMT3A status did not however independently predict prognosis in the only series for which multivariate analyses were documented, as survival effects were linked to increased rates of DNMT3A mutation in poor-risk, phenotypically immature disease.11 While that study did document a correlation between DNMT3A alteration and survival within the immature T-ALL subgroup, this finding was not corroborated in an independent cohort of early thymic precursor (ETP) ALL cases.12 The issue of whether DNMT3A mutation truly alters the biology of T-ALL is therefore only partially addressed by the currently available evidence. In particular, it is unclear whether the associated poor survival simply reflects the prosaic fact that patients with DNMT3A alterations are older,11,12 and therefore do not tolerate intensive ALL treatment as well as their younger counterparts. In order to address this question, we used next-generation sequencing (NGS) to evaluate the DNMT3A genotype of a large cohort of 198 adult T-ALL patients treated as part of the multinational GRAALL-2003 and -2005 studies. We found that DNMT3A mutation strongly correlated with disease relapse and shorter survival, and that these prognostic effects were independent of patientsâ&#x20AC;&#x2122; age. Furthermore, we report the presence of DNMT3A mutations in nonleukemic cells in a subset of patients, providing the first evidence of age-related clonal hematopoiesis in T-ALL.

Methods Patients Details of the GRAALL-2003 and -2005 studies are provided in the Online Supplementary Methods. Informed consent was obtained from all patients before inclusion into the trials. Both studies were conducted in accordance with the Declaration of Helsinki and approved by local and multicenter research ethical committees. The complete study protocols are detailed in the Online Data Supplement. Both trials were registered at http:/www.ClinicalTrials.gov (NCT00222027, NCT00327678). The criteria for inclusion in the current project were a diagnosis of TALL and the availability of diagnostic material for NGS analysis of DNMT3A genotype. Survival outcomes of the 198 patients (36 from GRAALL-2003 and 162 from GRAALL-2005) who fulfilled these criteria did not differ from those of the remaining 139 T-ALL patients of the study cohorts. As expected in retro1618

spective studies, initial white blood cell count (WBC) was higher in the study cohort. However, no differences in allogeneic stem cell transplant rate, disease-free survival, event-free survival, or overall survival were found. A full comparison of the clinical features of each group is shown in Online Supplementary Table S1.

Next-generation sequencing Nextera XT (Illumina) DNA Libraries were prepared according to the manufacturerâ&#x20AC;&#x2122;s instructions and sequenced using the Illumina MiSeq sequencing system. The custom NGS panel comprised genes coding for factors involved in molecular pathways known to be mutated in T-ALL, namely cytokine receptor and RAS signaling (NRAS, KRAS, JAK1, JAK3, STAT3, STAT5B, IL7R, BRAF, NF1, SH2B3, PTPN11), hematopoietic development (RUNX1, ETV6, GATA3, IKZF1, EP300), chemical modification of histones (SUZ12, EED, EZH2, KMT2A, KMT2D, SETD2) and DNA methylation (DNMT3A, IDH1, IDH2, TET2, TET3). This panel was originally inspired by the repertoire of genes found to be preferentially altered in pediatric ETP-ALL,23 and we have reported a subset of the results described in the current paper in a previous clinico-biological and genetic analysis of adult ETPALL.24 Sequencing reads were analyzed using in-house software (Polyweb, Institut Imagine, Paris, France), and additional inhouse custom filtering criteria (comprising minimum read counts and variant allele frequencies, and reference to external reference databases) were applied to minimize false-positive rates. Primers used to confirm mutations by direct sequencing are listed in Online Supplementary Table S2.

Outcome analyses Comparisons between groups were performed with the Fisher exact and Mann-Whitney tests for categorical and continuous variables respectively. Corticosteroid sensitivity was defined as clearance of peripheral blood circulating blasts (<1 x 109/L) following steroid prophase treatment. Complete remission was defined as clearance of bone marrow blasts (<5%) following induction treatment. Overall survival was calculated from the date of inclusion in the trial to the last follow-up date, censoring patients alive at that date. Event-free survival was calculated from date of inclusion in the trial to the date of induction failure, relapse, or death, censoring patients alive in first complete remission without relapse at the last follow-up date. Cumulative incidence of relapse was calculated in patients who attained complete remission, from the date of achieving the complete remission to the date of relapse, with death in first complete remission being considered as a competing event. Univariate and bivariate analyses assessing the impact of DNMT3A mutations and age were performed with a Cox model. Variables that were significantly associated with outcome in univariate analysis were considered as covariates in multivariate Cox models. The proportional-hazards assumption was checked before conducting multivariate analyses. Statistical analyses were performed with STATA software (STATA 12.0 Corporation, College Station, TX, USA). All P values were twosided, with P<0.05 denoting statistical significance.

Results Analysis of DNMT3A genotype in patients with T-cell acute lymphoblastic leukemia in the GRAALL studies We performed targeted NGS of a panel of genes, including DNMT3A, which have been described to be recurrently mutated in T-ALL. This panel included all exons of DNMT3A, thereby providing a comprehensive picture of haematologica | 2019; 104(8)

DNMT3A mutation predicts adverse outcome in adult T-ALL

the spectrum of alterations across this gene in T-ALL. Diagnostic DNA was available for 198 patients treated during the GRAALL-2003 and -2005 studies. A partial analysis of a subgroup of this cohort has been reported previously.24 We detected 21 DNMT3A mutations in 18 patients (9.1%). Most alterations occurred in regions coding for defined protein functional domains, including six mutations at the R882 hotspot1 (Figure 1A). Further details of patient-specific alterations are shown in Online Supplementary Table S3. Of note, the vast majority of detected mutations are predicted to be significantly damaging to protein function. In keeping with previous reports of DNMT3A-mutated

T-ALL which cited high rates of either compound heterozygosity or homozygosity,9,11 a significant proportion of cases (8/18) had either two separate alterations, or high variant allele frequencies that were suggestive of either homozygous mutation, concomitant deletion of the wildtype (WT) allele or copy-neutral loss of heterozygosity. Comparative genomic hybridization analyses were available for 85 of the cases in this study, including 6/18 patients with DNMT3A mutations. We detected only two deletions of the DNMT3A locus, which in each case were associated with concomitant DNMT3A mutation and elevated variant allele frequencies (cases 11 and 12 in Online Supplementary Table S3).

Figure 1. DNMT3A mutations in T-cell acute lymphoblastic leukemia. (A) Schematic representation of the 21 mutations detected in this study. Further patient-specific details are provided in Online Supplementary Table S3. (B) Comparison of the mutational genotypes of DNMT3A altered (n=18) and DNMT3A wild-type (n=180) T-cell acute lymphoblastic leukemia. Percentage frequencies in each group are depicted. Functional categories are listed in bold.

haematologica | 2019; 104(8)

1619

J. Bond et al.

The prevalence of other mutations detected by NGS is shown in Figure 1B. DNMT3A-altered cases had an increased frequency of alterations in other genes included in the NGS panel, compared with the rest of the cohort (88.9% DNMT3A mutated vs. 64.4% DNMT3A WT, P=0.036). There were no statistically significant differences in the prevalence of mutations in any specific functional gene category, namely factors involved in cytokine receptor and RAS signaling (61.1% DNMT3A mutated vs. 41.7% DNMT3A WT, P=0.113), hematopoiesis (38.5% DNMT3A mutated vs. 12.8% DNMT3A WT, P=0.082) and chemical modification of histones (50.0% DNMT3A mutated vs. 30.0% DNMT3A WT, P=0.082). However, we did observe significant co-occurrence of DNMT3A alterations and IDH2 mutations (27.8% DNMT3A mutated vs. 2.2% DNMT3A WT, P<0.001). This association has been described previously in both AML25 and myelodysplastic syndromes.26

Evidence of possible clonal hematopoiesis in DNMT3Amutated T-cell acute lymphoblastic leukemia DNMT3A is the most commonly altered gene in agerelated clonal hematopoiesis,27-29 and DNMT3A mutations have been detected in non-malignant cells in AML30,31 and peripheral T-cell lymphoma.14 We therefore tested whether DNMT3A mutations were present in nonleukemic hematopoietic cells in T-ALL patients. DNA from remission bone marrow was available for only three of the 18 patients with DNMT3A mutations. While two of these samples had a DNMT3A WT genotype (Online Supplementary Figure S1), one interesting case had evidence of DNMT3A alterations in non-leukemic bone marrow cells. At diagnosis, this patient (case 6 in Online Supplementary Table S3) had mutations in exons 14 and 15 of DNMT3A, a NOTCH1 PEST domain insertion, and an NRAS G12D substitution. Sequencing of remission DNA revealed mutation of DNMT3A exon 14 in nonleukemic cells, while NOTCH1, NRAS and DNMT3A exon 15 all presented wild-type genotypes (Figure 2A). We confirmed that the exon 14 mutation has never been reported as a polymorphic variant, while SIFT analysis (http://sift.jcvi.org/) predicted this M548T substitution to be highly deleterious to protein function, with a SIFT score of 0. These results suggest that this T-ALL may have developed on a background of DNMT3A-mutated clonal hematopoiesis, and that the other genetic alterations, including the second DNMT3A mutation, were acquired at leukemic transformation. In order to extend this analysis of non-leukemic DNMT3A mutation, we performed immunophenotypic sorting of two further diagnostic bone marrow samples, and extracted DNA from both the leukemic and the minor residual non-leukemic fractions. We detected a mutation in non-leukemic DNA in one patient (case 4 in Online Supplementary Table S3). Again, we confirmed that this mutation has not been reported as a polymorphism, and that the resultant P385L substitution is predicted to damage protein function, with a SIFT score of 0.02. Similar to the case with mutated remission DNA, this sample was negative for two NOTCH1 alterations detected at T-ALL diagnosis, confirming the specificity of DNMT3A mutation persistence (Figure 2B). The other tested nonleukemic DNA had a DNMT3A WT genotype (Online Supplementary Figure S2), giving an overall rate of nonleukemic DNMT3A mutant positivity of 2/5 samples from the GRAALL-2003 and -2005 studies. We also tested a fur1620

ther three T-ALL cases not included in this cohort, but found no evidence of non-leukemic DNMT3A mutation. This gives an overall incidence of possible clonal hematopoiesis in 2/8 T-ALL samples assessed in our laboratory. It was unfortunately not possible to obtain nonhematopoietic tissue from either of these patients, in order to exclude that these alterations were not constitutional, and to confirm definitively that these results reflect the persistence of a DNMT3A-mutated clonal hematopoietic population in these cases.

DNMT3A mutations are associated with older age and treatment resistance A clinico-biological comparison of cases with and without DNMT3A mutations is shown in Table 1. In keeping with previous reports,11,12 patients with mutations were considerably older than the rest of the T-ALL cohort (median age 43.9 years mutated vs. 29.4 years non-mutated, P<0.001). In addition, DNMT3A-mutated leukemias were more likely to have an immature T-receptor genotype32 (53.3% mutated vs. 24.4% non-mutated, P=0.016), although this did not correspond to a significantly higher incidence of an ETP-ALL immunophenotype33 (35.7% mutated vs. 20.3% non-mutated, P=0.184). DNMT3A mutation was notably associated with poor initial treatment response. We observed trends towards early corticosteroid resistance (66.7% mutated vs. 43.3% non-mutated, P=0.081) and induction failure (13.3% vs. 2.9%, P=0.096), and patients with DNMT3A mutations had significantly higher rates of death during induction (16.7% vs. 2.8%, P=0.027), and lower attainment of complete remission (72.2% mutated vs. 94.4% non-mutated, P=0.006). As only four patients with mutations were evaluated for minimal residual disease, we could not verify that molecular remission was similarly compromised. We found that the type of DNMT3A mutation did not significantly correlate with any individual clinico-biological parameter, suggesting that the alterations detected in this study are likely to have broadly similar biological consequences.

DNMT3A mutation correlates with poor outcome in T-cell acute lymphoblastic leukemia The median follow-up of the cohort was 5.5 years. Prognostic analyses revealed that DNMT3A mutation was associated with an increased 5-year cumulative incidence of relapse (53.9% mutated vs. 28.7% non-mutated, P=0.037) (Figure 3A) and with 5-year event-free survival [27.8% mutated vs. 61.0% non-mutated; hazard ratio (HR) 3.22, (95% confidence interval (95% CI): 1.81-5.72, P<0.001] (Figure 3B). Patients with DNMT3A mutations also had a markedly inferior 5-year overall survival (38.8% mutated vs. 68.7% non-mutated, HR 2.91, 95% CI: 1.565.43, P=0.001) (Figure 3C).

The poor prognosis of DNMT3A-mutated T-cell acute lymphoblastic leukemia is age-independent Our and othersâ&#x20AC;&#x2122; data11,12 have shown that the incidence of DNMT3A mutation in T-ALL increases with age, but previous reports have not documented whether this factor contributes to prognosis. As older patients treated during the GRAALL studies had worse outcomes due to impaired tolerance of intensive chemotherapy,34 we considered it critical to determine to what extent age was a confounding prognostic variable. haematologica | 2019; 104(8)

DNMT3A mutation predicts adverse outcome in adult T-ALL

We therefore performed bivariate analyses of the effects of DNMT3A mutations and age across a series of outcome measures. These results are shown in Online Supplementary Table S4. In each case, DNMT3A genotype was still associated with significantly increased cumulative incidence of relapse (HR 2.80, 95% CI: 1.12-6.97, P=0.034), and shorter event-free survival (HR 2.62, 95% CI: 1.45â&#x20AC;&#x201C;5.06, P=0.004) and overall survival (HR 2.05, 95% CI: 1.02-4.12, P=0.043). Since DNMT3A alterations were almost exclusively found in patients >40 years (16/18 cases), we also performed survival analyses that were restricted to the >40year old subgroup, which constituted a quarter of the total cohort of patients (50/198, 25.3%). Consistent with the

results of the bivariate analyses, DNMT3A mutation was associated with significantly worse 5-year cumulative incidence of relapse (58.3% mutated vs. 21.7% nonmutated, HR 3.90, 95% CI: 1.30-11.68, P=0.015) (Figure 4A), 5-year event-free survival (25.0% mutated vs. 56.7% non-mutated, HR 2.95, 95% CI: 1.37-6.32, P=0.005) (Figure 4B), and 5-year overall survival (37.5% mutated vs. 62.1% non-mutated, HR 2.35, 95% CI: 1.05-5.26, P=0.038) (Figure 4C). Finally, we carried out multivariate outcome analyses in the whole cohort using the risk factors that were used to stratify treatment during the GRAALL-2003 and -2005 studies, and which were found to significantly predict prognosis in the univariate analyses. Among age, log(WBC),

Figure 2. Evidence of DNMT3A mutations in non-leukemic DNA. (A) Direct sequencing of DNMT3A exons 14 and 15, NOTCH1 and NRAS in diagnostic (left panels) and remission (right panels) samples. (B) Mutational assessment of DNA extracted from leukemic and non-leukemic fractions of samples from patients with T-cell acute lymphoblastic leukemia. Sequencing results of DNMT3A and NOTCH1 in leukemic (left panels) and non-leukemic (right panels) DNA are shown. Cases are numbered according to the listing in Online Supplementary Table S3.

haematologica | 2019; 104(8)

1621

J. Bond et al.

corticosteroid sensitivity, early chemosensitivity, and DNMT3A genotype, only DNMT3A genotype was associated with cumulative incidence of relapse in univariate analysis (data not shown). As shown in Tables 2 and 3, age, log(WBC), corticosteroid resistance along with DNMT3A genotype were significantly associated with a poor eventfree survival and overall survival. In multivariate analysis adjusting for these covariates, DNMT3A mutation was still significantly associated with shorter event-free survival (HR 2.33, 95% CI: 1.06–4.04, P=0.02) (Table 2), although not with overall survival (HR 1.66, 95% CI: 0.82–3.37, P=0.16) (Table 3). Taken together, these results provide strong evidence that DNMT3A mutation, while mostly observed in older cases, predicts a poor prognosis that is not related to the patient’s age.

Discussion To our knowledge, this is the most extensive study of DNMT3A-mutated T-ALL yet reported. Our targeted NGS approach allowed comprehensive assessment of genotype across the entire DNMT3A locus, along with the prevalence of co-occurring genetic alterations. Our data additionally benefit from the analysis of a large cohort of patients who were uniformly treated as part of the GRAALL-2003 and 2005 studies, thereby allowing rigorous outcome comparisons between mutated and wild-type cases. Some of our results were expected, and the findings that

DNMT3A mutations are more commonly present in older patients and genotypically immature leukemias are consistent with previously published data.9,11-13 We did not, however, observe increased rates of ETP-ALL immunophenotype, as might have been predicted. We did not detect a clear association with any other geneticallydefined subgroup, and there was no link to increased HOXA expression, which we have previously shown to predict outcome in immature T-ALL.35 The detection of DNMT3A alterations in non-leukemic bone marrow suggests that some of these cases of T-ALL might have arisen from DNMT3A-mutated clonal hematopoiesis. While pre-leukemic NOTCH1 mutations have been detected in neonatal blood spot samples of pediatric patients with T-ALL,36 to our knowledge our data provide the first potential evidence of age-related clonal hematopoiesis in T-ALL. As it was not possible to obtain non-hematopoietic tissue from either of the patients with this finding, we cannot definitively exclude that these alterations are constitutional, or might even represent an inherited cancer predisposition. Further work is necessary to investigate the incidence of clonal hematopoiesis linked to alterations in DNMT3A and other genes in T-ALL. Non-leukemic DNMT3A mutations have been seen in AML,30,31 and it has been postulated that DNMT3A-altered immature T-ALL might arise from malignant transformation of a multipotent myeloid/lymphoid progenitor cell.12 In keeping with this, one of the cases with a non-leukemic DNMT3A mutation in this study had expression of myeloid cell surface markers as part of an ETP-ALL phe-

Table 1. Characteristics and outcome of the patients according to DNMT3A genotype. Total (%) Clinical subsets analyzed Male Median age (years)[IQR] WBC (109/L, median) CNS involvement T-cell receptor status Immature (IM0, IMD, IMG#) αβ lineage γδ lineage ETP immunophenotype# Oncogenetics HOXA positivity# NOTCH1/ FBXW7 mutated RAS/ PTEN mutated Risk classifier, high# Early treatment response Corticosteroid sensitivity Complete remission Induction death Induction failure 5-year treatment outcome Cumulative incidence of relapse Event-free survival Overall survival

DNMT3A Mutated

DNMT3A Wild-type

Total

P-value

18 (9.1%)

180 (90.9%)

198 (100%)

13 (72.2%) 43.9 [40.7–53.6] 41.1 3 (16.7%)

128 (71.1%) 29.4 [23.2–37.2] 31.9 21 (11.7%)

141 (71.2%) 30.5[23.4-40.4] 32.6 24 (12.1%)

0.921 <0.001 0.491 0.463

8 (53.3%) 3 (20.0%) 4 (26.7%) 5 (35.7%)

38 (24.4%) 104 (66.7%) 14 (9.0%) 32 (20.3%)

46 (26.9%) 107 (62.6%) 18 (10.5%) 37 (18.7%)

0.015 <0.001 0.033 0.184

4 (25.0%) 15 (83.3%) 5 (29.4%) 8 (44.4%)

41 (26.6%) 124 (68.9%) 33 (19.4%) 74 (42.3%)

45 (26.5%) 139 (70.2%) 38 (20.3%) 82 (42.5%)

1.000 0.282 0.365 1.000

6 (33.3%) 13 (72.2%) 3 (16.7%) 2/15 (13.3%)

102 (56.7%) 170 (94.4%) 5 (2.8%) 5/175 (2.9%)

108 (54.5%) 183 (92.4%) 8 (4.0%) 7/190 (3.7%)

0.081 0.006 0.027 0.097

53.9% 27.8% 38.8%

28.7% 61.0% 68.7%

30.5% 58% 66%

0.037 <0.001 0.001

# T-cell receptor status (n=171), early thymic precursor (ETP) immunophenotype (n=172), HOXA positivity (n=170) and Risk classifier based on NOTCH1, FBXW7, PTEN, NRAS and KRAS genotypes (n=193) were determined as previously described.32,33,35,37 For the Risk classifier, numbers categorized as high risk (NOTCH1/FBXW7 WT and/or NRAS/KRAS/PTEN altered) are shown. Statistically significant results are shown in bold.

1622

haematologica | 2019; 104(8)

DNMT3A mutation predicts adverse outcome in adult T-ALL

Figure 3. DNMT3A mutation correlates with poor outcome in T-cell acute lymphoblastic leukemia. Comparisons of outcomes for patients with (n=18) and without (n=180) DNMT3A mutations are shown for: (A) cumulative incidence of relapse; (B) event-free survival; and (C) overall survival. The 5-year results were as follows: cumulative incidence of relapse 53.9% mutated vs. 28.7% nonmutated; event-free survival 27.8% mutated vs. 61.0% non-mutated; overall survival 38.8% mutated vs. 68.7% non-mutated. P values are indicated.

Figure 4. DNMT3A genotype predicts outcome in the age group of patients at risk of mutation. Comparisons of outcomes for patients with (n=16) and without (n=34) mutations in patients >40 years are shown for: (A) cumulative incidence of relapse; (B) event-free survival; and (C) overall survival. The 5-year results were as follows: cumulative incidence of relapse 58.3% mutated vs. 21.7% nonmutated; event-free survival, 25.0% mutated vs. 56.7% non-mutated; overall survival 37.5% mutated vs. 62.1% non-mutated. P values are indicated.

haematologica | 2019; 104(8)

1623

J. Bond et al. Table 2. Prognostic impact of DNMT3A genotype on event-free survival.

EFS Age* Log(WBC)* Corticosteroid sensitivity Early chemosensitivity DNMT3A mutation

Univariate 95% CI

Multivariate 95% CI

1.03 1.62 0.52 0.90 3.22

1.01 – 1.05 1.12 – 2.34 0.34 – 0.81 0.68 – 1.18 1.81 – 5.72

0.009 0.011 0.003 0.436 <0.001

1.02 1.50 0.66 2.20

1.00 – 1.04 0.98 – 2.29 0.41 – 1.07 1.13 – 4.27

0.071 0.062 0.093 0.02

*Continuous variable. Statistically significant differences are highlighted in bold. EFS: event-free survival; HR: hazard ratio; 95% CI: 95% confidence interval; WBC: white blood cell count..

Table 3. Prognostic impact of DNMT3A genotype on overall survival.

OS Age* Log(WBC)* Corticosteroid sensitivity Early chemosensitivity DNMT3A mutation

Univariate 95% CI

Multivariate 95% CI

1.04 1.65 0.59 0.94 2.91

1.01 – 1.06 1.10 – 2.46 0.37 – 0.94 0.71 – 1.24 1.56 – 5.43

0.002 0.015 0.027 0.640 0.001

1.03 1.63 0.79 1.66

1.01 – 1.06 1.03 – 2.57 0.47 – 1.34 0.82 – 3.37

0.009 0.037 0.388 0.160

*Continuous variable. Statistically significant differences are highlighted in bold. OS: overall survival; HR: hazard ratio; 95% CI: 95% confidence interval; WBC: white blood cell count..

notype, while full immunophenotypic assessment was unfortunately not possible for the other patient. The factors that may dictate the acute leukemic phenotype in clonally mutated cases remain to be clarified. For example, this might be influenced by the differentiation capacity of the cell in which the initial DNMT3A mutation occurs. In addition, it is tempting to speculate that the acquisition of specific cooperative mutations, such as the NOTCH1 mutations observed in these T-ALL cases, might act as lineage determinants. Outcome analyses revealed that DNMT3A mutation correlated with poor prognosis independently of the patients’ age in bivariate analyses. Multivariable analyses using parameters that were used to stratify treatment in the GRAALL-2003 and -2005 studies showed that DNMT3A genotype independently predicted both event-free survival and cumulative incidence of relapse. DNMT3A mutation status also independently predicted event-free survival and overall survival in bivariable analyses that incorporated our recently described oncogenetic risk classifier37 (Online Supplementary Table S5). These results suggest that DNMT3A mutation is directly linked to aggressive T-ALL biology. As DNMT3A-altered T-ALL had higher mutation rates in other genes included in our targeted sequencing panel, it is also possible that increased genotype complexity may contribute to the more aggressive phenotype in these leukemias. This issue may be clarified by more comprehensive genomic assessment in future studies. The high rates of treatment failure observed in this study suggest that therapeutic intervention is warranted for DNMT3A-mutated cases, and that treatment intensification should be considered for the infrequent younger patients with mutations. Indeed, we have previously documented a benefit from allogeneic stem cell transplantation in first complete remission for ETP-ALL,24 which similarly exhibits high rates of intrinsically treatment-resistant disease. As only three of the 18 DNMT3A-mutated patients in this study underwent allogeneic stem cell 1624

transplantation (data not shown), we are unable to estimate the potential benefit of such treatment in this setting. We recently reported that treatment-related toxicity in the GRAALL-2005 study increased in proportion to the patients’ age,38 and further therapy intensification in elderly patients must therefore be considered of questionable benefit. The upper age of this study cohort was 60 years, but it is likely that the rate of mutations in older patients who do not tolerate such intensive chemotherapy is higher. Data reported for patients with AML suggest that DNMT3A mutation confers increased sensitivity to hypomethylating agents,39 providing a rationale for evaluation of these drugs in DNMT3A-mutated T-ALL. In the longer term, it is to be hoped that investigation of the molecular mechanisms by which DNMT3A mutation alters T-ALL biology will lead to better treatments and improved outcomes for these high-risk cases. Acknowledgments This manuscript was written on behalf of the Group for Research on Adult Acute Lymphoblastic Leukemia (GRAALL), which includes the former France-Belgium Group for Lymphoblastic Acute Leukemia in Adults (LALA), the French Western-Eastern Group for Lymphoblastic Acute Leukemia (GOELAL), and the Swiss Group for Clinical Cancer Research (SAKK). The authors would like to thank all participants in the GRAALL-2003 and GRAALL-2005 study groups for collection and provision of data and samples, and V. Lheritier for collection of clinical data. The GRAALL-2003 study was sponsored by the Hôpitaux de Toulouse, and the GRAALL-2005 study by the Assistance Publique-Hôpitaux de Paris. The SAKK was supported by the Swiss State Secretariat for Education, Research and Innovation (SERI). JB was supported by a Kay Kendall Leukaemia Fund Intermediate Research Fellowship and by the National Children's Research Centre, Children's Health Ireland at Crumlin, Dublin, Ireland. The Necker Laboratory is supported by the Association Laurette Fugain, La Ligue contre le Cancer and the INCa CARAMELE Translational Research and PhD programs. haematologica | 2019; 104(8)

DNMT3A mutation predicts adverse outcome in adult T-ALL

References 1. Ley TJ, Ding L, Walter MJ, et al. DNMT3A mutations in acute myeloid leukemia. N Engl J Med. 2010;363(25):2424-2433. 2. Renneville A, Boissel N, Nibourel O, et al. Prognostic significance of DNA methyltransferase 3A mutations in cytogenetically normal acute myeloid leukemia: a study by the Acute Leukemia French Association. Leukemia. 2012;26(6):1247-1254. 3. Gale RE, Lamb K, Allen C, et al. Simpson's paradox and the impact of different DNMT3A mutations on outcome in younger adults with acute myeloid leukemia. J Clin Oncol. 2015;33(18):20722083. 4. Marcucci G, Metzeler KH, Schwind S, et al. Age-related prognostic impact of different types of DNMT3A mutations in adults with primary cytogenetically normal acute myeloid leukemia. J Clin Oncol. 2012;30(7): 742-750. 5. Yan XJ, Xu J, Gu ZH, et al. Exome sequencing identifies somatic mutations of DNA methyltransferase gene DNMT3A in acute monocytic leukemia. Nat Genet. 2013;43(4):309-315. 6. Haferlach T, Nagata Y, Grossmann V, et al. Landscape of genetic lesions in 944 patients with myelodysplastic syndromes. Leukemia. 2014;28(2):241-247. 7. Stegelmann F, Bullinger L, Schlenk RF, et al. DNMT3A mutations in myeloproliferative neoplasms. Leukemia. 2011;25(7):12171219. 8. Jankowska AM, Makishima H, Tiu RV, et al. Mutational spectrum analysis of chronic myelomonocytic leukemia includes genes associated with epigenetic regulation: UTX, EZH2, and DNMT3A. Blood. 2011;118(14): 3932-3941. 9. Roller A, Grossmann V, Bacher U, et al. Landmark analysis of DNMT3A mutations in hematological malignancies. Leukemia. 2013;27(7):1573-1578. 10. Van Vlierberghe P, Ambesi-Impiombato A, Perez-Garcia A, et al. ETV6 mutations in early immature human T cell leukemias. J Exp Med. 2011;208(13):2571-2579. 11. Grossmann V, Haferlach C, Weissmann S, et al. The molecular profile of adult T-cell acute lymphoblastic leukemia: mutations in RUNX1 and DNMT3A are associated with poor prognosis in T-ALL. Genes Chromosomes Cancer. 2013;52(4):410-422. 12. Neumann M, Heesch S, Schlee C, et al. Whole-exome sequencing in adult ETP-ALL reveals a high rate of DNMT3A mutations. Blood. 2013;121(23):4749-4752. 13. Van Vlierberghe P, Ambesi-Impiombato A, De Keersmaecker K, et al. Prognostic relevance of integrated genetic profiling in adult T-cell acute lymphoblastic leukemia. Blood.

haematologica | 2019; 104(8)

2013;122(1):74-82. 14. Couronne L, Bastard C, Bernard OA. TET2 and DNMT3A mutations in human T-cell lymphoma. N Engl J Med. 2012;366(1):9596. 15. Sakata-Yanagimoto M, Enami T, Yoshida K, et al. Somatic RHOA mutation in angioimmunoblastic T cell lymphoma. Nat Genet. 2014;46(2):171-175. 16. Choi J, Goh G, Walradt T, et al. Genomic landscape of cutaneous T cell lymphoma. Nat Genet. 2015;47(9):1011-1019. 17. Ho PA, Kutny MA, Alonzo TA, et al. Leukemic mutations in the methylationassociated genes DNMT3A and IDH2 are rare events in pediatric AML: a report from the Children's Oncology Group. Pediatr Blood Cancer. 2011;57(2):204-209. 18. Huether R, Dong L, Chen X, et al. The landscape of somatic mutations in epigenetic regulators across 1,000 paediatric cancer genomes. Nat Commun. 2014;5:3630. 19. Shiba N, Taki T, Park MJ, et al. DNMT3A mutations are rare in childhood acute myeloid leukaemia, myelodysplastic syndromes and juvenile myelomonocytic leukaemia. Br J Haematol. 2012;156(3):413414. 20. Shen Y, Zhu YM, Fan X, et al. Gene mutation patterns and their prognostic impact in a cohort of 1185 patients with acute myeloid leukemia. Blood. 2011;118(20): 5593-5603. 21. Ribeiro AF, Pratcorona M, ErpelinckVerschueren C, et al. Mutant DNMT3A: a marker of poor prognosis in acute myeloid leukemia. Blood. 2012;119(24):5824-5831. 22. Tie R, Zhang T, Fu H, et al. Association between DNMT3A mutations and prognosis of adults with de novo acute myeloid leukemia: a systematic review and metaanalysis. PloS One. 2014;9(6):e93353. 23. Zhang J, Ding L, Holmfeldt L, et al. The genetic basis of early T-cell precursor acute lymphoblastic leukaemia. Nature. 2012;481(7380):157-163. 24. Bond J, Graux C, Lhermitte L, et al. Early response-based therapy stratification improves survival in adult early thymic precursor acute lymphoblastic leukemia: a group for research on adult acute lymphoblastic leukemia study. J Clin Oncol. 2017;35(23):2683-2691. 25. 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. 26. Lin ME, Hou HA, Tsai CH, et al. Dynamics of DNMT3A mutation and prognostic relevance in patients with primary myelodysplastic syndrome. Clin Epigenetics. 2018;10:42. 27. 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. 28. Jaiswal S, Fontanillas P, Flannick J, et al. Agerelated clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371(26):2488-2498. 29. 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. 30. Shlush LI, Zandi S, Mitchell A, et al. Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia. Nature. 2014;506(7488):328-333. 31. 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. 32. Asnafi V, Beldjord K, Boulanger E, et al. Analysis of TCR, pT alpha, and RAG-1 in Tacute lymphoblastic leukemias improves understanding of early human T-lymphoid lineage commitment. Blood. 2003;101(7): 2693-2703. 33. Coustan-Smith E, Mullighan CG, Onciu M, et al. Early T-cell precursor leukaemia: a subtype of very high-risk acute lymphoblastic leukaemia. Lancet Oncol. 2009;10(2):147156. 34. Huguet F, Leguay T, Raffoux E, et al. Pediatric-inspired therapy in adults with Philadelphia chromosome-negative acute lymphoblastic leukemia: the GRAALL-2003 study. J Clin Oncol. 2009;27(6):911-918. 35. Bond J, Marchand T, Touzart A, et al. An early thymic precursor phenotype predicts outcome exclusively in HOXA-overexpressing adult T-cell acute lymphoblastic leukemia: a Group for Research in Adult Acute Lymphoblastic Leukemia study. Haematologica. 2016;101(6):732-740. 36. Eguchi-Ishimae M, Eguchi M, Kempski H, Greaves M. NOTCH1 mutation can be an early, prenatal genetic event in T-ALL. Blood. 2008;111(1):376-378. 37. Trinquand A, Tanguy-Schmidt A, Ben Abdelali R, et al. Toward a NOTCH1/FBXW7/RAS/PTEN-based oncogenetic risk classification of adult T-cell acute lymphoblastic leukemia: a Group for Research in Adult Acute Lymphoblastic Leukemia study. J Clin Oncol. 2013;31 (34):4333-4342. 38. Huguet Fo, Leguay T, Thomas X, et al. The upper age limit for a pediatric-inspired therapy in younger adults with Ph-negative acute lymphoblastic leukemia (ALL)? Analysis of the GRAALL-2005 study. 2016:762-762. 39. Metzeler KH, Walker A, Geyer S, et al. DNMT3A mutations and response to the hypomethylating agent decitabine in acute myeloid leukemia. Leukemia. 2012;26(5): 1106-1107.

1625

ARTICLE Ferrata Storti Foundation

Non-Hodgkin Lymphoma

Human leukocyte antigen class II expression is a good prognostic factor in adult T-cell leukemia/lymphoma Mai Takeuchi,1 Hiroaki Miyoshi,1 Naoko Asano,2 Noriaki Yoshida,1,3 Kyohei Yamada,1 Eriko Yanagida,1 Mayuko Moritsubo,1 Michiko Nakata,1 Takeshi Umeno,1 Takaharu Suzuki,1 Satoru Komaki,1 Hiroko Muta,1 Takuya Furuta,1 Masao Seto1 and Koichi Ohshima1

Haematologica 2019 Volume 104(8):1626-1632

Department of Pathology, Kurume University School of Medicine, Kurume, Fukuoka; Department of Molecular Diagnostics, Nagano Prefectural Shinshu Medical Center, Suzaka, Nagano and 3Department of Clinical Studies, Radiation Effects Research Foundation, Hiroshima, Japan

1 2

ABSTRACT

Correspondence: KOICHI OHSHIMA ohshima_kouichi@med.kurume-u.ac.jp Received: August 28, 2018. Accepted: January 9, 2019. Pre-published: January 10, 2019. doi:10.3324/haematol.2018.205567 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1626 ©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.

1626

ttenuated human leukocyte antigen (HLA) class I expression is implicated as a major immune escape mechanism in several types of tumor. We previously reported that HLA class I/β2 microglobulin and programmed death ligand-1 expression are prognostic factors in adult T-cell leukemia/lymphoma. A recent report suggested that HLA class II expression is also an important prognostic factor for the clinical outcome of programmed death-1 blockade therapy in recurrent/refractory Hodgkin lymphoma. This prompted us to evaluate HLA class II expression in adult T-cell leukemia/lymphoma and to compare the findings with the patients’ clinicopathological features. Of the 132 biopsy specimens examined from newly diagnosed patients, lymphoma cells were positive for HLA class II expression in 44 patients (33.3%), whereas programmed death ligand-1 expression was observed on neoplastic cells from nine patients (6.9%) and on stromal cells in the tumor microenvironment in 83 cases (62.9%). HLA class II-positive cases showed a significantly better overall survival compared to the HLA class II-negative cases (P<0.0001). Patients positive for HLA class II and programmed death ligand-1 microenvironmental expression had significantly better prognosis than the other groups (P<0.0001). HLA class II-positive and HLA class II-negative groups also showed a significant difference in complete remission rate (P=0.0421), HLA class I/β2 microglobulin expression (P=0.0165), and the number of programmed death-1-positive tumor infiltrating cells (P=0.0020). HLA class II expression was a prognostic factor for overall survival both in univariate and multivariate analyses (P<0.0001 and P=0.0007, respectively). Our study reveals that HLA class II is a novel prognostic factor in adult T-cell leukemia/lymphoma.

Introduction Adult T-cell leukemia/lymphoma (ATLL), caused by human T-cell lymphotropic virus type 1 (HTLV-1), is an aggressive hematopoietic malignancy with a poor prognosis. HTLV-1 carriers are common in coastal areas of south-western Japan, and approximately 2-5% of carriers develop ATLL in their lifetime.1,2 According to the Shimoyama classification, there are four clinical subtypes of ATLL, namely acute, lymphoma, chronic, and smoldering.3 The acute and lymphoma types are typically associated with a poor prognosis.3 The chronic and smoldering types show better prognosis than the acute and lymphoma types.3 However, half of the cases of chronic ATLL have been reported to transform into leukemia.4 Programmed cell death ligand-1 (PD-L1) is expressed on various tumors and is considered to contribute to tumor immune escape.5-9 We previously reported that PD-L1 expression in tumor cells and stromal cells of the tumor microenvironment are prognostic factors in ATLL.10 Presentation of neoantigens by human leukocyte antigen (HLA) class Ι is required for CD8+ cytotoxic T cells to attack tumor cells. However, the expression of HLA class I and/or its component β2 microglobulin (β2M) is frequently reduced or lost haematologica | 2019; 104(8)

HLA class II is a good prognostic factor in ATLL

in some tumors, which plays an important role in immune escape and survival of tumors.11 Kataoka et al. found that more than half of ATLL patients had alterations in HLA class I genes.12 We have previously reported that ATLL expressing HLA class I/β2M had a good clinical outcome.13 Programmed cell death-1 (PD-1) blockade therapy is effective in various recurrent/refractory tumors. PD-1 is expressed on activated T cells and binds to PD-L1 on tumor cells. Binding of PD-L1 to PD-1 suppresses production of cytokines and inactivates T cells.14,15 HLA class I plays a key role in PD-1 blockade therapy because activated T cells attack tumor cells by recognizing neoantigens presented by HLA class I on tumor membranes. In fact, an interaction between neoantigens and neoantigen-specific T cells via HLA class I has been reported in patients with malignant melanoma who received PD-1 blockade therapy.16 It has been hypothesized that tumors lacking HLA class I and/or β2M could be resistant to PD-1 blockade therapy.17 However, a favorable clinical response to PD-1 blockade has been observed in patients with Hodgkin lymphoma, who usually lack HLA class I expression due to the loss-of-function mutation of β2M.18,19 From the standpoint of immunotherapy, these results are indicative that other immunological mechanisms exist, at least among patients with Hodgkin lymphoma. HLA class II is expressed on antigen-presenting cells including B cells, dendritic cells, and macrophages. HLA class II presents antigens to CD4+ T cells. HLA class II expression has been reported to be associated with a favorable prognosis in squamous cell carcinoma of the larynx, colorectal cancer, and diffuse large B-cell lymphoma.20-22 In contrast, HLA class II expression was not significantly associated with prognosis in malignant melanoma, Hodgkin lymphoma, and lung cancer treated with standard chemotherapies.23-25 Roemer et al. recently reported that HLA class II expression is more important than HLA class I expression for PD-1 blockade therapy in recurrent/refractory Hodgkin lymphoma.26 Although the precise mechanism remains unknown, it has been suggested that an interaction between HLA class II-expressing lymphoma cells and CD4+ T cells could contribute to tumor immunity in Hodgkin lymphoma. The prognosis of ATLL patients treated with standard chemotherapy remains poor. PD-1 blockade could be a good therapeutic strategy; however, a clinical trial showed that PD-1 blockade in ATLL patients induced disease progression.27 For a better understanding of tumor immunity associated with PD-L1/PD-1 in ATLL, we examined the clinico-pathological effect of HLA class II expression in ATLL and the correlation of HLA class II to HLA class I/β2M and PD-L1/PD-1 expression.

Methods Patients and samples All patients included in this study were the same as those in our previous reports.10,13 Fifty-nine patients were derived from the International Peripheral T-Cell Lymphoma Project, and 73 were newly diagnosed with ATLL in Kurume University between 2006 and 2012.28 A tissue microarray including 132 samples in formalinfixed paraffin-embedded blocks was created. The tissue microarray specimens used in this study were the same as those examined in our previous studies.10,13 The use of patients’ specimens and clinhaematologica | 2019; 104(8)

ical data was approved by the Research Ethics Committee of Kurume University and was carried out in accordance with the Declaration of Helsinki.

Immunohistochemistry Immunohistochemistry was performed as previously described.10,13 The following antibodies were used: HLA class II (CR3/43 M0775, DAKO, Glostrup, Denmark), HLA class I (EMR85; Abcam, Tokyo, Japan), β2M (HPA006361; Sigma-Aldrich, Tokyo, Japan), PD-L1 (EPR1 161(2); Abcam), and PD-1 (NAT105, Abcam). As in our previous studies, PD-L1 expression was determined on neoplastic cells (nPD-L1) and microenvironmental stromal cells (miPD-L1). nPD-L1 was considered positive when 50% or more neoplastic cells were positive for PD-L1.10 miPD-L1 were distinguished from nPD-L1 by irregular shaped morphology, low nuclear/cytoplasmic ratio, and nuclei without atypia (Online Supplementary Figure S1A, S1B).10 miPD-L1 was defined as positive in patients with ten or more PD-L1+ non-neoplastic stromal cells per high power field.10 Other markers, including HLA class II, were considered positive when 30% or more tumor cells were positive. PD-1+ tumor-infiltrating lymphocytes were detected as small lymphocytes without atypia and were distinguished from PD-1+ tumor cells (Online Supplementary Figure S1C, S1D). PD-1+ tumorinfiltrating lymphocytes were counted in five representative highpower fields, and the average number for each sample was calculated as previously reported.10

Statistical analysis

Clinico-pathological characteristics were compared by the χ2 test or Fisher two-sided exact test. Wilcoxon rank sum test was performed to compare age and PD-1+ tumor-infiltrating lymphocyte counts. Overall survival was defined as the period between the day of diagnosis and the day of death or the day of last follow up. Overall survival was analyzed by the Kaplan–Meier method, and the log-rank test was performed to determine significant differences. Univariate and multivariate Cox proportional regression models were used to examine the proposed prognostic factors. All statistical analyses were performed using JMP version 11.0. P values <0.05 were considered statistically significant.

Results Clinico-pathological characteristics The participants’ clinical information is summarized in Table 1. The median age of the participants was 67.5 years (range, 35-90), and the male to female ratio was 75:57. Seventy-eight of 126 (61.9%) patients had a high International Prognostic Index (high or high-intermediate), and 50/129 (38.8%) patients had high scores in the Japan Clinical Oncology Group Prognostic Index (JCOG-PI).29 According to the Shimoyama classification, most cases were acute type (53/108, 49.1%) or lymphoma type (52/108, 48.1%). Only three cases (2.8%) were smoldering ATLL and no cases of chronic ATLL were included in this study. Most patients received chemotherapy (115/131, 87.8%); radiation (14/129, 10.9%) and allogenic stem cell transplantation (17/129, 13.2%) were chosen in fewer cases. The chemotherapy regimens were not significantly different between the HLA class II-positive and negative groups (Online Supplementary Table S1). No PD-1 blockade therapy was given. Of the 115 patients, 32 (27.8%) achieved a complete remission. The median follow-up period was 10.96 months (range, 0.03-114.8 months). 1627

M. Takeuchi et al.

Figure 1. Immunohistochemistry of HLA class II in adult T-cell leukemia/lymphoma. (A, B) Representative images of human leukocyte antigen (HLA) class II-positive cases. Adult T-cell leukemia/lymphoma cells show membranes with/without cytoplasmic HLA class II expression. (C) A representative image of HLA class II-negative cases.

Table 1. Patients and treatment characteristics.

Features Age, median (range), years Sex (male/female) PS score 2-4 IPI risk high or high-intermediate JCOG-PI high Ann Arbor stage III or IV Shimoyama classification Acute type Lymphoma type Smoldering type Chronic type Treatment Chemotherapy Radiation Transplantation No treatment CR/CR(u)

Numbers

67.5 (35-90) 75/57 43/129 78/126 50/129 111/132

33.3 61.9 38.8 84.1

53/108 52/108 3/108 0/108

49.1 48.1 2.8 0.0

115/131 14/129 17/129 12/130 32/115

87.8 10.9 13.2 9.2 27.8

Features +

PS,: Performance Status; IPI: International Prognostic Index; JCOG-PI, Japan Clinical Oncology Group-Prognostic Index, CR: complete response; CR(u),: complete response (unconfirmed). Only the patients with available data were calculated.

Expression of HLA class II and its comparison with HLA class I and programmed death ligand-1 expression A summary of the immunohistochemical analysis is shown in Table 2. Neoplastic cells were positive for HLA class II (HLA class II+) in 44/132 (33.3%) cases, and HLA class II was expressed on the membrane with or without cytoplasm (Figure 1A,B). HLA class II was negative in 88/132 cases (66.7%) (Figure 1C). In 48 out of 120 cases, HLA class I and β2M were positive on the membrane (HLA class Im+/β2Mm+) (40.0%). Both HLA class II+ and HLA class Im+/β2Mm+ were found in 22/120 cases (18.3%). PD-L1 was positive on neoplastic cells (nPD-L1+) in 9/130 cases (6.9%). Five cases (3.8%) were both HLA class II+ and nPD-L1+. PD-L1-positive stromal cells in the tumor microenvironment (miPD-L1+) were observed in 83/132 cases (62.9%). Thirty-two of 132 cases (24.3%) were both HLA class II+ and miPD-L1+. 1628

Table 2. Immunohistochemical characteristics of patients wth adult Tcell leukemia/lymphoma. HLA class II HLA class IIHLA class Im+β2Mm+ HLA class Im- and/or β2MmHLA class II+/HLA class Im+β2Mm+ HLA class II+/HLA class Im-and/or β2MmHLA class II-/HLA class Im+β2Mm+ HLA class II-/HLA class Im-and/or β2MmnPD-L1+ nPD-L1HLA class II+ nPD-L1+ HLA class II+ nPD-L1HLA class II- nPD-L1+ HLA class II- nPD-L1miPD-L1+ miPD-L1HLA class II+ miPD-L1+ HLA class II+ miPD-L1HLA class II- miPD-L1+ HLA class II- miPD-L1-

Number

44/132 88/132 48/120 72/120 22 17 26 55 9/130 121/130 5 39 4 82 83/132 49/132 32 12 51 37

33.3 66.7 40.0 60.0 18.3 14.2 21.7 45.8 6.9 93.1 3.8 30.0 3.1 63.1 62.9 37.1 24.3 9.1 38.6 28.0

ATLL,: Adult T-cell leukemia/lymphoma; HLA: human leukocyte antigen; β2M: β2 microglobulin; nPD-L1: neoplastic programmed cell death ligand-1; miPD-L1: microenvironmental programmed cell death ligand-1.

Overall survival in patients with adult T-cell leukemia/lymphoma according to HLA class II expression HLA class II+ cases had significantly better overall survival than HLA class II− cases (median survival time, 36.53 months vs. 9.43 months, respectively; P<0.0001) (Figure 2).

Overall survival in patients with adult T-cell leukemia/lymphoma according to HLA class II and HLA class I/β2-microglobulin expression Compared to the other groups, HLA class II+ /HLA class I /β2Mm+ patients had the best clinical outcomes (P=0.0013), although HLA class II+/HLA class Im+/β2Mm+ patients also showed favorable outcomes (Figure 3). HLA class II− patients had a poor prognosis regardless of the HLA class I/β2M expression (Figure 3). m+

haematologica | 2019; 104(8)

HLA class II is a good prognostic factor in ATLL

Overall survival of patients with adult T-cell leukemia/lymphoma according to HLA class II and programmed death ligand-1 expression As we have previously reported, the nPD-L1+ group had significantly worse overall survival than the nPD-L1− group (median survival time, 5.93 months vs. 14.06 months, respectively; P=0.0119).10 Moreover, miPD-L1+ patients had a better prognosis than miPD-L1− patients (median survival time 15.63 months vs. 10.1 months, respectively) (P=0.0029).10 Patients in the HLA class II+/nPD-L1- group had a significantly better survival time than the other patients. (P <0.0001) (Figure 4A). The HLA class II+/miPD-L1+ group had significantly better overall survival than the other groups (P<0.0001) (Figure 4B).

Hodgkin lymphoma and lung cancer are often associated with PD-L1 expression on tumor cells and respond to immune checkpoint inhibitors. However, HLA class II was not associated with prognosis in patients with these cancers treated with standard chemotherapy.23-25 In our study ATLL cases co-expressing PD-L1 and HLA class II, had a very poor prognosis, although the number of patients analyzed was small (n=5, 3.8%). In the cases of Hodgkin lymphoma and malignant melanoma, HLA class II expression predicted the clinical outcome of PD-1 blockade therapy, suggesting that the prediction of prognosis by HLA class II expression depends on the expression of PD-L1.24,26 HLA class I/β2M is also a good prognostic marker in ATLL patients. However, HLA class I+/β2M+ patients without HLA class II expression had a poor prognosis in this

Clinico-pathological differences between HLA class II-positive and HLA class II-negative cases The clinicopathological characteristics of HLA class II+ and HLA class II− cases are summarized in Table 3. HLA class II+ cases showed a significantly higher complete response rate (P=0.0421), higher HLA class I+/β2M+ expression (P=0.0165), and increased number of PD-1+ tumor-infiltrating lymphocytes (P=0.0020). Performance Status, International Prognostic Index, JCOG-PI, clinical stage, and therapies were not significantly different between the two groups. The association of HLA class II, PD-L1, and PD-1 expression is shown in Online Supplementary Figure S2. There was no significant association among them.

Prognostic factors in patients with adult T-cell leukemia/lymphoma Table 4 presents results of univariate and multivariate analyses of prognostic factors in ATLL. In univariate analyses, age (>70 years) [hazard ratio (HR) 1.709; 95% confidence interval (95% CI): 1.106-2.608; P=0.0164], high JCOG-PI (HR 1.841; 95% CI: 1.186-2.819; P=0.0071), Ann Arbor stage III or IV (HR 1.801; 95% CI: 1.002-3.583; P=0.0491), HLA class II expression (HR 0.386; 95% CI: 0.233-0.617; P<0.0001), and HLA class II/miPD-L1 expression (HR 0.276; 95% CI: 0.146-0.482; P<0.0001) were significantly related to prognosis. In a multivariate analysis of HLA class II expression and other factors, HLA class II expression (HR 0.441; 95% CI: 0.263-0.714; P=0.0007) and high JCOG-PI (HR 1.841; 95% CI: 1.163-2.883; P=0.0096) showed significant prognostic value. We also performed a random forest analysis including all available clinical prognostic factors (JCOG-PI, age >70 years, elevated lactate dehydrogenase, extranodal involvement, sex, elevated C-reactive protein, B-symptoms, clinical stage, chemotherapy, radiation, and transplantation) and HLA class II expression. JCOG-PI, HLA class II expression, and age were associated with patients living for more than 2 years.

Figure 2. Association of HLA class II expression with overall survival in adult Tcell leukemia/lymphoma. The Kaplan–Meier plot depicts the survival difference between human leukocyte antigen (HLA) class II-positive (HLA class II+) and HLA class II-negative (HLA class II−) patients (P<0.0001; log-rank test).

Discussion In this report, we describe that HLA class II expression is an important prognostic factor in ATLL, as are both HLA class I/β2M and PD-L1 expression. HLA class II had a prognostic value in ATLL patients treated with standard chemotherapy in our study. Malignant melanoma, haematologica | 2019; 104(8)

Figure 3. Overall survival according to expression of HLA class II and HLA class I/β2 microglobulin in patients with adult T-cell leukemia/lymphoma. The patients with adult T-cell leukemia/lymphoma were stratified into four groups according to human leukocyte antigen (HLA) class II and HLA class I/β2 microglobulin (β2M) expression. Differences in overall survival among the four groups are shown in the Kaplan–Meier plot (P=0.0013; log-rank test).

1629

M. Takeuchi et al.

study. Recently, the importance of CD4+ T cells and HLA class II-restricted neoantigens has been focused on in cancer immunology. It has been suggested that CD4+ type 1 helper T cells can upregulate HLA class II expression on tumor cells by producing cytokines and that they can kill tumor cells directly, independently of other immune cells.30 Thus, CD4+ non-neoplastic T cells may also contribute to the improved prognosis of HLA class II+ ATLL patients. In addition, CD4+ helper T cells can help CD8+ T cells in tumor immunity.30 Thus, HLA class II expression

may be necessary for CD8+ T cells to attack HLA class I+ tumor cells. HLA class II expression was also reported to be a good prognostic factor in diffuse large B-cell lymphoma,22 a malignancy derived from B cells. Thus, diffuse large B-cell lymphoma originally has active HLA class II, which can present neoantigens, suggesting that loss of HLA class II on diffuse large B-cell lymphoma disrupts the interaction with CD4+ T cells. In our study, PD-L1-expressing stromal cells (miPD-L1+)

Figure 4. Overall survival associated with HLA class II and programmed death ligand-1 expression. (A) Patients were stratified into four groups according to the expression of human leukocyte antigen (HLA) class II and programmed death ligand-1 on neoplastic cells (nPD-L1). The differences in survival are depicted in the Kaplan–Meier plot (P<0.0001; log-rank test). (B) The patients were divided into four groups according to the expression of HLA class II and programmed death ligand-1 on microenvironmental cells (miPD-L1). Kaplan–Meier plots reveal the differences among the groups (P<0.0001; log-rank test).

Table 3. Clinicopathological comparison according to HLA class II expression.

HLA class II+ (n=44) Age, mean (range), years Sex, male/female Shimoyama classification Acute type Lymphoma type Smoldering type Chronic type PS score 2-4 IPI high or high-intermediate JCOG-PI high Ann Arbor stage III, IV Elevated LDH Treatment Chemotherapy Radiation Transplantation No treatment CR/CR(u) Immunohistochemistry HLA class Im+β2M m+ nPD-L1+ miPD-L1+ PD-1+ TIL (/hpf), average (range)

65.5 (35-80) 28/16

HLA class II– (n=88)

68.5 (40-90) 47/41

P 0.0697 0.3515

19/35 15/35 1/35 0/35 12/44 25/43 14/44 34/44 27/44

54.3 42.8 2.9 0.0 27.3 58.1 31.8 77.3 61.4

34/73 37/73 2/73 0/73 31/85 53/83 36/85 77/88 49/88

46.6 50.7 2.7 0.0 36.5 63.9 42.4 87.5 55.7

0.5387 0.5382 0.3297 0.5654 0.2601 0.1387 0.5788

36/44 4/43 4/43 7/44 15/36

81.8 9.3 9.3 15.9 41.7

79/87 10/86 13/86 5/86 17/79

90.8 11.6 15.1 5.8 21.5

0.1624 0.7731 0.4205 0.1049 0.0421*

22/39 5/44 32/44 3.95 (0-37.4)

56.4 11.4 72.7

26/81 4/86 51/88 1.83 (0-38.4)

32.1 4.7 58.0

0.0165* 0.1655 0.1266 0.0020*

HLA: human leukocyte antigen; PS: Performance Status; IPI: International Prognostic Index; JCOG-PI; Japan Clinical Oncology - Prognostic Index; LDH: lactate dehydrogenase; CR: complete response; CR(u),: complete response (unconfirmed); β2M: β2 microglobulin; nPD-L1: neoplastic programmed death ligand-1; miPD-L1: microenvironmental programmed death ligand-1; PD-1: programmed death-1; TIL: tumor infiltrating cells; hpf: high power field. *Statistically significant P value.

1630

haematologica | 2019; 104(8)

HLA class II is a good prognostic factor in ATLL Table 4. Prognostic factors associated with overall survival in patients with adult T-cell leukemia/lymphoma.

Univariate analysis Characteristics HLA class II+ HLA class II+/miPD-L1+ JCOG-PI (high) Age>70 years Ann Arbor stage III, IV Elevated LDH (>normal)

HR (95% CI)

0.386 (0.233-0.617) 0.276 (0.146-0.482) 1.841 (1.186-2.819) 1.709 (1.106-2.608) 1.801 (1.002-3.583) 0.982 (0.651-1.492)

<0.0001* <0.0001* 0.0071* 0.0164 0.0491* 0.9344

Multivariate analysis HR (95% CI) P 0.441 (0.263-0.714)

0.0007*

1.841 (1.163-2.883) 1.441 (0.925-2.217) 1.661 (0.915-3.325) 0.934 (0.605-1,447)

0.0096* 0.1048 0.0982 0.7582

ATLL: adult T-cell leukemia/lymphoma; HR: hazard ratio; 95% CI: 95% confidence interval; HLA: human leukocyte antigen; miPD-L1: microenvironmental programmed cell death ligand-1; JCOG-PI: Japan Clinical Oncology Group-Prognostic Index; LDH: lactate dehydrogenase..*Statistically significant P value.

seemed necessary for good clinical outcome of HLA class II+ patients. miPD-L1 has been reported in some cancers, but the precise role of miPD-L1 is unclear. It is possible that miPD-L1 suppressed PD-1-expressing tumor cells, but no significant association between miPD-L1 and neoplastic PD-1 expression was observed in the present study (data not shown). Further research is required to determine whether there are direct or indirect interactions between HLA class II+ tumor cells and miPD-L1. Ratner et al. recently reported rapid progression of ATLL after PD-1 blockade therapy.27 They attributed the disease progression to the tumor suppressive role of PD-1 in ATLL. It seems that the ATLL patients included in their study expressed PD-L1 on only <1% or 5% of tumor cells. Our current and previous data suggest that expression of HLA class I/class II and miPD-L1 can predict good clinical outcome in ATLL patients without massive nPD-L1 expression. HLA class I/class II expression might be associated with the efficacy of PD-1 blockade therapy in ATLL. Although the precise function of miPD-L1 remains to be determined, PD-1 blockade might disrupt tumor immunity associated with miPD-L1 in ATLL. HLA class II expression on activated normal human T cells is regulated by class II transactivator (CIITA).31 In Tcell malignancies, expression of CIITA also correlates with HLA class II expression.31 In ATLL, CIITA was reported to inhibit the nuclear factor κB pathway activated by the Tax1 oncoprotein of HTLV-1.32 Thus, the favorable clinical prognosis of HLA class II+ ATLL patients may correlate with upregulation of CIITA, which leads to activation of HLA class II expression as well as suppression of the NFκB signaling pathway. The rate of HLA class II positivity was low in our study compared to that in the published literature, in which

References 1. Ohshima K, Jaffe E, Yoshino T, Siebert R. Adult T-cell leukemia/lymphoma. In Swerdlow SH, Campo CE, Harris NL, et al. eds. World Health Organization Classification of Tumours. Revised 4th ed. Lyon, France: IARC Press, 2017;363-367. 2. Tajima K, Hinuma Y. Epidemiology of HTLV-I/II in Japan and the world. In Takatsuki K, Hinuma Y, Yoshida M eds. Advances in Adult T-cell Leukemia and

haematologica | 2019; 104(8)

chronic and smoldering ATLL were reported to show high expression of HLA class II whereas acute type ATLL was described as showing low expression of HLA class II.33 It was suggested that HLA class II expression on ATLL could be associated with progression of chronic ATLL.33 Our study included mostly patients with acute or lymphoma type ATLL. HLA class II expression on acute or lymphoma type ATLL may be correlated with better clinical outcome. In conclusion, we report, for the first time, that HLA class II expression is a good prognostic marker in ATLL. Both HLA class II and miPD-L1 were required for good clinical outcome. Our results help to understand tumor immunity in ATLL. Acknowledgments The authors thank Kazutaka Nakashima, Mayumi Miura, Kanoko Miyazaki, and Chie Kuroki for their technical assistance and collaborators from the following institutions for providing clinical data and specimens to the Kyushu Lymphoma Study Group: Department of Medicine and Biosystemic Science, Kyushu University Faculty of Medicine; Department of Hematology, Karatsu Red Cross Hospital; Department of Hematology, Sasebo City General Hospital; Department of Hematology, Atomic Bomb Disease and Hibakusha Medicine Unit, Atomic Bomb Disease Institute, Nagasaki University; and Department of Medicine and Bioregulatory Science, Graduate School of Medical Sciences, Kyushu University. This work was supported in part by Grantsin-Aid from the Japan Agency for Medical Research and Development (15ck0106015h0002) (MS); the Japan Society for the Promotion of Science (KAKENHI), grant numbers JP26460446 (KO) and JP17K17894 (MT); and the Japan Leukemia Research Fund (General Research Award 2013) (NA). The language and format of this manuscript were edited by Editage (https://www.editage.com).

HTLV-I Research. Tokyo: Japan Scientific Societies Press, 1992;129-149. 3. Shimoyama M. Diagnostic criteria and classification of clinical subtypes of adult T-cell leukaemia–lymphoma. A report from the Lymphoma Study Group (1984–87). Br J Haematol. 1991;79(3):428-437. 4. Takasaki Y, Iwanaga M, Imaizumi Y, et al. Long-term study of indolent adult T-cell leukemia–lymphoma. Blood. 2010;115(22); 4337-4343. 5. Gao Q, Wang XY, Qiu SJ, et al.

Overexpression of PD-L1 significantly associates with tumor aggressiveness and postoperative recurrence in human hepatocellular carcinoma. Clin Cancer Res. 2009;15(3):971-979. 6. Nomi T, Sho M, Akahori T, et al. Clinical significance and therapeutic potential of the programmed death-1 ligand/programmed death-1 pathway in human pancreatic cancer. Clin Cancer Res. 2007;13(7):2151-2157. 7. Ohigashi Y, Sho M, Yamada Y, et al. Clinical significance of programmed death-1 ligand-

1631

M. Takeuchi et al.

9. 10.

11.

12.

13.

14. 15.

16. 17.

1632

1 and programmed death-1 ligand-2 expression in human esophageal cancer. Clin Cancer Res. 2005;11(8):2947-2953. Thompson RH, Dong H, Kwon ED. Implications of B7-H1 expression in clear cell carcinoma of the kidney for prognostication and therapy. Clin Cancer Res. 2007;13(2);709-715. Zou W, Chen L. Inhibitory B7-family molecules in the tumour microenvironment. Nat Rev Immunol. 2008;8(6):467-477. Miyoshi H, Kiyasu J, Kato T, et al. PD-L1 expression on neoplastic or stromal cells is respectively a poor or good prognostic factor for adult T-cell leukemia/lymphoma. Blood. 2016;128(10):1374-1381. Hicklin DJ, Wang Z, Arienti F, et al. β2microglobulin mutations, HLA class I antigen loss, and tumor progression in melanoma. J Clin Invest. 1998;101(12);2720-2729. Kataoka K, Nagata Y, Kitanaka A, et al. Integrated molecular analysis of adult T cell leukemia/lymphoma. Nat Genet. 2015;47 (11):1304-1315. Asano N, Miyoshi H, Kato T, et al. Expression pattern of immunosurveillancerelated antigen in adult T cell leukaemia/ lymphoma. Histopathology. 2018;72(6): 945-954. Keir ME, Francisco LM, Sharpe AH. PD-1 and its ligands in T-cell immunity. Curr Opin Immunol. 2007;9(3):309-314. Pedoeem A, Azoulay-Alfaguter I, Strazza M, et al. Programmed death-1 pathway in cancer and autoimmunity. Clin Immunol. 2014;153(1):145-152. Melief CMJ. Precision T-cell therapy targets tumours. Nature. 2017;547(7662):217-222. Seliger B. Molecular mechanisms of HLA class I-mediated immune evasion of human tumors and their role in resistance to immunotherapies. HLA. 2016;88(5): 213220.

18. Ansell SM, Lesokhin AM, Borrello I, et al. PD-1 blockade with nivolumab in relapsed or refractory Hodgkin’s lymphoma. N Engl J Med. 2015; 372(4):311-319. 19. Reichel J, Chadbum A, Rubunstein PG, et al. Flow sorting and exome sequencing reveal the oncogene of primary Hodgkin and ReedSternberg cells. Blood. 2015;125(7):10611072. 20. Sconocchia G, Eppenberger-Castori S, Zlobec I, et al. HLA class II antigen expression in colorectal carcinoma tumors as a favorable prognostic marker. Neoplasia. 2014;16(1):31-42. 21. Esteban F, Ruiz-Cabello F, Concha A, et al. HLA-DR expression is associated with excellent prognosis in squamous cell carcinoma of the larynx. Clin Exp Metastasis. 1990;8(4):319-328. 22. Rimsza LM, Robberts RA, Campo E, 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. 23. He Y, Rozeboomb L, Rivardb CJ, et al. MHC class II expression in lung cancer. Lung Cancer. 2017;112:75-80. 24. Johnson DB, Estrada MV, Salgado R, et al. Melanoma-specific MHC-II expression represents a tumour-autonomous phenotype and predicts response to anti-PD1/PD-L1 therapy. Nat Commun. 2016;29 (7):10582. 25. Roemer MG, Advani RH, Redd RA, et al. Classical Hodgkin lymphoma with reduced β2M/MHC class I expression is associated with inferior outcome independent of 9p24.1 status. Cancer Immunol Res. 2016;4(11):910-916.

26. Roemer MGM, Redd RA, Cader FZ, et al. Major histocompatibility complex class II and programmed death ligand expression predict outcome after programmed death 1 blockade in classic Hodgkin lymphoma. J Clin Oncol. 2018; 36(10):942-950. 27. Ratner L, Waldman TA, Janakiram M, et al. Rapid progression of adult T-cell leukemialymphoma after PD-1 inhibitor therapy. N Engl J Med. 2018;378(20):1947-1948. 28. Weisenburger DD1, Savage KJ, Harris NL, et al. International Peripheral T-cell Lymphoma Project. Peripheral T-cell lymphoma, not otherwise specified: a report of 340 cases from the International Peripheral T-cell Lymphoma Project. Blood. 2011;117(12): 3402-3408. 29. Fukushima T, Nomura S, Shimoyama M, et al. Japan Clinical Oncology Group (JCOG) prognostic index and characterization of long term survivors of aggressive adult T cell leukaemia lymphoma (JCOG0902A). Br J Haematol. 2014;166(5):739-748. 30. Sun Z. Chen F, Meng F, Wei J, Liu B. MHC class II restricted neoantigen: a promising target in tumor immunotherapy. Cancer Let. 2017;392:17-25. 31. Holling TM, Schooten E, Langerak AW, van den Elsen PJ. Regulation of MHC class II expression in human T-cell malignancies. Blood. 2004;103(4):1438-1444. 32. Forlani G, Abdallah R, Accol RS, Tosi G. The major histocompatibility complex class II transactivator CIITA inhibits the persistent activation of NF-κB by the human T cell lymphptropic virus type 1 Tax-1 oncoprotein. J Virol. 2016;90(7):3708-3721. 33. Shirono K, Hattori T, Hata H, Nishimura H, Takatsuki K. Profiles of expression of activated cell antigens on perioheral blood and lymph node cells from different clinical stages of sdult T-cell leukemia. Blood. 1989;73(6):1664-1671.

haematologica | 2019; 104(8)

ARTICLE

Non-Hodgkin Lymphoma

The novel CD19-targeting antibody-drug conjugate huB4-DGN462 shows improved anti-tumor activity compared to SAR3419 in CD19-positive lymphoma and leukemia models Stuart W. Hicks,1* Chiara Tarantelli,2* Alan Wilhem,1 Eugenio Gaudio,2 Min Li,1 Alberto J. Arribas,2 Filippo Spriano,2 Roberta Bordone,3 Luciano Cascione,2,4 Katharine C. Lai,1 Qifeng Qiu,1 Monica Taborelli,5 Davide Rossi,2,3 Georg Stussi,3 Emanuele Zucca,3 Anastasios Stathis,3 Callum M. Sloss1 and Francesco Bertoni2

Ferrata Storti Foundation

Haematologica 2019 Volume 104(8):1633-1639

*These authors contributed equally to this work.

ImmunoGen Inc., Waltham, MA, USA; 2Università della Svizzera italiana, Institute of Oncology Research, Bellinzona, Switzerland; 3Oncology Institute of Southern Switzerland, Bellinzona, Switzerland; 4Swiss Institute of Bioinformatics, Lausanne, Switzerland and 5 Cytogenetics Laboratory, Ente Ospedaliero Cantonale, Bellinzona, Switzerland 1

ABSTRACT

ntibody-drug conjugates (ADC) are a novel way to deliver potent cytotoxic compounds to cells expressing a specific antigen. Four ADC targeting CD19, including SAR3419 (coltuximab ravtansine), have entered clinical development. Here, we present huB4-DGN462, a novel ADC based on the SAR3419 anti-CD19 antibody linked via sulfoSPDB to the potent DNA-alkylating agent DGN462. huB4-DGN462 had improved in vitro anti-proliferative and cytotoxic activity compared to SAR3419 across multiple B-cell lymphoma and human acute lymphoblastic leukemia cell lines. In vivo experiments using lymphoma xenografts models confirmed the in vitro data. The response of B-cell lymphoma lines to huB4DGN462 was not correlated with CD19 expression, the presence of BCL2 or MYC translocations, TP53 inactivation or lymphoma histology. In conclusion, huB4-DGN462 is an attractive candidate for clinical investigation in patients with B-cell malignancies.

Correspondence: FRANCESCO BERTONI/CALLUM SLOSS frbertoni@mac.com/ callum.sloss@immunogen.com Received: November 3, 2018. Accepted: February 7, 2019. Pre-published: February 7, 2019.

Introduction doi:10.3324/haematol.2018.211011 Lymphomas are among the most common cancers and causes of cancer-related cell deaths in both adults and children.1 In western countries, B-cell lymphomas, and diffuse large B-cell lymphoma (DLBCL) in particular, represent the vast majority of cases. Despite treatment improvements, a proportion of B-cell lymphoma patients still succumb because of chemorefractory disease, requiring the need to find novel therapeutic modalities to extend patients’ lives.2,3 These neoplastic B cells express many proteins on their cell surface that are potential therapeutic targets. 4 The clinical success of the anti-CD20 monoclonal antibody, rituximab, demonstrates the potential of targeting B-cell specific surface proteins.5 Antibody-drug conjugates (ADC) are an innovative way to deliver potent cytotoxic compounds to cells expressing a specific antigen, as shown by brentuximab vedotin in patients with CD30-positive lymphomas and for ado-trastuzumab emtansine in patients with HER2-positive breast cancer.4 Another plasma membrane target, CD19, is almost exclusively expressed on B cells, appearing at the pre-B-cell stage and remaining expressed until their terminal differentiation into plasma cells6 and unlike CD20, CD19 is rapidly internalized, which makes it better suited for ADC development.68 Four ADC targeting CD19, SAR3419 (coltuximab ravtansine, huB4-DM4),9-13 SGNCD19A (denintuzumab mafodotin),14,15 ADCT-402 (loncastuximab tesirine),16-18 and SGN-CD19B19 have entered clinical development. SAR3419 is composed of the maytansinoid payload DM4 attached to the anti-CD19 humanized monoclonal haematologica | 2019; 104(8)

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1633 ©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.

1633

S.W. Hicks et al.

IgG1 antibody, huB4, via an N-succinimidyl-4-(2pyridyldithio)butyrate (SPDB) cleavable linker.9 SAR3419 demonstrated clinical activity in patients with relapsed/refractory B-cell lymphoma as both a single agent and in combination with rituximab.10-13 To further, improve the clinical benefit of targeting CD19, second generation ADC can be developed incorporating new innovative linkers and payload moieties.20 For example, the addition of a sulfonate group in the disulfide linker SPDB ([N-succinimidyl 2-sulfo-4-(2-pyridyldithio)butanoate], sulfo-SPDB) can increase the potency of ADC bearing similar payloads.21 The indolinobenzodiazepine pseudodimer DGN462 is a potent DNA-alkylating agent with proven anti-tumor activity in preclinical models of solid tumors and acute myeloid leukemia (AML).22 These two innovations have been implemented in the recently developed IMGN779, an anti-CD33 ADC with promising preclinical23,24 and early clinical activity25 in AML. Here, we present huB4-DGN462, a novel ADC utilizing the huB4, antiCD19 antibody, linked via sulfo-SPDB to DGN462. The huB4-ADC incorporating sulfo-SPDB and DGN462 demonstrated improved in vitro and in vivo activity in lymphoma and leukemia models.

Methods Cell lines Lymphoma cell lines, all validated for their identity by short tandem repeat DNA fingerprinting (IDEXX BioResearch, Ludwigsburg, Germany), were used and cultured as previously described.26 BCL2, MYC and TP53 status were defined as reported in Online Supplementary Table S1. Additional cell lines used included the Burkitt's lymphoma Ramos (ACC-603; DSMZ), and the human acute lymphoblastic leukemia (ALL) cell lines RS4;11 (CRL-1873, ATCC), TOM-1 (ACC-578, DSMZ), and BALL-1 (JCRB0071, JCRB).

Antibody-drug conjugates and free payload Antibody-drug conjugates (SAR3419, huB4-DGN462, non-targeting IgG-DGN462 control) and cell-permeable unconjugated toxin S-Methyl-DGN462 (DGN462-SMe) were generated by ImmunoGen. huB4-DGN462 and IgG-DGN462 were produced as described previously.22 Briefly, a 10-fold molar excess of in situ mixture containing DGN462 and sulfo-SPDB was added to huB4 or IgG antibody in buffer (50 mM EPPS [4-(2-Hydroxyethyl)-1piperazinepropanesulfonic acid], pH 8.5) containing 15% dimethylacetamide. Upon completion of conjugation, the reaction mixtures were purified and buffer exchanged into 20 mM histidine, 50 mM sodium chloride, 8.5% w/v sucrose, 0.01% Tween-20, 50 μM sodium bisulfite pH 6.2 using NAP desalting columns (Illustra Sephadex G-25, GE Healthcare).

In vitro anti-tumor activity To determine in vitro cytotoxicity potency, B-cell lymphoma and B-ALL cell lines were treated with a 3-fold dilution series of conjugate for five days without or with a 100-fold concentration of unconjugated huB4 blocking antibody. The relative number of viable cells in each well was then determined using the WST-8 based Cell Counting Kit-8 (Dojindo Molecular Technologies Inc., Rockville, MD, USA). The surviving fraction of cells was plotted against conjugate concentration and the EC50 of activity was calculated using a non-linear regression analysis (GraphPad Prims 4.0). For higher-throughput in vitro cytotoxicity screening,

1634

lymphoma or leukemia cells (104) were seeded in 96-well plates and the indicated compounds were added to create 4-fold dilution series ranging from 200 nM to 0.19 pM and assayed by MTT following 72 hours (h) of treatment, as previously described.24 Apoptosis was assessed using the luminescence-base Caspase-Glo 3/7 assay kit (Promega) according to the manufacturer's instructions. Briefly, cells were seeded in 384-well plates, treated with ADC or free toxin at the following concentrations (50 and 1,000 pM for huB4-DGN462 and DGN462-SMe; 100 and 5,000 pM for SAR3419). Apoptosis was defined by at least a 1.5-fold increase in signal activation with respect to controls. Differences among groups were calculated using the Wilcoxon rank-sum test (Stata/SE 12.1 for Mac, Stata Corporation, College Station, TX, USA). P<0.05 was considered statistically significant.

CD19 expression Fluorescence-activated cell sorting (FACS) for CD19 antigen expression analysis was performed on fresh low passage lymphoma cell lines. Cells were washed with ice-cold FACS buffer (PBS + 0.5% BSA) and divided 1x106 cells/tube. A pretreatment with human FcR blocking (Miltenyi Biotec Inc., Auburn, CA, USA) was performed according to the manufacturer’s instructions. CD19-PE (5 μL; 1μg) or control isotype was incubated with cells at 4°C for 30 minutes. Cells were washed twice and re-suspended in FACS buffer. Flow-cytometry analysis was carried out with a FACS Canto II instrument (BD Biosciences). Median Fluorescence intensity (MFI) of each sample was determined using FacsDiva v.8.0.1 software (BD Biosciences, Allschwil, Switzerland). Unstained cells and cells stained with isotype control antibody were used as controls. RNA expression levels, obtained with the HumanHT-12 v.4 Expression BeadChip (Illumina, San Diego, CA, USA) and with the HTG EdgeSeq Oncology Biomarker panel (HTG Molecular Diagnostics Inc., Tucson, AZ, USA) were published previously.26 For leukemia cell lines, the median number of CD19 antibody binding sites (ABC) was determined as previously described.24 Pairwise correlations were assessed using Stata/SE 12.1 for Mac. P<0.05 was considered statistically significant.

In vivo experiments Experiments using subcutaneous (DOHH2) and disseminated (Farage) xenograft models in female CB.17 severe combined immunodeficient (SCID) mice (Charles River Laboratories, Wilmington, MA, USA) were run following previously described procedures.27 All animal procedures were performed in strict accordance with Immunogen’s Animal Care and Use Committee and the National Institutes of Health Guide for the Care and Use of Laboratory Animals. For the DOHH2 model, mice were inoculated subcutaneously with 107 cells and were randomized by tumor volume (TV) into treatment groups (6 mice/group), when the average TV reached 100 mm3. Tumor volumes were recorded 2-3 times weekly by caliper measurements of the height (H), length (L), and width (W) of the tumor, following the formula: TV = (H x L x W)/2 For the disseminated model, mice were inoculated intravenously (IV) via the lateral tail vein with 107 Farage cells and were randomized into treatment groups (8 mice/group) by body weight. Animal survival was followed; animals were removed from the study when advanced health distress signs were observed or measured (> 20% loss of body weight, hind leg paralysis, tumor growth on body, or moribund state).

haematologica | 2019; 104(8)

A novel anti-CD19 antibody drug conjugate

Results huB4-DGN462 is a novel anti-CD19 antibody-drug conjugate The humanized anti-CD19 antibody huB428 was conjugated with DGN462, a novel member of the DNA-alkylating indolinobenzodiazepine pseudodimers (IGNs) class22 via the cleavable disulfide linker, sulfo-SPDB, at a drug-toantibody ratio (DAR) of 2.8 (Figure 1A). This novel CD19targeting ADC, huB4-sulfo-SPDB-DGN462 or huB4-

DGN462, combines the anti-CD19 antibody used in SAR3419 (Figure 1B)9 and the potent linker/payload exploited in the anti-CD33 IMGN779.24 To evaluate the consequence of conjugation on antigen binding, the relative binding affinity of huB4-DGN462 and the unconjugated huB4 antibody to CD19 was determined by FACS analysis on the Burkitt's lymphoma Ramos cell line, which endogenously express human CD19 (Figure 1C). The huB4-DGN462 bound with similar high affinity (KD ~0.15 nM) as the unconjugated antibody, suggesting that

B A

Figure 1. huB4-DGN462 is a novel anti-CD19 antibody-drug conjugate (ADC). (A) Chemical structure of huB4-DGN462. (B) Chemical structure of SAR3419. (C) Relative binding affinities of huB4 and huB4-DGN462 on Ramous cells. Cells were cultured with the indicated concentrations of either the unconjugated antibody or intact ADC, and binding was detected by flow cytometry using a fluorescently labeled anti-human antibody. (D) In vitro potency of huB4-DGN462 and SAR3419 in diffuse large B-cell lymphoma (DLBCL) (top) or B-cell acute lymphoblastic leukemia (B-ALL) (bottom) cell lines with or without blocking antibody. ABC: antibodies bound per cell.

haematologica | 2019; 104(8)

1635

S.W. Hicks et al.

Table 1. Anti-tumor activity of huB4-DGN462 and SAR3419 in B-cell lymphoma cell lines.

Number of cell lines All GCB-DLBCL ABC-DLBCL MCL MZL CLL PMBCL

46 19 8 10 6 2 1

huB4-DGN462 Median IC50 95%CI 100 pM 100 pM 600 pM 45 pM 65 pM 2250 pM 200 pM

SAR3419

38-214 pM 3 pM-1170 pM 20 pM-3000 pM 5-84 pM 6-910 pM 1500-3000 pM n.d.

Median IC50

95%CI

Wilcoxon rank-sum P

37 nM 2 nM 6.5 nM 3 nM 5 nM 25 nM 10 nM

2.4-5.1 nM 0.7-3.6 nM 3.8-7.3 nM 1.3-9.7 nM 1.5-9.8 nM 10-40 nM n.d.

<0.0001 0.003 0.001 0.017 0.004 0.130 n.d.

GCB-DLBCL: germinal center B-cell type diffuse large B-cell lymphoma; ABC-DLBCL: activated B-cell like diffuse large B-cell lymphoma; MCL: mantle cell lymphoma; MZL: marginal zone lymphoma; CLL: chronic lymphocytic leukemia; PMBCL: primary mediastinal large B-cell lymphoma; n.d.: not determined.

the incorporation of the payload did not appreciably alter antibody binding affinity.

huB4-DGN462 has improved in vitro anti-lymphoma activity compared to SAR3419 To characterize the ability of huB4-DGN462 to kill tumor cells, the in vitro cytotoxicity of huB4-DGN462 was compared to SAR3419 in a subset of B-cell lymphoma and B-ALL tumor cell lines. The tumor cell killing was assayed five days after treatment. While SAR3419 had limited potency, huB4-DGN462 was highly active with IC50s ranging from 1-16 pM (Figure 1D). In addition, unconjugated antibody was able to block the activity of both huB4-DGN462 and SAR3419 confirming that the cytotoxicity of both conjugates is the result of specific CD19 antigen binding on tumor cells. To further characterize the in vitro potency of huB4DGN462, we performed a 72-h in vitro cytotoxicity screen on 46 B-cell lymphoma cell lines, mostly derived from DLBCL, exposed also to SAR3419 (Table 1 and Online Supplementary Table S1). huB4-DGN462 was potent with a median IC50 of 100 pM (95%CI: 38-214). The free payload, DGN462-SMe, had a median IC50 of 26 pM (95%CI: 1-186) (Online Supplementary Table S1). huB4DGN462 induced caspase 3/7 activation consistent with an apoptotic mechanism of action (Online Supplementary Table S1). In agreement, cell cycle analysis in two DLBCL cell lines showed a G2-M arrest with subG0 accumulation of cells after huB4-DGN462 and DGN462-SMe treatment (Online Supplementary Figure S1), similar to that reported for IMGN779, a DGN462-containing ADC.24 While the rank order of anti-proliferative activity for huB4-DGN462 and SAR3419 was similar across the cell lines tested (R=0.42, P=0.004) (Online Supplementary Figure S2), huB4-DGN462 was over 30-times more potent than SAR3419 (P<0.0001) (Table 1 and Online Supplementary Figure S3). Increased cytotoxic activity of huB4-DGN462 was also confirmed in six CD19+ B-ALL cell lines (Online Supplementary Table S2). While a moderate correlation was seen for SAR3419, the response of B-cell lymphoma lines to huB4-DGN462 was not correlated with either surface CD19 protein expression (as measured by FACS) or CD19 RNA expression (as measured by microarray or by targeted RNA-Seq) (Online Supplementary Figures S4 and S5). However, the activity of huB4-DGN462 was positively correlated with the potency of its free payload (R = 0.71, P<0.0001). Among DLBCL cell lines, the sensitivity to the two 1636

ADCs was not affected by TP53 inactivation (present in 16 of 23 cell lines), nor the presence of MYC or BCL2 translocations (detected in 8 of 24 and 13 of 21 cell lines, respectively). The sensitivity to SAR3419, but not the novel ADC, was lower in activated B-cell-like (ABC) than in germinal center B-cell type (GCB) DLBCL cell lines (P=0.02) (Table 1, Online Supplementary Table S1 and Online Supplementary Figure S6A) This was in agreement with the correlation between SAR3419 activity and CD19 expression and an observed higher CD19 expression in GCB than ABC DLBCL cell lines (Online Supplementary Figure S6B).

huB4-DGN462 has superior in vivo anti-lymphoma activity compared to SAR3419 To examine whether the enhanced antitumor activity of huB4-DGN462 observed in vitro would translate to improved efficacy in vivo, the anti-tumor activity of both of CD19-targeting ADC was evaluated in two DLBCL xenograft models. In the subcutaneous DoHH2 model, a single intravenous dose of huB4-DGN462 resulted in a significant, dose-dependent tumor growth delay and survival benefit at 1.7 mg antibody (Ab)/kg (5 of 6 partial responses, 3 of 6 complete responses, 3 of 6 tumor-free survivors) compared to a non-targeted control IgG-DGN462 ADC (1 of 6 partial responses, 1 of 6 complete responses, 1 of 6 tumorfree survivors) (Figure 2A). The anti-tumor activity of huB4DGN462 was better than that obtained in a similar experiment using SAR3419 given at higher doses. Both huB4DGN462 and SAR3419 were well-tolerated, with no significant loss of body weight observed (Online Supplementary Figure S7). In the disseminated Farage model, the 1.7 mg/kg dose of the non-specific IgG1-DGN462 control conjugate was inactive while there was a significant dose-dependent increase in survival observed in mice treated with as low as 0.17 mg Ab/kg of huB4-DGN462 (Figure 2B). At 1.7 mg Ab/kg, survival of tumor bearing mice was increased over 400%. As above, the efficacy of huB4-DGN462 was better than that achieved with SAR3419. Taken together and consistent with the in vitro findings, treatment with huB4DGN462 resulted in a substantial enhancement of antitumor activity compared to SAR3419.

Discussion Here we have compared the anti-lymphoma activity of a novel CD19-targeting ADC, huB4-DGN462, with haematologica | 2019; 104(8)

A novel anti-CD19 antibody drug conjugate

SAR3419, an ADC that demonstrated clinical activity as single agent10,29 and in combination with the anti-CD20 monoclonal antibody rituximab.13 SAR3419 comprises the maytansinoid microtubule disruptor, DM4, conjugated to the huB4 anti-CD19 antibody via the hydrophobic linker SPDB.6,9 Utilizing the same anti-CD19 antibody in SAR3419, we created huB4-DGN462, which incorporates

the ultra-potent DNA-alkylating payload, DGN462 linked by the more hydrophilic sulfo-SPDB linker. This sulfoSPDB-DGN462 linker/payload combination has previously been shown to have significantly greater potency (up to 3 logs) than the SPDB-DM4.22 As predicted, huB4DGN462 was more potent than SAR3419 in in vitro antiproliferative and apoptotic assays as well as in in vivo

Figure 2. huB4-DGN462 exhibits more potent in situ antitumor efficacy than SAR3419 in both disseminated (A) and subcutaneous (B) germinal center B-cell type diffuse large B-cell lymphoma (GCB DLBCL) xenograft models. (A) Nude mice bearing Farage xenografts were treated with vehicle, a single dose of huB4-DGN462 (0.17, 0.56 or 1.7 mg antibody /kg), or a single dose of non-targeting IgG1-DGN462 (1.7 mg antibody /kg), as indicated. (Right) Nude mice bearing Farage xenografts were treated with vehicle or a single dose of SAR3419 (2.5 or 5 mg antibody /kg) as indicated. The table summarizes the data for the experiments. (B) Nude mice with subcutaneous DOHH2 xenografts were treated with vehicle, a single dose of huB4-DGN462 (0.56 or 1.7 mg antibody /kg), or a single dose of non-targeting IgG1-DGN462 (1.7 mg antibody /kg), as indicated. (Right) Nude mice bearing DOHH2 xenografts were treated with vehicle or a single dose of SAR3419 (2.5 or 5 mg antibody /kg) as indicated. The table summarizes the data for the experiments. Ab: antibody. T/C: treated versus control. PR: partial response (defined when the tumor volume at any given measurement point was <50% of the initial pretreatment tumor volume).27 CR: complete response (defined when no palpable tumor could be detected).27 TGD: tumor growth delay.

haematologica | 2019; 104(8)

1637

S.W. Hicks et al.

DLBCL subcutaneous and systemic tumor models. The improved activity of huB4-DGN462 compared to SAR3419 can be attributed to the increased potency of the linker/payload combination of sulfo-SPDB-DGN462 compared to SPDB-DM4. HuB4-DGN462 was highly active in multiple cell lines derived from B-cell lymphomas. The antitumor activity was not affected by the lymphoma histology or DLBCL subtype or TP53, BCL2 or MYC gene status. Despite specificity for CD19, similar to other ADC, the anti-lymphoma activity of huB4-DGN462 did not correlate with the absolute expression level of CD19,30,31 but rather the sensitivity of the cell line to the free payload. The impressive potency of huB4-DGN462 observed is likely to be due to the fact that low target expression was sufficient to deliver a cytotoxic concentration of the novel linker/payload combination. Given the broad expression profile of CD19 in B-cell malignancies, CD19 is an attractive target for cancer therapy. In particular, CD19-targeting immunotherapies have seen recent successes with the US Food and Drug Administration (FDA) approvals of the bispecific T-cell engaging antibody blinatumomab (Blincyto)32-34 and the FDA approvals of two chimeric antigen receptors (CAR) T-cell therapies targeting CD19, tisagenlecleucel (Kymriah)35 and axicabtagene cilolecucel

References 1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2018. CA Cancer J Clin. 2018; 68(1):7-30. 2. Iacoboni G, Zucca E, Ghielmini M, Stathis A. Methodology of clinical trials evaluating the incorporation of new drugs in the firstline treatment of patients with diffuse large B-cell lymphoma (DLBCL): a critical review. Ann Oncol. 2018;29(5):1120-1129. 3. Stathis A, Iasonos A, Seymour JF, et al. th Report of the 14 International Conference on Malignant Lymphoma (ICML) Closed Workshop on Future Design of Clinical Trials in Lymphomas. Clin Cancer Res. 2018;24(13):2993-2998. 4. Moek KL, de Groot DJA, de Vries EGE, Fehrmann RSN. The antibody-drug conjugate target landscape across a broad range of tumour types. Ann Oncol. 2017; 28(12):3083-3091. 5. Karmali R, Kimby E, Ghielmini M, Flinn IW, Gordon LI, Zucca E. Rituximab: a benchmark in the development of chemotherapy-free treatment strategies for follicular lymphomas. Ann Oncol. 2018; 29(2):332-340. 6. Blanc V, Bousseau A, Caron A, Carrez C, Lutz RJ, Lambert JM. SAR3419: an antiCD19-Maytansinoid Immunoconjugate for the treatment of B-cell malignancies. Clin Cancer Res. 2011;17(20):6448-6458. 7. Nathwani N, Chen R. Antibody drug conjugates in lymphoma. Clin Investig (Lond). 2014;6(10):915-922. 8. Gerber HP, Kung-Sutherland M, Stone I, et al. Potent antitumor activity of the antiCD19 auristatin antibody drug conjugate hBU12-vcMMAE against rituximab-sensitive and -resistant lymphomas. Blood. 2009;113(18):4352-4361. 9. Hong EE, Erickson H, Lutz RJ, et al. Design of coltuximab ravtansine, a CD19-targeting antibody-Drug Conjugate (ADC) for the

1638

10.

11.

12.

13.

14.

15.

16.

(Yescarta).36 Despite the success of CD19-targeting immunotherapies, there still remains the need for more effective therapies, particularly in patients with relapsed/refractory lymphomas such as DLBCL. The clinical activity of CD19-targeting ADC following other antiCD19 directed therapeutic modalities32-39 will have to be determined. However, unlike immunotherapies that stimulate the patientâ&#x20AC;&#x2122;s immune systems to fight cancer, ADC like huB4-DGN462 kill tumor cells by delivering a cytotoxic payload directly to tumor cells. Given the distinct difference in mechanism of action of ADC to that of immunotherapies, patients have the potential to benefit from both types of therapies. This work demonstrates that huB4-DGN462 has superior preclinical antitumor activity compared to clinicallyvalidated SAR3419. Each of the individual components of huB4-DGN462 (antibody, linker and payload) have passed early clinical evaluation13,25 making huB4-DGN462 an attractive candidate for clinical investigation with the potential to extend the lives of patients with B-cell malignancies. Funding Partially supported by institutional research funds from ImmunoGen and the Gelu Foundation (to FB).

treatment of B-cell malignancies: structureactivity relationships and preclinical evaluation. Mol Pharm. 2015;12(6):1703-1716. Younes A, Kim S, Romaguera J, et al. Phase I multidose-escalation study of the antiCD19 maytansinoid immunoconjugate SAR3419 administered by intravenous infusion every 3 weeks to patients with relapsed/refractory B-cell lymphoma. J Clin Oncol. 2012;30(22):2776-2782. Ribrag V, Dupuis J, Tilly H, et al. A doseescalation study of SAR3419, an anti-CD19 antibody maytansinoid conjugate, administered by intravenous infusion once weekly in patients with relapsed/refractory B-cell non-Hodgkin lymphoma. Clin Cancer Res. 2014;20(1):213-220. Trneny M, Verhoef G, Dyer MJ, et al. Starlyte phase II study of coltuximab ravtansine (CoR, SAR3419) single agent: Clinical activity and safety in patients (pts) with relapsed/refractory (R/R) diffuse large B-cell lymphoma (DLBCL; NCT01472887). J Clin Oncol. 2014;32(15_suppl):8506-8506. Coiffier B, Thieblemont C, de Guibert S, et al. A phase II, single-arm, multicentre study of coltuximab ravtansine (SAR3419) and rituximab in patients with relapsed or refractory diffuse large B-cell lymphoma. Br J Haematol. 2016;173(5):722-730. Law C-L, Sutherland M, Miyamoto J, et al. Abstract 625: Preclinical characterization of an auristatin-based anti-CD19 drug conjugate, SGN-19A. Cancer Res. 2011;71(8 Supplement):625-625. Chen RW, Jacobsen ED, Kostic A, Liu T, Moskowitz CH. A randomized, phase 2 trial of denintuzumab mafodotin and RICE vs RICE alone in the treatment of patients (pts) with relapsed/refractory (r/r) diffuse large B-cell lymphoma (DLBCL) who are candidates for autologous stem cell transplant (ASCT). J Clin Oncol. 2016; 34(15_suppl):TPS7584-TPS7584. Zammarchi F, Corbett S, Adams L, et al.

17.

18.

19.

20.

21.

22.

23.

24.

ADCT-402, a PBD dimer-containing antibody drug conjugate targeting CD19expressing malignancies. Blood. 2018; 131(10):1094-1105. Caimi P, Kahl BS, Hamadani M, et al. Safety and efficacy of Adct-402 (loncastuximab tesirine), a novel antibody drug conjugate, in relapsed/refractory follicular lymphoma and mantle cell lymphoma: interim results from the Phase 1 first-in-human Study. Blood. 2018;132(Suppl 1):2874-2874. Radford J, Kahl BS, Hamadani M, et al. Interim Results from the First-in-Human Clinical Trial of Adct-402 (Loncastuximab Tesirine), a Novel PyrrolobenzodiazepineBased Antibody Drug Conjugate, in Relapsed/Refractory Diffuse Large B-Cell Lymphoma. Blood. 2018;132(Suppl 1):398398. Ryan MC, Palanca-Wessels MC, Schimpf B, et al. Therapeutic potential of SGNCD19B, a PBD-based anti-CD19 drug conjugate, for treatment of B-cell malignancies. Blood. 2017;130(18):2018-2026. Beck A, Goetsch L, Dumontet C, Corvaia N. Strategies and challenges for the next generation of antibody-drug conjugates. Nat Rev Drug Discov. 2017;16(5):315-337. Zhao RY, Wilhelm SD, Audette C, et al. Synthesis and evaluation of hydrophilic linkers for antibody-maytansinoid conjugates. J Med Chem. 2011;54(10):3606-3623. Miller ML, Fishkin NE, Li W, et al. A New Class of Antibody-Drug Conjugates with Potent DNA Alkylating Activity. Mol Cancer Ther. 2016;15(8):1870-1878. Whiteman KR, Noordhuis P, Walker R, et al. The Antibody-Drug Conjugate (ADC) IMGN779 Is Highly Active in Vitro and in Vivo Against Acute Myeloid Leukemia (AML) with FLT3-ITD Mutations. Blood. 2014;124(21):2321-2321. Kovtun Y, Noordhuis P, Whiteman KR, et al. IMGN779, a Novel CD33-Targeting Antibody-Drug Conjugate with DNA-

haematologica | 2019; 104(8)

A novel anti-CD19 antibody drug conjugate

25.

26.

27.

28.

Alkylating Activity, Exhibits Potent Antitumor Activity in Models of AML. Mol Cancer Ther. 2018;17(6):1271-1279. Cortes JE, Traer E, Wang ES, et al. IMGN779, a Next-Generation CD33Targeting Antibody-Drug Conjugate (ADC) Demonstrates Initial Antileukemia Activity in Patients with Relapsed or Refractory Acute Myeloid Leukemia. Blood. 2017;130(Suppl 1):1312-1312. Tarantelli C, Gaudio E, Arribas AJ, et al. PQR309 Is a Novel Dual PI3K/mTOR Inhibitor with Preclinical Antitumor Activity in Lymphomas as a Single Agent and in Combination Therapy. Clin Cancer Res. 2018;24(1):120-129. Hicks SW, Lai KC, Gavrilescu LC, et al. The Antitumor Activity of IMGN529, a CD37Targeting Antibody-Drug Conjugate, Is Potentiated by Rituximab in Non-Hodgkin Lymphoma Models. Neoplasia. 2017;19(9):661-671. Al-Katib AM, Aboukameel A, Mohammad R, Bissery MC, Zuany-Amorim C. Superior antitumor activity of SAR3419 to rituximab in xenograft models for non-Hodgkin's lymphoma. Clin Cancer Res. 2009;15(12):4038-4045.

haematologica | 2019; 104(8)

29. Kantarjian HM, Lioure B, Kim SK, et al. A Phase II Study of Coltuximab Ravtansine (SAR3419) Monotherapy in Patients With Relapsed or Refractory Acute Lymphoblastic Leukemia. Clin Lymphoma Myeloma Leuk. 2016;16(3):139-145. 30. Yu SF, Zheng B, Go M, et al. A Novel AntiCD22 Anthracycline-Based Antibody-Drug Conjugate (ADC) That Overcomes Resistance to Auristatin-Based ADCs. Clin Cancer Res. 2015;21(14):3298-3306. 31. Deckert J, Park PU, Chicklas S, et al. A novel anti-CD37 antibody-drug conjugate with multiple anti-tumor mechanisms for the treatment of B-cell malignancies. Blood. 2013;122(20):3500-3510. 32. Przepiorka D, Ko CW, Deisseroth A, et al. FDA Approval: Blinatumomab. Clin Cancer Res. 2015;21(18):4035-4039. 33. Pulte ED, Vallejo J, Przepiorka D, et al. FDA Supplemental Approval: Blinatumomab for Treatment of Relapsed and Refractory Precursor B-Cell Acute Lymphoblastic Leukemia. Oncologist. 2018;23(11):13661371. 34. Jen EY, Xu Q, Schetter A, et al. FDA Approval: Blinatumomab for Patients with B-cell Precursor Acute Lymphoblastic

35.

36.

37.

38.

39.

Leukemia in Morphologic Remission with Minimal Residual Disease. Clin Cancer Res. 2019;25(2):173-477. O'Leary MC, Lu X, Huang Y, et al. FDA Approval summary: tisagenlecleucel for treatment of patients with relapsed or refractory B-cell precursor acute lymphoblastic leukemia. Clin Cancer Res. 2018 Oct 11. [Epub ahead of print] Bouchkouj N, Kasamon YL, de Claro RA, et al. FDA Approval summary: axicabtagene ciloleucel for relapsed or refractory large Bcell lymphoma. Clin Cancer Res. 2018 Nov 9. [Epub ahead of print] Chow VA, Shadman M, Gopal AK. Translating anti-CD19 CAR T-cell therapy into clinical practice for relapsed/refractory diffuse large B-cell lymphoma. Blood. 2018;132(8):777-781. Jurczak W, Zinzani PL, Gaidano G, et al. Phase IIa study of the CD19 antibody MOR208 in patients with relapsed or refractory B-cell non-Hodgkin's lymphoma. Ann Oncol. 2018;29(5):1266-1272. Munshi PN, Ujjani C. The acceleration of CAR-T therapy in non-Hodgkin lymphoma. Hematol Oncol. 2018 Nov 14. [Epub ahead of print]

1639

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(8):1640-1647

Plasma Cell Disorders

Once-weekly versus twice-weekly carfilzomib in patients with newly diagnosed multiple myeloma: a pooled analysis of two phase I/II studies Sara Bringhen,1 Roberto Mina,1 Maria Teresa Petrucci,2 Gianluca Gaidano,3 Stelvio Ballanti,4 Pellegrino Musto,5 Massimo Offidani,6 Stefano Spada,1 Giulia Benevolo,7 Elena Ponticelli,1 Piero Galieni,8 Michele Cavo,9 Tommaso Caravita Di Toritto,10* Francesco Di Raimondo,11 Vittorio Montefusco,12 Antonio Palumbo,1** Mario Boccadoro1 and Alessandra Larocca1

Myeloma Unit, Division of Hematology, University of Torino, Azienda OspedalieroUniversitaria Città della Salute e della Scienza di Torino, Torino; 2Division of Hematology, Department of Cellular Biotechnologies and Hematology, Sapienza University of Rome, Rome; 3Division of Hematology, Department of Translational Medicine, Università del Piemonte Orientale, Novara; 4Sezione di Ematologia e Immunologia Clinica, Ospedale Santa Maria della Misericordia di Perugia, Perugia; 5Unit of Haematology and Stem Cell Transplantation, IRCCS-CROB, Referral Cancer Center of Basilicata, Rionero in Vulture; 6 Clinica di Ematologia, AOU Ospedali Riuniti di Ancona, Ancona; 7Hematology, Città della Salute e della Scienza, Turin; 8Division of Hematology, Ospedale “C. e G. Mazzoni”, ASUR Marche-AV5, Ascoli Piceno; 9"Seràgnoli" Institute of Hematology, Bologna University School of Medicine, Bologna; 10UOC Ematologia, Ospedale S. Eugenio, ASLRM2, Rome; 11Division of Hematology, AOU Policlinico-OVE, University of Catania, Catania and 12Hematology Department, Fondazione IRCCS Istituto Nazionale Tumori, Milano, Italy 1

*currently employed at UOSD Ematologia, P.O. S. Spirito e Nuovo Regina Margherita ASLRM1, Rome; **currently employed at Takeda Pharmaceuticals International, Switzerland

ABSTRACT

Correspondence: SARA BRINGHEN sarabringhen@yahoo.com Received: October 4, 2018. Accepted: February 7, 2019. Pre-published: February 7, 2019. doi:10.3324/haematol.2018.208272 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1640 ©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.

1640

wice-weekly carfilzomib is approved at 27 and 56 mg/m2 to treat relapsed multiple myeloma patients. In the phase III study ARROW, once-weekly 70 mg/m2 carfilzomib prolonged the median progression-free survival of relapsed multiple myeloma patients in comparison with twice-weekly 27 mg/m2 carfilzomib, without adding significant toxicity. Data were pooled from two phase I/II studies of newly diagnosed multiple myeloma patients who received nine induction cycles of carfilzomib (either 70 mg/m2 once-weekly or 36 mg/m2 twice-weekly), cyclophosphamide and dexamethasone, followed by carfilzomib maintenance. Overall, 121 transplant-ineligible patients with newly diagnosed multiple myeloma were analyzed (once-weekly, n=63; twice-weekly, n=58). We found no significant difference in median progression-free survival [35.7 months (95%CI: 23.7-not reached, NR) vs. 35.5 months (95%CI: 24.3-NR); HR: 1.39; P=0.26] and 3-year overall survival [70% [95%CI: 59%-84%) vs. 72% (95%CI: 60%85%); HR: 1.27; P=0.5] between once-weekly and twice-weekly carfilzomib. From the start of maintenance, 3-year progression-free survival [47% (95%CI: 33%-68%) vs. 51% (95%CI: 38%-70%); HR: 1.04; P=0.92] and overall survival [72% (95%CI: 58%-89%) vs. 73% (95%CI: 59%-90%); HR: 0.82; P=0.71] were similar in the onceversus twice-weekly carfilzomib. The rate of grade 3-5 hematologic (24% vs. 30%; P=0.82) and non-hematologic (38% vs. 41%; P=0.83) adverse events was similar in the two groups. Onceweekly 70 mg/m2 carfilzomib as induction and maintenance therapy for newly diagnosed multiple myeloma patients was as safe and effective as twice-weekly 36 mg/m2 carfilzomib and provided a more convenient schedule. The trials are registered at clinicaltrials.gov identifiers: 01857115 (IST-CAR561) and 01346787 (IST-CAR-506). haematologica | 2019; 104(8)

Once- vs. Twice-Weekly Carfilzomib in NDMM

Introduction In the last two decades, several novel agents of various classes have been developed and approved to treat multiple myeloma (MM), resulting in improved overall survival (OS) for both transplant-eligible and -ineligible patients.1 Among new agents, the immunomodulatory drugs (IMiD) thalidomide and lenalidomide, and the proteasome inhibitor (PI) bortezomib, have been included in the initial treatment for newly diagnosed (ND) MM patients. Bortezomib, a first-generation PI, proved to be a very effective anti-MM agent. It was initially approved for the relapse setting and then approved for upfront therapy. Despite the efficacy of bortezomib, its long-term administration is limited by the emergence of peripheral neuropathy, which was reported in 4-13% of patients (grade 3-4).2,3 Carfilzomib, a second-generation PI, showed significant activity among patients with relapsed and/or refractory (RR) MM and was approved by US Food and Drug Administration and the European Medicines Agency in combination with dexamethasone or lenalidomide-dexamethasone (Rd) for the treatment of RRMM patients. Given the efficacy displayed by carfilzomib in the relapse setting, several trials tested carfilzomib as part of upfront therapy for NDMM patients, either with Rd (KRd) or with alkylating agents, such as melphalan-prednisone (KMP) or cyclophosphamide-dexamethasone (KCyd).4-7 Carfilzomib is currently approved with the twice-weekly schedule at a dose of 27 mg/m2 over a 2-10-minute (min) infusion period when administered alone or in combination with lenalidomide and dexamethasone, or at a dose of 56 mg/m2 over a 30-min infusion period when given in combination with dexamethasone (Kd).

Nonetheless, other doses (up to 70 mg/m2) and schedules (once weekly) have been shown to be promising. The current twice-weekly schedule may not be very convenient for patients (particularly for elderly patients with limited access to hospital facilities), affecting their quality of life and treatment compliance. In order to improve the convenience of the carfilzomib schedule, preliminary studies tested higher doses of carfilzomib administered in a once-weekly schedule. The phase Ib/II CHAMPION-1 study tested different doses of once-weekly carfilzomib in RRMM patients to define its maximum tolerated dose (MTD) combined with dexamethasone.8 The MTD of once-weekly carfilzomib proved to be 70 mg/m2 over a 30-min infusion period, displaying good efficacy and tolerability. Based on these results, a phase III study (ARROW) was initiated to compare twice-weekly carfilzomib at the dose of 27 mg/m2 with once-weekly carfilzomib at the dose of 70 mg/m2.9 Among 578 RRMM patients, once-weekly carfilzomib improved the overall response rate (ORR; 62.9% vs. 40.8%) and prolonged median progression-free survival (PFS) as compared to twice-weekly carfilzomib (median PFS, 11.2 vs. 7.6 months), with a similar rate of grade 3-4 adverse events (68% vs. 62%). A major limitation of the ARROW study was the low dose (27 mg/m2) of carfilzomib in the twiceweekly arm as compared to the 70 mg/m2 dose adopted in the once-weekly arm. This low dose was determined according to the carfilzomib approval at the time of study design. We previously published data from two phase I/II (ISTCAR-561) and phase II (IST-CAR-506) studies investigating once-weekly (70 mg/m2) and twice-weekly (36 mg/m2) carfilzomib combined with cyclophosphamide and dex-

Table 1. Patients' characteristics.

Age, median (range) â&#x2030;Ľ75 years Sex, female Serum creatinine, mg/dL, median (range) ISS 1 2 3 FISH Standard risk High risk Missing Frailty Score Fit Intermediate fitness Frail LDH, [UI/mol] median (range) Missing

All patients N=121

IST-CAR-561 Once-weekly N=63

IST-CAR-506 Twice-weekly N=58

72 (55-86) 31 (26%) 68 (56%) 0.90 (0.46-3.7)

72 (60-85) 14 (22%) 37 (59%) 0.82 (0.5-3.7)

71 (55-86) 17 (29%) 31 (53%) 1.00 (0.46-2.92)

0.67 0.41 0.59 0.06

40 (33%) 38 (31%) 43 (36%)

24 (38%) 19 (30%) 20 (32%)

16 (28%) 19 (33%) 23 (40%)

0.46

57 (47%) 37 (31%) 27 (22%)

24 (38%) 19 (30%) 20 (32%)

33 (57%) 18 (31%) 7 (12%)

0.4

67 (55%) 40 (33%) 14 (12%) 296 (81-768) 20 (17%)

37 (59%) 22 (35%) 4 (6%) 306 (100-768) 2 (3%)

30 (52%) 18 (31%) 10 (17%) 278 (81-654) 18 (31%)

0.19

0.35

N: number; IQR: interquartile range; LDH: lactate dehydrogenase; ISS: International Staging System; FISH: fluorescence in situ hybridization.

haematologica | 2019; 104(8)

1641

S. Bringhen et al.

amethasone (KCyd) as initial treatment for transplantineligible NDMM patients.6,7 In both trials, KCyd was shown to be a safe and effective option for NDMM patients. Here we report the results of a pooled analysis of these two studies.

Methods Study design and participants For this analysis, we pooled together data from two phase I/II (IST-CAR-561; clinicaltrials.gov identifier: 01857115) and phase II (IST-CAR 506; clinicaltrials.gov identifier: 01346787) studies; these studies were led by the same co-operative groups. Patients were recruited from 14 sites across Italy (hospitals, clinics, oncology or medical centers). Both trials enrolled NDMM patients older than 65 years of age or younger but not eligible for autologous stem-cell transplantation. Inclusion and exclusion criteria are similar between the two source studies and have been previously published.6,7 Ethics committees or institutional review boards at the

study sites approved both studies, which were carried out in accordance with the Declaration of Helsinki. All patients provided written informed consent.

Procedures In both studies, patients received nine 4-week induction cycles with carfilzomib, cyclophosphamide (orally, 300 mg on days 1, 8 and 15) and dexamethasone (40 mg on days 1, 8, 15 and 22). In the IST-CAR 561 study, patients received once-weekly carfilzomib at the dose of 70 mg/m2 (on days 1, 8 and 15), while in the IST-CAR 506 study patients received twice-weekly carfilzomib at the dose of 36 mg/m2 (on days 1, 2, 8, 9, 15 and 16). After the induction phase, patients received maintenance treatment with carfilzomib as single agent, which was administered at the same dose and schedule of the induction phase and until progressive disease or intolerable toxicity. Details of study procedures have been previously published.6,7

Outcomes Focusing on patients who received onceversus twice-weekly

Figure 1. Analysis profile. N: number; AE: adverse events; PD: progressive disease; SPM: second primary malignancy.

1642

haematologica | 2019; 104(8)

Once- vs. Twice-Weekly Carfilzomib in NDMM

carfilzomib, the primary goals of this analysis were: (1) to compare PFS, PFS-2 and OS from the date of entry onto the trial in the intention-to-treat (ITT) population; (2) to compare PFS from start of maintenance therapy (PFS_m), PFS 2 from start of maintenance therapy (PFS-2_m) and overall survival from start of maintenance therapy (OS_m) in a population who completed the induction phase and started maintenance treatment. (Note that PFS-2 was calculated from the date of enrollment to the date of second relapse/progression or death or the date the patient was last known to be in remission.) Secondary end points were responses, time to response, and safety of onceversus twice-weekly carfilzomib. Responses were recorded at the beginning of every cycle, according to the International Myeloma Working Group (IMWG) criteria. All adverse events (AE) were assessed during each cycle and graded according to the National Cancer Institute's Common Terminology Criteria for Adverse Events (version 4.0).10 Fluorescence in situ hybridization (FISH) was centrally assessed with a 10% cut-off for numerical aberrations and a 15% cut-off for IgH translocations; high-risk FISH was defined by the presence of at least one of the following chromosomal abnormalities: del(17p), t(4;14) or t(14;16).11 Frailty status was evaluated according to the IMWG Frailty Score.12 The intention-to-treat population consisted of all the enrolled patients and was the basis for the analysis of efficacy end points. Patients were analyzed according to initial treatment assignment. The safety population was defined as all the enrolled patients who received at least one dose of carfilzomib, cyclophosphamide or dexamethasone, and was the basis for the analysis of the safety end points.

Statistical analysis Data from the two studies were pooled together and analyzed. Comparisons between different patient groups were investigated using Fisher exact test. Time to response was calculated from the start of treatment to the date of the first response [complete remission (CR), partial remission (PR)]. PFS was calculated from the date of enrollment to the date of progression or death or the date the patient was last known to be in remission. PFS-2 was calculated from the date of enrollment to the date of second relapse/progression or death or the date the patient was last known to be in remission. OS was calculated from the date of enrollment to the date of death or the date the patient was last known to be alive. PFS_m, PFS-2_m and OS_m were calculated from the date of the start of maintenance therapy. In order to account for potential confounders, the comparison onceversus twice-weekly carfilzomib was adjusted for age, International Staging System (ISS), FISH, Frailty Score, and, in relation to the maintenance analysis, for response to induction therapy. Time-to-event data were analyzed using the Kaplan-Meier method; survival curves were compared with the log-rank test. Results are presented as hazard ratios (HRs), 95% confidence intervals (95% CIs), and two-sided P-values. Data were censored on September 30th 2015 for the IST-CAR-506 study and on April 30th 2018 for the IST-CAR-561 study. Data were analyzed using R software (version 3.5.1).

Results One hundred and twenty-one transplant-ineligible NDMM patients were analyzed: 63 from the IST-CAR-

Table 2. Grade 3-5 treatment-related adverse events during induction and maintenance therapy.

Grade 3-5 AE

IST-CAR-561 Once-weekly N=63 (%)

IST-CAR-506 Twice-weekly N=56 (%) Overall

At least 1 hematologic AE Anemia Neutropenia Thrombocytopenia At least 1 non-hematologic AE Cardiac - Heart failure - Myocardial infarction - Atrial fibrillation/flutter - Sudden death Vascular - Hypertension Gastrointestinal Infection Nervous Respiratory - Pulmonary edema Fatigue Creatinine increase At least 1 dose reduction for carfilzomib Patients who discontinued carfilzomib due to AE

15 (24) 2 (3) 13 (21) 4 (6) 24 (38) 4 (6) 3 (5) 0 0 1 (2) 5 (8) 4 (6) 3 (5) 5 (8) 2 (3) 4 (6) 3 (5) 0 2 (3) 18 (29) 17 (27)

17 (30) 6 (11) 12 (21) 3 (5) 23 (41) 5 (9) 2 (4) 1 (2) 2 (4) 0 6 (11) 2 (4) 3 (5) 3 (5) 3 (5) 1 (2) 1 (2) 3 (5) 0 17 (30) 17 (30)

AE: adverse events.

haematologica | 2019; 104(8)

1643

S. Bringhen et al.

561 study (once-weekly carfilzomib) and 58 from the IST-CAR-506 study (twice-weekly carfilzomib). Patients' characteristics are listed in Table 1. The median age at diagnosis in the entire population was 72 years (range, 5586 years). Cytogenetic data were available in 94 patients: 37 (31%) had high-risk chromosomal abnormalities by FISH, including 10% of patients with t(4;14), 3% with t(14;16), and 18% with del(17p), while 57 patients (47%) were classified as standard-risk. No significant differences were observed in the two groups between the percentage of patients with ISS 3 disease (32% vs. 40%; P=0.45), high-risk FISH (30% vs. 31%; P=0.40) or Frailty Score (6% vs. 17%; P=0.09). The median follow up of the entire cohort was 39 months [interquartile range (IQR): 31-47], without any difference between the two groups. Overall, 119 of 121 patients enrolled in the studies started induction therapy (Figure 1): 63 in the once-weekly group and 56 in the twice-weekly group. Two patients did not start therapy in the twice-weekly group: one withdrew consent and one was lost to follow up. Ninety patients entered the maintenance phase: 47 (75%) and 43 (74%) in the once- and twice-weekly groups, respectively (Figure 1). In the ITT population, the median PFS from enrollment was 35.7 months (95%CI: 23.7-NR) in the once-weekly group and 35.5 months (95%CI: 24.3-NR) in the twice-

weekly group, with, respectively, 47% and 49% of patients alive and free from progression at three years (Figure 2A). When adjusting for age, ISS, FISH, and Frailty Score, no significant differences in the risk of progression or death were observed between the once-weekly and the twice-weekly carfilzomib groups (HR: 1.39; P=0.26). Median PFS-2 was similar in patients receiving onceweekly (48.6 months; 95%CI: 36.5-NR) and twice-weekly (48.5 months; 95%CI: 44.1-NR) carfilzomib (HR: 1.25; P=0.51) (Figure 2B). At three years, median OS was not reached in either group, with 70% and 72% of patients alive in the two groups, respectively (Figure 2C). No difference in the risk of death was observed between the once-weekly and the twice-weekly carfilzomib groups when adjusting for age, ISS, FISH and Frailty Score (HR: 1.27; P=0.50). We also assessed PFS and OS according to cytogenetic risk. No significant difference in 3-year PFS (52% vs. 43%; HR: 0.76; P=0.38) and 3-year OS (78% vs. 73%; HR: 0.71; P=0.36) was reported between standardand high-risk FISH patients, with a greater reduction in the risk of progression or death in the once-weekly (HR: 1.17; P=0.72) than in the twice-weekly carfilzomib group (HR: 0.52; P=0.12; interaction P=0.19). The median duration of maintenance was 17 months (IQR: 4-28) in the once-weekly and 20 months (IQR 7-32) in the twice-weekly group (P=0.17). At three years, PFS_m

Figure 2. Once-weekly versus twice-weekly carfilzomib in patients with newly diagnosed multiple myeloma. (A) Intention-to-treat progression-free survival (ITT PFS). (B) Intention-to-treat progression-free survival 2 (ITT PFS-2). (C) Intention-to-treat overall survival (ITT OS). (Note that PFS-2 was calculated from the date of enrollment to the date of second relapse/progression or death or the date the patient was last known to be in remission.)

1644

haematologica | 2019; 104(8)

Once- vs. Twice-Weekly Carfilzomib in NDMM

was 47% (95%CI: 33%-68%) and 51% (95%CI: 38%70%) in the once-weekly group and in the twice-weekly group, respectively (Figure 3A), with no significant difference in the risk of progression (HR: 1.04; P=0.92) within the two groups when adjusting for age, ISS, FISH, Frailty Score and response to induction. No differences in 3-year PFS-2_m (65% vs. 74%; HR: 0.85; P=0.74) and OS_m (72% vs. 73%; HR: 0.82; P=0.71) were observed between the two groups (Figure 3B and C). Overall, the proportion of patients achieving a PR or better was 92% in the once-weekly versus 90% in the twice-weekly group (P=0.76), including 22% and 29% of patients obtaining a CR or better (P=0.41). Responses were rapid: median time to PR or better was 1.9 months in the once-weekly group and 1.2 months in the twiceweekly group. Carfilzomib dose reduction was necessary in 18 (29%) patients receiving the once-weekly schedule and in 17 (30%) patients receiving the twice-weekly schedule. The median relative dose intensity of carfilzomib [once weekly 97.6% (IQR 88.3-100%); twice weekly 97.2% (IQR 90.4-100%)] was similar in the two groups (P=0.75). Dexamethasone dose reduction was necessary in 13 (21%) patients receiving the once-weekly schedule and in 18 (32%) patients receiving the twice-weekly schedule. The median relative dose intensity of dexamethasone

[once weekly 100% (IQR 82.6-100%); twice weekly 100% (IQR 88.5-100%)] was similar in the two groups (P=0.85). Cyclophosphamide dose reduction was necessary in 7 (11%) patients receiving the once-weekly schedule and in 15 (27%) patients receiving the twice-weekly schedule. Nevertheless, the median relative dose intensity of cyclophosphamide [once weekly 96.85% (IQR 90.8100%); twice weekly 96.75% (IQR 88.6-100%)] was similar in the two groups (P=0.97). The most common AE leading to carfilzomib dose reduction were acute kidney injury (1 patient in the once-weekly group and 2 patients in the twice-weekly group), infections (2 patients in each group), and hypertension (4 patients in the once-weekly group and none in the twice-weekly group). Treatmentrelated AE leading to the discontinuation of carfilzomib occurred in 17 (27%) patients in the once-weekly group and 17 (30%) patients in the twice-weekly group. The most common AE leading to carfilzomib discontinuation were cardiac injury (6 patients in the once-weekly group and 6 patients in the twice-weekly group), infections (3 patients in the once-weekly group and 3 patients in the twice-weekly group), and thromboembolism (2 patients in the once-weekly group and 1 in the twice-weekly group). Cardiac events leading to drug discontinuation during induction (3 and 2) and maintenance (3 and 4) occurred at similar rates in patients receiving onceversus

Figure 3. Analysis from start of maintenance therapy. (A) Progression-free survival from start of maintenance therapy (PFS_m). (B) Progression-free survival 2 from start of maintenance therapy (PFS-2_m). (C) Overall survival from start of maintenance therapy (OS_m). (Note that PFS-2 was calculated from the date of enrollment to the date of second relapse/progression or death or the date the patient was last known to be in remission.)

haematologica | 2019; 104(8)

1645

S. Bringhen et al.

twice-weekly carfilzomib, respectively. Similarly, the rates of infections leading to drug discontinuation were similar during induction (2 and 3) and maintenance (1 and 0) between the onceversus twice-weekly groups. Overall, the incidence of treatment-related grade 3-5 AE was similar in the once-weekly and the twice-weekly carfilzomib groups, both in terms of hematologic (24% vs. 30%; P=0.82) and non-hematologic (38% vs. 41%; P= 0.83) AE. The most frequent non-hematologic grade ≥3 AE were infections [5 (8%) in the once-weekly group vs. 3 (5%) in the twice-weekly group], respiratory [4 (6%) vs. 1 (2%)], cardiac [4 (6%) vs. 5 (9%)] and hypertension [4 (6%) vs. 2 (4%)]. The incidence of treatment-related grade 3-5 AE during carfilzomib maintenance was low, with comparable rates of hematologic (0% vs. 5%) and nonhematologic (21% vs. 23%) AE in the once-weekly and the twice-weekly groups. The most frequent ≥3 AE was hypertension [4 (9%) vs. none]. All AE are reported in Table 2.

Discussion In this pooled analysis of two phase I/II studies comparing two alternative schedules of carfilzomib, transplantineligible NDMM patients who received once-weekly carfilzomib at the dose of 70 mg/m2 showed similar response rates as compared to patients treated with twiceweekly carfilzomib at the dose of 36 mg/m2. Moreover, the analysis did not report differences in terms of PFS, PFS2, and OS. Administering high-dose carfilzomib (70 mg/m2) in a once-weekly schedule did not impair the safety profile of the KCyd combination in comparison with a lower (36 mg/m2) twice-weekly schedule. To date, two doses of twice-weekly carfilzomib, 27 mg/m2 and 56 mg/m2, have been approved for the treatment of RRMM patients, based on the results of the phase III ASPIRE and ENDEAVOR trials. In the ASPIRE study, carfilzomib was tested at the dose of 27 mg/m2.13 However, a higher dose of carfilzomib (36 mg/m2) had been investigated in combination with lenalidomide and dexamethasone, and was shown to be safe and effective for NDMM patients.4,14-16 In the ENDEAVOR trial, which compared Kd versus bortezomib-dexamethasone (Vd), carfilzomib was administered at the dose of 56 mg/m2.17,18 Despite the great results yielded by the introduction of carfilzomib, treatment compliance and quality of life of young active patients, as well as those of elderly patients with reduced mobility, are compromised by the need for frequent visits to the outpatient clinic for carfilzomib dosing. From this point of view, a shift from the current twice-weekly to a once-weekly dosing schedule would decrease by 50% patient visits to health care facilities, with a subsequent improvement in quality of life and a reduction in drug and health care costs. For these purposes, higher doses of carfilzomib, administered once-weekly, were tested in the relapse setting in a phase Ib/II study and in a subsequent phase III study.6-9 Once-weekly carfilzomib yielded a higher ORR as compared to twice-weekly carfilzomib, resulting in prolonged median PFS (11.2 vs. 7.6 months) without significantly increasing the rate of AE or the risk of treatment discontinuation due to AE. However, the major limitation of the ARROW study was the low dose adopted for the twice-weekly arm, which was chosen by the investigators because it was the 1646

approved dose at the time of trial design. Indeed, higher doses of carfilzomib (up to 36 mg/m2 when given in combination and 70 mg/m2 alone) have been safely delivered both in upfront and relapse settings.5,8,14-20 To our knowledge, this is the first analysis to compare two different schedules and doses of carfilzomib (70 mg/m2 once-weekly vs. 36 mg/m2 twice-weekly) as induction and maintenance therapies for elderly, transplant-ineligible NDMM patients. In the ITT analysis, we observed no significant differences in 3-year PFS (47% vs. 49%), PFS-2 (62% vs. 70%) and OS (70% vs. 72%) in patients receiving onceversus twice-weekly carfilzomib. The risks of dose reduction or treatment discontinuation were equal between the two groups. Of note, delivering 70 mg/m2 of carfilzomib in a single dose did not increase the risk of grade 3-5 hematologic (24% vs. 30%; P=0.82) and non-hematologic (38% vs. 41%; P=0.83) AE, as compared to a twice-weekly administration of 36 mg/m2 of carfilzomib. Importantly, no new cardiovascular safety risks were identified with once-weekly carfilzomib treatment at the 70 mg/m2 dose. The aim of continuous treatment is to prolong PFS and OS among NDMM patients without negatively affecting their quality of life. For this purpose, we compared onceversus twice-weekly maintenance with carfilzomib. Among patients who received carfilzomib maintenance, we did not observe any significant differences in terms of 3-year PFS_m (47% vs. 51%), PFS-2_m (65% vs. 74%), and OS_m (72% vs. 73%) between the once-weekly and twice-weekly schedules. Continuous treatment with single-agent carfilzomib was well tolerated and grade 3-5 AE were infrequent in both groups, although patients in the once-weekly arm were at higher risk of developing grade 3-5 hypertension (9% vs. 0%) as compared to patients in the twice-weekly group. As previously reported, carfilzomib is able to at least partially abrogate the unfavorable prognostic significance of high-risk FISH cytogenetic abnormalities.21 In this trial, we observed no difference between standard and highrisk FISH patients in terms of 3-year PFS (52% vs. 43%) and OS (78% vs. 73%), with a greater reduction in the risk of progression or death for high-risk FISH patients (as compared to standard-risk FISH patients) in the onceweekly (HR: 1.17) than in the twice-weekly (HR: 0.52) carfilzomib schedule. These results compared favorably with those observed in the ARROW trial.9 However, due to the high frequency of patients with unknown cytogenetic risk and the small number of patients, these results should be interpreted with caution and this evidence must be confirmed by further studies. The major limitation of our analysis was the non-randomized design of the two phase I/II studies. Indeed, the study populations were slightly different (e.g. frailty status, FISH data availability), but the most important inclusion and exclusion criteria, as well as the treatment schema, were identical. Furthermore, the reproducibility of the results in the community setting was limited both by the lower percentage of older patients (≥75 years) as compared to other studies (such as the FIRST and the ALCYONE), and by the fact that patients included in this analysis were treated in the context of clinical trials in a limited number of selected, experienced centers. With this limitation, our results should be interpreted with caution, even though they should be considered as the basis of future randomized trials. haematologica | 2019; 104(8)

Once- vs. Twice-Weekly Carfilzomib in NDMM

In conclusion, a once-weekly 70 mg/m2 infusion of carfilzomib was shown to be as safe and effective as a twice weekly 36 mg/m2 infusion for the initial treatment of elderly transplant-ineligible NDMM patients, both as induction therapy in combination with cyclophosphamide and dexamethasone and as single-agent maintenance. This analysis supports the use of high-dose onceweekly carfilzomib and provides the rationale for the investigation of onceversus twice-weekly carfilzomib as initial treatment for MM patients.

Neoplasie Sangue ONLUS (Italy). The IST-CAR-506 (NCT01346787) study was sponsored by the HOVON Foundation and co-sponsored by Fondazione Neoplasie Sangue ONLUS. Both trials were supported by funding from AMGEN (Onyx Pharmaceuticals), which had no role in study design, data collection, data analysis, data interpretation, writing of the report or publication of this article. The corresponding author had full access to all the data in the two studies and had final responsibility for the decision to prepare and submit this manuscript for publication, together with the other authors.

Funding The IST-CAR-561 (NCT01857115) study was sponsored by Stichting Hemato-Oncologie voor Volwassenen Nederland (HOVON, the Netherlands), in collaboration with Fondazione

Acknowledgments We thank the patients who participated in these studies and their families; the study co-investigators, nurses, and coordinators at each of the clinical sites.

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. Leukemia. 2014;28(5):1122-1128. 2. San Miguel JF, Schlag R, Khuageva NK, et al. Persistent Overall Survival Benefit and No Increased Risk of Second Malignancies With Bortezomib-Melphalan-Prednisone Versus Melphalan-Prednisone in Patients With Previously Untreated Multiple Myeloma. J Clin Oncol. 2013;31(4):448455. 3. Bringhen S, Larocca A, Rossi D, et al. Efficacy and safety of once-weekly bortezomib in multiple myeloma patients. Blood. 2010;116(23):4745-4753. 4. Dytfeld D, Jasielec J, Griffith KA, et al. Carfilzomib, lenalidomide, and low-dose dexamethasone in elderly patients with newly diagnosed multiple myeloma. Haematologica. 2014;99(9):e162-164. 5. Facon T, Lee JH, Moreau P, et al. Phase 3 Study (CLARION) of Carfilzomib, Melphalan, Prednisone (KMP) v Bortezomib, Melphalan, Prednisone (VMP) in Newly Diagnosed Multiple Myeloma (NDMM). Clin Lymphoma Myeloma Leuk. 2017;17(1):e26-e27. 6. Bringhen S, Petrucci MT, Larocca A, et al. Carfilzomib, cyclophosphamide, and dexamethasone in patients with newly diagnosed multiple myeloma: a multicenter, phase 2 study. Blood. 2014;124(1):63-69. 7. Bringhen S, D’Agostino M, De Paoli L, et al. Phase 1/2 study of weekly carfilzomib, cyclophosphamide, dexamethasone in newly diagnosed transplant-ineligible myeloma. Leukemia. 2018;32(4):979-985. 8. Berenson JR, Cartmell A, Bessudo A, et al.

haematologica | 2019; 104(8)

10.

11.

12.

13.

14.

15.

CHAMPION-1: a phase 1/2 study of onceweekly carfilzomib and dexamethasone for relapsed or refractory multiple myeloma. Blood. 2016;127(26):3360-3368. Moreau P, Mateos M-V, Berenson JR, et al. Once weekly versus twice weekly carfilzomib dosing in patients with relapsed and refractory multiple myeloma (A.R.R.O.W.): interim analysis results of a randomised, phase 3 study. Lancet Oncol. 2018; 19(7):953-964. National Cancer Institute [USA]. Common Terminology Criteria for Adverse Events (CTCAE) Version. 4.0. US Department of Health and Human Services. May 28, 2009. 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. Palumbo A, Bringhen S, Mateos M-V, et al. Geriatric assessment predicts survival and toxicities in elderly myeloma patients: An International Myeloma Working Group report. Blood. 2015;125(13):2068-2074. Stewart AK, Rajkumar SV, Dimopoulos MA, et al. Carfilzomib, lenalidomide, and dexamethasone for relapsed multiple myeloma. N Engl J Med. 2015;372(2):142-152. Jakubowiak AJ, Dytfeld D, Griffith KA, et al. A phase 1/2 study of carfilzomib in combination with lenalidomide and low-dose dexamethasone as a frontline treatment for multiple myeloma. Blood. 2012; 120(9):1801-1809. Gay F, Rota Scalabrini D, Belotti A, et al. Updated efficacy and MRD data according to risk-status in newly diagnosed myeloma patients treated with carfilzomib plus lenalidomide or cyclosphosphamide: Results from the FORTE trial. HemaSphere 2018;2(S1):6. [Abstract #S109, EHA 2018

23rd Congress]. 16. Gay F, Foà R, Musto P, et al. Updated efficacy data and MRD analysis according to risk status in newly diagnosed myeloma patients treated with carfilzomib + lenalidomide or cyclophosphamide (FORTE trial). J Clin Oncol. 2018;36(15_suppl) [Abstract #8009, ASCO 2018 Annual Meeting]. 17. Dimopoulos MA, Moreau P, Palumbo A, et al. Carfilzomib and dexamethasone versus bortezomib and dexamethasone for patients with relapsed or refractory multiple myeloma (ENDEAVOR): a randomised, phase 3, open-label, multicentre study. Lancet Oncol. 2016;17(1):27-38. 18. Dimopoulos MA, Goldschmidt H, Niesvizky R, et al. Carfilzomib or bortezomib in relapsed or refractory multiple myeloma (ENDEAVOR): an interim overall survival analysis of an open-label, randomised, phase 3 trial. Lancet Oncol. 2017;18(10):1327-1337. 19. Boccia RV, Bessudo A, Agajanian R, et al. A Multicenter, Open-Label, Phase 1b Study of Carfilzomib, Cyclophosphamide, and Dexamethasone in Newly Diagnosed Multiple Myeloma Patients (CHAMPION2). Clin Lymphoma Myeloma Leuk. 2017; 17(7):433-437. 20. Sonneveld P, Asselbergs E, Zweegman S, et al. Phase 2 study of carfilzomib, thalidomide, and dexamethasone as induction/consolidation therapy for newly diagnosed multiple myeloma. Blood. 2015; 125(3):449-456. 21. Chng W-J, Goldschmidt H, Dimopoulos MA, et al. Carfilzomib–dexamethasone vs bortezomib–dexamethasone in relapsed or refractory multiple myeloma by cytogenetic risk in the phase 3 study ENDEAVOR. Leukemia. 2017;31(6):1368-1374.

1647

ARTICLE Ferrata Storti Foundation

Platelet Biology & its DIsorders

Platelet glycoprotein VI and C-type lectin-like receptor 2 deficiency accelerates wound healing by impairing vascular integrity in mice Surasak Wichaiyo,1,2 Sian Lax,1 Samantha J. Montague,1 Zhi Li,3 Beata Grygielska,1 Jeremy A. Pike,4 Elizabeth J. Haining,1 Alexander Brill,1,5 Steve P. Watson,1,4,6 and Julie Rayes1

Institute of Cardiovascular Sciences, College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK; 2Department of Pharmacology, Faculty of Pharmacy, Mahidol University, Bangkok, Thailand; 3Institute of Immunology and Immunotherapy, College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK; 4 Centre of Membrane Proteins and Receptors (COMPARE), Universities of Birmingham and Nottingham, The Midlands, UK; 5Department of Pathophysiology, Sechenov First Moscow State Medical University, Moscow, Russia and 6Institute of Microbiology and Infection, College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK 1

Haematologica 2019 Volume 104(8):1648-1660

ABSTRACT

Correspondence: JULIE RAYES j.rayes@bham.ac.uk STEVE P. WATSON s.p.watson@bham.ac.uk Received: October 12, 2018. Accepted: January 28, 2019. Pre-published: February 7, 2019. doi:10.3324/haematol.2018.208363 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1648 ÂŠ2019 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.

1648

latelets promote wound healing by forming a vascular plug and by secreting growth factors and cytokines. Glycoprotein (GP)VI and Ctype lectin-like receptor (CLEC)-2 signal through a (hem)-immunoreceptor tyrosine-based activation motif, which induces platelet activation. GPVI and CLEC-2 support vascular integrity during inflammation in the skin through regulation of leukocyte migration and function, and by sealing sites of vascular damage. In this study, we investigated the role of impaired vascular integrity due to GPVI and/or CLEC-2 deficiency in wound repair using a full-thickness excisional skin wound model in mice. Transgenic mice deficient in both GPVI and CLEC-2 exhibited accelerated skin wound healing, despite a marked impairment in vascular integrity. The local and temporal bleeding in the skin led to greater plasma protein entry, including fibrinogen and clotting factors, was associated with increased fibrin generation, reduction in wound neutrophils and M1 macrophages, decreased level of tumor necrosis factor (TNF)-Îą, and enhanced angiogenesis at day 3 after injury. Accelerated wound healing was not due to developmental defects in CLEC-2 and GPVI double-deficient mice as similar results were observed in GPVI-deficient mice treated with a podoplanin-blocking antibody. The rate of wound healing was not altered in mice deficient in either GPVI or CLEC-2. Our results show that, contrary to defects in coagulation, bleeding following a loss of vascular integrity caused by platelet CLEC-2 and GPVI deficiency facilitates wound repair by increasing fibrin(ogen) deposition, reducing inflammation, and promoting angiogenesis.

Introduction Cutaneous wound repair is a complex process, which requires a well-regulated interplay between diverse cell types and molecules.1 Four inter-related phases of wound healing have been described, namely hemostasis, inflammation, proliferation, and remodeling.1 Immediately upon injury, platelets and fibrin generate a hemostatic plug to prevent excessive blood loss. Additionally, the fibrin clot forms a scaffold to promote migration of local cells surrounding the wound.2 Deficiency in clotting factors, including tissue factor (TF), factor (F)VII or FIX, results in persistent intra-tissue bleeding and delayed wound healing.3-5 Shortly after, neutrophils and inflammatory macrophages (M1) are recruited to eliminate microbes and cellular debris, driving the inflammatory phase.1,6 During the proliferative phase, re-epithelialization and angiogenesis promote cell growth and wound recovery. In addition, fibroblasts infiltrate the granulation tissue to produce extracellular matrix proteins, and to differentiate into myofibroblasts, mediating haematologica | 2019; 104(8)

Platelet GPVI and CLEC-2 in skin wound healing

wound contraction.1 In parallel, the number of M2 reparative macrophages increases, contributing to resolution of inflammation.1,6 Complete wound closure and re-organization of collagen fibers restore skin integrity during the remodeling phase.1 Platelets play several roles in wound healing during the hemostatic, inflammatory, and vascular repair phases.7 Platelets secrete chemoattractants and growth factors that mediate cell recruitment and tissue repair, respectively.1,7,8 This is illustrated by the delay in healing of corneal epithelial abrasion in thrombocytopenic and P-selectindeficient mice.9 Moreover, platelet-rich plasma, which contains growth factors, promotes skin wound healing in mice10 and is a possible therapeutic agent to facilitate wound repair. Platelet immunoreceptor tyrosine-based activation motif (ITAM) receptors, GPVI and CLEC-2, share a common Src/Syk/PLCÎł2-dependent signaling pathway leading to platelet activation.11 A primary role for GPVI and a secondary role for CLEC-2 in maintaining vascular integrity in the inflamed skin has been demonstrated.12,13 In addition, CLEC-2 and GPVI are key regulators of inflammation. GPVI promotes a pro-inflammatory phenotype during glomerulonephritis,14 arthritis,15 and dermatitis.16 The CLEC-2-podoplanin axis is anti-inflammatory and protects against organ damage during lung and systemic inflammation.17,18 Due to the complex interplay between platelets and inflammation during wound healing, we hypothesize that GPVI and/or CLEC-2 regulate vascular integrity during wound repair and alter the healing process. In the present study, we show that deletion of both CLEC-2 and GPVI accelerates wound healing in a mouse model of fullthickness excisional skin wound repair. This is associated with a transient and self-limited bleeding (i.e. due to impaired vascular integrity), fibrin(ogen) matrix deposition, a reduction in wound neutrophils and M1 macrophages, and increased angiogenesis during the inflammatory phase. Taken together, we show that impaired vascular integrity-induced bleeding is beneficial during a model of sterile wound healing.

Methods Animals Male and female wild-type (WT), platelet-specific CLEC-2deficient (Clec1bfl/flPf4-Cre), GPVI knockout (Gp6-/-), and CLEC2/GPVI double-deficient (Clec1bfl/flPf4-Cre/Gp6-/-; DKO) mice aged 8-10 weeks were used. All experiments were performed in accordance with UK laws (Animal Scientific Procedures Act 1986) with the approval of the local ethics committee and UK Home Office under PPL P0E98D513 and P14D42F37, respectively.

Full-thickness excisional skin wound model A single full-thickness excisional skin wound was made on the shaved flank skin of mice using a 4 mm-diameter biopsy punch (Kai Industries, Japan). Wounds were imaged using a Nikon COOLPIX B500 digital camera each day and wound size measured using calipers19 on a daily basis for up to nine days post injury. Wound area was calculated as described19 and presented as the percentage of initial wound size.10 In a second set of experiments, anti-podoplanin antibody (clone 8.1.1, 100 Îźg)13,17,20 or IgG isotype control were intravenously injected 24 haematologica | 2019; 104(8)

hours (h) before and again 24 h after wounding. Wound size was monitored for three days post injury.

Other associated methods For details of other associated methods see the Online Supplementary Appendix.

Results CLEC-2 and GPVI ligands are present within perivascular areas during skin wounding In unchallenged WT mouse skin, collagen was abundantly expressed throughout the dermis and hypodermis, including around blood vessels, and podoplanin was predominantly expressed on lymphatic endothelial cells and on perivascular cells (Online Supplementary Figure S1A and B). At day 3 post wounding, podoplanin expression was also observed surrounding blood vessels in WT and transgenic mice (Figure 1A). These podoplanin-expressing cells included pericytes (neuron-glial antigen 2; NG2+) (Figure 1A), fibroblasts (vimentin+), infiltrating monocytes (Ly6C+), and macrophages (F4/80+) (Figure 1B). In addition, podoplanin was up-regulated on migrating keratinocytes, and on stromal and infiltrating cells within the granulation tissue (Online Supplementary Figure S1C). The level of perivascular podoplanin was increased in Clec1bfl/flPf4-cre mice compared to WT mice (Figure 1A and B). In DKO mice, podoplanin expression was similar to that observed in WT and Gp6-/- mice (Figure 1A and B). Three days after wound injury, platelets (CD41+) were observed in close proximity to the vessel wall (surrounded by pericytes) in WT, Clec1bfl/flPf4-cre, and Gp6-/- mice (Figure 1A, arrow). In DKO mice, platelets were more widely distributed, including at the vessel wall and in the surrounding tissue (Figure 1A, star). The platelet count in Clec1bfl/flPf4-cre and DKO mice was approximately 25% and 40% lower than in WT, respectively (Online Supplementary Figure S1D) and was not altered by wound injury. These results demonstrate that platelets are present in perivascular areas during the initial wound healing process in mouse skin where the ligands for GPVI and CLEC-2 are also observed. Deletion of CLEC-2 from platelets is associated with increased podoplanin-expressing cells in the perivascular space during wound healing while concurrent ablation with GPVI reverses this phenotype.

Cutaneous wound healing is accelerated in mice deficient in platelet CLEC-2 and GPVI To determine the contribution of platelet ITAM receptors in wound repair, we monitored the time course of wound closure in WT and platelet ITAM receptor(s)-deficient mice. All mouse strains exhibited complete wound closure within nine days post injury (Figure 2A). However, mice deficient in both GPVI and CLEC-2 demonstrated accelerated repair (Figure 2B), characterized by the presence of a dark scab and redness around the wound (Figure 2A), especially within the first three days after injury compared to WT and single ITAM-deficient mice. There was no discernible difference in wound appearance in any of the mouse strains after four days post injury (Figure 2A) and there was no difference in macroscopic wound size at day 9 post injury (Figure 2B). 1649

S. Wichaiyo et al.

However, the morphometric analysis of skin histology at this time point revealed a significantly smaller scar in DKO mice, as shown by a shorter length of hyperplastic epidermis compared to WT and Clec1bfl/flPf4-cre mice (Figure 2C and D) and a narrower inter-subcutaneous gap than WT mice (Figure 2C and E). At day 3 post injury, re-epithelialization was observed in all groups (Figure 3A). However, this process was enhanced in DKO mice compared to WT and Clec1bfl/flPf4cre mice, but not Gp6-/- mice (Figure 3B). There was no sig-

nificant difference in wound contraction between all tested groups (Figure 3C). DKO mice also had a larger area of granulation tissue compared to WT mice (Figure 3A and D). Improved wound healing was associated with enhanced angiogenesis as assessed by the increase in CD31+ area in DKO animals at day 3 post injury (Figure 3E and F), although the density of blood vessels (CD31+) and lymphatic vessels (podoplanin+) in unchallenged skin was similar among all groups (Online Supplementary Figure S1E and F).

Figure 1. Podoplanin-expressing cells are present at perivascular area in contact with platelets during skin wound repair. (A) Immunofluorescence staining of NG2 (red), podoplanin (green) and CD41 (white) illustrates platelets and podoplanin-expressing pericytes (NG2+) around blood vessel at day 3 after injury (n=5-7). Hoechst counterstains nuclei (blue). Arrow points to platelets at perivascular site. Star indicates extravascular localization of platelets. Scale bar=20 Îźm. (B) Podoplanin (green) was double-stained with either vimentin (red; top) or Ly6C (red; middle) or F4/80 (red; bottom), which are located around blood vessel (surrounded by NG2+ pericytes) at day 3 after injury (n=4-5). Scale bar=20 Îźm. BV: blood vessel.

1650

haematologica | 2019; 104(8)

Platelet GPVI and CLEC-2 in skin wound healing

These results demonstrate that deletion of both CLEC2 and GPVI accelerates skin wound closure, enhances reepithelialization and angiogenesis, and reduces scar formation.

GPVI and CLEC-2 maintain vascular integrity during the inflammatory phase of wound repair The redness surrounding the wound in DKO mice in the inflammatory phase is indicative of increased vascular leakiness. Macroscopic examination of the skin at day 3 post injury demonstrated vasodilation and bleeding into the wound as well as into the surrounding skin in DKO mice, with less severe vascular leakage seen in Gp6-/- mice (Figure 4A). Hematoxylin & Eosin (H&E) staining confirmed the extravasation of red blood cells (RBC) in the dermis at the edge of the wound at day 3 post injury in

both DKO and Gp6-/- mice (Online Supplementary Figure S2A). Clearance of extravascular RBC was observed at day 9 post injury in all groups (Online Supplementary Figure S2B). Due to blood/lymphatic mixing phenotype in DKO, and to a lesser degree in Clec1bfl/flPf4-cre mice, RBC were also present in lymphatic vessels (podoplanin+) (Online Supplementary Figure S2C). These observations demonstrate marked impairment of vascular integrity during the inflammatory phase of wound repair in DKO mice, with a milder phenotype in Gp6-/- mice. Vascular leakage is diminished in later phases when inflammation subsides.

GPVI and CLEC-2 deficiency increases fibrin(ogen) deposition during the inflammatory phase of repair Increased vascular permeability results in leakage of

Figure 2. Deletion of platelet immunoreceptor tyrosine-based activation motif (ITAM) receptors accelerates skin wound repair process. Mice were subjected to a full-thickness excisional skin wound and wound closure was monitored for nine days after injury. (A) Macroscopic appearance of wound at indicated time points. (B) Changes of wound size over nine days post injury (n=10-13). (C) Hematoxylin & Eosin staining at day 9 post-injury (n=9-13). a: length of hyperplastic epidermis; b: inter-subcutaneous distance. Scale bar=200 μm. (D) Measurement of the length of hyperplastic epidermis. (E) Measurement of inter-subcutaneous distance. All graphs are presented as mean±Standard Error of Mean (SEM). Kinetics of wound closure (B) are analyzed by two-way ANOVA with Bonferroni’s multiple comparison -/test. *P<0.05; **P<0.01. *WT versus DKO. +Clec1bfl/flPf4-cre versus DKO. #Gp6 versus DKO. Other parameters are analyzed by one-way ANOVA with Bonferroni’s multiple comparison test. *P<0.05.

haematologica | 2019; 104(8)

1651

S. Wichaiyo et al. blood cells and plasma proteins into the wound21 where they come in contact with TF,5,22 which activates the extrinsic pathway of blood coagulation. An increase in fibrinogen deposition was particularly marked in the granulation tissue at day 3 post injury (Figure 4B and C) with no significant alteration in TF expression in DKO mice (Online Supplementary Figure S3A and B). Fibrin con-

tent in the wound scab was similar between WT and DKO mice at day 1 post injury (Online Supplementary Figure S3C and D), but was significantly increased in the DKO mice at day 3 post injury, compared to WT and Clec1bfl/flPf4-cre mice but not to Gp6-/- mice, which exhibited a more moderate vascular leakage (Figure 4D and E). At day 9 post injury, fibrin was mainly located on the

E F

Figure 3. Enhanced re-epithelialization and angiogenesis occur at the early phase of wound healing in the absence of GPVI and CLEC-2. (A) Hematoxylin & Eosin staining at day 3 post-injury (n=6-9). Dotted line indicates hyperplastic coverages. Black arrow points to wound edge. Red arrow indicates gap between epithelial tongues. S: scab; G: granulation tissue. Scale bar=500 μm. (B) Measurement of re-epithelialization. (C) Measurement of wound contraction. (D) Quantification of granulation tissue area. (E) Detection of endothelial cells (CD31+ cells; green) in wound area at day 3 post injury. Hoechst counterstains nuclei (blue). Scale bar=50 μm. (F) Quantification of CD31+ area within the wound at day 3 post injury (n=5-6). Graphs are presented as mean±Standard Error of Mean and analyzed by oneway ANOVA with Bonferroni’s multiple comparison test. *P<0.05.

1652

haematologica | 2019; 104(8)

Platelet GPVI and CLEC-2 in skin wound healing

upper part of the scar in all groups (Figure 4F) and was notably lower in the DKO mice compared to WT and Gp6-/- mice (Figure 4G). There was no significant difference in wound (myo)fibroblasts (Online Supplementary Figure S3E-G) and collagen content (Online Supplementary Figure S3H and I) between DKO mice and WT at day 3

and day 9 post injury. These results indicate that the loss of vascular integrity during the inflammatory phase increases extravasation of plasma proteins, including fibrinogen and clotting factors, into the wound of DKO mice. The increase in fibrinogen and fibrin deposition is associated with accelerated

B C

F G

Figure 4. Lack of platelet immunoreceptor tyrosine-based activation motif (ITAM) receptors causes local and temporal bleeding leading to fibrin(ogen) deposition during inflammatory phase of wound repair. (A) Macroscopic images of inner side of skin wound at day 3 post injury (n=4-6). Dotted circle indicates wound area. Arrow points to dilated vessel. Arrowhead shows bleeding into surrounding skin. (B) Fibrinogen staining (brown) of skin wound at day 3 post injury. (C) Quantification of fibrinogen content at day 3 post injury (n=6). (D) Martius scarlet blue (MSB) staining of skin wound at day 3 post-injury. Red: old fibrin; blue: collagen; yellow: red blood cells/fresh fibrin. (E) Quantification of fibrin content in the wound at day 3 post injury (n=6-9). (F) MSB staining of wound scar at day 9 post injury. (G) Quantification of fibrin content in the scar at day 9 post injury (n=9-13). Graphs are presented as mean±Standard Error of Mean and analyzed by one-way ANOVA with Bonferroni’s multiple comparison test. *P<0.05; **P<0.01. Scale bar=200 μm.

haematologica | 2019; 104(8)

1653

S. Wichaiyo et al.

wound healing and is then cleared from the healing wound.

Platelet immunoreceptor tyrosine-based activation motif receptor deficiency reduces wound neutrophils and M1 macrophages during the inflammatory phase The influx of inflammatory cells was investigated at days 1, 3, and 9 post injury. Neutrophil (Gr-1+) infiltration on day 1 post injury was similar between WT and DKO mice (Online Supplementary Figure S4A and B). At day 3 post injury, DKO mice showed a significant impairment in neutrophil infiltration compared to WT and Clec1bfl/flPf4-cre mice but not Gp6-/- mice (Figure 5A and B). The decrease in neutrophil infiltration was confirmed using anti-Ly6G antibody clone 1A8 (data not shown). A 2fold increase in wound neutrophils was observed in WT but not in DKO mice at this time relative to day 1 post

injury (Figure 5C). At day 9 post injury, DKO mice showed higher numbers of wound neutrophils than WT (Figure 5D and E), although neutrophil level was significantly decreased in both groups relative to day 1 and day 3 post injury (Figure 5C). Blood neutrophil counts were similar in unchallenged mice across all groups (Figure 5F). At day 3 post injury, the number of blood neutrophils was significantly decreased in WT and Clec1bfl/flPf4-cre (Figure 5F) while remaining unaltered during the time course of wound healing in DKO and Gp6-/- mice (Figure 5F). The decrease in neutrophil infiltration was not due to a defect of chemoattractants at the wound site as measured by the presence of chemokine CXCL-1 (Online Supplementary Figure S6A and B) or platelet factor 4 (PF4) (Online Supplementary Figure S6C and D). In vitro migration towards N-formyl-methionyl-leucyl-phenylalanine (fMLP) using bone marrow-derived neutrophils demonstrated that migration of neutrophils from WT mice was

Figure 5. Neutrophil influx is decreased during the inflammatory phase of wound healing following platelet CLEC-2 and GPVI double-deletion. (A) Detection of neutrophils (Gr-1 staining; brown) in wound at day 3 post injury. (B) Quantification of neutrophils (Gr-1+ cells) in wound at day 3 post injury. *P<0.05; **P<0.01. (C) Comparison of Gr1+ cells between day 1, day 3, and day 9 post injury in wild-type (WT) and DKO mice. P<0.05 in *WT and §DKO mice, compared to the data at day 1 post injury, respectively. The bracket shows P<0.05 for the comparison between day 3 and day 9 post injury in *WT and §DKO mice, respectively. (D) Detection of neutrophils (Gr-1 staining; brown) in wound at day 9 post injury. (E) Quantification of neutrophils (Gr-1+ cells) in wound at day 9 post injury. *P<0.05. (F) Comparison of blood neutrophil counts between baseline, day 3, and day 9 post injury in each mouse strain. P<0.05 in *WT and +Clec1bfl/flPf4-cre mice, compared to their control, respectively. Sample numbers in unchallenged control = 10, day 1 = 5, day 3 = 6-9, and day 9 post injury = 9-13, respectively. Graphs are presented as mean±Standard Error of Mean and analyzed by one-way ANOVA with Bonferroni’s multiple comparison test. Scale bar = 20 μm.

1654

haematologica | 2019; 104(8)

Platelet GPVI and CLEC-2 in skin wound healing

attenuated by fibrinogen, and strongly inhibited by crosslinked fibrin compared to the migration through collagen (Online Supplementary Figure S7). Fibrinogen and fibrin showed a similar degree of inhibition for the migration of neutrophils from DKO mice (Online Supplementary Figure S7). There was no significant difference in wound monocytes (Ly6C+) at day 1 post injury between DKO mice and WT (Online Supplementary Figure S4C and D). At day 3 post injury, DKO mice showed a significantly higher number of wound monocytes compared to all other strains (Figure 6A and B), which was reflected by a 4-fold increase in wound monocytes relative to day 1 post injury (Figure 6C). At day 9 post injury, monocytes within the wound remained at a low level in WT and single-knockout mice (Figure 6D and E). Wound monocytes in DKO mice were decreased at this time compared to day 3 post injury (Figure 6C), but remained higher than other groups

(Figure 6D and E). In blood, Clec1bfl/flPf4-cre exhibited an increase in baseline circulating monocytes compared to WT and DKO mice (Online Supplementary Figure S4G, left). Blood monocytes were greatly reduced at day 3 and day 9 post injury in all groups compared to their unchallenged controls (Figure 6F). However, the level of blood monocytes in DKO mice was significantly higher than Clec1bfl/flPf4-cre mice at day 3 post injury (Online Supplementary Figure S4G, middle) and then all groups at day 9 post injury (Online Supplementary Figure S4G, right). The influx of macrophages (F4/80+) in DKO mice was reduced at day 1 post injury compared to WT (Online Supplementary Figure S4E and F). At day 3 post injury, wound macrophages in Clec1bfl/flPf4-cre mice were elevated, compared to all other strains (Figure 7A and B). A 6fold increase in wound macrophages was observed in WT mice at day 3 relative to day 1 post injury (Figure 7C). DKO mice showed a significant reduction in

Figure 6. A higher number of wound monocytes is observed during the inflammatory phase of repair in mice that lack both GPVI and CLEC-2. (A) Detection of monocytes (Ly6C+ cells; brown) in wound at day 3 post-injury. (B) Quantification of Ly6C+ cells in wound at day 3 post-injury (n=5-7). **P<0.01. (C) Comparison of Ly6C+ cells between day 1, day 3, and day 9 post-injury in WT and DKO mice. The symbols * and § indicate P<0.05 in WT and DKO mice, compared to the data at day 1 post injury, respectively. The bracket shows P<0.05 for the comparison between day 3 and day 9 post injury in §DKO mice. (D) Detection of Ly6C+ cells (brown) in wound at day 9 post injury. (E) Quantification of Ly6C+ cells in wound at day 9 post injury (n=6). *P<0.05; **P<0.01. (F) Comparison of blood monocyte counts between baseline, day 3, and day 9 post injury in each mouse strain. The symbols *, +, #, and § indicate P<0.05 in WT, Clec1bfl/flPf4-cre, Gp6-/-, and DKO mice, compared to their control, respectively. Sample numbers in unchallenged control, n=10; day 1, n=5; day 3, n=6-9; day 9 post injury, n=10-13, respectively. Graphs are presented as mean±Standard Error of Mean and analyzed by one-way ANOVA with Bonferroni’s multiple comparison test. Scale bar = 20 μm.

haematologica | 2019; 104(8)

1655

S. Wichaiyo et al. macrophages compared to WT but not Gp6-/- mice at this time (Figure 7B and C). Inducible nitric oxide synthase (iNOS)-expressing macrophages (M1 phenotype) (Figure 7D and E) and TNF-α level (Figure 7G and H) within the wound of DKO mice were decreased whereas Fizz-1positive macrophages (a M2 marker) were similar to WT

(Online Supplementary Figure S5D and E) at day 3 post injury. At day 9 post injury, wound macrophages in DKO mice were similar to WT and Clec1bfl/flPf4-cre mice (Online Supplementary Figure S5D and E) with no difference in M1 (Figure 7D and F) and M2 macrophages (Online Supplementary Figure S5A and C) between WT and DKO

Figure 7. M1 macrophages and TNF-α level are reduced during the inflammatory phase of wound healing in platelet immunoreceptor tyrosine-based activation motif (ITAM) receptor-deficient mice. (A) Detection of macrophages (F4/80+ cells; brown) in wound at day 3 post injury. (B) Quantification of F4/80+ cells in wound at day 3 post injury (n=6-8). *P<0.05; **P<0.01. (C) Comparison of F4/80+ cells between day 1, day 3, and day 9 post injury in wild-type (WT) and DKO mice. The symbols * and § indicate P<0.05 in WT and DKO mice, compared to the data at day 1 post injury, respectively. The bracket shows P<0.05 for the comparison between day 3 and day 9 post injury in §DKO mice. Sample numbers at days post injury: at day 1, n=5; day 3, n=6-9; day 9, n=10-13, respectively. (D) Immunofluorescence double staining of iNOS (red) and F4/80 (green) in the wound of WT and DKO mice at day 3 (n=4) and day 9 (n=4) post injury. Hoechst counterstains nuclei (blue). (E) Quantification of M1 macrophages (iNOS+F4/80+ cells; yellow) at day 3 post injury (n=4). *P<0.05. (F) Quantification of M1 macrophages (iNOS+F4/80+ cells; yellow) at day 9 post injury (n=4). (G) Immunohistochemistry staining of TNF-α (brown) in the wound at day 3 post injury. (H) Quantification of TNF-α level in granulation tissue area at day 3 post injury (n=6). *P<0.05. Graphs are presented as mean±Standard Error of Mean and analyzed by either Student t-test (E, F) or one-way ANOVA with Bonferroni’s multiple comparison test (B, C, and H). Scale bar = 20 μm.

1656

haematologica | 2019; 104(8)

Platelet GPVI and CLEC-2 in skin wound healing

strains. Gp6-/- mice showed a lower number of macrophages within the wound scar than in Clec1bfl/flPf4cre mice at day 9 post injury (Online Supplementary Figure S5E). Overall, these data illustrate that CLEC-2 deletion promotes leukocyte sequestration in the wound during the inflammatory phase, especially macrophages. Deletion of both ITAM receptors leads to a significant reduction in wound neutrophils and M1 macrophages during this phase and a decrease in TNF-α expression in the tissue.

Inhibition of CLEC-2-podoplanin axis accelerates wound healing in GPVI-deficient mice Clec1bfl/flPf4-cre and DKO mice present with blood/lymphatic mixing and moderate thrombocytopenia.11,17 To investigate the influence of these defects on wound healing, we injected Gp6-/- mice with an antibody to podoplanin (anti-podoplanin-treated Gp6-/- mice) that blocks CLEC-2-podoplanin interaction. A significant acceleration in wound healing was observed in antipodoplanin-treated Gp6-/- mice at day 2 and day 3 post

Figure 8. Anti-podoplanin antibody injection in Gp6-/- mice (Gp6-/- + anti-PDPN) simulates the accelerated phenotype of skin wound repair observed in DKO mice. (A) Macroscopic appearance of wound at indicated time points. Arrow points to intra-skin bleeding around the wound at day 3 post injury. (B) Changes of wound size over 3 days post injury (n=5). (C) Hematoxylin & Eosin staining at day 3 post-injury (n=5). Arrow points the bleeding into surrounding skin. Scale bar = 20 μm. (D) Martius scarlet blue staining of skin wound at day 3 post injury. Red: old fibrin, blue: collagen, yellow: red blood cells/fresh fibrin. Scale bar = 200 μm. (E) Quantification of fibrin content (red) in the wound at day 3 post injury (n=5). (F) Staining of neutrophils (Gr-1; brown) in wound area at day 3 post injury. (G) Quantification of neutrophils (Gr-1+ cells) in wound area at day 3 post injury (n=5). (H) Detection of macrophages (F4/80 staining; brown) in wound area at day 3 post injury. (I) Quantification of macrophages (F4/80+ cells) in wound area at day 3 post-injury (n=5). All graphs are presented as mean±Standard Error of Mean. Kinetics of wound closure (B) are analyzed by two-way ANOVA with Bonferroni’s multiple comparison test. Other parameters are analyzed by Student t-test. *P<0.05.

haematologica | 2019; 104(8)

1657

S. Wichaiyo et al.

injury (Figure 8A and B), in association with enhanced vascular leakage (Figure 8A and C) and increased fibrin deposition (Figure 8D and E). Moreover, decreased wound neutrophils (Figure 8F and G) and macrophages (Figure 8H and I), and higher wound monocytes (Online Supplementary Figure S8A and B) but unaltered level of TNF-Îą (Online Supplementary Figure S8C and D) were observed in anti-podoplanin-treated Gp6-/- mice. The reepithelialization (Online Supplementary Figure S8E and F), granulation tissue formation (Online Supplementary Figure S8E and H), and angiogenesis (Online Supplementary Figure S8I and J) were also increased in anti-podoplanin-treated Gp6-/- mice. However, no significant change was seen in wound contraction (Online Supplementary Figure S8E and G). Similar to DKO mice, extravasation of platelets was detected together with the presence of anti-podoplanin antibody on pericytes and other perivascular cells in antipodoplanin-treated Gp6-/- mice at this time (Online Supplementary Figure S9). These data indicate that the accelerated wound healing in DKO mice is not due to developmental defects or thrombocytopenia, but due to a combined loss of the interaction of platelet CLEC-2 and GPVI with their respective ligands.

Discussion In this study, we show that combined deletion of platelet CLEC-2 and GPVI promotes healing of a fullthickness skin wound in mice compared to WT or singleknockout mice. The accelerated wound closure is accompanied by impairment of vascular integrity, rapid reepithelialization and an increase in granulation tissue formation, resulting in a smaller wound scar. This is also associated with elevated levels of fibrinogen and fibrin in the tissue, reduced infiltration of leukocytes, and enhanced angiogenesis. A proposed model for the multifactorial regulation of accelerated wound healing in the absence of the two ITAM receptors is shown in Online Supplementary Figure S10. Wound healing is a multistep process involving coagulation, vascular permeability changes, inflammation, cell proliferation, and cell migration.1,2,21 In our study, we show that deletion of CLEC-2 and GPVI in platelets leads to increased bleeding into the wound, in association with accelerated wound repair. These effects are not due to blood/lymphatic vessel mixing or the reduction in platelet count in the DKO mice as similar results were observed in anti-podoplanin-antibody treated Gp6-/- mice, which do not have these defects. Acceleration of wound healing is also observed in mice treated with histamine23 or serum fraction of the natural latex from rubber tree,21 which enhances vascular permeability, a milder form of vascular leakage. Therefore, impaired vascular integrity as seen in the DKO platelets or increased permeability leads to extravasation of growth factors, cytokines, and plasma proteins to the tissue, and promotion of wound healing. The role of blood coagulation in wound repair has been studied in mice expressing low TF5 or lacking FIX (hemophilia B mice).3,5 In these two models, the deficiency in clotting factors leads to a persistent hematoma formation (due to a prolonged subcutaneous bleeding) and a reduction in fibrin generation, both of which contribute to a delay in wound repair. Chronic hemorrhage also con1658

tributes to the characteristics of non-healing wounds in cancers, which promote tumor growth.24-26 In contrast, we show that in the presence of an intact coagulation cascade, the transient and self-limited bleeding into the wound (which is rich in TF5,22) caused by the impairment of vascular integrity at an initial stage, is associated with accelerated wound repair. This is most likely due to increased entry of platelets, clotting factors, and plasma proteins, leading to fibrinogen accumulation and fibrin generation. Thus, the mechanism of bleeding determines whether it is beneficial or detrimental to wound repair. Fibrinogen and fibrin are not only crucial for clot formation, but also act as a natural suture/sealant, providing a matrix for cell migration as well as a reservoir for growth factors and cytokines.2 For example, fibrin exposes plasminogen to migrating keratinocytes, which in turn convert plasminogen into plasmin to mediate fibrinolysis, allowing the cells to move along the fibrin(ogen) matrix.2729 Similarly, endothelial cells proliferate and migrate over fibrin(ogen), forming a capillary tube that contributes to angiogenesis during wound healing.30-32 In DKO mice, we therefore propose that fibrin(ogen) accumulation promotes re-epithelialization and angiogenesis, and accelerates wound repair. In contrast, fibrinogen33 and a high concentration of fibrin34 inhibit neutrophil migration as previously reported and in line with our in vitro data. These data may explain our observations that in DKO mice, wound neutrophils at day 3 did not increase over the amount seen at day 1 post injury. This is in contrast to wound neutrophil accumulation in WT, which was higher at day 3 than at day 1 post injury. Increased fibrin content in the wound was also observed in DKO mice compared to WT at day 3 post injury. Therefore, the physical obstruction and anti-migratory properties of the fibrin clot is likely to inhibit neutrophil wound entry observed in DKO mice. Moreover, given that, at this early phase, there was no significant difference in the level of blood neutrophils between WT and DKO mice, altered neutrophil turnover/apoptosis in DKO mice is unlikely to make a significant contribution to the reduction in wound neutrophils observed in DKO compared to WT at day 3 post injury. Consistent with our data, several lines of evidence demonstrate improved wound healing in neutrophildepleted conditions.35-38 Indeed, neutrophil infiltration mediates damage through a production of proteases,35 oxidative radicals, elastase,39 and neutrophil extracellular traps,37 all of which can delay the healing process. A similar effect of fibrin(ogen) might also apply to monocyte/macrophage recruitment since fibrinogen has an anti-adhesive effect against monocytes.40 A recent sterile wound model has demonstrated that Ly6C+ monocytes are present throughout wound healing and differentiate into M1 macrophages during inflammatory phase and M2 macrophages during the reparative phase.41 In addition, it has been shown that TNF-Îą secretion is increased during monocyte-to-M1-macrophage differentiation.41,42 The reduction in TNF-Îą during the inflammatory phase is associated with the increase in wound monocytes but not macrophages. This suggests an inhibition in monocyte-to-M1-macrophage transition in DKO mice, resulting in a significant change in immune cell infiltration in the wound. Depletion of macrophages in the first five days has previously been shown to delay the later stages of wound haematologica | 2019; 104(8)

Platelet GPVI and CLEC-2 in skin wound healing

closure, but not during the inflammatory phase (day 13).43 This is due to a decrease in M2 macrophages,43 supporting our observation that a reduction in wound macrophages, particularly M1 phenotype, in the early phase does not negatively affect wound closure but may contribute to the reduction in scar formation.43 The alteration in M2 macrophages was not observed in DKO mice although the previous dermatitis model has reported an increase number of M2 phenotypes in GPVI-deficient mice,16 suggesting other contributing factors for macrophage polarization during skin wound healing.6 The increased risk of wound contamination is a concern in the context of intra-tissue bleeding and reduced wound leukocytes. However, it has recently been shown that rapid formation of a fibrin film over the surface of the wound is protective against bacterial infection.44 This process might also reduce the need for leukocyte infiltration to kill microbes. Moreover, a previous study has reported that a 2-fold increase in wound neutrophils is driven by Staphylococcus aureus infection.45 Whether the beneficial potential of targeting GPVI and CLEC-2 might modulate the risk of wound infection requires further investigation. In conclusion, we show that deletion of platelet GPVI

References 1. Shaw TJ, Martin P. Wound repair at a glance. J Cell Sci. 2009;122(Pt 18):32093213. 2. Drew AF, Liu H, Davidson JM, Daugherty CC, Degen JL. Wound-healing defects in mice lacking fibrinogen. Blood. 2001;97(12):3691-3698. 3. Hoffman M, Harger A, Lenkowski A, Hedner U, Roberts HR, Monroe DM. Cutaneous wound healing is impaired in hemophilia B. Blood. 2006;108(9):30533060. 4. Xu Z, Xu H, Ploplis VA, Castellino FJ. Factor VII deficiency impairs cutaneous wound healing in mice. Mol Med. 2010; 16(5-6):167-176. 5. Monroe DM, Mackman N, Hoffman M. Wound healing in hemophilia B mice and low tissue factor mice. Thromb Res. 2010;125 Suppl 1:S74-77. 6. Sindrilaru A, Scharffetter-Kochanek K. Disclosure of the Culprits: MacrophagesVersatile Regulators of Wound Healing. Adv Wound Care (New Rochelle). 2013; 2(7):357-368. 7. Golebiewska EM, Poole AW. Platelet secretion: From haemostasis to wound healing and beyond. Blood Rev. 2015;29(3):153162. 8. Brill A, Elinav H, Varon D. Differential role of platelet granular mediators in angiogenesis. Cardiovasc Res. 2004;63(2):226-235. 9. Li Z, Rumbaut RE, Burns AR, Smith CW. Platelet response to corneal abrasion is necessary for acute inflammation and efficient re-epithelialization. Invest Ophthalmol Vis Sci. 2006;47(11):4794-4802. 10. Yang HS, Shin J, Bhang SH, et al. Enhanced skin wound healing by a sustained release of growth factors contained in platelet-rich plasma. Exp Mol Med. 2011;43(11):622629. 11. Bender M, May F, Lorenz V, et al.

haematologica | 2019; 104(8)

12.

13.

14.

15.

16.

17.

18.

19.

and CLEC-2 facilitates cutaneous wound repair through a local and temporal vascular leakage leading to increased fibrin(ogen) deposition and reduced leukocyte infiltration. Thus, impaired vascular integrity due to the loss of GPVI and CLEC-2 is beneficial to wound repair. This contrasts with results in coagulation-deficient mice, with differences explained by altered formation of fibrin and most likely alteration in immune cell trafficking. A shorter duration of healing lowers the risk of complications (e.g. infection) and the cost of caring for the wound.46 Based on our study, targeting CLEC-2 and GPVI at the wound site together with optimal wound care (e.g. aseptic dressing) might represent a new pathway to promote healing and reduce scar formation. Acknowledgments The authors would like to thank the BMSU at the University of Birmingham for technical support in animal experiments and the Technology Hub for imaging assistance. Funding This work was supported by the Ministry of Sciences and Technology of Thailand and the British Heart Foundation (RG/13/18/30563). SPW holds a BHF Chair (CH03/003).

Combined in vivo depletion of glycoprotein VI and C-type lectin-like receptor 2 severely compromises hemostasis and abrogates arterial thrombosis in mice. Arterioscler Thromb Vasc Biol. 2013;33(5):926-934. Gros A, Syvannarath V, Lamrani L, et al. Single platelets seal neutrophil-induced vascular breaches via GPVI during immune-complex-mediated inflammation in mice. Blood. 2015;126(8):1017-1026. Rayes J, Jadoui S, Lax S, et al. The contribution of platelet glycoprotein receptors to inflammatory bleeding prevention is stimulus and organ dependent. Haematologica. 2018;103(6):e256-e258. Devi S, Kuligowski MP, Kwan RY, et al. Platelet recruitment to the inflamed glomerulus occurs via an alphaIIbbeta3/GPVI-dependent pathway. Am J Pathol. 2010;177(3):1131-1142. Boilard E, Nigrovic PA, Larabee K, et al. Platelets amplify inflammation in arthritis via collagen-dependent microparticle production. Science. 2010;327(5965):580-583. Pierre S, Linke B, Suo J, et al. GPVI and Thromboxane Receptor on Platelets Promote Proinflammatory Macrophage Phenotypes during Cutaneous Inflammation. J Invest Dermatol. 2017;137 (3):686-695. Rayes J, Lax S, Wichaiyo S, et al. The podoplanin-CLEC-2 axis inhibits inflammation in sepsis. Nat Commun. 2017;8(1):2239. Lax S, Rayes J, Wichaiyo S, et al. Platelet CLEC-2 protects against lung injury via effects of its ligand podoplanin on inflammatory alveolar macrophages in the mouse. Am J Physiol Lung Cell Mol Physiol. 2017;313(6):L1016-L1029. Moreira CF, Cassini-Vieira P, da Silva MF, Barcelos LS. Skin Wound Healing Model Excisional Wounding and Assessment of Lesion Area. Bio-protocol. 2015;5(22): e1661.

20. Payne H, Ponomaryov T, Watson SP, Brill A. Mice with a deficiency in CLEC-2 are protected against deep vein thrombosis. Blood. 2017;129(14):2013-2020. 21. Mendonca RJ, Mauricio VB, Teixeira Lde B, Lachat JJ, Coutinho-Netto J. Increased vascular permeability, angiogenesis and wound healing induced by the serum of natural latex of the rubber tree Hevea brasiliensis. Phytother Res. 2010;24(5):764768. 22. Chen J, Kasper M, Heck T, et al. Tissue factor as a link between wounding and tissue repair. Diabetes. 2005;54(7):2143-2154. 23. Numata Y, Terui T, Okuyama R, et al. The accelerating effect of histamine on the cutaneous wound-healing process through the action of basic fibroblast growth factor. J Invest Dermatol. 2006;126(6):1403-1409. 24. McDonald DM, Baluk P. Significance of blood vessel leakiness in cancer. Cancer Res. 2002;62(18):5381-5385. 25. Dvorak HF. Tumors: wounds that do not heal-redux. Cancer Immunol Res. 2015;3 (1):1-11. 26. Yin T, He S, Liu X, et al. Extravascular red blood cells and hemoglobin promote tumor growth and therapeutic resistance as endogenous danger signals. J Immunol. 2015;194(1):429-437. 27. Kubo M, Van de Water L, Plantefaber LC, et al. Fibrinogen and fibrin are anti-adhesive for keratinocytes: a mechanism for fibrin eschar slough during wound repair. J Invest Dermatol. 2001;117(6):1369-1381. 28. Ronfard V, Barrandon Y. Migration of keratinocytes through tunnels of digested fibrin. Proc Natl Acad Sci U S A. 2001;98(8):4504-4509. 29. Geer DJ, Andreadis ST. A novel role of fibrin in epidermal healing: plasminogenmediated migration and selective detachment of differentiated keratinocytes. J Invest Dermatol. 2003;121(5):1210-1216. 30. Chalupowicz DG, Chowdhury ZA, Bach TL, Barsigian C, Martinez J. Fibrin II

1659

S. Wichaiyo et al.

31. 32.

33.

34.

35. 36.

1660

induces endothelial cell capillary tube formation. J Cell Biol. 1995;130(1):207-215. Laurens N, Koolwijk P, de Maat MP. Fibrin structure and wound healing. J Thromb Haemost. 2006;4(5):932-939. Sahni A, Francis CW. Vascular endothelial growth factor binds to fibrinogen and fibrin and stimulates endothelial cell proliferation. Blood. 2000;96(12):3772-3778. Higazi AA, Barghouti, II, Ayesh SK, Mayer M, Matzner Y. Inhibition of neutrophil activation by fibrinogen. Inflammation. 1994;18(5):525-535. Hanson AJ, Quinn MT. Effect of fibrin sealant composition on human neutrophil chemotaxis. J Biomed Mater Res. 2002;61 (3):474-481. Dovi JV, He LK, DiPietro LA. Accelerated wound closure in neutrophil-depleted mice. J Leukoc Biol. 2003;73(4):448-455. Mori R, Kondo T, Ohshima T, Ishida Y, Mukaida N. Accelerated wound healing in tumor necrosis factor receptor p55-defi-

37.

38.

39.

40.

41.

cient mice with reduced leukocyte infiltration. FASEB J. 2002;16(9):963-974. Wong SL, Demers M, Martinod K, et al. Diabetes primes neutrophils to undergo NETosis, which impairs wound healing. Nat Med. 2015;21(7):815-819. Stavrou EX, Fang C, Bane KL, et al. Factor XII and uPAR upregulate neutrophil functions to influence wound healing. J Clin Invest. 2018;128(3):944-959. Wilgus TA, Roy S, McDaniel JC. Neutrophils and Wound Repair: Positive Actions and Negative Reactions. Adv Wound Care (New Rochelle). 2013;2(7):379-388. Lishko VK, Burke T, Ugarova T. Antiadhesive effect of fibrinogen: a safeguard for thrombus stability. Blood. 2007;109(4):1541-1549. Crane MJ, Daley JM, van Houtte O, Brancato SK, Henry WL Jr, Albina JE. The monocyte to macrophage transition in the murine sterile wound. PLoS One. 2014; 9(1):e86660.

42. Francke A, Herold J, Weinert S, Strasser RH, Braun-Dullaeus RC. Generation of mature murine monocytes from heterogeneous bone marrow and description of their properties. J Histochem Cytochem. 2011; 59(9):813-825. 43. Lucas T, Waisman A, Ranjan R, et al. Differential roles of macrophages in diverse phases of skin repair. J Immunol. 2010; 184(7):3964-3977. 44. Macrae FL, Duval C, Papareddy P, et al. A fibrin biofilm covers blood clots and protects from microbial invasion. J Clin Invest. 2018;128(8):3356-3368. 45. Kim MH, Liu W, Borjesson DL, et al. Dynamics of neutrophil infiltration during cutaneous wound healing and infection using fluorescence imaging. J Invest Dermatol. 2008;128(7):1812-1820. 46. Lindholm C, Searle R. Wound management for the 21st century: combining effectiveness and efficiency. Int Wound J. 2016;13 Suppl 2:5-15.

haematologica | 2019; 104(8)

ARTICLE

Platelet Biology & its DIsorders

All-trans retinoic acid protects mesenchymal stem cells from immune thrombocytopenia by regulating the complement–interleukin-1β loop

Ferrata Storti Foundation

Xiaolu Zhu,1 Yanan Wang,1 Qian Jiang,1 Hao Jiang,1 Jin Lu,1 Yazhe Wang,1 Yuan Kong,1 Yingjun Chang,1 Lanping Xu,1 Jun Peng,2 Ming Hou,2 Xiaojun Huang1 and Xiaohui Zhang1

Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation, National Clinical Research Center for Hematologic Disease, Beijing and 2Department of Hematology, Qilu Hospital, Shandong University, Jinan, P.R. China 1

Haematologica 2019 Volume 104(8):1661-1675

ABSTRACT

nhanced peripheral complement activation has long been considered as one of the major pathogenic elements of immune thrombocytopenia. A dysfunctional bone marrow microenvironment, especially with regards to mesenchymal stem cells, has been observed in patients with immune thrombocytopenia. However, the potential role of the complement system in the dysfunctional bone marrow microenvironment remains poorly understood. In this study, bone marrow samples from patients with immune thrombocytopenia were divided into two groups based on whether or not complement components were deposited on the surfaces of their mesenchymal stem cells. The mesenchymal cells from the group with complement deposition were less numerous, dysfunctional, had a reduced capacity to proliferate, and showed increased apoptosis as well as abnormal secretion of interleukin-1β and C-X-C motif chemokine ligand 12. In vitro treatment with all-trans retinoic acid increased the number and improved the function of the complement-positive bone marrow mesenchymal stem cells by upregulating DNA hypermethylation of the interleukin-1β promoter. In vivo studies showed that all-trans retinoic acid could rescue the impaired mesenchymal stem cells to support the thrombopoietic niche in both patients with immune thrombocytopenia and a murine model of this disease. Taken together, these results indicate that impairment of mesenchymal stem cells, mediated by the complement–interleukin-1β loop, plays a role in the pathogenesis of immune thrombocytopenia. All-trans retinoic acid represents a promising therapeutic approach in patients with immune thrombocytopenia through its effect of repairing mesenchymal stem cell impairment.

Introduction Immune thrombocytopenia (ITP) is a common autoimmune disorder characterized by severe isolated thrombocytopenia.1 Increasing evidence suggests a role for complement activation in ITP.1-3 We previously characterized the abnormal enhanced complement activation in plasma samples from patients with ITP, as well as enhanced plasma complement activation/fixation capacity on immobilized heterologous platelets.4 Moreover, we confirmed the activation of both the classical and alternative complement pathways in the peripheral blood of patients with ITP.4 However, our knowledge regarding the involvement of the complement system in the bone marrow of patients with ITP is still very limited. Emerging evidence indicates that complement affects not only B-cell responses5,6 but also T-cell immunity during the induction, effector and contraction phases of an immune response.7-10 Interestingly, we and others have identified an imbalance between B-effector and T-regulatory networks involved in the pathogenesis of ITP.1,11-14 Mesenchymal stem cells (MSC) have been documented to play crucial haematologica | 2019; 104(8)

Partial results of this research were presented as an oral presentation at the 58th American Society of Hematology Annual Meeting and Exposition (ASH), San Diego, CA, 3 December 2016 (“Crucial Role of Complement Activation and IL-1β in Bone Marrow Niche of Immune Thrombocytopenia”, n. 166).

Correspondence: XIAOHUI ZHANG zhangxh100@sina.com Received: August 13, 2018. Accepted: January 21, 2019. Pre-published: January 24, 2019. doi:10.3324/haematol.2018.204446 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1661 ©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.

1661

X. Zhu et al.

roles in immune modulatory functions with effects on Tand B-cell activation.15,16 Notably, we found that MSC from ITP patients exhibited increased apoptosis and senescence, which was associated with the regulation of T-cell subsets.17-19 However, the underlying mechanisms of the dysfunction of MSC in ITP bone marrow remain unclear. We, therefore, wondered whether complement activation in bone marrow was associated with defective MSC in ITP. Complement components can enhance pro-inflammatory receptor-mediated signaling in phagocytes, leading to increased production of interleukin-1β (IL-1β).20-22 IL-1β is critically involved in several inflammatory diseases and its levels have been reported to be elevated in ITP.23 Interestingly, bone marrow MSC have been demonstrated to be capable of synthesizing and releasing IL-1β.24,25 All-trans retinoic acid (ATRA) has revolutionized the therapy of acute promyelocytic leukemia.26 We previously reported that the combination of ATRA with any one of methylprednisolone, danazol or cyclosporine A produced better responses in patients with corticosteroid-resistant or relapsed ITP (54th American Society of Hematology Annual Meeting and Exposition; Poster ID: 3338). Recently we also reported the findings of a multicenter, randomized, open-label, phase II trial, suggesting that ATRA represents a promising candidate treatment for patients with corticosteroid-resistant or relapsed ITP.27 Panzer and Pabinger positively appraised our findings of a high response rate to ATRA as well as the few, mild adverse events associated with this drug compared with other second-line treatments for ITP.28 However, few studies have focused on the mechanisms underlying the effects of ATRA.29 Furthermore, the role of ATRA in regulating MSC function in ITP bone marrow is poorly understood. It has not been elucidated whether the complement system and associated pro-inflammatory cytokines are targeted by ATRA. Here, we present evidence strongly suggesting that the complement-IL-1β loop mediates bone marrow MSC impairment in ITP. More importantly, ATRA protects MSC from dysfunction and apoptosis by upregulating DNA hypermethylation of the IL-1β promoter, which is conducive to restoring the thrombopoietic niche. We believe that these findings will serve to shift the focus of future studies on the complement system in the pathogenesis of ITP and interventions with ATRA to factors that regulate thrombocytopoiesis.

Methods Patients and study design The blood samples utilized in this study were collected between December 2016 and November 2017 from 58 consecutive, newly diagnosed ITP patients at the Institute of Hematology, Peking University People’s Hospital, Beijing, China. Approval to take blood and bone marrow samples from healthy volunteers and patients was granted by the Ethics Committee of Peking University People’s Hospital, and written informed consent was obtained from all subjects according to the Declaration of Helsinki. Only untreated patients over 18 years old at diagnosis with platelet counts <30×109/L were enrolled. ITP was diagnosed based on guidelines for ITP.30,31 Bone marrow samples were also taken from transplant donors (n=42), who were considered as 1662

healthy controls. The healthy control cohort comprised 17 males and 25 females, aged 18-59 years (median, 38 years). Patients in the group given combination therapy received 10 mg of oral ATRA twice daily and concomitant therapy with oral danazol at a relatively low dose of 200 mg twice daily consecutively. Initial response was assessed after 4, 8 and 12 weeks of treatment. The primary endpoint of the study was overall response. Secondary endpoints were complete response, response, time to response, peak platelet count, reduction in bleeding symptoms, and safety. A complete response was defined as a platelet count of at least 100×109/L. Response was defined as a platelet count between 30×109/L and 100×109/L and at least doubling of the baseline count. Overall response included complete responses and responses. No response was defined as a platelet count lower than 30×109/L or less than doubling of the baseline count.31 Additionally, we defined time to response as the time from starting treatment to the time to achieve the response. We defined bleeding in accordance with the World Health Organization’s bleeding scale (0 = no bleeding, 1 = petechiae, 2 = mild blood loss, 3 = gross blood loss, and 4 = debilitating blood loss). Lastly, we graded adverse events according to the Common Terminology Criteria for Adverse Events (version 4.0).

Animal model and treatment All animal experiments were approved by the Ethics Committee of Peking University People’s Hospital and undertaken in accordance with the Institutional Guidelines for the Care and Use of Laboratory Animals. The inclusion and exclusion criteria for patients, the methods for MSC isolation, RNA extraction and microarrays, the immunofluorescence assays, enzyme-linked immunosorbent assays (ELISA), cell proliferation assays, apoptosis assays, western blotting, determination of CXCR4 expression on megakaryocyte surfaces, analysis of CD34+ cells, analysis of colony-forming unitmegakaryocytes, the modified monoclonal antibody-specific immobilization of platelet antigens assay, analysis of DNA methylation, real-time polymerase chain reaction analysis, gene silencing, gene overexpression, immunostaining, in situ hybridization, animal model and treatment, and statistics are explained in the Online Supplement.

Results Complement activation in the bone marrow of patients with immune thrombocytopenia To illuminate the role of the complement system in the bone marrow of ITP patients, we used indirect ELISA to detect the deposition of complement proteins C1q, C4d, C3b and C5b-9 on the surface of MSC from ITP patients and healthy volunteers. Reference ranges of complement deposition on MSC from the healthy controls were determined: C1q (1.0 ± 0.1), C4d (0.9 ± 0.2), C3b (0.9 ± 0.1) and C5b-9 (1.0 ± 0.4). The cutoff values defining deposition of C1q, C4d, C3b and C5b-9 were 1.2, 1.3, 1.1 and 1.8, respectively. These cutoffs represent a level of complement deposition that falls approximately two standard deviations above the reference mean for the majority of complement components. Patients were categorized as having complement activation if the level of one or more of the measured complement components was equal to or more than the cutoff value. In this study, 26 of the 58 patients were assigned to the group with complement activation (MSC-ITP-C+ group), while the remaining 32 patients were haematologica | 2019; 104(8)

ATRA protects impaired BM MSC in ITP

assigned to the group without complement activation (MSC-ITP-C- group). Forty-two healthy volunteers served as a control group (MSC-control group) (Online Supplementary Table S1). To further distinguish ITP patients with complement activation in bone marrow, deposition of complement proteins was also determined by indirect immunofluorescence assays (Figure 1H, I). In accordance with the results of the indirect ELISA, MSC from ITP patients in the MSC-ITP-C+ group were evidently stained with the complement components, especially C4d and C5b-9 (Online Supplementary Figure S1).

Intrinsic mRNA alterations in complement-activated mesenchymal stem cells from patients with immune thrombocytopenia To investigate whether the MSC from patients in the

MSC-ITP-C+ group show transcriptome abnormalities and to better understand molecular pathways that may regulate MSC-ITP-C+ biology, we performed an mRNA expression analysis by microarray. Unsupervised clustering analysis comparing MSC-ITPC+ to MSC-ITP-C- and MSC-control showed distinctive gene expression signatures and we observed differential expression (fold change >2, P<0.05) of 2,978 probes between the groups, with 1,926 upregulated and 1,052 downregulated (Figure 1A). Bioinformatics analysis was used to identify complement, IL-1 family, IL-1 receptor family, CXC chemokine family, CXC chemokine receptor family, tumor necrosis factor (TNF) family and TNF receptor family-related genes that were differentially expressed in the MSC-ITP-C+ group (fold change >2.0, P<0.05) (Online Supplementary Figure S2A-D).

Figure 1. Mesenchymal stem cells with complement deposition from patients with immune thrombocytopenia are genetically and functionally abnormal. (A) Unsupervised clustering analysis of differentially selected probes. (B) Morphology of mesenchymal stem cells (MSC) from the three groups (MSC-ITP-C+: MSC from patients with immune thrombocytopenia with complement deposition on MSC; MSC-ITP-C- MSC from patients with immune thrombocytopenia without complement deposition on MSC; MSC-control: MSC from healthy subjects) under a light microscope (original magnification ×3200; scale bar: 200 μm). (C) The growth curves of the MSC-ITP-C+ (n=26), MSC-ITP-C- (n=32) and MSC-control (n=42) groups at passage 3 from four independent experiments. One-way analysis of variance (ANOVA). (D) Cell apoptosis of MSC determined by annexin V assays (MSC-ITP-C+, n=12; MSC-ITP-C-, n=21; MSC-control, n=12; one-way ANOVA). (E) The levels of tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), CXCL12, and the complement factors C3a and C5a in bone marrow supernatants from the MSC-control (n=42), MSC-ITP-C(n=32) and MSC-ITP-C+ (n=26) groups; one-way ANOVA). (F) The correlation between the level of IL-1β in culture supernatants and C5b-9 deposition on MSC (R2 = 0.7426, P<0.001, Spearman rank correlation rho). (G) Levels of factor H in MSC culture supernatants 12, 24, and 48 h after co-culture with TNF-α, determined by enzyme-linked immunosorbent assays (paired t-tests). (H, I) Intracellular expression of IL-1β and CXCL12 in MSC-ITP-C+, MSC-ITP-C- and MSC-control, determined by immunofluorescence assays. Scale bar: 100 μm. DAPI: 4′,6-diamidino-2-phenylindole.

haematologica | 2019; 104(8)

1663

X. Zhu et al.

Figure 2. Altered CXCL12 gradients and megakaryocyte marrow niche occupancy in immune thrombocytopenia patients with complement deposition on their mesenchymal stem cells. (A-C) Representative images of radioactive in situ hybridization with CXCL12 antisense probe [CXCL12 transcripts: brown pseudocolor; 4′,6diamidino-2-phenylindole (DAPI): blue] on bone marrow sections from the three groups [MSC-control: mesenchymal stem cells (MSC) from healthy subjects; MSCITP-C-: MSC from patients with immune thrombocytopenia (ITP) without complement deposition on MSC; MSC-ITP-C+: MSC from patients with ITP with complement deposition on MSC]. Scale bar: 100 μm. (D) Representative images of marrow immunohistochemistry for CD41 (megakaryocytes, green) and CD31 (vascular endothelium, red) from the MSC-control, MSC-ITP-C- and MSC-ITP-C+ groups (left: bone-associated marrow; right: central marrow). Scale bar: 100 μm. (E) Ratio of CXCL12 transcript area in the bone-associated region (between 0-100 μm from the endosteal surface within the diaphysis) compared to an immediately adjacent region of the same size (between 100-200 μm from the endosteal surface) for ITP patients (MSC-ITP-C- and MSC-ITP-C+) and healthy volunteers (MSC-control). (F) Quantification of CD41+ megakaryocytes (MKs) physically associated with CD31+ vessels (between 0-10 μm from the sinusoidal endothelium within the diaphysis) by immunohistochemistry. (G) Quantification of CD41+ megakaryocytes in the bone-associated region (within 100 μm of the endosteal surface within the diaphysis) by immunohistochemistry. (H) Quantification of vessel densities by immunofluorescence staining for CD31. (I) Quantification of CD41+ megakaryocytes per section. (EI) MSC-control, n=36; MSC-ITP-C-, n=36; MSC-ITP-C+, n=36; one-way analysis of variance).

1664

haematologica | 2019; 104(8)

ATRA protects impaired BM MSC in ITP

To evaluate the regulation of MSC-ITP-C+ on cell signaling, we performed a canonical pathway analysis. Canonical pathway annotation enabled us to assign differentially expressed genes to 28 pathways. The first nine enriched pathways were the unfolded protein response, protein ubiquitination pathway, cell cycle, IL-1 signaling, p53 signaling, endoplasmic reticulum stress pathway, ERK5 signaling, NRF2-mediated oxidative stress response, and p38 MAPK signaling, which might convey the differences between MSC-ITP-C+, MSC-ITP-C-, and MSC-control. The first 12 enriched pathways are shown in Online Supplementary Figure S2E. Disease and functional heatmaps revealed the activated or inhibitory relationship between differentially expressed genes and diseases and functions (Online Supplementary Figure S2F). “Organismal death” (Z-score = 16.497) and “Morbidity or mortality” (Z-score = 16.484) were significantly activated in MSC-ITP-C+.

Complement-activated mesenchymal stem cells from immune thrombocytopenia patients showed increased apoptosis and functional impairment Bone marrow MSC were successfully isolated. Flow cytometry analysis demonstrated that MSC from both healthy donors and ITP patients expressed CD105, CD73, and CD90 and lacked expression of CD14, CD19, CD34, CD45, and HLA-DR (Online Supplementary Figure S3). MSC-control expanded and acquired a spindle shape morphology during culture. In contrast, MSC-ITP-C+ expanded more slowly and appeared larger and flattened (Figure 1B). CCK8 proliferative assays were conducted on MSC at days 1, 3, 7 and 14 after the third passage. The growth curves showed a lower proliferative capacity of MSC-ITP-C+ (Figure 1C). We further assessed the apoptotic cell rate using annexin V. As shown in Figure 1D, the rates of both early apoptosis and late apoptosis were higher in the MSC-ITP-C+ group. Since the complement system has been shown to be associated with the inflammatory response, we determined cytokine levels in bone marrow supernatants from ITP patients and healthy controls using ELISA. The levels of the inflammatory factors IL-1β and TNF-α were both significantly higher (P<0.001) in the MSC-ITP-C+ group compared with the MSC-ITP-C- and MSC-control groups (Figure 1E), while the levels of CXCL12 were significantly lower (P<0.001) (Figure 1E). Intracellular expression of IL1β and CXCL12 was further confirmed by immunofluorescence assays (Figure 1H, I). Moreover, the levels of expression of complement activation fragments C3a and C5a were also significantly higher (P<0.001 for both) in MSC-ITP-C+ (Figure 1E). There was a positive correlation between the level of IL-1β in culture supernatant and C5b-9 deposition on MSC (R2 = 0.7426, P<0.001, Spearman rank correlation rho) (Figure 1F). In view of the diminished expression of complement factor H found by microarray analysis in the MSC-ITP-C+ group, we performed a TNF-α stimulation test. Factor H concentrations in MSC culture supernatants 12, 24, and 48 h after co-culture with TNF-α were measured by ELISA. Upon exposure to TNF-α, no significant effect on the secretion of factor H was seen in the MSC-ITP-C+ group, which was not consistent with the significantly increased factor H secretion observed in the MSC-ITP-Cand MSC-control groups (Figure 1G). haematologica | 2019; 104(8)

Altered CXCL12 gradients and megakaryocyte marrow niche occupancy in immune thrombocytopenia patients with mesenchymal stem cell complement deposition As the MSC-ITP-C+ group showed attenuated expression of both CXCL12 protein and its encoding gene, we investigated whether there are changes in bone marrow CXCL12 and their consequences for megakaryocytes. Given the importance of the location of CXCL12 production for this chemokine’s chemotactic function32 and the varied injury and recovery kinetics of different bone marrow cell populations in MSC-ITP-C+, we determined the location of CXCL12 transcripts in bone marrow sections of the MSC-ITP-C+, MSC-ITP-C- and MSC-control groups by radioactive in situ hybridization (Figure 2A-C). As expected, the MSC-ITP-C+ group showed an overall increase in the CXCL12 message in the bone-associated marrow and a decrease in the central marrow (Figure 2C). Differences in the distribution of CXCL12 in the bone marrow were quantified by comparing CXCL12 transcript levels within the bone-associated region to the levels in an adjacent region within the central marrow (Figure 2E). Interestingly, we detected the development of a CXCL12 gradient toward the endosteum in the MSCITP-C+ group (Figure 2E). Since megakaryocytes interact with sinusoidal endothelium, we examined the effects of observed shifts of CXCL12 expression on marrow vasculature by immunohistochemistry (Figure 2D). A noticeable, marked decrease in megakaryocytes associated with sinusoids was observed in the MSC-ITP-C+ group (Figure 2D, F). Given the development of a CXCL12 gradient toward the endosteum in the MSC-ITP-C+ group, we also quantified megakaryocyte distribution in the bone-associated region, defined as the region within 100 μm of the diaphyseal endosteum.33 Strikingly, there was an increase in megakaryocytes in the bone-associated marrow, coincident with the increased CXCL12 gradient (Figure 2D, G). Altogether, these results implicated spatial and temporal alterations in marrow CXCL12 of the MSC-ITP-C+ group in the observed changes in megakaryocyte niche occupancy. The overall numbers of megakaryocytes and the density of vessels were not significantly different between the MSC-control, MSC-ITP-C- and MSC-ITP-C+ groups (Figure 2H, I).

The C5b-9/interleukin-1β loop regulates mesenchymal stem cells from patients with immune thrombocytopenia On the basis of the finding of enhanced expression of IL-1β in the MSC-ITP-C+ group, we hypothesized that IL1β induced by complement activation plays an important role in the dysfunction of MSC. The production of pro– IL-1β and caspase-1 was significantly higher in MSC-ITPC+ than in MSC-ITP-C- and MSC-control (Figure 3A), indicating that complement induced IL-1β maturation and caspase-1 processing. To obtain further insight into the mechanisms underlying the increased expression of IL-1β, we then analyzed the role of signaling pathways in MSC from ITP patients and healthy volunteers. Expression of the interleukin-1 receptor (IL-1R) was upregulated on MSC-ITP-C+ (Figure 3B) and increased activation of MyD88 and NF-KB (p65) was detected in MSC-ITP-C+ (Figure 3C). Simultaneously, the expression levels of p-ERK1/2 and p-p38 MAPK were significantly higher in MSC-ITP-C+ than in the other two 1665

X. Zhu et al.

groups (Figure 3D). Collectively, these data indicate the involvement of IL-1R/MyD88/NF-kB, ERK1/2 and p38 MAPK signaling pathways in MSC-ITP-C+. To determine whether the autocrine IL-1β protein is crucial for the dysfunction of MSC, MSC-control were transfected with IL-1β-cDNA and signal sequences of IL1R antagonist. The secretion of IL-1β was markedly improved in MSC cell lysates on day 7, while CXCL12 expression was inhibited (Figure 3E, F). Moreover, phosphorylation of MyD88, ERK1/2, p38 MAPK and NF-κB was enhanced (Figure 3I). Lentiviruses carrying IL-1β-

siRNA were transfected into MSC-ITP-C+. The expression of IL-1β was markedly reduced in the cell lysates on day 7 (Figure 3G). Conversely, CXCL12 expression was upregulated (Figure 3H). IL-1β knockdown decreased the phosphorylation of MyD88, ERK1/2, p38 MAPK and NFκB (Figure 3I). As expected, CCK8 proliferative assays at days 1, 3 and 7 revealed altered proliferative capacity (Figure 3J, K). The apoptosis rate was higher in MSC-control transfected with IL-1β cDNA and lower in MSC-ITPC+ with IL-1β-siRNA (Figure 3L, M). Furthermore, bone marrow sections from patients in the MSC-ITP-C+ group

Figure 3. The C5b-9/interleukin-1β loop regulates mesenchymal stem cells from patients with immune thrombocytopenia. (A, B) Levels of pro–interleukin-1β (IL1β) and caspase 1 (A) and interleukin-1 receptor (IL-1R) (B) in the three groups [MSC-control: mesenchymal stem cells (MSC) from healthy subjects; MSC-ITP-C- MSC from patients with immune thrombocytopenia (ITP) without complement deposition on MSC; MSC-ITP-C+: MSC from patients with ITP with complement deposition on MSC] were detected by western blotting in MSC cell lysates. β-actin was used as the loading control (MSC-control, n=12; MSC-ITP-C-, n=12; MSC-ITP-C+, n=12; One-way analysis of variance (ANOVA). (C, D) Phosphorylation of MyD88, NF-κB, ERK1/2 and p38 MAPK signaling pathway proteins in MSC from ITP patients and healthy volunteers. β-actin was used as the loading control (MSC-control, n=12; MSC-ITP-C-, n=12; MSC-ITP-C+, n=12; one-way ANOVA). (E, F) Expression of IL-1β and CXCL12 in MSC-control cell lysates at day 1 and day 7 with transfection with lentivirus carrying the IL-1β-cDNA from two independent experiments (n=12; Student t-tests). (G, H) Expression of IL-1β and CXCL12 in MSC-ITP-C+ cell lysates 7 days after transfection with lentivirus carrying the IL-1β-siRNA from two independent experiments (n=12; Student t-tests). (I) Phosphorylation of MyD88, ERK1/2, p38 MAPK and NF-κB in transfected MSC by Western blotting (n=8; Student t-tests). (J, K) CCK8 proliferative assays of transfected MSC from three independent experiments (MSC-control, n=12; MSC-ITP-C+, n=12; Student t-tests). (L, M) Apoptosis rate of transfected MSC (MSC-control, n=12; MSC-ITP-C+, n=12; χ2 tests).

1666

haematologica | 2019; 104(8)

ATRA protects impaired BM MSC in ITP

revealed increased intensity of IL-1β fluorescence and number of TUNEL+CD90+ MSC in the central marrow (Online Supplementary Figures S4A and S5A). These results imply that the C5b-9/IL-1β loop contributes to the dysfunction of MSC-ITP-C+.

All-trans retinoic acid directed in vitro functional recovery of complement-activated mesenchymal stem cells from patients with immune thrombocytopenia We previously reported the therapeutic efficacy of ATRA in ITP patients.27 In order to ascertain the mecha-

Figure 4. All-trans retinoic acid directed in vitro functional recovery of mesenchymal stem cells from immune thrombocytopenia patients with complement deposition on their mesenchymal stem cells. (A) CCK8 proliferative assays of all-trans retinoic acid (ATRA) treated mesenchymal stem cells (MSC) from patients with immune thrombocytopenia (ITP) with complement deposition on their MSC (MSC-ITP-C+) from three independent experiments (n=12; Student t-tests). (B) Apoptosis rate of ATRA-treated MSC-ITP-C+ (n=12; Student t-tests). (C, D) Intracellular expression of interleukin-1β (IL-1β) and CXCL12 in untreated and treated MSC-ITP-C+ by immunofluorescence assays. (E, F) The levels of IL-1β and CXCL12 in MSC-ITP-C+ cell lysates at day 1 and day 7 with the administration of ATRA from two independent experiments (n=12; Student t-tests). (G) Phosphorylation of MyD88, ERK1/2, p38 MAPK and NF-κB in MSC-ITP-C+ before and after treatment with ATRA by western blotting (n=8; Student t-tests). (H, I) mRNA levels of IL-1β and CXCL12 in MSC from healthy controls (MSC-control), MSC from ITP without complement deposition on their MSC (MSC-ITP-C-) and MSC-ITP-C+ before and after treatment with ATRA. mRNA levels of IL-1β and CXCL12 were normalized to the mRNA levels of 18srRNA (n=12 independent experiments). Tukey tests for multiple comparisons among pre-treatment groups. Student t-tests for the comparisons between pre-treatment and post-treatment groups from independent experiments. (J) Promoter DNA methylation of IL-1β in MSC-control, MSC-ITP-C- and MSC-ITP-C+ before and after treatment with ATRA. Percentages of DNA methylation were obtained by analyzing the mean methylation of all the CpG sites present in the promoter (n=12 independent experiments). Tukey tests for multiple comparisons among pre-treatment groups. χ2 tests for the comparisons between pre-treatment and post-treatment groups from independent experiments. (K) The methylation status of -299, -256, -20 and +13 CpG sites in the IL-1β proximal promoter (n=12 independent experiments). (L) There was a linear relationship between DNA methylation of the IL-1β promoter and mRNA levels (R2 = 0.5925, P<0.001, Spearman rank correlation rho).

haematologica | 2019; 104(8)

1667

X. Zhu et al.

nisms underlying the response to ATRA, MSC from ITP patients and healthy volunteers were treated with ATRA. As was foreseeable, exposure of MSC-ITP-C+ to ATRA led to marked proliferation and a reduced rate of apoptosis (Figure 4A, B). Downregulated secretion of IL-1β and upregulated levels of CXCL12 were found in the treated MSC-ITP-C+ group (Figure 4C, D). Levels of IL-1β and CXCL12 in MSC cell lysates were regulated accordingly (Figure 4E, F). Distinctly, the treated group showed inhibited phosphorylation of MyD88, ERK1/2, p38 MAPK and NF-κB (Figure 4G). No significant differences in proliferative capacity, apoptosis rate, cytokine secretion and signaling pathways were observed in either the MSC-ITP-Cgroup (Online Supplementary Figure S6) or the MSC-control group (Online Supplementary Figure S7) before or after the administration of ATRA. We further examined the potential genetic mechanism

of the suppressive effect of ATRA on IL-1β in MSC-ITPC+. Analysis of the expression of the gene encoding IL-1β showed a dramatic, significant expression of IL-1β mRNA levels by MSC-ITP-C+ (Figure 4H). ATRA produced a modest decrease in IL-1β mRNA expression of MSC-ITPC+, while it had no effect on IL-1β mRNA levels of MSCITP-C- and MSC-control (Figure 4H). CXCL12 mRNA levels were downregulated in MSC-ITP-C+ and could be increased by the administration of ATRA (Figure 4I). The frequency of DNA hypermethylation of the IL-1β promoter in MSC-ITP-C+ was significantly lower than that in MSC-ITP-C- and MSC-control (Figure 4J). ATRA-treated MSC-ITP-C+ had higher DNA methylation of the IL-1β promoter (Figure 4J). We further analyzed the percentage methylation at the -299, -256, -20 and +13 CpG sites in the IL-1β promoter (Figure 4K). The percentage methylation was 50%, 40%, 15%, and 18% at the -299, -256, -20,

Figure 5. All-trans retinoic acid directed functional recovery of bone marrow mesenchymal stem cells from mice with immune thrombocytopenia. (A) Peripheral platelet counts of wildtype (WT) mice (n=15), immune thrombocytopenia (ITP) mice (n=15), ITP mice treated with a 5 mg/kg intraperitoneal injection of all-trans retinoic acid (ATRA) (n=11) and ITP mice treated with a 20 mg/kg intraperitoneal injection of ATRA (n=13). Tukey tests. (B) The growth curves of bone marrow mesenchymal stem cells (MSC) from WT mice (n=6), ITP mice (n=6), ITP mice treated with ATRA 5 mg/kg intraperitoneal injection (n=6) and ITP mice treated with ATRA 20 mg/kg intraperitoneal injection (n=6) at passages 7-12. Covariance analyses. (C, D) Cell apoptosis of bone marrow MSC determined by annexin V assays (WT mice, n=6; ITP mice, n=6; ITP mice treated with ATRA 5 mg/kg intraperitoneal injection, n=6; ITP mice treated with ATRA 20 mg/kg intraperitoneal injection, n=6; Tukey tests). (E) The phenotypes of mice bone marrow MSC identified by flow cytometer. (F) C5b-9 deposition (green) assayed by marrow immunohistochemistry of WT mice and ITP mice (CD31, vascular endothelium, red). S, sinusoid. Scale bar: 0.1 mm. (G) The levels of interleukin-1β (IL-1β) and CXCL12 in bone marrow supernatants (WT mice, n=6; ITP mice, n=6; ITP mice treated with ATRA 5 mg/kg intraperitoneal injection, n=6; ITP mice treated with ATRA 20 mg/kg intraperitoneal injection, n=6; Tukey tests). (Continued on next page)

1668

haematologica | 2019; 104(8)

ATRA protects impaired BM MSC in ITP

and +13 CpG sites, respectively. DNA methylation of the IL-1β promoter and mRNA levels were inversely correlated (R2 = 0.5925, P<0.001, Spearman rank correlation rho) (Figure 4L). Above all, in the presence of ATRA, IL-1β expression in MSC-ITP-C+ was alleviated and this modification was associated with hypermethylation.

All-trans retinoic acid directed the functional recovery of bone marrow mesenchymal stem cells of mice with immune thrombocytopenia CD61+ SCID mice engrafted with 5×104 splenocytes

from CD61 knockout (KO) mice immunized against CD61+-platelets exhibited profound thrombocytopenia34 (Online Supplementary Figure S8) and some animals died of bleeding within 2-3 weeks after transfer. Platelet counts were significantly lower in ITP mice than in wild-type (WT) mice (68.73±38.76×109/L vs. 1054.07±138.93×109/L, respectively; P=0.000) (Figure 5A). The phenotypes of mice bone marrow MSC were determined by flow cytometry. The results showed that these cells were homogenously positive for the mesenchymal markers CD29, CD90 and CD105, but negative for the hematopoi-

Figure 5. (H) Left: expression of CXCL12 (red) and CD31 (vascular endothelium, green) in the bone marrow niche assayed by marrow immunohistochemistry of WT mice, ITP mice, ITP mice treated with ATRA 5 mg/kg intraperitoneal injection and ITP mice treated with ATRA 20 mg/kg intraperitoneal injection. Right: Hematoxylin & eosin staining of marrow sections from WT mice, ITP mice, ITP mice treated with 5 mg/kg intraperitoneal injection and ITP mice treated with 20 mg/kg intraperitoneal injection. S, sinusoid; E, endosteum; MK, megakaryocyte. Scale bar: 0.1 mm. (I) Quantification of megakaryocytes in the bone-associated region (within 100 μm of the endosteal surface within the diaphysis) by immunohistochemistry (WT mice, n=27; ITP mice, n=27; ITP mice treated with ATRA 5 mg/kg intraperitoneal injection, n=15; ITP mice treated with ATRA 20 mg/kg intraperitoneal injection, n=21). (Continued from previous page). (J) Quantification of megakaryocytes physically associated with sinusoids (between 0-10 μm from the sinusoidal endothelium within the diaphysis) by immunohistochemistry (WT mice, n=27; ITP mice, n=27; ITP mice treated with ATRA 5 mg/kg intraperitoneal injection, n=15; ITP mice treated with ATRA 20 mg/kg intraperitoneal injection, n=21). (K) Quantification of the vascular area within marrow vessels by immunohistochemistry (WT mice, n=27; ITP mice, n=27; ITP mice treated with ATRA 5 mg/kg intraperitoneal injection, n=15; ITP mice treated with ATRA 20 mg/kg intraperitoneal injection, n = 21). (I-K) Tukey tests.

haematologica | 2019; 104(8)

1669

X. Zhu et al.

etic markers CD34 and CD45 (Figure 5E). CCK8 proliferative assays were conducted on MSC at days 1, 3, and 7 with passages 7-12. The growth curves showed a lower proliferative capacity of MSC from ITP mice (Figure 5B), while annexin V studies showed a higher rate of apoptosis (P<0.001) (Figure 5C, D). Enhanced complement activation was confirmed by marrow immunohistochemistry, which showed increased deposition of C5b-9 (Figure 5F). The levels of IL-1β were significantly higher in supernatants of bone marrow from ITP mice than in those from WT mice (P=0.000) (Figure 5G), while the levels of CXCL12 were significantly lower (P=0.011) (Figure 5G). Similar to the findings in the marrow of patients from the MSC-ITP-C+ group, CXCL12 fluorescence intensity was increased in the bone-associated marrow and decreased in the central marrow in ITP mice (Figure 5H),

implying a CXCL12 gradient toward the endosteal niche. Increased fluorescence intensity of IL-1β and numbers of TUNEL+CD90+ MSC in the central marrow were also identified in ITP mice (Online Supplementary Figures S4B and S5B). Notably, there was a marked vascular dilation in the area of the bone marrow occupied by the vasculature in ITP mice (Figure 5H, K). The vascular dilation was accompanied by a decrease in the megakaryocytes associated with sinusoids in ITP mice (Figure 5H-J). ATRA (5 mg/kg or 20 mg/kg dissolved in dimethyl sulfoxide) was administered for 10 days by intraperitoneal injection. The platelet counts in the group treated with 20 mg/kg ATRA recovered, although they remained below the platelet counts of WT mice (Figure 5A). This group also exhibited improved MSC proliferation (Figure 5B) and less MSC apoptosis (Figure 5C, D), decreased levels of

F E

Figure 6. In vivo treatment with all-trans retinoic acid improved the function of bone marrow mesenchymal stem cells from patients with immune thrombocytopenia. (A, B) The effect of all-trans retinoic acid (ATRA) on peripheral platelet counts of the two groups of patients with immune thrombocytopenia (ITP) [MSC-ITP-C+ patients with ITP with complement deposition on their mesenchymal stem cells (MSC); MSC-ITP-C-: patients with ITP without complement deposition on their MSC]. MSC-ITP-C+ group (n=26) and MSC-ITP-C- group (n=32). Paired t-tests. (C) The proportions of patients with an overall response (OR) or response (R) in the MSC-ITPC+ group (n=26) and the MSC-ITP-C- group (n=32). Student t-tests. (D) The growth curves of bone marrow MSC from patients in the MSC-ITP-C+ group (n=26) and the MSC-ITP-C- group (n=32) before and after treatment with ATRA. Covariance analyses. (E) Cell apoptosis of bone marrow MSC of the patients from the MSC-ITP-C+ group (n=26) and the MSC-ITP-C- group (n=32) before and after treatment with ATRA. χ2 tests for comparisons between pre-treatment groups. Paired t-tests for comparisons of samples of patients from pre- and post-treatment groups. (F, G) The levels of interleukin-1β (IL-1β) and CXCL12 in MSC lysates (MSC-ITP-C+ group, n=26; MSC-ITP-C- group, n=32; paired t-tests). (Continued on next page)

1670

haematologica | 2019; 104(8)

ATRA protects impaired BM MSC in ITP

IL-1β in bone marrow supernatants, as well as downregulated IL-1β fluorescence intensity and reduced TUNEL+CD90+ MSC in the central marrow (Online Supplementary Figures S4D and S5D). In contrast, the group treated with 5 mg/kg ATRA did not show marked improvement of the above-mentioned parameters (Figure 5A-D, G, Online Supplementary Figures S4C and S5C). The group treated with 20 mg/kg ATRA displayed increased CXCL12 fluorescence intensity in the central marrow and decreased intensity in the bone-associated marrow (Figure 5H), implying a reversal of the CXCL12 gradient toward the vascular niche. A reduction in vascular area and restoration of megakaryocytes associated with sinusoids in the marrow were also demonstrated in this group (Figure 5H-K). The group treated with 5 mg/kg did not show significant changes in vascular area or megakaryocyte occupation. These results mirrored the complement activation, CXCL12 and megakaryocyte shift

toward the bone-associated marrow in ITP mice. Administration of 20 mg/kg ATRA could regulate the CXCL12 and megakaryocyte shift toward the vasculature and promote platelet production.

Clinical responses to all-trans retinoic acid in patients with immune thrombocytopenia Patients from both the MSC-ITP-C+ and MSC-ITP-Cgroups were given oral ATRA (10 mg twice daily) and concomitant therapy with oral danazol (200 mg twice daily). Online Supplementary Table S2 shows the baseline characteristics of the ITP patients. At the 12-week follow up, the proportion of patients with an overall response was significantly higher in the MSC-ITP-C+ group (19 of 26 patients, 73.08%) than in the MSC-ITP-C- group (13 of 32 patients, 40.63%; P=0.013) (Figure 6C). There were no grade 3 or worse adverse events. No patients required a dose reduction because of adverse events. The main drug-related

Figure 6. (Continued from previous page). (H) Phosphorylation of MyD88, ERK1/2, p38 MAPK and NF-κB in MSC of patients from the MSC-ITP-C+ group (n=8) and the MSC-ITP-C- group (n=8) before and after treatment with ATRA. Student t-tests for comparisons between pre-treatment groups. Paired t-tests for comparisons of samples from patients in the pre- and post-treatment groups. (I) Representative images of radioactive in situ hybridization with CXCL12 antisense probe [CXCL12 transcripts: brown pseudocolor; 4′,6-diamidino-2-phenylindole (DAPI): blue] on marrow sections of the patients from the MSC-ITP-C+ and the MSC-ITP-C- groups. Ratio of CXCL12 transcript area in the bone-associated region (between 0-100 μm from the endosteal surface within the diaphysis) compared to an immediately adjacent region of the same size (between 100-200 μm from the endosteal surface) for ITP patients before and after ATRA treatment. (J) Representative images of bone marrow immunohistochemistry for CD41 (megakaryocytes, green) and CD31 (vascular endothelium, red) for patients from the MSC-ITP-C+ and MSC-ITP-C- groups. (K, L) Quantification of CD41+ megakaryocytes physically associated with CD31+ vessels (between 0-10 μm from the sinusoidal endothelium within the diaphysis) and in the bone-associated region (within 100 μm of the endosteal surface within the diaphysis). (M, N) CD41+ megakaryocytes per section and vessel densities by immunofluorescence staining for CD31. (I, K-N) The MSC-ITP-C+ group, n=36; the MSC-ITP-C- group, n=36; Scale bar: 100 μm; χ2 tests for comparisons between pre-treatment groups. Paired t-tests for comparisons of samples from patients in the pre- and post-treatment groups.

haematologica | 2019; 104(8)

1671

X. Zhu et al.

toxic effects were skin desquamation, headache and dizziness, liver injury, edema, amenorrhea, hypertension and gastrointestinal disorders in the follow-up. Among patients who achieved an overall response, the median time to the treatment response was 34.87 days and the peak platelet count was 105.72×109/L in the MSC-ITP-C+ group compared to 39.00 days and a peak platelet count of 80.62×109/L in the MSC-ITP-C- group (Table 1). To further clarify the mechanisms of the effect of ATRA in ITP, bone marrow MSC were isolated from ITP patients who had received 12 weeks of treatment. Compared with the baseline levels before ATRA therapy, the proliferative capacity, apoptosis, secretion of IL-1β and CXCL12, activation of IL-1R/MyD88/NF-κB, ERK1/2 and p38 MAPK signaling pathways were altered, with trends similar to those in the in vitro studies in the patients from the MSCITP-C+ group after treatment (Figure 6D-H). In contrast, no significant differences were found between pre-treatment and post-treatment levels in patients from the MSC-ITPC- group (Figure 6D-H). Unsurprisingly, patients from the MSC-ITP-C+ group after treatment had overall increases in CXCL12 and megakaryocytes in the central marrow and decreases in the bone-associated marrow (Figure 6I, J), implying a CXCL12 gradient toward the sinusoids again. No significant differences of marrow CXCL12 distribution and spatial occupation of megakaryocytes were exhibited in the patients from the MSC-ITP-C- group after treatment (Figure 6I, J). The numbers of megakaryocytes and the density of vessels did not differ before and after treatment (Figure 6J). These data support the hypothesis that ATRA therapy contributed to changes in the location of CXCL12 production and the altered megakaryocyte location in the bone marrow niche for platelet generation.

Discussion Thrombopoiesis, a complex biological process initiated by pluripotent hematopoietic stem cells in the bone marrow, involves interactions with many cell types that contribute to the bone marrow niche, including osteoblasts, perivascular cells, endothelial cells, MSC, and various mature immune cells. Abnormalities during any stage of thrombopoiesis can influence platelet production.35 In the current study, we describe MSC impairment in the bone marrow of patients with ITP resulting from complementderived perturbations in the IL-1β/IL-1R/NF-κB, ERK1/2 and p38 MAPK signaling pathways. Importantly, ATRA was found to decrease IL-1β mRNA expression and increase promoter DNA methylation. ATRA promoted functional recovery of MSC-ITP-C+ in vitro and in vivo, which facilitated the re-location of CXCL12 toward the vascular niche, enhanced megakaryocyte localization in the thrombopoietic niche and consequently promoted circulating platelet production (Figure 7). Platelet destruction in ITP occurs through a variety of different immune-mediated mechanisms.36 Current theories propose that antibodies against platelet-induced Fc-mediated phagocytosis, anti-GPIb/IX-mediated desialylation and the activation of cytotoxic lymphocytes are involved in the pathogenesis of ITP.1,37-39 The complement system has long been suspected to participate in platelet elimination in patients with ITP.1-4 Activation of the complement cascade was found to play a role in the pathogenesis of ITP 1672

Table 1. Responses and outcomes of immune thrombocytopenia patients with or without complement deposition on mesenchymal stem cells.

MSC-ITP-C+ (n = 26)

MSC-ITP-C(n = 32)

P value

Treatment response (%) Overall response 19 (73.08) 13 (40.63) Complete response 4 (15.38) 4 (12.50) Response 15 (57.69) 9 (28.13) No resonse 7 (26.92) 19 (59.37) Blood component transfusion (%) 16 (61.54) 15 (46.88) Significant bleeding *(%) For patients with a response/ 0 (0) 1 (3.13) complete response Time to response, days, median 34.89 (23-52) 39.00 (27-53) (range) Peak platelet count, ×109/L, median 105.72 (62-201) 80.62 (37-112) (range)

0.013

0.266 0.363 0.190 0.048

MSC-ITP-C+: mesenchymal stem cells with complement deposition from patients with immune thrombocytopenia; MSC-ITP-C-: mesenchymal stem cells without complement deposition from patients with immune thrombocytopenia. *Significant bleeding was defined as bleeding events with a bleeding score of 3-4..

via complement-mediated generation of platelet microparticles in the peripheral blood. Activation of the complement system in the bone marrow of ITP patients was explored for the first time in this study. The observed enhanced deposition of C4d on MSC from patients with ITP was consistent with activation of the classical complement pathway. The lack of observed increases in C3b deposition on MSC from ITP patients indicated that the alternative complement pathway was not activated in bone marrow or that the extent of activation was at least less than the test threshold. Moreover, of 26 patients in the MSC-ITP-C+ group, five (19.2%) did not have detectable platelet antibodies, whereas of 32 patients in the MSC-ITPC- group, 14 (43.7%) tested positive. Based on our results, there was not a one-to-one correspondence between the presence of autoantibodies and complement activation in the niche (Online Supplementary Table S3). MSC residing in the bone marrow have long been believed to participate in regulating the balance between hematopoietic stem cell self-renewal and differentiation.40 In addition to their self-renewal properties and potential to differentiate, MSC play crucial roles in immune modulatory functions.41 We and others recently reported that MSC from ITP patients lose their conventional proliferation capacity resulting in defective immunoregulation.17-19,42,43 However, the mechanism underlying these abnormalities remains unknown. In the present study, we not only investigated the enhanced complement deposition on the surface of MSC from ITP patients, but also explored the underlying genetic and molecular changes that give rise to deficiencies in MSC, which were associated with activation of the complement cascade in bone marrow. Complement components can enhance pro-inflammatory Toll-like receptor-mediated signaling in phagocytes, leading to increased production of IL-1β.20-22 IL-1β is critically involved in several inflammatory diseases and its levels are elevated in many conditions characterized by complement overactivation. Furthermore, Martino et al. reported that IL-1β/IL-1R1/MyD88 signaling impairs MSC proliferation, migration and differentiation.44 However, it is not known whether and how complement and IL-1β are linked and their roles in MSC remain elusive. The biology haematologica | 2019; 104(8)

ATRA protects impaired BM MSC in ITP

Figure 7. Proposed model for aberrantly enhanced complement activation and interleukin-1β expression in mesenchymal stem cells from patients with immune thrombocytopenia. C5b-9 deposition and the associated pro-inflammatory cytokine interleukin-1β (IL-1β) could induce dysfunction of bone marrow mesenchymal stem cells (MSC) in immune thrombocytopenia (ITP), and consequently cause dynamic changes in niche CXCL12 and alterations in the location of megakaryocytes in the bone marrow. All-trans retinoic acid (ATRA) has an inhibitory effect on IL-1β mRNA expression and increases promoter DNA methylation, further promoting in vitro and in vivo functional recovery of MSC from ITP patients with complement deposition on their MSC (MSC-ITP-C+), which facilitates the re-location of CXCL12 towards the vascular niche, and enhances megakaryocyte localization in thrombopoietic niche.

of IL-1β is complex with the release of functional IL-1β requiring two steps. The first, rate-limiting step is the transcription of its precursor, pro-IL-1β. The second step involves the conversion of the inactive precursor into the biologically active IL-1β. This process is mediated by inflammasomes that activate cysteine protease caspase1.45 IL-1β and caspase-1 are expressed by monocytes, as previously reported, and interestingly, these molecules are also expressed by MSC.24,25 The microarray data from this study showed an absolute increase in mRNA expression of pro-inflammatory cytokines, including IL-1β, in the MSC-ITP-C+ group. Further results suggested a complement-IL-1β loop in MSC, which contributed to the dysfunction and apoptosis of MSC in ITP. Interestingly, the numbers of hematopoietic stem cells and megakaryocyte progenitors were not significantly different between the MSC-ITP-C+, MSC-ITP-C- and MSC-control groups (Online Supplementary Figure S9). Whether MSC or osteoblasts (derived from MSC) are the primary target for attack by the complement-IL-1β loop in ITP remains to be elucidated. Several studies have implicated chemokine CXCL12 signaling, through the CXCR4 receptor, in the maturational localization of megakaryocytes to the vascular niche. Some cell types within the bone marrow produce CXCL12, including osteoblasts, endothelial cells, and mesenchymal stromal cell populations.46,47 Despite the growing body of evidence indicating a role for CXCL12/CXCR4 in megakaryopoiesis, the effects of deficient MSC on CXCL12 distribution and megakaryocyte localization in ITP remain unknown. Here, we observed decreased expression of CXCL12 mRNA and protein in haematologica | 2019; 104(8)

MSC cell lysates, accompanied by an overall increase in bone-associated/central marrow CXCL12 and a decrease in the megakaryocytes associated with sinusoids in MSCITP-C+. In addition to enlarging the sample size, further investigations, such as establishing a transgenic murine model, would specifically help to identify the distribution of CXCL12 in the bone marrow niche. CXCR4 expression on megakaryocyte surfaces was higher in patients with ITP, with no difference between the MSC-ITP-C+ and the MSC-ITP-C- groups (Online Supplementary Figure S10). ATRA has revolutionized the therapy of acute promyelocytic leukemia.26 ATRA also shows potential as an immune modulator, given that inflammatory cytokine production was affected in the presence of ATRA. In vitro studies have shown that natural regulatory T cells treated with ATRA are resistant to T-helper cell conversion and maintain FOXP3 expression under inflammatory conditions. ATRA could also promote transforming growth factor-β-induced regulatory T cells and inhibit the differentiation of T-helper-1 and T-helper-17 cells, indicating that ATRA is a novel treatment for immune-mediated diseases.48,49 In patients with ITP, treatment with ATRA restored decreased concentrations of regulatory T cells and IL-10, reduced FOXP3 expression and restored the balance of macrophages towards M2.29 In addition, previous studies demonstrated the ability of ATRA to inhibit IL-1β mRNA expression and its potential role in regulating DNA promoter methylation. Importantly, we previously reported the clinical efficacy of ATRA in patients with ITP,27 although its effect on primary ITP remains unknown. These findings led us to explore whether the mechanisms of action of ATRA in the treatment of ITP are 1673

X. Zhu et al.

associated with MSC function via the regulation of IL-1β synthesis. Expectedly, ATRA inhibited IL-1β mRNA expression in MSC via upregulation of promoter DNA methylation. Further studies are warranted to confirm these data in a larger group of patients with primary ITP. In conclusion, we have shown that enhanced complement activation and the associated pro-inflammatory cytokine IL-1β could induce dysfunction of bone marrow MSC in ITP and consequently cause dynamic changes in niche CXCL12 and alterations in the location of megakaryocytes in the bone marrow. ATRA is a promising therapeutic approach for repairing MSC dysfunction in primary ITP patients. The present study sheds light on complement activation in the pathogenesis of ITP and

References 1. Toltl LJ, Nazi I, Jafari R, Arnold DM. Piecing together the humoral and cellular mechanisms of immune thrombocytopenia. Semin Thromb Hemost. 2011;37(6):631-639. 2. Najaoui A, Bakchoul T, Stoy J, et al. Autoantibody-mediated complement activation on platelets is a common finding in patients with immune thrombocytopenic purpura (ITP). Eur J Haematol. 2012;88(2): 167-174. 3. Peerschke EI, Andemariam B, Yin W, Bussel JB. Complement activation on platelets correlates with a decrease in circulating immature platelets in patients with immune thrombocytopenic purpura. Br J Haematol. 2010;148 (4):638-645. 4. Zhu X, Zhang J, Wang Q, et al. Diminished expression of beta2-GPI is associated with a reduced ability to mitigate complement activation in anti-GPIIb/IIIa-mediated immune thrombocytopenia. Ann Hematol. 2018;97 (4):641-654. 5. Carroll MC. The complement system in regulation of adaptive immunity. Nat Immunol. 2004;5(10):981-986. 6. Hardy RR, Hayakawa K. B cell development pathways. Annu Rev Immunol. 2001;19: 595-621. 7. Heeger PS, Kemper C. Novel roles of complement in T effector cell regulation. Immunobiology. 2012;217(2):216-224. 8. Kemper C, Atkinson JP. T-cell regulation: with complements from innate immunity. Nat Rev Immunol. 2007;7(1):9-18. 9. Longhi MP, Harris CL, Morgan BP, Gallimore A. Holding T cells in check--a new role for complement regulators? Trends Immunol. 2006;27(2):102-108. 10. Kemper C, Chan AC, Green JM, Brett KA, Murphy KM, Atkinson JP. Activation of human CD4+ cells with CD3 and CD46 induces a T-regulatory cell 1 phenotype. Nature. 2003;421(6921):388-392. 11. Audia S, Mahevas M, Samson M, Godeau B, Bonnotte B. Pathogenesis of immune thrombocytopenia. Autoimmun Rev. 2017;16(6): 620-632. 12. Song Y, Wang YT, Huang XJ, Kong Y. Abnormalities of the bone marrow immune microenvironment in patients with immune thrombocytopenia. Ann Hematol. 2016;95 (6):959-965. 13. Zufferey A, Kapur R, Semple JW. Pathogenesis and therapeutic mechanisms in immune thrombocytopenia (ITP). J Clin Med. 2017;6(2):pii:E16.

1674

provides new shreds of evidence confirming the general applicability of ATRA in the treatment of ITP. Acknowledgments This work was supported by the National Natural Science Foundation of China (n. 81470343 and n. 81670116), a Key Program of the National Natural Science Foundation of China (n. 81730004), the Beijing Natural Science Foundation (n. 7171013), the Beijing Municipal Science and Technology Commission (n. Z171100001017084), the National Key Research and Development Program of China (n. 2017YFA0105500 and n. 2017YFA0105503) and the Foundation for Innovation Research Groups of the National Natural Science Foundation of China (n. 81621001).

14. Ji L, Zhan Y, Hua F, et al. The ratio of Treg/Th17 cells correlates with the disease activity of primary immune thrombocytopenia. PLoS One. 2012;7(12):e50909. 15. Glennie S, Soeiro I, Dyson PJ, Lam EW, Dazzi F. Bone marrow mesenchymal stem cells induce division arrest anergy of activated T cells. Blood. 2005;105(7):2821-2827. 16. Tabera S, Pérez-Simón JA, Díez-Campelo M, et al. The effect of mesenchymal stem cells on the viability, proliferation and differentiation of B-lymphocytes. Haematologica. 2008;93(9):1301-1309. 17. Zhang JM, Feng FE, Wang QM, et al. Platelet-derived growth factor-BB protects mesenchymal stem cells (MSCs) derived from immune thrombocytopenia patients against apoptosis and senescence and maintains MSC-mediated immunosuppression. Stem Cells Transl Med. 2016;5(12):16311643. 18. He Y, Xu LL, Feng FE, et al. Mesenchymal stem cell deficiency influences megakaryocytopoiesis through the TNFAIP3/NFkappaB/SMAD pathway in patients with immune thrombocytopenia. Br J Haematol. 2018;180(3):395-411. 19. Xu LL, Fu HX, Zhang JM, et al. Impaired function of bone marrow mesenchymal stem cells from immune thrombocytopenia patients in inducing regulatory dendritic cell differentiation through the Notch-1/Jagged1 signaling pathway. Stem Cells Dev. 2017;26(22):1648-1661. 20. Laudisi F, Spreafico R, Evrard M, et al. Cutting edge: the NLRP3 inflammasome links complement-mediated inflammation and IL-1beta release. J Immunol. 2013;191(3):1006-1010. 21. Fang C, Zhang X, Miwa T, Song WC. Complement promotes the development of inflammatory T-helper 17 cells through synergistic interaction with Toll-like receptor signaling and interleukin-6 production. Blood. 2009;114(5):1005-1015. 22. Zhang X, Kimura Y, Fang C, et al. Regulation of Toll-like receptor-mediated inflammatory response by complement in vivo. Blood. 2007;110(1):228-236. 23. Hua F, Li Y, Zhao X, et al. The expression profile of toll-like receptor signaling molecules in CD19(+) B cells from patients with primary immune thrombocytopenia. Immunol Lett. 2016;176:28-35. 24. Liu N, Wang H, Han G, Tian J, Hu W, Zhang J. Alleviation of apoptosis of bone marrowderived mesenchymal stem cells in the acute injured kidney by heme oxygenase-1 gene

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

modification. Int J Biochem Cell Biol. 2015;69:85-94. Romieu-Mourez R, Francois M, Boivin MN, Bouchentouf M, Spaner DE, Galipeau J. Cytokine modulation of TLR expression and activation in mesenchymal stromal cells leads to a proinflammatory phenotype. J Immunol. 2009;182(12):7963-7973. Cicconi L, Lo-Coco F. Current management of newly diagnosed acute promyelocytic leukemia. Ann Oncol. 2016;27(8):14741481. Feng FE, Feng R, Wang M, et al. Oral alltrans retinoic acid plus danazol versus danazol as second-line treatment in adults with primary immune thrombocytopenia: a multicentre, randomised, open-label, phase 2 trial. Lancet Haematol. 2017;4(10):e487e496. Panzer S, Pabinger I. Is oral all-trans retinoic acid plus danazol a refinement of secondline therapy for primary immune thrombocytopenia in adults? Lancet Haematol. 2017;4(10):e457-e458. Feng Q, Xu M, Yu YY, et al. High-dose dexamethasone or all-trans-retinoic acid restores the balance of macrophages towards M2 in immune thrombocytopenia. J Thromb Haemost. 2017;15(9):1845-1858. Provan D, Stasi R, Newland AC, et al. International consensus report on the investigation and management of primary immune thrombocytopenia. Blood. 2010; 115(2):168-186. Neunert C, Lim W, Crowther M, Cohen A, Solberg LJ, Crowther MA. The American Society of Hematology 2011 evidence-based practice guideline for immune thrombocytopenia. Blood. 2011;117(16):4190-4207. Laguri C, Arenzana-Seisdedos F, LortatJacob H. Relationships between glycosaminoglycan and receptor binding sites in chemokines-the CXCL12 example. Carbohydr Res. 2008;343(11):2018-2023. Niswander LM, Fegan KH, Kingsley PD, McGrath KE, Palis J. SDF-1 dynamically mediates megakaryocyte niche occupancy and thrombopoiesis at steady state and following radiation injury. Blood. 2014;124(2): 277-286. Chow L, Aslam R, Speck ER, et al. A murine model of severe immune thrombocytopenia is induced by antibody- and CD8+ T cellmediated responses that are differentially sensitive to therapy. Blood. 2010;115(6): 1247-1253. Wang L, Li Y, Hou M. Idiopathic thrombocytopenic purpura and dysmegakaryocy-

haematologica | 2019; 104(8)

ATRA protects impaired BM MSC in ITP

36.

37.

38.

39.

40.

topoiesis. Crit Rev Oncol Hematol. 2007;64(2):83-89. Cines DB, Bussel JB, Liebman HA, Luning PE. The ITP syndrome: pathogenic and clinical diversity. Blood. 2009;113(26):65116521. Zhang F, Chu X, Wang L, et al. Cell-mediated lysis of autologous platelets in chronic idiopathic thrombocytopenic purpura. Eur J Haematol. 2006;76(5):427-431. Li J, van der Wal DE, Zhu G, et al. Desialylation is a mechanism of Fc-independent platelet clearance and a therapeutic target in immune thrombocytopenia. Nat Commun. 2015;6:7737. Zhao C, Li X, Zhang F, Wang L, Peng J, Hou M. Increased cytotoxic T-lymphocyte-mediated cytotoxicity predominant in patients with idiopathic thrombocytopenic purpura without platelet autoantibodies. Haematologica. 2008;93(9):1428-1430. Almeida-Porada G, Porada CD, Tran N, Zanjani ED. Cotransplantation of human stromal cell progenitors into preimmune fetal sheep results in early appearance of

haematologica | 2019; 104(8)

41.

42.

43.

44.

human donor cells in circulation and boosts cell levels in bone marrow at later time points after transplantation. Blood. 2000;95 (11):3620-3627. English K, Ryan JM, Tobin L, Murphy MJ, Barry FP, Mahon BP. Cell contact, prostaglandin E(2) and transforming growth factor beta 1 play non-redundant roles in human mesenchymal stem cell induction of CD4+CD25(High) forkhead box P3+ regulatory T cells. Clin Exp Immunol. 2009;156(1):149-160. PĂŠrez-SimĂłn JA, Tabera S, Sarasquete ME, et al. Mesenchymal stem cells are functionally abnormal in patients with immune thrombocytopenic purpura. Cytotherapy. 2009;11(6):698-705. Zhang D, Li H, Ma L, et al. The defective bone marrow-derived mesenchymal stem cells in patients with chronic immune thrombocytopenia. Autoimmunity. 2014;47 (8):519-529. Martino MM, Maruyama K, Kuhn GA, et al. Inhibition of IL-1R1/MyD88 signalling promotes mesenchymal stem cell-driven tissue

regeneration. Nat Commun. 2016;7:11051. 45. de Mooij C, Netea MG, van der Velden W, Blijlevens N. Targeting the interleukin-1 pathway in patients with hematological disorders. Blood. 2017;129(24):3155-3164. 46. Wynn RF, Hart CA, Corradi-Perini C, et al. A small proportion of mesenchymal stem cells strongly expresses functionally active CXCR4 receptor capable of promoting migration to bone marrow. Blood. 2004;104 (9):2643-2645. 47. Ding L, Morrison SJ. Haematopoietic stem cells and early lymphoid progenitors occupy distinct bone marrow niches. Nature. 2013;495(7440):231-235. 48. Liu ZM, Wang KP, Ma J, Guo ZS. The role of all-trans retinoic acid in the biology of Foxp3+ regulatory T cells. Cell Mol Immunol. 2015;12(5):553-557. 49. Lu L, Lan Q, Li Z, et al. Critical role of alltrans retinoic acid in stabilizing human natural regulatory T cells under inflammatory conditions. Proc Natl Acad Sci U S A. 2014;111(33): E3432-E3440.

1675

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(8):1676-1681

Hemostasis

Recurrent stroke: the role of thrombophilia in a large international pediatric stroke population

Gabrielle deVeber,1* Fenella Kirkham,2,3* Kelsey Shannon,1 Leonardo Brandão,1 Ronald Sträter,4 Gili Kenet,5 Hartmut Clausnizer,6 Mahendranath Moharir,1 Martina Kausch,6 Rand Askalan,1 Daune MacGregor,1 Monika Stoll,7 Antje Torge,6 Nomazulu Dlamini,1 Vijeja Ganesan,2 Mara Prengler,2 Jaspal Singh3 and Ulrike Nowak-Göttl4/6

The Hospital for Sick Children, Toronto, Canada; 2Developmental Neurosciences Programme, UCL Great Ormond Street Institute of Child Health, London, UK; 3University Hospital Southampton, UK; 4Department of Paediatric Haematology/Oncology, University of Münster, Münster, Germany; 5Pediatric Coagulation Service, National Hemophilia Centre and Institute of Thrombosis and Hemostasis Sheba Medical Center, Tel-Hashomer, Israel; 6Institute of Clinical Chemistry, University Hospital Kiel-Lübeck, Kiel, Germany and 7Department of Genetic Epidemiology, University of Münster, Münster, Germany 1

*GdeV and FK contributed equally to this work.

ABSTRACT

Correspondence: ULRIKE NOWAK-GÖTTL leagottl@uksh.de Received: November 7, 2018. Accepted: January 22, 2019. Pre-published: January 24, 2019. doi:10.3324/haematol.2018.211433 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1676 ©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.

1676

isk factors for arterial ischemic stroke in children include vasculopathy and prothrombotic risk factors but their relative importance to recurrent stroke is uncertain. Data on recurrent stroke from databases held in Canada (Toronto), Germany (Kiel-Lübeck/Münster), and the UK (London/Southampton) were pooled. Data were available from 894 patients aged 1 month to 18 years at first stroke (median age, 6 years) with a median follow-up of 35 months. Among these 894 patients, 160 (17.9%) had a recurrence between 1 day and 136 months after the first stroke (median, 3.1 months). Among 288 children with vasculopathy, recurrence was significantly more common [hazard ratio (HR) 2.5, 95% confidence interval (95% CI) 1.92-3.5] compared to the rate in children without vasculopathy. Adjusting for vasculopathy, isolated antithrombin deficiency (HR 3.9; 95% CI: 1.4-10.9), isolated elevated lipoprotein (a) (HR 2.3; 95% CI: 1.3-4.1), and the presence of more than one prothrombotic risk factor (HR 1.9; 95% CI: 1.12-3.2) were independently associated with an increased risk of recurrence. Recurrence rates calculated per 100 person-years were 10 (95% CI: 3-24) for antithrombin deficiency, 6 (95% CI: 4-9) for elevated lipoprotein (a), and 13 (95% CI: 7-20) for the presence of more than one prothrombotic risk factor. Identifying children at increased risk of a second stroke is important in order to intensify measures aimed at preventing such recurrences.

Introduction Published estimates of the annual incidence of arterial ischemic stroke (AIS) in children range from 1.2 to 8 per 100,000 children.1-5 Prior diagnoses in children with symptomatic AIS are common and include cardiac disorders, hematologic conditions (sickle cell anemia, prothrombotic disorders), collagen tissue diseases, metabolic disorders, other chronic diseases, and acute illnesses.6-8 Childhood infections, including Varicella zoster virus, have been shown to be associated with an increased risk of AIS, with routine vaccination being protective against AIS.9,10 In addition, the presence of prothrombotic risk factors has been found in small case series and case control studies to be associated with ischemic stroke in children and this association has been confirmed by a meta-analysis.11 This is in contrast to perinatal stroke, in which recent studies have not shown an association with thrombophilia12 and recurrence is relatively rare.13,14 haematologica | 2019; 104(8)

Underlying pathologies in pediatric stroke

In children, stroke recurrence is common and associated with significant morbidity and mortality. Five-year recurrence rates are estimated to be between 6-20%, with rates as high as 66% in certain subgroups.15-20 Several studies have identified vasculopathy, in particular moyamoya, as an important factor in predicting recurrent stroke.8,15,20 There are some early data to suggest that prothrombotic states may also enhance the risk of recurrence, but many of these studies are limited by size and scope.7,11,15,16,21 We therefore conducted an international cohort study to investigate the relevance of prothrombotic risk factors, as well as underlying stroke subtypes, to a second stroke in pediatric patients.

‘silent’ recurrent strokes noted on follow-up imaging without clinical manifestations.

Methods

Results

Study population, design, and endpoints

From January 1990 to January 2016, a total of 990 inand out-patients consecutively reviewed at the study sites from Canada (n=308), Germany (n=461), and the UK (n=221), aged >1 month and without sickle cell anemia were enrolled and pooled in the pediatric stroke database located in Germany. After enrollment, we excluded 54 children with moyamoya, two patients with either congenital homozygous protein C or homozygous antithrom-

The present study is a multicenter cohort study to assess the rate of symptomatic stroke recurrence per 100 person-years following a first AIS. The core protocol was developed by the German collaborative group and was adopted by centers in Toronto and the UK; data were pooled across these sites to determine whether the data were generalizable and to increase the statistical power for analysis of the secondary study objective, i.e. the time to recurrence. From January 1990 to January 2016, consecutively admitted in- and out-patients from each study site, i.e. Canada (Toronto: single-center registry), Germany (KielLübeck/Münster: multicenter national registry: patients newly enrolled after 2002), and the UK (London/Southampton: two-center registry) were enrolled and pooled in the pediatric stroke database located in Germany. Consecutive patients with a first symptomatic AIS were recruited whether or not prothrombotic risk factors were present and recurrence was ascertained during follow up in survivors. Patients referred from other tertiary centers were excluded. Neonates <1 month of age and children with sickle cell anemia were not enrolled in the present dataset, as recurrence rates and risk factors differ markedly in these subjects from those in other subtypes of childhood AIS. After enrollment, children with moyamoya, and those with congenital homozygous protein C or antithrombin deficiency were excluded, since recurrence risk and therapy differ substantially from those in the remaining study cohort. In addition, we excluded children in whom thrombophilia screening was not performed and those lost to follow up. The first AIS was confirmed by standard imaging methods, i.e. magnetic resonance imaging and computed tomography.15,16 AIS was defined as an acute-onset neurological deficit with an acute focal infarct in a corresponding arterial vascular territory. Recurrence was defined as clinically symptomatic AIS events which presented as acute focal neurological deficits with infarction in a vascular distribution on neuroimaging and which began more than 24 h following the onset of the first stroke. The fixed study end date for the last follow up was January 1, 2017. The number of patients with recurrence, and type of antithrombotic (antiplatelet or anticoagulant) therapy administered prior to recurrence were recorded. The proportion of deaths following stroke recurrence was also recorded. Following discontinuation of antithrombotic treatment, asymptomatic pediatric patients were followed up every few months for the first year and at more prolonged intervals thereafter (at least yearly). All patients were seen at least once for a follow-up evaluation by a pediatric neurologist. Transient ischemic attacks, defined as acute-onset neurological deficits lasting <24 h and with no associated infarct on repeat neuroimaging, were excluded from the study endpoint, as were haematologica | 2019; 104(8)

Ethics This study was performed in accordance with the ethical standards laid down in the updated version of the 1964 Declaration of Helsinki and was either approved or the requirement for approval was waived, by Research Ethics Boards at the Hospital for Sick Children, Toronto, the Great Ormond Street Hospital, (and UK National Health Service), and the University of Münster. Details of the stroke subtypes,11,22 treatment modalities,21-25 laboratory work-up26-28 and statistical methods applied29,30 are summarized in the Online Supplement.

Figure 1. Flow chart of patients’ distribution through the study. AIS: arterial ischemic stroke.

1677

G. deVeber et al.

bin deficiency, five children who were lost to follow up and 35 children in whom thrombophilia screening was not performed (Figure 1). We therefore studied 894 consecutively recruited children aged 1 month to â&#x2030;¤18 years who survived a first episode of AIS (Figure 1, Table 1) and were followed for a median (minimum-maximum) duration of 35 (1-256) months. These patientsâ&#x20AC;&#x2122; clinical characteristics are summarized in Table 1. In total, 160 children (17.9%) experienced a recurrent AIS at a median time from the initial stroke of 3.1 months (minimum-maximum: 0.1-136). Of the 160 children who had a second AIS, 15 (9.4%) died at the time of the second stroke. The overall recurrence rate, with the 95% confidence interval (95% CI), calculated per 100 person-years was 5 (95% CI: 4-6) with a yearly incidence of 0.05%. Data on antithrombotic prophylaxis prior to a second AIS, i.e. anticoagulation with low molecular weight heparin or a vitamin K antagonist or antiplatelet therapy (with acetylsalicylic acid or clopidogrel), were available for a subgroup of 122 out of the 160 cases on an exploratory basis: antithrombotic prophylaxis was administered independently of the presence or absence of thrombophilia (P=0.89) or different stroke subtypes (P=0.2). In children with vasculopathy prior to the second AIS, 34 of 72 index patients were on antiplatelet agents (acetylsalicylic acid alone), 19 were being treated with low molecular weight heparin, and one child developed a second stroke while taking a vitamin K antagonist. In patients without vasculopathy 15 out of 50 children were on acetylsalicylic acid, in two cases combined with clopidogrel, seven were being treated with low molecular weight heparin and three were on oral anticoagulation with a vitamin K antagonist. Fortyfour of the 122 patients did not receive anticoagulation or antiplatelet therapy prior to a second AIS (vasculopathy group n=19; non-vasculopathy group n=25; P=0.05).

Stroke subtypes Stroke subtypes are presented in Figure 1 and Table 1. Of the 894 incident pediatric stroke cases, 288 (32.2%) had vasculopathy. The sub-classification of the types of vascular stroke is shown in the Online Supplement. Imaging-confirmed recurrent stroke occurred in 160/880 enrolled children, for (i) cardiac disease 23/109 (21.1%), (ii) vasculopathy 82/284 (28.9%), for (iii) cryptogenic 41/401 (10.2%) and for (iv) other stroke types 14/78 (17.9%). Recurrent AIS was significantly more frequent in patients with vasculopathy (HR 2.5, 95% CI: 1.92-3.5; P<0.001) than in patients with cryptogenic stroke. Calculated per 100 person-years in those with vasculopathy, the recurrence rate was 8 (95% CI: 6-10) with a yearly incidence rate of 7.7%. The time to recurrence, calculated as the probability of AIS-free survival, comparing pediatric AIS patients with and without vasculopathy, is depicted in Figure 2 (log rank P-value <0.001). No statistically signifi-

Vascular territory of the second arterial ischemic stroke In the majority of cases (75%) the same vascular territory was involved as in the first AIS. The anterior circulation was involved in 62.5% of cases and the posterior circulation in the other 37.5%.

Figure 2. Arterial ischemic stroke-free survival in children with vasculopathy compared with that in the remaining children with a normal arterial examination (P<0.001). Two additional patients suffering from cardiac disease, autoimmune disorder along with short vessel stenosis are included.

Table 1. Clinical characteristics of the 894 children with a first arterial ischemic stroke studied for stroke recurrence.

Characteristics at first stroke onset

Canada number (%)

Germany number (%)

UK number (%)

Total number (%)

Ethnicity Caucasian Black Asian First nations/Aboriginal Central/South American Mixed ethnicity Unknown Median age, years [min-max] Male Children with a first cryptogenic AIS Children with a first vascular stroke Children with a first cardiac stroke Children with a recurrent AIS Patients on treatment prior to second AIS

294 (100.0) 166 (56.5) 17 (05.8) 44 (15.0) 2 (0.70) 5 (01.7) 9 (03.1) 51 (17.3) 4.8 [0.1-17.7) 183 (59.4)

379 (100.0) 377 (99.5) 2 (0.95)

221 (100.0) 190 (86.0) 4 (2.0) 27 (12.2) -

894 (100.0) 733 (81.2)

7.1 [0.2 -18] 203 (54.7)

4.5 [0.1-16.7] 125 (56.5)

6 (0.1-21] 511 (56.8)

43 (14.6) 89 (30.3) 92 (31.3) 76 (25.9) 54 (71.1)

264 (69.5) 86 (22.7) 8 (2.1) 27 (7.1) 17 (63.0)#

96 (43.4) 113 (51.1) 9 (4.1) 57 (25.8) 30 (52.6)

403 (45.1) 288 (32.2) 109 (12.2) 160 (17.9) 101 (63.1)

Cohort data previously published in part.25

1678

haematologica | 2019; 104(8)

Underlying pathologies in pediatric stroke Table 2. Univariable analysis: association between prothrombotic risk factors and a second stroke. A single prothrombotic disorder was detected in 269 children.

Risk factor

Fasting Lp(a) >30 mg/dL Fibrinogen Fasting homocysteine Antithrombin-deficiency Protein C deficiency Protein S deficiency Factor 5 at rs6025 Factor 2 at rs1799963 Combined defects

AIS first onset: numbers with abnormal test /numbers tested (%)

AIS recurrence: numbers with abnormal test /numbers tested (%)

Reference: no thrombophilia 115/580 (19.8) 43/787 (5.5) 16/708 (2.3) 23/750 (3.1) 26/778 (3.3) 28/708 (4.0) 71/726 (9.8) 21/631 (3.3) 88/848 (10.4)

Chi-squared P-value

23/115 (20.0) 13/43 (30.2) 7/16 (43.8) 7/23 (30.4) 5/26 (19.2) 7/28 (28.0) 8/71 (11.3) 3/21 (15.0) 23/88 (26.1)

0.04 0.04 0.01 0.15 0.9 0.2 0.17 0.84 0.04

AIS: arterial ischemic stroke; Lp: lipoprotein

cant association was found between recurrent AIS and the remaining stroke subgroups: cardiac stroke (HR 1.15, 95% CI: 0.7-2.0; P=0.61); non-vascular/non-cardiac/non-idiopathic (HR 1.01, 95% CI: 0.5-2.0; P=0.95).

Table 3. Risk contribution to a second arterial ischemic stroke adjusted for age at onset, gender and study center (Cox proportional hazards model).

Prothrombotic risk factors

Stroke subtypes Reference: cryptogenic stroke Vascular stroke 2.5 Cardiac stroke 1.15 Non-vascular, non-cardiac non-cryptogenic 1.01 Thrombophilia Reference: no thrombophilia Fasting Lp(a) >30 mg/dL 2.3 Fibrinogen 0.9 Fasting homocysteine 3.6 Heterozygous antithrombin-deficiency 3.9 Protein C deficiency 1.3 Protein S deficiency 2.2 Factor 5 at rs6025 0.7 Factor 2 at rs1799963 1.8 Combined prothrombotic risk factors 1.9

Results derived from univariable analyses are shown in Table 2. A single prothrombotic disorder was detected in 269 children, whereas more than one prothrombotic risk factor was diagnosed in 88 cases. Heterozygous antithrombin deficiency, high lipoprotein (a) [Lp(a)], high fibrinogen, high fasting homocysteine and the presence of more than one prothrombotic disorder were associated with recurrence. Of the 7/23 patients with heterozygous antithrombin deficiency who experienced a recurrent stroke, none was on unfractionated heparin or vitamin K antagonist treatment: four were prescribed acetylsalicylic acid at the time of recurrence, one patient was on low molecular weight heparin, while two subjects were taking no prophylaxis immediately prior to the recurrent stroke (noncompliance was not excluded). In six of 23 (26%) patients with combined defects and a second stroke, the factor 5 mutation at rs6025 was involved. Interestingly, however, the factor 2 mutation at rs1799963 did not play a role in children with combined defects. Examination of the roles of different stroke subtypes and prothrombotic risk factors, using multivariable Cox proportional hazards regression of variables with a Pvalue â&#x2030;¤0.15 in the univariable analyses, adjusted for age, gender and study center, demonstrated that the presence of vasculopathy (HR 2.5), antithrombin deficiency (HR 3.9), elevated Lp(a) (HR 2.3) and the presence of more than one prothrombotic risk factor (HR 1.9) were independently associated with an increased risk of recurrent stroke (Table 3). The time to recurrence, i.e. recurrence-free survival comparing pediatric AIS patients with elevated Lp(a) to those with normal Lp(a) levels is illustrated in Figure 3 (log rank P-value <0.039). Recurrence rates calculated per 100 person-years were 10 (95% CI: 3-24) for antithrombin deficiency, yearly incidence rate 0.1%, 6 (95% CI: 4-9) for elevated Lp(a), yearly incidence rate 0.13%, and 13 (95% CI: 7-20) for the presence of more than one prothrombotic risk factor, yearly incidence rate 0.13%. Exposure time of the children was 2437 years for those with antithrombin haematologica | 2019; 104(8)

Risk factor

Hazard ratio

95% confidence interval

1.92-3.5 0.7-2.0 0.5-2.0

1.3-4.1 0.3-2.8 0.8-15.8 1.4-10.9 0.3-5.5 0.5-9.8 0.23-1.91 0.4-7.8 1.12-3.2

AIS: arterial ischemic stroke; Lp: lipoprotein

deficiency, 1938 years for those with elevated Lp(a) and 2887 years for patients with more than one prothrombitic risk factor. Kaplan-Meier survival curves for children with multiple thrombophilic factors and for those with normal thrombophilic status are illustrated in Online Supplementary Figure S1. Based on data from the Caucasian pediatric population, the number-needed-to-screen to detect one patient with elevated Lp(a) was 10 and that to detect children with more than one prothrombotic abnormality was 20.

Discussion In our study cohort of 894 Canadian, English and German 1679

G. deVeber et al.

pediatric stroke patients > 1 month of age, a second AIS event was diagnosed in 17.9% of patients within a median period of 3.1 months after the first stroke. In this international cohort of children who had had a stroke we found that the presence of more than one prothrombotic risk factor was associated with AIS recurrence. Specifically, as demonstrated recently in children with recurrent deep venous thrombosis and thromboembolic stroke, heterozygous antithrombin deficiency is a major risk factor for second AIS events.31 In addition, data presented here confirm an increased risk for recurrent stroke events in the subgroup of patients with an underlying vasculopathy.15,22,32-35 The rates of recurrent cerebral thrombo-embolic events vary widely across published studies. Differences are likely related to: (i) the variable inclusion of neonates, known to have a very low rate of recurrence; (ii) the ethnicity of the patients enrolled and other variables in the patient populations; (iii) the definitions of recurrence; and (iv) the duration of follow up. Some studies mix transient ischemic attacks and recurrent stroke in reporting the recurrence risk.14-20 Adverse outcomes resulting from recurrent AIS are certainly more ominous than those for a transient ischemic attack alone, with a mortality rate after second AIS of 9.4% in our patients. Keeping in mind differences in patient populations, underlying diseases as well as treatment modalities applied, and assuming that inclusion of transient ischemic attacks would approximately double the recurrence risk,16 the recurrence rate reported by us for AIS alone is within the lower rate of approximately 20% reported by other authors.14-20 It is possible that our lower rate of recurrence reflects the use of standardized treatment protocols and institutional pediatric stroke programs in our centers. Specialized stroke care is likely to lower the rate of recurrence through experienced selection of patients for antiplatelet/anticoagulant treatment and more consistent use of any preventative treatment. With respect to the inherited prothrombotic risk factors investigated, multivariable analysis provided evidence that the presence of vasculopathy, heterozygous antithrombin deficiency, increased Lp(a) and more than one cause of thrombophilia are risk factors for recurrent ischemic stroke in pediatric patients.14,36-38 The numbers needed to screen to detect one patient each with elevated Lp(a) or combined thrombophilic abnormalities in the present cohort were 10 and 20, respectively. The benefit of recognizing an underlying thrombophilic condition, as well as the numbers need to screen to detect a carrier at risk, should be balanced against clinical impact, cost and potential insurance implications. In contrast to the literature on recurrent venous thromboembolism in a similar population of children, as well as an association with first AIS onset,11,38 the presence of isolated mutations in factor 5 at rs6025, factor 2 at rs1799963, protein Cand protein S were not individually significantly associated with recurrent AIS in this multicenter cohort. Despite the recurrence rate of 17.9% (160/894 patients), our study may have been underpowered to find these associations. Alternatively, these factors may have been more aggressively treated with antithrombotic treatment, reducing the risk of recurrence in affected patients. Importantly, combinations of prothrombotic risk factors, including the factor 5 mutation at rs6025, in 26% of cases were associated with recurrence, emphasizing the importance of comprehensive investigation and appropriate management. Our study has several limitations. First, the long duration of the study means that many children were enrolled more than a decade ago, when the selection of treatments may 1680

Figure 3. Arterial ischemic stroke-free survival in children with elevated lipoprotein (a) compared with that in the remaining children with normal lipoprotein (a) levels (P=0.039).

have been different from that currently recommended. While this long duration provided us with the opportunity to monitor children for longer-term recurrent AIS, our inclusion of some children with only a brief follow-up duration, as short as 1 month, could also have resulted in our underestimating recurrence risk, and our rate of recurrence should therefore be viewed as a minimum estimate. Secondly, the proportions of children with vascular, cardiac or cryptogenic stroke varied across countries, likely representing either differences in assignment of patients to categories (e.g. inclusion or exclusion of occlusion alone as ‘vasculopathy’) or different referral patterns to the centers (Toronto, London/Southampton and Kiel-Lübeck/Münster) in the three countries. Thirdly, we were unable to include pediatric stroke drug therapy as a main focus of this study since recommendations on antithrombotic and antiplatelet agents for children with stroke are derived from non-randomized pediatric trials and small case series with a low level of evidence, without adjustment for treatable prothrombotic risk factors such as heterozygous antithrombin deficiency21-25 or routine drug monitoring to detect resistance to acetylsalicylic acid or non-drug compliance. Finally, the thrombophilia testing was done over time and in three different laboratory settings. However, since laboratory parameters were investigated with standard laboratory techniques and assays and were only classified as abnormal when (i) abnormal on a repeat test and (ii) confirmed by family studies or the identification of an underlying gene mutation,26-28 it is likely that our results are reliable. In summary, recurrent AIS is relatively frequent, and is associated with significant mortality. Risk is enhanced in children who have vasculopathy, even when moyamoya is excluded, and in those with certain isolated thrombophilic risk factors or more than one prothrombotic disorder. The results of this study emphasize the value of pooling individual patients’ data across geographic regions. Future studies should seek to validate our findings in additional cohorts of children with a first AIS with stroke subtypes clearly defined according to new pediatric stroke classifications, such as CASCADE.39,40 Of note, our finding that recurrence of childhood AIS is comparable across European and North American centers supports the feasibility of multinational recruitment strategies to provide sufficient power to ranhaematologica | 2019; 104(8)

Underlying pathologies in pediatric stroke

domized treatment studies, which could include the development of stroke recurrence prediction models. Such studies should be focused on prevention of recurrent stroke in subpopulations of pediatric patients with the highest risks of recurrent AIS. In addition, from the data reported here, prediction models could be derived combining non-moyamoya vasculopathy with the presence of multiple thrombophilic risk factors of interest. On the background of regional differences with respect to prevalence rates of thrombophilic risk factors across study populations, the

References 14. 1. Agrawal N, Johnston SC, Wu YW, Sidney S, Fullerton HJ. Imaging data reveal a higher pediaric stroke incidence than prior US estimates. Stroke. 2009;40(11):3415-3421. 2. Tuckuviene R, Christensen AL, Helgestad J, Johnsen SP, Kristensen SR. Paediatric arterial ischaemic stroke and cerebral sinovenous thrombosis in Denmark 1994-2006: a nationwide population-based study. Acta Paediatr. 2011;100(4):543-549. 3. Mallick AA, Ganesan V, Kirkham FJ, et al. Childhood arterial ischaemic stroke incidence, presenting features, and risk factors: a prospective population-based study. Lancet Neurol. 2014;13(1):35-43. 4. Suppiej A, Gentilomo C, Saracco P, et al. First report from the Italian Registry of Pediatric Thrombosis (R. I. T. I., Registro Italiano Trombosi Infantili). Thromb Haemost. 2015;113(6):1270-1277. 5. deVeber GA, Kirton A, Booth FA, et al. Epidemiology and outcomes of arterial ischemic stroke in children: the Canadian Pediatric Ischemic Stroke Registry. Pediatr Neurol. 2017;69:58-70. 6. Mackay M T, Wiznitzer M, Benedict SL, et al. Arterial ischemic stroke risk factors: the International Pediatric Stroke Study. Ann Neurol. 2011;69(1):130-140. 7. Rodan L, McCrindle BW, Manlhiot C, et al. Stroke recurrence in children with congenital heart disease. Ann Neurol. 2012;72(1): 103-111. 8. Fullerton H J, deVeber GA, Hills NK, et al. Inflammatory biomarkers in childhood arterial ischemic stroke: correlates of stroke cause and recurrence. Stroke. 2016;47(9): 2221-2228. 9. Fullerton HJ, Hills NK, Elkind MS, et al. Infection, vaccination, and childhood arterial ischemic stroke: results of the VIPS study. Neurology. 2015;85(17):1459-1466. 10. Elkind MS, Hills NK, Glaser CA, et al. Herpesvirus infections and childhood arterial ischemic stroke: results of the VIPS study. Circulation. 2016;133(8):732-741. 11. Kenet G, Lütkhoff LK, Albisetti M, et al. Impact of thrombophilia on risk of arterial ischemic stroke or cerebral sinovenous thrombosis in neonates and children: a systematic review and meta-analysis of observational studies. Circulation. 2010;121(16): 1838-1847. 12. Curtis C, Mineyko A, Massicotte P, et al. Thrombophilia risk is not increased in children after perinatal stroke. Blood. 2017;129(20):2793-2800. 13. Kurnik K, Kosch A, Sträter R, Schobess R, Heller C, Nowak-Göttl U; Childhood Stroke Study Group. Recurrent thromboembolism in infants and children suffering from symptomatic neonatal arterial

haematologica | 2019; 104(8)

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

numbers need to screen to detect carriers at risk will allow investigators to power future pediatric stroke trials adequately. It is important to keep in mind, however, that (i) antithrombotic and/or antiplatelet therapy may also have a significant impact on the risk of AIS recurrence in children, and (ii) up to now, due to the lack of randomized controlled trials, pediatric stroke treatment modalities are recommended on the basis of low-level evidence.21-25 Further efforts must be made to address the latter issue, as well.

stroke: a prospective follow-up study. Stroke. 2003;34(12): 2887-2892. Lehman LL, Beaute J, Kapur K, et al. Workup for perinatal stroke does not predict recurrence. Stroke. 2017;48(8):2078-2083. Sträter R, Becker S, von Eckardstein, A, et al. Prospective assessment of risk factors for recurrent stroke during childhood--a 5-year follow-up study. Lancet. 2002;360(9345): 1540-1545. Ganesan V, Prengler Wade MA, Kirkham FJ. Clinical and radiological recurrence after childhood arterial ischemic stroke. Circulation. 2006;114(20):2170-2177. Fullerton HJ, Wu YW, Sidney S, Johnston SC. Risk of recurrent childhood arterial ischemic stroke in a population-based cohort: the importance of cerebrovascular imaging. Pediatrics. 2007;119(3):495-501. Braun KP, Bulder MM, Chabrier S, et al. The course and outcome of unilateral intracranial vasculopathy in 79 children with ischaemic stroke. Brain. 2009;132(Pt 2):544-557. Sultan SM, Beslow LA, Vossough A, et al. Predictive validity of severity grading for cerebral steno-occlusive vasculopathy in recurrent childhood ischemic stroke. Int J Stroke. 2015;10(2):213-218. Fullerton HJ, Wintermark M, Hills NK, et al. Risk of recurrent arterial ischemic stroke in childhood: a prospective international study. Stroke. 2016;47(1):53-59. Goldenberg NA, Bernard TJ, Fullerton HJ, Gordon A, deVeber G and International Pediatric Stroke Study. Antithrombotic treatments, outcomes, and prognostic factors in acute childhood-onset arterial ischaemic stroke: a multicentre, observational, cohort study. Lancet Neurol. 2009;8(12):1120-1127. Lanthier S, Carmant L, David M, Larbrisseau A, deVeber G. Stroke in children: the coexistence of multiple risk factors predicts poor outcome. Neurology. 2000;54(2):371-378. Andrew M, deVeber G: Pediatric Thromboembolism and Stroke Protocols. B.C. Decker Inc., Hamilton, 1st edition 1997, 2nd edition 1999 Sträter R, Kurnik K, Heller C, Schobess R, Luigs P, Nowak-Göttl U. Aspirin versus lowdose low-molecular-weight heparin: antithrombotic therapy in pediatric ischemic stroke patients: a prospective follow-up study. Stroke. 2001;32(11):2554-2558. Bernard TJ, Goldenberg NA, Tripputi M, Manco-Johnson MJ, Niederstadt T, NowakGöttl U. Anticoagulation in childhoodonset arterial ischemic stroke with nonmoyamoya vasculopathy: findings from the Colorado and German (COAG) collaboration. Stroke. 2009;40(8):2869-2871. Limperger V, Franke A, Kenet G, et al. Clinical and laboratory characteristics of paediatric and adolescent index cases with

27.

28.

29.

30. 31.

32.

33.

34.

35.

36.

37. 38.

39. 40.

venous thromboembolism and antithrombin deficiency. An observational multicentre cohort study. Thromb Haemost. 2014;112 (3):478-485. Limperger V, Klostermeier UC, Kenet G, et al. Clinical and laboratory characteristics of children with venous thromboembolism and protein C-deficiency: an observational Israeli-German cohort study. Br J Haematol. 2014;167(3):385-393. Klostermeier UC, Limperger V, Kenet G, et al. Role of protein S deficiency in children with venous thromboembolism. An observational international cohort study. Thromb Haemost. 2015;113(2):426-433. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol. 1996;49(12):1373-1379. Mayer D. Essential Evidence-Based Medicine. Cambridge Univ Press 2004; 119120. Brüwer G, Limperger V, Kenet G, et al. Impact of high risk thrombophilia status on recurrence among children and adults with VTE: an observational multicenter cohort study. Blood Cells Mol Dis. 2016; 62:24-31. Nguyen P, Reynaud J, Pouzol P, Munzer M, Richard O, Francois P. Varicella and thrombotic complications associated with transient protein C and protein S deficiencies in children. Eur J Pediatr. 1994;153(9):646-649. Chiche L, Bahnini A, Koskas F, Kieffer E. Occlusive fibromuscular disease of arteries supplying the brain: results of surgical treatment. Ann Vasc Surg. 1997;11(5):496-504. Askalan R, Laughlin S, Mayank S, et al. Chickenpox and stroke in childhood: a study of frequency and causation. Stroke. 2001;32(6):1257-1262. Delsing BJ., Catsman-Berrevoets CE, Appel IM. Early prognostic indicators of outcome in ischemic childhood stroke. Pediatr Neurol. 2001;24(4):283-289. Haywood S, Liesner R, Pindora S, V. Ganesan V. Thrombophilia and first arterial ischaemic stroke: a systematic review. Arch Dis Child. 2005;90(4):402-405. Milionis HJ, Winder AF, Mikhailidis DP. Lipoprotein (a) and stroke. J Clin Pathol. 2000;53(7):487-496. Wityk RJ, Kittner SJ, Jenner JL, et al. Lipoprotein (a) and the risk of ischemic stroke in young women. Atherosclerosis. 2000;150(2):389-396. Stacy A, Toolis C, Ganesan V. Rates and risk factors for arterial ischemic stroke recurrence in children. Stroke. 2018;49:842-847. Chabrier S, Sebire G, Fluss J. Transient cerebral arteriopathy, postvaricella arteriopathy, and focal cerebral arteriopathy or the unique susceptibility of the M1 segment in children with stroke. Stroke. 2016;47:24392441.

1681

ARTICLE Ferrata Storti Foundation

Haematologica 2019 Volume 104(8):1682-1688

Stem Cell Transplantation

Fecal microbiota transplantation before or after allogeneic hematopoietic transplantation in patients with hematologic malignancies carrying multidrug-resistance bacteria Giorgia Battipaglia,1,2 Florent Malard,1,3 Marie Therèse Rubio,1,4,5 Annalisa Ruggeri,1 Anne Claire Mamez,1 Eolia Brissot,1,3 Federica Giannotti,1 Remy Dulery,1 Anne Christine Joly,6 Minh Tam Baylatry,6 Marie Jeanne Kossmann,7 Jacques Tankovic,8 Laurent Beaugerie,6,9 Harry Sokol and 3,9,10,11* Mohamad Mohty1,3* *These authors contributed equally to this work as co-senior authors

Department of Hematology, Saint Antoine Hospital, Paris, France; 2Federico II University, Hematology Department, Naples, Italy; 3Sorbonne Universités, UPMC Univ Paris 06, INSERM, Centre de Recherche Saint-Antoine (CRSA), F-75012 Paris, France; 4Service d’Hématologie, Hôpital Brabois, CHRU Nancy, France; 5CMRS UMR 7563, IMoPa, Biopole de l’Université de Lorraine, France; 6Microbiote Transplant Préparations Unit, Pharmacy Department, Saint Antoine Hospital, Paris, France; 7Unité d’Hygiène et de Lutte Contre les Infections Nosocomiales, Saint Antoine Hospital, Paris, France; 8Department of Bacteriology, Saint Antoine Hospital, Paris, France; 9Department of Gastroenterology, Saint Antoine Hospital, AP-HP, Paris, France; 10Sorbonne Université, École Normale Supérieure, PSL Research University, CNRS, INSERM, AP-HP, Hôpital Saint-Antoine, Laboratoire de Biomolécules, LBM, F-75005 Paris, France and 11INRA, UMR1319 Micalis & AgroParisTech, Jouy en Josas, France 1

ABSTRACT

Correspondence: GIORGIA BATTIPAGLIA giorgia.battipaglia@aphp.fr Received: May 29, 2018. Accepted: January 31, 2019. Pre-published: February 7, 2019. doi:10.3324/haematol.2018.198549 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/8/1682 ©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.

1682

ecal microbiota transplantation is an effective treatment in recurrent Clostridium difficile infection. Promising results to eradicate multidrugresistant bacteria have also been reported with this procedure, but there are safety concerns in immunocompromised patients. We report results in ten adult patients colonized with multidrug-resistant bacteria, undergoing fecal microbiota transplantation before (n=4) or after (n=6) allogeneic hematopoietic stem cell transplantation for hematologic malignancies. Stools were obtained from healthy related or unrelated donors. Fecal material was delivered either by enema or via nasogastric tube. Patients were colonized or had infections from either carbapenemase-producing bacteria (n=8) or vancomycin-resistant enterococci (n=2). Median age at fecal microbiota transplantation was 48 (range, 16-64) years. Three patients needed a second transplant from the same donor due to initial failure of the procedure. With a median follow up of 13 (range, 4-40) months, decolonization was achieved in seven of ten patients. In all patients, fecal microbiota transplantation was safe: one patient presented with constipation during the first five days after FMT and two patients had grade I diarrhea. One case of gut grade III acute graft-versus-host disease occurred after fecal microbiota transplantation. In patients carrying or infected by multidrugresistant bacteria, fecal microbiota transplantation is an effective and safe decolonization strategy, even in those with hematologic malignancies undergoing hematopoietic stem cell transplantation.

Introduction During the last decades, the prevalence of multidrug-resistant bacteria (MDRB) has largely increased, becoming a serious worldwide problem.1 Under physiological conditions, commensal microbiota prevents gut colonization from MDRB. However, in particular conditions, such as in patients with hematologic malignancies, use of chemotherapeutic agents and broad spectrum antibiotics may favor selection of resistant pathogens through the alterations of the gastrointestinal barrier and the consequent dysbiosis.2 Patients undergoing allogeneic hematopoietic stem cell transplantation (allo-HSCT) are at even higher risk of dysbiosis due to haematologica | 2019; 104(8)

FMT in carriers of MDRB undergoing allo-HSCT

their profound immune depression.3 In cases of bloodstream infections from MDRB, outcomes are even poorer, leading to increased mortality.4 For example, an Italian study showed that carbapenemase producing (CP-) bacteria, including Pseudomonas aeruginosa, were independent predictors of death in patients diagnosed with acute leukemia, while this was not observed in cases of extended-spectrum β-lactamase (ESBL) Enterobacteriaceae.5 In order to prevent bacteria spreading to other patients, preventive measures are required, including isolating patients, limitating transfer to other healthcare centers, and management by dedicated staff. These measures result in an increase in healthcare costs, which cannot always be met.6 According to French recommendations, patients colonized with MDRB may be denied access to healthcare facilities if dedicated staff are not available.7 New classes of antibiotics are under study to treat infections related to MDRB, and research is ongoing to find effective decolonization strategies.8 The use of oral gentamicin was initially proposed in some MDR-gram negative strains but failure is common, and management of gentamicin-resistant strains may also be an issue.9,10 Fecal microbiota transplantation (FMT) is a procedure that has been proved to be effective and safe in the treatment of recurrent Clostridium difficile infection (CDI), and it is now a recommended therapy in this setting.11 Use of FMT in patients carrying MDRB is still at an investigational stage, but there are reports and case series showing its efficacy in this setting.12,13 Many concerns were initially raised about the feasibility of FMT in immunocompromised patients, such as those affected by hematologic malignancies, because of the theoretical potential for local and bloodstream infections. However, recent case reports revealed the efficacy and safety in this particular population.14-16 Recently, Bilinski et al. reported the results of a prospective study evaluating FMT in 20 patients with MDRB gut colonization and contemporarily affected by hematologic malignancies. Overall 25 FMT were performed and 15 of 20 patients experienced complete MDRB decolonization,17 including some of them with graft-versus-host disease (GvHD) after allo-HSCT. In this retrospective study, we report our experience with FMT in patients diagnosed with hematologic malignancies and undergoing FMT either before or after alloHSCT.

Methods In this single-center study, we retrospectively analyzed data on all consecutive adult patients diagnosed with hematologic malignancies who underwent FMT before or after allo-HSCT due to MDRB colonization. In our center, microbiological screening is performed weekly in all inpatients, with consequent preventive measures in positive patients in order to limit MDRB spread, according to national guidelines.7 (See Online Supplementary Appendix for details). This study was approved by the institutional ethics committees. The treatment plan was discussed in advance by a multidisciplinary team (hematologist, gastroenterologist, pharmacist) who approved the procedure. The decision was made on a patient-topatient basis. All patients signed informed consent explaining the theoretical risks of the procedure because of the current investigational use of FMT in the field of MDRB and in patients with hematologic malignancies. haematologica | 2019; 104(8)

Eligibility criteria included: asymptomatic carriers or systemic infections from vancomycin-resistant enterococci (VRE), carbapenemase-producing Enterobacteriaceae (CPE) or CP-Pseudomonas aeruginosa. The rationale for FMT and MDRB decolonization were mainly to limit infectious complications related to these bacteria and to facilitate patient transfer to other departments, such as intensive care units or rehabilitation centers. Contemporary colonization from ESBL-producing bacteria was also registered in patients undergoing FMT. We, therefore, subsequently evaluated whether FMT also allowed decolonization from these MDRB. For the purpose of this retrospective analysis, we also classified MDRB as multi-drug (MDR), extensive-drug (XDR) and pan-drugresistant (PDR), according to the definition proposed by Magiorakos et al.;18 MDR was defined as the presence of acquired non-susceptibility to at least one agent in three or more antimicrobial categories, XDR as non-susceptibility to at least one agent in all but two or fewer antimicrobial categories (i.e. bacterial isolates remain susceptible to only one or two categories), and PDR as non-susceptibility to all agents in all antimicrobial categories. Details on donor selection, microbiological testing, fecal material preparation and delivery are available in the Online Supplementary Appendix. Decolonization from VRE, CPE or CP-Pseudomonas aeruginosa after negative results on a minimum of three consecutive microbiological cultures (performed weekly) was defined as “major decolonization”, while “persistent decolonization” was defined as the persistence of negative rectal swab until last follow up after a first or second FMT, whenever this was feasible. In patients concomitantly colonized by ESBL-producing Enterobacteriaceae, “concomitant decolonization” was defined as negative results on at least three consecutive rectal swabs after FMT. The safety of the procedure was also registered. For all patients, data on significant infections (defined as bacteriemia or sepsis occurring during the first 90 days after FMT) were also collected. Although febrile neutropenia or fever of unknown origin was not considered a significant infectious episode, such events were also recorded. A second attempt could be proposed in those patients presenting either a relapse of MDRB colonization or experiencing FMT failure.

Results During the period between 2014 and 2017, ten patients underwent FMT: seven due to gut colonization without systemic infection by either CPE (Escherichia coli, n=1; Citrobacter freundii, n=2; Klebsiella pneumoniae, n=1), or CPPseudomonas aeruginosa (n=1) or VRE (n=2), and three after having experienced systemic infections from CPPseudomonas aeruginosa. Median age at FMT was 48 (range, 16-64) years. Four patients underwent FMT as a decolonization strategy before allo-HSCT, with a median interval from FMT to transplant of 28 (range, 9-46) days. Of note, one patient was contemporarily colonized by three different CPE. Two patients started conditioning regimen three days after FMT and the other two after a month. Six patients underwent FMT after allo-HSCT, with a median time from allo-HSCT to FMT of 163 (range, 98-344) days. Of note, all patients undergoing FMT after allo-HSCT were still on immunosuppressive therapy at the time of FMT, with only one of six presenting active grade IV steroid-dependent gut GvHD. Overall, six patients were also colonized by ESBL-producing Enterobacteriaceae. All ESBL-producing bacteria were classified as MDR. A frozen product was used in eight of ten patients and 1683

G. Battipaglia et al.

Figure 1. Results of fecal microbiota transplantation. FMT: fecal microbiota transplantation; allo-HSCT: allogeneic hematopoietic stem cell transplantation; CPE: carbapenemase-producing Enterobacteriaceae; CP-Pseudomonas Aeruginosa: carbapenemase producing Pseudomonas Aeruginosa; VRE: vancomycin resistant enterococci. *A third patient achieved decolonization from vancomycin-resistant enterococci and then experienced recurrence of colonization 20 months after fecal microbiota transplantation, concomitantly to disease relapse.

enema was the preferred method of administration in all but one patient who presented a compromised neurological status due to a cerebral toxoplasmosis and was not considered eligible for enema. Median donor stool quantity was 84 g (range, 43-104 g). At the time of FMT, neutrophil count was >1x109/L in all patients but one who had a neutrophil count of 0.17x109/L (with steroid-resistant GvHD). Platelet count was count >20x109/L in all patients. Three patients required a second FMT. In one patient, after initial efficacy, VRE was again detectable two months after the first FMT. This patient developed multiple infectious episodes (particularly sinusitis and pneumonia), prompting the frequent use of large spectrum antibiotics; this probably led to recurrence of VRE colonization. In the other two patients, a second attempt was made due to the failure of the first procedure. In one patient, failure was mainly attributable to incorrect preparation with PEG (insufficient intake). After a second attempt with a correct preparation, VRE eradication was achieved and maintained until 20 months after FMT, at which point, VRE was detectable at the same time as recurrence of hematologic disease. In the last patient, administration of first and second FMT was mainly for compassionate use to treat active grade IV gut GvHD and multiple infectious episodes which made withdrawal of antibiotics impossible, even during the 72 hours (h) following FMT (see below). Globally, major decolonization (three consecutive negative microbiological cultures) was achieved in seven of ten patients, including two patients after a second FMT (Figure 1). Persistent decolonization (negative microbiological cultures at last follow up) was achieved in six of ten patients after a median follow up of 13 months (range, 4-40 months) from FMT. Indeed, as already mentioned, one patient presented a positive rectal swab for ERV 20 months after FMT contemporary to disease relapse. She later died due to hematologic progression. Failure occurred in the remaining three patients. The patient undergoing FMT for compassionate use had presented multiple infectious episodes from CP-Pseudomonas aeruginosa, making it impossible to stop antibiotics during the 72 h after FMT. Moreover, grade IV gut GvHD was associated with intestinal occlusion, 1684

requiring aspiration via a nasogastric tube, at time of FMT. Despite two attempts with FMT, the procedure was a failure and the patient later died. In the second patient, due to the problems encountered in the positioning of a nasogastric tube, FMT was administered by enema and the patient was not able to retain the product for the advised 2-3 h; she refused a second attempt. The third patient underwent FMT by enema from an unrelated donor and the reason given for FMT failure was that she had not received sufficient stool quantity (43 g); however, this is not logical given that decolonization from concomitant ESBL-producing Enterobacteriaceae had been achieved. A second attempt in this patient was not possible due to the unavailability of additional material. Among the six patients concomitantly colonized from ESBL-producing Enterobacteriaceae, three obtained concomitant decolonization. Details on FMT performed before or after allo-HSCT are reported in Table 1. As an example of successful FMT, Figure 2 shows the case of the patient undergoing FMT from nasogastric tube, after experiencing breakthrough infectious episodes related to colonization from CP-Pseudomonas aeruginosa requiring continuous hospitalization for the first year after allo-HSCT. After FMT, this patient did not experience any other infectious episode and could finally be cared for as an outpatient. With regards to the safety of FMT procedure, one patient presented constipation during the first five days after FMT which was favorably resolved after the use of laxatives, while two patients presented grade I diarrhea the day after FMT. No other major adverse events were observed. Only one patient undergoing FMT before allo-HSCT developed a grade III acute gut GvHD at day +30 after allo-HSCT and at day +51 after FMT. A differential diagnosis with CMV colitis was made and she responded favorably to both antiviral and steroid treatment. When looking at severe infectious episodes during the 90 days following FMT, in two of those patients undergoing FMT before allo-HSCT, documented bacteriemia without sepsis occurred early after allo-HSCT; these responded favorably to the introduction of large-spectrum antibiotics. In particular, one patient experienced a documented bacteriemia from multi-sensitive Pseudomonas haematologica | 2019; 104(8)

FMT in carriers of MDRB undergoing allo-HSCT

aeruginosa at day +80 after allo-HSCT while the other patient experienced a documented bacteriemia from an ESBL-producing Escherichia Coli at day 60 after allo-HSCT. The additional two patients undergoing FMT before alloHSCT also received a large spectrum antibiotic such as piperacillin-tazobactam or cephalosporins for febrile neutropenia without documentation. Interestingly, despite

the use of large spectrum antibiotics, no cases of MDRB recurrence were observed in these four patients. Fungal and viral infections were observed in only one patient more than six months after FMT, but these were not considered to be related to FMT because this patient was under systemic immunosuppressive treatments for a cortico-resistant extensive GvHD (lung, skin, mucosal)

Table 1. Characteristics of patients undergoing fecal microbiota transplantation before (A) or after (B) hematopoietic stem cell transplantation.

Patient sex Age at time of FMT, years

M 64

M 42

F 45

M 47

B Hematologic malignancy AML AML AML Identified MDRB CP- Pseudomonas aeruginosa CP-Pseudomonas aeruginosa CPE Antimicrobial resistance category XDR MDR MDR Concomitant MDR-ESBL-producing bacteria colonization, bacteria Y N Y Systemic infections due to MDRB before FMT Y N N Time from FMT to allo-HSCT (days) 41 46 16 FMT donor Daughter Sister Husband Way of administration Enema Enema Enema Major decolonization Y Y Y Persistent decolonization Y Y Y Concomitant ESBL-producing bacteria decolonization Y N/A N Follow up after FMT, days 820 368 148 Follow up after allo-HSCT, days 779 322 132 Status Alive Dead Alive Cause of death N/A Disease progression N/A Patient sex F M F F F Age at time of FMT, years 50 54 16 19 62 Hematologic malignancy MPN MPN AML ALL MPN Identified MDRB CP- Pseudomonas CP- Pseudomonas VRE VRE CPE aeruginosa aeruginosa CPE Antimicrobial resistance category PDR XDR XDR XDR MDR Concomitant MDR-ESBL-producing bacteria colonization N Y Y Y N Systemic infections due to MDRB before FMT Y Y N N N Time from allo-HSCT to FMT 324 344 98 160 123 FMT donor Husband Unrelated Mother Mother Brother Way of administration Nasogastric tube Nasogastric tube Enema Enema Enema Second FMT N Y Y Y N Time from first to second FMT, days N/A 27 118 84 N/A Major decolonization Y N Y Y N Persistent decolonization Y N/A Y N N/A Concomitant ESBL-producing bacteria decolonization N/A N N Y N/A Colonization relapse N N/A N Y N/A Follow-up after FMT, days 678 33 1220 595 184 Follow-up after allo-HSCT, days 1002 404 1436 839 307 Status Alive Dead Alive Dead Alive Cause of death N/A Uncontrolled N/A Disease N/A GvHD and infection progression

BPDCN CPE° MDR N N 9 Sister Enema Y Y N/A 399 390 Alive N/A F 54 ALL

XDR Y N 167 Unrelated Enema N N/A N N/A Y N/A 307 474 Alive N/A

°Three different types: Citrobacter freundii, Klebsiella Pneumoniae, Enterobacter Cloacae. F: female; M: male; FMT: fecal microbiota transplantation; AML: acute myeloid leukemia; BPDCN: blastic plasmacytoid dendritic cell neoplasm; MDRB: multidrug-resistant bacteria; CP: carbapenemase-producing; CPE: carbapenemase-producing Enterobacteriaceae; XDR: extensively-drug resistant; MDR: multi-drug resistant; ESBL: extended-spectrum β-lactamase; Y: yes, N: no; allo-HSCT: allogeneic hematopoietic stem cell transplantation; GvHD: graft-versus-host disease; N/A: not applicable; MPN: myeloproliferative neoplasm; ALL: acute lymphoblastic leukemia; VRE: vancomycin-resistant enterococci; PDR: pandrug resistant.

haematologica | 2019; 104(8)

1685

G. Battipaglia et al.

Figure 2. Evolution of Patient n. 5 as model of successful fecal microbiota transplantation. CP: carbapenemase-producing; FMT: fecal microbiota transplantation; HSCT: hematopoietic stem cell transplantation.

and infectious episodes exacerbated during immunosuppressive treatment. None of the other patients presented fungal or viral infections.

Discussion The increasing emergence and diffusion of MDRB represents a major public health problem, with higher mortality in patients experiencing infections. This involves high costs of prolonged in-hospital care and preventive measures used to limit diffusion to other patients.6,19 Human gut microbiota, also named as â&#x20AC;&#x153;gut resistomeâ&#x20AC;?, is the primary site for MDRB acquisition and colonization, being an important reservoir of antibiotic resistant genes.20 Patients diagnosed with hematologic malignancies are at high risk of colonization from MDRB. In fact, conditioning regimens for allo-HSCT and intensive chemotherapy significantly alter the gastrointestinal barrier and this modifies the composition of intestinal microbiota. Moreover, patients affected by hematologic malignancies or undergoing allo-HSCT are at particular risk for MDRB colonization or infection due to the large, prolonged and, sometimes, improper use of large spectrum antibiotics.2 Most bloodstream infections in hematologic patients derive from the gut, and infections are even more severe in those patients undergoing allo-HSCT, with high mortality rates of 36-95%.3,4 It has been widely reported that microbioma modifications are associated to worse survival, and higher risk of infections and GvHD in patients undergoing allo1686

HSCT.21,22 Therefore, efficacious decolonization strategies in this particular setting of patients are urgently needed. Fecal microbiota transplantation is a fascinating decolonization strategy that has been proved to be efficacious in patients with recurrent CDI.23,24 On the other hand, concerns were initially raised for the use of FMT as a decolonization strategy in immunocompromised patients, due to the possible risk of local or systemic infections after the inoculum of microbiota pathogens. Recently, DeFilipp et al. investigated the use of third-party FMT with the use of oral capsules as a strategy to restore microbioma diversity in patients undergoing allo-HSCT. The authors support the safety and feasibility of this procedure, underlining the possibility that microbiome restoration early after allo-HSCT may be of benefit.25 Here we describe the results of FMT in ten patients diagnosed with hematologic malignancies and undergoing FMT for MDRB colonization, namely CPE, CP-Pseudomonas aeruginosa or VRE, either before or after allo-HSCT. Decolonization was achieved in seven of ten patients, this being persistent at last follow up in six of ten patients. Our retrospective study not only suggests the efficacy of this procedure, but also its safety in patients with hematologic malignancies undergoing allo-HSCT. Interestingly, despite not being a selection criterion for FMT, we also registered patients concomitantly colonized from ESBL-producing enterobacteriaceae, with decolonization in three of six cases. We also showed that, in patients experiencing failure or relapse of MDRB colonization, a second FMT is feasible and efficacious. Interestingly, only three patients experienced significant infections after FMT. haematologica | 2019; 104(8)

FMT in carriers of MDRB undergoing allo-HSCT

Moreover, it is worth underlining the significant benefit of major decolonization in the patient who had experienced multiple infectious episodes due to a CP-Pseudomonas aeruginosa, limiting breakthrough infections. Our results also highlight that, despite the fact that administration of large spectrum antibiotics may hypothetically represent a risk for decolonization failure, the procedure remained effective in the majority of patients, without recurrence of MDRB in the majority of them despite use of broad spectrum antibiotics early after FMT. Interestingly, in one patient, VRE was detectable again at the time of disease relapse, despite no large-spectrum antibiotics having been used just before. One can speculate that disease relapse may probably have been associated to dysbiosis favoring selection of VRE, but conclusions cannot be drawn on just one case. Despite the initial aforementioned concerns in immunocompromised patients, results of FMT in this setting are promising in terms of both efficacy and safety.4,15,16 A recent prospective study showed that FMT allowed total eradication of MDRB in 60% of cases, without any significant adverse event after the procedure.17 This is the only prospective study published to date using FMT in 20 patients with blood disorders and colonized with MDRB. Differently from our series, in this study, all types of MDRB were included and only a few patients underwent allo-HSCT. In our Center, we only chose patients colonized with highly resistant bacteria, and in particular those classified as eXDR according to French guidelines or those known to cause a significant higher risk of systemic infection with very poor prognosis (i.e. CP-Pseudomonas aeruginosa). So far, no specific guidelines have been defined as to the ideal timing, the best preparation of stools for FMT, and the best method of administration. In our experience, FMT was successfully undertaken either before or after allo-HSCT and, interestingly, it was also successful in two patients starting conditioning regimen for allo-HSCT three days after FMT. As for stool preparation, frozen material was preferred in our center for logistical reasons, although in two cases fresh stools were used; this did not modify the results of FMT. It has recently been reported in a meta-analysis of patients receiving FMT for CDI, that the success rate of FMT was similar when using frozen or fresh stools.26 In contrast to most of the reported series of FMT for MDRB decolonization, we preferred enema as a method of administration as this is associated with lower risk of inhalation as compared to nasogastric administration. The mechanisms underlying the efficacy of FMT for MDRB decolonization are still not clear. Recent studies showed that recipient stool assumed donor-like taxonomic and functional composition immediately following FMT.27 Therefore, we hypothesize that FMT for MDRB decolonization works through the restoration of a more physiological microbiome, thus increasing the ecological pressure on MDRB. However, given the absence of translational studies on antibiotic resistance genes and microbiota composition on the patientâ&#x20AC;&#x2122;s stool after FMT, we cannot exclude the possibility that FMT works through lowering MDRB below the threshold of detection rather than through true elimination. In our series, after FMT, almost all patients had no major haematologica | 2019; 104(8)

infectious complications during the first three months after FMT. Interestingly, in those patients subsequently undergoing allo-HSCT, no severe infectious bacterial complications occurred during the early transplant phase. Regarding the impact of FMT on GvHD, only one of our patients had a grade IV acute gut GvHD concomitant to a carbapenemase-producing Pseudomonas aeruginosa at the time of FMT. In this specific case, the procedure was not efficacious either for MDRB or for GvHD. However, it is worth underlining that FMT was performed at a very late stage for compassionate use, and this may also explain the failure of the procedure. Importantly, among the nine remaining patients, only one experienced grade III acute gut GvHD after FMT (with a possible differential diagnosis with CMV colitis). A role for FMT in causing GvHD in this patient cannot formally be excluded and this point may be addressed in a prospective clinical trial. Early studies in mice and humans suggested a link between gut microbiota and propensity to GvHD, with mice treated with gut-decontaminating antibiotics developing GvHD less often.28,29 Recent results of a pilot study also highlight the possible advantage of microbiota modulation with FMT in patients affected by steroid-refractory or steroid-dependent GvHD.30 With regards to donor choice, when available, people living in the same household of the patient were preferred as they widely share the same pathogens and environmental exposure, thus reducing the risk of transferring additional infectious agents from the donor to the recipient. In line with previous reports, we consider that targeting gut microbiota in patients with impaired immune reconstitution in an attempt to reinstate a more equilibrated flora may favor stable eradication of the carrier status and prevent subsequent life-threatening infections. This study has some limitations, including its retrospective nature, low number of patients, heterogeneous inclusion criteria, and differences in FMT procedure, making it difficult for any definitive conclusions to be drawn. However, we consider that our results support the use of FMT as a promising strategy to manage the considerable potential risks associated with the MDRB carrier status in immunocompromised patients with intestinal dysbiosis and in those patients having experienced single or multiple systemic infections. The majority of patients experienced no breakthrough infections after decolonization or MDRB recurrence despite the use of broad spectrum antibiotics. Furthermore, our results provide fresh evidence of the safety of the procedure in this population, despite previous concerns in immunocompromised patients. These preliminary results underline the need for further prospective studies on the safety and efficacy of FMT. Acknowledgments The authors thank Prof. Junia V. Melo (University of Adelaide, Australia, and Imperial College, London) for medical editing of this manuscript. Funding This study was supported by educational grants from the â&#x20AC;&#x153;Association for Training, Education and Research in Hematology, Immunology and Transplantationâ&#x20AC;? (ATERHIT, Nantes, France). 1687

G. Battipaglia et al.

References 1. Spellberg B, Guidos R, Gilbert D, et al. The Epidemic of Antibiotic-Resistant Infections: A Call to Action for the Medical Community from the Infectious Diseases Society of America. Clin Infect Dis. 2008;46(2):155-164. 2. Montassier E, Gastinne T, Vangay P, et al. Chemotherapy-driven dysbiosis in the intestinal microbiome. Aliment Pharmacol Ther. 2015;42(5):515-528. 3. Samet A, Śledzińska A, Krawczyk B, et al. Leukemia and risk of recurrent Escherichia coli bacteremia: Genotyping implicates E. Coli translocation from the colon to the bloodstream. Eur J Clin Microbiol Infect Dis. 2013;32(11):1393-1400. 4. Bilinski J, Robak K, Peric Z, et al. Impact of Gut Colonization by Antibiotic-Resistant Bacteria on the Outcomes of Allogeneic Hematopoietic Stem Cell Transplantation: A Retrospective, Single-Center Study. Biol Blood Marrow Transplant. 2016; 22(6):10871093. 5. Cattaneo C, Zappasodi P, Mancini V, et al. Emerging resistant bacteria strains in bloodstream infections of acute leukaemia patients: results of a prospective study by the Rete Ematologica Lombarda (Rel). Ann Hematol. 2016;95(12):1955-1963. 6. Birgand G, Leroy C, Nerome S, et al. Costs associated with implementation of a strict policy forcontrolling spread of highly resistant microorganisms in France. BMJ Open. 2016;6(1):1-9. 7. Lepelletier D, Berthelot P, Lucet J, et al. French recommendations for the prevention of “emerging extensively drug-resistant bacteria ” (eXDR). J Hosp Infect. 2015;90(3):186195. 8. Bassetti M, Giacobbe DR, Giamarellou H, et al. Management of KPC-producing Klebsiella pneumoniae infections. Clin Microbiol Infect. 2018;24(2):133-144. 9. Tascini C, Sbrana F, Flammini S, et al. Oral gentamicin gut decontamination for prevention of KPC-producing Klebsiella pneumoniae infections: Relevance of concomitant systemic antibiotic therapy. Antimicrob Agents Chemother. 2014; 58(4):1972-1976. 10. Lübbert C, Faucheux S, Becker-Rux D, et al.

1688

11.

12.

13.

14.

15.

16.

17.

18.

19. 20.

Rapid emergence of secondary resistance togentamicin and colistin following selective digestive decontamination in patients with KPC-2- producing Klebsiella pneumoniae: A single-centre experience. Int J Antimicrob Agents. 2013;42(6):565-570. Rossen NG, MacDonald JK, De Vries EM, et al. Fecal microbiota transplantation as novel therapy in gastroenterology: A systematic review. World J Gastroenterol. 2015;21(17):5359- 5371. Manges AR, Steiner TS, Wright AJ. Fecal microbiota transplantation for the intestinal decolonization of extensively antimicrobialresistant opportunistic pathogens: A review. Infect Dis (Lond). 2016;48(8):587-592. Stalenhoef JE, Terveer EM, Knetsch CW, et al. Fecal microbiota transfer for multidrugresistant gram-negatives: A clinical success combined with microbiological failure. Open Forum Infect Dis. 2017;4(2):2-5. Bilinski J, Grzesiowski P, Muszynski J, et al. Fecal microbiota transplantation inhibits multidrug-resistant gut pathogens: preliminary report performed in an immunocompromised host. Arch Immunol Ther Exp (Warsz). 2016;64(3):255-258. de Castro CG Jr, Ganc AJ, Ganc RL, Petrolli MS, Hamerschlack N. Fecal microbiota transplant after hematopoietic SCT: report of a successful case. Bone Marrow Transplant. 2014;50(1):145. Kelly CR, Ihunnah C, Fischer M, et al. Fecal Microbiota Transplant for Treatment of Clostridium diffi cile Infection in Immunocompromised Patients. Am J Gastroenterol. 2014;109(7):1065-1071. Bilinski J, Grzesiowski P, Sorensen N, et al. Fecal microbiota transplantation in patients with blood disorders inhibits gut colonization with antibiotic-resistant bacteria: results of a prospective, single-center study. Clin Infect Dis. 2017; 65(3):364-370. Magiorakos A, Srinivasan A, Carey RB, et al. bacteria: an international expert proposal for interim standard definitions for acquired resistance. Clin Microbiol Infect. 2011;18(3):268- 281. WHO. Antimicrobial resistance. Global Report on Surveillance. Bull World Health Organ. 2014;61(3):383-394. Schaik W Van. The human gut resistome.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

Philos Trans R Soc Lond B Biol Sci. 2015;370(1670):20140087. Taur Y, Jenq RR, Perales M, et al. The effects of intestinal tract bacterial diversity on mortality following allogeneic hematopoietic stem cell transplantation. Blood. 2017;124 (7):1174-1183. Jenq RR, Taur Y, Devlin SM, et al. Intestinal Blautia Is Associated with Reduced Death from Graft-versus-Host Disease. Biol Blood Marrow Transplant. 2015;21(8):1373-1383. Austin M, Mellow M, Tierney WM. Fecal microbiota transplantation in the treatment of clostridium difficile infections. Am J Med. 2014;127(6):479-483. Webb BJ, Brunner A, Ford CD, Gazdik MA, Petersen FB. Fecal microbiota transplantation for recurrent Clostridium difficile infection in hematopoietic stem cell transplant recipients.Transpl Infect Dis. 2016;18(4): 628-633. Defilipp Z, Peled JU, Li S, et al. Third-party fecal microbiota transplantation following allo-HCT reconstitutes microbiome diversity. Blood Adv. 2018;2(7):745-753. Tang G, Yin W, Liu W. Is frozen fecal microbiota transplantation as effective as fresh fecal microbiota transplantation in patients with recurrent or refractory Clostridium difficile infection: A meta-analysis? Diagn Microbiol Infect Dis. 2017;88(4):322-329. Moss EL, Falconer SB, Tkachenko E, et al. Long-term taxonomic and functional divergence from donor bacterial strains following fecal microbiota transplantation in immunocompromised patients. PLos One. 2017;12 (8):e0182585. Jones JM, Wilson R, Bealmear PM. Mortality and gross pathology of secondary disease in 17 germfree mouse radiation chimeras. Radiat Res. 1971;45(3):577-588. Shono Y, Docampo MD, Peled JU, et al. Increased GVHD-related mortality with broadspectrum antibiotic use after allogeneic hematopoietic stem cell transplantation in human patients and mice Yusuke. Sci Transl Med. 2016;8(339):1-16. Kakihana K, Fujioka Y, Suda W, et al. Fecal microbiota transplantation for patients with steroid-resistant acute graft-versus-host disease of the gut. Blood. 2016;128(16):20832088.

haematologica | 2019; 104(8)

haematologica — Vol. 104 n. 8 — August 2019 — 1503-1688