Haematologica, Volume 103, Issue 11 by Haematologica

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

Ancient Greek

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

Scientific Latin

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

Scientific Latin

haematologicus (adjective) = related to blood

Modern English

haematologica (adjective, plural and neuter, used as a noun) = hematological subjects The oldest hematology journal, publishing the newest research results. 2017 JCR impact factor = 9.090

Haematologica, as the journal of the European Hematology Association (EHA), aims not only to serve the scientific community, but also to promote European cultural identify.

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

Information for readers, authors and subscribers Haematologica (print edition, pISSN 0390-6078, eISSN 1592-8721) publishes peer-reviewed papers on all areas of experimental and clinical hematology. The journal is owned by a non-profit organization, the Ferrata Storti Foundation, and serves the scientific community following the recommendations of the World Association of Medical Editors (www.wame.org) and the International Committee of Medical Journal Editors (www.icmje.org). Haematologica publishes editorials, research articles, review articles, guideline articles and letters. Manuscripts should be prepared according to our guidelines (www.haematologica.org/information-for-authors), and the Uniform Requirements for Manuscripts Submitted to Biomedical Journals, prepared by the International Committee of Medical Journal Editors (www.icmje.org). Manuscripts should be submitted online at http://www.haematologica.org/. Conflict of interests. According to the International Committee of Medical Journal Editors (http://www.icmje.org/#conflicts), “Public trust in the peer review process and the credibility of published articles depend in part on how well conflict of interest is handled during writing, peer review, and editorial decision making”. The ad hoc journal’s policy is reported in detail online (www.haematologica.org/content/policies). Transfer of Copyright and Permission to Reproduce Parts of Published Papers. Authors will grant copyright of their articles to the Ferrata Storti Foundation. No formal permission will be required to reproduce parts (tables or illustrations) of published papers, provided the source is quoted appropriately and reproduction has no commercial intent. Reproductions with commercial intent will require written permission and payment of royalties. Detailed information about subscriptions is available online at www.haematologica.org. Haematologica is an open access journal. Access to the online journal is free. Use of the Haematologica App (available on the App Store and on Google Play) is free. For subscriptions to the printed issue of the journal, please contact: Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, E-mail: info@haematologica.org). Rates of the International edition for the year 2018 are as following: Print edition

Institutional Euro 600

Personal Euro 150

Advertisements. Contact the Advertising Manager, Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, e-mail: marketing@haematologica.org). Disclaimer. Whilst every effort is made by the publishers and the editorial board to see that no inaccurate or misleading data, opinion or statement appears in this journal, they wish to make it clear that the data and opinions appearing in the articles or advertisements herein are the responsibility of the contributor or advisor concerned. Accordingly, the publisher, the editorial board and their respective employees, officers and agents accept no liability whatsoever for the consequences of any inaccurate or misleading data, opinion or statement. Whilst all due care is taken to ensure that drug doses and other quantities are presented accurately, readers are advised that new methods and techniques involving drug usage, and described within this journal, should only be followed in conjunction with the drug manufacturer’s own published literature. Direttore responsabile: Prof. Edoardo Ascari; Autorizzazione del Tribunale di Pavia n. 63 del 5 marzo 1955. Printing: Press Up, zona Via Cassia Km 36, 300 Zona Ind.le Settevene - 01036 Nepi (VT)

haematologica calendar of events

Journal of the European Hematology Association Published by the Ferrata Storti Foundation 13th Congress of the Albanian Association of Hematology Albanian Association of Hematology Chairs: A Ivanaj, M Dimopoulos, G Gaidano, X Pivot November 5-6, 2018 Tirana, Albania EBMT Severe Aplastic Anaemia and Autoimmune Diseases Working Parties - Joint Educational Course EBMT Chairs: C Dufour, J Snowden, RP de la Tour, A Tobias November 15-17, 2018 Firenze, Italy 7th National Symposium "Young Hematologist" "Hemostasis and Thrombosis" Bulgarian Medical Society of Hematology Chairs: V Kaleva, M Guenova November 16-17, 2018 Saints Constantine and Helena, Bulgaria The American Society of Hematology 60th ASH Annual Meeting and Exposition American Society of Hematology (ASH) Chairs: A Thompson, M Crowther, M Sekeres, J Crispino, M Sola-Visner December 1-4, 2018 San Diego, USA EHA-SLCH Hematology Tutorial on Myeloid Malignancies and MDS Chairs: G Ossenkoppele, V Gunawardena, HW Goonasekera February 8-9, 2019 Colombo, Sri Lanka

EHA-SWG Scientific Meeting on Immunotherapy: The Advent of CAR T Cell Therapy for Hematological Malignancies Chair: H Einsele, Co-chair: M Hudecek February 14-16, 2019 Amsterdam, The Netherlands EHA-AORK Hematology Tutorial on Lymphoma and Multiple Myeloma Chair: D Kaidarova, Co-chairs: S Gabbasova & B Afanasyev March 14-16, 2019 Almaty, Kazakhstan EHA Pediatric Course 2019 April 3-6, 2019 Sorrento, Italy EHA-ROHS-RHS Hematology Tutorial on Real World Challenges and Opportunities in Diagnostics and Management of Onco-Hematological Patients Today Chairs: I Poddubnaya, E Parovnichikova April 12-13, 2019 Moscow, Russia 24th Congress of EHA European Hematology Association June 13-16, 2019 Amsterdam, The Netherlands

Calendar of Events updated on October 11, 2018

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

Table of Contents Volume 103, Issue 11: November 2018 Cover Figure

Bone marrow smear from a patient with myelodysplastic syndrome showing a group of micromegakaryocytes. Courtesy of Prof. Rosangela Invernizzi.

Editorials 1753

Loss-of-function ferroportin disease: novel mechanistic insights and unanswered questions L. Tom Vlasveld and Dorine W. Swinkels

1756

To what extent can mathematical modeling inform the design of clinical trials? The example of safe dose reduction of tyrosine kinase inhibitors in responding patients with chronic myeloid leukemia Joshua T. Schiffer and Charles A. Schiffer

1758

Older adults with acute myeloid leukemia treated with intensive chemotherapy: “old” prognostic algorithms may not apply Richard M. Stone and Coleman Lindsley

Review Article 1760

Preclinical human models and emerging therapeutics for advanced systemic mastocytosis Michel Arock et al.

Guideline Article 1772

European Myeloma Network recommendations on tools for the diagnosis and monitoring of multiple myeloma: what to use and when Jo Caers et al.

Articles Hematopoiesis

1785

Alas1 is essential for neutrophil maturation in zebrafish Junwei Lian et al.

Iron Metabolism & its Disorders

1796

The SLC40A1 R178Q mutation is a recurrent cause of hemochromatosis and is associated with a novel pathogenic mechanism Chandran Ka et al.

Bone Marrow Failure

1806

Novel lineage depletion preserves autologous blood stem cells for gene therapy of Fanconi anemia complementation group A Jennifer E. Adair et al.

Phagocyte Biology & its Disorders

1815

Prognostic factors of Erdheim–Chester disease: a nationwide survey in Japan Takashi Toya et al.

Chronic Myeloid Leukemia

1825

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 Artur C. Fassoni et al.

1835

Treatment-free remission after two-year consolidation therapy with nilotinib in patients with chronic myeloid leukemia: STAT2 trial in Japan Naoto Takahashi et al.

Acute Myeloid Leukemia

1843

Inhibition of protein disulfide isomerase induces differentiation of acute myeloid leukemia cells Justyna Chlebowska-Tuz et al.

Haematologica 2018; vol. 103 no. 11 - November 2018 http://www.haematologica.org/

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation 1853

Genetics of acute myeloid leukemia in the elderly: mutation spectrum and clinical impact in intensively treated patients aged 75 years or older Victoria V. Prassek et al.

1862

MDM2- and FLT3-inhibitors in the treatment of FLT3-ITD acute myeloid leukemia, specificity and efficacy of NVP-HDM201 and midostaurin Katja Seipel et al.

Acute Lymphoblastic Leukemia

1873

Radiation exposure from computerized tomography and risk of childhood leukemia: Finnish register-based case-control study of childhood leukemia (FRECCLE) Atte NikkilĂ¤ et al.

Chronic Lymphocytic Leukemia

1881

Adherence to the Western, Prudent, and Mediterranean dietary patterns and chronic lymphocytic leukemia in the MCC-Spain study Marta Solans et al.

1889

Safety of obinutuzumab alone or combined with chemotherapy for previously untreated or relapsed/refractory chronic lymphocytic leukemia in the phase IIIb GREEN study VĂŠronique Leblond et al.

Non-Hodgkin-Lymphoma

1899

Inferior survival in high-grade B-cell lymphoma with MYC and BCL2 and/or BCL6 rearrangements is not associated with MYC/IG gene rearrangements Ellen D. McPhail et al.

1907

PD-L1+ tumor-associated macrophages and PD-1+ tumor-infiltrating lymphocytes predict survival in primary testicular lymphoma Marjukka Pollari et al.

Stem Cell Transplantation

1915

Coagulation & its Disorders

1925

N-linked glycosylation modulates the immunogenicity of recombinant human factor VIII in hemophilia A mice Jesse D. Lai et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/103/11.toc

e496

L-leucine increases translation of RPS14 and LARP1 in erythroblasts from del(5q) myelodysplastic syndrome patients Erica Bello et al. http://www.haematologica.org/content/103/11/e496

e501

The long non-coding RNA landscape in juvenile myelomonocytic leukemia Mattias Hofmans et al. http://www.haematologica.org/content/103/11/e501

e505

Loss of DEP-1 (Ptprj) promotes myeloproliferative disease in FLT3-ITD acute myeloid leukemia Anne Kresinsky et al. http://www.haematologica.org/content/103/11/e505

e510

Concomitant WT1 mutations predict poor prognosis in acute myeloid leukemia patients with double mutant CEBPA Feng-Ming Tien et al. http://www.haematologica.org/content/103/11/e510

Haematologica 2018; vol. 103 no. 11 - November 2018 http://www.haematologica.org/

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation e514

Phase 3 results for vosaroxin/cytarabine in the subset of patients ≥60 years old with refractory/early relapsed acute myeloid leukemia Farhad Ravandi et al. http://www.haematologica.org/content/103/11/e514

e519

Reducing mortality in newly diagnosed standard-risk acute promyelocytic leukemia in elderly patients treated with arsenic trioxide requires major reduction of chemotherapy: a report by the French Belgian Swiss APL group (APL 2006 trial) Ramy Rahmé et al. http://www.haematologica.org/content/103/11/e519

e522

Transcriptional activities of DUX4 fusions in B-cell acute lymphoblastic leukemia Yosuke Tanaka et al. http://www.haematologica.org/content/103/11/e522

e527

Pentraxin-3 polymorphisms and invasive mold infections in acute leukemia patients receiving intensive chemotherapy Anne-Sophie Brunel et al. http://www.haematologica.org/content/103/11/e527

e531

A reversible carnitine palmitoyltransferase (CPT1) inhibitor offsets the proliferation of chronic lymphocytic leukemia cells Elena Gugiatti et al. http://www.haematologica.org/content/103/11/e531

e537

In vitro and in vivo activity of a new small-molecule inhibitor of HDAC6 in mantle cell lymphoma Montserrat Pérez-Salvia et al. http://www.haematologica.org/content/103/11/e537

e541

Lenalidomide plus bendamustine-rituximab does not overcome the adverse impact of TP53 mutations in mantle cell lymphoma Christian Winther Eskelund et al. http://www.haematologica.org/content/103/11/e541

e544

Mutational screening of newly diagnosed multiple myeloma patients by deep targeted sequencing Yanira Ruiz-Heredia et al. http://www.haematologica.org/content/103/11/e544

Case Reports Case Reports are available online only at www.haematologica.org/content/103/11.toc

e549

A double Philadelphia chromosome-positive chronic myeloid leukemia patient, co-expressing P210BCR-ABL1 and P195BCR-ABL1 isoforms Raquel Vinhas et al. http://www.haematologica.org/content/103/11/e549

e553

Clonally related diffuse large B-cell lymphoma and interdigitating dendritic cell sarcoma sharing MYC translocation Yotaro Ochi et al. http://www.haematologica.org/content/103/11/e553

e557

Comprehensive molecular characterization of a heavy chain deposition disease case Sébastien Bender et al. http://www.haematologica.org/content/103/11/e557

Haematologica 2018; vol. 103 no. 11 - November 2018 http://www.haematologica.org/

EDITORIALS Loss-of-function ferroportin disease: novel mechanistic insights and unanswered questions L. Tom Vlasveld1 and Dorine W. Swinkels2 1

Department of Internal Medicine, Haaglanden Medical Center, Location Bronovo, The Hague and 2Translational Metabolic Laboratory, Department of Laboratory Medicine, Radboud University Medical Center, Nijmegen, the Netherlands E-mail: dorine.swinkels@radboudumc.nl doi:10.3324/haematol.2018.203315

erroportin, a 571 amino acid cation transporter encoded by the SLC40A1 gene, is the only known human cellular iron exporter and primarily expressed in the basolateral membrane of duodenal enterocytes, macrophages, and hepatocytes.1,2 By regulating ferroportin-mediated iron export in these cells, the hepatocyte-derived peptide hormone hepcidin plays a central role in intestinal iron resorption, macrophage iron recycling and hepatic iron storage (Figure 1A).3 Functional studies reveal the internalization of ferroportin upon hepcidin binding with subsequent ubiquitination of the protein resulting in the diminution of cellular iron export. The mechanism of the inhibition of cellular iron efflux is not fully elucidated.4 Genetic ferroportin variants result in autosomal dominantly inherited hereditary hemochromatosis (HH) type 4 or ferroportin disease which is traditionally divided into two entities primarily based on the pattern of cellular iron distribution. Classical ferroportin disease (type 4A) is characterized by macrophage iron retention and decreased circulating iron availability for erythropoiesis, and is clinically recognized by the presence of elevated serum ferritin concentrations with low to normal transferrin saturation (TSAT) and poor tolerance to phlebotomy. Non-classical (atypical, type 4B or ferroportinassociated HH) ferroportin disease is characterized by parenchymal (hepatocellular) iron overload with both elevated serum ferritin and TSAT and is indistinguishable from other forms of HH.5 To assess the pathogenicity of a ferroportin variant functional studies on iron export and/or ferroportin expression are performed in cell-lines (mostly HEK293T, a human embryonic kidney cell line) transfected with the variant. However, interpretation of results may be hampered by these studies being done in non-enterocyte and non-macrophage cell lines. Moreover, comparability between reports may have shortcomings, since iron loading and hepcidin exposure protocols, the mode of iron export measurements and the determination of ferroportin expression differ largely between studies. Nevertheless, the results obtained are quite consistent, in that patients with the type 4A phenotype are mostly associated with loss-of-function ferroportin variants which display decreased membrane expression and/or iron transport in functional assays, while patients with the type 4B phenotype have gain-of-function variants with preserved membrane expression and iron export capacity even after hepcidin exposure. Notwithstanding a different pattern of cellular iron distribution, a review of numerous case series of ferroportin disease reveals that in those patients who underwent magnetic resonance imaging (MRI) or liver biopsy, hepatic iron content is increased for both types of ferroportin disease. These observations indicate that loss-of-function variants also result in body iron overload. Variety in iron parameters reported for patients with ferroportin variants is large and the correlation between iron export capacity in functional studies and clinical phenotype is poor, haematologica | 2018; 103(11)

especially on an individual level. The establishment of the pathogenicity of ferroportin variants may not only be impeded by the presence of iron homeostasis modulating co-morbidities such as alcohol consumption and liver steatosis, but also by the limited applicability of in silico prediction models in the absence of a fully elucidated three-dimensional (3D) structure of ferroportin. In the most widely accepted secondary structure, ferroportin comprises 12 helices located in 12 transmembrane (TM) domains bound via six extracellular (ES) and five intracellular (IS) segments with a large intracellular loop between the 6th and 7th transmembrane helix and an intracellularly located N- and C-terminus. The available 3D models are based on a comparison with membrane transport proteins from a wide range of other species with only a 10 - 24% sequence homology and a maximal 40% similarity. Two studies, using E. coli glycerol-3-phosphate transporter GlpT and lactose permease LacY or the Bdellovibrio bacteriovorus Bd2019 iron transporter as template, respectively, reveal an open inward and an open outward structure with an intra- and extracellular gate between the 6th and 7th transmembrane domain.6,7 Site mutagenetic and conformational studies revealed the residues, located at IS1, IS2 and IS5, that are important in intracellular gate interaction and the residues, located at TM1, TM12 and ES1 and ES4, that are involved in extracellular gate interaction. These studies also identified ferroportin residues that are essential for binding and ubiquitination of hepcidin and may be involved in iron binding and egress. The unequivocally hepcidin-resistant gain-of-function ferroportin variants Cys326Ser, Tyr501Cys, Asp504Asn and Tyr64Asn and His507Arg are reported to have impaired hepcidin binding and hepcidindependent ubiquitination, respectively. In the open outward structure, variants causing impaired hepcidin binding were found to be localized within the extracellular gate, while variants causing impaired ubiquitination were found in the periphery of the molecule, suggesting that these latter variants interfere with appropriate folding after hepcidin binding.8 These findings provide a molecular basis for the observed cellular distribution pattern of the type 4B iron overload in patients and the behavior in functional tests performed for these hepcidinresistant gain-of-function variants (Figure 1B). The study by Ka et al., published in this issue, provides interesting insights into the pathophysiologic mechanisms involved in ferroportin disease caused by loss-of-function variants. They describe 22 patients from six independent families with hyperferritinemia, a normal TSAT and the heterozygous presence of the Arg178Gln variant. The serum hepcidin levels determined in two patients were above the reference range. The hepatic iron content, estimated by MRI methodology in three of the patients, was mildly elevated and a liver biopsy, performed in one patient, revealed predominant iron deposition in Kupffer cells. Although Arg178Gln displayed a reduced export of 55Fe out of transfected HEK293T cells, the variant was properly localized on the cellular membrane with disappearance after 1753

Editorials

hepcidin exposure. Based on both the clinical phenotype and functional tests the variant was assigned as a hepcidin sensitive loss-of-function variant. In the open outward conformation of the 3D structure, using Bdellovibrio bacteriovorus Bd2019 as the template, the demonstrated non-covalent interaction between Arg178 and Asp473 (located on the Nand C-lobe, respectively) was assumed to be involved in the stabilization of the open outward conformation needed to preserve iron egress. Indeed, the Asp473Ala ferroportin

variant was also properly expressed on the membrane with a nearly total loss of iron export capacity. In most loss-offunction variants the abolished iron transport is attributed to defective expression of ferroportin on the membrane. Ka et al., however, provide evidence for interference in the stabilization of the conformation state, i.e., the open outward state, in the Arg178Gln variant as an alternative mechanism for diminished iron transport, despite the proper membrane expression of this variant. Further exploration of the amino-

Figure 1. Model displaying cellular iron flows in patients with loss-of-function ferroportin variants as compared to physiological conditions and patients with gainof-function ferroportin variants and IRIDA. Ferroportin activity in the enterocyte, macrophage and hepatocyte in physiological steady state conditions (A), hepcidinresistant gain-of-function (GOF) (B), iron refractory iron deficiency anemia (IRIDA) (C), loss-of-function (LOF) with mislocalisation on the membrane (D), membrane expressed destabilized and hepcidin sensitive LOF (E), membrane expressed destabilized LOF (F). Physiologically, iron availability for erythropoiesis (in the form of transferrin bound iron ( ) is systemically regulated by the inhibitory effects of the hepatocyte-derived hormone hepcidin ( ) on the activity of ferroportin ( ). This regulatory system contributes to the regulation of intracellular iron ( ) and ferritin bound iron ( ) levels, serum iron ( ) concentration, transferrin saturation (TSAT) ( ) and serum ferritin ( ) concentration. In hepcidin-resistant GOF, there is an increased non-impressionable iron export out of macrophages with continuous enterocyte iron transport with iron deposition in hepatocytes. In IRIDA, ferroportin action is inhibited by (for body iron status) inappropriately elevated hepcidin leading to decreased enterocyte iron resorption and diminished ferric-transferrin availability for the erythroblast with the sequestering of some iron, derived from erythrophagocytosis, within the macrophage. In patients with LOF variants there is decreased iron transport out of the macrophage leading to iron sequestration in these cells, but the mechanism of the apparently relatively increased enterocyte iron export with subsequent body iron overload is not yet fully elucidated. Clinically observed changes in serum iron, and ferritin concentration and TSAT in the various conditions are depicted by the size of the rectangles in the lower right-hand side of the panels. DMT: divalent metal transporter; tfr: transferrin receptor-1; fe: iron. Transferrin ( ).

1754

haematologica | 2018; 103(11)

Editorials

acid interaction network in various ferroportin conformation states may reveal additional sites essential for stabilization and provide molecular explanation for defective iron egress in variants without defective membrane expression, such as Arg88Gly, Leu129Pro, Ile152Phe and Asn174Ile.9-11 Notwithstanding the importance of this finding, it does not clarify the iron overload observed in these patients. Moreover, as reported by Ka and colleagues, it is notable that the patients with a loss-of-function variant ferroportin, that completely disappears from the cellular membrane after exposure to hepcidin, have no anemia and intensive phlebotomy regimens were reported to be well tolerated. In patients with defective iron export transport due to a lossof-function variant, either due to mislocalization or improper folding, one might expect diminished enterocyte iron absorption leading to a phenotype similar to that described for iron-refractory iron deficiency (IRIDA). In IRIDA, mutations in the TMPRSS6 gene result in inappropriately elevated serum hepcidin levels with diminished ferroportin activity and a variable degree of iron deficiency anemia, unresponsive to oral iron treatment, and intra-enterocyte iron retention in experimental animals (Figure 1C).12,13 Experimental animals with diminished ferroportin activity due to monoallelic wild-type ferroportin expression also display iron deficiency anemia.2 This is in contrast to the situation in patients with diminished ferroportin activity due to loss-of-function ferroportin variants which is characterized by iron overload, normal circulating hemoglobin concentrations and absence of intra-enterocyte iron (evaluated in a limited number of patients) (Figure 1D,E).14 It appears that in the enterocyte diminished ferroportin activity due to decreased activity of wild-type ferroportin has a different effect on iron export capacity than decreased ferroportin activity caused by a loss-of-function variant. In addition, the impact of loss-of-function variants on iron export capacity may be different between macrophages and enterocytes, and the results of the functional studies in HEK293T cells are more predictive for macrophage iron handling. It has been suggested that in patients with lossof-function variants the remaining monoallelic expressed wild-type ferroportin protein would be sufficient to preserve iron export in cells with low iron turnover, such as enterocytes, but insufficient to maintain iron export capacity in cells with high iron turnover, such as macrophages, with subsequent intracellular iron retention in these cells.5 However, this theory may not provide a full explanation for the state of iron overload in these patients. While the activity of ferroportin in both enterocytes and macrophages in the systemic regulation of iron homeostasis is mediated by hepcidin, there is a considerable amount of experimental data indicating that the regulation of ferroportin expression and activity differ between enterocytes and macrophages. These data include differences in intracellular regulatory mechanisms that fine-tune ferroportin membrane expression as well as differences in ferroportin

haematologica | 2018; 103(11)

activity upon various systemic stimuli.15-20 The study by Ka et al. contributes to our understanding of the pathogenic mechanisms involved in decreased ferroportin activity by loss-of-function variants. Studies focused on the consequences of these variants on enterocyte iron handling are warranted to further comprehend the pathophysiology of this intriguing iron overload disorder.

References 1. Abboud S, Haile DJ. A novel mammalian iron-regulated protein involved in intracellular iron metabolism. J Biol Chem. 2000;275 (26):19906-19912. 2. Donovan A, Lima CA, Pinkus JL, et al. The iron exporter ferroportin/Slc40a1 is essential for iron homeostasis. Cell Metab. 2005;1(3):191-200. 3. Ganz T, Nemeth E. Hepcidin and iron homeostasis. Biochim Biophys Acta. 2012;1823(9):1434-1443. 4. 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. 5. Pietrangelo A. The ferroportin disease: pathogenesis, diagnosis and treatment. Haematologica. 2017;102(12):1972-1984. 6. Bonaccorsi di Patti MC, Polticelli F, Cece G, et al. A structural model of human ferroportin and of its iron binding site. FEBS J. 2014;281 (12):2851-2860. 7. Taniguchi R, Kato HE, Font J, et al. Outward- and inward-facing structures of a putative bacterial transition-metal transporter with homology to ferroportin. Nat Commun. 2015;6:8545. 8. Aschemeyer S, Qiao B, Stefanova D, et al. Structure-function analysis of ferroportin defines the binding site and an alternative mechanism of action of hepcidin. Blood. 2018;131(8):899-910. 9. Moreno-Carralero MI, Munoz-Munoz JA, Cuadrado-Grande N, et al. A novel mutation in the SLC40A1 gene associated with reduced iron export in vitro. Am J Hematol. 2014;89(7):689-694. 10. Le Gac G, Ka C, Joubrel R, et al. Structure-function analysis of the human ferroportin iron exporter (SLC40A1): effect of hemochromatosis type 4 disease mutations and identification of critical residues. Hum Mutat. 2013;34(10):1371-1380. 11. Callebaut I, Joubrel R, Pissard S, et al. Comprehensive functional annotation of 18 missense mutations found in suspected hemochromatosis type 4 patients. Hum Mol Genet. 2014;23(17):4479-4490. 12. De Falco L, Sanchez M, Silvestri L, et al. Iron refractory iron deficiency anemia. Haematologica. 2013;98(6):845-853. 13. Folgueras AR, de Lara FM, Pendas AM, et al. Membrane-bound serine protease matriptase-2 (Tmprss6) is an essential regulator of iron homeostasis. Blood. 2008;112(6):2539-2545. 14. Corradini E, Montosi G, Ferrara F, et al. Lack of enterocyte iron accumulation in the ferroportin disease. Blood Cells Mol Dis. 2005;35(3):315-318. 15. Zhang DL, Ghosh MC, Rouault TA. The physiological functions of iron regulatory proteins in iron homeostasis - an update. Front Pharmacol. 2014;5:124. 16. Drakesmith H, Nemeth E, Ganz T. Ironing out ferroportin. Cell Metab. 2015;22(5):777-787. 17. Canonne-Hergaux F, Donovan A, Delaby C, et al. Comparative studies of duodenal and macrophage ferroportin proteins. Am J Physiol Gastrointest Liver Physiol. 2006;290(1):G156-163. 18. 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. 19. Jacolot S, Ferec C, Mura C. Iron responses in hepatic, intestinal and macrophage/monocyte cell lines under different culture conditions. Blood Cells Mol Dis. 2008;41(1):100-108. 20. Chaston T, Chung B, Mascarenhas M, et al. Evidence for differential effects of hepcidin in macrophages and intestinal epithelial cells. Gut. 2008;57(3):374-382.

1755

Editorials

Vaccine and Infectious Diseases and Clinical Research Divisions, Fred Hutchinson Cancer Research Center, Seattle, WA; 2Department of Medicine, University of Washington, Seattle, WA and 3Department of Oncology Wayne State University School of Medicine, Karmanos Cancer Institute, Detroit, MI, USA E-mail: schiffer@karmanos.org doi:10.3324/haematol.2018.201897

he development of the tyrosine kinase inhibitors (TKIs) that inhibit the BCR/ABL oncoprotein driving the growth and persistence of chronic myeloid leukemia (CML) is one of the most remarkable advances in anti-cancer treatment in recent decades, and has served as the model, if not the Platonic ideal, for targeted therapies of other cancers. It remains standard practice to administer TKIs indefinitely because of concern about relapse should compliance be erratic or therapy stopped. However, a number of recent trials have demonstrated that approximately 50% of patients whose transcript levels were either extremely low or undetectable for at least 2-3 years using sensitive polymerase chain reaction (PCR) assays, have not relapsed after therapy was stopped with many patients relapse free for more than five years after discontinuation.1,2 Remarkably, almost all relapses occurred within the first 68 months after therapy cessation. The mechanism(s) by which CML remains dormant is not known, although immunological explanations are proposed, and the only predictors of continued remission are longer durations of TKI therapy and PCR negativity. The benefits of these so-called “treatment-free remissions”3 are obvious, and include substantially reduced costs (CML-directed TKIs also served as the prototype for the obscene pricing of “targeted” agents),4 decreases in the potential for serious longer term organ toxicities (fortunately very rare with imatinib5), elimination of the often bothersome, low-grade chronic side-effects such as fatigue and diarrhea, as well as the ability to plan for pregnancies in younger patients. But, it is estimated that less than 20% of patients who are started on TKI treatment will be able to successfully discontinue therapy. It is likely, however, that some of these same benefits would be achieved if it were possible to reduce the dose of the TKIs without loss of disease control. Indeed, the selection of doses and schedule for many drugs is often based on very small, under-powered trials with relatively few subsequent attempts to determine whether lower doses might result in similar outcomes. In this regard, it was recently shown in a large phase II trial in newly diagnosed patients in chronic phase that a starting dose of 50 mg of dasatinib seems to produce the same response rate as the “standard” dose of 100 mg with perhaps fewer side-effects.6,7 In this issue of the Journal, Fassoni et al. propose what they consider to be a safe strategy of dose reduction of TKIs for CML patients in chronic phase based on mathematical models generated from large recently generated clinical trial data.8 The authors evaluated patients who were in major molecular response (MMR) for at least one year, defined as 1756

a greater than 3 log reduction in transcripts from baseline, and who had received imatinib therapy for more than three years. Their model concludes that, for most patients, halving the dose of the TKI will maintain the current level of response, albeit with a high likelihood of a temporary increase in transcript levels which return to baseline after continued treatment with the reduced dose. Fassoni et al.’s paper8 highlights the crucial role that mathematical models can play in providing a mechanistic underpinning of observed data, and in influencing therapeutic strategies. As a general principle, differential equationbased models are a necessary tool to capture non-linearity in biomedical datasets. CML treatment involves the coupling of mechanisms governing tumor cell division, transition rates between progenitor cells and terminally differentiated tumor cells, anti-tumor immune responses and the selective pressure of TKIs. The non-linearity inherent in a model which captures the dynamics of this system can yield counterintuitive and useful predictions, such as Fassoni et al.’s calculation that the increase in BCR/ABL levels that may occur following TKI dose reduction is transitory.8 Fassoni et al. hypothesize that an increase in transcripts is due to a self-limited increase in proliferating leukemia stem cells (LSCs), rather than mutations leading to TKI resistance or changes in the underlying dynamics of quiescent, non-proliferating LSCs.8 Mathematical modeling as a field would benefit from some demystification. Well-designed models represent nothing more than in silico experiments where models with competing structures and mechanistic assumptions are compared for their ability to fit to observed data. Fassoni et al.’s model recapitulates data from selected patients from the IRIS and CML IV trials in whom the pattern of second phase decrease in transcripts is thought to result from slow depletion of quiescent LSCs. This depletion is predicted to occur independently of TKI dose reduction because the transition from quiescent to proliferating LSC, rather than incomplete efficacy of TKIs, is the rate-limiting step for CML eradication during MMR. The model’s generalizability to patients with an adequate treatment response is further supported by the scalability of its key parameters governing transitions between quiescent and proliferating LSCs across all patients. Finally, the model’s main prediction, that TKI dose reduction will not usually lead to loss of MMR, is qualitatively consistent with interim analyses from the DESTINY trial, a study in which 174 participants underwent 50% dose reduction as the initial phase of a “stopping” trial.9 It is not known whether the transcript increases seen in some of these patients at the reduced dose would haematologica | 2018; 103(11)

Editorials

have resolved, because the protocol called for restarting therapy if MMR was lost. Thus, to implement their recommendation in a subsequent clinical trial, clinicians would have to resist the current dogma to resume the higher dose or to switch to another TKI should the levels increase. How then should the model’s enticing predictions be interpreted? Similar to in vitro models or animal models which aim to recapitulate key features of human disease, the primary role of mathematical models is to generate a hypothesis for a definitive test of concept studies in humans, rather than providing “proof” of the concept. Accordingly, mathematical model simulations of clinical trials can critically inform the design of clinical trials themselves, but should never replace them. Indeed, the fact that Fassoni et al.’s model neglects several potential reasons for treatment failure in patients following dose reduction, such as de novo loss of immunological control or TKI resistance, highlights the point that the model’s novel hypotheses need to be tested. Fassoni et al.’s simulations validate the concept that effective TKI dose reduction is biologically plausible. Moreover, the output of the model provides useful information for TKI dose selection and sampling frequency in subsequent clinical trials. Model output suggests that the amount of TKI dose reduction may be individualized, based partly on observed BCR/ABL kinetics during the initial phase of therapy, which can then be used to derive two key parameters: 1) leukemic stem cell activation; and 2) TKI efficacy. Estimation of these parameters will require frequent BCR/ABL measurements during both primary and second phases of TKI therapy, an important consideration for trial design. Most importantly, the modeling provides informed criteria for assessing treatment failure after dose reduction, and specifically recommends that investigators should not immediately classify increases in BCR-ABL transcripts as treatment failures. Rather, the model suggests that increases in BCR-ABL ratio could be permitted for as long as a year, at which point second phase decay would be expected to occur. Again, sampling following dose reduction must be frequent enough to allow model fitting that precisely characterizes this phenomenon. A third critical parameter, the LSC proliferation rate, can only be estimated following TKI dose reduction and could theoretically be leveraged to

haematologica | 2018; 103(11)

make dose adjustments in real time. Finally, because the model predicts effect size, it could also theoretically be used to inform power calculations to project an adequate sample size for ‘proof of concept’ trials. To understand the importance of these findings, one only need consider the situation in which a priori mathematical modeling is not performed, and TKI dose reduction is formally tested and labeled a failure based on what could have been only a temporary increase in BCR/ABL transcript levels. This would represent a spurious rejection of an important and valid scientific hypothesis, as well as a waste of resources. It is likely that these types of errors are not uncommon in clinical trial design, and that many could be predicted a priori with the use of strategic mathematical modeling.

References 1. Mahon FX, Réa D, Guilhot J, et al. Discontinuation of imatinib in patients with chronic myeloid leukaemia who have maintained complete molecular remission for at least 2 years: the prospective, multicentre Stop Imatinib (STIM) trial. Lancet Oncol. 2010;11(11):10291035. 2. 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):515-522. 3. Hughes TP, Ross DM. Moving treatment-free remission into mainstream clinical practice in CML. Blood. 2016;128(1):17-23. 4. 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. 5. Gambacorti-Passerini C, Antolini L, Mahon FX, et al. Multicenter independent assessment of outcomes in chronic myeloid leukemia patients treated with imatinib. J Natl Cancer Inst. 2011;103(7):553-561. 6. Naqvi K, Jabbour E, Skinner J, et al. Early results of lower dose dasatinib (50 mg daily) as frontline therapy for newly diagnosed chronic-phase chronic myeloid leukemia. Cancer. 2018;124(13):2740-2747. 7. Schiffer CA. The evolution of dasatinib dosage over the years and its relevance to other anticancer medications. Cancer. 2018;124(13):26872689. 8. 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. Hematologica. 2018 Jun 28. [Epub ahead of print] 9. 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.

1757

Editorials

Older adults with acute myeloid leukemia treated with intensive chemotherapy: “old” prognostic algorithms may not apply Richard M. Stone and Coleman Lindsley Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, USA E-mail: rstone@partners.org doi:10.3324/haematol.2018.201848

he notion that older adults with acute myeloid leukemia (AML) represent a major therapeutic challenge due to host biological factors and intrinsic disease features leading to poor outcomes is accepted by consensus. Older adults have diminished stem cell reserve, more frequent comorbid medical conditions, and decreased ability to excrete chemotherapy compared to younger patients, in part leading to a relatively increased rate of mortality from induction treatment.1 Moreover, AML that arises in the older population can exhibit features of intrinsic chemo-resistance exemplified by a higher expression of proteins that extrude chemotherapy, a higher proportion of adverse cytogenetic abnormalities,2 and a higher frequency of gene mutations associated with antecedent myeloid disease, which are, in turn, associated with an elevated rate of disease resistance.3 Based on this unfavorable balance of response and toxicity, many believe that older patients, especially those over the age of 75, should as a whole be treated with less intensive therapies: usually a hypomethylating agent (azacytidine or decitabine) in the USA or lowdose cytarabine.4 In counterpoint, a population-based study from Sweden suggested that older adults may benefit from intensive treatment, while others have shown that the genetic characteristics5 of AML in older patients may discriminate subgroups with distinct clinical responses to intensive therapy.3 For example, patients with de novo AML genetic alterations have a higher likelihood of responding to induction therapy, while both ‘secondary-type’ AML mutations (those associated with AML arising after a prodromic myelodysplastic syndrome) and TP53 mutations are associated with more resistant disease.3 The challenge facing the field is how to incorporate contrasting data from large studies of heterogeneous patient populations that may lack sufficient data with regards to genetic variables and smaller studies with detailed clinical information and genetic annotation, but limited statistical power to identify independent genetic effects. The debate about ‘3+7’ chemotherapy versus non-intensive therapy for older adults with AML continues.6 How do the data provided by the German Austrian AML Study Group7 add to the discussion? Their study has several important strengths. First, only patients older than the age of 74 were included, enabling a distinct and focused evaluation of this under-studied population. Second, all patients received the same intensive induction chemotherapy on an AML Cooperative Group protocol,8 which compared standard induction chemotherapy with thioguanine, daunorubicin, and cytarabine (TAD9) with a high-dose cytarabine/mitoxantrone induction. Patients whose bone marrow samples showed residual AML on day 15 were to receive high-dose cytarabine/mitoxantrone. The doses of chemotherapy were attenuated in older adults. Third, the 1758

authors performed targeted, next-generation sequencing of 64 genes known to be mutated in AML, paired with focused assessment of FLT3-ITD, NPM1, and CEBPA mutations by standard laboratory-validated methodologies. They looked at outcomes according to number and type of mutations along with other potentially relevant data such as patient’s age, cytogenetics, risk group by one of two classifications systems, and randomized treatment assignment. More details about why these older adults received chemotherapy as opposed to other therapies would have been welcome. Were they particularly fit? Some actually had a performance status worse than 2, which is an adverse prognostic factor for response to chemotherapy.2 Obviously, no comparative group of patients treated contemporaneously with less intensive chemotherapy can be provided. The analysis highlights that some commonly employed prognostic factors in younger adult AML patients, including FLT3-ITD and NPM1, did not have the same relevance in the older patients. The European LeukemiaNet9 and Medical Research Council10 prognostic algorithms simply were not developed based on data from intensively treated patients in this age group. Given the distinct clinical and molecular features in this age group, one can begin to grasp why the ‘usual prognostic rules’ noted in younger patients may not apply. The findings in this cohort of 151 patients aged over 74 years old confirm those of published studies showing that there is a preponderance of gene mutations associated with myelodysplastic syndromes and clonal hematopoiesis of indeterminate potential in AML in older individuals.3,11,12 In a previous study, outcomes after 3+7based chemotherapy were particularly poor in patients with histologically confirmed secondary AML or de novo AML harboring myelodysplastic syndrome-associated gene mutations.3 Although few patients in this study were older tha 74, it raised the idea that gene mutations suggestive of antecedent myelodysplastic syndromes (SRSF2, SF3B1, U2AF1, ZRSR2, BCOR, EZH2, ASXL1, STAG2) may highlight a subgroup of older patients with more chemoresistant disease. By contrast, older patients without such mutations had a much more favorable induction success rate and a 50% event-free survival.3 In the cohort reported by Prassek et al.,7 35% of the patients died during induction. This finding has important clinical implications, as older patients among a group of seemingly fit-for-chemotherapy patients may experience elevated treatment-related mortality due to a poor stem cell reserve, subtle organ dysfunction, and increased co-morbid disease, which may sometimes overwhelm the effect of disease biology. This finding offers important context for the interpretation of genetic results. In AML, the most prominent and reproducible associations between specific haematologica | 2018; 103(11)

Editorials

gene mutations (FLT3, TP53, CEBPA, NPM1) and overall survival are related to the mutations’ differential impact on relapse risk.13 In fact, there are limited data available regarding the potential association between disease genetics and induction mortality. Notably, ten out of 13 patients with IDH1 mutations experienced early death during induction, while just two had refractory disease. This detail is central to the interpretation of the results and to contemplating clinical actionability. Do IDH1 mutations drive specific chemoresistance in older patients, thereby suggesting augmentation of induction with IDH1-directed therapy? Are IDH1 mutations causally or non-causally associated with clinical features that have been linked to increased transplant-related mortality?14 A combination of the two? In younger adults the prognostic impact of IDH mutations is unclear. One study suggested that those with both NPM1 and IDH mutations enjoyed a high event-free survival rate when treated with aggressive chemotherapy,15 while others showed a neutral or negative impact.16,17 The simplest explanation for the disparate literature regarding IDH1 mutations is that related to the random noise in small samples. Alternatively, the clinical impact of IDH mutations may be ‘context-dependent’, and different depending on its place in the clonal hierarchy (progression mutation in some cases and a founder or ‘early’ mutation in others), the co-occurring gene mutations, or undefined clinical characteristics. Does this group of patients in their 70’s and 80’s have IDH1 mutations that reflect a distinct group with particularly chemoresistant disease? If so, one could speculate that such individuals might have fared better with an IDH inhibitor than with chemotherapy. Indeed the single-agent studies with these drugs,18,19 mainly conducted in patients with advanced disease, include some previously untreated patients with comparable outcomes. In summary, the German Austrian AML Study Group research is important because it confirms the commonality of secondary or stem cell type mutations in adults aged over 74 with AML. It further reminds us that prognostic characteristics cannot necessarily be applied with abandon across the age spectrum. However, we remain in the same conundrum about the utility of aggressive versus less intensive therapy in this age group.

References 1. Klepin HD, Geiger AM, Tooze JA, et al. Geriatric assessment predicts survival for older adults receiving induction chemotherapy for acute myelogenous leukemia. Blood. 2013;121(21):4287-4294. 2. Leith CP, Kopecky KJ, Godwin J, et al. Acute myeloid leukemia in the

haematologica | 2018; 103(11)

3. 4.

5. 6. 7.

9. 10.

11. 12. 13. 14.

15. 16. 17.

18. 19.

elderly: assessment of multidrug resistance (MDR1) and cytogenetics distinguishes biologic subgroups with remarkably distinct responses to standard chemotherapy. A Southwest Oncology Group study. Blood. 1997;89(9):3323-3329. Lindsley RC, Mar BG, Mazzola E, et al. Acute myeloid leukemia ontogeny is defined by distinct somatic mutations. Blood. 2015;125(9):1367-1376. Ferrara F, Barosi G, Venditti A, et al. Consensus-based definition of unfitness to intensive and non-intensive chemotherapy in acute myeloid leukemia: a project of SIE, SIES and GITMO group on a new tool for therapy decision making. Leukemia. 2013;27(5):997-999. Juliusson G, Antunovic P, Derolf A, et al. Age and acute myeloid leukemia: real world data on decision to treat and outcomes from the Swedish Acute Leukemia Registry. Blood. 2009;113(18):4179-4187. Ossenkoppele G, Lowenberg B. How I treat the older patient with acute myeloid leukemia. Blood. 2015;125(5):767-774. Prassek VV, Rothenberg-Thurley M, Sauerland MC, et al. Genetics of acute myeloid leukemia in the elderly: mutation spectrum and clinical impact in intensively treated patients aged 75 years or older. Haematologica. 2018;103(11):1853-1861. Buchner T, Berdel WE, Haferlach C, et al. Long-term results in patients with acute myeloid leukemia (AML): the influence of high-dose AraC, G-CSF priming, autologous transplantation, prolonged maintenance, age, history, cytogenetics, and mutation status. Data of the AMLCG 1999 trial. Blood. 2009;114(22):485. Döhner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129(4):424-447. Grimwade D, Hills RK, Moorman AV, et al. Refinement of cytogenetic classification in acute myeloid leukemia: determination of prognostic significance of rare recurring chromosomal abnormalities among 5876 younger adult patients treated in the United Kingdom Medical Research Council trials. Blood. 2010;116(3):354-365. 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. Metzeler KH, Herold T, Rothenberg-Thurley M, et al. Spectrum and prognostic relevance of driver gene mutations in acute myeloid leukemia. Blood. 2016;128(5):686-698. Marcucci G, Haferlach T, Döhner H. Molecular genetics of adult acute myeloid leukemia: prognostic and therapeutic implications. J Clin Oncol. 2011;29(5):475-486. Walter RB, Othus M, Borthakur G, et al. Prediction of early death after induction therapy for newly diagnosed acute myeloid leukemia with pretreatment risk scores: a novel paradigm for treatment assignment. J Clin Oncol. 2011;29(33):4417-4423. Patel JP, Gönen M, Figueroa ME, et al. Prognostic relevance of integrated genetic profiling in acute myeloid leukemia. N Engl J Med. 2012;366(12):1079-1089. Green CL, Evans CM, Zhao L, et al. The prognostic significance of IDH2 mutations in AML depends on the location of the mutation. Blood. 2011;118(2):409-412. 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. Stein EM, DiNardo CD, Pollyea DA, et al. Enasidenib in mutant IDH2 relapsed or refractory acute myeloid leukemia. Blood. 2017;130(6):722731. 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.

1759

REVIEW ARTICLE Ferrata Storti Foundation

Preclinical human models and emerging therapeutics for advanced systemic mastocytosis

Michel Arock,1,2 Ghaith Wedeh,1 Gregor Hoermann,3,4 Siham Bibi,1 Cem Akin,5 Barbara Peter,4,6 Karoline V. Gleixner,4,6 Karin Hartmann,7 Joseph H. Butterfield,8 Dean D. Metcalfe9 and Peter Valent4,6

LBPA CNRS UMR8113, Ecole Normale Supérieure Paris-Saclay, Cachan, France; Laboratory of Hematology, Pitié-Salpêtrière Hospital, Paris, France; 3Department of Laboratory Medicine, Medical University of Vienna, Austria; 4Ludwig Boltzmann Cluster Oncology, Medical University of Vienna, Austria; 5Michigan Medicine Allergy Clinic, University of Michigan, Ann Arbor, MI, USA; 6Department of Medicine I, Division of Hematology and Hemostaseology, Medical University of Vienna, Austria; 7Department of Dermatology, University of Luebeck, Germany; 8Mayo Clinic, Division of Allergic Diseases, Rochester, MN, USA and 9Laboratory of Allergic Diseases, NIAID, NIH, Bethesda, MD, USA 1

Haematologica 2018 Volume 103(2):1760-1771

ABSTRACT

Correspondence: arock@ens-cachan.fr

Received: May 4, 2018. Accepted: June 27, 2018. Pre-published: July 5, 2018. doi:10.3324/haematol.2018.195867 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1760

astocytosis is a term used to denote a group of rare diseases characterized by an abnormal accumulation of neoplastic mast cells in various tissues and organs. In most patients with systemic mastocytosis, the neoplastic cells carry activating mutations in KIT. Progress in mastocytosis research has long been hindered by the lack of suitable in vitro models, such as permanent human mast cell lines. In fact, only a few human mast cell lines are available to date: HMC-1, LAD1/2, LUVA, ROSA and MCPV-1. The HMC-1 and LAD1/2 cell lines were derived from patients with mast cell leukemia. By contrast, the more recently established LUVA, ROSA and MCPV-1 cell lines were derived from CD34+ cells of non-mastocytosis donors. While some of these cell lines (LAD1/2, LUVA, ROSAKIT WT and MCPV-1) do not harbor KIT mutations, HMC-1 and ROSAKIT D816V cells exhibit activating KIT mutations found in mastocytosis and have thus been used to study disease pathogenesis. In addition, these cell lines are increasingly employed to validate new therapeutic targets and to screen for effects of new targeted drugs. Recently, the ROSAKIT D816V subclone has been successfully used to generate a unique in vivo model of advanced mastocytosis by injection into immunocompromised mice. Such a model may allow in vivo validation of data obtained in vitro with targeted drugs directed against mastocytosis. In this review, we discuss the major characteristics of all available human mast cell lines, with particular emphasis on the use of HMC-1 and ROSAKIT D816V cells in preclinical therapeutic research in mastocytosis.

©2018 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.

1760

Introduction Mast cells (MC) are tissue-fixed cells found in all vascularized organs. These cells are involved in a number of physiological processes, such as adaptive and innate immune responses.1 Moreover, MC play a central role in many pathological conditions, including allergic reactions and mastocytosis.2 MC develop from bone marrow CD34+/CD117+ progenitor cells,3 which enter the circulation and migrate into tissues, where they mature into MC in response to their major growth factor, stem cell factor (SCF), the ligand of KIT, also known as CD117. KIT is a transmembrane receptor with intrinsic tyrosine kinase activity (Figure 1).4 Besides, mature tissue MC express the high affinity receptor for IgE (FcεRI) and can be activated through this receptor during allergic reactions.5 haematologica | 2018; 103(11)

Preclinical models of mastocytosis

Mastocytosis designates a group of rare disorders characterized by a pathological accumulation of MC in one or more organs.6 Clinical presentations of mastocytosis range from skin-limited disease (cutaneous mastocytosis) occurring mainly in childhood and often regressing spontaneously, to systemic disease categories (systemic mastocytosis; SM), mostly seen in adults. SM variants usually involve the bone marrow and sometimes other internal organs, such as the spleen, liver, and/or gastrointestinal tract. According to the World Health Organization (WHO), mastocytosis can be classified into three major categories: cutaneous mastocytosis, the most common variant, followed by SM, and MC sarcoma, a rare localized MC tumor (Online Supplementary Table S1).7 SM is subdivided into five distinct categories: indolent SM (ISM), smoldering SM (SSM), SM with an associated hematologic neoplasm (SM-AHN), aggressive SM (ASM) and MC leukemia (MCL) (Online Supplementary Table S1).7 While patients with ISM have a normal or near-normal life expectancy, patients with SM-AHN, ASM or MCL, collectively termed advanced SM, share a poor prognosis.8 The diagnosis of SM is based on WHO criteria and is established when one major criterion and one minor criterion or at least three minor criteria are present (Online Supplementary Table S2).9 Once the diagnosis of SM has been established, patients are further graded according to the presence of B-findings reflecting a high MC burden, and of C-findings reflecting organ damage related to MC infiltration (Online Supplementary Table S3).10 The pathophysiology of mastocytosis is complex and if acquired activating mutations in KIT (mostly KIT D816V: NM_000222.2(KIT):c.2447A>T, p.Asp816Val) seem to be major drivers of disease in ISM, the same cannot be said for advanced SM in which, in addition to KIT mutants, KIT-independent signaling pathways are activated and additional genetic defects are frequently found. Given the complex pathophysiology of mastocytosis, in vitro models mimicking neoplastic MC found in SM patients could be useful for developing new therapeutic approaches. To date only a few human MC lines have been described, namely HMC-111 and its subclones (HMC-1.1 and HMC-1.2),12 LAD (subclones 1 through 5),13 LUVA,14 ROSAKIT WT and its subclone ROSAKIT D816V,15 and MCPV-1.1 through MCPV-1.4.16 While LAD, LUVA and ROSAKIT WT cells express KIT wild-type (WT), HMC1.1, HMC-1.2 and ROSAKIT D816V cells harbor KIT activating mutations,15,17 and MCPV-1 are RAS-mutated cells.16 Although these molecular aberrancies do not recapitulate all the characteristics of neoplastic MC found in advanced SM, these last four cell lines are currently the best available models for identifying molecular targets and defining the effects of several interventional (targeted) drugs currently used to treat advanced SM.

Pathophysiology of mastocytosis The pathophysiology of mastocytosis is governed by the presence of KIT activating mutations in neoplastic MC.18 Indeed, various KIT activating mutations have been described, initially in patients with SM,19 then in children with cutaneous mastocytosis.20 In adult SM patients, KIT mutations affect primarily exon 17 encoding for the phosphotransferase domain, usually D816V (>80% of all patients) (Figure 1).21 Other less frequent mutations affect exons 2, 8 and 9 encoding for the extracellular domain or haematologica | 2018; 103(11)

exons 13 and 14 encoding for kinase domain 1.21 By contrast, in children, KIT mutations are found in nearly 75% of biopsies of skin lesions, but the KIT D816V mutation is found in only 30% of all cases.20 Indeed, a significant percentage of children present with KIT mutants located in the extracellular domain (codons 8 and 9) (Figure 1).20 In KIT D816V+ SM patients, the development of neoplastic MC is principally governed by the PI3K/AKT and JAK/STAT5 signaling pathways activated downstream of KIT.22,23 Indeed, AKT and STAT5 are constitutively acti-

Figure 1. Normal structure of the KIT receptor and KIT mutations described in human mast cell leukemia-like cell lines and in patients with mastocytosis. In humans, KIT, located on chromosome 4q12, contains 21 exons transcribed/translated into a transmembrane receptor with tyrosine kinase activity (145 kDa and 976 amino acids). KIT is presented in its monomeric form, but dimerizes as a result of stem cell factor (SCF) ligation. The extracellular domain, in yellow, comprises five immunoglobulin (Ig)-like subunits where the ligand binding site (SCF for KIT) and the dimerization site are located. The cytoplasmic region contains a transmembrane domain (TMD) made by a single helix, in blue. The intracellular portion of KIT, in gray, contains an auto-inhibitory juxtamembrane domain (JMD) and a kinase domain (in dark gray) which is split into two parts: an ATP-binding domain (ABD) and a phosphotransferase domain (PTD) linked by a large kinase insert (KI) domain of ~ 60â&#x20AC;&#x201C;100 residues. For the sake of clarity and readability, only the most frequent mutations found in patients and/or in human MC lines are represented. In red, mutants found in adult patients and in cell lines (KIT D816V found in >80% of adult patients with systemic mastocytosis and 30% of children with cutaneous mastocytosis, as well as in HMC-1.2 and ROSAKIT D816V MCL-like cell lines, and KIT V560G found in the MCLlike cell lines HMC-1.1 and HMC-1.2, but only in a very few adult patients). In black, three of the KIT defects most frequently found in pediatric patients: KIT Del419 (NM_000222.2(KIT):c.1255_1257del, p.Asp419del), KIT ITD501-502 (NM_000222.2(KIT):c.1500_1505dup, p.Ser501_Ala502dup) and KIT ITD502503 (NM_000222.2(KIT):c.1503_1508dup, p.Ala502_Tyr503dup) and in brown, the KIT K509I mutant found in several familial cases of the disease. For a complete overview of the various KIT mutations found in pediatric and adult mastocytosis patients, see Valent et al.7 Del, deletion; ITD, internal tandem duplication.

1761

M. Arock et al.

vated in neoplastic MC in such patients and in KITmutated MCL-like cell lines, and inhibition of these pathways induces growth arrest in such cells.22,24 Other intracellular pathways and molecules, such as the Feline sarcoma oncoprotein,25 or the mechanistic target of rapamycin (mTOR) complex,26 are also potential triggers of oncogenesis. In addition, the KIT mutant activates ERK independently of SRC, in contrast to KIT WT.27 Finally, LYN and BTK are found activated in neoplastic MC in a KIT-independent manner.28 Because KIT-activating mutations are found in most SM patients, several KIT-targeted tyrosine kinase inhibitors (KIT-TKI) have been developed. However, the nature of the mutation influences the sensitivity of the mutant to these TKI. For instance, the KIT D816V mutant is insensitive to imatinib.29 By contrast, patients presenting with KIT WT, or KIT mutant outside exon 17, may potentially respond to imatinib.30 While in ISM the KIT D816V mutant seems to be the unique molecular abnormality found, additional and recurrent somatic mutations of myeloid malignancy-related genes have been reported in advanced SM. The genes most frequently affected are TET2, SRSF2, ASXL1, RUNX1, JAK2, N/KRAS and CBL,31-35 while EZH2, IDH2, ETV6, U2AF or SF3B1 are less often affected.36 All these mutations may be co-expressed with KIT D816V in the same cells or may be expressed in other myeloid cells but not in MC, especially in (A)SM-AHN with TET2, SRSF2 and ASXL1 mutants, in which acquisition of KIT D816V is often a late event conferring a mastocytosis phenotype on a pre-existing clonal condition.37 These defects, and particularly the SRSF2, ASXL1 and RUNX1 (S/A/R) mutations, have a negative impact on the disease prognosis.31-34,35 Thus, targets other than KIT and drug combinations might be considered in order to develop more effective therapies for advanced SM. The potential of new targets and of new targeted drugs or drug combinations is currently investigated using the available human MCL-like MC lines, which will be described hereafter.

Major characteristics of available human mast cell leukemia-like mast cell lines It is beyond the scope of this review to develop all the applications of the available human MC lines. We will only provide a detailed description of the cell lines that qualify as MCL-like, namely HMC-1, ROSA and MCPV1. Indeed, these cell models have been and are still used in vitro and/or in vivo to evaluate the potential effects of drugs and drug combinations to treat mastocytosis. Table 1 shows a summary of the major characteristics of all available human MC lines, while Tables 2 and 3 provide detailed phenotypic information.

The HMC-1 cell line and subclones (HMC-1.1 and HMC-1.2) The origin, major characteristics and phenotype of HMC-1 cells are presented in Tables 1 through 4. HMC-1 cells are metachromatic cells containing histamine and tryptase.11 The original cell line presented with a complex karyotype (Table 1), which might have potentially played a role in the cells’ immortalization. However, despite this fact, HMC-1 cells remained sensitive to KIT inhibitors, in favor of a critical role of the KIT mutants in their maintenance. Indeed, in HMC-1 cells, KIT is constitutively phosphorylated on tyrosine residues in the absence of SCF.17 Sequencing of the coding region of KIT cDNA revealed that KIT in HMC-1 cells is composed of a normal WT 1762

allele and a mutant allele with two point mutations, KIT V560G (NM_000222.2(KIT):c.1679T>G, p.Val560Gly) and KIT D816V.38 Seven years later, two HMC-1 subclones, namely HMC-1.1 and HMC-1.2, were described.12 Both subclones have a heterozygous KIT V560G mutation,12 but only HMC-1.2 cells display the KIT D816V mutation.12 In both subclones, KIT was found constitutively phosphorylated in the absence of SCF, although the presence of the KIT D816V mutation seemed to confer a slight growth advantage to HMC-1.2 cells over HMC-1.1 cells.12 HMC-1 cells have been extensively used to study KIT mutant-related and KIT mutant-independent signaling pathways and to evaluate anti-neoplastic effects of cytoreductive/targeted drugs developed to treat advanced SM.

The ROSAKIT WT and ROSAKIT D816V subclones The SCF-dependent ROSAKIT WT cell line was established from a CD34+ fraction of normal umbilical cord blood cells.15 CD34+ cord blood cells were cultured in the presence of human SCF and, after an 8-week culture period, cells continued to proliferate, with virtually all cells being MC. The doubling time was relatively short (48-72 h) in the presence of SCF.15 ROSAKIT WT cells are round cells with a relatively high nuclear-to-cytoplasm ratio and metachromatic cytoplasmic granules.15 ROSAKIT WT cells stain strongly positive for tryptase and KIT, but express only little if any chymase.15 ROSAKIT WT cells were found to express FcεRI, KIT (CD117), CD33, CD4, CD9, CD203c, and CD300a, consistent with a MC phenotype, while they did not express CD2 or CD25 (Tables 2 and 3).15 Moreover, similar to primary cord blood-derived MC, incubation of ROSA cells with interleukin-4 and IgE for 4-5 days enhanced surface expression of FcεRI. In addition, ROSA cells sensitized with interleukin-4 and IgE were fully activated by antiIgE.15 However, over long periods of continuous culture, expression of FcεRI tends to fade on the cells, which become less sensitive to FcεRI cross-linking (unpublished observation). ROSAKIT WT cells have a normal KIT structure, but harbor a complex karyotype, with a derivative chromosome 1 [der(1)inv(1)(p31q21)del(1)(q24q32)]. In fact, the cell line consists of two subclones: one minor subclone carrying a complete trisomy 5, and the other predominant subclone carrying a partial trisomy 5 [+del(5)(q14q34)].15 In addition, molecular studies revealed that both ROSAKIT WT subclones have a P53 deletion and a hot spot K700E mutation in SF3B1 (unpublished observation). We assume that these alterations contributed to the immortalization of ROSAKIT WT cells and provide a premalignant (permissive) cellular background sufficient to trigger proliferation when a driver, such as KIT D816V, is introduced. Indeed, when ROSAKIT WT cells were further transfected with a lentivirus encoding for GFP + KIT D816V, the resulting subclone proliferated independently of SCF. This subclone, termed ROSAKIT D816V, has the same doubling time as ROSAKIT WT cells cultured in SCF.15 Similar to their parental cells, ROSAKIT D816V cells have a rather mature morphology with numerous cytoplasmic granules.15 ROSAKIT D816V cells exhibit the same complex karyotype and the same SF3B1 K700E mutation as ROSAKIT WT cells.15 Moreover, the phenotype of ROSAKIT D816V cells is similar to that of the parental cell line, including expression of the FcεRI and negativity for CD2 and CD25 (detailed in haematologica | 2018; 103(11)

Preclinical models of mastocytosis

Tables 1-4), although, in contrast to ROSAKIT WT cells, repeated attempts to activate ROSAKIT D816V cells by crosslinking FcεRI failed in our hands.15 Interestingly, KIT (CD117) is expressed at higher levels in ROSAKIT D816V cells than in ROSAKIT WTcells.15 While KIT phosphorylation in ROSAKIT WT cells needs the presence of SCF, KIT is constitutively phosphorylated in ROSAKIT D816V cells.15 In addition, STAT5 and AKT are constitutively phosphorylated in ROSAKIT D816V cells, as in primary neoplastic MC.22,24,39 Interestingly, inhibition of AKT or STAT5 decreases ROSAKIT D816V cell proliferation.15 As expected, ROSAKIT WT cells responded to imatinib, while ROSAKIT D816V cells were resistant to imatinib, but sensitive to dasatinib or midostaurin (PKC412),15 making these couple of cell lines a convenient tool for determining the relative selectivity of TKI towards the two forms of KIT (WT versus D816V).

Of note, ROSAKIT D816V cells were reported to engraft NOD/SCID IL-2Rγ−/− (NSG) mice efficiently, giving rise to an ASM/MCL-like disease in vivo,15 described later in this manuscript. Thus, the ROSAKIT D816V cell line is a unique model of human KIT D816V+ ASM/MCL useful for in vitro and in vivo studies. Finally, ROSA cells also appear well suited to investigating the transforming potential of KIT mutants found in other categories of mastocytosis. For example, starting from ROSAKIT WT, we created ROSA subclones stably expressing the mutant KIT Del417-419insY ( N M _ 0 0 0 2 2 2 . 2 ( K I T ) : c . 1 2 4 9 _ 1 2 5 5 d e l i n s T, p.Thr417_Asp419delinsTyr), or the mutant KIT K509I (NM_000222.2(KIT):c.1526A>T, p.Lys509Ile), both found in pediatric patients.20 In each case, the cells became SCF-independent, and KIT was found constitu-

Table 1. Major characteristics of the available human mast cell lines.

Cell line

HMC-1

Date of first description

Origin

Doubling time

1988

KIT status

SCFdependence

FcεRI expression

Authentication by DNA fingerprinting

KIT V560G and KIT D816V PB of a patient with MCL

Short (48-72 h)

Complex (46 < 2n > XX, ins(10;16) (q25;q22q12), add(13)(q33))

Presence of non-KIT somatic mutation(s)

u.k.

Subclones HMC-1.1 HMC-1.2

2003

LAD subclones (1-5)*

2003

BM cells of a patient with KIT D816G+ MCL

Long (2 weeks)

KIT WT

Yes

Complex (41-72,XXY, -[7],+2[11], +4[2],+5[6],+7[2], +16[3], +18[2], -21[14] cp15)§

u.k.

LUVA

2011

PB of a patient with allergic disease

Short (48-72 h)

KIT WT

Yes

u.k.

ROSAKIT WT

2014

Normal CB-derived CD34+ cells

KIT WT

Yes Yes

KIT V560G and KIT D816V

ROSAKIT D816V

MCPV-1 subclones (1.1-1.4)

KIT V560G

Karyotype

Short (48-72 h)

ROSAKIT WT cells transduced with a KIT D816V construct 2014

Normal CB-derived MC progenitors transduced with HRAS G12V

Short (48-72 h)

KIT D816V

KIT WT

Complex (see SF3B1 K700E description in the text)

u.k.

HRAS G12V

BM: bone marrow; CB: cord blood; MC: mast cell; MCL: mast cell leukemia; PB: peripheral blood; SCF: stem cell factor; u.k.: unknown. *Only the LAD2 subclone has been widely distributed since its description. §Karyotype of the LAD2 subclone.

haematologica | 2018; 103(11)

1763

M. Arock et al. Table 2. Expression of activation-induced antigens, cytokine receptors and molecular targets† by the human mastocytosis-like mast cell lines.

CD CD16 CD23 CD26 CD27 CD32 CD59 CD63 CD64 CD66a CD69 CD71 CD95 CD105 CD114 CD115 CD116 CD127 CD129 CD135 CD138 CD164 CD184 CD203c CD213a1 CD218a CD243 CD304 CD309

Name

HMC-1

ROSAKIT D816V

MCPV-1.1

FcγRIII FcεRII DPP4 TNFRSF7 FcγRII MIRL LAMP-3 FcγRI BGP-1 AIM TfR1 FAS ENDOGLIN G-CSFR M-CSFR GM-CSFRa IL-7R IL-9R FLT3 SYND1 MGC-24 CXCR4 ENPP3 IL-13Ra1 IL-18Ra MDR-1 NRP1 VEGFR2

_ _ _ _ + + + _ + +/+ + _ _ _ _ _ _ _ _ + +/_ _ _ _ _ _

_ _ n.t. n.t. + n.t. + +/+ + + _ _ _ _ _ _ n.t. n.t. _ +/n.t. + _ _ n.t. n.t. _

_ _ _ _ + + + _ _ + + + _ +/_ _ _ + +/_ + _ + _ _ _ + _

some of the related antigens are also listed in Table 4. n.t.: not tested; +: strong expression; +/-: weak expression; -: no expression.

†

tively phosphorylated. In addition, both subclones remained, as expected, sensitive to the growth inhibitory effects of imatinib (unpublished observations). These additional data demonstrate that ROSA cells are reasonable tools for investigating the oncogenic potential of newly discovered KIT mutants as well as for screening for their sensitivity to TKI.

The MCPV-1 subclones The human MCPV-1 subclones (MCPV-1.1 through 1.4) were generated from cord blood-derived CD34+ progenitors by culturing these cells with SCF and interleukin6 for 8 weeks and then stably transducing HRAS G12V, SV40 TAg and TERT.16 Single-cell clones were then isolated and cultured for more than 2 years to demonstrate immortalization. Light microscopy of Wright-Giemsastained MCPV-1.1 cells reveals large, immature cells with bi-, tri-, or multi-lobed (often cloverleaf-like-shaped) nuclei characteristic of MC precursors.16 MCPV-1 cells contain a basophilic cytoplasm, cytoplasmic protrusions and sparse granulation. Moreover, MCPV-1.1 cells exhibit an immunophenotype consistent with MC progenitors (Tables 2 and 3).16 MCPV-1 cells express tryptase but lack 1764

Table 3. Expression of lineage-related markers and adhesion molecules† by the human mastocytosis-like mast cell lines.

CD CD2 CD3 CD4 CD5 CD8 CD9 CD10 CD14 CD15 CD17 CD19 CD20 CD22 CD24 CD31 CD38 CD45 CD48 CD50 CD54 CD56 CD58 CD90 CD96 CD133 CD134 CD144 CD146 CD150 CD153 CD166 CD326

Name

HMC-1

ROSAKIT D816V

MCPV-1.1

LFA2 TCR T4 T1 T8 MRP-1 CALLA LPSR LeX LacCer B4 B1 SIGLEC-2 BA-1 PECAM-1 T10 LCA BLAST1 ICAM-3 ICAM-1 NCAM LFA-3 THY1 TACTILE AC133 OX-40 VE-CADHERIN MUC18 SLAM CD30L ALCAM EPCAM

+ _ _ _ _ + _ _ _ _ _ _ _ + + _ + + + + _ + _ _ _ _ _ _ _ _ + _

n.t. _ + n.t. + + + n.t. + n.t. n.t. _ _ _ n.t. _ + + + + + n.t. + _ _ _ _ _ _ _ _ _

_ _ _ _ _ _ _ _ _ _ _ _ _ _ + _ _ + + + _ + _ + _ _ _ _ _ _ + _

† some of the related antigens are also listed in Table 4. n.t.: not tested; +: strong expression; +/-: weak expression; -: no expression.

surface FcεRI.16 MCPV-1 cells grow independently of SCF and produce a MCL-like disease in NSG mice.

Human mast cell leukemia-like cell lines as models for in vitro testing of growth-inhibiting drugs Treatment of ISM mainly aims at symptomatic relief of MC mediator symptoms.40 By contrast, treatment of advanced SM is challenging and relies principally on nontargeted and/or targeted cytoreductive therapy. 41 In unusual cases (rare KIT-mutant forms or WT KIT) the disease may respond to imatinib or masitinib.30,42,43 In a subgroup of patients with slowly progressing ASM, low-dose prednisolone and interferon-a may be efficacious.44 In addition, low-dose methylprednisolone and cyclosporine A may show some (usually minor) effects in ASM patients.45 Cladribine (2CdA) is often recommended as first-line therapy in patients with advanced SM with multi-organ involvement and slow progression.46,47 A haematologica | 2018; 103(11)

Preclinical models of mastocytosis

forthcoming new standard of therapy in advanced SM is midostaurin (PKC412).48,49 This drug was approved for treatment of advanced SM by the American Food and Drug Administration and the European Medicines Agency in 2017. For ASM/MCL patients with rapid progression and those resistant to 2CdA or midostaurin, poly-chemotherapy is usually recommended, followed, when possible, by allogeneic hematopoietic stem cell transplantation.50 Almost all drug-based cytoreductive therapies have been validated preclinically in vitro using MCL-like MC lines. The most widely used cells for this purpose have been (and still remain) the two HMC-1 subclones. However, the newly emerging MCL-like human MC lines, ROSA and MCPV-1, have also been used repeatedly in such drug-testing studies. A summary of drug-testing approaches and of results obtained with these cell lines is provided in the following paragraphs.

Table 4. Expression by the human mastocytosis-like mast cell lines of antigens aberrantly expressed or overexpressed on malignant mast cells and/or their neoplastic progenitors in patients with systemic mastocytosis.

CD CD13 CD25 CD30 CD33 CD34 CD44 CD52 CD87 CD117 CD123

Name

HMC-1

ROSAKIT D816V

MCPV-1

APN IL-2Ra KI-1 SIGLEC-3 HPCA-1 PGP-1 CAMPATH-1 UPAR KIT IL-3Ra

+ + + +/+ + -

+ + + + + + +/-

+ + + + + + -

+: strong expression; +/-: weak expression; -: no expression.

HMC-1 cell lines and their responses to targeted and non-targeted cytoreductive drugs Numerous antineoplastic drugs have been tested for their effects on HMC-1 cells.51 Among conventional antineoplastic drugs, doxorubicin and cytosine arabinoside were the most active agents.51 Other effective agents were vinblastine, etoposide and mitomycin.51 The potent effects of these chemotherapy-type drugs, otherwise used to treat acute myeloid leukemia, formed the basis to suggest treatment of patients with rapidly progressing ASM and MCL as well as patients with SM-acute myeloid leukemia with standard induction chemotherapy, often as preparation for allogeneic stem cell transplantation. The effects of interferon(s) on the growth of HMC-1 cells have also been analyzed.52 HMC-1 cell numbers decreased in the presence of interferon-Îł but were unaffected by interferon-a,52 contrasting with the activity of interferon-a in a subset of patients with advanced SM.44 This example highlights the fact that not all drug effects observed in vitro can be translated into clinical practice and that in each case, drugs and drug combinations need to be tested in additional disease models and finally in interventional clinical trials. Studies of the in vitro anti-proliferative activity of 2CdA on HMC-1 cells were published after this drug was used in vivo to treat patients with advanced SM. Indeed, the first reports on the in vivo effects of 2CdA in patients were published between 2001 and 2004,53-55 but it was not until 2006 that the in vitro effects of 2CdA on HMC-1 cells were described.56 While 2CdA alone produced growthinhibitory effects on HMC-1 cells, the drug was also found to cooperate with midostaurin.56 The observation that midostaurin can induce apoptosis and growth inhibition in HMC-1 cells and that efficacy was identical in HMC-1.1 and HMC-1.2 cells prompted further investigations and led to the initiation of clinical trials.48,49

Human mast cell leukemia-like cell lines as models of drug resistance Because most patients with SM harbor an activating point mutation in KIT (mostly KIT D816V) which is associated with disease pathology, considerable efforts have been made to identify drugs capable of inhibiting the kinase activity of the KIT mutant. The effects of imatinib, a drug targeting KIT WT, on cell lines harboring various KIT mutations, were investigated soon after the drug was haematologica | 2018; 103(11)

found to block growth of leukemic cells in Philadelphia chromosome-positive chronic myeloid leukemia. In 2000, Ma et al. reported that imatinib inhibited KIT WT at low concentrations, without significant effects on the KIT D816V mutant.57 In 2003, these findings were confirmed using HMC-1.2 cells and patient-derived KIT D816V+ MC.29,58 More recently, it was also confirmed that ROSAKIT D816V cells are insensitive to imatinib.15 Masitinib, another TKI active on KIT WT, although devoid of activity on KIT D816V in vitro,59 was administered in a randomized, double-blind, placebo-controlled, phase 3 study in a cohort of severely symptomatic ISM or SSM patients resistant to classical anti-mediator therapy.60 Interestingly, masitinib improved mediator-related symptoms in a subset of patients as compared to placebo-treated patients, regardless of KIT mutational status.60 This clinical activity was linked to the in vitro inhibitory effects of masitinib on two molecules involved in MC activation, namely LYN and FYN.59 Given the inefficacy of imatinib on the KIT D816V mutant, several other TKI have been evaluated in vitro (and for a few of them in vivo) for their potential activity in the SM context. Dasatinib is a multikinase inhibitor highly active on BCR-ABL1, KIT and PDGFRa.61,62 The potential activity of this drug against KIT D816V was investigated in vitro in HMC-1 cells, SM patient-derived KIT D816V+ cells and ROSA cells.15,63 In each instance, dasatinib exerted in vitro cytotoxic effects at relatively low half maximal inhibitory concentrations (IC50), although the IC50 for dasatinib was higher in KIT D816V+ cells than in KIT D816Vâ&#x20AC;&#x201C; cells.15,63 However, when evaluated in vivo in clinical trials or in individual SM patients, dasatinib unexpectedly demonstrated only marginal activity.64-66 While the in vivo effects of dasatinib have been disappointing, midostaurin, a potent multikinase inhibitor, has proven to be highly promising. Indeed, midostaurin decreased the proliferation of KIT D816V+ cell lines at pharmacological concentrations.15,56,63,67 In addition, the drug abrogated KIT phosphorylation in MCL-like cell lines harboring KIT D816V and induced their apoptosis.15,56,68 Moreover, midostaurin suppressed the growth of primary human KIT D816V+ neoplastic MC.68 Finally, midostaurin was found to block IgE-dependent histamine release from MC and basophils.67,69,70 Based on these data, 1765

M. Arock et al.

clinical trials have been conducted to determine the efficacy of midostaurin in patients with advanced SM, with promising results.48,49,71 An overview of the effects of midostaurin and of several other TKI on the growth of HMC-1 and ROSA cells is presented in Figure 2 and Online Supplementary Table S4. Several other TKI with different mechanisms of action were also found to exert antineoplastic effects in vitro on KIT D816V+ neoplastic MC, including HMC-1 cells. These drugs include 17-AAG (17-allylamino-17demethoxygeldanamycin),72 EXEL-0862 (a TKI active against fibroblast growth factor receptors, vascular endothelial growth factor receptors, platelet-derived growth factor receptors, FLT3 and KIT),73 triptolide (a diterpenoid),74 ponatinib (a multi-kinase blocker),68 and bosutinib (a LYN/BTK-inhibiting TKI), 28 which was administered to a patient with advanced SM, with no benefit.75 Nilotinib, which showed some effects in vitro on mutant KIT,76 was recently administered to 61 SM patients, with transient activity in some patients.77 More recently, several new KIT-TKI with inhibitory activity in vitro on several KIT mutants, including KIT D816V, have been developed. DCC-2618 (Deciphera Inc.) is a switch control type II KIT inhibitor, which arrests KIT in an inactive state, regardless of activating mutations, such as KIT D816V.78 In a recent study, it was found that DCC-2618 inhibits proliferation and survival of HMC1.1, HMC-1.2 and ROSAKIT D816V cells at IC50 <1 mM.79 BLU285 is a TKI developed by Blueprint Medicines. At low concentrations, BLU-285 selectively inhibited KIT D816V enzymatic activity (IC50 = 0.27 nM).80 The cellular activity of BLU-285 on this mutant was also measured by autophosphorylation in HMC-1.2 cells with an IC50 of 4.0 nM.80 Finally, BLU-285 inhibited in vitro the proliferation

of KIT D816V+ HMC-1.2 cells with an IC50 of 125 nM, while being less active on KIT D816V– HMC-1.1 cells (IC50 = 344 nM).81

Human mastocytosis-like mast cell lines and drugs targeting KITdependent or KIT-independent signaling pathways Since KIT D816V is equally present in ISM and advanced SM patients, who have different life expectancies,82 the current assumption is that additional, KIT-independent pathways and pro-oncogenic hits and lesions are responsible for disease progression in advanced SM. Such pathways and pro-oncogenic molecules include LYN, BTK, STAT5, PI3-K, mTOR and members of the BCL-2 family. For instance, LYN and BTK are phosphorylated in HMC-1.1 and HMC-1.2 subclones independently of KIT, and short interfering RNA against LYN and BTK decreased the survival of both subclones.28 In the same set of experiments, dasatinib blocked not only the kinase activity of KIT, but also LYN and BTK activation in neoplastic MC, while bosutinib inhibited LYN and BTK activation without decreasing KIT kinase activity.28 Another molecule, STAT5, seems critical for KIT D816V-driven proliferation in MCL-like MC lines as well as in neoplastic MC in SM patients.24,83 Chaix et al. reported that the KIT D816V mutant can directly phosphorylate STAT5 in vitro.83 Interestingly, STAT5 is transcriptionally active in the HMC-1 cell line and in ROSAKIT D816V cells,15,24 and drugs targeting STAT5 are effective in decreasing the growth rate of these cells.84,85 Figure 3 shows representative curves of dose-dependent inhibition of the viability of MC lines by STAT5 inhibitors. Gabillot-Carre et al. reported constitutive activation of the mTOR signaling pathway in both HMC-1 sub-

Figure 2. Dose-dependent inhibition of the proliferation of wild-type and mutant KIT human mast cell lines by various tyrosine kinase inhibitors in vitro. Human mast cell lines expressing KIT D816V (HMC-1.2 or ROSAKIT D816V; black lines) or lacking KIT D816V (HMC-1.1 or ROSAKIT WT; gray lines) were incubated in control medium (Co) or in medium containing various concentrations of tyrosine kinase inhibitors (as indicated) at 37°C for 48 h. Thereafter, 3H-thymidine uptake was measured. Results are expressed as percent of 3H-thymidine uptake compared to the control and represent the mean ± standard deviation of three different experiments.

1766

haematologica | 2018; 103(11)

Preclinical models of mastocytosis

clones.86 However, the mTOR inhibitor rapamycin induced apoptosis only in HMC-1.2 cells.86 To support this unexpected selectivity, the authors demonstrated that rapamycin inhibited the phosphorylation of 4E-BP1, a downstream substrate of the mTOR pathway, only in HMC-1.2 cells.86 More recently, it was reported that the dual PI3-kinase/mTOR blocker NVP-BEZ235 has similar growth inhibitory effects in HMC-1.1 and HMC-1.2 cells.87 However, despite these encouraging data, no objective response was observed in a study in which everolimus, an oral mTOR inhibitor, was given at a dose of 10 mg daily to ten SM patients.88 Finally, aberrant accumulation of neoplastic MC in SM might result from deregulation of apoptosis pathways.89 Indeed, the anti-apoptotic molecules BCL-2, BCL-xL and MCL-1 are overexpressed in KIT D816V+ neoplastic MC in SM patients,90-92 while the expression of the pro-apoptotic molecule BIM is suppressed in these cells.93 It has also been reported that MCL-1 is detectable in HMC-1.1 and HMC-1.2 cells.93 Moreover, exposure of these cells to MCL-1-specific antisense oligonucleotides or to MCL-1specific short interfering RNA resulted in reduced cell survival and increased apoptosis.93 In further studies, evidence was provided that the pan-BCL-2 family blocker obatoclax inhibited the proliferation of HMC-1 cells, together with increased expression of PUMA, NOXA, and BIM mRNA, and apoptosis.94

Human mastocytosis-like mast cell lines and drugs targeting surface antigens or epigenetic regulators Although drugs targeting KIT D816V have demonstrated activity on MC in vitro and in vivo, these agents do not cure patients with advanced SM.48,49,71 Apart from several different mechanisms of resistance, neoplastic cells in these patients may exhibit a complex pattern of genetic alterations together with, or even often preceding the appearance of, the KIT mutant, as it is the case for TET2, SRSF2 and ASXL1 mutants,37 which could explain resistance to TKI. For this reason, attention has been focused recently on alternative targets which could help to overcome such resistance, namely surface antigens specifically or aberrantly expressed by neoplastic MC and epige-

netic targets. Antibodies or drugs directed against these targets may also be able to overcome intrinsic neoplastic stem cell resistance, often associated with quiescence and altered drug influx or rapid drug efflux. Antibody-based drugs may exert antineoplastic effects independently of such mechanisms of resistance. Targeting surface antigens Several antigens are aberrantly expressed or overexpressed on neoplastic MC and on their progenitors in SM, including CD13, CD25, CD30, CD33, CD44, CD52, CD117 and CD123,95-97 and might, therefore, be considered as potential therapeutic targets. Table 4 provides an overview of cell surface targets expressed on human MCL-like cell lines. CD30 is aberrantly expressed by neoplastic MC in a subset of patients with SM, but not by normal/reactive MC.98 In a recent study, it was observed that the MCPV1.1 subclone expressed high levels of CD30, while HMC1.1 cells expressed low CD30 levels, and HMC-1.2 cells did not express CD30.99 The CD30-targeting antibodyconjugate brentuximab-vedotin inhibited the in vitro proliferation of neoplastic MC, with lower IC50 values obtained for MCPV-1.1 cells (10 mg/mL) than for HMC1.2 cells (>50 mg/mL).99 In addition, brentuximab-vedotin produced apoptosis in primary CD30+ neoplastic MC.99 However, overall, the effects of brentuximab-vedotin on neoplastic MC are relatively weak. Correspondingly, no major clinical activity has been reported in clinical trials to date. In addition, neoplastic stem cells in advanced SM usually lack CD30 (personal information, PV). In contrast to normal MC and MC from ISM patients, CD52 is abundantly expressed on neoplastic MC in most patients with advanced SM.16 Recently, it was reported that the CD52-targeting antibody alemtuzumab counteracts growth of MCPV-1.1 cells.16 Additionally, MCPV-1.1 cells were injected into NSG mice which were then treated with alemtuzumab or control vehicle. The alemtuzumab-treated mice had increased survival compared to controls, and reduced organ infiltration by neoplastic MC.16 Given that neoplastic (leukemic) stem cells identified in advanced SM may also express CD52,100 it can be

Figure 3. Dose-dependent inhibition of the proliferation of KIT D816V+ human mast cell lines by STAT5 inhibitors in vitro. ROSAKIT D816V and HMC-1.2 cells were cultured in 96-well plates for 72 h in control medium containing 0.1% dimethylsulfoxide (DMSO) or with increasing concentrations (between 1.0 and 50.0 mM) of SF-1066, a very weak STA5 inhibitor (Ki >25 mM on STAT5), and of more specific and potent STAT5 inhibitors BP-1-102 and BP-1-108 (Ki >10 mM on STAT5).74 Viability was calculated in each condition by the MTT method. Results are expressed as percent of control and represent the mean Âą standard deviation of triplicate experiments. The half maximal inhibitory concentration (IC50) at day 3 of each compound for each cell line was calculated using Prism GraphPad 4.0 software after plotting log concentration versus response. As expected, while the IC50 for SF-1066 was >50 mM, IC50 values for the more STAT5-specific compounds were lower: 11 mM for BP-1-102 on both KIT D816V+ neoplastic human MC lines and 34 and 22 mM for BP-1-108 for, respectively, HMC-1.2 and ROSAKIT D816V cells. Although these values are still irrelevant at the pharmacological level, they open hopes that drug optimization might lead in a near future to the design of more potent small molecules inhibiting STAT5 activity.

haematologica | 2018; 103(11)

1767

M. Arock et al.

hypothesized that a combination of a KIT-TKI and a monoclonal antibody against CD52 might help to achieve major antineoplastic effects in advanced SM. Another potential surface target is CD33. In fact, CD33 is invariably expressed on neoplastic MC and their stem cells in patients with advanced SM.95 In addition, it has been described that the CD33-targeted antibody-construct gemtuzumab-ozogamicin is able to suppress growth and survival of neoplastic MC.101 In the light of the revival of gemtuzumab-ozogamicin, its clinical efficacy in patients with acute myeloid leukemia and its effects on neoplastic MC,102 it might be reasonable to propose clinical trials testing the effects of gemtuzumab-ozogamicin alone or in combination with other antineoplastic drugs or stem cell transplantation in advanced SM. Targeting epigenetic regulators The epigenetic reader bromodomain-containing 4 protein (BRD4), a member of the BET family proteins, has recently been identified as a promising target in acute myeloid leukemia.103,104 In addition, highly selective BET bromodomain inhibitors, including JQ1,105 I-BET151,106,107 and I-BET762,106,108 have demonstrated in vitro and in vivo activity against several hematopoietic malignancies. It has also been evaluated whether BRD4 might be a target in advanced SM. Indeed, BRD4 was found to be expressed in HMC-1.1, HMC-1.2, ROSAKIT WT and ROSAKIT D816V cells as well as in primary neoplastic MC.109 Independently of the grade or variant of disease, neoplastic MC exhibit nuclear BRD4.109 However, in ASM and MCL, neoplastic MC also express substantial amounts of cytoplasmic BRD4.109 In line with this observation, HMC1 and ROSA cells express cytoplasmic and nuclear BRD4 as well.109 The KIT-TKI midostaurin and dasatinib suppressed the expression of BRD4 in all MC lines.109 BRD4specific short hairpin RNA and JQ1 decreased the proliferation of HMC-1 and ROSA cells.109 Based on these data, 1768

Figure 4. Engraftment of ROSAKIT D816V-Gluc cells in NSG mice. Increasing numbers of ROSAKIT D816V-Gluc cells (1x106, 5x106 or 10x106) were injected into the tail vein of NOD-SCID IL-2RÎł/(NSG) mice (3 groups; 6 mice per group) 24 h after irradiation at 2.5 Gy from a cesium-137 source.113 After 10 weeks, engraftment was assessed using (A) quantitative measurements of Gluc activity in plasma and (B) in vivo bioluminescence imaging (IVIS) on engrafted mice. Ten weeks after engraftment, mice were sacrificed and (C) bone marrow and (D) spleen sections from the three groups of mice were stained by immunohistochemistry with an anti-human tryptase antibody. Staining was visualized by a Histomouse Kit, showing human MC with a brownish stain, which massively invaded the bone marrow and to a lesser extent the spleen. (C) and (D) are from one representative mouse from the group injected with 10x106 ROSAKIT D816V-Gluc cells. Original magnification, x100.

BRD4 is a promising target in advanced SM, although this needs to be confirmed in forthcoming clinical studies.

Human mast cell leukemia-like cell lines as tools to develop in vivo models In vivo models have been developed in order to understand the pathophysiology of SM better. In addition to transgenic mouse models,110,111 another approach is to create a SM-like disease in vivo by transplanting human neoplastic MC into immunodeficient mice. The HMC-1 cell line engrafts immunodeficient mice after intravenous or subcutaneous injection, giving rise to subcutaneous tumors after 2 to 5 months.11, 112 The reason why intravenous injection does not give rise to a MCLlike disease is unknown, but limits the usefulness of HMC-1 cells to establish an in vivo model of advanced SM. More recently, an in vivo model of advanced SM was established using ROSA cells. Indeed, we engineered a ROSA subclone, termed ROSAKIT D816V-Gluc, which naturally secretes Gaussia princeps luciferase (Gluc), used as a reporter.113 In this study, intravenous injection of NSG mice with ROSAKIT D816V-Gluc cells led to an advanced SM phenotype, with neoplastic MC invading the bone marrow, spleen and liver, as testified by the quantification of engrafting cells by measuring Gluc reporter activity in peripheral blood and by an in vivo imaging system (IVIS).113 The detailed characteristics of this in vivo model are presented in Figure 4. All in all, this in vivo model of advanced SM is potentially the best available to date for in vivo testing of drugs previously identified as active in vitro on neoplastic MC.

Summary and future perspectives Despite decades of intensive research, only a few human MC lines have been established to date: HMC-1, LAD-2, haematologica | 2018; 103(11)

Preclinical models of mastocytosis

LUVA, ROSA and MCPV-1. While none of these cell lines simultaneously expresses the KIT D816V mutant and a functional FcεRI, making them useless for testing MC-stabilizing drugs or drugs interfering with FcεRI-induced signaling in the context of KIT D816V+ SM, some of these cell lines may qualify as MCL-like since they harbor SM-related KIT variants and/or other oncogenic molecules relevant to SM. Among all MC lines, HMC-1 cells have been most frequently used, but other more recently established MC lines, such as ROSA and MCPV-1, are now available and are being used in various preclinical studies. For example, these cell lines have been used to analyze in vitro the growth-inhibitory effects of antineoplastic drugs, including various KIT-TKI, on neoplastic MC. However, because neoplastic MC in advanced SM are triggered by KIT-independent signaling pathways and additional genetic lesions that confer resistance against KIT-TKI, it might be interesting to establish in vitro models of multi-mutated neoplastic MC, starting from established human KIT mutant-positive MC lines in which additional lesions, such as the S/A/R

References 1. DeBruin EJ, Gold M, Lo BC, et al. Mast cells in human health and disease. Methods Mol Biol. 2015;1220:93-119. 2. Galli SJ, Tsai M. IgE and mast cells in allergic disease. Nat Med. 2012;18(5):693-704. 3. Kirshenbaum AS, Metcalfe DD. Growth of human mast cells from bone marrow and peripheral blood-derived CD34+ pluripotent progenitor cells. Methods Mol Biol. 2006;315:105-112. 4. Galli SJ, Tsai M, Wershil BK, Tam SY, Costa JJ. Regulation of mouse and human mast cell development, survival and function by stem cell factor, the ligand for the c-kit receptor. Int Arch Allergy Immunol. 1995;107(1-3):5153. 5. Brown JM, Wilson TM, Metcalfe DD. The mast cell and allergic diseases: role in pathogenesis and implications for therapy. Clin Exp Allergy. 2008;38(1):4-18. 6. Arock M, Valent P. Pathogenesis, classification and treatment of mastocytosis: state of the art in 2010 and future perspectives. Expert Rev Hematol. 2010;3(4):497-516. 7. Valent P, Akin C, Hartmann K, et al. Advances in the classification and treatment of mastocytosis: current status and outlook toward the future. Cancer Res. 2017;77 (6):1261-1270. 8. Pardanani A. Systemic mastocytosis in adults: 2017 update on diagnosis, risk stratification and management. Am J Hematol. 2016;91 (11):1146-1159. 9. Valent P, Horny HP, Escribano L, et al. Diagnostic criteria and classification of mastocytosis: a consensus proposal. Leuk Res. 2001;25(7):603-625. 10. Valent P, Escribano L, Broesby-Olsen S, et al. Proposed diagnostic algorithm for patients with suspected mastocytosis: a proposal of the European Competence Network on Mastocytosis. Allergy. 2014;69(10):12671274. 11. Butterfield JH, Weiler D, Dewald G, Gleich GJ. Establishment of an immature mast cell line from a patient with mast cell leukemia. Leuk Res. 1988;12(4):345-355. 12. Sundstrom M, Vliagoftis H, Karlberg P, et al.

haematologica | 2018; 103(11)

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

combination of molecular lesions might be introduced. Such multi-mutated neoplastic MC lines should be useful to test combination therapies in vitro, and later in clinical trials in patients with advanced SM. With these approaches, new therapeutic concepts should be established in order to improve therapy in advanced SM. Acknowledgments The authors would like to thank Dr Patrick T. Gunning (Department of Medical Biophysics, University of Toronto, Ontario, Canada) for kindly providing the STAT-specific inhibitors presented in the manuscript. We also express our deep thanks to Dr Fawzia Louache (Inserm UMRS-1170, CNRS GDR 3697 Micronit, Institut Gustave Roussy, Université ParisSud, Université Paris-Saclay, Villejuif, France) for her invaluable help with the in vivo studies presented in the manuscript. K.H. is supported by a research grant from the German Research Council (DFG; HA 2393/6-1). DM is supported by the DIR, NIAID, and PV is supported by the Austrian Science Fund (FWF) grants F 4701-B20 and F 4704-B20.

Functional and phenotypic studies of two variants of a human mast cell line with a distinct set of mutations in the c-kit protooncogene. Immunology. 2003;108(1):89-97. Kirshenbaum AS, Akin C, Wu Y, et al. Characterization of novel stem cell factor responsive human mast cell lines LAD 1 and 2 established from a patient with mast cell sarcoma/leukemia; activation following aggregation of FcepsilonRI or FcgammaRI. Leuk Res. 2003;27(8):677-682. Laidlaw TM, Steinke JW, Tinana AM, et al. Characterization of a novel human mast cell line that responds to stem cell factor and expresses functional FcepsilonRI. J Allergy Clin Immunol. 2011;127(3):815-822. Saleh R, Wedeh G, Herrmann H, et al. A new human mast cell line expressing a functional IgE receptor converts to tumorigenic growth by KIT D816V transfection. Blood. 2014;124(1):111-120. Hoermann G, Blatt K, Greiner G, et al. CD52 is a molecular target in advanced systemic mastocytosis. FASEB J. 2014;28(8):35403551. Kanakura Y, Furitsu T, Tsujimura T, et al. Activating mutations of the c-kit protooncogene in a human mast cell leukemia cell line. Leukemia. 1994;8 (Suppl 1):S18-22. Cruse G, Metcalfe DD, Olivera A. Functional deregulation of KIT: link to mast cell proliferative diseases and other neoplasms. Immunol Allergy Clin North Am. 2014;34 (2):219-237. Longley BJ, Tyrrell L, Lu SZ, et al. Somatic cKIT activating mutation in urticaria pigmentosa and aggressive mastocytosis: establishment of clonality in a human mast cell neoplasm. Nat Genet. 1996;12(3):312-314. Bodemer C, Hermine O, Palmerini F, et al. Pediatric mastocytosis is a clonal disease associated with D816V and other activating c-KIT mutations. J Invest Dermatol. 2010;130(3):804815. Arock M, Sotlar K, Akin C, et al. KIT mutation analysis in mast cell neoplasms: recommendations of the European Competence Network on Mastocytosis. Leukemia. 2015;29(6):1223-1232. Harir N, Boudot C, Friedbichler K, et al. Oncogenic Kit controls neoplastic mast cell

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

growth through a Stat5/PI3-kinase signaling cascade. Blood. 2008;112(6):2463-2473. Bibi S, Arslanhan MD, Langenfeld F, et al. Co-operating STAT5 and AKT signaling pathways in chronic myeloid leukemia and mastocytosis: possible new targets of therapy. Haematologica. 2014;99(3):417-429. Baumgartner C, Cerny-Reiterer S, Sonneck K, et al. Expression of activated STAT5 in neoplastic mast cells in systemic mastocytosis: subcellular distribution and role of the transforming oncoprotein KIT D816V. Am J Pathol. 2009;175(6):2416-2429. Voisset E, Lopez S, Dubreuil P, De Sepulveda P. The tyrosine kinase FES is an essential effector of KITD816V proliferation signal. Blood. 2007;110(7):2593-2599. Smrz D, Kim MS, Zhang S, et al. mTORC1 and mTORC2 differentially regulate homeostasis of neoplastic and non-neoplastic human mast cells. Blood. 2011;118(26): 6803-6813. Sun J, Pedersen M, Ronnstrand L. The D816V mutation of c-Kit circumvents a requirement for Src family kinases in c-Kit signal transduction. J Biol Chem. 2009;284(17):11039-11047. Gleixner KV, Mayerhofer M, Cerny-Reiterer S, et al. KIT-D816V-independent oncogenic signaling in neoplastic cells in systemic mastocytosis: role of Lyn and Btk activation and disruption by dasatinib and bosutinib. Blood. 2011;118(7):1885-1898. Zermati Y, De Sepulveda P, Feger F, et al. Effect of tyrosine kinase inhibitor STI571 on the kinase activity of wild-type and various mutated c-kit receptors found in mast cell neoplasms. Oncogene. 2003;22(5):660-664. Alvarez-Twose I, Matito A, Morgado JM, et al. Imatinib in systemic mastocytosis: a phase IV clinical trial in patients lacking exon 17 KIT mutations and review of the literature. Oncotarget. 2017;8(40):6895068963. Traina F, Visconte V, Jankowska AM, et al. Single nucleotide polymorphism array lesions, TET2, DNMT3A, ASXL1 and CBL mutations are present in systemic mastocytosis. PLoS One. 2012;7(8):e43090. Soucie E, Hanssens K, Mercher T, et al. In aggressive forms of mastocytosis, TET2 loss

1769

M. Arock et al.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47. 48.

1770

cooperates with c-KITD816V to transform mast cells. Blood. 2012;120(24):4846-4849. Damaj G, Joris M, Chandesris O, et al. ASXL1 but not TET2 mutations adversely impact overall survival of patients suffering systemic mastocytosis with associated clonal hematologic nonmast-cell diseases. PLoS One. 2014;9(1):e85362. Hanssens K, Brenet F, Agopian J, et al. SRSF2-p95 hotspot mutation is highly associated with advanced forms of mastocytosis and mutations in epigenetic regulator genes. Haematologica. 2014;99(5):830-835. Jawhar M, Schwaab J, Schnittger S, et al. Additional mutations in SRSF2, ASXL1 and/or RUNX1 identify a high-risk group of patients with KIT D816V(+) advanced systemic mastocytosis. Leukemia. 2016;30(1): 136-143. Bibi S, Langenfeld F, Jeanningros S, et al. Molecular defects in mastocytosis: KIT and beyond KIT. Immunol Allergy Clin North Am. 2014;34(2):239-262. Jawhar M, Schwaab J, Schnittger S, et al. Molecular profiling of myeloid progenitor cells in multi-mutated advanced systemic mastocytosis identifies KIT D816V as a distinct and late event. Leukemia. 2015;29(5):1115-1122. Furitsu T, Tsujimura T, Tono T, et al. Identification of mutations in the coding sequence of the proto-oncogene c-kit in a human mast cell leukemia cell line causing ligand-independent activation of c-kit product. J Clin Invest. 1993;92(4):1736-1744. Zuluaga Toro T, Hsieh FH, Bodo J, Dong HY, Hsi ED. Detection of phospho-STAT5 in mast cells: a reliable phenotypic marker of systemic mast cell disease that reflects constitutive tyrosine kinase activation. Br J Haematol. 2007;139(1):31-40. Valent P, Akin C, Escribano L, et al. Standards and standardization in mastocytosis: consensus statements on diagnostics, treatment recommendations and response criteria. Eur J Clin Invest. 2007;37(6):435453. Arock M, Akin C, Hermine O, Valent P. Current treatment options in patients with mastocytosis: status in 2015 and future perspectives. Eur J Haematol. 2015;94(6):474490. Georgin-Lavialle S, Lhermitte L, Suarez F, et al. Mast cell leukemia: identification of a new c-Kit mutation, dup(501-502), and response to masitinib, a c-Kit tyrosine kinase inhibitor. Eur J Haematol. 2012;89(1):47-52. Alvarez-Twose I, Matito A, Morgado JM, et al. Imatinib in systemic mastocytosis: a phase IV clinical trial in patients lacking exon 17 KIT mutations and review of the literature. Oncotarget. 2016;8(40):6895068963. Simon J, Lortholary O, Caillat-Vigneron N, et al. Interest of interferon alpha in systemic mastocytosis. The French experience and review of the literature. Pathol Biol (Paris). 2004;52(5):294-299. Kurosawa M, Amano H, Kanbe N, et al. Response to cyclosporin and low-dose methylprednisolone in aggressive systemic mastocytosis. J Allergy Clin Immunol. 1999;103(5):S412-420. Barete S, Lortholary O, DamaJ G, et al. Long-term efficacy and safety of cladribine (2-CdA) in adult patients with mastocytosis. Blood. 2015;126(8):1009-1016. Akin C. Cladribine for mastocytosis: benefits and risks. Blood. 2015;126(8):931-932. Gotlib J, Kluin-Nelemans HC, George TI, et al. Efficacy and safety of midostaurin in advanced systemic mastocytosis. N Engl J

Med. 2016;374(26):2530-2541. 49. Chandesris MO, Damaj G, Canioni D, et al. Midostaurin in advanced systemic mastocytosis. N Engl J Med. 2016;374(26): 26052607. 50. Ustun C, Gotlib J, Popat U, et al. Consensus opinion on allogeneic hematopoietic cell transplantation in advanced systemic mastocytosis. Biol Blood Marrow Transplant. 2016;22(8):1348-1356. 51. Butterfield JH, Weiler DA. In vitro sensitivity of immature human mast cells to chemotherapeutic agents. Int Arch Allergy Appl Immunol. 1989;89(2-3):297-300. 52. Kirshenbaum AS, Worobec AS, Davis TA, Goff JP, Semere T, Metcalfe DD. Inhibition of human mast cell growth and differentiation by interferon gamma-1b. Exp Hematol. 1998;26(3):245-251. 53. Tefferi A, Li CY, Butterfield JH, Hoagland HC. Treatment of systemic mast-cell disease with cladribine. N Engl J Med. 2001;344(4): 307-309. 54. Kluin-Nelemans HC, Oldhoff JM, Van Doormaal JJ, et al. Cladribine therapy for systemic mastocytosis. Blood. 2003;102(13): 4270-4276. 55. Pardanani A, Hoffbrand AV, Butterfield JH, Tefferi A. Treatment of systemic mast cell disease with 2-chlorodeoxyadenosine. Leuk Res. 2004;28(2):127-131. 56. Gleixner KV, Mayerhofer M, Aichberger KJ, et al. PKC412 inhibits in vitro growth of neoplastic human mast cells expressing the D816V-mutated variant of KIT: comparison with AMN107, imatinib, and cladribine (2CdA) and evaluation of cooperative drug effects. Blood. 2006;107(2):752-759. 57. Ma Y, Zeng S, Metcalfe DD, et al. The c-KIT mutation causing human mastocytosis is resistant to STI571 and other KIT kinase inhibitors; kinases with enzymatic site mutations show different inhibitor sensitivity profiles than wild-type kinases and those with regulatory-type mutations. Blood. 2002;99(5):1741-1744. 58. Akin C, Brockow K, D'Ambrosio C, et al. Effects of tyrosine kinase inhibitor STI571 on human mast cells bearing wild-type or mutated c-kit. Exp Hematol. 2003;31(8):686692. 59. Dubreuil P, Letard S, Ciufolini M, et al. Masitinib (AB1010), a potent and selective tyrosine kinase inhibitor targeting KIT. PLoS One. 2009;4(9):e7258. 60. Lortholary O, Chandesris MO, Bulai Livideanu C, et al. Masitinib for treatment of severely symptomatic indolent systemic mastocytosis: a randomised, placebo-controlled, phase 3 study. Lancet. 2017;389 (10069):612-620. 61. Schittenhelm MM, Shiraga S, Schroeder A, et al. Dasatinib (BMS-354825), a dual SRC/ABL kinase inhibitor, inhibits the kinase activity of wild-type, juxtamembrane, and activation loop mutant KIT isoforms associated with human malignancies. Cancer Res. 2006;66(1):473-481. 62. Shah NP, Lee FY, Luo R, Jiang Y, Donker M, Akin C. Dasatinib (BMS-354825) inhibits KITD816V, an imatinib-resistant activating mutation that triggers neoplastic growth in most patients with systemic mastocytosis. Blood. 2006;108(1):286-291. 63. Gleixner KV, Mayerhofer M, Sonneck K, et al. Synergistic growth-inhibitory effects of two tyrosine kinase inhibitors, dasatinib and PKC412, on neoplastic mast cells expressing the D816V mutated oncogenic variant of KIT. Haematologica. 2007;92(11):1451-1459. 64. Aichberger KJ, Sperr WR, Gleixner KV, Kretschmer A, Valent P. Treatment respons-

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

es to cladribine and dasatinib in rapidly progressing aggressive mastocytosis. Eur J Clin Invest. 2008;38(11):869-873. Purtill D, Cooney J, Sinniah R, et al. Dasatinib therapy for systemic mastocytosis: four cases. Eur J Haematol. 2008;80 (5):456-458. Verstovsek S, Tefferi A, Cortes J, et al. Phase II study of dasatinib in Philadelphia chromosome-negative acute and chronic myeloid diseases, including systemic mastocytosis. Clin Cancer Res. 2008;14(12): 3906-3915. Peter B, Winter GE, Blatt K, et al. Target interaction profiling of midostaurin and its metabolites in neoplastic mast cells predicts distinct effects on activation and growth. Leukemia. 2016;30(2):464-472. Gleixner KV, Peter B, Blatt K, et al. Synergistic growth-inhibitory effects of ponatinib and midostaurin (PKC412) on neoplastic mast cells carrying KIT D816V. Haematologica. 2013;98(9):1450-1457. Krauth MT, Mirkina I, Herrmann H, Baumgartner C, Kneidinger M, Valent P. Midostaurin (PKC412) inhibits immunoglobulin E-dependent activation and mediator release in human blood basophils and mast cells. Clin Exp Allergy. 2009;39(11):1711-1720. Valent P, Akin C, Hartmann K, et al. Midostaurin: a magic bullet that blocks mast cell expansion and activation. Ann Oncol. 2017;28(10):2367-2376. DeAngelo DJ, George TI, Linder A, et al. Efficacy and safety of midostaurin in patients with advanced systemic mastocytosis: 10-year median follow-up of a phase II trial. Leukemia. 2018;32(2):470-478. Fumo G, Akin C, Metcalfe DD, Neckers L. 17-Allylamino-17-demethoxygeldanamycin (17AAG) is effective in down-regulating mutated, constitutively activated KIT protein in human mast cells. Blood. 2004;103(3):1078-1084. Pan J, Quintas-Cardama A, Kantarjian HM, et al. EXEL-0862, a novel tyrosine kinase inhibitor, induces apoptosis in vitro and ex vivo in human mast cells expressing the KIT D816V mutation. Blood. 2007;109(1):315322. Jin Y, Chen Q, Shi X, et al. Activity of triptolide against human mast cells harboring the kinase domain mutant KIT. Cancer Sci. 2009;100(7):1335-1343. Randall N, Courville EL, Baughn L, Afrin L, Ustun C. Bosutinib, a Lyn/Btk inhibiting tyrosine kinase inhibitor, is ineffective in advanced systemic mastocytosis. Am J Hematol. 2015;90(4):E74. Manley PW, Drueckes P, Fendrich G, et al. Extended kinase profile and properties of the protein kinase inhibitor nilotinib. Biochim Biophys Acta. 2010;1804(3):445453. Hochhaus A, Baccarani M, Giles FJ, et al. Nilotinib in patients with systemic mastocytosis: analysis of the phase 2, open-label, single-arm nilotinib registration study. J Cancer Res Clin Oncol. 2015;141(11):20472060. Bai Y, Bandara G, Ching Chan E, et al. Targeting the KIT activating switch control pocket: a novel mechanism to inhibit neoplastic mast cell proliferation and mast cell activation. Leukemia. 2013;27(2):278-285. Schneeweiss M, Peter B, Bibi S, et al. The KIT and PDGFRA switch-control inhibitor DCC2618 blocks growth and survival of multiple neoplastic cell types in advanced mastocytosis. Haematologica. 2018;103(5): 799-809.

haematologica | 2018; 103(11)

Preclinical models of mastocytosis 80. Evans E, Gardino A, Hodous B, et al. Blu-285, a potent and selective inhibitor for hematologic malignancies with KIT exon 17 mutations. Blood. 2015;126(23):568-568. 81. Evans EK, Gardino AK, Kim JL, et al. A precision therapy against cancers driven by KIT/PDGFRA mutations. Sci Tran Med. 2017;9(414). 82. Lim KH, Tefferi A, Lasho TL, et al. Systemic mastocytosis in 342 consecutive adults: survival studies and prognostic factors. Blood. 2009;113(23):5727-5736. 83. Chaix A, Lopez S, Voisset E, Gros L, Dubreuil P, De Sepulveda P. Mechanisms of STAT protein activation by oncogenic KIT mutants in neoplastic mast cells. J Biol Chem. 2011;286(8):5956-5966. 84. Page BD, Khoury H, Laister RC, et al. Small molecule STAT5-SH2 domain inhibitors exhibit potent antileukemia activity. J Med Chem. 2012;55(3):1047-1055. 85. Peter B, Bibi S, Eisenwort G, et al. Druginduced inhibition of phosphorylation of STAT5 overrides drug resistance in neoplastic mast cells. Leukemia. 2018;32(4):1016-1022. 86. Gabillot-Carre M, Lepelletier Y, Humbert M, et al. Rapamycin inhibits growth and survival of D816V-mutated c-kit mast cells. Blood. 2006;108(3):1065-1072. 87. Blatt K, Herrmann H, Mirkina I, et al. The PI3-kinase/mTOR-targeting drug NVPBEZ235 inhibits growth and IgE-dependent activation of human mast cells and basophils. PLoS One. 2012;7(1):e29925. 88. Parikh SA, Kantarjian HM, Richie MA, Cortes JE, Verstovsek S. Experience with everolimus (RAD001), an oral mammalian target of rapamycin inhibitor, in patients with systemic mastocytosis. Leuk Lymphoma. 2010;51(2):269-274. 89. Baldus SE, Zirbes TK, Thiele J, Eming SA, Henz BM, Hartmann K. Altered apoptosis and cell cycling of mast cells in bone marrow lesions of patients with systemic mastocytosis. Haematologica. 2004;89(12):1525-1527. 90. Cervero C, Escribano L, San Miguel JF, et al. Expression of Bcl-2 by human bone marrow mast cells and its overexpression in mast cell leukemia. Am J Hematol. 1999;60(3):191195. 91. Hartmann K, Artuc M, Baldus SE, et al. Expression of Bcl-2 and Bcl-xL in cutaneous and bone marrow lesions of mastocytosis. Am J Pathol. 2003;163(3):819-826.

haematologica | 2018; 103(11)

92. Aichberger KJ, Mayerhofer M, Gleixner KV, et al. Identification of MCL1 as a novel target in neoplastic mast cells in systemic mastocytosis: inhibition of mast cell survival by MCL1 antisense oligonucleotides and synergism with PKC412. Blood. 2007;109(7): 3031-3041. 93. Aichberger KJ, Gleixner KV, Mirkina I, et al. Identification of proapoptotic Bim as a tumor suppressor in neoplastic mast cells: role of KIT D816V and effects of various targeted drugs. Blood. 2009;114(26):5342-5351. 94. Peter B, Cerny-Reiterer S, Hadzijusufovic E, et al. The pan-Bcl-2 blocker obatoclax promotes the expression of Puma, Noxa, and Bim mRNA and induces apoptosis in neoplastic mast cells. J Leukoc Biol. 2014;95(1): 95-104. 95. Valent P, Cerny-Reiterer S, Herrmann H, et al. Phenotypic heterogeneity, novel diagnostic markers, and target expression profiles in normal and neoplastic human mast cells. Best Pract Res Clin Haematol. 2010;23(3): 369-378. 96. Teodosio C, Mayado A, Sanchez-Munoz L, et al. The immunophenotype of mast cells and its utility in the diagnostic work-up of systemic mastocytosis. J Leukoc Biol. 2015;97(1):49-59. 97. Pardanani A, Lasho T, Chen D, et al. Aberrant expression of CD123 (interleukin3 receptor alpha) on neoplastic mast cells. Leukemia. 2015;29(7):1605-1608. 98. Sotlar K, Cerny-Reiterer S, Petat-Dutter K, et al. Aberrant expression of CD30 in neoplastic mast cells in high-grade mastocytosis. Mod Pathol. 2011;24(4):585-595. 99. Blatt K, Cerny-Reiterer S, Schwaab J, et al. Identification of the Ki-1 antigen (CD30) as a novel therapeutic target in systemic mastocytosis. Blood. 2015;126(26):2832-2841. 100. Florian S, Sonneck K, Hauswirth AW, et al. Detection of molecular targets on the surface of CD34+/CD38-- stem cells in various myeloid malignancies. Leuk Lymphoma. 2006;47(2):207-222. 101. Krauth MT, Bohm A, Agis H, et al. Effects of the CD33-targeted drug gemtuzumab ozogamicin (Mylotarg) on growth and mediator secretion in human mast cells and blood basophils. Exp Hematol. 2007;35(1): 108-116. 102. Alvarez-Twose I, Martinez-Barranco P, Gotlib J, et al. Complete response to gem-

tuzumab ozogamicin in a patient with refractory mast cell leukemia. Leukemia. 2016;30(8):1753-1756. 103. Zuber J, Shi J, Wang E, et al. RNAi screen identifies Brd4 as a therapeutic target in acute myeloid leukaemia. Nature. 2011;478(7370):524-528. 104. Herrmann H, Blatt K, Shi J, et al. Small-molecule inhibition of BRD4 as a new potent approach to eliminate leukemic stem- and progenitor cells in acute myeloid leukemia AML. Oncotarget. 2012;3(12):1588-1599. 105. Filippakopoulos P, Qi J, Picaud S, et al. Selective inhibition of BET bromodomains. Nature. 2010;468(7327):1067-1073. 106. Chaidos A, Caputo V, Gouvedenou K, et al. Potent antimyeloma activity of the novel bromodomain inhibitors I-BET151 and IBET762. Blood. 2014;123(5):697-705. 107. Wyspianska BS, Bannister AJ, Barbieri I, et al. BET protein inhibition shows efficacy against JAK2V617F-driven neoplasms. Leukemia. 2014;28(1):88-97. 108. Mirguet O, Gosmini R, Toum J, et al. Discovery of epigenetic regulator I-BET762: lead optimization to afford a clinical candidate inhibitor of the BET bromodomains. J Med Chem. 2013;56(19):7501-7515. 109. Wedeh G, Cerny-Reiterer S, Eisenwort G, et al. Identification of bromodomain-containing protein-4 as a novel marker and epigenetic target in mast cell leukemia. Leukemia. 2015;29(11):2230-2237. 110. Zappulla JP, Dubreuil P, Desbois S, et al. Mastocytosis in mice expressing human Kit receptor with the activating Asp816Val mutation. J Exp Med. 2005;202(12):16351641. 111. Gerbaulet A, Wickenhauser C, Scholten J, et al. Mast cell hyperplasia, B-cell malignancy, and intestinal inflammation in mice with conditional expression of a constitutively active Kit. Blood. 2011;117(6):2012-2021. 112. Schumacher U, van Damme EJ, Peumans WJ, Butterfield JH, Mitchell BS. Lectin histochemistry of human leukaemic mast cells (HMC-1) transplanted into severe combined immunodeficient (scid) mice. Acta Histochem. 1998;100(1):1-9. 113. Bibi S, Zhang Y, Hugonin C, et al. A new humanized in vivo model of KIT D816V+ advanced systemic mastocytosis monitored using a secreted luciferase. Oncotarget. 2016;7(50):82985-83000.

1771

GUIDELINE ARTICLE Ferrata Storti Foundation

Haematologica 2018 Volume 103(11):1772-1784

European Myeloma Network recommendations on tools for the diagnosis and monitoring of multiple myeloma: what to use and when

Jo Caers,1,2 Laurent Garderet,3 K. Martin Kortüm,4 Michael E. O’Dwyer,5 Niels W.C.J. van de Donk,6 Mascha Binder,7 Sandra Maria Dold,8 Francesca Gay,9 Jill Corre,10 Yves Beguin,1,2 Heinz Ludwig,11 Alessandra Larocca,9 Christoph Driessen,12 Meletios A. Dimopoulos,13 Mario Boccadoro,9 Martin Gramatzki,14 Sonja Zweegman,6 Hermann Einsele,4 Michele Cavo,15 Hartmut Goldschmidt,16,17 Pieter Sonneveld,18 Michel Delforge,19 Holger W. Auner,20 Evangelos Terpos13 and Monika Engelhardt8

Department of Hematology, University Hospital of Liege, Belgium; 2Laboratory of Hematology, GIGA-I3, University of Liège, Belgium; 3Department of Hematology, Hopital Saint Antoine, Paris, France; 4Department of Internal Medicine II, University Hospital of Wuerzburg, Germany; 5Department of Hematology, National University of Ireland Galway, Ireland; 6 Department of Hematology, VU University Medical Center, Amsterdam, the Netherlands, 7Department of Internal Medicine II, University Medical Center HamburgEppendorf, Hamburg, Germany; 8Department of Medicine I, Hematology, Oncology & Stem Cell Transplantation, Medical Center, Faculty of Medicine, University of Freiburg, Germany; 9 Department of Hematology-Oncology, University Hospital Città della Salute e della Scienza, Torino, Italy; 10Unit for Genomics in Myeloma, Institut Universitaire du Cancer – Oncopole, Toulouse, France; 11Wilhelminen Cancer Research Institute, Vienna, Austria; 12Department of Oncology and Hematology, Cantonal Hospital St. Gallen, Switzerland; 13School of Medicine, National and Kapodistrian University of Athens, Greece; 14Division of Stem Cell Transplantation and Immunotherapy, University of Kiel, Germany; 15Seragnoli 'Institute of Hematology, Bologna University School of Medicine, Italy; 16Department of Hematology, Rheumatology and Oncology, University Hospital Heidelberg, Germany; 17National Center for Tumor Diseases, Heidelberg Medical University, Germany; 18Department of Hematology, Erasmus MC Cancer Institute, Rotterdam, the Netherlands; 19Department of Hematology, University Hospital Leuven, Belgium and 20Centre for Haematology, Hammersmith Hospital, Imperial College London, UK 1

Correspondence: jo.caers@chu.ulg.ac.be

Received: January 21, 2018. Accepted: August 27, 2018. Pre-published: August 31, 2018.

doi:10.3324/haematol.2018.189159 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1772 ©2018 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.

1772

ABSTRACT

he diagnosis of multiple myeloma can be challenging, even for experienced physicians, and requires close collaboration between numerous disciplines (orthopedics, radiology, nuclear medicine, radiation therapy, hematology and oncology) before the final diagnosis of myeloma is made. The definition of multiple myeloma is based on the presence of clinical, biochemical, histopathological, and radiological markers of disease. Specific tests are needed both at presentation and during follow-up in order to reach the correct diagnosis and characterize the disease precisely. These tests can also serve prognostic purposes and are useful for follow-up of myeloma patients. Molecular analyses remain pivotal for defining high-risk myeloma and are used in updated patient stratifications, while minimal residual disease assessment via flow cytometry, molecular techniques and radiological approaches provides additional prognostic information on patients’ long-term outcome. This pivotal information will guide our future treatment decisions in forthcoming clinical trials. The European Myeloma Network group updated their guidelines on different diagnostic recommendations, which should be of value to enable appropriate use of the recommendations both at diagnosis and during follow-up.

haematologica | 2018; 103(11)

EMN recommendations on MM diagnosis and monitoring

Introduction The classification and differential diagnosis of monoclonal gammopathies is based on clinical, biological and radiological criteria but remains challenging in certain cases. Multiple myeloma (MM) is the most common malignant gammopathy and is associated with a wide spectrum of signs and symptoms.1 In the past decade, the treatment options for patients with MM have increased considerably. Together with improved supportive care, these new regimens significantly prolong the survival of both younger and older patients.2 The 2014 revision of the diagnostic criteria for MM allows the initiation of treatment in patients defined only by biomarkers, annotated as SLIM criteria [bone marrow (BM) infiltration >60%, involved/uninvolved serum free light-chain (SFLC) ratio >100 or >1 focal lesion >5 mm as determined by magnetic resonance imaging (MRI)], without waiting for conventional CRAB criteria (hypercalcemia, renal impairment, anemia, bone disease) to occur.3,4 Both the SLIM biomarker and CRAB criteria are listed in Figure 1. Given the recent evolution in diagnosis and response assessment, members of the European Myeloma Network (EMN) agreed to review and recommend diagnostic and response criteria to allow their discriminating use in daily practice and current care of patientsâ&#x20AC;&#x2122; .

Methodology These recommendations were developed by a panel of clinical experts on MM based on evidence of published data through August 2017. Expert consensus was used to suggest recommendations, where sufficient data were lacking. The final recommendations were classified based

on the GRADE criteria,5 which incorporates the strength and quality of evidence (Online Supplementary Table S1). Based on discussions at the 2017 EMN Trialist meeting (Baveno, Italy) guidelines were prepared and circulated among all panel members. The manuscript subsequently underwent revision in three rounds until the EMN experts reached consensus. In line with the guidelines of the International Committee of Medical Journal Editors, authorship was based on active contribution during discussion, writing and revision of the guidelines.

European Myeloma Network recommendations Diagnostic tools Blood tests A defining feature of plasma cell (PC) disorders is the secretion of monoclonal immunoglobulins, often referred to as a monoclonal M-protein, which can be used as a diagnostic marker, but also for the follow-up of the disease. Its heavy- and light-chain components can be identified by immunofixation and further quantified by serum protein electrophoresis and/or a serum free light-chain (SFLC) assay. It should be kept in mind that with persisting disease and possible de-differentiation of myeloma cells, the capacity to produce M-proteins may decrease or be completely lost (light-chain escape). In those patients, low M-protein levels, the presence of light chains only, or even complete absence of M-proteins and light chains may be mistaken as an ongoing or evolving response.6 Of note, immunofixation is approximately 10-fold more sensitive than serum protein electrophoresis and is required at diagnosis to characterize the phenotype of the M-protein, and for confirmation of a complete response, which is defined as being immunofixation-negative.7

Figure 1. The differential diagnosis between monoclonal gammopathy of undetermined significance, smoldering myeloma and multiple myeloma. The discrimination between these monoclonal gammopathies is based on: (i) the plasma cell infiltration in the bone marrow, (ii) the presence of clinical symptoms related to myeloma disease and (iii) the existence of biomarkers of disease that allow initiation of treatment. MGUS: monoclonal gammopathy of undetermined significance; SMM: smoldering multiple myeloma; MM: multiple myeloma; BM: bone marrow; PC: plasma cells; FLC: free light chain; MRI: magnetic resonance imaging.

haematologica | 2018; 103(11)

1773

J. Caers et al.

Serum electrophoresis and immunofixation may not be able to detect light-chain aberrations in patients with oligo-secretory disease, such as light-chain MM. Due to their low molecular weight, these SFLC are rapidly cleared by the kidneys. In such cases, the monoclonal burden should be measured in a 24 h urine collection or in the serum by an automated SFLC immunoassay (Grade 1A), the latter having a higher sensitivity to detect and quantify the involved free light chains.7 In concordance with the International Myeloma Working Group (IMWG), we recommend the performance of serum immunofixation and electrophoresis on serum and urine samples and a SFLC assay for the diagnosis of a monoclonal PC disorder (Grade 1A). Additional laboratory tests should be performed for the diagnosis and follow-up of MM patients, such as a complete blood count to evaluate possible cytopenias, blood smears to look for circulating PC and general biochemistry tests (renal and liver function tests, calcium, phosphate, uric acid, albumin, creatinine, lactate dehydrogenase, C-reactive protein, β2-microglobulin). Quantification of serum immunoglobulins by nephelometry enables an indirect measurement of the M-protein or the recognition of a secondary hypogammaglobinemia and is recommended for any patient presenting with a gammopathy (oligo-, poly- or monoclonal) (Grade 1A). The HevyLite® immunoassay quantifies both the involved and uninvolved intact immunoglobulin chains and quantifies them separately (IgGκ/λ, IgAκ/λ, IgMκ/λ).8 This assay has prognostic value for progression-free and overall survival9 and seems particularly useful when the Mprotein is difficult to measure via serum protein electrophoresis. This assay is not yet part of the routine workup of MM patients, but may be of value in the followup of patients and has been included in clinical trials.10

Urine analysis Proteinuria should be assessed on urine samples from all patients at diagnosis and during the follow-up. If proteinuria is present, it should be quantified in a 24 h urine collection. Total 24 h protein and Bence-Jones proteinuria should be evaluated by densitometry, electrophoresis and immunofixation. To detect low amounts of monoclonal proteins, it is recommended that the urine is concentrated 200-fold.11 These 24 h urine collections are often inconsistently performed, resulting in incomplete urine collection. In addition, renal function can influence the accuracy of the results, a fact which should be taken into consideration when interpreting laboratory values. The SFLC assay can be used for the follow-up of patients with light-chain MM. A recent French study, focusing on patients with light-chain MM, demonstrated that the SFLC assay is superior to 24 h urine collection for: (i) identifying patients with measurable disease, (ii) following their response to initial therapy, and (iii) giving a prognostic indication of the patients’ response and overall survival.12 This study included 113 patients with light-chain MM, all of whom had an abnormal SFLC ratio and measurable disease parameters in serum, while only 64% patients had measurable M-proteins in the urine, as determined by urine protein electrophoresis. Similar results were found in 576 patients with light-chain MM from the UK Myeloma IX and XI trials. The disease burden of the patients with light-chain MM could be measured and monitored by urine protein electrophoresis in 80% of 1774

cases. Of the remaining patients 113 (97%) had involved free light chains >100 mg/L, which was sufficient to measure response to therapy.13 These two studies confirmed the importance of SFLC measurements to diagnose and monitor patients with light-chain or oligosecretory MM. The replacement of urine studies by the SFLC assay for all myeloma patients remains controversial, since an Eastern Cooperative Oncology Group study on 399 MM patients (of whom only a minority had lightchain MM disease) found only a weak correlation between results of the SFLC assay and 24 h protein analysis.14 In line with the IMWG guidelines,15 we recommend the SFLC assay for the diagnosis and monitoring of patients with oligosecretory disease (Grade 2B). However, for patients with measurable urinary M-proteins, MM should be monitored by 24 h urine collections. When albumin is the dominant protein found in the urine, a glomerulopathy (such as AL-amyloidosis or light-chain deposition disease) should be excluded. The 24 h urine collection remains important when results are discordant.

Bone marrow studies A BM aspirate enables quantification of infiltrating PC and cytogenetic studies on purified PC. Unfortunately, dilution by peripheral blood during aspiration or the presence of patchy disease (uneven distribution of MM cells throughout the BM) may result in an underestimation of tumor infiltration.16 We therefore recommend an additional BM trephine biopsy, which may generate complementary information (Grade 1B). A BM biopsy correctly identified MM disease in 95% of symptomatic patients with a low PC count on the initial BM smears.16 The correct quantification of BM PC is also important because of the 60% cut-off as a biomarker of malignancy. The IMWG earlier recommended retaining the highest PC infiltration in case of discrepancy. Finally, the monoclonality of PC in the diagnostic sample should be confirmed by multiparameter flow cytometry or by immunohistochemistry confirming light-chain restriction. Flow cytometry of bone marrow cells In cases of monoclonal gammopathies, the most relevant information provided by multiparameter flow cytometry is the identification and enumeration of neoplastic versus polyclonal BM PC. Regardless of the disease category, these neoplastic PC share similar immunophenotypic features, which are distinct from those of normal PC. Typically, CD38, CD138 and CD45 (together with light scatter characteristics) are the best backbone markers for the discrimination of PC. In addition, expression of CD19, CD56, CD117, CD20, CD28, CD27 and CD81, together with cytoplasmic immunoglobulin light-chain restriction, allows a clear discrimination between normal/reactive versus monoclonal PC17 and was used by the EuroFlow consortium to create a standardized panel allowing the quantification and immunophenotypic characterization of neoplastic PC.18 Due to dilution and the sometimes patchy disease distribution, multiparameter flow cytometry often underestimates the infiltration but remains important for detection of monoclonal PC in the peripheral blood and for the detection of minimal residual disease (MRD) in the BM. The Mayo Clinic group reported on the prognostic importance of circulating neoplastic cells in patients with haematologica | 2018; 103(11)

EMN recommendations on MM diagnosis and monitoring

newly diagnosed or relapsing MM.19,20 They recently monitored circulating MM cells at diagnosis and after induction therapy by multiparameter flow cytometry and confirmed inferior progression-free and overall survival for patients with persistent circulating MM cells before transplantation.21

Molecular studies Cytogenetics MM remains a heterogeneous disease with some patients progressing rapidly, while others survive more than 10 years. This clinical diversity is mainly driven by genetic abnormalities affecting the biological characteristics of MM cells.22 These alterations, summarized in Table 1, are important prognostic factors and can be divided into primary, disease-initiating abnormalities (hyperdiploidy and translocations involving the IGH locus) and secondary events, related to further progression of the disease.23 Fluorescence in situ hybridization on interphase cells, performed after purification of CD138+ cells or after counterstaining for the monoclonal light chains, is the technique required to detect these abnormalities.24 Alternative techniques that can be used are singlenucleotide polymorphism arrays, which are able to detect loss of heterozygosity and numerical chromosome abnormalities, and comparative genomic hybridization arrays, which mainly reveal numerical abnormalities. Up to 65% of patients with MM have translocations that involve the immunoglobulin heavy chain gene (IGH) on chromosome 14q32. The prevalence and prognostic impact of these IGH translocations vary according to the partner chromosome (Table 1). Hyperdiploidy generally consists of numerical gains (of the odd chromosomes)

with a few structural changes, and is usually associated with longer overall survival. Not all trisomies have the same prognostic impact: trisomy 21 impairs, while trisomies 3 and 5 improve overall survival and may partially abrogate the negative impact of del17p and t(4;14).25 The most recurrent secondary alterations are deletion/monosomy of chromosome 13, deletion of chromosome 17p13, chromosome 1 abnormalities (1p deletions and 1q gains/amplifications), and C-MYC translocations. Deletion 17p13 is considered the most detrimental prognostic factor (due to short remission after high-dose therapy and an increased incidence of extramedullary disease) and is present in 8-10% of untreated patients.26 This deletion becomes clinically relevant when identified in the majority of PC. Different percentages (varying from 10%-60%) have been proposed to define a threshold that is associated with an impaired prognosis.27-29 The presence of a biallelic inactivation (i.e. by an additional mutation) of TP53 may particularly shorten overall survival.30 Aberrations of chromosome 1 (either 1q21 gains/amplifications or deletions of 1p32) are common and associated with shorter progression-free and overall survival, particularly the less frequent del(1p32).31 Patients with adverse cytogenetics may have additional aberrations: in the British MRC IX trial patients with two adverse cytogenetic lesions had a median overall survival of 2 years, while the survival of patients with three aberrations (an adverse IGH translocation, +1q21 and del17p13) was only 9 months.32 This inferior survival of patients with additional genetic abnormalities was also found in an Intergroupe Francophone du Myelome study that focused on patients with either del17p13 or t(4;14): for patients harboring t(4;14), multivariate analyses showed

Table 1. Recommended cytogenetic studies with implicated gene alterations and related prognosis.

Cytogenetics Del17p13

P53

Genetic event Frequency

Prognosis

5-15% Independent marker, with negative impact on PFS and OS

t(4;14)(p16.3;q32)

Gain 1q21

CKS1B

FGR3 MMSET

15%

Independent marker, with negative impact on PFS and OS

34-40% Independent marker,

Response to PI

Response to IMiD

Remarks

Negative prognostic factor

Negative prognostic factor Pomalidomide seems beneficial Unfavorable for any IMiD

Most important prognostic factor

Negative prognostic factor

Might be directly 101 implicated in bortezomib resistance 102

Good prognosis

103 Sensitive to venetoclax Considered as a 104 negative prognostic factor, but not confirmed in IFM study May neutralize the 25 negative prognostic impact of del17p or t(4;14)

Improves survival compared to classic agents Negative prognostic factor

CDKN2C

7-17%

t(11;14)(q13;q32)

CCND1

20%

with negative impact on PFS and OS Independent marker, with negative impact on PFS and OS Good prognosis

t(14;16)(q32;q23)

CMAF

2-3%

Controversial

60%

Standard prognosis, Standard prognostic unless associated with factor other negative prognostic markers

Del 1p32, Del 1p22

Hyperdiploidy of odd chromosomes

Negative prognostic factor Good prognosis

Ref 31, 60, 97

27, 98-100

PFS: progression free survival, OS overall survival, FGFR3: fibroblast growth factor receptor 3, MMSET: multiple myeloma SET domain, PI proteasome inhibitor, IMiD: immunomodulatory drug, CKS1B: CDC28 protein kinase regulatory subunit 1B, CDKN2C: cyclin-dependent kinase 4 inhibitor C, CCND1: cyclin-D1, MAF: musculoaponeurotic fibrosarcoma, IFM: Intergroupe Francophone du Myelome

haematologica | 2018; 103(11)

1775

J. Caers et al.

a shorter overall survival for patients with a combined del(13q14) or del(1p32). Among patients with del17p13, overall survival was shorter in those with del(1p32).33 Next-generation genome sequencing Next-generation sequencing allows the detection of baseline clonal heterogeneity,34 clonal tiding35 and linear and branching evolution and contributes to a better understanding of MM disease biology.36 The availability of more than 2000 sequenced MM genomes has essentially defined the genomic landscape. These data revealed a high incidence of clinically relevant genomic aberrations, including oncogenic RAS mutations, but also a number of rarer and potentially actionable lesions, such as BRAF mutations.37,38 Of note, the vast majority of available genomic data in MM is still derived from samples obtained at diagnosis and does not, therefore, necessarily reflect the situation during disease progression. In addition, the clinical relevance of most mutations has not yet been determined and is undergoing investigation in large sequencing programs (CoMMpass, The Myeloma Genome Project and others).39,40 No mutation screening has yet been implemented in standard clinical workflows, but mutational analyses may help to identify potential therapeutic targets (such as BRAF mutations) and to stratify of patients in clinical trials. Gene expression profiling Based on microarrays to study mRNA expression, gene expression profiling gives a global snapshot of disease biology and may help clinicians to classify patients into separate groups with distinct outcomes. The University of Arkansas pioneered this technique to stratify MM patients

and to characterize individuals’ disease at the molecular level.41 They identified gene expression profiling patterns that allowed MM patients to be grouped in seven disease classes. Further correlation of their microarray results with survival data of individual patients identified a list of 70 genes (GEP70) that had strong prognostic information.41 Similarly, the HOVON group identified a 92-gene signature (termed SKY92), based on the gene expression profiling results of the Hovon-65 trial.42,43 Other gene expression profiling-based risk models have been developed, such as the IFM-15 and MRC-IX-6 gene signatures44,45 Although not routinely determined in the majority of laboratories within or outside Europe, both the GEP70 and the SKY92 profiles are commercially available.

Imaging Traditionally, osteolytic bone disease was investigated by conventional skeletal radiography. The 2014 IMWG disease criteria also considered small osteolytic lesions (≥5 mm), detected by computed tomography (CT) or combined 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET/CT) as symptoms of myeloma-induced bone disease.3 Taking into account these definitions, in 2015 the EMN proposed a relevant algorithm for guiding the choice of imaging technique.46,47 Different European centers have integrated CT into their diagnostic work-up based on its superior sensitivity and its ease of operation This choice was supported by the recent implementation as a national standard of care in the diagnostic workup of patients with suspected MM in the UK and elsewhere.48 Whole-body CT has also been integrated into the diagnostic work-up of the European Society of Medical Oncology49 and the upcoming IMWG guidelines.

Table 2. Recommendations on further examinations at diagnosis, for response assessment, during follow-up and at relapse.

Diagnostic site

Bone marrow

Blood

Urine

Imaging

Tool

Diagnosis

At response

At follow-up

At relapse

BM cytology and biopsy to confirm plasmacytosis and monoclonality Flow cytometry Cytogenetics Advanced techniques: GEP, NGS Blood count and blood smear Serum electrophoresis and IF Serum free light chain Serum immunoglobulin levels Renal and liver function tests Calcium Lactate dehydrogenase Albumin, β2-microglobulin Urine sample to check for proteinuria and Bence-Jones proteins 24 h urine collection Low dose whole-body CT PET/CT Whole-body MRI

Obligatory

Obligatory*

Not required

Obligatory**

Recommended Obligatory Optional Obligatory Obligatory Recommended *** Obligatory Obligatory Obligatory Obligatory Obligatory Obligatory

Optional Not required Not required Obligatory Obligatory Recommended *** Obligatory Obligatory Obligatory Obligatory Recommended Obligatory

Not required Not required Not required Obligatory Obligatory Recommended *** Obligatory Obligatory Obligatory Obligatory Recommended Obligatory

Optional Optional Not required Obligatory Obligatory Recommended *** Obligatory Obligatory Obligatory Obligatory Obligatory Obligatory

Recommended† Recommended†† Optional Optional

Recommended† Not required Optional††† Not required

Recommended† When symptomatic When symptomatic When symptomatic

Recommended† Recommended Optional Optional

BM: bone marrow; GEP: gene expression profiling; IF: immunofixation; NGS: next generation sequencing; CT: computed tomography; PET: positron emission tomography; MRI: magnetic resonance imaging; *Obligatory for patients in complete response. **Obligatory for patients with light chain escape, oligosecretory disease, *** SFLC monitoring is obligatory for patients with light-chain disease. †Obligatory in the case of proteinuria. ††Obligatory when radiographs do not show osteolytic lesions †††PET/CT is required for confirmation of minimal residual disease negativity.

1776

haematologica | 2018; 103(11)

EMN recommendations on MM diagnosis and monitoring

The risk of pathological fractures or neurological complications should be assessed in patients with lytic lesions. In this regard, MRI is the preferred examination to detect spinal cord compression. If whole-body CT is not available, conventional radiographs can still be used but must be interpreted with their limited sensitivity in mind. In asymptomatic patients without lytic lesions, axial MRI or whole-body MRI should be considered to assess the presence of focal lesions (Grade 1B). Addition of dynamic contrast enhancement or diffusion weighted imaging to a whole-body MRI protocol provides additional information on BM vascularization, cellularity, and composition and improves the sensitivity of MRI.50,51 Two or more focal lesions on MRI are considered as a MM-defining biomarker.52 18F-FDG PET/CT can replace whole-body CT, if image acquisition of CT allows a detailed evaluation of the bone structures from vertex to knees, including both arms (Grade 1B).53 18F-FDG PET/CT is important to assess the presence of extramedullary disease, known to be an independent prognostic factor.54 The integration of MRI, PETCT and whole-body CT always requires experience, interdisciplinary consensus and reflection and needs to be correlated with blood, urine and BM results. Finally, baseline 18 F-FDG PET/CT scans enable post-treatment follow-up of hypermetabolic regions with a greater sensitivity than MRI.54,55 PET/CT is also useful in confirming MRD.56 European Myeloma Network recommendations for the diagnosis of multiple myeloma: The initial work-up should include: complete blood count, kidney function tests, serum protein electrophoresis with immunofixation, serum albumin, β2-microglobulin, lactate dehydrogenase, C-reactive protein, calcium, serum free light chains (especially useful in the case

of light-chain multiple myeloma), 24 h protein collection with protein quantification, electrophoresis and urine immunofixation, and bone marrow (aspiration only is acceptable) studies to quantify and characterize abnormal plasmacytosis (Table 2). The intervals between follow-up studies depend on the response obtained and the patients’ characteristics, as proposed in Table 3 (Grade 2C). After CD138+ plasma cell sorting, fluorescence in situ hybridization analysis should include at least t(4;14) and del17p; analysis of t(14;16), 1q21 gain and del(1p32) are also recommended. In addition, bone integrity needs to be evaluated with whole-body computed tomography and/or whole-body magnetic resonance imaging (at least axial). Quantification of the level of plasma cell infiltration, serum free light chains and magnetic resonance imaging assessment are required to assess the SLIM-CRAB biomarkers that define early active multiple myeloma. At relapse, the extent of myeloma-induced bone disease should be re-evaluated, especially if the relapse occurs late after the initial diagnosis.

Staging and prognosis Disease-specific prognostic scores The variable outcome of MM patients depends on differences in disease biology, global disease burden and the health status of the patient. Researchers have developed clinical scoring systems in order to estimate individual prognosis. The degree of anemia, renal failure and osteolysis were the first disease-related prognostic biomarkers described in MM and were all included in the Salmon & Durie staging system. Subsequently, β2-microglobulin, albumin and C-reactive protein levels, and proliferative activity of MM cells were found to be additional prognostic factors, and albumin and β2-microglobulin levels were incorporated in the International Staging System (ISS) in

Table 3. Follow-up of multiple myeloma patients according to response and patients’ characteristics (general strength of these recommendation GRADE 2C).

Patient risk group

General myeloma population

Response to prior treatment

Blood and urine tests

Imaging

In CR or VGPR

Follow-up with blood and urine samples, initially every 1-3 months* with gradually increasing intervals (max 6 months) Follow-up with blood and urine samples, initially every 1-2 months* with gradually increasing intervals (max 6 months) Regular follow-up with blood and urine samples, initially every month. Consider treatment initiation for patients with high-risk disease

Imaging studies should be performed when there are signs of bone disease. Consider PET/CT for high-risk patients

In PR

With biological progression

Frail patients Special patient groups

LC-MM and patients with renal failure Patients with extramedullary disease

Follow-up with blood and urine samples, every 2 months; if stable increase intervals to 3 months. Involve family doctor for follow up Blood, SFLC and 24 h urine collection. Follow-up with SPEP, UPEP and SFLC every month; if stable, increase intervals to 3 months

Imaging studies should be performed when there are signs of bone disease. Consider PET/CT for high-risk patients. Imaging using the whole body approach is recommended.

Imaging directed to the affected region only when signs of progressive bone disease. Imaging studies should be performed when there are signs of bone disease PET/CT is the preferented technique for follow-up. Recommended every 6 months and obligatory in the case of progression

CR: complete response;VGPR: very good partial response; PR: partial response, PET/CT: positron emission tomography–computed tomograph; max: maximum , SPEP: serum protein electrophoresis; UPEP: urine protein electrophoresis; SFLC: serum free light chain, LC-MM: light-chain multiple myeloma, 24 h: 24 hours * The monitoring intervals are generally shorter for patients with high-risk disease (monthly follow-up) than for patients with standard-risk disease (every 2 to 3 months).

haematologica | 2018; 103(11)

1777

J. Caers et al. Table 4. The Revised-International Staging System is one of the best stratification methods; it is based on routinely available cytogenetic and biochemistry tests (Palumbo et al.).56

R-ISS definitions

Determinants

R-ISS stage I

ISS stage I, no high-risk CA, and normal LDH Other combinations ISS stage III plus high-risk CA or high LDH

R-ISS stage II R-ISS stage III

Number

OS (5 years)

Median OS

PFS (5 years)

Median PFS

871 (28%)

82%

55%

66 months

1894 (62%)

62%

83 months

36%

42 months

295 (10%)

40%

43 months

24%

29 months

R-ISS Revised- International Staging System; ISS: International Staging System; OS: overall survival; PFS: progression free survival; CA; cytogenetic abnormalities; LDH: lactate dehydrogenase; NR: not reported.

2005. Other biomarkers have been identified and include markers of tumor cell proliferation, cytokines, pro-angiogenic factors, indicators of bone remodeling, circulating (exosomal) miRNA and imaging abnormalities. Another promising biomarker is the serum level of shed B-cell maturation antigen, which correlates well with BM PC infiltration and declines according to tumor response.57 Follow-up of serum B-cell maturation antigen levels seems of interest in patients with non-secretory disease, for whom we lack reliable parameters in the blood; future studies are therefore warranted. Apart from the often still reported Salmon & Durie staging system, the ISS and revised (R)-ISS are frequently used as staging systems; in the latter systems, the β2microglobulin and albumin levels reflect patients' tumor burden, turnover rate, presence of renal impairment, and nutritional and performance status.58 In order to improve the prognostic performance of the ISS score, the IMWG updated it, adding high-risk cytogenetics [t(4;14), t(14;16), and del17p determined by interphase fluorescence in situ hybridization] and elevated serum lactate dehydrogenase (Table 4).59 These factors had been previously identified as relevant risk factors for early progression after autologous transplantation.60 Of note, the ISS and R-ISS give prognostic information at diagnosis, but have not been validated in relapsed/refractory MM.

Frailty and co-morbidities Once a treatment has started, adherence to the established protocol remains a major clinical concern in elderly and frail patients. This requires an individualized approach in which therapeutic decisions should be driven by both disease features and the patient’s characteristics. As in other malignancies, comprehensive geriatric assessments have been evaluated to assess patients’ functional, cognitive and mental status, comorbidities, nutrition and presence of geriatric symptoms. Palumbo and co-workers developed a retrospective simplified geriatric assessment, named the IMWG-Frailty Index, in which age, the Charlson Comorbidity Index, activities of daily living and instrumental activities of daily living were used to discriminate between fit, intermediate-fit and frail patients, showing different incidences of severe adverse events, progression-free survival and overall survival.61 Expectedly, more severe adverse events and treatment discontinuations were reported in frail patients. The most extensive retrospective and prospective tests and validation analyses were performed within the German study group, who prospectively assessed the IMWG-Frailty Index with the revised Myeloma Comorbidity Index (R-MCI) and other comorbidity indices.62 A second prospective German 1778

study including 801 MM patients determined that impaired renal and pulmonary function, poorer Karnofsky performance status, frailty and age were independent, multivariate risk factors for overall survival. Addition of cytogentic abnormalities resulted in the weighted revised Myeloma Comorbidity Index, which is able to assess patients' physical condition accurately and is simple to apply in the clinic.63 Although not yet proven via randomized treatment algorithms, there is circumstantial evidence that limited induction therapy, careful dose modifications and reductions, sensible use of supportive care and watchful surveillance of unfit and frail patients may improve patients’ outcome further.64 The EMN insists on developing trials, specifically designed for frail patients, for further refinement of frailty-related diagnostics and best treatment selection.65

Drug-related biomarkers While prognostic factors regarding disease evaluation are listed above, drug-related biomarkers are being assessed to predict response to treatment and, possibly, to facilitate optimal treatment while avoiding ineffective therapies and unnecessary toxicity. Recent pharmacogenomic studies revealed gene signatures that could predict the clinical outcome after treatments based on immunomodulatory drugs66 or bortezomib.67 The expression of cereblon, an intracellular binding partner of immunomodulatory drugs has been intensively studied as a biomarker and initial studies correlated cereblon levels with the outcome of MM patients receiving treatment with such drugs.68-71 These studies used quantitative real time PCR analysis, gene expression profiling or immunohistochemistry to quantify cereblon expression and showed that loss of cereblon expression was associated with resistance to immunomodulatory drugs. Further investigations revealed limitations of these assays, because both splice variants of cereblon and point mutations were described.72,73 When exploring predictors for tumor responses to daratumumab, higher CD38 expression was found on MM cells of responsive patients. However, good responses were also seen in patients with lower CD38 expression, an observation confirmed in a second study that showed that CD38 expression level was not necessarily predictive of response in advanced MM; nevertheless attempts to assess agents that keep CD38 upregulation increased, e.g., all-trans retinoic acid and histone deacetylase inhibitors, are being pursued pre-clinically and clinically.74,75 Finally, expression of anti-apoptotic proteins, BCL-2, BCL-XL or MCL-1, measured by quantitative real-time PCR, predict pharmacological responses to the bcl-2 inhibitor venetoclax, which is mostly active in patients harboring t(11,14) translocations.76 haematologica | 2018; 103(11)

EMN recommendations on MM diagnosis and monitoring

European Myeloma Network recommendations: The International Staging System score and, whenever possible, the Revised International Staging System score, should be determined at diagnosis to assess prognosis. At least a minimal frailty assessment should be performed to aid the choice of induction therapy, dose amendments and supportive care. Although of interest due to their prognostic and predictive value, biomarkers, such as cereblon and CD38 protein expression, are not routinely assessed in daily multiple myeloma care, while fluorescence in situ hybridization for t(11;14) should be performed if treatment with venetoclax is a clinical option.

Response assessment The implication of the results of an SFLC assay and MRD assessment prompted the IMWG to update MM response criteria.77 In 2011, two new categories, stringent complete response and very good partial response, were added. Correct disease assessment is not only crucial for reporting in clinical trials, it also indicates prognosis in individual cases.77,78 It is well known that patients who obtain a complete response following induction have improved progression-free and overall survival after intensive treatment.79 Patients should, therefore, be evaluated before initiation of each treatment cycle to determine their response to therapy. For MM patients with intact immunoglobulins, the recommended method for monitoring is quantification of serum and urinary M-protein. Whether all serum (and urine) parameters have to be checked after each cycle, rather than after every two or three cycles is left to the discretion of each physician, taking into account disease aggressiveness, organ (i.e. renal) impairment and various other factors. To confirm a stringent complete response, normalization of the SFLC values and disappearance of monoclonal PC infiltration in the BM should be added to negative immunofixation on serum and urine samples. The BM must be evaluated in order to confirm a complete response, but this can be done at some time after the end of treatment, allowing full recovery of the BM. Of note, BM infiltration can be heterogeneous with persisting focal lesions in an otherwise recovered BM (earlier referred to as patchy disease). The follow-up of patients with light-chain MM and measurable M-protein levels in urine should include 24 h urine collections. The SFLC assay generally allows response assessment in patients with oligosecretory disease with unmeasurable serum and urine M-protein levels [serum M-protein <1 g/dL (10 g/L) or urine M-protein <200 mg/24 h]. If the SFLC assay is not informative, BM plasmacytosis should be assessed.77 The definition of relapse applies to a patient in complete response who experiences reappearance of MM, while progression refers to patients with an increasing disease burden from a baseline or persistent residual disease. An additional assessment for confirmation is mandatory before initiating a new line of therapy. MM progression can be determined biochemically (increase in an existing monoclonal peak), or by radiological and clinical criteria. The interested reader can find the criteria for relapse and disease progression recently described by the IMWG.77 Response assessment can be challenging, especially in cases of deep response after the use of monoclonal antibodies, which may interfere with quantification of the M-protein and may require specific assays.80

Minimal residual disease Current induction regimens, in association with autologous stem cell transplantation, achieve very high haematologica | 2018; 103(11)

response rates and the responses are often deep. Unfortunately, however, MM often recurs due to residual MM cells, drug resistance and/or persistence of resistant dormant subclones.81 MRD can be assessed by multiparameter flow cytometry, polymerase chain reaction (PCR)based methods or next-generation sequencing to identify persistent clonal cells. Recent studies, listed in Table 5, confirmed the prognostic impact of MRD status as an independent variable for outcome.82 In the future, MRD will be more widely used in clinical trials to guide treatment choices and probably as a surrogate marker for progression-free and overall survival.56 Conventional flow-MRD approaches, based on multiple institutional non-standardized protocols, can reliably identify malignant PC and discriminate aberrantly expressed cell surface markers in approximately 90% of patients (with a sensitivity of detecting 10−4 atypical PC in normal BM). Recent studies conducted by Spanish and UK groups have shown that negative MRD by multiparameter flow cytometry is predictive for both progressionfree survival and overall survival, even in patients who achieved a complete response.83,84 Recent technical advances have increased the sensitivity of next-generation flow cytometry protocols down to the 10–6 range.85 MRD analysis by PCR detects persistent residual tumor cells through the amplification of a tumor-specific molecular marker. The IGH rearrangement is used as a marker of clonality in various B-cell malignancies.86 Allele-specific oligonucleotide PCR with primers complementary to the heavy chain variable sequence remains one of the most sensitive approaches to detect residual malignant PC, reaching a sensitivity of 10−5.87 Unfortunately, it is a laborious, time-consuming approach that is not widely available because of its dependence on patient-specific primers and probes for quantitative PCR. Next-generation sequencing of the IGH rearrangement segments provides insights into the architecture of the Blineage repertoire with consensus primers. Since the Blineage repertoire includes the malignant PC clone in BM, next-generation sequencing of IGH enables a quantitative determination of MRD, without per-patient customization, provided that the malignant clone was identified in a diagnostic sample or a sample taken during active disease.88 Results from next-generation sequencing are highly concordant with flow-based MRD detection, highly reproducible and reach a sensitivity of 10−6.89-91 A lack of standardization and limited commercial availability are the main restraints for next-generation sequencing. Flow cytometry and molecular techniques both require an appropriate BM sample. Heterogeneous BM infiltration and peripheral blood dilution can be major hurdles to the evaluation of MRD. Since neither of these techniques is able to detect extramedullary disease, they should be combined with imaging studies. Imaging is a third approach to evaluate MRD in MM. Both PET/CT and MRI have been evaluated in this setting.53,55 Regarding PET/CT, two large studies assessed the prognostic value of negative PET/CT after induction and autologous stem cell transplantation.55,92 Both studies found that PET/CT-negative patients had a better progression-free and overall survival compared to PET/CTpositive patients (52 versus 38 months and 5-year estimates of 90% versus 71%, respectively).92 In the French IMAJEM study, MRD was evaluated in 86 patients via PET-CT and flow cytometry. Although the concordance 1779

J. Caers et al. Table 5. Recent studies on minimal residual disease and the implications for progression-free and overall survival of patients with multiple myeloma.

N. Reference

Number of Method Study patients used for question MRD assessment

Patient cohorts

Time point of MRD assessment

MRD– status surpasses CR in all risk groups. MRD negativity most relevant endpoint for elderly fit pts with ASCT. 3 months post-ASCT 56% MRD– with superior OS MRD status important and PFS; del17 pts no difference markers for survival but in PFS and OS between differs according to sCR and MRD–; t(4;14) pts superior cytogenetics PFS and OS in MRD– pts than sCR 6 months MRD– in 85% 6 months after MRD and sCR excellent post-ASCT transplantation, sCR markers for PFS and OS in 56% and CR in 20% At ID after 9,18 Determines 3 MRD groups, MRD is a relevant cycles with significant longer prognostic factor in PFS and OS for MRD– group elderly and MRD status correlates with OS and PFS At baseline and OS and PFS longer in MRD surpasses at suspected CR MRD– vs. MRD+ pts conventional SFLC and with BM CR. Same results for BM CR MRD– in VGPR or PR group ID + pre-ASCT, 58% in CR, 68% MRD–. None of the MRD– group post-ASCT, relapsed during a FU of 39 post-consolidation, months end of treatment 6 cycles of 22% MRD– with longer PFS MRD surpasses CR and is induction and OS. VMP better than VTP. 70% a prognostic factor for OS in CR after VMP also MRD– only and PFS 45% in VTP group. After 3 and 6 Lower MRD in ASCT vs. CRD pts. MRD identifies a low-risk courses of Differences between HR vs. SR group, response maintenance and and relapse vs. non-relapse pts independently, better then every characterizes activity 6 months till PD of treatment. ID and after MRD assessed in 103 pts.; 54% ASO RQ-PCR less treatment MRD+ by PCR, 46% by MFC. applicable than MFC but MRD– pts had prolonged PFS powerful to assess and OS treatment efficacy and risk stratification ID, study entry, OS 72% at 8 years median FU Long-term MRD after 2 cycles in MRD– and 48% in MRD+ monitoring is useful and CTD, end of pts. PFS for MRD– 38 months maintenance therapy treatment, every and MRD+ 9 months ensures responses 6 months until PD After induction for MRD– pts had significantly MRD surpasses CR and elderly, after longer PFS and OS; median depth of MRD level ASCT for PFS: MRD≥10-3 27 months, showed significant <65 years 10-3-10-5 48 months; differences in pts with CR <10-5 80 months. Pts in CR or 29% of pts in CR are MRD–; MRD surpasses CR better after treatment ID, achievement nCR or better pts were High rates of MRD of CR and/or MRD– in 100% NDMM /92% negativity with longer PFS completion of SMM by MFC, 67%/75% by in NDMM/HR SMM cycles 8, 20 and 32, NGS and 41%/26% by end of treatment FGD-PET/CT

Lahuerta et al., JCO 2017.105

609

MFC

PFS and OS in MRD– pts

NDMM

Chakraborty et al., 185 Biol Blood Marrow Transplant 2017.106

MFC

PFS and OS in post-transplant sCR patients with HR cytogenetic MM

NDMM

Nadiminti et al., OncoTargets and Therapy 2017.107 Paiva et al.; Blood 2016.108

100

MFC

NDMM and pre-treated

162

MFC

Toxicity, safety, PFS and OS of VTD + HD melphalan Monitor MRD in transplant-ineligible pts

NDMM elderly

Ludwig et al.; BJH 2015.109

MFC

VTD and VTCD induction in NDMM

Untreated MM, elderly

Roussel et al.; JCO 2014.110

MFC

VRD induction and consolidation for ASCT pts

Untreated MM, <65 years

Mateos et al.; Blood 2014.111

260

MFC

VMP vs. VTP as induction

NMDD, >65 years

Oliva et al., Oncotarget 2017.112

Puig et al.; Leukemia 2014.113

170

ASO RQPCR; MFC

#10 Ferrero et al.; Leukemia 2015.114

PCR

#11 Martinez-Lopez 133 et al.; Blood 2014.90

#12 Chari et al.; Blood 2017.115

103

#13 Korde et al.; 45 JAMA Oncol. 2015.116

MFC and Consolidation with ASO-RQ-PCR ASCT or CRD plus R maintenance → MRD

MFC and NGS

NGS

Applicability, sensitivity and prognostic value of ASO RQ-PCR MRD kinetics’ impact on survival

Prognostic value of MRD in pts with VGPR after front-line therapy Safety, OS/PFS and MRD

MFC, NGS, Tolerability and FGD-PET/CT impact on MRD negativity

NDMM

NDMM with and without ASCT

NDMM with ASCT

NDMM

≥2 treatment lines NDMM and HR SMM

9 months after treatment

Results

Lesson learnt

MRD– prolonged PFS and OS; MRD+ in CR similar PFS and OS to MRD+ in nCR and PR.

continued on the next page

1780

haematologica | 2018; 103(11)

EMN recommendations on MM diagnosis and monitoring

continued from the previous page

N. Reference

Number of Method Study patients used for question MRD assessment

Patients cohort

Time points of MRD assessment Baseline, during post-treatment FU, PD

#14 Zagmani et al.; Clin Cancer Res 2015.92

FGD-PET/CT Role of FGD-PET/CT on PFS and OS

NDMM

#15 Patriarca et al., Biol Blood Marrow Transplant 2015.117

FGD-PET/CT Prognostic significance of PET/CT

AlloSCT pts Before and/or within 6 months after alloSCT.

#16 Lapa et al.; Oncotarget 2014.118

FGD-PET/CT Prognostic Pretreated, At relapse, value of FGD-PET/CT and after treatment progression after SCT

Results

PET-negativity post-treatment in 70%, CR in only 53%. PET-negativity influenced PFS and OS favorably. Persistence of EMD at transplantation → poor PFS. EMD and <CR or VGPR after alloSCT →shorter PFS/OS. PET CR pts prolonged PFS/OS. Absence of foci positive factor for PFS/OS. EMD and intense uptake associated with shorter PFS/OS. 30% of pts' management changed after PET/CT

Lesson learnt

PET-CT more careful evaluation of CR. PET-negativity independent predictor of prolonged PFS/OS PET/CT imaging significantly associated with outcome.

Prognostic value of PET/CT in post-SCT relapse patients important and significant impact on pt management

AlloSCT: allogenic stem cell transplantation; ASCT: autologous stem cell transplantation; ASO RQ-PCR: allele-specific oligonucleotide real-time quantitative polymerase chain reaction; BM: bone marrow; CR: complete remission; CRD: cyclophosphamide, lenalidomide, dexamethasone; d: day; EMD: extramedullary disease; FGD-PET/CT: fluorodeoxyglucose positron emission tomographycomputed tomography; FU: follow-up; HD: high dose; HR: high risk; ID: initial diagnosis; MFC: multiparameter flow cytometry; MM: multiple myeloma; MRD: minimal residual disease; nCR: nearcomplete remission; NGS: next generation sequencing; NDMM: newly diagnosed multiple myeloma; OS: overall survival; PCR: polymerase chain reaction; PD: progressive disease; PFS: progression free survival; PR: partial remission; pts: patients; R: lenalidomide; sCR: stringent complete remission; SCT: stem cell transplantation; SFCL: serum free light chain; SMM: smouldering multiple myeloma; SR: standard risk; TTP: time to tumor progression; VGPR: very good partial remission; VMP: bortezomib, melphalan, prednisone; VRD: bortezomib, lenalidomide, dexamethasone; vs.: versus; VTCD: bortezomib, thalidomide, cyclophosphamide, dexamethasone; VTD: bortezomib, thalidomide, dexamethasone; VTP: bortezomib, thalidomide, prednisone.

between the two tests was low, progression-free survival was better in patients who were negative according to both techniques compared to those who were positive by PET and/or flow cytometry (3-year progression-free survival, 86.8% versus 52.9%), indicating that both techniques are complementary. A major advantage of PET/CT is its capacity to assess MRD outside the BM; its disadvantages are high cost and the lack of reimbursement in certain countries, insufficient standardization and reduced tracer uptake in some MM patients. Evaluation via PET/CT has been incorporated into the new IMWG MRD criteria.77 In the future, the increased capabilities of diffusion weighted MRI to detect small lesions and diffuse infiltration may offer advantages that merit prospective evaluation in MRD assessment studies.50 In addition, recent studies have demonstrated that circulating DNA fragments carrying tumor-specific sequence alterations can be detected and quantified in the blood of patients with solid tumors.93,94 In MM, various studies have provided evidence that - much like in solid tumors - MM-specific alterations (VDJ rearrangements or somatic genomic alterations) can also be identified and tracked in cell-free DNA circulating in blood.95,96 European Myeloma Network recommendations: Response assessment is an essential part of myeloma management. Patients under treatment should be evaluated before the initiation of each cycle and according to international guidelines. Minimal residual disease testing is not currently recommended in routine follow-up of patients but is likely to be incorporated in standard response/progression evaluation soon. Valid options for the assessment of minimal residual disease are based on bone marrow cells (next-generation flow cytometry) or molecular analysis (next-

haematologica | 2018; 103(11)

generation sequencing), often also combined with an imaging-based evaluation. These methods require appropriate expertise.

Conclusion While novel agents have certainly improved the outcomes of patients with myeloma, prompt diagnosis and close follow-up of MM patients remain highly relevant and contribute to better survival. In most cases, the diagnosis of MM is straightforward, being based on biological and radiological evidence when evocative clinical signs are present. During response assessment, the evaluation of MRD will become increasingly important and, within the next few years, will guide treatment choices in clinical trials and possibly also outside trial scenarios. International efforts are needed to standardize the different techniques that can be used to evaluate MRD. Guidelines on appropriate follow-up and patient-tailored monitoring have been updated in this EMN consensus paper and should help to improve the outcome and prognosis of our patients. Acknowledgments The authors thank distinguished IMWG, EMN, DSMM and GMMG experts for their advice and recommendations that have helped us to improve this paper. This work was supported by the Deutsche Krebshilfe (grants 1095969 and 111424 to ME), the Foundation against Cancer, the Fonds National de la Recherche Scientifique and the Fonds d'Investissement de Recherche Scientifique (FIRS) du CHU de Liège (grants to JC), the NIHR Imperial Biomedical Research Centre and the Cancer Research UK Imperial Centre (grants to HWA).

1781

J. Caers et al.

References 1. Kumar SK, Rajkumar V, Kyle RA, et al. Multiple myeloma. Nat Rev Dis Primers. 2017;3:17046. 2. 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. 3. Rajkumar SV, Dimopoulos MA, Palumbo A, et al. International Myeloma Working Group updated criteria for the diagnosis of multiple myeloma. Lancet Oncol. 2014;15(12):e538-548. 4. Caers J, Fernández de Larrea C, Leleu X, et al. The changing landscape of smoldering multiple myeloma: a European perspective. Oncologist. 2016;21(3):333-342. 5. Guyatt G, Oxman AD, Akl EA, et al. GRADE guidelines: 1. Introduction-GRADE evidence profiles and summary of findings tables. J Clin Epidemiol. 2011;64(4):383-394. 6. Kuhnemund A, Liebisch P, Bauchmuller K, et al. 'Light-chain escape-multiple myeloma'an escape phenomenon from plateau phase: report of the largest patient series using LCmonitoring. J Cancer Res Clin Oncol. 2009;135(3):477-484. 7. Jenner E. Serum free light chains in clinical laboratory diagnostics. Clin Chim Acta. 2014;427:15-20. 8. Bradwell AR, Harding SJ, Fourrier NJ, et al. Assessment of monoclonal gammopathies by nephelometric measurement of individual immunoglobulin κ/λ ratios. Clin Chem. 2009;55(9):1646-1655. 9. Ludwig H, Milosavljevic D, Berlanga O, et al. Suppression of the noninvolved pair of the myeloma isotype correlates with poor survival in newly diagnosed and relapsed/refractory patients with myeloma. Am J Hematol. 2016;91(3):295-301. 10. Greil C, Ihorst G, Gaiser F, et al. The serum heavy/light chain immunoassay: a valuable tool for sensitive paraprotein assessment, risk, and disease monitoring in monoclonal gammopathies. Eur J Haematol. 2017;99 (5):449-458. 11. Katzmann JA, Kyle RA, Benson J, et al. Screening panels for detection of monoclonal gammopathies. Clin Chem. 2009;55(8): 1517-1522. 12. Dejoie T, Corre J, Caillon H, et al. Serum free light chains, not urine specimens, should be used to evaluate response in light-chain multiple myeloma. Blood. 2016;128(25):29412948. 13. Heaney JLJ, Campbell JP, Griffin AE, et al. Diagnosis and monitoring for light chain only and oligosecretory myeloma using serum free light chain tests. Br J Haematol. 2017;178(2):220-230. 14. Dispenzieri A, Zhang L, Katzmann JA, et al. Appraisal of immunoglobulin free light chain as a marker of response. Blood. 2008;111(10):4908-4915. 15. Dispenzieri A, Kyle R, Merlini G, et al. International Myeloma Working Group guidelines for serum-free light chain analysis in multiple myeloma and related disorders. Leukemia. 2008;23(2):215-224. 16. Lee N, Moon SY, Lee Jh, et al. Discrepancies between the percentage of plasma cells in bone marrow aspiration and BM biopsy: impact on the revised IMWG diagnostic criteria of multiple myeloma. Blood Cancer J. 2017;7:e530. 17. Rawstron AC, Orfao A, Beksac M, et al. Report of the European Myeloma Network on multiparametric flow cytometry in mul-

1782

18.

19.

20.

21.

22. 23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

tiple myeloma and related disorders. Haematologica. 2008;93(3):431-438. van Dongen JJM, Lhermitte L, Bottcher S, et al. EuroFlow antibody panels for standardized n-dimensional flow cytometric immunophenotyping of normal, reactive and malignant leukocytes. Leukemia. 2012;26(9):1908-1975. Gonsalves WI, Rajkumar SV, Gupta V, et al. Quantification of clonal circulating plasma cells in newly diagnosed multiple myeloma: implications for redefining high-risk myeloma. Leukemia. 2014;28(10):2060-2065. Gonsalves WI, Morice WG, Rajkumar V, et al. Quantification of clonal circulating plasma cells in relapsed multiple myeloma. Br J Haematol. 2014;167(4):500-505. Chakraborty R, Muchtar E, Kumar SK, et al. Serial measurements of circulating plasma cells before and after induction therapy has an independent prognostic impact in patients with multiple myeloma undergoing upfront autologous transplantation. Haematologica. 2017;102(8):1439-1445. Robiou du Pont S, Cleynen A, Fontan C, et al. Genomics of multiple myeloma. J Clin Oncol. 2017;35(9):963-967. Kuehl WM, Bergsagel PL. Molecular pathogenesis of multiple myeloma and its premalignant precursor. J Clin Invest. 2012;122(10):3456-3463. Ross FM, Avet-Loiseau H, Ameye G, et al. Report from the European Myeloma Network on interphase FISH in multiple myeloma and related disorders. Haematologica. 2012;97(8):1272-1277. Chretien M-L, Corre J, Lauwers-Cances V, et al. Understanding the role of hyperdiploidy in myeloma prognosis: which trisomies really matter? Blood. 2015;126(25):2713-2719. Billecke L, Murga Penas EM, May AM, et al. Cytogenetics of extramedullary manifestations in multiple myeloma. Br J Haematol. 2013;161(1):87-94. Avet-Loiseau H, Leleu X, Roussel M, et al. Bortezomib plus dexamethasone induction improves outcome of patients with t(4;14) myeloma but not outcome of patients with del(17p). J Clin Oncol. 2010;28(30):46304634. Neben K, Lokhorst HM, Jauch A, et al. Administration of bortezomib before and after autologous stem cell transplantation improves outcome in multiple myeloma patients with deletion 17p. Blood. 2012;119(4):940-948. Merz M, Hielscher T, Seckinger A, et al. Baseline characteristics, chromosomal alterations, and treatment affecting prognosis of deletion 17p in newly diagnosed myeloma. Am J Hematol. 2016;91(11):E473-E477. Thanendrarajan S, Tian E, Qu P, et al. The level of deletion 17p and bi-allelic inactivation of TP53 has a significant impact on clinical outcome in multiple myeloma. Haematologica. 2017;102(9):e364-e367. Avet-Loiseau H, Attal M, Campion L, et al. Long-term analysis of the IFM 99 trials for myeloma: cytogenetic abnormalities [t(4;14), del(17p), 1q gains] play a major role in defining long-term survival. J Clin Oncol. 2012;30(16):1949-1952. Boyd KD, Ross FM, Chiecchio L, et al. A novel prognostic model in myeloma based on co-segregating adverse FISH lesions and the ISS: analysis of patients treated in the MRC Myeloma IX trial. Leukemia. 2012;26(2):349-355. Hebraud B, Magrangeas F, Cleynen A, et al. Role of additional chromosomal changes in the prognostic value of t(4;14) and del(17p)

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

in multiple myeloma: the IFM experience. Blood. 2015;125(13):2095-2100. Keats JJ, Chesi M, Egan JB, et al. Clonal competition with alternating dominance in multiple myeloma. Blood. 2012;120(5):10671076. Egan JB, Shi C-X, Tembe W, et al. Wholegenome sequencing of multiple myeloma from diagnosis to plasma cell leukemia reveals genomic initiating events, evolution, and clonal tides. Blood. 2012;120(5):10601066. Melchor L, Brioli A, Wardell CP, et al. Singlecell genetic analysis reveals the composition of initiating clones and phylogenetic patterns of branching and parallel evolution in myeloma. Leukemia. 2014;28(8):1705-1715. Lohr Jens G, Stojanov P, Carter Scott L, et al. Widespread genetic heterogeneity in multiple myeloma: implications for targeted therapy. Cancer Cell. 25(1):91-101. Bolli N, Avet-Loiseau H, Wedge DC, et al. Heterogeneity of genomic evolution and mutational profiles in multiple myeloma. Nat Commun. 2014;5:2997. Bolli N, Li Y, Sathiaseelan V, et al. A DNA target-enrichment approach to detect mutations, copy number changes and immunoglobulin translocations in multiple myeloma. Blood Canc J. 2016;6:e467. Johnson DC, Weinhold N, Mitchell JS, et al. Genome-wide association study identifies variation at 6q25.1 associated with survival in multiple myeloma. Nat Commun. 2016;7:10290. Shaughnessy JD, Zhan F, Burington BE, et al. A validated gene expression model of highrisk multiple myeloma is defined by deregulated expression of genes mapping to chromosome 1. Blood. 2007;109(6):2276-2284. Kuiper R, Broyl A, de Knegt Y, et al. A gene expression signature for high-risk multiple myeloma. Leukemia. 2012;26(11):24062413. Kuiper R, van Duin M, van Vliet MH, et al. Prediction of high- and low-risk multiple myeloma based on gene expression and the International Staging System. Blood. 2015;126(17):1996-2004. Decaux O, Lodé L, Magrangeas F, et al. Prediction of survival in multiple myeloma based on gene expression profiles reveals cell cycle and chromosomal instability signatures in high-risk patients and hyperdiploid signatures in low-risk patients: a study of the Intergroupe Francophone du Myélome. J Clin Oncol. 2008;26(29):4798-4805. Dickens NJ, Walker BA, Leone PE, et al. Homozygous deletion mapping in myeloma samples identifies genes and an expression signature relevant to pathogenesis and outcome. Clin Cancer Res. 2010;16(6):18561864. Terpos E, Kleber M, Engelhardt M, et al. European Myeloma Network guidelines for the management of multiple myeloma-related complications. Haematologica. 2015;100 (10):1254-1266. Hillengass J, Moulopoulos LA, Delorme S, et al. Whole-body computed tomography versus conventional skeletal survey in patients with multiple myeloma: a study of the International Myeloma Working Group. Blood Canc J. 2017;7:e599. Excellence NIfHaC. Myeloma: diagnosis and management 2016 [cited; Available from: https://www.nice.org.uk/guidance/ ng35/resources/myeloma-diagnosis-andmanagement-pdf-1837394042821 Moreau P, San Miguel J, Sonneveld P, et al. Multiple myeloma: ESMO clinical practice

haematologica | 2018; 103(11)

EMN recommendations on MM diagnosis and monitoring

50. 51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2017;28(suppl_4):iv52iv61. Messiou C, Kaiser M. Whole body diffusion weighted MRI – a new view of myeloma. Br J Haematol. 2015;171(1):29-37. Hillengass J, Bauerle T, Bartl R, et al. Diffusion-weighted imaging for non-invasive and quantitative monitoring of bone marrow infiltration in patients with monoclonal plasma cell disease: a comparative study with histology. Br J Haematol. 2011;153(6):721-728. Hillengass J, Fechtner K, Weber M-A, et al. Prognostic significance of focal lesions in whole-body magnetic resonance imaging in patients with asymptomatic multiple myeloma. J Clin Oncol. 2010;28(9):16061610. Cavo M, Terpos E, Nanni C, et al. Role of 18F-FDG PET/CT in the diagnosis and management of multiple myeloma and other plasma cell disorders: a consensus statement by the International Myeloma Working Group. Lancet Oncol. 2017;18(4):e206-e217. Zamagni E, Patriarca F, Nanni C, et al. Prognostic relevance of 18-F FDG PET/CT in newly diagnosed multiple myeloma patients treated with up-front autologous transplantation. Blood. 2011;118(23):5989-5995. Moreau P, Attal M, Caillot D, et al. Prospective evaluation of magnetic resonance imaging and [18F]fluorodeoxyglucose positron emission tomography-computed tomography at diagnosis and before maintenance therapy in symptomatic patients with multiple myeloma included in the IFM/DFCI 2009 trial: results of the IMAJEM Study. J Clin Oncol. 2017;35(25):2911-2918. Anderson KC, Auclair D, Kelloff GJ, et al. The role of minimal residual disease testing in myeloma treatment selection and drug development: current value and future applications. Clin Cancer Res. 2017;23(15):39803993. Sanchez E, Li M, Kitto A, et al. Serum B‐cell maturation antigen is elevated in multiple myeloma and correlates with disease status and survival. Br J Haematol. 2012;158(6):727-738. Greipp PR, San Miguel J, Durie BG, et al. International Staging System for multiple myeloma. J Clin Oncol. 2005;23(15):34123420. Palumbo A, Avet-Loiseau H, Oliva S, et al. Revised International Staging System for multiple myeloma: a report from International Myeloma Working Group. J Clin Oncol. 2015;33(26):2863-2869. Moreau P, Cavo M, Sonneveld P, et al. Combination of International Scoring System 3, high lactate dehydrogenase, and t(4;14) and/or del(17p) identifies patients with multiple myeloma (MM) treated with front-line autologous stem-cell transplantation at high risk of early MM progression– related death. J Clin Oncol. 2014;32(20): 2173-2180. Palumbo A, Bringhen S, Mateos MV, et al. Geriatric assessment predicts survival and toxicities in elderly myeloma patients: an International Myeloma Working Group report. Blood. 2015;125(13):2068-2074. Engelhardt M, Dold SM, Ihorst G, et al. Geriatric assessment in multiple myeloma patients: validation of the International Myeloma Working Group (IMWG) score and comparison with other common comorbidity scores. Haematologica. 2016;101(9): 1110-1119. Engelhardt M, Domm A-S, Dold SM, et al. A concise revised Myeloma Comorbidity

haematologica | 2018; 103(11)

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

Index as a valid prognostic instrument in a large cohort of 801 multiple myeloma patients. Haematologica. 2017;102(5):910921. Mateos MV, Oriol A, Martinez-Lopez J, et al. Outcomes with two different schedules of bortezomib, melphalan, and prednisone (VMP) for previously untreated multiple myeloma: matched pair analysis using longterm follow-up data from the phase 3 VISTA and PETHEMA/GEM05 trials. Ann Hematol. 2016;95(12):2033-2041. Larocca A, Dold SM, Zweegman S, et al. Patient-centered practice in elderly myeloma patients: an overview and consensus from the European Myeloma Network (EMN). Leukemia. 2018 Apr 25. [Epub ahead of print] Bhutani M, Zhang Q, Friend R, et al. Investigation of a gene signature to predict response to immunomodulatory derivatives for patients with multiple myeloma: an exploratory, retrospective study using microarray datasets from prospective clinical trials. Lancet Haematol. 2017;4(9):e443e451. Van Vliet MH Kuiper R, Broijl A, Van Duin M, et al. Proteasome inhibitor treatment response can be predicted by gene expression profiling in multiple myeloma. Haematologica. 2014;99(S1):494. Broyl A, Kuiper R, van Duin M, et al. High cereblon expression is associated with better survival in patients with newly diagnosed multiple myeloma treated with thalidomide maintenance. Blood. 2013;121(4):624-627. Heintel D, Rocci A, Ludwig H, et al. High expression of cereblon (CRBN) is associated with improved clinical response in patients with multiple myeloma treated with lenalidomide and dexamethasone. Br J Haematol. 2013;161(5):695-700. Schuster SR, Kortuem KM, Zhu YX, et al. The clinical significance of cereblon expression in multiple myeloma. Leuk Res. 2014;38(1):23-28. Zhu YX, Braggio E, Shi C-X, et al. Identification of cereblon-binding proteins and relationship with response and survival after IMiDs in multiple myeloma. Blood. 2014;124(4):536-545. Gandhi AK, Mendy D, Waldman M, et al. Measuring cereblon as a biomarker of response or resistance to lenalidomide and pomalidomide requires use of standardized reagents and understanding of gene complexity. Br J Haematol. 2014;164(2):233-244. Kortüm KM, Mai EK, Hanafiah NH, et al. Targeted sequencing of refractory myeloma reveals a high incidence of mutations in CRBN and Ras pathway genes. Blood. 2016;128(9):1226-1233. Nijhof IS, Casneuf T, van Velzen J, et al. CD38 expression and complement inhibitors affect response and resistance to daratumumab therapy in myeloma. Blood. 2016;128(7):959-970. Pick M, Vainstein V, Goldschmidt N, et al. Daratumumab resistance is frequent in advanced-stage multiple myeloma patients irrespective of CD38 expression and is related to dismal prognosis. Eur J Haematol. 2018;100(5):494-501. Punnoose EA, Leverson JD, Peale F, et al. Expression profile of BCL-2, BCL-XL, and MCL-1 predicts pharmacological response to the BCL-2 selective antagonist venetoclax in multiple myeloma models. Mol Cancer Ther. 2016;15(5):1132-1144. Kumar S, Paiva B, Anderson KC, et al. International Myeloma Working Group consensus criteria for response and minimal

78.

79.

80.

81.

82.

83.

84.

85.

86. 87.

88.

89.

90.

91.

residual disease assessment in multiple myeloma. Lancet Oncol. 2016;17(8):e328e346. Durie BGM, Harousseau JL, Miguel JS, et al. International uniform response criteria for multiple myeloma. Leukemia. 2006;20(9): 1467-1473. Lahuerta JJ, Mateos MV, Martínez-López J, et al. Influence of pre- and post-transplantation responses on outcome of patients with multiple myeloma: sequential improvement of response and achievement of complete response are associated with longer survival. J Clin Oncol. 2008;26(35):5775-5782. van de Donk Niels WCJ, Otten Henny G, El Haddad O, et al. Interference of daratumumab in monitoring multiple myeloma patients using serum immunofixation electrophoresis can be abrogated using the daratumumab IFE reflex assay (DIRA). Clin Chem Lab Med, 2016:1105. Paiva B, Corchete LA, Vidriales M-B, et al. Phenotypic and genomic analysis of multiple myeloma minimal residual disease tumor cells: a new model to understand chemoresistance. Blood. 2016;127(15):18961906. de Tute RM, Rawstron AC, Gregory WM, et al. Minimal residual disease following autologous stem cell transplant in myeloma: impact on outcome is independent of induction regimen. Haematologica. 2016;101(2): e69-e71. Paiva B, Martinez-Lopez J, Vidriales M-B, et al. Comparison of immunofixation, serum free light chain, and immunophenotyping for response evaluation and prognostication in multiple myeloma. J Clin Oncol. 2011;29(12):1627-1633. Rawstron AC, Child JA, de Tute RM, et al. Minimal residual disease assessed by multiparameter flow cytometry in multiple myeloma: impact on outcome in the Medical Research Council Myeloma IX study. J Clin Oncol. 2013;31(20):2540-2547. Flores-Montero J, Sanoja-Flores L, Paiva B, et al. Next generation flow for highly sensitive and standardized detection of minimal residual disease in multiple myeloma. Leukemia. 2017;31:2094. Tonegawa S. Somatic generation of antibody diversity. Nature. 1983;302(5909):575581. van der Velden VHJ, Cazzaniga G, Schrauder A, et al. Analysis of minimal residual disease by Ig//TCR gene rearrangements: guidelines for interpretation of realtime quantitative PCR data. Leukemia. 2007;21(4):604-611. Logan AC, Gao H, Wang C, et al. Highthroughput VDJ sequencing for quantification of minimal residual disease in chronic lymphocytic leukemia and immune reconstitution assessment. Proc Natl Acad Sci. 2011;108(52):21194-21199. Martinez-Lopez J, Fernández-Redondo E, García-Sánz R, et al. Clinical applicability and prognostic significance of molecular response assessed by fluorescent-PCR of immunoglobulin genes in multiple myeloma. Results from a GEM/PETHEMA study. Br J Haematol. 2013;163(5):581-589. Martinez-Lopez J, Lahuerta JJ, Pepin F, et al. Prognostic value of deep sequencing method for minimal residual disease detection in multiple myeloma. Blood. 2014;123(20): 3073-3079. Avet-Loiseau H, Corre J, Lauwers-Cances V, et al. Evaluation of minimal residual disease (MRD) by next generation sequencing (NGS) is highly predictive of progression free survival in the IFM/DFCI 2009 Trial.

1783

J. Caers et al. Blood. 2015;126(23):191-191. 92. Zamagni E, Nanni C, Mancuso K, et al. PET/CT improves the definition of complete response and allows to detect otherwise unidentifiable skeletal progression in multiple myeloma. Clin Cancer Res. 2015;21(19):4384-4390. 93. Dawson S-J, Tsui DWY, Murtaza M, et al. Analysis of circulating tumor DNA to monitor metastatic breast cancer. N Engl J Med. 2013;368(13):1199-1209. 94. Diehl F, Schmidt K, Choti MA, et al. Circulating mutant DNA to assess tumor dynamics. Nat Med. 2008;14(9):985-990. 95. Oberle A, Brandt A, Voigtlaender M, et al. Monitoring multiple myeloma by next-generation sequencing of V(D)J rearrangements from circulating myeloma cells and cell-free myeloma DNA. Haematologica. 2017;102 (6):1105-1111. 96. Mithraprabhu S, Khong T, Ramachandran M, et al. Circulating tumour DNA analysis demonstrates spatial mutational heterogeneity that coincides with disease relapse in myeloma. Leukemia. 2017;31(8):1695-1705. 97. Leleu X, Karlin L, Macro M, et al. Pomalidomide plus low-dose dexamethasone in multiple myeloma with deletion 17p and/or translocation (4;14): IFM 2010-02 trial results. Blood. 2015;125(9):1411-1417. 98. Avet-Loiseau H, Attal M, Moreau P, et al. Genetic abnormalities and survival in multiple myeloma: the experience of the Intergroupe Francophone du Myélome. Blood. 2007;109(8):3489-3495. 99. An G, Acharya C, Deng S, et al. Cytogenetic and clinical marks for defining high-risk myeloma in the context of bortezomib treatment. Exp Hematol. 2015;43(3):168176.e162. 100. Klein U, Jauch A, Hielscher T, et al. Chromosomal aberrations +1q21 and del(17p13) predict survival in patients with recurrent multiple myeloma treated with lenalidomide and dexamethasone. Cancer. 2011;117(10):2136-2144. 101. Nahi H, Våtsveen TK, Lund J, et al.

1784

Proteasome inhibitors and IMiDs can overcome some high-risk cytogenetics in multiple myeloma but not gain 1q21. Eur J Haematol. 2016;96(1):46-54. 102. Hebraud B, Leleu X, Lauwers-Cances V, et al. Deletion of the 1p32 region is a major independent prognostic factor in young patients with myeloma: the IFM experience on 1195 patients. Leukemia. 2014;28(3):675-679. 103. Lakshman A, Moustafa MA, Rajkumar SV, et al. Natural history of t(11;14) multiple myeloma. Leukemia. 2018;32(1):131-138. 104. Avet-Loiseau H, Malard F, Campion L, et al. Translocation t(14;16) and multiple myeloma: is it really an independent prognostic factor? Blood. 2011;117(6):2009-2011. 105. Lahuerta JJ, Paiva B, Vidriales MB, et al. Depth of response in multiple myeloma: a pooled analysis of three PETHEMA/GEM clinical trials. J Clin Oncol. 2017;35(25): 2900-2910. 106. Chakraborty R, Muchtar E, Kumar SK, et al. Impact of post-transplant response and minimal residual disease on survival in myeloma with high-risk cytogenetics. Biol Blood Marrow Transplant. 2017;23(4):598-605. 107. Nadiminti K, Singh Abbi KK, Mott SL, et al. VTD-melphalan is well tolerated and results in very high rates of stringent CR and MRDnegative status in multiple myeloma. Onco Targets Ther. 2017;10:217-226. 108. Paiva B, Cedena M-T, Puig N, et al. Minimal residual disease monitoring and immune profiling in multiple myeloma in elderly patients. Blood. 2016;127(25):3165-3174. 109. Ludwig H, Greil R, Masszi T, et al. Bortezomib, thalidomide and dexamethasone, with or without cyclophosphamide, for patients with previously untreated multiple myeloma: 5-year follow-up. Br J Haematol. 2015;171(3):344-354. 110. Roussel M, Lauwers-Cances V, Robillard N, et al. Front-line transplantation program with lenalidomide, bortezomib, and dexamethasone combination as induction and consolidation followed by lenalidomide maintenance in patients with multiple myeloma: a

phase II study by the Intergroupe Francophone du Myélome. J Clin Oncol. 2014;32(25):2712-2717. 111. Mateos M-V, Oriol A, Martínez-López J, et al. GEM2005 trial update comparing VMP/VTP as induction in elderly multiple myeloma patients: do we still need alkylators? Blood. 2014;124(12):1887-1893. 112. Oliva S, Gambella M, Gilestro M, et al. Minimal residual disease after transplantation or lenalidomide-based consolidation in myeloma patients: a prospective analysis. Oncotarget. 2017;8(4):5924-5935. 113. Puig N, Sarasquete ME, Balanzategui A, et al. Critical evaluation of ASO RQ-PCR for minimal residual disease evaluation in multiple myeloma. A comparative analysis with flow cytometry. Leukemia. 2014;28(2):391-397. 114. Ferrero S, Ladetto M, Drandi D, et al. Longterm results of the GIMEMA VEL-03-096 trial in MM patients receiving VTD consolidation after ASCT: MRD kinetics' impact on survival. Leukemia. 2015;29(3):689-695. 115. Chari A, Suvannasankha A, Fay JW, et al. Daratumumab plus pomalidomide and dexamethasone in relapsed and/or refractory multiple myeloma. Blood. 2017;130(8):974981. 116. Korde N, Roschewski M, Zingone A, et al. Treatment with carfilzomib-lenalidomidedexamethasone with lenalidomide extension in patients with smoldering or newly diagnosed multiple myeloma. JAMA Oncol. 2015;1(6):746-754. 117. Patriarca F, Carobolante F, Zamagni E, et al. The role of positron emission tomography with 18F-fluorodeoxyglucose integrated with computed tomography in the evaluation of patients with multiple myeloma undergoing allogeneic stem cell transplantation. Biol Blood Marrow Transplant. 2015;21(6):1068-1073. 118. Lapa C, Luckerath K, Malzahn U, et al. 18 FDG-PET/CT for prognostic stratification of patients with multiple myeloma relapse after stem cell transplantation. Oncotarget. 2014;5(17):7381-7391.

haematologica | 2018; 103(11)

ARTICLE

Hematopoiesis

Alas1 is essential for neutrophil maturation in zebrafish

Ferrata Storti Foundation

Junwei Lian,1 Jiakui Chen,1 Kun Wang,1 Lingfeng Zhao,1 Ping Meng,1 Liting Yang,1 Jiayi Wei,1 Ning Ma,1 Jin Xu,2 Wenqing Zhang2 and Yiyue Zhang1,2

Key Laboratory of Zebrafish Modeling and Drug Screening for Human Diseases of Guangdong Higher Education Institutes, Department of Developmental Biology, School of Basic Medical Sciences, Southern Medical University and 2Division of Cell, Developmental and Integrative Biology, School of Medicine, South China University of Technology, Guangzhou, P.R. China 1

Haematologica 2018 Volume 103(11):1785-1795

ABSTRACT

eutrophils play essential roles in innate immunity and are the first responders to kill foreign micro-organisms, a function that partially depends on their granule content. The complicated regulatory network of neutrophil development and maturation remains largely unknown. Here we utilized neutrophil-deficient zebrafish to identify a novel role of Alas1, a heme biosynthesis pathway enzyme, in neutrophil development. We showed that Alas1-deficient zebrafish exhibited proper neutrophil initiation, but further neutrophil maturation was blocked due to heme deficiency, with lipid storage and granule formation deficiencies, and loss of heme-dependent granule protein activities. Consequently, Alas1-deficient zebrafish showed impaired bactericidal ability and augmented inflammatory responses when challenged with Escherichia coli. These findings demonstrate the important role of Alas1 in regulating neutrophil maturation and physiological function through the heme. Our study provides an in vivo model of Alas1 deficiency and may be useful to evaluate the progression of heme-related disorders in order to facilitate the development of drugs and treatment strategies for these diseases.

Correspondence: yiyue@smu.edu.cn or mczhangyy@scut.edu.cn or mczhangwq@scut.edu.cn

Introduction Neutrophils are the most abundant leukocytes in the circulation and the first responders to sites of infection, where they attack pathogens by phagocytosis, degranulation, and by generating neutrophil extracellular traps.1,2 Neutrophil development is highly conserved in vertebrates, making zebrafish a suitable model for investigation. Neutrophils are derived from granulocyte-monocyte progenitors, and undergo determination and differentiation from myeloblasts to mature neutrophils.1,3 During neutrophil differentiation and maturation, neutrophil granules are formed and assembled.1 Anti-microbial proteins are thought to be the major constituents of neutrophil granules, and they play important roles in neutrophil diapedesis, chemotaxis, and the phagocytosis of micro-organisms.1,4 Several transcription factors have been reported to be involved in neutrophil development and physiological function in mammals, including SPI.1/PU1 and C/EBP-Îľ.3 A recent study of embryonic myelopoiesis revealed that the Pu.1-Runx1 regulatory loop controlled embryonic myeloid cell fate in zebrafish.5,6 We previously demonstrated that c-Myb and Cebp1 co-operatively acted in parallel to govern neutrophil maturation.7 In addition to these transcription factors, zebrafish deficient for the neutrophil granule protein myeloperoxidase (Mpx) or the neutrophilspecific marker nephrosin (Npsn) have altered neutrophil maturation and inflammatory responses to fungal and bacterial infection, respectively.8,9 Nevertheless, the complicated regulatory network of neutrophil maturation, as well as the impact on physiological function remain poorly understood. Heme (iron protoporphyrin IX) functions as a prosthetic group on various proteins, so-called hemoproteins, such as hemoglobin, myoglobin, cytochromes, catalases, and peroxidases.10 Hemoproteins are involved in diverse biological functions, including oxygen transport, energy metabolism, and drug biotransformation.11 Moreover, heme also plays important roles in the regulation of transcription,12-14 haematologica | 2018; 103(11)

Received: March 27, 2018. Accepted: June 27, 2018. Pre-published: June 28, 2018 doi:10.3324/haematol.2018.194316 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1785 ÂŠ2018 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.

1785

J. Lian et al. translation,15,16 protein degradation,17,18 and microRNA processing.19 The accumulation of excess heme and its precursors in tissues can cause oxidative damage via the generation of reactive oxygen species.11 Thus, cellular heme homeostasis must be tightly controlled. The 5-aminolevulinate synthase 1 (ALAS1) is a mitochondrial enzyme that catalyzes the condensation of glycine and succinyl-CoA, forming 5-aminolevulinic acid. ALAS1 is the first and rate-limiting enzyme of the heme biosynthetic pathway, which is conserved from lower to higher organisms.20 ALAS1 (also called hepatic ALAS or non-specific ALAS) is ubiquitously expressed throughout the body, whereas another isoform, ALAS2 (also called ALAS-E), is predominantly expressed in erythroid cells, to meet the need of the large amounts of heme required for hemoglobin synthesis.11 It has been reported that human ALAS2 mutations cause X-linked sideroblastic anemia;21 ALAS2-deficient mice and zebrafish also display severe anemia,22,23 revealing the major contribution of ALAS2 to erythroid heme biosynthesis, and hence how it is essential to erythroid differentiation. In contrast to ALAS2, there are no reported human diseases directly caused by mutations in ALAS1. In mice, Alas1-null embryos are lethal by embryonic day 8.5 (E8.5),24 thus, the in vivo physiological role of ALAS1 is unclear. Using GFP knock-in mice (Alas1+/GFP), ALAS1 was found to be highly expressed in the liver, exocrine, endocrine glands, and myeloid cells, where large amounts of heme are required to meet the needs of tissue-specific hemoproteins, such as MPO, NADPH oxidase, and CYP450.24 Notably, Alas1 is also expressed higher in neutrophils than macrophages,24 suggesting cell-specific roles in neutrophils. However, the function of ALAS1 in neutrophils is still unknown. Taking advantage of their transparent body, we can observe neutrophil morphology to trace neutrophil behaviors in live zebrafish. Zebrafish is an ideal model for studying neutrophil biology.25,26 Here we report a novel role of Alas1 in regulating neutrophil development using a neutrophil lineage-deficient mutant zebrafish line (previously named smu350) that was isolated from our ENU mutant collection.27 We demonstrated that alas1 was the causative gene for the mutant and revealed that alas1 was essential for neutrophil development and physiological function.

Methods Fish maintenance Zebrafish were raised and maintained under standard conditions.28 Embryos were maintained in egg water containing 0.2 mM N-phenylthiourea (Sigma-Aldrich, St. Louis, MO, USA) to prevent pigment formation. All work involving zebrafish was approved by Southern Medical University Animal Ethics Committee. The following strains were used: AB, Tg(lyz:DsRed),29 Tg(gata1:DsRed), vltm651/+ and alas1smu350/+.

Treatment with succinylacetone Succinylacetone (SA) (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in egg water. Zebrafish embryos were placed in culture dishes containing 1 mM SA at 5 hours (h) post fertilization (hpf) till the desired stage.

Bacterial infection The eGFP-labeled E. coli strain XL10 was cultured in LB broth with 50 mg mL-1 ampicillin at 37°C until reaching an optical den1786

sity at 600 nm between 0.5-0.8. Bacteria were washed with sterile phosphate buffered saline (PBS) three times, harvested by centrifugation at 5000 x g for 5 minutes (min) and resuspended in sterile PBS. The working concentration of E. coli was 2x109 mL-1, and approximately 0.5 nL of bacterial suspension was subcutaneously injected over a somite into 3-day-post-fertilization (dpf) embryos with 0.02% tricaine using a PLI-100A Pico-Injector (Warner Instruments, Hamden, CT, USA) as previously described.26,30 For bacterial colony forming assays, every injected embryo was washed with sterile PBS three times, and then homogenized in 200 mL of sterile PBS at the desired time points. Then, 10 mL of homogenate was plated on LB medium with ampicillin and cultured at 37°C overnight. The results are the average of two separate experiments.

Statistical analysis Data were recorded and analyzed using GraphPad Prism 7 and IBM SPSS v.23. Two-tailed Student t-test and Mann-Whitney U test were used for comparisons between parametric and non-parametric data, respectively. One-way analysis of variance (with Bonferroni or Dunnett T3 post-test adjustment) was used for parametric data to make multi-comparisons. Differences were considered significant at P<0.05. Data are expressed as the mean±Standard Deviation (SD).

Results Neutrophil deficiency in smu350 mutant zebrafish To identify new regulators of neutrophil development, we conducted a genetic screen for neutrophil-deficient zebrafish mutants using Sudan black B (SB) staining. From this screen, we isolated the neutrophil-deficient smu350 mutant, which lacked the SB signal as early as 36 hpf (Figure 1A). The early loss of the SB signal in smu350 mutants suggested defects in embryonic neutrophils, as SB+ cells represent embryonic neutrophils that are initiating from embryonic myelopoietic tissue.31 Because myeloid progenitors that are derived from rostral blood islands will progress to neutrophils during embryonic myelopoiesis, we first determined if there were defects in the formation of myeloid progenitors. The results showed that pu.1 expression at 22 hpf was normal (Figure 1B), suggesting the presence of myeloid progenitors in smu350 mutants. Therefore, we speculated that the loss of the SB signal was due to defects in neutrophil maturation. To test this possibility, we detected the transcript and protein activity of the myeloid-specific peroxidase (Mpx), which is an abundant granule protein in neutrophils.2 Wholemount in situ hybridization (WISH) showed that mpx mRNA expression was intact (Figure 1C), suggesting the presence of neutrophils in smu350 mutants. To further examine Mpx enzyme activity, diaminobenzidine (DAB, a peroxidase substrate) staining was performed. The results showed that while the signal in the yolk sac (representing hemoglobin peroxidase activity32) was present, signals representing neutrophil peroxidase activity were absent in smu350 mutants (Figure 1D), suggesting that Mpx lost its catalytic activity. As neutrophil granules are abundant with Mpx, we directly monitored neutrophil granule morphology via video-enhanced differential interference contrast (VE-DIC) analyses of live embryos. The results showed that neutrophils from siblings (alas1smu350/+ and alas1+/+ embryos from a heterozygous alas1smu350/+ in-cross) had abundant visible and highly mobilized granules, while neutrophils from smu350 mutants lacked such granules haematologica | 2018; 103(11)

Regulation of neutrophil maturation in zebrafish

(Figure 1E and Online Supplementary Appendix, Movies 1 and 2). These results suggest that neutrophil maturation is defective in smu350 mutants.

The alas1 mutation was responsible for neutrophil defects of smu350 mutant Positional cloning was then performed to identify the causative gene in smu350 mutants. Initial mapping with bulk segregation analysis located the mutated site to link-

age group 11 (data not shown), then fine mapping placed the mutated gene within a region between two simple sequence length polymorphism markers, CU633745-M and CU929297-M, from the Massachusetts General Hospital panel (Figure 2A). The mutation was then mapped to a 100-kb region partly covered by two bacterial artificial chromosomes (BACs) (Figure 2A). There were 9 predicted genes in this region (Figure 2A). By sequencing, we found a T-to-A mutation in alas1 intron 7 next to the

haematologica | 2018; 103(11)

Figure 1. Neutrophil deficiency in smu350 mutants. (A) The Sudan black B (SB) signal was totally absent in smu350 mutants. SB staining in siblings (left) and smu350 mutants (right) at 3 days post fertilization (dpf). The boxed regions are magnified in the lower right-hand corner. (B) Whole-mount in situ hybridization (WISH) of pu.1 expression in siblings (left) and smu350 mutants (right) at 22 hours post fertilization (hpf). (C) WISH of mpx expression in siblings (left) and smu350 mutants (right) at 2 dpf. (D) DAB staining in siblings (left) and smu350 mutants (right) at 2 dpf. Signal points representing neutrophil peroxidase activity (black arrowheads) were absent in smu350 mutants. Red arrowheads indicate the hemoglobin signal. The boxed regions are magnified in the lower right-hand corner. (E) Neutrophil granules were absent in smu350 mutants. In vivo imaging of neutrophils in 2-dpf Tg(lyz:DsRed);smu350 embryos by VE-DIC microscopy. Neutrophils of siblings had abundant visible and highly mobilized granules (29 of 31 embryos), while neutrophils of smu350 mutants lacked granules (12 of 12 embryos). White arrowheads indicate neutrophil granules. (Left) Bright-field DIC image; (right) an overlay of bright-field DIC and fluorescent images. See also Online Supplementary Appendix, Movies 1 and 2. (F) Quantifications of SB+ cells in the posterior blood island (PBI), pu.1+ cells (B), mpx+ cells in the PBI (C), and DAB+ cells in the PBI (D). MeanÂąStandard Deviation (SD), n>15; Student t-test: ****P<0.0001. ns: not significant; nd: not detectable. Scale bars: 200 mm (A-D) and 5 mm (E).

1787

J. Lian et al.

Figure 2. The alas1 gene was mutated in smu350 mutants. (A) The mutated gene in smu350 mutants mapped to a 100-kb region between two simple sequence length polymorphism markers, CU633745-M (two recombinants in 6160 smu350 mutant embryos) and CU929297-M (three recombinants in 6160 smu350 mutant embryos), on linkage group 11. The 100-kb region, partly covered by two bacterial artificial chromosomes (BACs) (CU633745 and CU929297), contains 9 predicted genes. (B) The structure of the zebrafish alas1 gene. The red asterisk indicates a T-to-A mutation in intron 7 of alas1 in smu350 mutants. The black arrow indicates the position of the CRISPR/Cas9 target in alas1. Numbers of constitutive exons are indicated. (C) Agarose gel electrophoresis of alas1 RT-qPCR amplification products from 3-day post fertilization (dpf) wild-type zebrafish, siblings, and smu350 mutants. Four major products (indicated by black arrows) were identified in smu350 mutants compared with wild-type transcripts (461 bp). The actb2 was used as an internal control. (D) The mutated alas1 transcripts and their predicted translation products in smu350 mutants. The blue arrow indicates the position of the transcriptional start site. Black arrows indicate the RT-qPCR primers used in (C), and black boxes indicate the wild-type peptides. Red boxes indicate the incorrect peptides generated by the altered splicing. Pink boxes indicate the pre-sequence domain (pfam09029). Green boxes indicate the aspartate aminotransferase superfamily domain (fold type I) of pyridoxal phosphate-dependent enzymes (cl18945). Boxes with zigzag edges indicate truncated regions. Blue boxes with white numbers indicate exons. Black and red numbers denote distances to the start codon in wildtype and mutants, respectively. (E) alas1 expression was up-regulated in smu350 mutants. Relative expression of alas1 transcript assessed by RT-qPCR in smu350 mutants (gray column) and wild-type (black column) at 2, 3, 5, and 7 dpf [meanÂąStandard Deviation (SD); n=10 in each group, performed in triplicate]. Statistical significance was determined using Student t-test, ***P<0.001. (F) Alas1 protein was absent in smu350 mutants. Alas1 protein expression in the whole fish body assessed by western blotting at 5 dpf. GAPDH was used as the loading control.

1788

haematologica | 2018; 103(11)

Regulation of neutrophil maturation in zebrafish

exon-intron boundary in smu350 mutants (Figure 2B), which is likely to be a splicing mutation. By amplifying alas1 cDNA from smu350 mutants, we found at least four unexpected alas1 transcripts (Figure 2C), which were confirmed by sequencing analysis following TA cloning. These unexpected transcripts were predicted to produce truncated Alas1 proteins or in-frame-insertion Alas1 proteins, all of which would interrupt the enzyme activity domain of Alas1 (Figure 2D). Expression analyses by reverse transcription quantitative real-time polymerase chain reaction (RT-qPCR) showed elevated alas1 mRNA expression in smu350 mutants compared with siblings throughout development (Figure 2E). However, we found that neither the wild-type form nor the abnormal variants of Alas1 protein were present in smu350 mutants, as determined by western blotting (Figure 2F). These data strongly suggest that this alas1 mutation is responsible for the smu350 mutant (hereafter named alas1smu350/smu350) phenotype and that the alas1smu350/smu350 mutant is a loss-of-function mutant. To confirm that the alas1smu350/smu350 mutant phenotype was indeed caused by the alas1 mutation, we used CRISPR/Cas9 to create alas1-knock-out mutants. A homozygous alas1 mutant (alas1Δ2/Δ2) with a 2-bp deletion

within exon 7 of alas1 was obtained, and the mutation resulted in a frameshift of the alas1 product, causing a loss of Alas1 protein in the alas1Δ2/Δ2 mutant (Figure 3A and B). Similar to the alas1smu350/smu350 mutant, the alas1Δ2/Δ2 homozygous mutant and the alas1smu350/Δ2 bi-allelic mutant also showed loss of SB staining (Figure 3C and D), indicating that alas1 is indeed the causative gene for the altered neutrophil development phenotype.

The alas1 mutation caused heme deficiency ALAS1 is the first and rate-limiting enzyme for heme biosynthesis, and heme negatively regulates ALAS1 expression through a feedback mechanism.10 As the Alas1 protein was undetectable (Figure 2F), we postulated that the heme levels of alas1smu350/smu350 mutants might be decreased. To test this hypothesis, we measured heme levels in alas1smu350/smu350 mutants using a fluorescence heme assay. Surprisingly, total heme levels of alas1smu350/smu350 mutants were abnormally elevated (Figure 4A). As erythroid tissue is the major site of heme production in the body and depends on the isozyme Alas2,33 we then checked if this elevated heme was derived from erythrocytes. We first compared the erythrocyte numbers between 4-dpf alas1smu350/smu350 mutants and their siblings and

Figure 3. The alas1 was the causative gene of the smu350 mutant. (A) Sequencing analysis revealed a 2-bp deletion within exon 7 of alas1 in CRISPR/Cas9 generated alas1Δ2/Δ2 mutants. Uppercase sequences highlighted blue indicate exons; sequences in lowercase indicate introns. Sequences underlined in red indicate the CRISPR/Cas9 target in alas1. The red asterisk indicates the smu350 mutation site. Black boxes indicate the wild-type peptides, and the red box indicates the incorrect peptide generated by the altered splicing. The pink box indicates the pre-sequence domain (pfam09029). The green box indicates the aspartate aminotransferase superfamily (fold type I) domain of pyridoxal phosphate-dependent enzymes (cl18945). The box with zigzag edges indicates the truncated region. Black and red numbers denote distances to the start codon in wild type and mutants, respectively. (B) Alas1 protein was absent in alas1Δ2/Δ2 mutants. Examination of Alas1 protein expression in the whole fish body by western blotting at 5 days post fertilization (dpf). GAPDH was used as the loading control. (C) The Sudan black B (SB) signal was absent in alas1Δ2/Δ2 mutants. SB staining in siblings (upper) and alas1Δ2/Δ2 mutants (lower) at 3 dpf. Boxed regions are magnified in the lower right-hand corner. Scale bars: 200 mm. (D) SB signal was absent in alas1smu350/Δ2 mutants. SB staining in siblings (upper) and alas1smu350/Δ2 mutants (lower) at 3 dpf. Boxed regions are magnified in the lower right-hand corner. Scale bars: 200 mm.

haematologica | 2018; 103(11)

1789

J. Lian et al.

Figure 4. The alas1 mutation caused heme deficiency. (A) Whole fish heme levels of alas1smu350/smu350 mutants were significantly higher than those of siblings. Relative whole fish heme levels in siblings (circles) and alas1smu350/smu350 mutants (squares) at 4 days post fertilization (dpf). Lines show mean±Standard Deviation (SD), 6 individual data points in each group, each data point was based on 3 measurements for 5 embryos; Student t-test, ***P<0.001. The relative heme level was normalized to per-fish level. (B) No significant differences in erythrocyte numbers between siblings and alas1smu350/smu350 mutants. gata1:DsRed+ erythrocyte numbers were measured by flow cytometric analysis using 4-dpf self-progeny of Tg(gata1:DsRed)+/+;alas1smu350/+ transgenic line. Numbers were normalized to per-fish level in siblings (black column) and alas1smu350/smu350 mutants (gray column). Mean±SD; performed in triplicate; Student t-test, ns: not significant. (C) Cellular heme levels of alas1smu350/smu350 mutant neutrophils were decreased compared with those of siblings. Relative cellular heme levels in sorted neutrophils (lyz:DsRed+ cells) of siblings (black column) and alas1smu350/smu350 mutants (gray column) at 4 dpf. Mean±SD; performed at least in triplicate; Student t-test, *P<0.05. The relative heme level was normalized to per-cell level. (D) Cellular heme levels of alas1smu350/smu350 mutant erythrocytes were increased compared with those of siblings. Relative cellular heme levels in sorted erythrocytes (gata1:DsRed+ cells) of siblings (black column) and alas1smu350/smu350 mutants (gray column) at 4 dpf. Mean±SD; performed in triplicate; Student t-test, *P<0.05. The relative heme level was normalized to per-cell level. (E) Aberrant whole fish heme increment was dampened in alas1smu350/smu350 mutants without erythrocytes. Relative whole fish heme levels in 4-dpf alas1;gata1a siblings, alas1 single mutants, gata1a single mutants, and alas1;gata1a double mutants from alas1smu350/+;gata1am651/+ in-cross. Lines show mean±SD, 4 individual data points in each group, each data point was based on 3 measurements for 5 embryos; one-way ANOVA followed by Dunnett T3 post test, *P<0.05. ns: not significant; nd: not detectable. The relative heme level was normalized to per-fish level. (F) Relative expressions of alas2 and genes related to heme degradation and transport. The assay was performed by RT-qPCR in 4-dpf siblings (black column) and alas1smu350/smu350 mutants (gray column). Mean±SD; n=8 in each group, performed in triplicate. Statistical significance was determined using Student t-test; ****P<0.0001, **P<0.01, *P<0.05. ns: not significant.

1790

haematologica | 2018; 103(11)

Regulation of neutrophil maturation in zebrafish

found no significant differences (Figure 4B), indicating the elevated whole heme in mutants was not due to increased erythrocyte numbers. We next isolated neutrophils and erythrocytes of alas1smu350/smu350 mutants and their siblings by fluorescence-activated cell sorting (FACS) using 4-dpf selfprogeny of Tg(lyz:DsRed)+/+;alas1smu350/+ and +/+ smu350/+ transgenic lines, respectively, Tg(gata1:DsRed) ;alas1 to directly measure heme levels in the two cell types. By comparing relative heme levels in neutrophils or erythrocytes between alas1smu350/smu350 mutants and their siblings, we found that heme was less abundant in neutrophils of alas1smu350/smu350 mutants than that of siblings (Figure 4C), while in erythrocytes, heme was more accumulated in erythrocytes of mutants (Figure 4D). These data indicate that alas1 mutation results in heme insufficiency in neutrophils but abnormal accumulation in erythrocytes. To further confirm that the elevated heme of the whole body was derived from erythrocytes in alas1smu350/smu350 mutants, we introduced vltm651,34 (a gata1a mutant with a 'bloodless' phenotype having no erythrocytes but intact white blood cells) into the smu350 mutant background to eliminate the effect of erythrocytes. As expected, whole fish heme levels of alas1smu350/smu350 mutants were almost undetectable compared with those of siblings in the gata1a mutant background (Figure 4E). These data indicate that the aberrant heme accumulation of the whole body is indeed derived from erythrocytes in alas1smu350/smu350 mutants. Alas2, the other isozyme of Alas1, is essential for the heme biosynthesis in erythrocytes and predominantly expressed in erythrocytes.10 To test whether the erythroid heme increment resulted from the elevated alas2 expres-

sion, we checked alas2 expression in alas1smu350/smu350 mutants. The data showed that alas2 was not altered compared with that in siblings (Figure 4F), suggesting that the erythroid heme accumulation was not due to the compensatory of alas2, at least at the transcription level. Since heme content is tightly controlled by the homeostasis of heme biosynthesis, degradation, and transport pathways,35 we then detected the expression of heme oxygenase enzymes (hmox1a and hmox2a),10,36 which encode the rate-limiting enzymes for heme degradation. The results showed that both gene expressions were down-regulated in alas1smu350/smu350 mutants (Figure 4F), suggesting the impaired heme degradation in the absence of alas1. The results suggest that the dysregulation of heme biosynthesis affects the heme degradation in alas1smu350/smu350 mutants, and the elevated heme might be attributed to the reduced heme degradation. The elevated heme in erythrocytes could not be utilized by heme-deficient neutrophils in alas1 mutants, which is likely due to the fact that the synthesized heme in erythrocytes could not be transported to neutrophils. To test this hypothesis, we further detected the expressions of genes encoding heme transporters. Flvcr1a is reported to export heme out of the cell as a plasma membrane heme exporter.37 Hpx, a high-affinity heme-binding protein, is reported to interact with FLVCR in heme transfer.35 HRG-1 is reported to deliver heme to the cytosol,38 which is encoded by slc48a1a (heme transporter hrg1-B) and slc48a1b (heme transporter hrg1-A) in zebrafish. MRP-5/ABCC5 is reported to reside on the plasma membrane and endosomal compartments and regulate the export of cytosolic heme.39 RT-qPCR showed that the

Figure 5. Heme was essential for neutrophil maturation. (A) Whole fish heme levels of succinylacetone (SA)-treated embryos were significantly decreased than those of untreated control. Relative whole fish heme levels in untreated control (circles) and SA-treated embryos (triangles) at 2 days post treatment (dpt). Lines show MeanÂąStandard Deviation (SD), 6 individual data points in each group, each data point was based on 3 measurements for 5 embryos; Student t-test: ***P<0.001. The relative heme level was normalized to per-fish level. (B) The o-Dianisidine signal was totally absent in SA-treated embryos. o-Dianisidine staining in untreated wild-type (left, 11 of 11 embryos) and SA-treated (right, 10 of 10 embryos) embryos at 2 dpt. (C) The Sudan black B (SB) signal was totally absent in SA-treated embryos. SB staining in untreated wild-type (left, 20 of 20 embryos) and SA-treated (right, 17 of 17 embryos) embryos at 2 dpt. Black arrowheads indicate the neutrophil signals. The boxed regions are magnified in the lower right-hand corner. (D) The DAB signal was totally absent in SA-treated embryos. DAB staining in untreated wild-type (left, 18 of 18 embryos) and SA-treated (right, 14 of 14 embryos) embryos at 2 dpt. Black arrowheads indicate the neutrophil peroxidase signals. The boxed regions are magnified in the lower right-hand corner. (E) WISH of lyz expression in untreated wild-type (left, 21 of 21 embryos) and SA-treated (right, 25 of 25 embryos) embryos at 2 dpt. (F) WISH of mpx expression in untreated wild-type (left, 23 of 23 embryos) and SA-treated (right, 15 of 15 embryos) embryos at 2 dpt. Scale bars: 200 mm (B-F).

haematologica | 2018; 103(11)

1791

J. Lian et al.

Figure 6. alas1 deficiency caused impaired host immunity against E. coli infection. (A) Arrowhead indicates the site of bacteria injection. The imaged region is boxed. (B-E) The in vivo behavior of neutrophils against E. coli was monitored by confocal microscopy. eGFP+ E. coli were subcutaneously injected over one somite into 3-day post fertilization (dpf) sibling (upper panels) and alas1smu350/smu350 mutant (lower panels) larva of the Tg(lyz:DsRed) background. Neutrophil behavior was analyzed through live imaging at 0.5 hours post injection (hpi) (B), 3 hpi (C), 5 hpi (D), and 24 hpi (E). All images are maximum-intensity projections from 25 steps x 2 mm. Scale bars: 50 mm. (F) Bacterial burden of embryos injected with E. coli. Significantly more bacterial cells were detected in alas1smu350/smu350 mutants (gray column) compared with siblings (black column) at 5 and 24 hpi. Mean±Standard Deviation (SD), n>10 in each group; Mann-Whitney U test: ****P<0.0001, **P<0.01, ns: not significant. (G) Quantification of recruited DsRed+ neutrophils at infection sites in live embryos injected with E. coli. Significantly more neutrophils were observed at the infection sites of alas1smu350/smu350 mutants (gray column) compared with siblings (black column) at 24 hpi. Mean±SD; n>19 in each group; Student t-test: ****P<0.0001. nd: not detectable, ns: not significant. (H) Relative il1b expression assessed by RT-qPCR. il1b expression was up-regulated in alas1smu350/smu350 mutants (gray column) compared with siblings (black column) at 3, 5, and 24 hpi. Mean±SD; n=4 in each group, performed in triplicate. Expression levels were adjusted for trauma [phosphate buffered saline (PBS) injection only]. Statistical significance was determined using Student t-test: **P<0.01, *P<0.05. ns: not significant. (I) Relative cxcl8a expression assessed by RT-qPCR. cxcl8a expression was up-regulated in alas1smu350/smu350 mutants (gray column) compared with siblings (black column) at 3 and 5 hpi. Mean±SD; n=4 in each group, performed in triplicate. Expression levels were adjusted for trauma (PBS injection only). Statistical significance was determined using Student t-test: **P<0.01. ns: not significant.

1792

haematologica | 2018; 103(11)

Regulation of neutrophil maturation in zebrafish

expressions of these heme transporter genes were decreased in alas1smu350/smu350 mutants compared with their siblings (Figure 4F), probably due to the feedback regulation of aberrant heme contents in alas1smu350/smu350 mutants. Thus, the heme transport deficiency might be one of the reasons that the elevated erythroid heme could not be utilized by neutrophils in alas1smu350/smu350 mutants.

Heme was essential for neutrophil maturation To confirm whether the neutrophil maturation defects were caused by inadequate heme levels, we next treated wild-type zebrafish embryos with SA, an inhibitor of Î´-aminolevulinic acid dehydratase, which catalyzes the second step in heme biosynthesis pathway,40 to inhibit the endogenous heme levels. Total heme levels of SA-treated embryos were significantly decreased compared with untreated control embryos (Figure 5A). As reported, the o-Dianisidine staining signal was decreased as hemoglobin synthesis is inhibited without heme (Figure 5B).23,40 When we monitored the neutrophil phenotypes, we found that SA-treated embryos showed loss of SB and DAB staining but intact lyz and mpx expression (Figure 5C-F), which mimics the neutrophil maturation defects in Alas1-deficient mutants. These data suggest that the neutrophil defects in alas1smu350/smu350 mutant are indeed caused by inadequate effective heme in neutrophils.

Neutrophil bactericidal defects in alas1smu350/smu350 mutants Neutrophils play key roles in various functions, including action against certain infections, largely depending on granule proteins.1 The neutrophil granule defects suggest that the anti-infection ability of alas1smu350/smu350 mutant neutrophils may be attenuated. To detect whether alas1 deficiency affected neutrophil bactericidal function, alas1smu350/smu350 mutants were challenged with a bacterial infection.26,30 We subcutaneously injected eGFP labeled E. coli over one somite in 3-dpf Tg(lyz:DsRed);alas1smu350/+ intercrossed embryos, in which DsRed was expressed specifically in neutrophils (Figure 6A). Neutrophil behavior and immune responses were then monitored. We first monitored in vivo bacterial growth and detected the kinetic curves of the bacterial burden of infected embryos. In sibling embryos, bacteria growth was inhibited effectively in the host, as green fluorescent bacteria decreased rapidly in the infection site (Figure 6B-E). By further plating the homogenized embryos/larvae on LB medium for quantification, we found that bacterial colonies were gradually decreased from 5 h post injection (hpi) and eventually became almost absent at 24 hpi (Figure 6F), suggesting the inhibition of bacterial growth in the host. In mutant embryos, the eGFP+ bacterial load was similar to siblings within the first 3 hpi (Figure 6B and C), but fluorescent bacteria were still accumulating at 5 hpi and persisted at 24 hpi, when clearance had been completed in the siblings (Figure 6D-E). Quantification data consistently showed that the plated colony numbers from mutants were similar to those of siblings within the first 3 hpi, but the numbers were significantly higher at 5 and 24 hpi than those of the siblings (Figure 6F), suggesting antimicrobial activity was impaired in alas1smu350/smu350 mutants. We further counted the number of neutrophils recruited to the infection site. Within the first 5 hpi, alas1smu350/smu350 embryos showed similar neutrophil recruitment to sibling embryos (Figure 6Bhaematologica | 2018; 103(11)

D, G). However, neutrophils were still accumulating at the infection site in alas1smu350/smu350 larvae at 24 hpi, when recruited neutrophils had almost completely disappeared in the siblings (Figure 6E and G), confirming the defect in neutrophil-specific antibacterial response in alas1smu350/smu350 mutants. To gain further insight into the infection-induced inflammatory alterations, we detected the expression of inflammatory factors il1b and cxcl8a,41,42 which induce neutrophil chemotaxis and promote immune responses. Expression analyses showed that both genes were significantly up-regulated in infected alas1smu350/smu350 mutants compared with siblings. The il1b and cxcl8a expressions peaked in the mutants at 5 hpi, at which point their expressions had already been down-regulated in the siblings (Figure 6H and I), suggesting a more dramatic inflammatory response in alas1smu350/smu350 mutants. Taken together, these data demonstrate that alas1 deficiency causes impaired immune responses to bacterial infection.

Discussion In this study, we showed a role for the heme biosynthesis pathway enzyme Alas1 in regulating neutrophil maturation and function. Neutrophils in Alas1-deficient zebrafish had heme deficiency, which led to the loss of heme-related granule protein activities, defective granule formation, and altered immune responses against pathogenic bacteria. Here, we found that the heme dysregulation caused by the alas1 mutation led to neutrophil defects in zebrafish. Given the important role of neutrophils in immunity, it was expected that Alas1-deficient zebrafish would show impaired bactericidal ability. Inflammatory factors, such as il1b and cxcl8a,41,42 mediate neutrophil recruitment and promote immune responses. The over-elevated expression of inflammatory factors in infected alas1smu350/smu350 mutants might be explained by the ineffectiveness of killing bacteria, thereby leading to greater pathogen growth and stronger immune responses in the host. We also noticed that the inflammatory responses eventually subsided while the bacterial burden and recruited neutrophils were still present at 24 hpi in Alas1-deficient zebrafish, so it was likely that the inflammatory factors had been excessively depleted. As far as we know, the regulatory genes and pathways of heme biosynthesis, degradation and transport are largely conserved between mammals and zebrafish in general.20,23,35,36,43 Vertebrates contain two ALAS isozymes encoded by 2 distinct genes located on different chromosomes; ALAS2 is expressed in erythrocytes, whereas ALAS1 is ubiquitously expressed.11 By searching the integrated RNA-seq database on BloodSpot,44 we found that, in humans and mice, ALAS1 is highly expressed in myeloid cells, which is consistent with our zebrafish data and partly explains the importance of alas1 for neutrophils. In mice, ALAS1 is also highly expressed in the liver, exocrine, and endocrine glands, suggesting specific roles in those tissues.24 Accordingly, even though alas1 is ubiquitously expressed, we suspect that alas1 may play specific roles in certain tissues to meet the need of hemoproteins. It is reported that ALAS1-null mice died in utero until E8.5, with a severely retarded morphology, indicating that ALAS1 is essential for the early development of mouse embryos.24 1793

J. Lian et al.

It is likely that because of the lethality resulting from ALAS1 deficiency, no reported human diseases directly caused by mutations in ALAS1 have been reported so far. Although the Alas1-deficient zebrafish were indistinguishable on morphology from wild type at embryonic stages, the Alas1-deficient zebrafish are not viable past 8 dpf and showed some morphological defects from 4 dpf onwards, such as delayed disappearance of the yolk sac, abnormal liver, and failed swim bladder formation (data not shown). Thus, the specific functions of ALAS1 in different tissues and organs remain to be clarified. Alas1-deficient zebrafish showed impaired heme levels in neutrophils but elevated heme levels in erythrocytes. It is known that in addition to mitochondrial heme synthesis within all cells, heme can also be transported in and out of cells from the plasma.11 Intracellular heme levels are tightly controlled through the co-ordination of heme synthesis, degradation, and trafficking.35 In this study, we found that the expression of alas2, encoding the other rate-limiting enzyme of heme synthesis, was not changed, while the expressions of heme degradation and transporter genes were down-regulated in alas1smu350/smu350 mutants. In Alas1-deficient zebrafish, cellular heme levels were unexpectedly elevated in erythrocytes, partly due to the decrease in heme degradation. The excessive heme could not be effectively used by neutrophils, probably because the heme transport was impaired. Similar to mice,24 the over-produced heme by alas2 in erythrocytes

References 1. Amulic B, Cazalet C, Hayes GL, Metzler KD, Zychlinsky A. Neutrophil function: from mechanisms to disease. Annu Rev Immunol. 2012;30(1):459-489. 2. Odobasic D, Kitching AR, Holdsworth SR. Neutrophil-mediated regulation of innate and adaptive immunity: The role of myeloperoxidase. J Immunol Res. 2016; 2016(6):2349817. 3. Cowland JB, Borregaard N. Granulopoiesis and granules of human neutrophils. Immunol Rev. 2016;273(1):11-28. 4. Levy O. Antimicrobial proteins and peptides: anti-infective molecules of mammalian leukocytes. J Leukoc Biol. 2004; 76(5):909-925. 5. Bennett CM, Kanki JP, Rhodes J, et al. Myelopoiesis in the zebrafish, Danio rerio. Blood. 2001;98(3):643-651. 6. Xu J, Du L, Wen Z. Myelopoiesis during zebrafish early development. J Genet Genomics. 2012;39(9):435-442. 7. Jin H, Huang Z, Chi Y, et al. c-Myb acts in parallel and cooperatively with Cebp1 to regulate neutrophil maturation in zebrafish. Blood. 2016;128(3):415-426. 8. Wang K, Fang X, Ma N, et al. Myeloperoxidase-deficient zebrafish show an augmented inflammatory response to challenge with Candida albicans. Fish Shellfish Immunol. 2015;44(1):109-116. 9. Di Q, Lin Q, Huang Z, et al. Zebrafish nephrosin helps host defence against Escherichia coli infection. Open Biol. 2017; 7(8). 10. Furuyama K, Kaneko K, Vargas PD. Heme as a magnificent molecule with multiple mis-

1794

11.

12.

13.

14.

15.

16.

17.

18.

could not compensate for the function of alas1 in zebrafish, indicating the essential roles of alas1. The feedback mechanisms for heme homeostasis remain unclear, so future studies will be needed to elucidate the molecular mechanisms of heme metabolisms and trafficking. Impaired heme biosynthesis or heme deficiency leads to heme-related disorders, such as anemias, acute porphyrias, and leukemia.11 Acute intermittent porphyria is characterized by the accumulation and/or excretion of excess heme precursors.11 As Alas1 is the key enzyme in heme biosynthesis, repressing Alas1 activity by RNAi is now being used to prevent acute porphyria attacks.45,46 Thus, Alas1-deficient zebrafish may serve as an in vivo animal model for evaluating the risks of therapeutic strategies, since Alas1 deficiency causes neutrophil defects, as well as other potential defects in Alas1-abundant tissues. This study may also contribute to the development of new drugs or treatment strategies for hemerelated diseases. Acknowledgments We thank Dr. Zilong Wen for sharing the Tg(gata1:DsRed) line and eGFP-labeled E. coli strain XL10. Funding This work was supported by the National Natural Science Foundation of China (31471378) and the Team Program of the Guangdong Natural Foundation (2014A030312002).

sions: Heme determines its own fate and governs cellular homeostasis. Tohoku J Exp Med. 2007;213(1):1-16. Tsiftsoglou AS, Tsamadou AI, Papadopoulou LC. Heme as key regulator of major mammalian cellular functions: Molecular, cellular, and pharmacological aspects. Pharmacol Ther. 2006;111(2):327345. Yamamoto M, Hayashi N, Kikuchi G. Evidence for the transcriptional inhibition by heme of the synthesis of delta-aminolevulinate synthase in rat liver. Biochem Biophys Res Commun. 1982;105(3):985990. Suzuki H, Tashiro S, Hira S, et al. Heme regulates gene expression by triggering Crm1dependent nuclear export of Bach1. EMBO J. 2004;23(13):2544-2553. Tahara T, Sun JY, Nakanishi K, et al. Heme positively regulates the expression of betaglobin at the locus control region via the transcriptional factor Bach1 in erythroid cells. J Biol Chem. 2004;279(7):5480-5487. Yamamoto M, Hayashi N, Kikuchi G. Translational inhibition by heme of the synthesis of hepatic delta-aminolevulinate synthase in a cell-free system. Biochem Biophys Res Commun. 1983;115(1):225-231. Chen JJ, London IM. Regulation of protein synthesis by heme-regulated eIF-2 alpha kinase. Trends Biochem Sci. 1995;20(3):105108. Ishikawa H, Kato M, Hori H, et al. Involvement of heme regulatory motif in heme-mediated ubiquitination and degradation of IRP2. Mol Cell. 2005;19(2):171-181. Kubota Y, Nomura K, Katoh Y, Yamashita R, Kaneko K, Furuyama K. Novel mechanisms

19. 20.

21.

22. 23.

24.

25.

26.

27.

for heme-dependent degradation of ALAS1 protein as a component of negative feedback regulation of heme biosynthesis. J Biol Chem. 2016;291(39):20516-20529. Faller M, Matsunaga M, Yin S, Loo JA, Guo F. Heme is involved in microRNA processing. Nat Struct Mol Biol. 2007;14(1):23-29. Ferreira GC. Heme biosynthesis: biochemistry, molecular biology, and relationship to disease. J Bioenerg Biomembr. 1995;27(2):147-150. May A, Bishop DF. The molecular biology and pyridoxine responsiveness of X-linked sideroblastic anaemia. Haematologica. 1998;83(1):56-70. Yamamoto M, Nakajima O. Animal models for X-linked sideroblastic anemia. Int J Hematol. 2000;72(2):157-164. Brownlie A, Donovan A, Pratt SJ, et al. Positional cloning of the zebrafish sauternes gene: a model for congenital sideroblastic anaemia. Nat Genet. 1998; 20(3):244-250. Okano S, Zhou L, Kusaka T, et al. Indispensable function for embryogenesis, expression and regulation of the nonspecific form of the 5-aminolevulinate synthase gene in mouse. Genes Cells. 2010;15(1):7789. Henry KM, Loynes CA, Whyte MKB, Renshaw SA. Zebrafish as a model for the study of neutrophil biology. J Leukoc Biol. 2013;94(4):633-642. Colucci-Guyon E, Tinevez JY, Renshaw SA, Herbomel P. Strategies of professional phagocytes in vivo: unlike macrophages, neutrophils engulf only surface-associated microbes. J Cell Sci. 2011;124(18):30533059. Wang K, Huang Z, Zhao L, et al. Large-scale

haematologica | 2018; 103(11)

Regulation of neutrophil maturation in zebrafish

28.

29.

30.

31.

32.

33.

34.

forward genetic screening analysis of development of hematopoiesis in zebrafish. J Genet Genomics. 2012;39(9):473-480. Westerfield M. The zebrafish book. A guide for the laboratory use of zebrafish (Danio rerio). 4th ed. Univ. of Oregon Press, Eugene, 2000. Hall C, Flores MV, Storm T, Crosier K, Crosier P. The zebrafish lysozyme C promoter drives myeloid-specific expression in transgenic fish. BMC Dev Biol. 2007;7(1):42. Benard EL, van der Sar AM, Ellett F, Lieschke GJ, Spaink HP, Meijer AH. Infection of zebrafish embryos with intracellular bacterial pathogens. J Vis Exp. 2012;15(61):3781. Le Guyader D, Redd MJ, Colucci-Guyon E, et al. Origins and unconventional behavior of neutrophils in developing zebrafish. Blood. 2008;111(1):132-141. Wang SF, Yu XP, Lin ZH, et al. Hemoglobins likely function as peroxidase in blood clam Tegillarca granosa hemocytes. J Immunol Res. 2017;2017(8):7125084. Sadlon TJ, Dell'Oso T, Surinya KH, May BK. Regulation of erythroid 5-aminolevulinate synthase expression during erythropoiesis. Int J Biochem Cell Biol. 1999;31(10):11531167. Lyons SE, Lawson ND, Lei L, Bennett PE, Weinstein BM, Liu PP. A nonsense mutation in zebrafish gata1 causes the bloodless phe-

haematologica | 2018; 103(11)

35.

36.

37.

38.

39.

notype in vlad tepes. Proc Natl Acad Sci USA. 2002;99(8):5454-5459. Khan AA, Quigley JG. Control of intracellular heme levels: heme transporters and heme oxygenases. Biochim Biophys Acta. 2011;1813(5):668-682. Holowiecki A, O'Shields B, Jenny MJ. Spatiotemporal expression and transcriptional regulation of heme oxygenase and biliverdin reductase genes in zebrafish (Danio rerio) suggest novel roles during early developmental periods of heightened oxidative stress. Comp. Biochem. Physiol. C Toxicol. Pharmacol. 2017;191:138-151. Mercurio S, Petrillo S, Chiabrando D, Bassi ZI, Gays D, Camporeale A, et al. The heme exporter Flvcr1 regulates expansion and differentiation of committed erythroid progenitors by controlling intracellular heme accumulation. Haematologica. 2015; 100(6):720729. Rajagopal A, Rao AU, Amigo J, Tian M, Upadhyay SK, Hall C, et al. Haem homeostasis is regulated by the conserved and concerted functions of HRG-1 proteins. Nature. 2008;453(7198):1127-1131. Korolnek T, Zhang J, Beardsley S, Scheffer GL, Hamza I. Control of metazoan heme homeostasis by a conserved multidrug resistance protein. Cell Metab. 2014; 19(6):1008-1019.

40. Beru N, Sahr K, Goldwasser E. Inhibition of heme synthesis in bone marrow cells by succinylacetone: effect on globin synthesis. J Cell Biochem. 1983;21(2):93-105. 41. Dinarello CA. Interleukin-1 in the pathogenesis and treatment of inflammatory diseases. Blood. 2011;117(14):3720-3732. 42. de Oliveira S, Reyes-Aldasoro CC, Candel S, Renshaw SA, Mulero V, Calado A. Cxcl8 (IL8) mediates neutrophil recruitment and behavior in the zebrafish inflammatory response. J Immunol. 2013;190(8):43494359. 43. Cao C, Fleming MD. The ins and outs of erythroid heme transport. Haematologica. 2015;100(6):703. 44. Bagger FO, Sasivarevic D, Sohi SH, et al. BloodSpot: a database of gene expression profiles and transcriptional programs for healthy and malignant haematopoiesis. Nucleic Acids Res. 2016;44(D1):D917-924. 45. Chan A, Liebow A, Yasuda M, et al. Preclinical development of a subcutaneous ALAS1 RNAi therapeutic for treatment of hepatic porphyrias using circulating RNA quantification. Mol Ther Nucleic Acids. 2015;4(11):e263. 46. Balwani M, Wang B, Anderson KE, et al. Acute hepatic porphyrias: recommendations for evaluation and long-term management. Hepatology. 2017; 66(4):1314-1322.

1795

ARTICLE

Iron Metabolism & its Disorders

Ferrata Storti Foundation

Haematologica 2018 Volume 103(11):1796-1805

The SLC40A1 R178Q mutation is a recurrent cause of hemochromatosis and is associated with a novel pathogenic mechanism Chandran Ka,1,2,3* Julie Guellec,1,3,4* Xavier Pepermans,5 Caroline Kannengiesser,3,6,7 Cécile Ged,7,8 Wim Wuyts,9 David Cassiman,10 Victor de Ledinghen,11 Bruno Varet,12 Caroline de Kerguenec,13 Claire Oudin,6 Isabelle Gourlaouen,1,3 Thibaud Lefebvre,6 Claude Férec,1,2,7 Isabelle Callebaut14 and Gérald Le Gac1,2,3,7

UMR1078, INSERM, Université Bretagne Loire – Université de Bretagne Occidentale, Etablissement Français du Sang – Bretagne, Institut Brestois Santé-Agro-Matière, Brest, France; 2Laboratoire de Génétique Moléculaire et Histocompatibilité, CHRU de Brest, Hôpital Morvan, France; 3Laboratory of Excellence GR-Ex, Paris, France; 4Association Gaetan Saleun, Brest, France; 5Center for Human Genetics, University Hospital of StLuc, Brussels, Belgium; 6UMR1149, INSERM, Centre de Recherche sur l'Inflammation, Université Paris Diderot, AP-HP, Hôpital Bichat, Département de Génétique, France; 7On behalf of the French National Network for the Molecular Diagnosis of Inherited Iron Overload Disorders (J. Rochette, E. Cadet, C. Kannengiesser, H. Puy, C. Ged, H. de Verneuil, G. Le Gac, C. Férec, S. Pissard, V. Gérolami), Brest, France; 8INSERM U1035, BMGIC, CHU de Bordeaux, Laboratoire de Biochimie et Biologie Moléculaire, France; 9 Department of Medical Genetics, University and University Hospital of Antwerp, Edegem, Belgium; 10Department of Gastroenterology-Hepatology and Metabolic Center, University Hospital of Leuven, Belgium; 11Department of Gastroenterology and Digestive Oncology, University Hospital of Bordeaux, France; 12Université Paris Descartes et AP-HP, Hôpital Necker, Service d’Hématologie, France; 13AP-HP, Hôpital Beaujon, Département d'Hépatologie, Clichy, France and 14UMR7590, CNRS, Sorbonne Universités, Université Pierre et Marie Curie-Paris, France 1

*CK and JG contributed equally to this work

ABSTRACT

Correspondence: gerald.legac@univ-brest.fr

Received: January 29, 2018. Accepted: July 6, 2018. Pre-published: July 12, 2018. doi:10.3324/haematol.2018.189845 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1796 ©2018 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.

1796

emochromatosis type 4 is one of the most common causes of primary iron overload, after HFE-related hemochromatosis. It is an autosomal dominant disorder, primarily due to missense mutations in SLC40A1. This gene encodes ferroportin 1 (FPN1), which is the sole iron export protein reported in mammals. Not all heterozygous missense mutations in SLC40A1 are disease-causing. Due to phenocopies and an increased demand for genetic testing, rare SLC40A1 variations are fortuitously observed in patients with a secondary cause of hyperferritinemia. Structure/function analysis is the most effective way of establishing causality when clinical and segregation data are lacking. It can also provide important insights into the mechanism of iron egress and FPN1 regulation by hepcidin. The present study aimed to determine the pathogenicity of the previously reported p.Arg178Gln variant. We present the biological, clinical, histological and radiological findings of 22 patients from six independent families of French, Belgian or Iraqi decent. Despite phenotypic variability, all patients with p.Arg178Gln had elevated serum ferritin concentrations and normal to low transferrin saturation levels. In vitro experiments demonstrated that the p.Arg178Gln mutant reduces the ability of FPN1 to export iron without causing protein mislocalization. Based on a comparative model of the 3D structure of human FPN1 in an outward facing conformation, we argue that p.Arg178 is part of an interaction network modulating the conformational changes required for iron transport. We conclude that p.Arg178Gln represents a new category of loss-of-function mutations and that the study of “gating residues” is necessary in order to fully understand the action mechanism of FPN1.

haematologica | 2018; 103(11)

Ferroportin structure and function

Introduction Hemochromatosis type 4 (OMIM #606069), also called ferroportin disease, is an inborn error of iron metabolism transmitted through autosomal dominant inheritance and associated with mutations in the gene encoding the solute-carrier family 40 member 1 (SLC40A1). Although rare, hemochromatosis type 4 is observed in different ethnic groups and is considered to be the second most common cause of hereditary iron overload after HFE-related hemochromatosis.1,2 SLC40A1, also known as ferroportin 1 (UniProt accession number Q9NP59), is the sole iron export-protein reported in mammals. It is expressed in all types of cells that handle major iron flow, including macrophages, duodenal enterocytes, hepatocytes and placenta syncytiotrophoblasts.3 Expression of ferroportin 1 on the cell surface is predominantly regulated by the liver-derived peptide hepcidin, which induces internalization and degradation of ferroportin 1 and thereby decreases the delivery of iron to plasma.4 The hepcidin-ferroportin axis plays an important role in the pathogenesis of inherited and acquired iron metabolism disorders, including iron overload diseases and iron-restricted anemia.5 Mutations that alter ferroportin 1 function are expected to produce stronger effects in reticuloendothelial macrophages than in enterocytes or hepatocytes. These macrophages acquire most of their iron by recycling senescent red blood cells and account for >80% of the daily iron flux within the body.6 In line with this expectation, and at variance with HFE-related hemochromatosis, patients with loss-of-function mutations usually present with mesenchymal or mixed iron overload (corresponding to early iron deposition within Kupffer cells) and markedly elevated serum ferritin levels, contrasting with normal or low transferrin saturation values.7,8 Aggressive phlebotomy regimens can be a problem in the early stages of the disease, when patients often display borderline anemia.9,10 Although a majority of SLC40A1 mutations reported as being causally linked to hemochromatosis type 4 are true “pathogenic variants”,11 there may be some doubt in the case of variants for which phenotypic, population, segregation, functional and/or computational data are lacking or not fully convincing. The problem is not specific to hemochromatosis type 4, but includes all Mendelian disorders associated with large allelic heterogeneity, and can be mimicked by non-genetic conditions.12,13 Assessing the pathogenicity of 18 non-synonymous SLC40A1 variants found in 44 suspected hemochromatosis 4 patients, we previously demonstrated that eight very rare missense mutations had no noticeable effects on ferroportin 1 function or interaction with hepcidin.14 All these variants were identified in single cases showing moderate serum ferritin elevation and normal transferrin saturation, a biological condition that is common in clinical practice and is largely related to lifestyle and environmental factors.15 The present study provides strong evidence that the SLC40A1 p.Arg178Gln missense mutation is recurrent in the SLC40A1 gene of patients showing typical reticuloendothelial iron overload. We further demonstrate that the p.Arg178Gln ferroportin 1 mutant shows reduced ability to export iron out of the cell. This is likely a direct consequence of salt bridge disruption between Arg178 and Asp473, thereby affecting the stable formation of the haematologica | 2018; 103(11)

intracellular gate present in the ferroportin 1 outward facing state. Such a molecular mechanism of pathogenesis has never been reported in the context of hemochromatosis type 4.

Methods Genetic studies DNA was extracted from the peripheral blood of patients and unaffected family members. The complete coding sequence of SLC40A1 and intron/exon boundaries was investigated by Sanger sequencing in the probands, while family members were only assessed for exon 6 (containing codon p.Arg178). All probands were negative for genotypes known to cause hemochromatosis types 1-3 (in the HFE, HFE2, HAMP and TFR2 genes) and for mutations in the FTL and BMP6 genes (hyperferritinemia cataract syndrome: OMIM#600886; hyperferritinemia without iron overload and cataract: OMIM#134790; BMP6-related iron overload: OMIM#112266). Polymerase chain reaction and sequencing conditions are available upon request. A total of 734 DNA samples from healthy subjects, exclusively from north-western France (Brittany), were investigated to control the frequency of the SLC40A1 p.Arg178Gln variant. Informed consent for molecular studies was obtained from all patients and family members, in accordance with the Declaration of Helsinki; in line with French ethical guidelines, the Clinical Research Ethics Committee of the University Hospital of Brest approved the study on October 25, 2010.

Hepcidin measurement in human serums Serum hepcidin concentrations were measured using liquid chromatography coupled with tandem mass spectrometry (LCMS/MS), as previously described.16 The 95% reference interval obtained for normal hepcidin (200 serum samples from healthy subjects) ranged from 1.0 and 20.8 ng/mL (mean: 8.2 ng/mL).

Human-25 hepcidin synthesis and secretion by T-Rex-293 cells Human HAMP cDNA was amplified with reverse-transcription polymerase chain reaction (RT-PCR) from total ribonucleic acid (RNA) isolated from human liver hepatocellular carcinoma HepG2 cells. The PCR product was cloned into the PCR2.1 vector using the TA Cloning Kit (Thermo Fisher Scientific), subcloned into the pcDNA4™4/TO tetracycline-regulated mammalian expression vector (Thermo Fisher Scientific), and checked by sequencing. T-Rex-293 cells (Thermo Fisher Scientific) were stably transfected with calcium phosphate, and colonies were selected in the presence of 1.5 mg/ml blasticidin and 100 mg/ml zeocin for four weeks. Tetracycline (Sigma, St. Louis, MO, USA) was used to induce expression of the 84 hepcidin pre-propeptide amino acids from 1.106 T-Rex-293 cells. After 48h the cell supernatant was collected, filtered through a hydrophilic nylon membrane (pore size: 0.2 mm), and measured for hepcidin-25 levels using a commercially available competitive enzyme-linked immunosorbent assay kit (ELISA; Peninsula Laboratories International, San Carlos, CA, USA). The supernatant was stored at -20°C until used.

In vitro experiments In vitro investigations were performed as previously described14,17 and are presented in detail in the Online Supplementary Data.

3D structure modeling and analysis Models of the 3D structure of human ferroportin 1 were built using Modeller v9.15,18 considering the sequence alignment 1797

C. Ka et al.

Table 1. Biological and clinical data of index cases and their relatives.

FamilyCountry 1

2 3

5 6

Family Gender Age* SF TS HIC AST ALT GGT CRP RBC Hb Ht relationships (years) (mg/L) (%) ( mol/g) (IU/L) (IU/L) (IU/L) (mg/L)(Tera/L) (g/dL) (%)

Belgium Index case Father Daughter Son France Index case Son France Index case Brother Father Grandmother Uncle First cousin First cousin Belgium Index case Daughter Son Brother Godson Goddaughter Iraq Index case France Index case Son

M M F M M M M M M F M F M M F M M M F M M M

39 61 11 7 55 20 12 6 35 72 45 21 22 54 29 25 59 35 32 71 57 27

1873 1190 322 552 1452 976 773 382 1200 284 487 367 584 1114 263 613 1164 1354 372 2306 1386 1066

36 27 33

180 85 60

37 54

58 98

61 23

1.5 2.3

4.75 5.73

15,7 15,2 12,6 12,9 15.6 18.4

250

38 25

30 11

43 22

6 3

14,4 15,4

MCV (fL)

45 42,5

Clinical Alcohol observations consumption** Fatigue

45,8 52.1

96,5 90.9

45,3

97 91

Moderate

Abstinent Abstinent Digestive problems Abstinent Obesity Abstinent Abstinent Abstinent

28 26 21 28 36 25 29 43 33

Fatigue Hepatomegaly

Abstinent Moderate

*Age at diagnosis. **Moderate: up to 1 drink per day for women and up to 2 drinks per day for men. SF: serum ferritin (normal value ≤ 200 μg/L in females, ≤ 300 in males); TS: transferrin saturation (normal value < 45%); HIC: hepatic iron concentration (normal value < 36 mmol/g); AST: aspartate aminotransferase (normal range: 5-50 IU/L); ALT: alanine aminotransferase (normal range: 5-50 IU/L); GGT: gamma-glutamyl transferase (normal range: 5-55 IU/L); CRP: C-reactive protein; RBC: red blood cell (normal range: 4.0-5.7 x 1012/L); Hb: hemoglobin (normal range: 12.018.0 g/dL); Ht: hematocrit (normal range: 37-52%); MCV: mean corpuscular volume (normal range: 80-95 fL).

reported by Taniguchi et al. in 201519 and, as templates, the experimental 3D structures of Bdellovibrio bacteriovorus (Bb) iron transporter Bd2019 in outward and inward facing conformations (pdb 5aym and 5ayo,19 respectively).

Statistical analysis Data are presented as scatter dot plots and means. Comparisons used a one-tailed Student’s t-test.

Results Clinical data and segregation analysis The SLC40A1 p.Arg178Gln (c.533G>A) variant was identified in 22 patients from six independent families (Figure 1). It co-segregated with hyperferritinemia (defined as serum ferritin > 300 mg/L in males and > 200 mg/L in females) in all the tested patients, irrespective of age at testing, and was absent in three non-affected members of families 3 and 4. Index cases comprised six males, aged 12-72 years at diagnosis, who displayed significant hyperferritinemia and non-elevated transferrin saturation (Table 1). Hepatic Iron Concentration (HIC) was evaluated using the Gandon’s magnetic resonance imaging (MRI) method20 in three patients with values between 60 (youngest case) and 250 mmol/g (oldest case) dry weight. Liver biopsy, performed in the index case of family 1, confirmed iron 1798

overload, with a predominance of iron deposition in Kupffer cells (Figure 2). The patient showed no clinical symptoms, except for persistent fatigue. The platelet count (178x109/L; normal range: 120-369x109/L) was not suggestive of a fibrotic liver disease. The patient started a phlebotomy program. After 24 phlebotomies (450 mL every three weeks), the hemoglobin level dropped to 13.6 g/dL, from 15.7 g/dL at diagnosis (normal range: 13.5 – 17.5). In total, the cohort consisted of 17 males and five females. Women had lower serum ferritin levels than men (263-372 versus 487-2306 mg/L), while four children exhibited increased iron store between the ages of six and 12. Carrier testing in families identified 16 patients for whom no clinical manifestations of iron overload were reported. Serum hepcidin measurements were obtained before therapeutic phlebotomy in two affected individuals of family 3 (3.II.5 and 3.III.4). They were above the reported range in healthy individuals for the assay used (1.0-21 ng/mL; liquid chromatography coupled with tandem mass spectrometry).16 The proband’s grandmother (3.I.1) started therapeutic phlebotomy (1-2 venesections per year; well tolerated) several years before family screening. She showed moderately increased ferritin (280 µg/L) at the time of serum hepcidin evaluation (41.6 ng/mL). In contrast, normal serum hepcidin levels (8.2 ng/mL) were detected in the proband’s brother who was negative for haematologica | 2018; 103(11)

Ferroportin structure and function

Figure 1. Family studies and pedigrees with the p.Arg178Gln missense mutation. Arrows indicate the index case. Biological data of family members are presented, when available. TS: transferrin saturation; SF: serum ferritin.

haematologica | 2018; 103(11)

1799

C. Ka et al.

Figure 2. Liver histology of the index case of family 1 at diagnosis. Perlsâ&#x20AC;&#x2122; stain of liver biopsy shows significant amounts of stainable iron in Kupffer macrophages, while mild iron overload is observed in hepatocytes.

the SLC40A1 p.Arg178Gln missense mutation and had normal iron indices (3.III.1). It is noteworthy that the index case of family 2 and his son had two co-existing conditions of hyperferritinemia: hemochromatosis type 4 and hepatic steatosis. Transient Elastography-based Controlled Attenuation Parameter (TE-CAP) measurements revealed grade 3 severe steatosis in both family members (CAP scores: 396 and 363 dB/m, respectively). The index case was a 55-year-old male with 36% transferrin saturation and a serum ferritin level of 1,452 mg/L at diagnosis. He presented with a waist circumference of 104 cm, in a context of subnormal laboratory metabolic and liver tests: uric acid (453 mol/L; normal range: 240-420), total cholesterol (7 mmol/L; normal range: 3.5-6.0), alanine aminotransferase (58 IU/L; normal range: 5-50) and gamma-glutamyl transferase (61 IU/L; normal range: 5-55). Blood pressure, aspartate aminotransferase, triglycerides, HDL cholesterol and fasting blood glucose were within normal ranges. Abdominal magnetic MRI showed significantly reduced liver signal intensity, consistent with advanced iron overload (HIC: 180 Âľmol/g). The patient started venesection therapy when he turned 60 years old, after being diagnosed with prostate cancer and colon polyps. The phlebotomy program (500 mL every two weeks for eight months, then monthly for four months) was well tolerated. His son, who became overweight during infancy (waist circumference at diagnosis: 105 cm; Body Mass Index: 31.7 kg/m2), presented 1800

similar iron indices at the age of 20 years (transferrin saturation: 27%; serum ferritin: 976 mg/L). MRI, however, revealed a moderate increase in hepatic iron store (HIC: 85 mmol/g).

Functional characterization of the ferroportin 1 p.Arg178Gln variant The functional significance of the p.Arg178Gln variant was determined by first investigating its subcellular localization. Wild-type (WT) and mutant ferroportin 1-V5 constructs were expressed in HEK293T cells, and plasma membrane localization of the V5-tagged proteins was assayed by Western blot and densitometry. HLA-A was used as the control and standard for normalization, being a cell-surface protein with no known role in iron metabolism. The p.Ala77Asp missense mutation, which significantly damages ferroportin 1 structure and is known to prevent cell-surface localization,17,21 was used as the negative control. The p.Arg178Gln mutant was properly localized on the cell surface, comparable to the WT protein (Figure 3A). Next, the iron-exporting function of the ferroportin 1 p.Arg178Gln variant was assessed using radioactively labeled iron. HEK293T cells were grown in 20 mg/L 55 Fe-transferrin for 24 hr, washed, transiently transfected, and placed in a serum-free medium. The amount of 55Fe exported into the supernatant was measured after a period of 36 hr using cells transfected with the commercial haematologica | 2018; 103(11)

Ferroportin structure and function

pcDNA3.1-V5-His vector as negative control (no FPN1). As shown in Figure 3B, cells transfected with WT ferroportin 1-V5 displayed a 3-fold increase in iron release. The p.Arg178Gln variant was not able to export iron 55Fe in amounts comparable with WT ferroportin 1, but was more active than the p.Ala77Asp control; Studentâ&#x20AC;&#x2122;s t-tests highlighted significant differences between both variants and WT ferroportin 1 (P<0.001 and 0.0001, respectively) and between the two variants (P<0.0001). To investigate whether the p.Arg178Gln missense mutation could modify response to hepcidin, transiently transfected HEK293T cells were cultured for 24 hr with conditioned media derived from T-Rex-293 cells stably transfected with full-length human HAMP cDNA. Supernatant human-25 hepcidin concentration was determined using a competitive enzyme-linked immunosorbent assay. Two known ferroportin 1 mutants served as positive controls: p.Asn144His, which shows partial resistance to hepcidin inhibition,22 and p.Cys326Tyr, which abolishes hepcidin binding to ferroportin 1 and is responsible for complete resistance.23 As expected, the addition of hepcidin to cells expressing WT ferroportin 1 resulted in the disappearance of the iron exporter from the plasma membrane. The Western blot pattern of the p.Asp178Gln variant, unlike the p.Cys326Tyr and p.Asn144His mutants, was similar to that of the WT protein (Figure 4).

Structural and functional investigation of the intermolecular interaction between the N and C lobes of human ferroportin 1 â&#x20AC;&#x201C; 3D structure form I (outward facing) In order to understand the possible impact of the p.Arg178Gln mutation, we modeled the 3D structure of human ferroportin in both the outward and inward facing conformations, based on the recent 3D structures of the Bb iron transporter Bd2019, which shares 24% sequence identity with human ferroportin 1.19 As illustrated in Figure 5, Arg178 forms an inter-lobe salt bridge with Asp473 in the outward facing conformation of human ferroportin 1. We hypothesized that this noncovalent interaction between helix TM5 and helix TM8, located respectively in the N and C lobes, might be important in stabilizing the outward facing conformation of ferroportin 1, and that its disruption might cause a significant reduction in iron egress. To check this hypothesis, we replaced arginine 178 and aspartic acid 473 with alanine, which is the smallest amino acid after glycine and is neutral, being non-polar and devoid of any strong hydrophobic character. Moreover, it has the highest propensity of the 20 amino acids for the a-helical state;24 thus this modification was likely to have limited impact on local structure. As shown in Figure 6, the Asp473Ala mutant did not cause obvious mislocalization of the protein, which was, however, totally inoperative for iron export. Indeed, cells expressing the Asp473Ala mutant retained 55Fe in amounts comparable to cells expressing the two known p.Asn77Asp and p.Val162del loss-of-function mutations. The Arg178Ala mutant reduced cell surface expression to half that of the WT, but with less influence in iron export ability. We then examined whether charge swapping could restore the iron export function of ferroportin 1, and whether Arg178Gln dysfunction could be corrected by a p.Asp473Arg mutation. The p.Arg178Asp and haematologica | 2018; 103(11)

Figure 3. The ferroportin p.Arg178Gln mutant shows normal cell surface expression, but substantial loss of iron export. (A) HEK293T cells were transiently co-transfected with plasmids encoding either a V5-tagged ferroportin protein (WT or variant) or a V5-tagged HLA-A protein. Human leukocyte antigen (HLA)-A was used as control and standard for normalization, being a cell surface protein with no known role in iron metabolism. At 24h after transfection, cell surface proteins were selectively purified and analyzed by Western blotting using a peroxidase conjugated mouse anti-V5 antibody. Densitometric scans of SLC40A1 levels (normalized to HLA-A) are shown in the lower part of the figure. The results of three independent experiments are presented. (B) HEK293T cells were grown in 20 mg/mL 55Fe-transferrin for 24h before being washed and transiently transfected with WT or mutated SLC40A1-V5 expression plasmids. After 15h, cells were washed and then serum-starved. The 55Fe exported into the supernatant was collected at 36h. Data are presented as percentage of cellular radioactivity at time zero. Each point represents the average value (from triplicate) of five independent experiments. P values were calculated with the Studentâ&#x20AC;&#x2122;s t-test; **P<0.01 and ****P<0.0001. WT: wild-type.

p.Asp473Arg substitutions almost abolished cell surface expression of the protein. Introducing the two p.Arg178Asp and p.Asp473Arg missense mutations did not rescue the membrane expression level decreased by single mutations. The Arg178Gln/Asp473Arg double mutant also resulted in a strong reduction in cell surface expression (Online Supplementary Figure S1). Taken together, these results suggested that the salt bridge between arginine 178 and aspartic acid 473 is essential for ferroportin 1 iron export function. Any subtle changes in charge or size on the side chains may cause loss of function. 1801

C. Ka et al.

Discussion The p.Arg178Gln missense mutation is typical of the ferroportin 1 variants for which evidence remains inadequate and clinical pathogenicity doubtful. Herein, we provide a comprehensive genotype-phenotype analysis, and argue

that p.Arg178Gln substitution abolishes a salt bridge between the N and C lobes of human ferroportin 1, leading to a less stable outward facing state and, thus, to an altered equilibrium between the different conformational states. The p.Arg178Gln missense mutation was first reported in a 70-year-old female with hyperferritinemia and normal

Figure 4. The ferroportin p.Arg178Gln mutant is not resistant to hepcidin. HEK293T cells were transiently co-transfected with plasmids expressing HLA(A)-V5 and either wild-type SLC40A1-V5 (WT) or SLC40A1-V5 variants. At 16h post-transfection, the cells were incubated in the presence or absence of hepcidin for 3h. Plasma membrane proteins were purified and analyzed by Western blotting and densitometry. Data are expressed as the percentage of ferroportin in cells not treated by hepcidin, according to the formula 100 x (SLC40A1 â&#x20AC;&#x201C; hepcidin / SLC40A1 + hepcidin). Error bars are the standard deviation of three independent experiments.

TM6

TM1

TM5

TM3 TM4

TM8 R178 N174

Q481

D157

D473

R88 E486

R489 TM2

TM10

TM9

TM7 TM12

D84 TM11

Figure 5. Ribbon representation of a human ferroportin-1 3D structure model in an outward facing conformation, with atomic representation of amino acids involved in non-covalent bonds, likely leading to the stabilization of this conformation. The iron within the iron-binding site is represented as a red ball at the top of the figure. This figure was drawn using Chimera.38

1802

haematologica | 2018; 103(11)

Ferroportin structure and function

transferrin saturation,25 and then observed with incomplete penetrance in members of an independent Greek family. Serum ferritin was found to be markedly elevated in the proband (a 25-year-old female), but only slightly in her 53-year-old mother and within the normal range in her 87-year-old grandfather.26,27 We observed the p.Arg178Gln missense mutation in 11 adult males with high serum ferritin concentrations (>970 mg/L) and normal plasma iron levels (transferrin saturation: 26-43%). Three adult females from the larger pedigrees (3.III.4, 4.II.1 and 4.II.6) displayed a mild phenotype, with serum ferritin concentrations ranging between 200 and 400 mg/L at diagnosis, while two males presented an intermediate phenotype (584 and 613 mg/L) in their third decade (3.III.6 and 4.II.3). Liver biopsy was available for the index case of family 1, and showed tissue iron overload, with iron deposits primarily observed in non-parenchymal cells. The index case

had two children who exhibited increased iron stores at 11 and seven years of age, respectively. Similar phenotypes were observed in two children from family 3 (3.III.2 and 3.III.3). None of the patients were reported to have developed significant fibrosis or cirrhosis. Taken together, these data indicate that the p.Arg178Gln substitution is responsible for the classic form of ferroportin disease, or hemochromatosis type 4A. The expressivity of the SLC40A1 p.[Arg178Gln];[=] genotype is variable, with milder phenotypes observed in women. High serum ferritin levels can be observed in young patients, highlighting the fact that tissue iron overload may appear early in life and that, in contrast to HFE hemochromatosis, diagnosis of ferroportin disease should not be restricted to adults.28 This recapitulates some previous observations on the relationship between ferroportin loss-of-function mutations and mild to severe reticuloendothelial iron overload.2,11,29

Figure 6. Effect of p.Arg178Ala and p.Asp473Ala ferroportin variants on cell surface expression and iron export. (A) HEK293T cells were transiently co-transfected with plasmids encoding either a V5-tagged ferroportin protein (WT or variant) or a V5-tagged HLA-A protein. At 24h after transfection, cell surface proteins were selectively purified and analyzed by Western blotting using peroxidase-conjugated mouse anti-V5 antibody. Densitometric scans of SLC40A1 levels (normalized to HLA-A) are shown. The error bars represent the standard deviation of three independent experiments. (B) HEK293T cells were grown in 20 mg/mL 55Fe-transferrin for 24h before being washed and transiently transfected with WT or mutated SLC40A1-V5 expression plasmids. After 15h, cells were washed and then serum-starved for up to 36h. 55Fe exported into the supernatant was collected at various time points. Data are presented as a percentage of cellular radioactivity at time zero. Each point represents the mean standard deviation; n=3 in each group. The data are representative of three separate experiments. WT: wild-type.

haematologica | 2018; 103(11)

1803

C. Ka et al.

In contrast to all other types of hemochromatosis, which are characterized by hepcidin deficiency, we found increased serum hepcidin levels in three affected individuals from family 3. This recapitulates previous observations in seven patients with the recurrent and well-characterized p.Val162del loss-of-function ferroportin 1 mutation.11,30,31 M. Speletas et al. and S. Cunat et al. failed to detect the p.Arg178Gln missense mutation in the DNA of 253 bone marrow donors from central Greece27 and 50 French controls.25 The results presented herein of 730 DNA samples from healthy subjects born in the western part of France (Brittany) were identical. The variant was also absent in GnomAD, which is an extension of the Exome Aggregation Consortium (ExAC) database and includes 123,136 exome sequences and 15,496 whole-genome sequences from unrelated individuals sequenced as part of various disease-specific and population genetic studies. The p.Arg178Gln missense mutation can thus be expected to be very rare; nevertheless, it has now been associated with hyperferritinemia in 25 patients from France, Belgium, Greece and Iraq. This provides another indication of the pathogenicity of the SLC40A1 p.Arg178Gln allele, which is not restricted to European populations. We and others have previously shown that the p.Arg88Gly, p.Ile152Phe and p.Asn174Ile clinical mutations are defective in terms of iron egress while being normally addressed to the cell surface.14,17,32,33 In the present study, we demonstrated that the p.Arg178Gln substitution follows an identical trend (Figure 3). This was not a typical situation of loss of ferroportin 1 function, which is usually associated with protein mislocalization.14,21 This prompted us to look at the 3D structure of human ferroportin 1 and examine the molecular mechanism responsible for reduced iron export. Ferroportin 1 is a member of the major facilitator superfamily (MFS),17 which is the largest group of secondary active membrane transporters, essential for the movement of a wide range of substrates across biological membranes.34 In recent years, the number of experimental 3D structures has increased dramatically, leading to a better appreciation of the conformational changes that are needed for effective MFS-mediated transport.35,36 All MFS transporters share a common and characteristic core fold that is organized in two similar domains (N and C lobes), each consisting of six consecutive transmembrane segments (TM1-TM6 and TM7-TM12). They progress through a conformational cycle that involves at least four conformational states: inward open state, the ligandbound and ligand-free occluded states, and outward open state.36 These conformational changes are orchestrated by

References 1. Wallace DF, Subramaniam VN. The global prevalence of HFE and non-HFE hemochromatosis estimated from analysis of nextgeneration sequencing data. Genet Med. 2016;18(6):618-626. 2. Pietrangelo A. Ferroportin disease: pathogenesis, diagnosis and treatment. Haematologica. 2017;102(12):1972-1984.

1804

a set of specific residues that mediate interactions between the N and C lobes.35 The recent report of the crystal structures of a putative bacterial homologue of ferroportin 1 (BbFPN) and the description of inter- and intra-domain conformational rearrangements during the transport of iron open up new avenues for predicting the atomic details of the organization of human ferroportin transmembrane helices and elucidating the detailed mechanisms of iron egress.19 Such structural investigation has already been conducted, using more distant 3D structures as a template.14,17,37 In the present study, we built an outward-open conformation of human ferroportin 1, using the experimental structure of BbFPN as a template. This allowed us to specifically investigate interactions between TM3, TM4 and TM5 of the N lobe and TM8 and TM9 of the C lobe (Figure 5). We demonstrated that Arg178 (TM5) forms a salt bridge with Asp473 (TM8). This bond may act in the same way as those observed between Asn174 (TM5) and Gln481 (TM10), or between Arg88 (TM3) and Glu486 (TM11) and between Asp157 (TM4) and Arg489 (TM11), thus extending the definition of an interaction network on the intracellular side of the outward facing structure of BbFPN.19 That the interaction between Arg178 and Asp473 is important in stabilizing human ferroportin 1 in the outward facing state is further supported by the results presented in Figure 6, where replacing arginine 178 or aspartic acid 473 by alanine strongly decreased the ability of ferroportin 1 to export iron. The reason why the Asp473Ala mutant showed a smaller impact on protein trafficking than Arg178Ala, despite its stronger effect on iron egress, remains to be elucidated. One possibility is that the interruption of the charge-helix dipole interaction (Arg178 â&#x20AC;&#x201C; Asp 181) may destabilize the local structure of TM3. To conclude, the present study demonstrates the causality of the p.Arg178Gln missense mutation, which is now considered to be one of the most frequent SLC40A1 loss-offunction mutations. It also reveals a new molecular mechanism of disease, involving residues that participate in the stabilization of the different conformational states and thus mediate iron export. These findings can be extended to the functional interpretation of other rare missense mutations that are associated with typical reticuloenthelial iron overload but do not significantly alter the cell surface expression of ferroportin 1. They also confirm that it is essential to identify so-called â&#x20AC;&#x153;gating residuesâ&#x20AC;? in order to fully understand the action mechanism of MFS transporters.35 Funding The authors would like to thank the French Hospital Clinical Research Program (Progamme Hospitalier de Recherche Clinique 2009) for funding; Brest University Hospital UF0857.

3. Donovan A, Lima CA, Pinkus JL, et al. The iron exporter ferroportin/Slc40a1 is essential for iron homeostasis. Cell Metab. 2005; 1(3):191-200. 4. Drakesmith H, Nemeth E, Ganz T. Ironing out ferroportin. Cell Metab. 2015; 22(5):777787. 5. Girelli D, Nemeth E, Swinkels DW. Hepcidin in the diagnosis of iron disorders. Blood. 2016;127(23):2809-2813.

6. Ganz T. Macrophages and iron metabolism. Microbiol Spectr. 2016;4(5). 7. Pietrangelo A. The ferroportin disease. Blood Cells Mol Dis. 2004;32(1):131-138. 8. Cremonesi L, Cemonesi L, Forni GL, et al. Genetic and clinical heterogeneity of ferroportin disease. Br J Haematol. 2005; 131(5):663-670. 9. Pietrangelo A, Montosi G, Totaro A, et al. Hereditary hemochromatosis in adults

haematologica | 2018; 103(11)

Ferroportin structure and function

10.

11.

12.

13.

14.

15.

16.

17.

without pathogenic mutations in the hemochromatosis gene. N Engl J Med. 1999;341(10):725-732. McDonald CJ, Wallace DF, Ostini L, Bell SJ, Demediuk B, Subramaniam VN. G80Slinked ferroportin disease: classical ferroportin disease in an Asian family and reclassification of the mutant as iron transport defective. J Hepatol. 2011;54(3):538-544. Mayr R, Janecke AR, Schranz M, et al. Ferroportin disease: a systematic metaanalysis of clinical and molecular findings. J Hepatol. 2010;53(5):941-949. MacArthur DG, Manolio TA, Dimmock DP, et al. Guidelines for investigating causality of sequence variants in human disease. Nature. 2014;508(7497):469-476. 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. Callebaut I, Joubrel R, Pissard S, et al. Comprehensive functional annotation of 18 missense mutations found in suspected hemochromatosis type 4 patients. Hum Mol Genet. 2014;23(17):4479-4490. Adams PC, Barton JC. A diagnostic approach to hyperferritinemia with a nonelevated transferrin saturation. J Hepatol. 2011;55(2):453-458. Lefebvre T, Dessendier N, Houamel D, et al. LC-MS/MS method for hepcidin-25 measurement in human and mouse serum: clinical and research implications in iron disorders. Clin Chem Lab Med. 2015; 53(10): 1557-1567. Le Gac G, Ka C, Joubrel R, et al. Structurefunction analysis of the human ferroportin iron exporter (SLC40A1): effect of hemochromatosis type 4 disease mutations and identification of critical residues. Hum Mutat. 2013;34(10):1371-1380.

haematologica | 2018; 103(11)

18. Martí-Renom MA, Stuart AC, Fiser A, Sánchez R, Melo F, Sali A. Comparative protein structure modeling of genes and genomes. Annu Rev Biophys Biomol Struct. 2000;29:291-325. 19. Taniguchi R, Kato HE, Font J, et al. Outwardand inward-facing structures of a putative bacterial transition-metal transporter with homology to ferroportin. Nat Commun. 2015;6:8545. 20. Gandon Y, Olivié D, Guyader D, et al. Noninvasive assessment of hepatic iron stores by MRI. Lancet. 2004;363(9406):357-362. 21. Schimanski LM, Drakesmith H, Merryweather-Clarke AT, et al. In vitro functional analysis of human ferroportin (FPN) and hemochromatosis-associated FPN mutations. Blood. 2005;105(10):4096-4102. 22. Drakesmith H, Schimanski LM, Ormerod E, et al. Resistance to hepcidin is conferred by hemochromatosis-associated mutations of ferroportin. Blood. 2005;106(3):1092-1097. 23. Fernandes A, Preza GC, Phung Y, et al. The molecular basis of hepcidin-resistant hereditary hemochromatosis. Blood. 2009; 114(2): 437-443. 24. Callebaut I, Labesse G, Durand P, et al. Deciphering protein sequence information through hydrophobic cluster analysis (HCA): current status and perspectives. Cell Mol Life Sci. 1997;53(8):621-645. 25. Cunat S, Giansily-Blaizot M, Bismuth M, et al. Global sequencing approach for characterizing the molecular background of hereditary iron disorders. Clin Chem. 2007; 53(12):20602069. 26. Speletas M, Kioumi A, Loules G, et al. Analysis of SLC40A1 gene at the mRNA level reveals rapidly the causative mutations in patients with hereditary hemochromatosis type IV. Blood Cells Mol Dis. 2008;40(3):353359. 27. Speletas M, Onoufriadis E, Kioumi A, Germenis AE. SLC40A1-R178G mutation and ferroportin disease. J Hepatol.

2011;55(3):730–731; author reply 731–732. 28. Galicia-Poblet G, Cid-París E, López-Andrés N, et al. Pediatric ferroportin disease. J Pediatr Gastroenterol Nutr. 2016; 63(6):e205-e207. 29. Le Lan C, Mosser A, Ropert M, et al. Sex and acquired cofactors determine phenotypes of ferroportin disease. 2011;140(4): 1199-1207.e2. 30. Papanikolaou G, Tzilianos M, Christakis JI, et al. Hepcidin in iron overload disorders. Blood. 2005;105(10):4103-4105. 31. Zoller H, McFarlane I, Theurl I, et al. Primary iron overload with inappropriate hepcidin expression in V162del ferroportin disease. Hepatology. 2005;42(2):466-472. 32. Girelli D, De Domenico I, Bozzini C, et al. Clinical, pathological, and molecular correlates in ferroportin disease: a study of two novel mutations. J Hepatol. 2008;49(4):664671. 33. De Domenico I, McVey Ward D, Nemeth E, et al. Molecular and clinical correlates in iron overload associated with mutations in ferroportin. Haematologica. 2006; 91(8):10921095. 34. Law CJ, Maloney PC, Wang D-N. Ins and outs of major facilitator superfamily antiporters. Annu Rev Microbiol. 2008; 62:289-305. 35. Quistgaard EM, Löw C, Guettou F, Nordlund P. Understanding transport by the major facilitator superfamily (MFS): structures pave the way. Nat Rev Mol Cell Biol. 2016;17(2):123-132. 36. Yan N. Structural biology of the major facilitator superfamily transporters. Annu Rev Biophys. 2015;44:257-283. 37. Bonaccorsi di Patti MC, Polticelli F, Cece G, et al. A structural model of human ferroportin and of its iron binding site. FEBS J 2014;281(12):2851-2860. 38. Pettersen EF, Goddard TD, Huang CC, et al. UCSF Chimera--a visualization system for exploratory research and analysis. J Comput Chem. 2004;25(13):1605-1612.

1805

ARTICLE

Bone Marrow Failure

Ferrata Storti Foundation

Novel lineage depletion preserves autologous blood stem cells for gene therapy of Fanconi anemia complementation group A Jennifer E. Adair,1,2* Devikha Chandrasekaran,1 Gabriella Sghia-Hughes,1 Kevin G. Haworth,1 Ann E. Woolfrey,1,2 Lauri M. Burroughs,1,2 Grace Y. Choi,1 Pamela S. Becker1,2 and Hans-Peter Kiem1,2*

Haematologica 2018 Volume 103(11):1806-1814

*JEA and H-PK contributed equally to this work.

Fred Hutchinson Cancer Research Center and 2University of Washington School of Medicine, Seattle, WA, USA

ABSTRACT

Correspondence: hkiem@fredhutch.org or jadair@fredhutch.org

Received: April 6, 2018. Accepted: July 4, 2018. Pre-published: July 5, 2018.

hallmark of Fanconi anemia is accelerated decline in hematopoietic stem and progenitor cells (CD34+) leading to bone marrow failure. Long-term treatment requires hematopoietic cell transplantation from an unaffected donor but is associated with potentially severe side-effects. Gene therapy to correct the genetic defect in the patient’s own CD34+ cells has been limited by low CD34+ cell numbers and viability. Here we demonstrate an altered ratio of CD34Hi to CD34Lo cells in Fanconi patients relative to healthy donors, with exclusive in vitro repopulating ability in only CD34Hi cells, underscoring a need for novel strategies to preserve limited CD34+ cells. To address this need, we developed a clinical protocol to deplete lineage+ (CD3+, CD14+, CD16+ and CD19+) cells from blood and marrow products. This process depletes >90% of lineage+ cells while retaining ≥60% of the initial CD34+ cell fraction, reduces total nucleated cells by 1-2 logs, and maintains transduction efficiency and cell viability following gene transfer. Importantly, transduced lineage– cell products engrafted equivalently to that of purified CD34+ cells from the same donor when xenotransplanted at matched CD34+ cell doses. This novel selection strategy has been approved by the regulatory agencies in a gene therapy study for Fanconi anemia patients (NCI Clinical Trial Reporting Program Registry ID NCI2011-00202; clinicaltrials.gov identifier: 01331018).

Introduction doi:10.3324/haematol.2018.194571 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1806 ©2018 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.

1806

Fanconi anemia (FA) is a rare monogenic disease with a wide array and variable presence of clinical symptoms, the hallmark of which is bone marrow (BM) failure.1 The genetic basis of FA is a mutation in any one of 21 genes2 whose protein components make up the FA/breast cancer pathway responsible for DNA repair of interstrand crosslinks through nucleotide excision followed by homologous recombination. Resulting compromises in genetic integrity are associated with a nearly uniform decline in hematopoietic stem and progenitor cells (HSPCs), a 50% incidence of myelodysplastic syndrome or acute myeloid leukemia by adolescence, and a 25% lifetime incidence of head and neck squamous cell carcinoma or gynecological cancer.3 In some patients, blood cell clones demonstrate spontaneous reversion to wild type (i.e. somatic mosaicism) leading to improved and stable blood cell counts for up to 27 years.4-6 Thus, correction of the FA hematopoietic defect could significantly alter the disease’s clinical course, and this has driven decades of research in HSPC gene therapy for FA. While FA was recognized as an early candidate disorder for gene therapy, several obstacles have been identified that have delayed clinical success.3 Initial clinical trials demonstrated a dramatic approximately 50-fold reduction in the number of true HSPCs in FA patients relative to other gene therapy patients, such as those treated for primary immune deficiencies.7 Moreover, FA HSPCs were exceptionally fragile when manipulated ex vivo for gene transfer. No treated patient has demonstrated haematologica | 2018; 103(11)

Obtaining blood stem cells for gene transfer in FA Table 1. Clinical characteristics of 3 patients with Fanconi Anemia A genetic defect enrolled in clinical trial NCT01331018.

Patient

Age (years)

Weight (kg)

Sex

Baseline average ANC/platelet count (thousand/mL)

Marrow cellularity

FANCA defect

70.7

Male

1.3/65

10-30%

2 3

10 5

19.6 14.7

Male Male

1.0/62 1.7/32

~20% ~30%

Exon22 splice variant (c. 1827-1 G>A) Exons6-31 Not determined

One adult and 2 pediatric patients were treated with lentivirus gene therapy for Fanconi Anemia-A (FA-A) defect. Three patients demonstrated steadily declining: absolute neutrophil count (ANC) and platelet counts in the peripheral blood prior to treatment and less than 30% marrow cellularity. Molecular characterization of the FANCA gene defect performed by gene sequencing demonstrated that Patient 1 was homozygous in the FANCA gene for the splicing variant. For Patient 2, Multiplex Ligation-dependent Probe Amplification (MLBA) on the FANCA gene identified a homozygous gross deletion of exons 6-31. No sequence analysis was performed for Patient 3, but complementation testing confirmed FANCA defect.

stable improvements in blood cell counts with long-term persistence of gene-corrected blood cells. These studies highlighted two needs for innovation in FA gene therapy: 1) to increase the number of available HSPCs for gene transfer and infusion; and 2) to increase the engraftment potential of these cells after gene transfer and infusion. Following the recommendations of the International FA Gene Therapy Working Group,8 we launched a phase I clinical trial of gene therapy for FA complementation group A (FA-A) patients in 2011 (clinicaltrials.gov identifier: 01331018). This trial design incorporates several features aiming to improve HSPC numbers and fitness. These include: i) a self-inactivating (SIN) lentiviral vector (LV) for transfer of the FANCA cDNA regulated by a human phosphoglycerate kinase (hPGK) promoter; ii) a short, overnight transduction to minimize ex vivo manipulation, as well as addition of the antioxidant N-acetylcysteine (NAC) throughout manipulation; and iii) culture under reduced oxygen (5%) to limit oxidative DNA damage.9 The target HSPC population for gene transfer expresses the CD34 cell surface protein (CD34+). When stained with fluorophore-conjugated antibody against CD34 and analyzed by flow cytometry, a small proportion of BM cells are CD34+, representing both primitive stem cells and more committed progenitors.10 The standard clinical procedure for isolating these cells first involves either BM collection or mobilization of the cells into circulation through cytokine stimulation with granulocyte colony stimulating factor (G-CSF) or, in certain clinical scenarios, a combination of G-CSF and the chemokine receptor CXCR4 antagonist plerixafor, followed by peripheral blood leukapheresis (mAPH). Initial isolation technologies relied on CD34 antigen expression on the cell surface and utilized biotinavidin affinity, panning, or immununomagnetic beadbased approaches. Expected yields were 50% of available CD34+ cells with highly variable purities, ranging from 2090% across techniques.11 Of these, immunomagnetic bead-based positive selection is the most widely-applied today, with the first US Food and Drug Administration (FDA) approval of a clinical device for human use in 2014. Advances in this technology to include automation have improved reliability in recovery to a mean yield of 70% with purities regularly over 90%.12,13 However, these values are based on BM and mAPH products wherein 1-3% of total cells express CD34 antigen, and the majority of these cells display high levels of CD34. For FA patients, the frequency of CD34+ cells is much lower: 0.1-1.5% in BM.14,15 This implies that non-standard processes may be haematologica | 2018; 103(11)

required to preserve the limited numbers of HSPCs for gene transfer in FA. Here we report HSPC collection results for the first 3 patients treated on our study. Initially, this protocol proposed direct isolation of CD34+ cells from BM without prior attempts at mobilization. The addition of a mobilization regimen with subsequent leukapheresis collections has permitted the evaluation of CD34 expression patterns in both product types and provided evidence for the need for alternative HSPC isolation strategies.

Methods Patient selection This study was approved by an Institutional Review Board at Fred Hutchinson Cancer Research Center (Fred Hutch) in accordance with the Declaration of Helsinki and the FDA, and conformed to the National Institutes of Health Guidelines for Research Involving Recombinant DNA Molecules. Informed consent was obtained from all patients or guardians. FA patients aged 4 years or over were diagnosed by a positive test for increased sensitivity to chromosomal breakage with mitomycin C (MMC) or diepoxybutane. Correction of melphalan hypersensitivity following retroviral transduction of the FANCA cDNA identified Patient 3 as belonging to the FA-A complementation group. (Online Supplementary Table S1). FA-A patients who demonstrated normal karyotype in BM analyses as defined in the trial were considered eligible for the study. Characteristics of enrolled patients are available in Table 1.

Lentiviral vectors All SIN lentiviral (LV) vectors were produced with a third-generation split packaging system and pseudotyped with vesicular stomatitis virus glycoprotein. LV used to transduce healthy donor cells encoded either an enhanced green fluorescent protein (eGFP) transgene (pRSC-PGK.eGFP-sW) or the full-length FANCA cDNA (pRSC-PGK.FANCA-sW), both regulated by an hPGK promoter. Research-grade vectors were produced by the Fred Hutch Vector Production Core (Principal Investigator: HPK). Clinical-grade LV (pRSC-PGK. FANCA-sW), was produced by the Indiana University Vector Production Facility (IUVPF, IN, USA) using a large-scale, validated process following Good Manufacturing Practices standards under an approved Drug Master File held by IUVPF. Infectious titer was determined by serial transduction of HT1080 human fibrosarcoma-derived cells and evaluated either by flow cytometry for eGFP expression or by quantitative polymerase chain reaction (qPCR). 1807

J.E. Adair et al. Table 2. Isolation and lentiviral vector transduction of autologous Fanconi Anemia A genetic defect HSPC.

Patient

Product collected

Product Total CD34+ CD34+ cell CD34+ cells volume cells purification transduced processed collected method

1278 mL

1.69E+08

2 3

BM mAPH

534 mL 513 mL

1.59E+07 9.5E+07

Direct enrichment None Lineage depletion

CD34+ cell dose/kg

CD34Hi cell dose/kg

CD34Lo cell dose/kg

5.34E+06

3.57E+04

1.77E+04

1.65E+04

8.15E+06 5.28E+07

3.57E+05 2.44E+06

4.62E+04 3.69E+05

3.15E+05 2.06E+06

CD34Hi: high CD34 expression; CD34Lo: low CD34 expression; BM: bone marrow product; mAPH: mobilized apheresis product.

Table 3. Transduction efficiency.

Patient

MOI (IU/cell)

Viability of infusion product (%)

VCN of infusion product

Plating efficiency in CFCs, 0 nM MMC (%)

Plating efficiency in CFCs, 10 nM MMC (%)

% Gene transfer in CFCs, 0 nM MMC (%)

% Gene transfer in CFCs, 10 nM MMC (%

1 2 3

10 IU/ cell 10 IU/ cell 5 IU/ cell

79 99.5 99.3

0.33 1.83 0.67

4.14 3.8 2.75

0.22 0.03 0.04

17.7 42.73 26.23

80 100 100

Viability of the infusion product was determined by trypan blue dye exclusion. The vector copy number (VCN) in the bulk transduced population was determined by quantitative polymerase chain reaction (qPCR) method against a reference standard curve. Plating efficiency of the infusion product was determined as the percentage of CD34+ cells plated with colony-forming capacity. Functional correction of the FANCA gene defect was determined by calculating the plating efficiency under stress of various concentrations of MMC. Gene transfer in colony-forming cells (CFCs) was determined as the percentage of colonies analyzed positive for the presence of lentivirus backbone by PCR analysis on DNA extracted from individual colonies. MOI: multiplicity of infection.

Study design and HSPC isolation

Transplantation in NSG mice

Patients underwent either BM harvest with a target collection goal of 15 cc/kg body weight or were administered daily G-CSF (filgrastim; 16 mg/kg BID; days 1-6) and plerixafor (240 mg/kg/day; days 4-6) subcutaneously to mobilize CD34+ cells. Mobilized patients were subjected to large volume leukapheresis when circulating CD34+ blood cell counts were ≥5 cells/mL. Healthy donor blood products were purchased from a commercial source (BM products; StemExpress, Folsom, CA, USA) or institutional shared resources (mAPH products). Immunomagnetic beads were from Miltenyi Biotech, GmbH (Auburn, CA, USA). For BM products, RBC were debulked by hetastarch sedimentation prior to labeling on a CliniMACS Prodigy™ device (Miltenyi Biotec GmbH, Germany). For mAPH products, an initial platelet wash was performed prior to labeling. Custom programming for lineage depletion was designed and executed on the CliniMACS Prodigy™ device (Miltenyi Biotec, GmbH). Complete processing methods are included in the Online Supplementary Materials and Methods.

All animal work was performed under protocol 1864 approved by the Fred Hutch Institutional Animal Care and Use Committee. NOD.Cg-PrkdcscidIL2rγtmlWj/Szj (NOD/SCID/IL2rγnull, NSG) mice were housed at Fred Hutch in pathogen-free conditions approved by the American Association for Accreditation of Laboratory Animal Care. 8-12-week old mice received 275 cGy total body irradiation (TBI) from a Cesium source. Four hours after TBI, 1x106 gene-modified total nucleated cells (TNCs) re-suspended in 200 mL phosphate buffered saline (D-PBS, Life Technologies Corporation, Grand Island, NY, USA) containing 1% heparin (APP) were infused via tail vein. Blood samples were collected into ethylenediaminetetraacetic acid (EDTA) Microtainers (BD Bioscience, San Jose, CA, USA) by retro-orbital puncture and diluted 1:1 with PBS prior to analysis. At necropsy, spleen and BM were collected. Tissues were filtered through 70 mm mesh (BD Bioscience) and washed with Dulbecco’s PBS (D-PBS).

Transduction

Colony-forming cell assays

CD34-enriched cells were cultured on RetroNectin (Takara Bio, Mountain View, CA, USA)-coated culture flasks at a density of 1x106 cells/mL and 2.9x105 cells/cm2 in StemSpanTM ACF media (StemCell Technologies, Vancouver, BC, Canada), supplemented with 4 mg/mL of protamine sulfate (American Pharmaceutical Partners; APP, East Shaumburg, IL, USA), 100 ng/mL each of recombinant human stem cell factor (rhSCF), thrombopoietin (rhTPO) and Flt-3 ligand (rhFLT3L) (all from CellGenix GmbH, Freiburg, Germany), and 1 mM NAC (Cumberland Pharmaceuticals, Nashville, TN, USA). Cells were immediately transduced at a multiplicity of infection (MOI) of 510 infectious units (IU)/cell. Following 12-24 hours of incubation at 37°C, 5% CO2 and 5% O2 , cells were harvested for infusion and/or analyses.

Transduced cell products were seeded in standard CFC assays in methylcellulose media (H4230, Stem Cell Technologies) as previously described16 with the following exceptions: to assess FANCA gene function, MMC (Sigma Aldrich, St. Louis, MO, USA) was added at concentrations of 0 nM, 5 nM, 10 nM, or 20 nM. Complete colony DNA extraction and PCR methods are included in the Online Supplementary Materials and Methods.

1808

Quantitative real-time PCR-based measurement of vector copy number Vector copy number (VCN) per genome equivalent was assessed by TaqMan 5' nuclease quantitative real-time PCR assay in duplicate reactions with an LV-specific primer/probe combination [forward, 5'-TGAAAGCGAAAGGGAAACCA; haematologica | 2018; 103(11)

Obtaining blood stem cells for gene transfer in FA

Figure 1. Diminished CD34Hi hematopoietic cells from Fanconi Anemia A genetic defect (FA-A) patients. CD34 expression in baseline bone marrow (BM) (Patients 1, 2, 3, and healthy donor 1) or mobilized leukapheresis (mAPH) (Patient 3 and healthy donor 2) products was determined by fluorescence staining and flow cytometry analysis. Positive cell fractions are gated based on unstained and isotype stained control samples into two levels of CD34 expression: low expression, CD34Lo, or high expression, CD34Hi. The average mean fluorescence intensity (MFI) of CD34Lo population = 3453; standard error of the mean (SEM) = 516 and CD34Hi population = 19731; SEM = 4103.

reverse, 5'-CCGTGCGCGCTTCAG; probe, 5'-AGCTCTCTCGACGCAGGACTCGGC (Integrated DNA Technologies; IDT, Coralville, IA, USA)] and in a separate reaction with a β-globinspecific primer/probe combination [forward, 5’-CCTATCAGAAAGTGGTGGCTGG; reverse, 5'-TTGGACAGCAAGAAAGTGAGCTT; probe, 5'-TGGCTAATGCCCTGGCCCACAAGTA (IDT)]. Two standard curves were established by serial dilution of gDNA isolated from a human cell line (HT1080) confirmed to contain a single integrant of the same LV backbone and from peripheral leukocytes collected from a healthy donor using both primer-probe sets independently. Individual colony gDNA samples were subjected to multiplex real-time TaqMan qPCR to amplify the LV-specific product and an endogenous control (TaqMan Copy Number Reference assay RNaseP, Thermo Fisher Scientific, Pittsburgh, PA, USA). Samples with an average VCN ≥0.5 were considered transduced.

Flow cytometry analysis of hematopoietic subsets Stained cells were acquired on a FACSCantoTM II, FACSAriaTM II or FACS LSR II (all from BD Bioscience) and analyzed using FlowJo software v.10.0.8 (Tree Star Inc., Ashland, OR, USA). Analysis was performed on up to 20,000 cells. Gates were established using Full Minus One stained controls. Antibodies included anti-human CD34 (clone 563), CD16 (clone 3G8), CD3 (clone UCHT1), CD4 (clone L200), CD8 (clone RPA-T8), all from BD Biosciences; CD14 (clone 61D3, Thermo Fisher Scientific, Pittsburgh, PA, USA); CD19 (clone 4G7, BD Pharmingen, San Diego, CA, USA); CD90 (clone 5E10), CD20 (clone 2H7), CD15 (clone W6D3), all from Biolegend (San Diego, CA, USA); CD133 (clone 293C3, Miltenyi Biotec, GmbH); CD45 (clone D058-1283) and CD45RA (clone 5H9), both from BD Horizon (San Jose, CA, USA). For mouse samples, antibodies were anti-mouse CD45-V500 (561487, clone 30-F11), anti-human CD45-PerCP (347464, clone 2D1), CD3-FITC (555332, clone UCHT1), CD4-V450 (560345, clone RPA-T4), CD8-APCCy7 (557834, clone SK1), CD20-PE (555623, clone 2H7), and CD14-APC (555824, clone 581), all from BD Biosciences. haematologica | 2018; 103(11)

Results Diminished CD34Hi expressing cells in FA-A BM and mAPH Two enrolled patients underwent BM harvest to collect available CD34+ HSPCs (Patients 1 and 2). The third patient underwent mobilization with filgrastim and plerixafor followed by peripheral blood leukapheresis (Patient 3). All 3 patients demonstrated reduced CD34 expression and estimated numbers of CD34+ cells in screening BM aspirate samples prior to collection and treatment, relative to healthy donor BM products, as well as in cell products collected for CD34+ cell isolation and gene transfer (Figure 1). Two levels of CD34 expression were observed, CD34Lo [mean fluorescence intensity (MFI)=3453±516], and CD34Hi (MFI=19731±4103). Notably, the proportion of CD34Hi cells were markedly reduced in FA-A patients relative to those observed in healthy donors (Figure 1).

FA-A CD34Hi cells, but not CD34Lo cells, demonstrate in vitro repopulating capacity To determine which CD34+ cells demonstrated repopulation potential, we used colony-forming cell (CFC) potential as a surrogate. This required sufficient blood product to flow-sort CD34Lo and CD34Hi cells for in vitro assays. Only the mAPH product collected from Patient 3 was sufficient for this study. For direct comparison, we sort-purified CD34Lo and CD34Hi cells from a healthy donor mAPH product. Only CD34Hi cells from the FA-A patient demonstrated colony-forming potential (Figure 2A). In the healthy donor, CD34Hi cells also demonstrated the majority of CFC capacity in comparison with CD34Lo cells, and at much higher levels as compared to the FA-A patient (Figure 2B). These data suggest repopulating capacity is restricted to CD34Hi cell fractions, underscoring the need to preserve as many of these cells as possible for gene transfer processes. 1809

J.E. Adair et al.

Extensive loss of FA-A CD34Hi cells with direct clinical purification protocols The current clinical standard for CD34+ cell enrichment is optimized for collection of CD34Hi cells. However, in Patient 1, direct enrichment of CD34+ cells using this protocol was inefficient, resulting in an approximately 3% yield and only 5.34x106 total CD34+ cells available for gene transfer (Table 2). Moreover, the purity of the enriched cell product was only 58.9%, and approximately 47% loss in viable cells was observed during culture and gene transfer. Resulting gene-modified cells retained colony-forming capacity and demonstrated acquired resistance to the potent DNA crosslinking agent MMC following LV-mediated FANCA gene transfer (Table 3). In Patient 2, estimated losses during direct CD34 enrichment and gene transfer were expected to reduce the cell product available for transduction to a level lower than observed for Patient 1. Thus, an urgent amendment was filed with the FDA to permit elimination of the direct CD34 enrichment steps and allow transduction of the entire red blood cell (RBC)-depleted BM product. This processing change preserved more CD34+ cells (Table 2), with improved transduction and viability (Table 3). Together, these data suggested that minimal manipulation of target CD34+ cells from FA-A patients could improve yield, gene transfer efficiency, and function in vivo.

Development of a novel strategy to deplete lineage+ cells We hypothesized that depleting non-target mature B cells, T cells, monocytes, and granulocytes would retain precious CD34+ cells with minimal manipulation, since CD34-expressing cells would not be directly labeled, selected, or washed (Figure 3). Building on our previous work automating cell selection and gene transfer using the CliniMACS Prodigy™ device,17 we designed a customized, automated RBC debulking and immunomagnetic bead-based lineage specific depletion strategy (Online Supplementary Materials and Methods). Four different beadconjugated antibody reagents were used in this approach: anti-CD3 (T-cell removal), anti-CD14 (monocyte removal), anti-CD16 (granulocyte and NK-cell removal), and anti-CD19 (B-cell removal). This protocol was designed for both BM and mAPH products.

Supplementary Figure S2). Approximately 24% of BM CD34+ cells were colony-forming in a standard methylcellulose assay, while 51% of mAPH CD34+ cells formed colonies (Figure 4C and Online Supplementary Figure S3). However, following LV transduction of these cells using the same protocol proposed for FA-A patient cells, we observed consistent 50% rates of gene transfer into CFCs from both cell product types (Figure 4D). Analysis of single colonies demonstrated an average VCN per CFC of 0.7 for BM CD34+ cells and 1.6 for mAPH CD34+ cells. VCN was also assessed in bulk transduced cells cultured for ten days in vitro, demonstrating an average value of 5 for both BM and mAPH products (Figure 4E). Final cell products tested for mycoplasma and sterility were negative, and endotoxin testing demonstrated values within criteria for patient infusion. Lineage-depleted and transduced cells from six mAPH and BM products each were infused into immunodeficient (NSG) mice at a target cell dose of 1x106 TNC per mouse. On average, the CD34+ cell dose per mouse for BM products was 2.86x104 CD34+ cells [standard error of the mean (SEM)=6.67x103] and for mAPH products was 1.08x105 CD34+ cells (SEM=1.45×104). Flow cytometry analysis on peripheral blood was used to evaluate engraftment (human CD45+) and lineage development into T cells (human CD3+), B cells (human CD20+),

Lineage depletion preserves available CD34 cells for gene transfer +

A total of nine BM and ten mAPH products were processed to establish process validity. An average 60% of BM CD45+ cells and 50% of mAPH CD45+ cells expressed one of the four target markers (CD3, CD14, CD16, or CD19) (Online Supplementary Figure S1A and B, respectively). CD34+ cell content in these products ranged from 0.351.4% in BM and 0.06-0.9% in mAPH products. The average process run time for BM products was ten hours, whereas mAPH products were processed over 13 hours. Observed total nucleated cell (TNC) reduction was approximately 1 log for both BM and mAPH products following lineage depletion (Figure 4A). All target lineage+ cells were depleted to less than 10% of initial numbers, and CD34+ cells were retained at 94.62±4.61% for BM products and 70.69±11.4% for mAPH products (Figure 4B). Retention of available CD34Hi and CD34Lo cells was observed and comparable or superior to that observed for the same products by direct CD34-enrichment (Online 1810

Figure 2. In vitro repopulation potential restricted to CD34Hi hematopoietic cells. Mobilized leukapheresis from FA-A Patient 3 (Panel A) and a healthy donor (Panel B) were in parallel fluorescence stained with anti-CD34 antibody and sort-purified for CD34Hi and CD34Lo cells. Total nucleated cells (TNC) equivalent to 1500 CD34-expressing cells were seeded in CFC assays. Percentage of CD34+ cells seeded in the assay that gave rise to colonies is represented as the % of colony-forming cells.

haematologica | 2018; 103(11)

Obtaining blood stem cells for gene transfer in FA

and monocytes (human CD14+) over time (Figure 4F). Both mAPH and BM products demonstrated long-term engraftment over 20 weeks of monitoring. Engraftment levels were comparable to results reported by Wiekmeijer et al. with CD34+ cells purified from BM and infused at similar cell doses.18

Lineage-depleted cell products xenoengraft equivalently to CD34-enriched products In this experiment, healthy donor BM products were divided into two aliquots. One was lineage-depleted and the other CD34-enriched. Resulting cell populations were transduced with the same LV vector under identical conditions and infused into NSG mice at matched CD34+ cell doses. We observed higher CD34+ cell retention with lineage depletion compared to CD34 selection, with no differences in transduction efficiency or colony-forming potential (Figure 5A and B). We observed slightly higher, but not significantly different, levels of human CD45+ blood cell engraftment in mice receiving transduced, lineage-depleted cells relative to mice receiving CD34-selected cells. We also observed more stability of T- and B-cell engraftment in mice receiving lineage-depleted cell products relative to mice receiving CD34-selected cell products (Figure 5C).

Lineage depletion protocol preserves limited FA CD34Hi cells These data collectively suggest that lineage-specific depletion preserved available CD34+ cells without compromising transduction efficiency or cell fitness. Under

FDA approval, the clinical protocol was modified to include both BM and/or mAPH products, with lineage depletion as the method of CD34+ cell enrichment. Patient 3 (the first treated under the modified protocol) was a 5year old male with FA-A confirmed by complementation studies. Baseline neutrophils averaged 1.7x109/L and baseline platelets averaged 32x109/L in the six months prior to treatment, with declining neutrophils and platelets over the prior 2-year interval (Online Supplementary Figure S4). Mobilization of â&#x2030;Ľ10 CD34+ cells/mL peripheral blood was achieved (Online Supplementary Figure S5A), and two successive apheresis collections resulted in 8.5x1010 TNC containing a total 1.6x108 CD34+ cells (Table 2). The patient required a total of two platelet transfusions and two packed red blood cell transfusions during mobilization and leukapheresis (Online Supplementary Figure S5B). Due to column limitations, 5x1010 TNC (equivalent to 9.5x107 total CD34+ cells) were subjected to lineage depletion, and the remainder were cryopreserved. Lineage depletion resulted in a 94% reduction in TNC and a 56% retention of available CD34+ cells. CD34 purity was 1.6%, representing a 1-2 log-fold increase in the total number of CD34+ cells per kg available for transduction and infusion relative to Patients 1 and 2 (Table 2). A total of 52.8x106 CD34+ cells were transduced at an MOI of 5 IU/cell, resulting in a final cell dose of 2.4x106 total CD34+ cells per kg with 99.3% viability based on trypan blue dye exclusion. Approximately 26% of CFCs in this cell product were transduced, displaying a mean VCN of approximately 1 (0.9) (Table 3). Thus, limited numbers of available CD34+ cells were indirectly enriched using lineage deple-

Figure 3. Direct CD34 enrichment versus depletion of lineage positive (+) cells. Products can include bone marrow (BM) or mobilized apheresis product (mAPH) (1). BM products were first processed through hetastarch sedimentation to deplete red blood cells (RBCs). Leukapheresis products were first subjected to several washes to deplete platelets. For direct CD34+ cell selection, anti-CD34 antibody-bound immunomagnetic beads (microbeads) are used, whereas for lineage depletion antiCD3+, CD14+, CD16+, and CD19+, microbeads are used (2). In both cases, microbead-bound cells are retained on the column and subjected to wash steps. When lineage depletion is used, CD34-expressing cells undergo minimal manipulation during purification. Following purification, cells are cultured and transduced with a VSV-G pseudotyped lentiviral vector at a multiplicity of infection (MOI) of 5â&#x20AC;&#x201C;10 IU/ cell (3). Following ~16 hours of incubation cells are harvested (4). *These processes were performed on the CliniMACS ProdigyTM device from Miltenyi Biotec GmbH.

haematologica | 2018; 103(11)

1811

J.E. Adair et al. B

Figure 4. Multi-lineage engraftment of lineage depleted and transduced bone marrow (BM) and mobilized apheresis products (mAPH) in NSG mice. Recovery of total nucleated cells (TNC), CD34+ cells and lineage positive (+) cells (A and B). (C-E) Gene transfer efficiency. The colony-forming potential of transduced cells in standard CFC assays is defined as the plating efficiency (TNC). The colony-forming potential normalized to the number of CD34+ cells seeded is depicted as plating efficiency (CD34+). The percentage of colonies analyzed positive for the presence of lentivirus (LV) backbone by PCR analysis on DNA extracted from individual colonies is depicted as transduction efficiency. The vector copy number per cell in the bulk transduced population is depicted as VCN. The average VCN per cell in the individual CFC is depicted as single colony VCN. Data are representative of the average of 9 healthy BM products and 10 healthy mAPH products. Error bars represent the standard error of the mean. (F) Engraftment of human CD45+ cells and lineage development into T cells (CD3+ ), monocytes (CD14+ ) and B cells (CD20+) was determined by flow cytometry over 20 weeks following infusion of lineage-depleted cell products. Data are representative of 36 mice from 6 mAPH donors and 42 mice from 6 BM donors, respectively. Error bars represent the Standard Error of the Mean.

tion on a blood product from an FA-A patient without compromising transduction efficiency.

Discussion Here we confirm prior reports of inefficient CD34+ cell enrichment from FA patient blood products by direct, immunomagnetic bead-based separation, which is the current standard protocol for isolating HSPCs.15,19-21 We also demonstrate substantially reduced levels of CD34Hi cells in FA patients relative to healthy donors, which likely contributes to poor positive selection results in blood products from FA patients. Colony seeding assays demonstrate that only CD34Hi cells contribute to in vitro colonyforming potential in both FA and healthy donor blood products, underscoring the need to preserve as many available CD34+ cells as possible during ex vivo manipulation for gene transfer. We demonstrate a clinically viable procedure for depleting lineage positive cells to indirectly enrich for CD34+ cells that preserves the limited numbers of these cells in FA-A patients without compromising viability, gene transfer, or engraftment potential. Importantly, the phenotype of limiting CD34+ cell numbers is not restricted to FA alone. Sickle cell disease (SCD) patients treated with hydroxyurea also display reduced CD34+ cell frequencies in BM, and there is a contraindication to mobilization of available CD34+ cells owing to an increased risk of vaso-occlusive crisis.22 Other inherited BM failure syndromes such as dyskeratosis congenita also 1812

are associated with abnormal CD34+ cell frequencies and behavior.23 As a larger number of disease targets become relevant for gene therapy, additional patient populations will likely display variable CD34+ cell frequency and antigen expression. These disease targets could also benefit from clinically viable alternative selection procedures such as we have developed here. Our observation of CFC potential in only the CD34Hi fraction in both FA and healthy samples suggests that CD34Lo cells may not be contributing to hematopoietic reconstitution. Notably, our data are from mAPH samples not BM, and we will need more patients for confirmation. Additionally, the standard colony-forming assay best defines progenitor cells, more so than true long-term repopulating hematopoietic stem cells.24 Alternatively, xenotransplant of purified cells into immunodeficient mice could provide the most robust evidence for CD34+ cell function in vivo, but the very small numbers of these cells may prove problematic to achieving relevant cell doses needed for these experiments. Another in vitro assay, such as the long-term culture-initiating cell assay,25 may provide additional insight into the desired target CD34+ subpopulations for gene therapy if they are present in either the CD34Hi or CD34Lo populations in FA patients. In this regard, we recently demonstrated that the CD34HiCD45RAâ&#x20AC;&#x201C;CD90+ phenotype is responsible for hematopoietic repopulation in non-human primates in the autologous, myeloablative setting,16 and evaluation of this phenotype in the enrolled FA patients is ongoing. Critically, our strategy of depleting cells expressing mature haematologica | 2018; 103(11)

Obtaining blood stem cells for gene transfer in FA

B LinCD34+

Figure 5. Multi-lineage engraftment levels of lineage-depleted cell products in NSG (NOD/SCID/IL2rgnull) mice is comparable to CD34-enriched cell products from the same donor. (A) Graph depicts percent recovery of total nucleated cells (TNC) and CD34+ cells from each arm following depletion or enrichment. (B) Numbers of total and transduced colony-forming cells (CFC) normalized to 1x108 cells processed to each arm, and vector copy number (VCN) in the bulk transduced cells following ten days of culture. Data are representative of 2 healthy donor bone marrow products. Error bars represent the Standard Error of the Mean. (C) Engraftment of human CD45+ cells and lineage development into T cells (CD3+), monocytes (CD14+) and B cells (CD20+) was determined by flow cytometry over 26 weeks following infusion. Data are representative of 9 mice for the lineage depleted (Lin-) arm and 6 mice for the CD34 enriched (CD34) arm, respectively.

blood cell lineage markers preserves all CD34+ cell phenotypes for gene transfer and infusion, as demonstrated by Patient 3, whose infused CD34+ cell dose was the largest received to date. One characteristic of lineage-depleted cell products requiring additional study is the presence and impact of other supporting cells on engraftment. Especially for BMderived products, our procedure does not include a marker to deplete mesenchymal stem cells (MSC). While the engraftment potential of MSC manipulated ex vivo in CD34+ cell supportive media is unexplored, two recent reports suggest that these cells are integral to BM function in FA, and can be LV-transduced and functionally corrected to facilitate hematopoietic recovery and function in a mouse model of FA.26,27 For mAPH-derived products, such as that infused into Patient 3, additional follow up will be required to determine if a selective advantage is observed haematologica | 2018; 103(11)

in vivo. The improved transduction efficiency of lineagedepleted cell products could reflect non-repopulating CD34â&#x20AC;&#x201C; cell uptake of LV. However, we still observed a benefit in transduction of hematopoietic CFC, even at the lower MOI of 5 IU/cell. One other possible explanation is the age and clinical condition of Patient 3. To address this concern we compared our results in Patient 3 to the 4 FA patients enrolled in the FANCOSTEM clinical trial in Spain (clinicaltrials.gov identifier: 02931071).28 These 4 patients were aged 3-7 years and demonstrated higher baseline blood cell counts at the time of collection. All 4 patients received the same mobilization regimen as Patient 3 reported here, but resulting mAPH products were subjected to direct CD34 enrichment prior to transduction at an MOI of 100 IU/cell. The reported mean VCN was 0.4Âą0.1 and ranged from 0.1 to 0.4 copies in individual CFC. Our data with a higher VCN at lower 1813

J.E. Adair et al.

MOI suggest that the mixed cell culture supports transduction of hematopoietic progenitor cells, at least. In conclusion, we describe an alternative strategy to a direct, immunomagnetic bead-based selection of CD34-expressing cells that overcomes current barriers in isolation of blood stem and progenitor cells especially for diseases like FA. Our novel approach to preserve available CD34+ cells during initial blood product processing has the potential to improve gene therapy and gene editing in settings of limited CD34+ cell availability, including FA and other diseases in which direct CD34 enrichment has proven inefficient, such as SCD. Acknowledgments We thank our patients and their families. We thank the Fanconi Anemia Research Fund for sponsoring patient travel and housing arrangements associated with this study. We especially thank the Fred Hutch Cell Processing Facility and Seattle Cancer Care Alliance for their support in enrollment and treatment of clinical trial patients. We thank H. Crawford and K. Gonzalez for help in preparing the manuscript. We

References 1. Nalepa G, Clapp DW. Fanconi anaemia and cancer: an intricate relationship. Nat Rev Cancer. 2018;18(3):168-185. 2. Mamrak NE, Shimamura A, Howlett NG. Recent discoveries in the molecular pathogenesis of the inherited bone marrow failure syndrome Fanconi anemia. Blood Rev. 2017;31(3):93-99. 3. Adair JE, Sevilla J, Heredia CD, Becker PS, Kiem HP, Bueren J. Lessons learned from two decades of clinical trial experience in gene therapy for Fanconi anemia. Curr Gene Ther. 2017;16(5):338-348. 4. Asur RS, Kimble DC, Lach FP, et al. Somatic mosaicism of an intragenic FANCB duplication in both fibroblast and peripheral blood cells observed in a Fanconi anemia patient leads to milder phenotype. Mol Genet Genomic Med. 2018;6(1):77-91. 5. Gregory JJ, Jr., Wagner JE, Verlander PC, et al. Somatic mosaicism in Fanconi anemia: evidence of genotypic reversion in lymphohematopoietic stem cells. Proc Natl Acad Sci USA. 2001;98(5):2532-2537. 6. Mankad A, Taniguchi T, Cox B, et al. Natural gene therapy in monozygotic twins with Fanconi anemia. Blood. 2006; 107(8):30843090. 7. Verhoeyen E, Roman-Rodriguez FJ, Cosset FL, Levy C, Rio P. Gene therapy in Fanconi anemia: A matter of time, safety and gene transfer tool efficiency. Curr Gene Ther. 2017;16(5):297-308. 8. Tolar J, Adair JE, Antoniou M, et al. Stem cell gene therapy for Fanconi anemia: report from the 1st International Fanconi Anemia Gene Therapy Working Group meeting. Mol Ther. 2011;19(7):1193-1198. 9. Becker PS, Taylor JA, Trobridge GD, et al. Preclinical correction of human Fanconi ane-

1814

10.

11. 12.

13.

14.

15. 16.

17.

18.

19.

also thank J. Chen and C. Ironside for excellent support in our mouse studies. This work was primarily funded by a Sponsored Research Agreement between Fred Hutch and Rocket Pharmaceuticals. Funding This work was also supported in part by grants and contracts to H-PK from the NIH HHSN2680004 and HL122173 and by funds to JEA from the Fred Hutch. This research was also funded in part through the NIH/NCI Cancer Center Support Grant P30 CA015704. H-PK is a Markey Molecular Medicine Investigator and received support as the inaugural recipient of the José Carreras/E. Donnall Thomas Endowed Chair for Cancer Research and the Fred Hutch Endowed Chair for Cell and Gene Therapy. JEA and H-PK are co-inventors on U.S. Provisional Patent Applications #62/491,116 and #62/503,801 “Novel Manufacturing of Gene Corrected Autologous Blood Cells for Gene Therapy.” JEA has previously served as a consultant for Rocket Pharmaceuticals. H-PK is an active consultant for Rocket Pharmaceuticals. The other authors declare that they have no competing interests.

mia complementation group A bone marrow cells using a safety-modified lentiviral vector. Gene Ther. 2010;17(10):1244-1252. Syrjala M, Ruutu T, Jansson SE. A flow cytometric assay of CD34-positive cell populations in the bone marrow. Br J Haematol. 1994;88(4):679-684. Collins RH Jr. CD34+ selected cells in clinical transplantation. Stem Cells. 1994; 12(6):577585. Spohn G, Wiercinska E, Karpova D, et al. Automated CD34+ cell isolation of peripheral blood stem cell apheresis product. Cytotherapy. 2015;17(10):1465-1471. Avecilla ST, Goss C, Bleau S, Tonon JA, Meagher RC. How do I perform hematopoietic progenitor cell selection? Transfusion. 2016;56(5):1008-1012. Muller LU, Williams DA. Finding the needle in the hay stack: hematopoietic stem cells in Fanconi anemia (Review). Mutat Res. 2009;668(1-2):141-149. Kelly PF, Radtke S, von Kalle C, et al. Stem cell collection and gene transfer in Fanconi anemia. Mol Ther. 2007;15(1):211-219. Radtke S, Adair JE, Giese MA, et al. A distinct hematopoietic stem cell population for rapid multilineage engraftment in nonhuman primates. Sci Transl Med. 2017; 9(414). Adair JE, Waters T, Haworth KG, et al. Semiautomated closed system manufacturing of lentivirus gene-modified haematopoietic stem cells for gene therapy. Nat Commun. 2016;7:13173. Wiekmeijer AS, Pike-Overzet K, Brugman MH, et al. Sustained engraftment of cryopreserved human bone marrow CD34(+) cells in young adult NSG mice. Biores Open Access. 2014;3(3):110-116. Liu JM, Kim S, Read EJ, et al. Engraftment of hematopoietic progenitor cells transduced with the Fanconi anemia group C gene (FANCC). Hum Gene Ther. 1999;

10(14):2337-2346. 20. Croop JM, Cooper R, Fernandez C, et al. Mobilization and collection of peripheral blood CD34+ cells from patients with Fanconi anemia. Blood. 2001;98(10):29172921. 21. Larghero J, Marolleau JP, Soulier J, et al. Hematopoietic progenitor cell harvest and functionality in Fanconi anemia patients. Blood. 2002;100(8):3051. 22. Uchida N, Fujita A, Hsieh MM, et al. Bone marrow as a hematopoietic stem cell source for gene therapy in sickle cell disease: Evidence from rhesus and SCD patients. Hum Gene Ther Clin Dev. 2017;28(3):136144. 23. Balakumaran A, Mishra PJ, Pawelczyk E, et al. Bone marrow skeletal stem/progenitor cell defects in dyskeratosis congenita and telomere biology disorders. Blood. 2015; 125(5):793-802. 24. Coulombel L. Identification of hematopoietic stem/progenitor cells: strength and drawbacks of functional assays. Oncogene. 2004;23(43):7210-7222. 25. Miller CL, Eaves CJ. Long-term culture-initiating cell assays for human and murine cells. Methods Mol Med. 2002;63:123-141. 26. Jacome A, Navarro S, Rio P, et al. Lentiviralmediated genetic correction of hematopoietic and mesenchymal progenitor cells from Fanconi anemia patients. Mol Ther. 2009; 17(6):1083-1092. 27. Zhou Y, He Y, Xing W, et al. An abnormal bone marrow microenvironment contributes to hematopoietic dysfunction in Fanconi anemia. Haematologica. 2017; 102(6):1017-1027. 28. Rio P, Navarro S, Guenechea G, et al. Engraftment and in vivo proliferation advantage of gene-corrected mobilized CD34(+) cells from Fanconi anemia patients. Blood. 2017;130(13):1535-1542.

haematologica | 2018; 103(11)

ARTICLE

Phagocyte Biology & its Disorders

Prognostic factors of Erdheim–Chester disease: a nationwide survey in Japan

Ferrata Storti Foundation

Takashi Toya,1* Mizuki Ogura,1* Kazuhiro Toyama,2 Akihide Yoshimi,1 Aya Shinozaki-Ushiku,3 Akira Honda,1 Kenjiro Honda,4 Noriko Hosoya,5 Yukako Murakami,6 Hiroyuki Kawashima,7 Yasuhito Nannya,1 Shunya Arai,1 Fumihiko Nakamura,1 Yusuke Shinoda,8 Masaomi Nangaku,4 Kiyoshi Miyagawa,5 Masashi Fukayama,3 Akiko Moriya-Saito,9 Ichiro Katayama,6 Takashi Ogura10 and Mineo Kurokawa1,2

Department of Hematology & Oncology, Graduate School of Medicine, The University of Tokyo; 2Department of Cell Therapy and Transplantation Medicine, The University of Tokyo Hospital; 3Department of Pathology, Graduate School of Medicine, The University of Tokyo; 4Division of Nephrology and Endocrinology, The University of Tokyo Graduate School of Medicine; 5Laboratory of Molecular Radiology, Center for Disease Biology and Integrative Medicine, Graduate School of Medicine, The University of Tokyo; 6 Department of Dermatology, Osaka University Graduate School of Medicine; 7Division of Orthopedic Surgery, Niigata University Graduate School of Medical and Dental Sciences; 8Department of Rehabilitation Medicine Graduate School of Medicine, The University of Tokyo; 9Clinical Research Center, National Hospital Organization Nagoya Medical Center, Aichi and 10Department of Respiratory Medicine, Kanagawa Cardiovascular and Respiratory Center, Yokohama, Japan 1

Haematologica 2018 Volume 103(11):1815-1824

*TT and MO contributed equally to this work

ABSTRACT

rdheim–Chester disease is a rare histiocytosis with insufficient clinical data. To clarify the clinical features and prognostic factors of Erdheim–Chester disease, we conducted a nationwide survey to collect the detailed data of 44 patients with Erdheim–Chester disease in Japan. The median age of onset of the participants was 51 (range: 23–76) years, and the median number of involved organs per patient was 4 (range: 1–11). The existence of central nervous system disease was correlated with older age (P=0.033), the presence of cardiovascular lesions (P=0.015), and an increased number of involved organs (P=0.0042). The median survival from the onset was 10.4 years, and >3.0 mg/dL C-reactive protein level at onset was associated with worse outcome (median survival, 14.6 vs. 7.4 years; P=0.0016). In a multivariate analysis, age >60 years (hazard ratio, 25.9; 95% confidence interval, 2.82–237; P=0.0040) and the presence of digestive organ involvement (hazard ratio, 4.74; 95% confidence interval, 1.05–21.4; P=0.043) were correlated with worse survival. Fourteen patients had available histological samples of Erdheim– Chester disease lesions. BRAFV600E mutation was detected in 11 patients (78%) by Sanger sequencing. A correlation between BRAF mutation status and clinical factors was not observed. Our study revealed that age and digestive organ involvement influence the outcome of Erdheim–Chester disease patients, and an inflammatory marker, such as C-reactive protein, might reflect the activity of this inflammatory myeloid neoplasm.

Introduction Erdheim–Chester disease (ECD) is a rare non-Langerhans histiocytosis that was first reported by Jakob Erdheim and William Chester in 1930.1 The number of reports has drastically increased recently, perhaps due to the increased recognition of the disease, and approximately 650–1000 cases have been reported.2-4 ECD typically develops among middle-aged males, and bilateral cortical osteosclerosis occurs in more than 95% of ECD patients.5 Furthermore, some patients experience involvements of the central nervous system (CNS), cardiovascular system, and various other organs.6,7 haematologica | 2018; 103(11)

Correspondence: kurokawa-tky@umin.ac.jp

Received: February 8, 2018. Accepted: July 4, 2018. Pre-published: July 5, 2018. doi:10.3324/haematol.2018.190728 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1815 ©2018 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.

1815

T. Toya et al.

The pathogenesis of ECD is still unclear, and whether this condition is a type of neoplasia or inflammation is a topic of debate. The high prevalence of mutations in BRAF, NRAS, and various genes involved in the MAPK pathway and the dramatic efficacy of BRAF inhibitors suggest the important role played by the BRAF and MAPK pathways in ECD development.8-11 Recently, Haroche et al. speculated that ECD could be redefined as “inflammatory myeloid neoplasia”, similar to Langerhans cell histiocytosis (LCH).2,12 A mouse model of LCH showed that genetic mutations in hematopoietic stem cells or immature myeloid progenitors induce misguided differentiation, and increase in pathological dendritic cells, which recruit and activate additional inflammatory cells (so-called “innocent bystanders”).13 Cavalli et al. suggested the significance of oncogene-induced senescence (OIS), a protective reaction against oncogenesis, in the recruitment of circulating normal leukocytes in ECD lesions.14,15 In the context of ECD, the local production of pro-inflammatory chemokines and Th1-associated cytokines by BRAFV600E mutated cells attracts circulating normal leukocytes to the sites and contributes to the inflammatory activation and formation of ECD. Further basic studies are warranted to investigate the exact mechanisms of ECD. The clinical course of ECD is quite heterogeneous, but most cases are progressive and become fatal within a few years. Little is known about the prognostic factors of ECD except that of the involvement of CNS.16 ECD diagnosis is mainly dependent on pathological findings, such as CD68+CD1a− foamy histiocyte infiltration, but the consensus guidelines also demand the clinical and radiological contexts for an appropriate diagnosis of the disease.5 However, our knowledge about the clinical profiles of ECD is unsatisfactory, and more information to consolidate this “context” is required. Herein, we conducted a nationwide survey on ECD in Japan and clarified some clinical features of this disease. Strikingly, the poor outcome observed among elderly patients and the prognostic impact of digestive organ involvement encouraged the spread of novel treatment strategies, such as vemurafenib, or other molecular targeted therapies.17 Additionally, the prognostic value of the increased C-reactive protein (CRP) level at onset suggested the significance of the inflammatory nature of ECD pathophysiology.

Guidelines for Biomedical Research Involving Human Subjects enforced on March 29, 2001. This study was approved by the ethics committees of the University of Tokyo and each participating institution. Written informed consent was obtained from all patients whose ECD samples were collected. No definitive diagnostic criteria for ECD have been published; therefore, the diagnoses were self-reported by each institute based on the pathological, radiological, and clinical findings. The digestive organ means gastrointestinal tract plus the accessory organs of digestion (the pancreas, liver, and gallbladder).

Methods

Results

Study design and participants

Patient characteristics and affected organs

We conducted a postal questionnaire-based, multicenter retrospective study on ECD. A questionnaire was simultaneously sent to the hematology, dermatology, respiratory medicine, orthopedics, and pathology departments of the involved hospitals in Japan in July 2014, which were certified by each Japanese society. All cases were diagnosed according to the histopathological findings consistent with ECD, which typically contain infiltration of foamy or lipid-laden histiocytes and which were positive for CD68, CD163 and negative for CD1a and Langerin on immunohistochemical staining. The questionnaire was also sent to hospital departments containing a member who published a paper or presented at an academic conference on ECD cases. The samples of the ECD lesions from the peripheral blood and/or bone marrow were also collected, if available. This study was performed in accordance with the Declaration of Helsinki and the Ethical

The questionnaire was sent to 3850 departments, of which 52% (2007 departments) responded. We confirmed that in Japan 75 patients have ECD, and detailed data were collected from 45 patients. One patient was excluded from the analyses because of insufficient pathological validity. Table 1 shows the clinical characteristics of the remaining 44 patients. The first signs of the disease are described in Online Supplementary Table S1. The median age at initial onset was 51 (range: 23–76) years. Among the 44 patients, 28 were males (63.6%) and 16 were females (36.4%). Five, eight, nine, and 22 patients were diagnosed before 1999, between 2000 and 2004, between 2005 and 2009, and after 2010, respectively. No association between time of diagnosis and clinical features, other than treatment choice, was observed. Three patients were diag-

1816

Mutation analysis Genomic DNA was extracted from each formalin-fixed, paraffin-embedded tissue and peripheral blood specimen using the QIAamp DNA FFPE Tissue Kit and DNA Mini Kit (Qiagen, Hilden, Germany), respectively. Polymerase chain reaction (PCR) for the detection of BRAF V600E mutation was performed using primers (TACCTAAACTCTTCATAATGCTTGC, GTAACTCAGCAGCATCTCAGGG) as previously reported.18 The products were purified with Illustra ExoStar (GE Healthcare, Tokyo, Japan), and Sanger sequencing was conducted using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) and the ABI Prism 3100 Genetic Analyzer (Life Technologies, Carlsbad, CA, USA).

Statistical analysis The numerical and categorical variables were compared using the t-test and Fisher’s exact test, respectively. Three patients diagnosed at autopsy were excluded from the analyses of the interval from onset to diagnosis. The survival time was calculated through the Kaplan–Meier method, and the log-rank test was utilized for the evaluation of the significant differences. The effect of various parameters on survival was evaluated through univariate and multivariate analyses with the use of the Cox proportional hazards regression model. ECD-related death was defined as death associated with ECD (such as heart failure due to cardiac involvement and increased intracranial pressure caused by brain mass etc.). The cumulative incidence of ECD-related death was calculated in a competing risks model. The factors with P<0.05 in the univariate analyses were included in the multivariate analysis of the survival. Hazard ratio (HR) was estimated with 95% confidence intervals (CI), and the respective P values were reported from these analyses. Differences were considered statistically significant at P values <0.05. Statistical analyses were performed using R version 3.3.2 (The R Foundation for Statistical Computing, Vienna, Austria).

haematologica | 2018; 103(11)

Prognostic factors of ECD

nosed at autopsy or after death, and the median and mean time from onset to diagnosis was 12 (range: 1–89) and 20.7 months (Online Supplementary Figure S1). No significant difference between interval from onset to diagnosis and presence of organ involvement, age, year of diagnosis, and any other clinical features were detected. Fifteen patients (33.3%) were treated with interferon a (IFN-a) or pegylated IFN. Twenty-five patients (57.8%) were administered with corticosteroid. Patients who were diagnosed after 2005 were more frequently administered IFN than those diagnosed before 2005 (43.8% vs. 7.7%, P=0.034). No patient was administered with vemurafenib or other BRAF inhibitors. The median number of involved organs per patient was four (range: 1–11). The lesions were mainly found in the bone (86.4%), the CNS (50.0%), the cardiovascular system (52.3%), the kidney and retroperitoneal organs (50.0%), the endocrine organs (40.9%), the skin (40.9%), the lungs (34.1%), and the digestive organs (13.6%). Oropharyngeal lesions in three patients (6.8%), lymphadenopathy in three patients (6.8%), splenomegaly in two patients (4.5%), and breast lesions in two patients (4.5%) were also detected. In one patient, the testes, prostate, bladder, spinal cord, glottis, and bone marrow were also affected. These unusually affected organs were exclusively observed in patients with multiple (three or more) lesions and, in many cases, who were diagnosed at autopsy. Thirty-eight patients had skeletal lesions, of which 22 had bone pain. Four patients (9.1%) had simultaneous or heterochronic LCH lesions. No patient was diagnosed as having immunoglobulin (Ig)G4-related disease as the clinical and pathological findings did not correspond to the disease, although specific examinations such as serum IgG-4 levels and immunohistochemistry for IgG4 were not carried out on the majority of patients. Among the 18 patients with endocrine involvement, 15 had diabetes insipidus (DI), two had pituitary tumors without obvious DI and one had hypothyroidism. Twenty-three patients had cardiovascular involvement including 19 with cardiac involvement and 11 with aortitis (seven had both). Among the 22 patients with CNS involvement, six had exophthalmos, 12 had parenchymal brain mass and four had meningeal involvement. Two patients had no radiological finding of CNS but their neurological symptoms (delirium and cerebellar ataxia) were considered to be symptoms of ECD. In addition, one patient had low-risk myelodysplastic syndrome (MDS) prior to ECD. He did not receive any treatment for MDS and one year later, after the detection of an abnormal pulmonary shadow, ECD was diagnosed through partial pneumonectomy. The shadow shrank after cyclosporine and prednisolone therapy, but he died of pneumonia 4.5 years after the ECD diagnosis.

Survival The median follow-up period of survivors was 5.17 (range: 1.0–21.1) years. Eighteen out of 44 patients died during the clinical course of their illness, and the cause of death was associated with ECD in 13 patients (72.2%, Online Supplementary Table S2). Moreover, two other patients died from infection, possibly due to the use of corticosteroid therapy as a treatment for ECD. The median survival from the initial onset, which was determined using the Kaplan–Meier method, was 10.42 years (Figure 1A). The log-rank test results revealed that haematologica | 2018; 103(11)

Table 1. Clinical characteristics of ECD patients

Characteristics Sex (male–female) Median age at onset, y (range) Median age at diagnosis, y (range) Mean time for diagnosis, m (range) Year of diagnosis, no. (%) Before 2000 2000–2004 2005–2009 2010 and later Coexistence of LCH, no. (%) Median number of involved organs (range) Skeletal involvement, no. (%) Bone pain CNS involvement, no. (%) Parenchymal brain mass Exophthalmos Meningeal involvement Without any radiological findings Cardiovascular involvement, no. (%) Cardiac involvement Aortitis Retroperitoneal involvement, no. (%) Hydronephrosis Hairy kidney Endocrine involvement, no. (%) Diabetes insipidus Pituitary tumor without diabetes insipidus Hypothyroidism Cutaneous involvement, no. (%) Xanthelasma Other Pulmonary involvement, no. (%) Pleural thickening Fibrosis Interlobular septal thickening Centrilobular shadow Digestive involvement, no. (%) Pancreas Liver Gastrointestinal Gallbladder CRP at onset (mg/dL), median (range) Treatments, no. (%) IFN a or PEG-IFN Corticosteroid Irradiation Imatinib mesylate Cyclophosphamide Cladribine Etoposide Cyclosporine Prostaglandin I2 CHOP JLSG-02 Bisphosphonate Lung transplantation

All patients (n = 44) 28–16 51 (23–76) 53 (23–77) 20.7 (1–89) 5 (11.4) 8 (18.2) 9 (20.5) 22 (50.0) 4 (9.1) 4 (1–11) 38 (86.4) 22 22 (50.0) 12 6 4 2 23 (52.3) 19 11 22 (50.0) 10 8 18 (40.9) 15 2 1 18 (40.9) 11 7 15 (34.1) 5 3 2 1 6 (13.6) 3 2 2 1 2.61 (0.1–14.1) 15 (34.1) 25 (56.8) 5 (11.4) 5 (11.4) 4 (9.1) 1 (2.3) 2 (4.5) 1 (2.3) 1 (2.3) 1 (2.3) 1 (2.3) 10 (22.7 1 (2.3)

ECD: Erdheim–Chester disease; LCH: Langerhans cell histiocytosis; CNS: central nervous system; CRP: C-reactive protein; IFN-a: interferon a; PEG: pegylated; CHOP: cyclophosphamide, adriamycin, vincristine, and prednisolone; JLSG-02: a regime protocol for LCH consisting of cytarabine, vincristine, and prednisolone.

1817

T. Toya et al.

the prognostic factors for this disease were age >60 years old at the onset (P<0.0001), lack of bone lesions (P<0.0001), and the involvement of the CNS (P=0.0010) and digestive organs (P=0.017) (Figure 1B-E). In contrast, the presence of cardiovascular disease, endocrinosis, and pulmonary, kidney/retroperitoneal and cutaneous lesions was not considered as a prognostic factor (Online

Supplementary Figure S2). Patients with only one ECD lesion, all of whom had a bone lesion, tended to have better survival after the onset of the disease than those with multiple lesions, although this difference was not statistically significant, perhaps because the number of patients who had a single lesion was only five (P=0.064). The univariate analyses, which were conducted using

Figure 1. Survival curves of patients with Erdheimâ&#x20AC;&#x201C;Chester disease. Kaplanâ&#x20AC;&#x201C; Meier estimation for survival from onset of (A) all 44 patients based on (B) age and presence of (C) bone lesions, (D) central nervous system involvement, and (E) digestive involvement. CNS: central nervous system; DO: digestive organ.

1818

haematologica | 2018; 103(11)

Prognostic factors of ECD

Cox proportional hazards regression model, revealed that age >60 years (P=0.00035), the presence of CNS involvement (P=0.0021) and digestive organ disease (P=0.0022), and the absence of bone lesions (P=0.0042) were significant poor prognostic factors for survival from the onset (Table 2). Sex, year of diagnosis, and existence of cardiovascular, endocrine, cutaneous, or pulmonary involvement and kidney/retroperitoneal lesions did not have an effect on survival.

The multivariate analysis showed that age >60 years (HR, 25.9; 95% CI, 2.82–237; P=0.0040) and the presence of digestive organ involvement (HR, 4.74; P=0.043) were correlated with worse survival.

ECD-related death Next, we analyzed the association between the cumulative incidence of ECD-related death and clinical factors (Figure 2A-D). ECD-related death was significantly associ-

Table 2. Univariate and multivariate analyses for survival.

Age, years Sex Year of diagnosis Skeletal involvement CNS involvement Cardiovascular involvement Retroperitoneal involvement Endocrine involvement Cutaneous involvement Pulmonary involvement Digestive involvement

≥ 60 < 60 Male Female Before 2005 2005 or later Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No

No. of patients

Univariate analysis P

Multivariate analysis HR (95% CI)

15 29 28 16 13 31 38 6 22 22 23 21 22 22 18 26 18 26 15 29 6 38

0.00035

25.9 (2.82–237) 1

P 0.0040

0.52 0.38 0.0042 0.0021

1 2.16 (0.660–7.09) 3.63 (0.844–15.6) 1

0.20 0.083

0.14 0.58 0.86 0.77 0.94 0.0022

4.74 (1.05–21.4) 1

0.043

CNS: central nervous system; HR: hazard ratio; CI: confidence interval.

Table 3. Skeletal involvement and clinical factors.

Characteristics Sex (male–female) Median age at onset, y (range) Diagnosed after 2005, no. (%) Coexistence of LCH, no. (%) Median number of involved organs (range) CNS involvement, no. (%) Cardiovascular involvement, no. (%) Retroperitoneal involvement, no. (%) Endocrine involvement, no. (%) Diabetes insipidus, no. (%) Cutaneous involvement, no. (%) Pulmonary involvement, no. (%) Digestive involvement, no. (%) CRP at onset (mg/dL), median (range)

Skeletal (−) (n = 6)

Skeletal (+) (n = 38)

3–3 59 (40–67) 2 (33.3) 0 (0) 3 (2–8) 5 (83.3) 3 (50.0) 3 (50.0) 3 (50.0) 1 (16.7) 4 (66.7) 1 (16.7) 1 (16.7) 5.38 (3.30–7.46)

25–13 49 (23–76) 29 (76.3) 4 (11.1) 4 (1–11) 17 (44.7) 20 (52.6) 19 (50.0) 15 (39.5) 14 (36.8) 14 (36.8) 14 (36.8) 5 (13.2) 2.45 (0.10–14.1)

0.65 0.43 0.053 1 0.77 0.19 1 1 0.68 0.65 0.21 0.65 0.56 0.49

LCH: Langerhans cell histiocytosis; CRP: C-reactive protein; CNS: central nervous system.

haematologica | 2018; 103(11)

1819

T. Toya et al.

ated with age >60 years (P<0.0001), digestive organ (P<0.0001), CNS (P=0.0062), and skeletal involvement (P=0.0072). Cardiovascular (P=0.071), cutaneous (P=0.26), retroperitoneal (P=0.27), respiratory (P=0.28), endocrine involvement (P=0.83) and sex (P=0.89) were not significantly associated with ECD-related death.

ECD without skeletal lesion Considering that the prevalence of skeletal lesions in our study was relatively low compared with that of previous studies in Western countries (>95%),5,19 we compared the clinical data of ECD patients with and without skeletal lesions. No obvious clinical difference based on skeletal involvement was observed, except that patients with skeletal lesions had a better outcome than those without (Figure 1C and Table 3).

ECD with digestive involvement Given that ECD with digestive involvement was associated with worse survival, we examined the characteristics of ECD patients with digestive disease (Table 4). The digestive organs concerned were the liver (50.0%), pancreas (50.0%), gastrointestinal tract (33.3%), and gallbladder (16.7%). Causes of death were arrhythmia in one patient with cardiac tamponade, heart and renal failure in

one patient with pericardial effusion and hydronephrosis, sepsis due to rectal perforation in one patient with gastrointestinal involvement of ECD, and ECD (details unknown) in one patient. An autopsy was carried out in four patients who died during the clinical course of digestive disease, and the involvement of ECD cells was histologically proven. Histological samples were available in two patients and both were BRAFV600E positive. No correlation was found between the existence of digestive organ disease and the CRP level at onset (P=0.99), age at onset (P=0.11), sex (P=0.39), year of diagnosis (P=1), and the presence of specific organ involvement.

CNS involvement and associated clinical factors Considering that the CNS was revealed to be a “risk organ”, based on the result of our analysis and that of a previous study,16 we compared the characteristics of patients with and without CNS disease to determine those were are at risk of CNS involvement (Table 5). Patients with CNS disease displayed significantly higher age at onset than those without (median, 62 [range, 23–76] years vs. 45 [range, 25–70] years; P=0.033). Additionally, the existence of CNS involvement was correlated with the presence of cardiovascular lesions (P=0.015). No statisti-

Figure 2. Cumulative incidence of ECD-related death. Competing risks models revealed the cumulative incidence of ECD-related death based on (A) age and presence of (B) digestive organs, (C) CNS, (D) skeletal involvement. DO: digestive organ; CNS: central nervous system.

1820

haematologica | 2018; 103(11)

Prognostic factors of ECD

cally significant association between the presence of CNS involvement and sex (P=0.75), year of diagnosis (P=1.0), and existence of bone (P=0.19) and skin lesions (P=0.36), endocrinosis (P=0.12), pulmonary (P=0.20) and digestive organ involvement (P=0.66), or kidney/retroperitoneal disease (P=0.13) was detected.

CRP level and outcome The CRP level at onset was higher than the upper normal limit in 29 (85.3%) out of 34 patients with sufficient laboratory data, with a median CRP level at onset of 2.61 mg/dL (range: 0.10–14.1). Interestingly, a >3.0 mg/dL CRP level at onset was associated with a worse outcome (median survival, 14.6 vs. 7.4 years; P=0.016; Figure 3A). The cumulative incidence of ECD-related death was also associated with a higher CRP level at onset (P=0.043; Figure 3B). We also compared the CRP level at onset with that after the administration of first-line therapy in 25 patients whose clinical data were available. The results of the Wilcoxon signed-rank test revealed that CRP levels tended to decline following the initial treatment (median at onset and after initial therapy: 2.38 [range, 0.10–14.1] mg/dL and 1.16 [range, 0.01–34.6] mg/dL, respectively; P=0.051; Figure 3C). In addition, patients whose CRP levels reduced by more than 2.0 mg/dL after first-line therapy had a better outcome compared with patients whose CRP levels dropped by less than 2.0 mg/dL or increased after first-line treatment (median survival, 11.3 vs. 7.4 years; P=0.045) (Figure 3D).

BRAFV600E mutation Thirteen patients had available histological samples on ECD lesions (Online Supplementary Table S3). The genomic DNA was extracted from the samples, which were then subjected to direct sequencing. A BRAFV600E mutation was detected in 11 out of 14 patients (78%). None of the BRAF mutations were detected through Sanger sequencing of the genomic DNA extracted from the peripheral blood or bone marrow samples. Moreover, no correlation between BRAF mutation status and age, CRP level at onset, and other clinical factors were observed.

Discussion Our nationwide study broadly investigated ECD patients and analyzed the clinical data of 44 patients. ECD is so rare that few reports on multiple ECD patient studies have been published and little evidence about the clinical characteristics or prognostic factors of this disease is available. The study herein is one of the largest in terms of the number of patients with ECD involved in our research.2,6,7,16 In this study, the duration between onset and diagnosis seems to be shorter compared with previous studies, in which many patients were diagnosed several years after initial onset. It might reflect an increased familiarity with ECD in recent years, although the exact reason is unclear.20 IFN has recently been recommended as first-line therapy, and BRAF inhibitors are also strong candidates for the treatment of ECD.5,16,21 In our study, IFN was administered to a very small proportion of patients during the clinical course of the disease, and no patients received BRAF inhibitors, which is partially attributed to the Japanese insurance system. Instead, many patients were prescribed corticosteroid, which is believed to temporarily alleviate the symptoms, although it is not recommended by the consensus guidelines.5 Two patients died from infection (one pneumonia and one invasive pulmonary aspergillosis), possibly due to immunosuppression induced by corticosteroid administration for the treatment of ECD. In our study, the mortality rate was relatively high compared with a recent report which showed a five year survival rate of 82.7 %.4 However, the outcome in our cohort was slightly better than that in patients who were not administered IFN in a previous report,16 perhaps due to the improvement of supportive care. To improve the prognosis of patients with ECD, more detailed analyses and prospective studies of the pathophysiology of ECD are required. Given that future studies on ECD may not include patients who were not administered with IFN and/or BRAF inhibitors, this study could serve as an important physician’s compass that reveals the baseline clinical behaviors of ECD. CNS involvement was a significant poor prognostic fac-

Table 4. Digestive involvement and clinical factors.

Characteristics Sex (male–female) Median age at onset, y (range) Diagnosed after 2005, no. (%) Coexistence of LCH, no. (%) Median number of involved organs (range) Skeletal involvement, no. (%) CNS involvement, no. (%) Exophthalmos, no. (%) Cardiovascular involvement, no. (%) Retroperitoneal involvement, no. (%) Endocrine involvement, no. (%) Cutaneous involvement, no. (%) Pulmonary involvement, no. (%) CRP at onset (mg/dL), median (range)

Digestive (−) (n=38)

Digestive (+) (n=6)

23–15 49 (23–76) 27 (71.1) 4 (11.1) 3 (1–7) 33 (91.7) 18 (47.4) 5 (13.2) 19 (50.0) 17 (44.7) 13 (34.2) 16 (42.1) 12 (31.6) 2.52 (0.10–14.1)

5–1 66 (46–68) 4 (66.6) 0 (0) 8 (3–11) 5 (83.3) 4 (66.7) 1 (16.7) 4 (66.7) 5 (83.3) 3 (50.0) 2 (33.3) 3 (50.0) 2.70 (0.13–7.46)

0.39 0.11 1 1 0.015 1 0.66 1 0.67 0.19 0.68 1 0.39 0.99

LCH: Langerhans cell histiocytosis; CRP: C-reactive protein; CNS: central nervous system.

haematologica | 2018; 103(11)

1821

T. Toya et al.

tor in our series, which seems to be consistent with the previous study reporting that CNS involvement was associated with resistance to IFN.16 Our study revealed that ECD patients with CNS lesions were significantly older and had cardiovascular lesions more frequently. These factors might affect the efficacy of IFN, as a cardiovascular lesion has also been suggested to contribute to the ineffectiveness of IFN therapy.22 Furthermore, the side effects of IFN, such as delirium, is known to be more frequently observed in older patients than in younger ones, despite intense psychiatric support.23 Future study is necessary to identify patients who are expected to benefit the most from treatment with IFN. The prevalence of affected organs, median age at onset and male predominance were roughly comparable with that of previous reports in Western countries,5,7 excepting the article by Cavalli et al. which found that CNS manifestations in younger patients are relatively common,7 and the fact that the percentage of patients with skeletal disease in our cohort was relatively low.5 The coexistence of LCH was also relatively rare in our series. Considering the close relationship between ECD and LCH, perhaps some

cases were overlooked.24 This dissociation might be attributed to coincidence, ethnicity, or other genetic/epigenetic diversity. No statistically significant clinical difference was found between ECD patients with and without bone involvement, which suggests that ECD cases have common features regardless of this factor. We were unable to obtain evidence that demonstrates whether ECD without bone lesions should be classified as a distinctive entity. The digestive organ was detected as a risk organ for patients with ECD in this study. The prognostic impact of digestive involvement was not found in previous reports,5,6 perhaps due to the small number of patients with digestive ECD lesions. Although the liver and pancreas were the most commonly affected organs, none of these patients with hepatic/pancreatic involvements suffered from the relevant organ failure. Four of six patients with digestive organ involvement died of ECD, however, only one patient died of rectal perforation due to ECD involvement and the other three patients died of other affected organs. A further accumulation of cases is necessary in order to confirm the prognostic impact of digestive involvement.

Figure 3. C-reactive protein (CRP) at onset and clinical outcome. (A) Kaplanâ&#x20AC;&#x201C;Meier estimation for survival from onset and (B) the cumulative incidence of ECD-related death of 34 patients with sufficient clinical data. (C) Comparison of the CRP level before and after administration of first-line therapy. (D) Kaplanâ&#x20AC;&#x201C;Meier estimation for survival from onset according to the decline of CRP levels after first-line therapy.

1822

haematologica | 2018; 103(11)

Prognostic factors of ECD

The CRP level was elevated in many patients with ECD, and declined after IFN or vemurafenib treatment,8,22 similar to the finding in our cohort. In this study, we firstly revealed that the CRP level at onset as well as the drop in CRP levels after initial therapy predict the patient outcome. We have to interpret this result with caution because initial treatments in our cohort were quite heterogeneous and most of them were not standard therapy. Recently, ECD has been considered as an inflammatory myeloid neoplasm. Although the MAPK pathway and other mutations, such as PIK3CA, tend to attract attention to the neoplastic aspect of the disease, the association between CRP levels and poor prognosis evokes the necessity of vigorous research on the inflammatory characteristic of Janus-faced ECD because these mutations themselves do not necessarily induce elevated CRP. Some inflammatory markers, such as interleukin (IL)-1, IL-6 and tumor necrosis factor, have also been reported as activation markers of ECD.25-28 However, CRP levels are relatively easy to measure compared with inflammatory cytokines and might be suitable as a convenient marker to detect patients at risk. Interestingly, low-risk MDS and ECD coexisted in one patient in our cohort. Papo et al. reported that adult histiocytic neoplasms frequently co-occur with other myeloid neoplasms (9.5%), and MAPK pathway-associated muta-

tions (BRAF, NRAS, and other mutations) could coexist with MDS-associated mutations (IDH1/2, ASXL1, and TET2 mutations) in the hematopoietic stem/progenitor cells of ECD patients.29 Unfortunately, we were unable to acquire a sample from the patient, and the genetic information was unavailable. Due to the retrospective nature of our study, we could not analyze some of the most important clinical information of many of the patients, such as osteosclerosis of the sinuses and BRAF or other mutations.30 It is now mandatory to systematically collect clinical findings, radiological and laboratory data, and above all histological samples for patients with ECD. A more sophisticated method of detection for the BRAF mutation is also warranted. Sanger sequencing is a classic method and still plays an important role in detecting ECD, partially owing to its low cost and convenience. However, immunohistochemistry for the BRAF V600E protein and hypersensitive methods, such as droplet digital PCR and/or targeted sequencing techniques, are becoming crucial because only a small fraction of cells in an ECD lesion actually harbor the BRAF mutation, and some studies have reported that a BRAF mutation in the peripheral blood could only be detected with ultrasensitive techniques as opposed to Sanger sequencing.15 In summary, our nationwide survey revealed the clinical characteristics of patients with ECD, and clarified the

Table 5. CNS involvement and clinical factors.

Characteristics

CNS (−) (n=22)

CNS (+) (n=22)

Sex (male–female) Median age at onset, y (range) Diagnosed after 2005, no. (%) Coexistence of LCH, no. (%) Median number of involved organs (range) Skeletal involvement, no. (%) Cardiovascular involvement, no. (%) Retroperitoneal involvement, no. (%) Endocrine involvement, no. (%) Cutaneous involvement, no. (%) Pulmonary involvement, no. (%) Digestive involvement, no. (%) CRP at onset (mg/dL), median (range) Treatments, no. (%) IFN a or PEG-IFN Steroid Radiation Imatinib mesylate Cyclophosphamide Cladribine Etoposide Cyclosporine Prostaglandin I2 CHOP JLSG-02 Lung transplantation

13–9 45 (25–70) 16 (72.7) 2 (9.1) 3 (1–8) 21 (95.5) 7 (31.8) 8 (36.4) 6 (27.3) 11 (50.0) 5 (22.7) 2 (9.1) 1.37 (0.10–14.1)

15–7 62 (23–76) 15 (68.2) 2 (9.1) 5 (2–11) 17 (77.3) 16 (72.7) 14 (63.6) 12 (54.5) 7 (31.8) 10 (45.5) 4 (18.2) 3.64 (0.12–12.6)

0.75 0.033 1 1 0.0042 0.19 0.015 0.13 0.12 0.36 0.20 0.66 0.34

5 (22.7) 8 (36.4) 4 (18.2) 2 (9.1) 3 (13.6) 0 (0) 0 (0) 0 (0) 1 (4.5) 0 (0) 0 (0) 1 (4.5)

10 (45.5) 17 (77.3) 1 (4.5) 3 (13.6) 1 (4.5) 1 (4.5) 2 (9.1) 1 (4.5) 0 (0) 1 (4.5) 1 (4.5) 0 (0)

0.20 0.014 0.34 1 0.61 – – – – – – –

LCH: Langerhans cell histiocytosis; CNS: central nervous system; CRP: C-reactive protein; IFN-a: interferon a; PEG: pegylated; CHOP: cyclophosphamide, adriamycin, vincristine, and prednisolone. JLSG-02: a regimen protocol for LCH consisting of cytarabine, vincristine, and prednisolone

haematologica | 2018; 103(11)

1823

T. Toya et al.

prognostic value of age and digestive involvement. The CRP level at onset also predicts the patientsâ&#x20AC;&#x2122; survival, perhaps reflecting the importance of the inflammatory status in the ECD condition. Additional large-scale studies are required in order to clarify the pathophysiology of ECD and improve the prognosis of the patients. Acknowledgments We would like to acknowledge K. Tanaka and Y. Nakada for their technical assistance. Furthermore, we thank M. Fukuda (Asoka Hospital), H. Hiraga (Hokkaido Cancer Center), T. Yamauchi (Toki General Hospital), K. Takeuchi (Tochigi Cancer Center), Y. Shido (Hamamatsu University School of Medicine), S. Nogawa (Tokyo Dental College Ichikawa General Hospital), T. Hamada (Okayama University), N. Hayashi (Wakayama Rosai Hospital), S. Yokokura (Shin-Yamanote Hospital), T. Makiishi (Japanese Red Cross Otsu Hospital), T. Kitani (Ehime University), Y. Shibuya (Showa University Northern Yokohama Hospital), M. Yoshimitsu (Kagoshima University), J. Mitsushima (Matsue Seikyo General Hospital), S. Yano (Matsue Medical Center), M. Susa (Keio University), K. Ono (Tosei General

References 1. Chester W. Uber lipoidgranulomatose. Virchows Arch Pathol Anat 1930;279:561602. 2. Haroche J, Abla O. Uncommon histiocytic disorders: Rosai-Dorfman, juvenile xanthogranuloma, and Erdheim-Chester disease. Hematology Am Soc Hematol Educ Program. 2015;2015:571-578. 3. Haroche J, Cohen-Aubart F, Charlotte F, et al. The histiocytosis Erdheim-Chester disease is an inflammatory myeloid neoplasm. Expert Rev Clin Immunol. 2015;11(9):10331042. 4. Haroche J, Cohen-Aubart F, Rollins BJ, et al. Histiocytoses: emerging neoplasia behind inflammation. Lancet Oncol. 2017; 18(2):e113-e125. 5. Diamond EL, Dagna L, Hyman DM, et al. Consensus guidelines for the diagnosis and clinical management of Erdheim-Chester disease. Blood. 2014;124(4):483-492. 6. Estrada-Veras JI, O'Brien KJ, Boyd LC, et al. The clinical spectrum of Erdheim-Chester disease: an observational cohort study. Blood Adv. 2017;1(6):357-366. 7. Cavalli G, Guglielmi B, Berti A, Campochiaro C, Sabbadini MG, Dagna L. The multifaceted clinical presentations and manifestations of Erdheim-Chester disease: comprehensive review of the literature and of 10 new cases. Ann Rheum Dis. 2013; 72(10):1691-1695. 8. Haroche J, Cohen-Aubart F, Emile JF, et al. Dramatic efficacy of vemurafenib in both multisystemic and refractory ErdheimChester disease and Langerhans cell histiocytosis harboring the BRAF V600E mutation. Blood. 2013;121(9):1495-1500. 9. Haroche J, Charlotte F, Arnaud L, et al. High prevalence of BRAF V600E mutations in Erdheim-Chester disease but not in other non-Langerhans cell histiocytoses. Blood. 2012;120(13):2700-2703. 10. Emile JF, Diamond EL, Helias-Rodzewicz Z, et al. Recurrent RAS and PIK3CA mutations in Erdheim-Chester disease. Blood. 2014; 124(19):3016-3019.

1824

Hospital), K. Tanaka (Okayama University), M. Yamamoto (Kochi University), K. Sato (Saitama Medical University), H. Sugano (Kochi Health Sciences Center), T. Kondo (Kyoto University), H. Takahama (National Cerebral and Cardiovascular Center), J. Nagano (Seirei Hamamatsu General Hospital), T. Baba (Kanagawa Cardiovascular and Respiratory Center), Y. Fukui (Juntendo University), K. Takenaka (Tsuchiura Kyodo General Hospital), K. Takada (Sapporo Medical University), S. Izumi (National Center for Global Health and Medicine), J. Kikuchi (Keio University), K. Takahashi (Kameda Medical Center), K. Shimizu (Higashihiroshima Medical Center), K. Takemura (Tokyo Teishin Hospital), and I. Choi (National Hospital Organization Kyushu Cancer Center) for providing the clinical data and/or patient samples. We also thank H. Kawano (Teikyo University) and the members of the Japanese Musculoskeletal Oncology Group for cooperating in the collection of clinical data and/or patient samples. Funding This study is supported by Grants-in-Aid of the Ministry of Health, Labor and Welfare of Japan (H28-Nanchi-Ippan-002 to MS, TO, IK, and MK).

11. Milne P, Bigley V, Bacon CM, et al. Hematopoietic origin of Langerhans cell histiocytosis and Erdheim Chester disease in adults. Blood. 2017;130(2):167-175. 12. Berres ML, Merad M, Allen CE. Progress in understanding the pathogenesis of Langerhans cell histiocytosis: back to histiocytosis X? Br J Haematol. 2015;169(1):313. 13. Durham BH, Roos-Weil D, Baillou C, et al. Functional evidence for derivation of systemic histiocytic neoplasms from hematopoietic stem/progenitor cells. Blood. 2017;130(2):176-180. 14. Cavalli G, Biavasco R, Borgiani B, Dagna L. Oncogene-induced senescence as a new mechanism of disease: the paradigm of erdheim-chester disease. Front Immunol. 2014;5:281. 15. Cangi MG, Biavasco R, Cavalli G, et al. BRAFV600E-mutation is invariably present and associated to oncogene-induced senescence in Erdheim-Chester disease. Ann Rheum Dis. 2015;74(8):1596-1602. 16. Arnaud L, Hervier B, Neel A, et al. CNS involvement and treatment with interferon-alpha are independent prognostic factors in Erdheim-Chester disease: a multicenter survival analysis of 53 patients. Blood. 2011;117(10):2778-2782. 17. Cohen Aubart F, Emile JF, Carrat F, et al. Targeted therapies in 54 patients with Erdheim-Chester disease, including followup after interruption (the LOVE study). Blood. 2017;130(11):1377-1380. 18. Arcaini L, Zibellini S, Boveri E, et al. The BRAF V600E mutation in hairy cell leukemia and other mature B-cell neoplasms. Blood. 2012;119(1):188-191. 19. Emile JF, Abla O, Fraitag S, et al. Revised classification of histiocytoses and neoplasms of the macrophage-dendritic cell lineages. Blood. 2016;127(22):2672-2681. 20. Haroche J, Amoura Z, Dion E, et al. Cardiovascular involvement, an overlooked feature of Erdheim-Chester disease: report of 6 new cases and a literature review. Medicine (Baltimore). 2004;83(6):371-392. 21. Haroche J, Cohen-Aubart F, Emile JF, et al.

22.

23.

24.

25.

26.

27.

28.

29.

30.

Reproducible and sustained efficacy of targeted therapy with vemurafenib in patients with BRAF(V600E)-mutated ErdheimChester disease. J Clin Oncol. 2015; 33(5):411-418. Haroche J, Amoura Z, Trad SG, et al. Variability in the efficacy of interferonalpha in Erdheim-Chester disease by patient and site of involvement: results in eight patients. Arthritis Rheum. 2006; 54(10):3330-3336. Raison CL, Demetrashvili M, Capuron L, Miller AH. Neuropsychiatric adverse effects of interferon-alpha: recognition and management. CNS Drugs. 2005;19(2):105123. Hervier B, Haroche J, Arnaud L, et al. Association of both Langerhans cell histiocytosis and Erdheim-Chester disease linked to the BRAFV600E mutation. Blood. 2014; 124(7):1119-1126. Arnaud L, Gorochov G, Charlotte F, et al. Systemic perturbation of cytokine and chemokine networks in Erdheim-Chester disease: a single-center series of 37 patients. Blood. 2011;117(10):2783-2790. Berti A, Cavalli G, Guglielmi B, et al. Tocilizumab in patients with multisystem Erdheim-Chester disease. Oncoimmunology. 2017;6(6):e1318237. Dagna L, Corti A, Langheim S, et al. Tumor necrosis factor alpha as a master regulator of inflammation in Erdheim-Chester disease: rationale for the treatment of patients with infliximab. J Clin Oncol. 2012; 30(28):e286-290. Aouba A, Georgin-Lavialle S, Pagnoux C, et al. Rationale and efficacy of interleukin-1 targeting in Erdheim-Chester disease. Blood. 2010;116(20):4070-4076. Papo M, Diamond EL, Cohen-Aubart F, et al. High prevalence of myeloid neoplasms in adults with non-Langerhans cell histiocytosis. Blood. 2017;130(8):1007-1013. Drier A, Haroche J, Savatovsky J, et al. Cerebral, facial, and orbital involvement in Erdheim-Chester disease: CT and MR imaging findings. Radiology. 2010; 255(2):586-594.

haematologica | 2018; 103(11)

ARTICLE

Chronic Myeloid Leukemia

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

Ferrata Storti Foundation

Artur C. Fassoni,1,2 Christoph Baldow,2 Ingo Roeder2,3* and Ingmar Glauche2*

Instituto de Matemática e Computação, Universidade Federal de Itajubá, Brazil; Institute for Medical Informatics and Biometry, Faculty of Medicine Carl Gustav Carus, Technische Universität Dresden, Germany and 3National Center for Tumor Diseases (NCT), Partner Site Dresden, Germany 1 2

*IR and IG contributed equally to this work.

Haematologica 2018 Volume 103(11):1825-1834

ABSTRACT

ontinuing tyrosine kinase inhibitor (TKI)-mediated targeting of the BCR-ABL1 oncoprotein is the standard therapy for chronic myeloid leukemia (CML) and allows for a sustained disease control in the majority of patients. While therapy cessation for patients appeared as a safe option for about half of those patients with optimal response, no systematic assessment of long-term TKI dose de-escalation has been made. We use a mathematical model to analyze and consistently describe biphasic treatment responses from TKI-treated patients from two independent clinical phase III trials. Scale estimates reveal that drug efficiency determines the initial response while the long-term behavior is limited by the rare activation of leukemic stem cells. We use this mathematical framework to investigate the influence of different dosing regimens on the treatment outcome. We provide strong evidence to suggest that TKI dose de-escalation (at least 50%) does not lead to a reduction of long-term treatment efficiency for most patients, who have already achieved sustained remission, and maintains the secondary decline of BCR-ABL1 levels. We demonstrate that continuous BCR-ABL1 monitoring provides patient-specific predictions of an optimal reduced dose without decreasing the anti-leukemic effect on residual leukemic stem cells. Our results are consistent with the interim results of the DESTINY trial and provide clinically testable predictions. Our results suggest that dose-halving should be considered as a long-term treatment option for CML patients with good response under continuing maintenance therapy with TKIs. We emphasize the clinical potential of this approach to reduce treatment-related side-effects and treatment costs. Introduction In tyrosine kinase inhibitors (TKI)-treated chronic myeloid leukemia (CML), the proportion of BCR-ABL1 mRNA is used to monitor the individual treatment response.1-3 Most patients show a typical bi-phasic response with a steep, initial decline (slope a), followed by a slower, secondary decline (slope β) of BCR-ABL1 levels.4-6 Whereas the initial decline is attributed to the eradication of proliferating leukemic cells (LC), the secondary decline has been suggested to result from a slower eradication of quiescent leukemic stem cells (LSCs).5-7 About two-thirds of the patients achieve major molecular remission (MMR), i.e. a reduction of three logs from the baseline (MR3), while one-third of these even achieve deep molecular remission (DMR, i.e. MR4.5) within five years of treatment.3,6,8 Recently, TKI cessation and, thus, treatment-free remission has been established as an important therapeutic goal.9,10 However, about 50% of the patients with good response experience a molecular relapse after stopping TKI, pointing towards persisting residual LCs that cannot be controlled by patient-specific immunological mechanisms. As these mechanisms underlying the currently unpredictable individual molecular relapse risk remain controversial, complementary approaches to haematologica | 2018; 103(11)

Correspondence: ingmar.glauche@tu-dresden.de

Received: March 29, 2018. Accepted: June 26, 2018. Pre-published: June 28, 2018. doi:10.3324/haematol.2018.194522 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1825 ©2018 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.

1825

A.C. Fassoni et al.

minimize side-effects associated with continuous TKI therapy are required. While a number of studies evaluate TKI cessation, strategies to apply long-term dose reductions are currently under-appreciated, although the potential benefits of dose de-escalation are at hand: besides reducing treatment-related side-effects and increasing patientsâ&#x20AC;&#x2122; quality of life, it also reduces the treatment-related costs.11-14 The DESTINY trial (clinicaltrials.gov identifier: 01804985)15,16 is an ongoing study addressing TKI dose deescalation. However, with its primary end point (molecular relapse risk after TKI cessation preceded by a one year dose de-escalation) this trial focuses on stopping TKI, rather than on long-term outcomes under continuous but reduced TKI treatment. Here, we describe a systematic, conceptual analysis of the impact of dose de-escalation on the long-term disease kinetics. Our simulation study relies on a mathematical description of TKI-treatment, which builds on a previously published CML model.5,17 In contrast to the earlier approach, we use a simplified model which allows for a stringent analytical formulation of the disease dynamics without changing the overall qualitative system properties. The model parameters are estimated from available, patient-specific BCR-ABL1 kinetics, determined within controlled clinical phase III trials [IRIS (clinicaltrials.gov identifier: 0000634318), CML IV (clinicaltrials.gov identifier: 0005587419)]. Our results support the rationale for TKI dose de-escalation in patients who have already reached sustained remission. We provide strong evidence that the long-term depletion of residual LSCs in remission phase is not affected by defined TKI-dose reduction. Furthermore, we propose a strategy to determine patient-specific optimal TKI doses, and predict that dose-halving is a safe treatment option for the majority of patients in sustained molecular remission. The suggested dose optimization can contribute to the prevention of severe side-effects (e.g. cardiovascular complications, pleural effusion) and to a reduction of overall treatment costs.

Methods Patient data and parametrization The presented results are based on a secondary analysis of previously published data from the IRIS (clinicaltrials.gov identifier: 00006343)18 and CML IV (clinicaltrials.gov identifier: 00055874) trials.19 In particular, we used 69 patients from the German imatinib (400 mg) arm of the IRIS trial and 280 patients from the 400 mg imatinib monotherapy arm of the CML-IV study, for which IS corrected BCR-ABL1/ABL1 time courses were available at the time of our primary model analysis.5,17 As described in the original publications,20,21 both clinical trials were conducted in accordance with the Declaration of Helsinki and applicable regulatory requirements. The protocols were approved by the institutional review board or ethics committee of each participating center. All patients or guardians gave written informed consent before participation. For each individual patient, the treatment response at time đ?&#x2018;Ą (đ??żđ?&#x2018;&#x201A;đ??ľđ?&#x2018;&#x2020;(đ?&#x2018;Ą)), measured in the form 100% Ă&#x2014; BCR-ABL1/ABL1, is further described according to a biphasic characteristic, i.e.

1826

For each patient, parameters A, B, a and Î˛ are determined using maximum likelihood estimation. For the model analysis, we selected patients with (i) sufficient time points for model fitting (>4) which (ii) do not show a long-term increase in BCRABL1/ABL1 ratio (Î˛<0) but are characterized by (iii) a biphasic decline, (a<Î˛<0). We also excluded 2 additional patients with measured BCR-ABL1 ratios of more than 500%, which indicates a pronounced non-linearity between BCR-ABL1 abundance and tumor load, resulting in n=55 (IRIS cohort) and n=134 (CML-IV cohort) (Online Supplementary Table S1). Median follow up of this patient cohort is 4.3 years [IQR (2.8,6.3)]; 98.4% of the patients achieved BCR-ABL1 ratio of less than 1%, while 91% achieved MR3 at least once. We also tested the robustness of our model results with respect to the reliability of high BCR-ABL1 values (Online Supplementary Table S2). For comparisons with the DESTINY trial (clinicaltrials.gov identifier: 01804985),15,16 only patients treated with TKI for at least three years and with BCR-ABL1 levels below MR3 for at least the last year of treatment were used. Therefore, we excluded from the study 53 patients treated for less than three years (excluding n=4 for IRIS; n=49 for CML-IV) and 14 patients with no MR3 in the entire last year of treatment (excluding n=4 for IRIS; n=10 for CML-IV). The time courses of the remaining 122 patients [(n=47 IRIS, median follow up 6.5 years [IQR(5.9;6.9)]; n=75 CML-IV, median follow up 4.6 years [IQR(3.9;6.1)] are available in Online Supplementary Figure S1. Following the DESTINY trial, these patients were further split into an MR4 and an MR3 cohort, depending on whether their BCR-ABL1 levels in the last year were below MR4 or not. These selection and classification procedures were based on the individual bi-exponential fit đ??żđ?&#x2018;&#x201A;đ??ľđ?&#x2018;&#x2020;(đ?&#x2018;Ą) of each patient (also shown in Online Supplementary Figure S1).

Mathematical model We apply a mechanistic model that describes TKI response as a dynamic process resulting from the interplay between tumor growth, activation/deactivation of LSCs, and cytotoxic TKI action on proliferating, but not on quiescent LSCs (Figure 1A). The model is a simplification of our previous computational CML model5,17 and formally related to a model proposed by Komarova and Wodarz.22 It considers three leukemic cell types: quiescent LSCs (đ?&#x2018;&#x2039;), proliferating LSCs (đ?&#x2018;&#x152;), and fully differentiated LCs (đ?&#x2018;&#x160;). Mathematically, the model is described by the following set of differential equations:

The activation of dormant LSCs and deactivation of proliferating LSCs are described by rate constants đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152; and đ?&#x2018;?đ?&#x2018;&#x152;đ?&#x2018;&#x2039; respectively. LSCs proliferate with a rate constant đ?&#x2018;?đ?&#x2018;&#x152; . During therapy, a cytotoxic TKI effect acts on proliferating LSCs, described by the rate đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??ź>0. Differentiation of proliferating LSCs is quantified by đ?&#x2018;?đ?&#x2018;&#x160; and the limited life-time of differentiated LCs is modeled by a mortality rate đ?&#x2018;&#x;đ?&#x2018;&#x160;. We define a net leukemia reduction đ?&#x2018;&#x17E;=đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??źâ&#x2C6;&#x2019;đ?&#x2018;?Y with đ?&#x2018;&#x17E;>0 for effective treatment. For illustrating simulations we used parameter values corresponding to the median values of the selected patients (Online Supplementary Text S1-S3). haematologica | 2018; 103(11)

TKI dose reduction in CML

Reimplementation of the model in a stochastic version using a Gillespie algorithm ensured that there are no distinct differences resulting from small cell numbers (Online Supplementary Text S4). In contrast to previous models,5,17 competition between normal cells and LCs is described only implicitly, by assuming constant total cell numbers, đ?&#x2018;&#x2021;đ?&#x2018;&#x152;, đ?&#x2018;&#x2021;X , đ?&#x2018;&#x2021;W in each cell compartment (Figure 1D, Online Supplementary Text S1 and Online Supplementary Figure S2 complementing Figure 1C on the level of absolute cell numbers). The actual tumor load, corresponding to BCR-ABL1 levels, is modeled as the percentage of LCs with respect to the total cell number. Figure 1C demonstrates that the

haematologica | 2018; 103(11)

modeled BCR-ABL1 levels of proliferating LSCs behave exactly like the BCR-ABL1 levels in the peripheral blood (PB). Therefore, only the dynamics of proliferating LSCs will be considered.

Results The long-term effect of TKI is limited by the rare activation of quiescent LSCs We apply a simple mathematical model that describes the time course of TKI response in CML as a dynamic

Figure 1. Mathematical model for chronic myeloid leukemia (CML) treatment and mechanistic interpretation of the bi-phasic decline. (A) Schematic model representation with three cell types: quiescent (đ?&#x2018;&#x2039;, blue) and proliferating (đ?&#x2018;&#x152;, red, turnover with rate đ?&#x2018;?đ?&#x2018;&#x152;) leukemic stem cells (LSCs), and differentiated leukemic cells (LCs), denoted by đ?&#x2018;&#x160; (green, generated with rate đ?&#x2018;?đ?&#x2018;&#x160;, decaying with rate rđ?&#x2018;&#x160;). The model assumes (i) mechanisms of activation/deactivation of quiescent/proliferating LSCs with rates đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152; and đ?&#x2018;?đ?&#x2018;&#x152;đ?&#x2018;&#x2039; and (ii) a cytotoxic effect of TKI on proliferating LSCs with intensity đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??ź. (B) The mechanistic model parameters [(TKI net effect (đ?&#x2018;&#x17E;=đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??źâ&#x2C6;&#x2019;đ?&#x2018;?Y), activation rate of quiescent LSCs ( đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152;), deactivation rate of proliferating LSCs (đ?&#x2018;?đ?&#x2018;&#x152;đ?&#x2018;&#x2039;)] were fitted to individual patient data from the IRIS and CML-IV trials.18,19 The resulting distributions reveal an intrinsic scaling between them, which are dispersed over different orders of magnitude. (C) Model simulation with median parameter values obtained from IRIS and CML-IV data illustrating the equivalence between tumor load (in terms of BCR-ABL1 levels) in the peripheral blood (green) and within the proliferating LSCs (red). Values on the y-axis indicate the relative abundance of BCR-ABL1 positive cells in each specific cell compartment [see equation (SE1) in Online Supplementary Text S1], which corresponds to the tumor load in terms of PCR-based measurements of the BCR-ABL1/ABL1 ratio. We adopted this scheme for all corresponding figures throughout the manuscript. Using the intrinsic scaling (B), the slopes in the bi-exponential decline of the BCR-ABL1 levels simplify to đ?&#x203A;źâ&#x2030;&#x2C6;â&#x2C6;&#x2019;đ?&#x2018;&#x17E; and đ?&#x203A;˝â&#x2030;&#x2C6;â&#x2C6;&#x2019;đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152;. The abundance of quiescent LSCs follows a monophasic decline approximated by đ?&#x203A;˝â&#x2030;&#x2C6;â&#x2C6;&#x2019;đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152;. See Online Supplementary Text S3 for parameter values used in all model simulations. (D) During the initial phase (upper panel, â&#x20AC;&#x153;1st slopeâ&#x20AC;?), eradication of the proliferating LSCs (red) with effective rate q is the dominating process (large black arrow). After the strong initial reduction, few proliferating cells remain (lower panel, â&#x20AC;&#x153;2nd slopeâ&#x20AC;?) and eradication is now limited by the activation rate đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152; (small black arrow) of quiescent LSCs (blue). Normal cells are shown in gray. See also Online Supplementary Figure S2.

1827

A.C. Fassoni et al.

process resulting from the interplay between tumor growth, activation/deactivation of LSCs, and cytotoxic TKI action (Figure 1A, and see Methods). In brief, the model describes three LC types: quiescent LSC (X), proliferating LSC (Y) and fully differentiated LCs (W). The activation of dormant LSCs and deactivation of proliferating LSCs are described by rate constants đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152; and đ?&#x2018;?đ?&#x2018;&#x152;X while LSCs proliferate with a rate constant đ?&#x2018;?đ?&#x2018;&#x152;. During therapy, we assume a cytotoxic TKI effect on proliferating LSCs,

described by the rate constant đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??ź>0. We obtain an exact solution for the model, in which the patient-specific response can be expressed in terms of the mechanistic model parameters [equation (SE5) in Online Supplementary Text S2]. In other words, a patientâ&#x20AC;&#x2122;s bi-phasic BCR-ABL1 decline characterized by the slopes a, Î˛ is expressed in terms of the resulting net cytotoxic effect đ?&#x2018;&#x17E;=đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??źâ&#x2C6;&#x2019;đ?&#x2018;?đ?&#x2018;&#x152; (difference between TKI toxicity and LSC proliferation) and the effects of LSC activation/deactivation đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152;

Figure 2. Model predictions on dose de-escalation and dose escalation. (A) The long-term treatment efficiency, defined as the magnitude of the second slope đ?&#x203A;˝, is shown as a function of the dose reduction. The threshold for optimal favorable reduction, đ?&#x2018;&#x201C;đ?&#x2018;&#x201A;đ?&#x2018;&#x192;đ?&#x2018;&#x2021; (â&#x2030;&#x2C6;25% in this example, i.e. using median parameters as in Figure 1C) indicates how much the standard dose can be reduced without losing treatment efficiency. đ?&#x2018;&#x201C;đ?&#x2018;&#x201A;đ?&#x2018;&#x192;đ?&#x2018;&#x2021; can be calculated for each patient (see main text). Any other favorable dose reductions (dose fraction đ?&#x2018;&#x201C;>đ?&#x2018;&#x201C;đ?&#x2018;&#x201A;đ?&#x2018;&#x192;đ?&#x2018;&#x2021; , green region) also retain the long-term treatment efficiency, while unfavorable dose reductions (dose fraction đ?&#x2018;&#x201C;<đ?&#x2018;&#x201C;đ?&#x2018;&#x201A;đ?&#x2018;&#x192;đ?&#x2018;&#x2021; , red region) are predicted to lead to a severe decrease in the long-term treatment efficiency. (B-E) Simulations of favorable (B), optimal favorable (C) and unfavorable (D and E) dose reductions after 36 months under standard dose. After favorable dose reductions, a transient increase in proliferating leukemic stem cells (LSCs) (red) is followed by a return to the original decrease rate đ?&#x203A;˝â&#x2030;&#x2C6;â&#x2C6;&#x2019;đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152;, while the dynamics of quiescent LSCs remains unchanged (blue lines). In the case of unfavorable reduction, an impaired scenario is observed. See also Online Supplementary Figures S3-S6. (F) dose escalation to đ?&#x2018;&#x201C;=200% after three years of treatment; although a deeper level is reached in the BCR-ABL1 levels of proliferating LSCs, the dynamics of quiescent LSCs remains unchanged.

1828

haematologica | 2018; 103(11)

TKI dose reduction in CML

and đ?&#x2018;?đ?&#x2018;&#x152;đ?&#x2018;&#x2039;. Taking into account the intrinsic scaling between the mechanistic parameters (Figure 1B), which indicates that LSC deactivation, activation, and depletion by TKI occur at different time scales, those expressions for đ?&#x203A;ź and đ?&#x203A;˝ can be further simplified, allowing us to dissect the prominent processes governing each treatment phase (Figure 1C and D, Online Supplementary Texts S5-S7 and Online Supplementary Figure S2). We found that slope đ?&#x203A;ź can be expressed as đ?&#x203A;źâ&#x2030;&#x2C6;â&#x2C6;&#x2019;đ?&#x2018;&#x17E;=đ?&#x2018;?đ?&#x2018;&#x152;â&#x2C6;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??ź, thereby confirming that the initial treatment phase is dominated by the cytotoxic TKI effect on proliferating LSCs, which leads to a rapid reduction in BCR-ABL1 levels. Dose-escalation studies for imatinib substantiate this result by indicating that a higher TKI dose leads to a more rapid response.23,24 Similarly, slope đ?&#x203A;˝ can be approximated as đ?&#x203A;˝â&#x2030;&#x2C6;â&#x2C6;&#x2019;đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152;, implying that after depletion of initially abundant proliferating LSCs, the treatment response is bounded by the rare activation of quiescent LSCs. This provides a consistent explanation for the slower long-term decrease in proliferating and quiescent LSCs. These analytical conclusions confirm previous findings,5-7 but also allow further predictions on the effect of dose deescalation to be made.

A wide range of reduced TKI doses is predicted to induce the same long-term response as standard dose Due to the bounded activation of quiescent LSCs after the initial therapy response, there is a range of favorable reduced TKI doses where the long-term efficiency (defined by the magnitude of slope đ?&#x203A;˝) remains almost constant with the same overall efficiency as that achieved when applying the standard dose (green region in Figure 2A, and Online Supplementary Figure S3). In this case, the resulting, although reduced, cytotoxic TKI effect is still sufficient to target the abundant proliferating LSCs once a patient has reached sustained remission. This range of â&#x20AC;&#x2DC;favorableâ&#x20AC;&#x2122; reduced doses spans from the standard full-dose to a certain threshold, i.e. an â&#x20AC;&#x2DC;optimal favorableâ&#x20AC;&#x2122; dose (green dashed line in Figure 2A), below which an accelerated decrease in long-term treatment efficiency is observed. Therefore, dose reductions below this optimal dose are considered as â&#x20AC;&#x2DC;unfavorableâ&#x20AC;&#x2122; (red region in Figure 2A). We look for a mathematical expression, which allows us to estimate this optimal dose reduction for each patient in terms of the model parameters. Although there is a minimal required plasma concentration for TKI cytotoxicity,

Figure 3. Step-wise treatment optimization. (A) Step 1: initial treatment with standard dose until the patient shows a clearly identifiable second slope (approx. 18 months) and determination of the bi-exponential parameters (A, B, a, Î˛). (B) Step 2: reduction of tyrosine kinase inhibitor (TKI) dose by half and continuous monitoring of the treatment response until the new intercept Bâ&#x20AC;˛ can be inferred (approx. 18 months). (C) Step 3: reduction to the optimal dose calculated from values of the identified parameters (A, B, Bâ&#x20AC;&#x2122;, a, Î˛; see main text). (D) The long-term follow up using optimal dose shows that the response to this adaptive treatment adheres to the original slope Î˛ for the eradication of residual leukemic stem cells (LSCs) as standard dose treatment. Although the adapted treatment leads to a delay in the reduction of BCR-ABL1 levels in proliferating LSCs and, therefore also in the peripheral blood, the treatment dynamics in the residual quiescent LSCs are unaltered while drug intake is drastically reduced.

haematologica | 2018; 103(11)

1829

A.C. Fassoni et al.

which sets a lower limit to the dose de-escalation, a linear dose-response relationship is generally accepted over a wide range of treatment-relevant doses above this threshold.25,26 Thus, we deduced an explicit expression for the patient-specific optimal favorable dose reduction fraction đ?&#x2018;&#x201C;đ?&#x2018;&#x201A;đ?&#x2018;&#x192;đ?&#x2018;&#x2021; (Online Supplementary Text S8), given by

This optimal fraction đ?&#x2018;&#x201C;đ?&#x2018;&#x201A;đ?&#x2018;&#x192;đ?&#x2018;&#x2021; corresponds to the minimal favorable dose which still maintains the original long-term reduction rate of both proliferating and quiescent LSCs. Therefore, as long as the TKI dose is not reduced below this threshold, our model predicts no impaired long-term efficiency, while an over-reduction compromises the overall treatment success. We conclude from equation (E5) that đ?&#x2018;&#x201C;đ?&#x2018;&#x201A;đ?&#x2018;&#x192;đ?&#x2018;&#x2021; is a patientspecific, fixed quantity determined by the proliferation rate of LSCs, đ?&#x2018;?đ?&#x2018;&#x152;, their activation rate đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152;, as well as the toxicity of the TKI, đ?&#x2018;&#x17E;=đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??źâ&#x2C6;&#x2019;đ?&#x2018;?đ?&#x2018;&#x152;. For the median parameters of the available dataset, the optimal favorable reduction fraction is

đ?&#x2018;&#x201C;đ?&#x2018;&#x201A;đ?&#x2018;&#x192;đ?&#x2018;&#x2021; =0.247, and corresponds to a long-term treatment efficiency of 98.4% compared to the standard dose. Therefore, for the â&#x20AC;&#x2DC;median patientâ&#x20AC;&#x2122; in our analysis, a reduction to 24.7% of the original dose would lead to a marginal decrease of only 1.6% in the long-term treatment efficiency, given that a minimal required plasma concentration for TKI cytotoxicity is maintained. Our model predicts transient increases in BCR-ABL1 levels of proliferating LSCs when applying different dose reductions after the first decline, i.e. once a substantial reduction in BCR-ABL1 levels had been achieved (Figure 2). However, for favorable reductions, BCR-ABL1 levels decrease again with the original long-term treatment efficiency (slope đ?&#x203A;˝) after a few months (Figure 2B and C). For the example of a â&#x20AC;&#x2DC;median patientâ&#x20AC;&#x2122;, dose reductions at month 36 of treatment will maintain MR3, while BCR-ABL1 levels are predicted to return to their original values at de-escalation after about 20 months (in case of favorable reduction with đ?&#x2018;&#x201C;=0.5) or 58 months (for the optimal favorable reduction with đ?&#x2018;&#x201C;=0.25). Importantly, the transient increase of proliferating LSCs and, therefore, of BCR-ABL1 levels in the PB, does not lead to either relevant differences in the overall response of quiescent LSCs or in the total

Figure 4. Model predictions on dose de-escalation and comparison with clinical data. (A) Comparison of DESTINY interim results with model simulations of 50% dose de-escalation applied to IRIS and CML-IV patient data (assuming the same protocol and patient selection criteria of DESTINY). We simulated dose de-escalation starting from the individually predicted remission level at the time of the last BCR-ABL1 measurement of each patient, and evaluated the fraction of patients above MR3 one year after de-escalation. Error bars indicate 90% confidence intervals. (B) Model estimates of the risk of losing MR3 within one year after de-escalation, depending on the patientâ&#x20AC;&#x2122;s individual predicted remission level just before de-escalation. Patients with remission level above MR3.5 are very likely to lose MR3 at least transiently. (C) Model simulation illustrating the transient relapse above MR3 three months after de-escalation (highlighted time interval). De-escalation of 50% was implemented for a hypothetical patient of the DESTINY trial one year after reaching MR3. The simulation of a continuing half-dose regimen predicts that after about nine months the BCR-ABL1 levels fall below MR3 and the response regains the original slope đ?&#x203A;˝. (D) Simulation results showing the predicted relative increase/decrease in the number of patients without molecular relapse two years after cessation. We use the standard treatment scenario (full-dose for one year) as the reference (corresponding to the dashed line at 0%) to compare it with: i) half-dose for one year (the DESTINY protocol; red), and ii) half-dose for two years (blue). Relapse is defined as loss of MR3.

1830

haematologica | 2018; 103(11)

TKI dose reduction in CML

LSCs population when compared to standard dose (Online Supplementary Figure S4). Our model also predicts that returning to the full dose regimen at a later point completely restores the original response levels of proliferating LSCs within a few months (Online Supplementary Figure S5).

Patient-specific optimal dose can be estimated after an initial dose reduction The calculation of patient-specific optimal TKI doses requires the knowledge of model parameters đ?&#x2018;?đ?&#x2018;&#x152;đ?&#x2018;&#x2039;, đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152;, đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??ź, and đ?&#x2018;?đ?&#x2018;&#x152;. Whereas đ?&#x2018;?đ?&#x2018;&#x152;đ?&#x2018;&#x2039;, đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152;, and đ?&#x2018;&#x17E;=đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??źâ&#x2C6;&#x2019;đ?&#x2018;?đ?&#x2018;&#x152; can be estimated from time course data, the LSC proliferation rate đ?&#x2018;?đ?&#x2018;&#x152; is confounded with the individual TKI-effect đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??ź and cannot be deduced directly. Therefore, we propose to estimate đ?&#x2018;?đ?&#x2018;&#x152; by observing the transient increase in proliferating LSCs occurring after a first favorable dose reduction. Technically, we suggest a moderate initial reduction to fraction đ?&#x2018;&#x201C; of the standard dose, after the patientâ&#x20AC;&#x2122;s response under standard dose has been sufficiently quantified in terms of the kinetic parameters (đ?&#x203A;ź,đ?&#x203A;˝,đ??´,đ??ľ) (Figure 3A). Although dose-halving (i.e. đ?&#x2018;&#x201C;=0.5) seems to be safe for this first de-escalation (see below), the proposed approach is valid for any reasonable reduction. After a transient increase in the BCR-ABL1 levels, a new intercept đ??ľâ&#x20AC;˛ can be observed approximately 18 months after the dose reduction (Figure 3B). Based on the difference of intercepts đ??ľ and đ??ľâ&#x20AC;˛, and the reduction fraction đ?&#x2018;&#x201C;, the proliferation rate đ?&#x2018;?đ?&#x2018;&#x152; can be estimated (Online Supplementary Text S9) by

and the individual optimal dose reduction fraction đ?&#x2018;&#x201C;đ?&#x2018;&#x201A;đ?&#x2018;&#x192;đ?&#x2018;&#x2021; can be calculated using equation (E5). This dose is predicted to retain the original long-term treatment efficiency (Figure 3C and D).

A population-based estimate predicts that the majority of patients in sustained remission retain the long-term treatment efficiency after dose-halving As pointed out above, the LSC proliferation rate đ?&#x2018;?đ?&#x2018;&#x152;, as an indicator of the â&#x20AC;&#x2DC;aggressivenessâ&#x20AC;&#x2122; of the untreated disease, is intrinsically unknown. In order to circumvent this limitation, we also used a population-based estimate derived from CML latency times, i.e. the time between the first leukemic transformation and diagnosis, given that no secondary events change the kinetics of disease emergence (Online Supplementary Figure S6A). Sampling from a distribution of CML latency times as reported by Radivoyevitch et al.27 [median latency time = 6.9 years, IQR (5.0,10.1)] (Online Supplementary Figure S6B) and taking into account the observed TKI response, we obtained an individual distribution of possible proliferation rates đ?&#x2018;?đ?&#x2018;&#x152; for each patient (Online Supplementary Figure S6C). In other words, we fit our mathematical model several times under different, plausible assumptions about the aggressiveness of the underlying leukemia. For each of those hypothetical but realistic scenarios we calculate the reduction level đ?&#x2018;&#x201C;đ?&#x2018;&#x201A;đ?&#x2018;&#x192;đ?&#x2018;&#x2021; (Online Supplementary Figure S6D). Based on these different possible scenarios (i.e. different leukemia growth parameters), we calculated for each patient the fraction of its individual values of đ?&#x2018;&#x201C;đ?&#x2018;&#x201A;đ?&#x2018;&#x192;đ?&#x2018;&#x2021; which are below 0.5 (Online Supplementary Figure S6F). This fraction indicates whether dose-halving appears as a suitable treatment option or not. With these estimates, our model haematologica | 2018; 103(11)

predicts that 90% of the patients in the German IRIS cohort and 81% of the patients in the CML-IV trial who once achieved MR3, could have safely decreased their TKI dose by at least 50%, while maintaining the overall therapy effect on quiescent LSCs (Online Supplementary Figure S7A). Therefore, dose-halving is expected to be safe for the majority of patients in sustained remission and might serve as the initial step to estimate the optimal individual dose. Our results also suggest that the ratio đ?&#x203A;ź/đ?&#x203A;˝ can be used to identify patients who are likely to benefit from dose reduction. We predict that patients with đ?&#x203A;ź/đ?&#x203A;˝>15 are very likely to retain the original long-term treatment efficiency after a 50% dose reduction (Online Supplementary Figure S7B and C). Furthermore, we derived a condition to identify patients who do not obtain sufficient TKI dose initially. Specifically, we found that patients with slopes đ?&#x203A;ź/đ?&#x203A;˝<2 would benefit from dose escalation, while only patients with đ?&#x203A;ź/đ?&#x203A;˝â&#x2030;Ť2 benefit from dose de-escalation (Online Supplementary Figure S8).

The model predictions are similar to results from the DESTINY trial and support the design of new, informative trials We compared our model results with findings from the DESTINY trial, which studies dose-halving in 174 TKItreated patients with CML (being either in MR4 or MR3 for at least 12 months) before TKI cessation. The published DESTINY interim-analysis indicates that 93% of the patients showed no loss of MR3 within 12 months post dose reduction.15 We simulated this scenario by predicting virtual treatment responses from BCR-ABL1 measurements in the IRIS/CML-IV trials. In particular, we identified 122 patient time courses fulfilling the inclusion criteria of the DESTINY protocol (> 3 years under TKI, > 1 year in MR3) and simulated a virtual TKI dose reduction according to the DESTINY protocol at the end of the available follow up for each of those patients. Using the same distribution of latency times as above, we calculated for each patient the fraction of values of đ?&#x2018;?đ?&#x2018;&#x152;, which lead to loss of MR3 within one year after de-escalation (Online Supplementary Figure S6E). This fraction can be interpreted as an estimate for the patient-specific risk of a molecular relapse. We also calculated the expected proportion of relapsed patients within the overall population, as well as in the corresponding subcohorts of patients being in either MR3 or MR4 within the last year before dose reduction (Figure 4A). Although a quantitative comparison should be considered with caution due to potential differences in the study populations and patient compliance, the results predicted for the IRIS/CML-IV patients show qualitatively similar relapse rates as observed in the DESTINY trial. Our findings also suggest that the individual relapse probability is related to the remission level before de-escalation, with patients below MR3.5 having a very low probability of relapse (Figure 4B). Furthermore, we predict that most of the observed relapses are transient, i.e. MR3 regain is expected when continuing the half-dose regimen (Figure 4C). Therefore, we argue that the current focus on exceeding MR3 as an indicator for a potential relapse might be reconsidered in the context of dose de-escalation strategies, while closer monitoring of the disease dynamics following dose reductions should be applied to distinguish transient from permanent BCRABL1 regrowth dynamics. 1831

A.C. Fassoni et al.

Although our model does not yet reflect immunological effects, which are proposed to be important determinants of disease dynamics post TKI-cessation, we speculated about the relative impact of dose reductions with respect to treatment stop. Specifically, we simulated the DESTINY cessation protocol for the IRIS and CML-IV patients in our data-set and evaluated the primary DESTINY end point, i.e. the proportion of patients who can de-escalate TKI dose by 50% for one year and then stop treatment completely for two years without losing MR3. Our simulations indicate higher rates of molecular relapse in comparison to a structurally identical control group receiving full-dose before TKI stop (Figure 4D) if no additional immunological control mechanisms are considered. However, comparing dose reductions of different duration, we predict a beneficial effect if patients remain at half-dose for longer before stopping TKI. We conclude that the same cumulative dose is more efficient if applied over a longer time period, thereby emphasizing that the full benefit of TKI dose de-escalation appears in the long term.

Discussion Our study supports the concept that TKI dose reduction in maintenance therapy can be a safe option for many CML patients who have already achieved a sustained remission. In particular, we dissect the typical biphasic response pattern under continuing TKI treatment and conclude that the TKI effect during the secondary treatment phase is limited by the rare activation of quiescent LSCs. Our simulations predict that the overall treatment effect is maintained for most patients, even if the TKI dose is reduced. Based on their initial treatment response under full dose, we identify patients who most likely benefit from a reduction scheme and present a strategy to estimate a patient-specific, optimal dose. Our results suggest a treatment strategy that could considerably reduce cumulative drug intake and, therefore, decrease drug-mediated side-effects. It might also increase compliance of patients to adhere to the prescribed treatment schedule. On the population level, the long overall survival times of continuously treated CML-patients add a distinct economical aspect to the outlined strategy, as the high treatment costs could be substantially reduced. The proposed strategy is not restricted to a particular TKI. Although our model encompasses all TKI pharmacokinetic parameters and doses within a single parameter đ?&#x2018;&#x2019;đ?&#x2018;&#x2021;đ??žđ??ź, we have found that this simple model, using a relative reduction compared to a standard dose of the respective TKI, is equivalent to a more elaborate formulation explicitly considering the daily TKI intake (Online Supplementary Text S10 and Online Supplementary Figure S9). Although a linear dose-response relationship appears to be an appropriate model assumption,25,26 we emphasize that the potential for dose reductions might be limited by a necessary TKI plasma concentration to ensure drug activity. Based on in vitro data for imatinib,28 we estimate this limit to be in the order of 25% of the original dose (Online Supplementary Text S11). Furthermore, recent results from clinical trials15,29 indicate that a 50% dose reduction is therapeutically active. Several clinical studies have addressed the potential of dose reductions in various settings. While Naqvi et al. 1832

report good results from a cohort of newly diagnosed CPCML patients treated with low-dose dasatinib (50 mg daily) first line,29 Russo et al. study the effect of a onemonth-on/on-month-off imatinib regimen in elderly patients after at least two years of initial full-dose therapy.30 In the later study, one-third of the patients lost their previous remission levels. Based on our model, we speculate whether the extended treatment interruptions might have added to this outcome as no therapeutically active TKI concentrations were achieved during this time. Furthermore, we have seen that several clinical studies reported a clear advantage of more potent TKI to achieve molecular response earlier as compared to standard-dose imatinib.8,31,32 However, the corresponding advantage in long-term survival is less pronounced, and we suggest that it is not the drug potency but the rare activation of LSCs that marginalizes the survival benefit. In order to test our predictions in a controlled clinical setting, we suggest an approach in which the dynamics of initial treatment response to standard therapy are sufficiently monitored to obtain reliable estimates of the relevant slopes for an informed treatment adaptation (measurements every 3-4 months within the first year and at least every six months thereafter). Only after a clinically relevant remission level (at least MR3) is reached can dose reductions (e.g. < 50% of the initial dose) be considered and these should be accompanied by a detailed follow-up monitoring of BCR-ABL1 levels to guarantee patient safety and also inform on the validity of our model approach. Further dose reductions might be considered as a second step towards approaching an optimally reduced dose that retains the therapeutic threshold. Our modeling results predict a transient increase of the number of proliferating LCs after dose reduction. However, because this is only a transient and expected effect, we suggest that the clinical criteria for molecular relapse after dose de-escalation would benefit from considering a follow-up period rather than focusing on fixed thresholds. Although the transient increase in proliferating LCs might increase the chance of acquiring secondary mutations, we reason that this is a marginal effect which needs to be compared with the benefits of reducing treatment-related side-effects. Indeed, assuming that the risk of acquiring a secondary mutation is proportional to the number of proliferating LSCs divisions, this risk increases by only 1.5% when half-dose is applied for three years (after an initial period of three years under standard dose) in comparison with the full-dose scenario (Online Supplementary Text S12 and Online Supplementary Figure S10). We acknowledge that our estimates are based on the assumptions that there are no direct dose-dependent resistance mechanisms. This is supported by the observation that re-starting TKI treatment after relapse in cessation studies proved overall successful and did not suggest a higher tendency for TKI resistances.33,34 Our current model does not consider any immunological effects or other more detailed competition mechanism between LCs and their environment. Therefore, our predictions on the risk of molecular relapse after TKI cessation are solely based on the fraction of LCs at the time of TKI stop and the proliferation rate estimated from CML latency times. This implies that a lower residual LSC number ultimately results in a lower relapse risk. Because of this limitation, our model does not reflect potentially beneficial effects resulting from mildly increased abundance haematologica | 2018; 103(11)

TKI dose reduction in CML

of proliferating LCs which potentially stimulate a patientâ&#x20AC;&#x2122;s immune response. Although TKI treatment response is still substantially heterogeneous with respect to many clinical parameters, the well-defined CML phenotype and the accessibility of kinetic data on treatment response has made CML a primary target for mathematical model approaches in oncology.4-6,22 It has also been recognized that the regulation of stem cell quiescence under continuing TKI therapy affects the kinetics of disease eradication.5,22 Here, we consider a conservative scenario, which only assumes a direct cytotoxic TKI effect on proliferating LCs, but does not alter LSCs quiescence. If assuming an additional TKI-dependent reduction of the activation rate đ?&#x2018;?đ?&#x2018;&#x2039;đ?&#x2018;&#x152;, as previously discussed,5,17,35-37 any favorable dose de-escalation would lead to an even better long-term response (Online Supplementary Figure S3). Recently, using an analytical approach very similar to our formulation, Werner et al. developed a mathematical model that allows an estimation of LSC fractions to be made from longitudinal measurements of tumor load.38 However, this model assumed no direct therapeutic effect on LSCs, which are, therefore, increasing even during treatment. In summary, we show that the systematic assessment of available clinical data by means of mathematical models has direct clinical implications, but also reveals underlying disease and treatment mechanisms. Our results are substantiated by and support the interim findings of the ongo-

References 10. 1. Pasic I, Lipton JH. Current approach to the treatment of chronic myeloid leukaemia. Leuk Res. 2017;55:65-78. 2. Rosti G, Castagnetti F, Gugliotta G, Baccarani M. Tyrosine kinase inhibitors in chronic myeloid leukaemia: which, when, for whom? Nat Rev Clin Oncol. 2017;14(3):141-154. 3. Apperley JF. Chronic myeloid leukaemia. Lancet. 2015;385(9976):1447-1459. 4. Michor F, Hughes TP, Iwasa Y, et al. Dynamics of chronic myeloid leukaemia. Nature. 2005;435(7046):1267-1270. 5. Roeder I, Horn M, Glauche I, Hochhaus A, Mueller MC, Loeffler M. Dynamic modeling of imatinib-treated chronic myeloid leukemia: functional insights and clinical implications. Nat Med. 2006;12(10):11811184. 6. Stein AM, Bottino D, Modur V, et al. BCRABL transcript dynamics support the hypothesis that leukemic stem cells are reduced during imatinib treatment. Clin Cancer Res. 2011;17(21):6812-6821. 7. Tang M, Gonen M, Quintas-Cardama A, et al. Dynamics of chronic myeloid leukemia response to long-term targeted therapy reveal treatment effects on leukemic stem cells. Blood. 2011;118(6):1622-1631. 8. Cortes JE, Saglio G, Kantarjian HM, et al. Final 5-Year Study Results of DASISION: The Dasatinib Versus Imatinib Study in Treatment-Naive Chronic Myeloid Leukemia Patients Trial. J Clin Oncol. 2016;34(20):2333-2340. 9. Saussele S, Richter J, Hochhaus A, Mahon FX. The concept of treatment-free remis-

haematologica | 2018; 103(11)

11.

12.

13.

14.

15.

16.

17.

ing DESTINY trial, thereby suggesting a change in current clinical practice and the consideration of TKI dose de-escalation strategies in maintenance therapy. As with any theoretical prediction, our results represent hypotheses that need to be validated in clinical trials. However, modeling approaches can substantially support the design of informative trials. We consider this simulation study as a proofof-concept for the use of systems medicine to optimize treatment efficacy and to minimize health care costs. Acknowledgments Our analysis is based on previously published data from the German cohort of the IRIS trial and the CML IV study. We would like to acknowledge the PIs, study groups and reference labs involved in these studies for generating and providing the data. In particular, we thank Andreas Hochhaus, RĂźdiger Hehlman and Martin Mueller. Funding The research of ACF was supported by the Excellence Initiative of the German Federal and State Governments (Dresden Junior Fellowship) and by CAPES/PĂłs-Doutorado no Exterior Grant number 88881.119037/2016-01. This work was further supported by the German Federal Ministry of Education and Research (www.bmbf.de/en/), Grant number 031A424 â&#x20AC;&#x153;HaematoOptâ&#x20AC;? to IR and Grant number 031A315 â&#x20AC;&#x153;MessAgeâ&#x20AC;? to IG. We acknowledge support from the Open Access Publication Funds of the SLUB/TU Dresden.

sion in chronic myeloid leukemia. Leukemia. 2016;30(8):1638-1647. Mahon FX. Discontinuation of TKI therapy and 'functional' cure for CML. Best Pract Res Clin Haematol. 2016;29(3):308-313. Caldemeyer L, Dugan M, Edwards J, Akard L. Long-Term Side Effects of Tyrosine Kinase Inhibitors in Chronic Myeloid Leukemia. Curr Hematol Malig Rep. 2016;11(2):71-79. Abboud C, Berman E, Cohen A, et al. 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. Hartmann JT, Haap M, Kopp HG, Lipp HP. Tyrosine kinase inhibitors - a review on pharmacology, metabolism and side effects. Curr Drug Metab. 2009;10(5):470481. Flynn KE, Atallah E. Quality of Life and Long-Term Therapy in Patients with Chronic Myeloid Leukemia. Curr Hematol Malig Rep. 2016;11(2):80-85. 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. De- Escalation and Stopping Treatment of Imatinib, Nilotinib or sprYcel in Chronic Myeloid Leukaemia. Available from: https://ClinicalTrials.gov/show/NCT01804 985. Horn M, Glauche I, Muller MC, et al. Model-based decision rules reduce the risk of molecular relapse after cessation of tyro-

18.

19.

20.

21.

22. 23.

24.

sine kinase inhibitor therapy in chronic myeloid leukemia. Blood. 2013;121(2):378384. STI571 Compared With Interferon Alfa Plus Cytarabine in Treating Patients With Newly Diagnosed Chronic Myelogenous Leukemia. Available from: h t t p s : / / C l i n i c a l Tr i a l s . g o v / s h o w / NCT00006343. Imatinib Mesylate With or Without Interferon Alfa or Cytarabine Compared With Interferon Alfa Followed by Donor Stem Cell Transplant in Treating Patients With Newly Diagnosed Chronic Phase Chronic Myelogenous Leukemia. Available from: https:// ClinicalTrials.gov/show/ NCT00055874. O'Brien SG, Guilhot F, Larson RA, et al. Imatinib compared with interferon and low-dose cytarabine for newly diagnosed chronic-phase chronic myeloid leukemia. N Engl J Med. 2003;348(11):994-1004. Hehlmann R, Lauseker M, Jung-Munkwitz S, et al. Tolerability-adapted imatinib 800 mg/d versus 400 mg/d versus 400 mg/d plus interferon-alpha in newly diagnosed chronic myeloid leukemia. J Clin Oncol. 2011;29(12):1634-1642. Komarova NL, Wodarz D. Effect of cellular quiescence on the success of targeted CML therapy. PLoS One. 2007;2(10):e990. Cortes JE, Kantarjian HM, Goldberg SL, et al. High-Dose Imatinib in Newly Diagnosed Chronic-Phase Chronic Myeloid Leukemia: High Rates of Rapid Cytogenetic and Molecular Responses. J Clin Oncol. 2009;27(28):4754-4759. Kalmanti L, Saussele S, Lauseker M, et al. Safety and efficacy of imatinib in CML over a period of 10 years: data from the random-

1833

A.C. Fassoni et al.

25.

26.

27.

28. 29.

1834

ized CML-study IV. Leukemia. 2015;29 (5):1123-1132. Deininger MW, O'Brien SG, Ford JM, Druker BJ. Practical management of patients with chronic myeloid leukemia receiving imatinib. J Clin Oncol. 2003;21 (8):1637-1647. La Rosee P, Johnson K, Corbin AS, et al. In vitro efficacy of combined treatment depends on the underlying mechanism of resistance in imatinib-resistant Bcr-Abl-positive cell lines. Blood. 2004;103(1):208-215. Radivoyevitch T, Hlatky L, Landaw J, Sachs RK. Quantitative modeling of chronic myeloid leukemia: insights from radiobiology. Blood. 2012;119(19):4363-4371. Peng B, Lloyd P, Schran H. Clinical pharmacokinetics of imatinib. Clin Pharmacokinet. 2005;44(9):879-894. Naqvi K, Jabbour E, Skinner J, et al. Early results of lower dose dasatinib (50 mg daily) as frontline therapy for newly diagnosed chronic-phase chronic myeloid leukemia. Cancer. 2018;35(15):e18551.

30. Russo D, Martinelli G, Malagola M, et al. Effects and outcome of a policy of intermittent imatinib treatment in elderly patients with chronic myeloid leukemia. Blood. 2013;121(26):5138-5144. 31. 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. 32. Hochhaus A, Saglio G, Hughes TP, et al. Long-term benefits and risks of frontline nilotinib vs imatinib for chronic myeloid leukemia in chronic phase: 5-year update of the randomized ENESTnd trial. Leukemia. 2016;30(5):1044-1054. 33. Mahon FX, Rea D, Guilhot J, et al. Discontinuation of imatinib in patients with chronic myeloid leukaemia who have maintained complete molecular remission for at least 2 years: the prospective, multicentre Stop Imatinib (STIM) trial. Lancet Oncol. 2010;11(11):1029-1035.

34. 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. 35. Graham SM, Jorgensen HG, Allan E, et al. Primitive, quiescent, Philadelphia-positive stem cells from patients with chronic myeloid leukemia are insensitive to STI571 in vitro. Blood. 2002;99(1):319-325. 36. Glauche I, Horn M, Roeder I. Leukaemia stem cells: hit or miss? Br J Cancer. 2007;96(4):677-678. 37. Besse A, Lepoutre T, Bernard S. Long-term treatment effects in chronic myeloid leukemia. J. Math. Biol. 2017;75(3):733-758. 38. Werner B, Scott JG, Sottoriva A, Anderson AR, Traulsen A, Altrock PM. The Cancer Stem Cell Fraction in Hierarchically Organized Tumors Can Be Estimated Using Mathematical Modeling and PatientSpecific Treatment Trajectories. Cancer Res. 2016;76(7):1705-1713.

haematologica | 2018; 103(11)

ARTICLE

Chronic Myeloid Leukemia

Treatment-free remission after two-year consolidation therapy with nilotinib in patients with chronic myeloid leukemia: STAT2 trial in Japan

Naoto Takahashi,1 Kaichi Nishiwaki,2 Chiaki Nakaseko,3,4 Nobuyuki Aotsuka,5 Koji Sano,2 Chikako Ohwada,4 Jun Kuroki,6 Hideo Kimura,7 Michihide Tokuhira,8 Kinuko Mitani,9 Kazuhisa Fujikawa,10 Osamu Iwase,11 Kohshi Ohishi,12 Fumihiko Kimura,13 Tetsuya Fukuda,14,15 Sakae Tanosaki,16 Saori Takahashi,17 Yoshihiro Kameoka,17 Hiroyoshi Nishikawa,18 Hisashi Wakita5,19 and the STAT study group

Department of Hematology, Nephrology and Rheumatology, Akita University Graduate School of Medicine; 2Department of Oncology and Hematology, Jikei University Kashiwa Hospital; 3 Department of Hematology, International University of Health and Welfare School of Medicine, Narita; 4Department of Hematology, Chiba University Hospital; 5 Department of Hematology and Oncology, Japanese Red Cross Narita Hospital; 6 Department of Internal Medicine, Yuri General Hospital, Yurihonjo; 7Department of Hematology, Northern Fukushima Medical Center, Date; 8Department of Hematology, Saitama Medical Center, Saitama Medical University, Kawagoe; 9Department of Hematology and Oncology, Dokkyo Medical University, Tochigi; 10Department of Hematology, Chibaken Saiseikai Narashino Hospital; 11Department of Hematology, Tokyo Medical University Hachioji Medical Center; 12Transfusion Medicine and Cell Therapy, Mie University Hospital, Tsu; 13Division of Hematology, National Defense Medical College, Tokorozawa; 14Department of Hematology, Tokyo Medical and Dental University Hospital; 15Department of Hematology, Tottori University Hospital, Yonago; 16Department of Hematology, The Fraternity Memorial Hospital, Tokyo; 17Clinical Research Promotion and Support Center, Akita University Hospital; 18Division of Cancer Immunology, Research Institute / Exploratory Oncology Research & Clinical Trial Center, National Cancer Center, Tokyo/Kashiwa and 19Japanese Red Cross Chiba Blood Center, Funabashi, Japan

Ferrata Storti Foundation

Haematologica 2018 Volume 103(11):1835-1842

Correspondence: ABSTRACT

he purpose of this trial was to evaluate the efficacy of 2-year consolidation therapy with nilotinib, at a dose of 300 mg twice daily, for achieving treatment-free remission in chronic myeloid leukemia patients with a deep molecular response (BCR-ABL1IS ≤0.0032%). Successful treatment-free remission was defined as no confirmed loss of deep molecular response. We recruited 96 Japanese patients, of whom 78 sustained a deep molecular response during the consolidation phase and were therefore eligible to discontinue nilotinib in the treatment-free remission phase; of these, 53 patients (67.9%; 95% confidence interval: 56.4–78.1%) remained free from molecular recurrence in the first 12 months. The estimated 3-year treatment-free survival was 62.8%. Nilotinib was readministered to all patients (n=29) who experienced a molecular recurrence during the treatment-free remission phase. After restarting treatment, rapid deep molecular response returned in 25 patients (86.2%), with 50% of patients achieving a deep molecular response within 3.5 months. Tyrosine kinase inhibitor withdrawal syndrome was reported in 11/78 patients during the early treatment-free remission phase. The treatment-free survival curve was significantly better in patients with undetectable molecular residual disease than in patients without (3-year treatment-free survival, 75.6 versus 48.6%, respectively; P=0.0126 by the log-rank test). There were no significant differences in treatment-free survival between subgroups based on tyrosine kinase inhibitor treatment before the nilotinib consolidation phase, tyrosine kinase inhibitorwithdrawal syndrome, or absolute number of natural killer cells. The haematologica | 2018; 103(11)

naotot@doc.med.akita-u.ac.jp

Received: April 4, 2018. Accepted: June 28, 2018. Pre-published: July 5, 2018. doi:10.3324/haematol.2018.194894 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1835 ©2018 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.

1835

N. Takahaski et al.

results of this study indicate that it is safe and feasible to stop tyrosine kinase inhibitor therapy in patients with chronic myeloid leukemia who have achieved a sustained deep molecular response with 2 years of treatment with nilotinib. This study was registered with UMIN-CTR (UMIN000005904).

Introduction Nilotinib is a second-generation tyrosine kinase inhibitor (TKI) that has been shown to be highly efficacious as a first- or second-line treatment for patients with Philadelphia chromosome-positive chronic myeloid leukemia (CML) in chronic phase. In patients newly diagnosed with CML in chronic phase superior rates of deep molecular response (DMR) were achieved with nilotinib in comparison with imatinib, which is a first-generation TKI currently used as the standard treatment for this disease.1-3 In addition, switching to nilotinib after a minimum of 2 years on imatinib led to increased DMR rates compared to remaining on imatinib.4 Recently, treatment-free remission (TFR) has been proposed as a goal for CML treatment.5-7 Indeed, prospective trials have indicated that imatinib therapy can be successfully discontinued in CML patients who have maintained a DMR for at least 2 years.8-10 In these prospective trials, the TFR rate was 43% [95% confidence interval (CI): 33–52%] at 6 months9 and 41% (95% CI: 29–52%) at 12 months in the STIM1 trial,8 while the TWISTER study revealed a TFR rate of 47.1% (95% CI: 31.5–62.7%) at 24 months.10 Moreover, the first TFR study of second-generation TKI, the DADI trial reported by Imagawa et al., showed that second-generation TKI therapy can be successfully discontinued.11 In this trial, all patients received dasatinib consolidation therapy for at least 1 year. The estimated TFR rate was 49% (95% CI: 36–61%) at 6 months.11 On the other hand, the ENESTfreedom study, which is a TFR study following frontline nilotinib treatment, required that all patients sustained DMR during the consolidation phase with nilotinib for 1 year. The TFR rate at 48 weeks was 51.6% (95% CI: 44.2–58.9%).12 Although the DMR in the consolidation phase with a second-generation TKI was sustained in both the DADI trial and the ENESTfreedom study, the TFR rate was not superior to those in the previously reported imatinib TFR studies.8-10 Most relapses occurred within 6 months of discontinuing second-generation TKI or imatinib therapy, and there was no disease progression in patients with molecular relapse after discontinuation.8-12 All patients who relapsed remained sensitive to TKI re-treatment in these TFR studies.8-12 Compared to imatinib, nilotinib may enable a greater proportion of patients with CML in chronic phase to achieve successful TFR if they receive nilotinib consolidation therapy for 2 years to sustain DMR; this is the same length of time required to achieve TFR in imatinib studies.8,9 The aim of this STAT2 trial (Stop Tasigna® Trial) was to evaluate the efficacy of 2-year consolidation treatment with nilotinib for achieving successful TFR in patients with chronic phase CML.

chronic phase, age ≥16 years, an Eastern Cooperative Oncology Group performance status of 0–2, and no severe primary organ dysfunction. Patients who had accelerated phase or blast crisis CML, a T315I mutation, or who had received allogeneic hematopoietic stem-cell transplantation were excluded from this study. Patients with a DMR (BCR-ABL1IS ≤0.0032% or a molecular response, MR4.5, defined as a 4.5-log reduction in BCR-ABL1 transcripts according to the international scale], assessed by realtime quantitative polymerase chain reaction (RQ-PCR), under treatment with imatinib or a second-generation TKI following imatinib were eligible for the STAT2 trial. Nilotinib (300 mg) was administered twice daily (600 mg/day) for 2 years in the consolidation phase. Patients who maintained a MR4.5 during the 2-year consolidation phase were eligible to enter the TFR phase and cease nilotinib treatment. Molecular recurrence was defined as the loss of a major molecular response (MMR: BCR-ABL1IS ≤0.1%) or confirmed loss of MR4.5 (at two consecutive assessments within 4 weeks) after discontinuing nilotinib, based on criteria used both in the STIM1 trial8 and the TWISTER study.9 Patients with molecular recurrence during the TFR phase restarted nilotinib 300 mg twice daily, thus entering the re-treatment phase.

Endpoints and assessments The primary endpoint of the STAT2 trial was the 12-month TFR rate after discontinuing nilotinib treatment; secondary endpoints were the 24-month TFR rate after discontinuing nilotinib treatment, the 3-year treatment-free survival, and the MR4.5 rate and time to MR4.5 achieved by nilotinib in the re-treatment phase. Safety profiles, especially vascular adverse events in the consolidation phase or symptoms related to TKI withdrawal syndrome in the TFR phase, were evaluated. Adverse events were assessed according to the Common Terminology Criteria for Adverse Events (CTCAE) version 4.03. MR was evaluated by BCR-ABL1IS RQ-PCR analysis upon study entry and every 3 months thereafter in the consolidation phase. After discontinuing nilotinib in the TFR phase, molecular recurrence was monitored by monthly BCR-ABL1IS RQ-PCR testing in the first year, bi-monthly testing in the second year, then every 3 months thereafter. In the re-treatment phase, BCR-ABL1IS was monitored by monthly RQ-PCR testing. The study protocol was terminated when MR4.5 was re-achieved, or when BCR-ABL1IS increased twice consecutively in the re-treatment phase. BCR-ABL1IS RQ-PCR was performed using a Molecular MD One-Step qRT-PCR BCR-ABL kit (BML Inc., Kawagoe, Japan). To validate BCR-ABL1 amplification, ABL1 was used as an internal control. A MMR was defined as a 3-log reduction in the BCRABL1 transcript according to the international scale (BCR-ABL1IS ≤0.1%), MR4.5 was defined as a 4.5-log reduction in the BCR-ABL1 transcript (BCR-ABL1IS ≤0.0032%), and MR5 was defined as a 5log reduction in the BCR-ABL1 transcript (BCR-ABL1IS ≤0.001%), as described above. Undetectable molecular residual disease was defined as undetectable BCR-ABL1 transcript with MR5 (UMRD with MR5). At least 100,000 control genes (ABL1) were required for a sample to be considered as adequate.

Methods Ethics Patients and study design The eligibility criteria for this multicenter, phase II, single-treatment arm, open-label clinical trial included: patients with CML in 1836

Forty-six institutions participated in this study. The study was conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from each parhaematologica | 2018; 103(11)

TFR after 2-year nilotinib consolidation

ticipant before enrollment. The study was approved by the Ethics Committee of Akita University (N. 786) and by all institutional ethic committees that participated in this study. The study was registered with UMIN-CTR (UMIN000005904).

Results Patients and treatment Between July 2011 and December 2012, 96 patients who achieved MR4.5 were enrolled in the STAT2 trial. These patients started treatment in the consolidation phase and were defined as the safety analysis set. Seventy-eight patients entered the TFR phase and were analyzed as the full analysis set for TFR. The baseline demographics of the safety and full analysis sets are shown in Table 1. The median age in the safety analysis set was 55.5 years (range, 20–83). Thirty-six (37.5%) patients were female and 16 (16.7%) patients had been administered interferona before TKI treatment. Based on TKI therapy prior to entry into the current study, patients were classified into three groups: 50 patients (52.1%) who had received only imatinib (‘imatinib only’), 40 patients (41.7%) who had received nilotinib following imatinib (‘nilotinib following imatinib’, including patients from the STAT1 study), and six patients (6.3%) who had received other therapy Table 1. Baseline demographics of all patients included in the study.

SAS (n=96) before entering the consolidation phase

FAS (n=78) before entering the TFR phase

55.5 (20–83)

57.0 (22–85)

60 (62.5) 36 (37.5)

45 (57.7) 33 (42.3)

56 (59.6) 20 (21.3) 18 (19.1) 16 (16.7)

44 (57.1) 17 (22.1) 16 (20.8) 12 (15.4)

50 (52.1) 40 (41.7) 6 (6.3) 75.6 (8–131)

40 (51.3) 33 (42.3) 5 (6.4) 99.0 (25–156)

73.6 (11–126) 7.7 (0–35) 5.6 (1–63) 14.5 (3–103) 48.1 (6–128)

74.9 (11–125) 25.3 (23–60) 5.5 (1–63) 14.2 (3–103) 47.9 (6–128)

70 (73.7) 25 (26.3)

58 (75.3) 19 (25.7)

96 (100) 29 (30.2)

78 (100) 41 (52.6)

Age (years) Sex Male Female Sokal risk Low Intermediate High Prior interferon-a Prior TKI before the trial Only imatinib Nilotinib following imatinib Other Total duration of TKI treatment (months) Duration of imatinib (months) Duration of nilotinib (months) Time to CCyR (months) Time to MMR (months) Time to MR4.5 (months) Treatment at achieving MR4.5 Imatinib Nilotinib Molecular status MR4.5 UMRD with MR5

Data are given as median (range) or n (%). SAS: safety analysis set; FAS: full analysis set; TFR: treatment free remission; TKI: tyrosine kinase inhibitor; CCyR: complete cytogenetic response; MMR: major molecular response; MR4.5: 4.5-log reduction of BCRABL1 transcripts by IS-PCR; UMRD: undetectable BCR-ABL1 transcript; MR5: 5-log reduction of BCR-ABL1 transcripts by IS-PCR.

haematologica | 2018; 103(11)

(‘other’). The STAT1 study (Switch to Tasigna® Trial) is a clinical trial to evaluate the efficacy of 2-year consolidation treatment with nilotinib for achieving DMR in chronic phase CML patients with MMR, recently reported by our study group.13 Among 40 patients in STAT2 treated by nilotinib following imatinib, 21 patients joined this study from STAT1 since they achieved MR4.5 in the STAT1 study. The reasons for switching from imatinib to nilotinib in the 40 patients included imatinib resistance in 7.5%, imatinib intolerance in 20.0%, and upon patients’ request in 72.5% (Online Supplementary Table S1). The median duration of imatinib or nilotinib treatment was 73.6 months (range, 11–126) or 7.7 months (range, 0–35), respectively. All patients showed MR4.5 at the time of entry into the study, and the median time to MR4.5 on TKI therapy was 48.1 months (range, 6–128). Among 96 patients in the safety analysis set, 70 (73.7%) achieved the MR4.5 by prior imatinib treatment and 25 (26.3%) by prior nilotinib treatment. Comparing patients in different subgroups based on prior TKI therapy (‘imatinib only’ versus ‘nilotinib following imatinib’), there were no significant differences except in the duration of imatinib therapy and time to MR4.5 (Online Supplementary Table S1). Of these, 40 patients in the ‘imatinib only’ group, 33 patients in the ‘nilotinib following imatinib’ group, and five patients in the ‘other’ group entered the TFR phase. Nilotinib was taken twice daily (600 mg/day) for 2 years in the consolidation phase (median dose intensity, 600 mg/day). Patients’ outcomes at the end of the consolidation phase at 24 months are summarized in Table 2. Among the 96 patients in the safety analysis set, 18 (18.8%) discontinued the study treatment. The most frequent reason for discontinuation was adverse events; disease progression was not observed in any of the patients.

Treatment-free remission after nilotinib discontinuation Among the 96 patients in the safety analysis set, 78 with sustained MR4.5 during the consolidation phase were eligible for nilotinib discontinuation in the TFR phase. The median follow-up of patients in the TFR phase was 35.4 months (range, 1.8–44.2). Among the 78 patients, 53 remained in the TFR phase without a confirmed loss of MR4.5 in the first 12 months; the 12-month TFR primary endpoint was 67.9% (95% CI: 56.4–78.1%), exceeding the targeted 40% success rate. The 24-month TFR rate was 62.8% (95% CI: 51.1–73.5%). The Kaplan-Meier curve for

Table 2. Patients’ outcomes and dose intensity at the end of the 2-year nilotinib consolidation phase.

SAS (n=96) Dose reduction Interruption Treatment discontinuation Adverse events Confirmed loss of MR4.5 Withdrawal of consent Lost to follow-up Other Duration of nilotinib (days) Dose intensity of nilotinib (mg/day)

14 (14.6) 13 (13.5) 18 (18.8) 6 (6.3) 3 (3.1) 4 (4.2) 1 (1.5) 6 (6.3) 737.5 (14–859) 600 (204.6–674.5)

Data are given as n (%) or median (range). SAS, safety analysis set; MR4.5, 4.5-log reduction of BCR-ABL1 transcripts by IS-PCR.

1837

N. Takahaski et al.

treatment-free survival is shown in Figure 1A. The estimated 3-year treatment-free survival was 62.8%. Among patients with a confirmed loss of MR4.5, 25 patients lost the MR4.5 within the first 6 months after discontinuing nilotinib (median, 3.4 months; range, 1.8–5.8 months), and the remaining four patients lost the MR4.5 between 16 and 20 months within the TFR phase (Figure 1B). In a subanalysis of groups based on prior TKI before entry into the STAT2 trial, the 12-month TFR rate was 62.5% (95% CI: 45.8–77.3%) in the ‘imatinib only’ group and 69.7% (95% CI: 51.3–84.4%) in the ‘nilotinib following imatinib’ group. In another subanalysis based on positivity of molecular residual disease (MRD) before the TFR phase, the 12-month TFR rate was 56.8% (95% CI: 39.5–

72.9%) in the MRD group and 78.0% (95% CI: 62.4– 89.4%) in the UMRD with MR5 group; the corresponding 24-month TFR rates were 48.6% (95% CI: 31.9–65.6%) and 75.6% (95% CI: 59.7–87.6%). An analysis of baseline factors as predictors of TFR at 12 months was conducted. With the exception of detectable MRD before entering the TFR phase (odds ratio, 0.369; 95% CI: 0.138–0.988; P=0.0473), there were no significant predictors in the univariate logistic regression analysis, including the absolute number of natural killer cells and natural killer cell cytotoxicity (Table 3). Multivariate analysis was not, therefore, performed. On the other hand, the treatment-free survival curve was significantly better in the UMRD with MR5 group than in the MRD group (estimated

A Figure 1. Treatment-free remission after 2-year consolidation with nilotinib. (A) Kaplan-Meier estimates of treatment-free survival after discontinuation of nilotinib (n=78). (B) The kinetics of BCR-ABL1 transcripts in the treatment-free remission (TFR) phase. Twenty-five patients lost MR4.5 within 12 months and eight patients showed fluctuations in the amounts of BCR-ABL1 transcript around the MR4.5 level. Among the eight patients with fluctuations, four lost MR4.5 after 16, 16, 17, and 20 months. (C) Kaplan-Meier estimates of treatment-free survival after discontinuation of nilotinib according to the molecular response [molecular residual disease (MRD) positive or undetectable MRD (UMRD with MR5)] at enrollment in the TFR phase. UMRD with MR5 was defined as undetectable BCR-ABL1 transcripts by IS-PCR in which at least 100,000 control genes (ABL1) were required for the sensitivity of MR5. (D) Cumulative incidence of MR4.5 in patients who lost MR4.5 during the TFR phase and were subsequently readministered nilotinib (n=29). (E) Cumulative incidence of reacquisition of major molecular response (MMR) in patients who lost MMR and were readministered nilotinib (n=16).

1838

haematologica | 2018; 103(11)

TFR after 2-year nilotinib consolidation

3-year treatment-free survival, 75.6 versus 48.6%; P=0.0126 by the log-rank test) (Figure 1C). There were no significant differences in the treatment-free survival curves between subgroups based on TKI treatment prior to the consolidation phase by nilotinib (‘imatinib only’ versus ‘nilotinib following imatinib’ group, P=0.9508 by the log-rank test), presence of TKI withdrawal syndrome (P=0.4096 by the log-rank test), or absolute number of natural killer cells (≥ median versus < median, P=0.4527 by the log-rank test).

Response to nilotinib treatment re-initiation Nilotinib was readministered to all 29 patients with a molecular recurrence during the TFR phase. After recommencing treatment, MR4.5 rapidly returned in 25 patients (86.2%), with 50% of patients achieving MR4.5 within 3.5 months (Figure 1D). Four out of 29 patients discontinued treatment: one patient discontinued at 1 month because of a nilotinib-associated rash, one at 2.5 months following the patient’s request, and two at 10 and 11 months because of increasing BCR-ABL1IS. Despite discontinuation of the study, all patients, including the two patients with increasing BCR-ABL1IS during the re-treatment phase, achieved DMR. The clinical course details of these four patients are shown in Online Supplementary Figure S1. Among 29 patients with a molecular recurrence, 16 had lost their

MMR at the first or second assessment of BCR-ABL1IS using RQ-PCR during the TFR phase. After starting treatment again, a MMR rapidly returned in all 16 patients (100%) in 3 months and 50% of patients achieved the MMR within 2 months (Figure 1E).

Safety, vascular adverse events, and tyrosine kinase inhibitor withdrawal syndrome No patients progressed to accelerated phase or blast crisis CML, or died during this study. Adverse events (all grades) were reported in 55 patients (57.3%) in the safety analysis set in the consolidation phase, and 30 patients (38.7%) in the full analysis set in the TFR phase; the incidence of grade 3/4 adverse events was 14.6%, and 2.6% in the safety analysis set and full analysis set, respectively (Online Supplementary Table S2). Vascular adverse events of any grade were reported in six patients (6.2%) during the nilotinib consolidation phase (Table 4). Ischemic heart disease (acute coronary syndrome or angina pectoris) was reported in three patients and cerebral infarctions in three patients, but peripheral arterial occlusive disease was not reported in any patient. Among the six patients with vascular adverse events, four had at least one traditional risk factor for such events (e.g., hypertension, hyperlipidemia, diabetes mellitus, smoking, or

Table 3. Univariate analysis of predictive factors for treatment-free remission at 12 months. Age (years) Height (cm) Body weight (kg) Comorbidity Platelet count Blast cells in PB (%) Eosinophils in PB (%) Basophils in PB (%) Spleen size (cm) Sokal risk (low) Hasford (low) EUTOS (low) Prior interferon-a Duration of imatinib (days) Time to CCyR (days) Time to MMR (days) Time to MR4.5 (days) MRD positive Total nilotinib dose (mg) Duration of nilotinib (days) Nilotinib dose intensity (mg/day) T cells (CD3+CD8+) T-LGL (CD3+CD57+) NK cells (CD3-CD56+) NK cells (CD16+CD56+) NK cell activity E/T ratio 10:1 NK cell activity E/T ratio 20:1

Coefficient

P-value

95% CI

0.0168 -0.0387 -0.0122 0.3325 0.0083 -0.1340 -0.0803 0.0343 -0.1052 -0.5124 -0.2131 -0.1178 0.4283 0.0000 -0.0005 0.0001 0.0126 -0.9966 0.0000 0.0066 -0.0056 -0.0335 -0.0221 -0.0010 -0.0191 -0.0220 -0.0112

0.0190 0.0265 0.0226 0.5125 0.0055 0.1683 0.0680 0.0466 0.0839 0.5933 0.5112 0.6687 0.7163 0.0002 0.0009 0.0004 0.0082 0.5025 0.0000 0.0097 0.0051 0.0338 0.0306 0.0013 0.0265 0.0208 0.0150

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

0.3769 0.1447 0.5876 0.5165 0.1322 0.4259 0.2379 0.4612 0.2099 0.3878 0.6768 0.8602 0.5499 0.8169 0.5240 0.6993 0.1227 0.0473 0.3515 0.4956 0.2713 0.3207 0.4714 0.4305 0.4723 0.2915 0.4527

1.017 0.962 0.988 1.395 1.008 0.875 0.923 1.035 0.900 0.599 0.808 0.889 1.535 1.000 0.999 1.000 1.013 0.369 1.000 1.007 0.994 0.967 0.978 0.999 0.981 0.978 0.989

0.980–1.055 0.913–1.013 0.945–1.033 0.511–3.808 0.997–1.019 0.629–1.216 0.808–1.054 0.945–1.134 0.764–1.061 0.187–1.916 0.297–2.201 0.240–3.297 0.377–6.248 1.000–1.000 0.998–1.001 0.999–1.001 0.997–1.029 0.138-0.998 1.000–1.000 0.988–1.026 0.984–1.004 0.905–1.033 0.921–1.039 0.997–1.001 0.931–1.033 0.939–1.019 0.960–1.018

SE: standard error; OR: odds ratio; 95% CI: 95% confidence interval; PB: peripheral blood; CCyR: complete cytogenetic response; MMR: major molecular response; MR4.5: 4.5-log reduction of BCR-ABL1 transcripts by IS-PCR; MRD: molecular residual disease; T-LGL: T-cell large granular lymphocytes; NK: natural killer; E/T: effector-target cell.

haematologica | 2018; 103(11)

1839

N. Takahaski et al.

chronic kidney disease). Percutaneous intervention was performed in three patients with ischemic heart disease. All patients, except one, with vascular adverse events recovered or improved. Although three patients with vascular adverse events stopped nilotinib treatment and switched to another TKI, three patients continued the study treatment and entered the TFR phase, eventually achieving successful TFR. No patients developed new vascular adverse events during the TFR phase. Arthralgia was reported only in the TFR phase (Online Supplementary Table S2). Eleven patients reported musculoskeletal pain events during the early phase of the TFR; these events were categorized as TKI withdrawal syndrome. The characteristics of the patients with TKI withdrawal syndrome are described in Table 5. The median time of onset of the TKI withdrawal syndrome in the TFR phase was 1 month (range, 0â&#x20AC;&#x201C;6). All patients recovered completely with or without treatment. Of the 11 patients with the syndrome, eight (73%) maintained TFR at 12 months and remained in remission throughout the 36month follow-up period. There were no significant differences in the TFR survival curves between subgroups based on TKI withdrawal syndrome.

Discussion The design of the STAT2 trial resembled that of the STIM18,9 and TWISTER trials,10 with the aim of administering a TKI during a 2-year consolidation phase to obtain a sustained DMR before TKI discontinuation. In this trial, in contrast to the aforementioned studies, imatinib was replaced with nilotinib as the consolidation TKI therapy. However, the assessment of BCR-ABL1 and the definition of molecular recurrence in this trial are identical to those in the aforementioned studies.8-10 Nilotinib was administered instead of imatinib because previous studies had indicated that nilotinib could induce DMR in a greater number of patients than imatinib,1-3 thereby potentially increasing the number of patients who achieve successful TFR. The 2-year period of consolidation therapy in this study was selected as the STIM18,9 and TWISTER10 trials required 2 years of sustained DMR before stopping imatinib therapy. Additionally, our retrospective study revealed that a DMR duration of at least 2 years was a significant predictive factor for successful TFR in Japanese CML patients.14 Therefore, the treatment duration for this trial involved a consolidation phase of 2 years, and sus-

Table 4. Treatment-related vascular adverse events in the consolidation phase.

Age (decade)

Sex (M/F)

Duration of imatinib (months)

Duration of nilotinib (months)

Traditional risk factors

Vascular adverse events

CTCAE grade

Outcome

Treatment outcome

40s

99.4

22.6

Hypertension

Improvement

ACS AP CI

3 3 2

Improvement by PCI Improvement by PCI Recovery

CI ACS

2 3

No change Recovery by PCI

Nilotinib discontinuation, switched TKI Entered TFR phase Entered TFR phase Nilotinib discontinuation, switched TKI Entered TFR phase Nilotinib discontinuation, switched TKI

70s 70s 40s

F F F

41.5 87.2 70.3

22.4 52.6 10.5

None Hypertension None

80s 70s

M M

45.7 124.6

24.0 17.9

Hypertension DM, CKD

M: male; F: female; CTCAE: Common Terminology Criteria for Adverse Events; DM: diabetes mellitus; CKD: chronic kidney disease; CI: cerebral infarction; AP: angina pectoris; ACS: acute coronary syndrome; PCI: percutaneous intervention; TFR: treatment-free remission; TRI: tyrosine kinase inhibitor.

Table 5. Tyrosine kinase inhibitor withdrawal syndrome in the treatment-free remission phase.

Age Sex (decade) (M/F) 50s 50s 50s 30s 60s 30s 50s 50s 40s 50s 60s

F F F F F M F F M M F

Duration Muscle Time CTCAE of TKI pain of WS grade (months) before DC (months) 103 107 124 51 94 120 140 139 99 135 88

Yes No No No No Yes No No No No No

0 1 1 1 1 1 2 3 3 5 6

1 2 2 1 3 1 2 1 1 2 2

Localization of symptoms

WS therapy

Duration of WS

TFR at 12 months (months)

Myalgia, whole body Arms Hands, wrists, arms, shoulders, legs Hands, elbows Hands, feet Hands, wrists, shoulders, lower back, legs Shoulders, legs, knees Hands, wrists, elbows, arms Hands, wrists, elbows, arms Hands Hands, wrists, elbows, lower back

None NSAID NSAID NSAID Prednisolone None Prednisolone None None Prednisolone NSAID

5 10 5 7 19 13 26 17 14 15 11

Yes Yes Yes No No Yes Yes No Yes Yes Yes

M: male; F: female; TKI: tyrosine kinase inhibitor; DC: discontinuation of nilotinib; WS: withdrawal syndrome; CTCAE: Common Terminology Criteria for Adverse Events (for the highest grade in the TFR phase); TFR: treatment-free remission; NSAID: non-steroidal anti-inflammatory drugs.

1840

haematologica | 2018; 103(11)

TFR after 2-year nilotinib consolidation

tained DMR without loss of MR4.5 was confirmed by regular BCR-ABL1IS RQ-PCR testing for 2 years. Among 78 patients who entered the TFR phase, 53 remained in TFR in the first 12 months; the 12-month TFR primary endpoint was, therefore, 67.9% (95% CI: 56.4â&#x20AC;&#x201C; 78.1%). Although the TFR rate is higher than that in the STIM1 trial (41%; 95% CI: 29â&#x20AC;&#x201C;52%), it is difficult to compare this result with those in other TFR studies with varying designs. Among 29 patients with a molecular recurrence, MR4.5 was rapidly regained in 25 patients who were readministered nilotinib, and no patients progressed to accelerated phase or blast crisis CML in this study. Thus, our findings suggest that nilotinib therapy may allow the majority of patients to achieve successful TFR after discontinuation of nilotinib; this result is comparable to those of previous TFR studies with imatinib.8-10 In this study, the identified predictive factor for successful 12-month TFR was UMRD with MR5 before discontinuation of nilotinib. In the EURO-SKI trial, there were no differences in MMR status at 6 months after treatment stop between depths of molecular response (MR4.5 versus no MR4.5).15 However, our finding suggests that a deeper MR favors successful achievement of TFR in CML patients, which is consistent with data from previous studies.16,17 Among four patients without TFR in the late TFR phase, in whom MR4.5 loss occurred at 16, 16, 17, and 20 months, three patients had MRD before entering the TFR phase. It is difficult to determine the quantity of BCR-ABL1 mRNA below MR4.5/MR5 because of the sensitivity of IS-RQPCR.18 However, UMRD with MR5 before TFR is one of the minimum requirements for TFR and the sustained duration of DMR might be a surrogate marker for the magnitude of the MR or the eradication of MRD during TKI treatment.19,20 Meanwhile, 1-year consolidation with either dasatinib or nilotinib, which are both second-generation TKI, was proposed in the DADI trial11 and the ENEStop study,21 respectively. Although the 1-year consolidation enabled identification of enrolled patients who had sustained MR4.5 and were eligible to stop treatment,21 it is still unknown whether 1 year of consolidation with a secondgeneration TKI is sufficient to achieve DMR/UMRD below MR4.5/MR5 for TFR. Of the 96 patients in the safety analysis set, 78 achieved a sustained MR4.5 on nilotinib consolidation and entered the TFR phase, including 33 patients (42.3%) who were treated with nilotinib following imatinib prior to enrollment in STAT2. Most patients (n=29, 72.5%) were switched from imatinib to nilotinib at the patientsâ&#x20AC;&#x2122; request prior to the trial, primarily to obtain a deeper and more sustained MR, despite not having imatinib resistance or intolerance. Similarly, the ENESTcmr study showed that a significant minority of patients who continued with imatinib therapy did later achieve DMR.4 In the subanalysis evaluating prior TKI exposure, the 12-month TFR was almost identical between patients treated with only imatinib (62.5%) and those treated with nilotinib following imatinib (69.7%). This suggests that, regardless of the specific TKI, DMR is the first step for achieving successful TFR in patients with CML. The proportion of natural killer cells in peripheral blood has been reported as a predictive marker, which might be related to a previously reported immuno-oncological effect.22,23 However, it was beyond the remit of this trial to

haematologica | 2018; 103(11)

identify the significance of either the activity or the proportion or natural killer cells in peripheral blood. TKI withdrawal syndrome is the most common musculoskeletal pain-related adverse event in imatinib TFR studies, being first reported in the EURO-SKI trial.24 TKI withdrawal syndrome was detected in 11 patients in the TFR phase in this study. Rousselot et al. suggested that prolonged inhibition of c-Kit signaling by imatinib may modulate nociceptive sensitivity, and that the sudden discontinuation of imatinib may reverse this phenomenon.25 As nilotinib targets the same tyrosine kinases as imatinib, including BCR-ABL kinase and c-Kit, albeit with differing potencies, nilotinib may also result in TKI withdrawal syndrome via the same mechanisms. Although TKI withdrawal syndrome was reported as an independent predictive factor for successful TFR by a Korean group,26 there was no significant relationship between TKI withdrawal syndrome and TFR identified in this nilotinib TFR study. However, because of the limited number of events during the TFR phase, univariate analysis is definitely limited in identifying a significant relationship between TKI withdrawal syndrome and TFR. Further examination with larger numbers of patients will be necessary to identify biomarkers for successful TFR. Although the safety of nilotinib is generally regarded as being acceptable, vascular adverse events are an important concern in nilotinib therapy. The frequency of such events in this study was similar to the frequency in the nilotinib 300 mg twice daily arm in the ENESTnd trial.3 The incidence of vascular adverse events in patients treated with nilotinib 300 mg twice daily was estimated to be 2.8 per 100 patient-years in a meta-analysis.27 In the STAT2 trial, all patients with vascular adverse events, except one, either recovered or improved with intervention or supportive care after nilotinib discontinuation. Moreover, three patients achieved TFR after the occurrence of vascular adverse events during the TFR phase of STAT2. Since four of the six patients had at least one traditional risk factor for vascular adverse events, patients should be carefully screened for risk factors prior to nilotinib administration, with appropriate treatment or supportive care of comorbidities to avoid the development of vascular adverse events. After considering the risk of vascular adverse events in patients given consolidation with nilotinib, we conclude that this therapeutic agent can be safely administered to achieve a successful TFR in CML patients. In conclusion, although previous evidence regarding TFR after discontinuation of second-line nilotinib therapy is limited, our study suggests that a 2-year consolidation period of nilotinib therapy can safely induce higher TFR rates in patients with MR4.5. Thus, 2-year consolidation therapy with this agent may be an effective strategy for achieving TFR in large numbers of CML patients. Acknowledgments This study was supported by research funding from Novartis Pharmaceuticals to NT. The authors would like to thank all study participants and their families, and the study investigators at participating study sites. We also thank Professor Takuhiro Yamaguchi for some advice as a biostatician, the STAT data center (EPS, Co.) for monitoring the clinical trial, and Dr. Toshihiro Miyamoto, Dr. Yosuke Minami, and Dr. Hidetaka Niitsu for their cooperation as members of the data and safety monitoring committee.

1841

N. Takahaski et al.

References 10. 1. Kantarjian HM, Hochhaus A, Saglio G, et al. Nilotinib versus imatinib for the treatment of patients with newly diagnosed chronic phase, Philadelphia chromosomepositive, chronic myeloid leukaemia: 24month minimum follow-up of the phase 3 randomised ENESTnd trial. Lancet Oncol. 2011;12(9):841–851. 2. Larson RA, Hochhaus A, Hughes TP, et al. Nilotinib vs imatinib in patients with newly diagnosed Philadelphia chromosome-positive chronic myeloid leukemia in chronic phase: ENESTnd 3-year follow-up. Leukemia. 2012;26(10):2197–2203. 3. Hochhaus A, Saglio G, Hughes TP, et al. Long-term benefits and risks of frontline nilotinib vs imatinib for chronic myeloid leukemia in chronic phase: 5-year update of the randomized ENESTnd trial. Leukemia. 2016;30(5):1044–1054. 4. Hughes TP, Lipton JH, Spector N, et al. Deep molecular responses achieved in patients with CML-CP who are switched to nilotinib after long-term imatinib. Blood. 2014;124(5):729–736. 5. Hughes TP, Ross DM. Moving treatmentfree remission into mainstream clinical practice in CML. Blood. 2016;128(1):17–23. 6. Dulucq S, Mahon FX. Deep molecular responses for treatment-free remission in chronic myeloid leukemia. Cancer Med. 2016;5(9):2398–2411. 7. Rea D, Cayuela JM. Treatment-free remission in patients with chronic myeloid leukemia. Int J Hematol 2017 Jul 8. [Epub ahead of print] 8. Mahon FX, Rea D, Guilhot J, et al. Discontinuation of imatinib in patients with chronic myeloid leukaemia who have maintained complete molecular remission for at least 2 years: the prospective, multicentre Stop Imatinib (STIM) trial. Lancet Oncol. 2010;11(11):1029–1035. 9. Ross DM, Branford S, Seymour JF, et al. Safety and efficacy of imatinib cessation for CML patients with stable undetectable

1842

11.

12.

13.

14.

15.

16.

17.

18.

minimal residual disease: results from the TWISTER study. Blood. 2013;122:515–522. 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. 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. 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. Noguchi S, Nakaseko C, Nishiwaki K, et al. Switching to nilotinib is associated with deeper molecular responses in chronic myeloid leukemia chronic phase with major molecular responses to imatinib: STAT1 trial in Japan. Int J Hematol. 2018;108(2):176-183. Takahashi N, Kyo T, Maeda Y, et al. Discontinuation of imatinib in Japanese patients with chronic myeloid leukemia. Haematologica. 2012;97(6):903–906. Mahon F, Richter J, Guilhot J, et al. Cessation of tyrosine kinase inhibitors treatment in chronic myeloid leukemia patients with deep molecular response: results of the Euro-Ski trial. Blood. 2016;128(22):787. Mori S, Vegge 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. 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. Int J Hematol. 2018;107(2):185–193. Deininger M. Hematology: curing CML

19.

20.

21.

22.

23.

24.

25. 26.

27.

with imatinib--a dream come true? Nat Rev Clin Oncol. 2011;8(3):127–128. Takahashi N. Predictive factors of successful treatment-free remission for patients with chronic myeloid leukemia. Rinsho Ketsueki. 2014;55(5):489–496. Saussele S, Richter J, Hochhaus A, Mahon FX. The concept of treatment-free remission in chronic myeloid leukemia. Leukemia. 2016;30(8):1638–1647. Mahon FX, Boquimpani C, Kim DW, et al. Treatment-free remission after second-line nilotinib treatment in patients with chronic myeloid leukemia in chronic phase: results from a single-group, phase 2, open-label study. Ann Intern Med. 2018;168(7):461470. Mizoguchi I, Yoshimoto T, Katagiri S, et al. Sustained upregulation of effector natural killer cells in chronic myeloid leukemia after discontinuation of imatinib. Cancer Sci. 2013;104(9):1146–1153. 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):1108– 1116. Richter J, Soderlund S, Lubking A, et al. Musculoskeletal pain in patients with chronic myeloid leukemia after discontinuation of imatinib: a tyrosine kinase inhibitor withdrawal syndrome? J Clin Oncol. 2014;32(25):2821–2823. Rousselot P, Charbonnier A, ConyMakhoul P, et al. Reply to J. Richter et al. J Clin Oncol. 2014;32(25):2823–2825. Lee SE, Choi SY, Song HY, 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. Chai-Adisaksopha C, Lam W, Hillis C. Major arterial events in patients with chronic myeloid leukemia treated with tyrosine kinase inhibitors: a meta-analysis. Leuk Lymphoma. 2016;57(6):1300–1310.

haematologica | 2018; 103(11)

ARTICLE

Acute Myeloid Leukemia

Inhibition of protein disulfide isomerase induces differentiation of acute myeloid leukemia cells

Justyna Chlebowska-Tuz,1,2,3 Olga Sokolowska,1,2,4 Pawel Gaj,1,5 Michal Lazniewski,6,7 Malgorzata Firczuk,1 Karolina Borowiec,1 Hanna Sas-Nowosielska,8 Malgorzata Bajor,1 Agata Malinowska,9 Angelika Muchowicz,1 Kavita Ramji,1 Piotr Stawinski,10 Mateusz Sobczak,2 Zofia Pilch,1 Anna Rodziewicz-Lurzynska,11 Malgorzata Zajac,12 Krzysztof Giannopoulos,12 Przemyslaw Juszczynski,13 Grzegorz W. Basak,14 Dariusz Plewczynski,6,15 Rafal Ploski,10 Jakub Golab1,16 and Dominika Nowis1,2,17

Department of Immunology, Medical University of Warsaw; 2Laboratory of Experimental Medicine, Center of New Technologies, University of Warsaw; 3Institute of Genetics and Biotechnology, Faculty of Biology, University of Warsaw; 4Postgraduate School of Molecular Medicine, Medical University of Warsaw; 5Laboratory of Human Cancer Genetics, Center of New Technologies, University of Warsaw; 6Laboratory of Functional and Structural Genomics, Center of New Technologies, University of Warsaw; 7 Department of Physical Chemistry, Faculty of Pharmacy, Medical University of Warsaw; 8 Laboratory of Imaging Tissue Structure and Function, Nencki Institute of Experimental Biology, Polish Academy of Sciences, Warsaw; 9Laboratory of Mass Spectrometry, Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Warsaw; 10 Department of Medical Genetics, Center of Biostructure Research, Medical University of Warsaw; 11Department of Laboratory Diagnostics, Faculty of Health Sciences, Medical University of Warsaw; 12Department of Experimental Hematooncology, Medical University of Lublin; 13Department of Experimental Hematology, Institute of Hematology and Transfusion Medicine, Warsaw; 14Department of Hematology, Oncology and Internal Diseases, Medical University of Warsaw; 15Faculty of Mathematics and Information Science, Warsaw University of Technology, Warsaw; 16Center for Preclinical Research and Technology, Medical University of Warsaw and 17Genomic Medicine, Medical University of Warsaw, Poland 1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(11):1843-1852

Correspondence: ABSTRACT

cute myeloid leukemia is a malignant disease of immature myeloid cells. Despite significant therapeutic effects of differentiation-inducing agents in some acute myeloid leukemia subtypes, the disease remains incurable in a large fraction of patients. Here we show that SK053, a thioredoxin inhibitor, induces differentiation and cell death of acute myeloid leukemia cells. Considering that thioredoxin knock-down with short hairpin RNA failed to exert antiproliferative effects in one of the acute myeloid leukemia cell lines, we used a biotin affinity probe-labeling approach to identify potential molecular targets for the effects of SK053. Mass spectrometry of proteins precipitated from acute myeloid leukemia cells incubated with biotinylated SK053 used as a bait revealed protein disulfide isomerase as a potential binding partner for the compound. Biochemical, enzymatic and functional assays using fluorescence lifetime imaging confirmed that SK053 binds to and inhibits the activity of protein disulfide isomerase. Protein disulfide isomerase knockdown with short hairpin RNA was associated with inhibition of cell growth, increased CCAAT enhancer-binding protein a levels, and induction of differentiation of HL-60 cells. Molecular dynamics simulation followed by the covalent docking indicated that SK053 binds to the fourth thioredoxin-like domain of protein disulfide isomerase. Differentiation of myeloid precursor cells requires the activity of CCAAT enhancer-binding protein a, the function of which is impaired in acute myeloid leukemia cells through various mechanisms, including translational block by protein disulfide isomerase. haematologica | 2018; 103(11)

d.nowis@cent.uw.edu.pl or jakub.golab@wum.edu.pl Received: February 3, 2018. Accepted: July 10, 2018. Pre-published: July 12, 2018. doi:10.3324/haematol.2018.190231 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1843 ÂŠ2018 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.

1843

J. Chlebowska-Tuz et al.

SK053 increased the levels of CCAAT enhancer-binding protein a and upregulated mRNA levels for differentiation-associated genes. Finally, SK053 decreased the survival of blasts and increased the percentage of cells expressing the maturation-associated CD11b marker in primary cells isolated from bone marrow or peripheral blood of patients with acute myeloid leukemia. Collectively, these results provide a proof-of-concept that protein disulfide isomerase inhibition has potential as a therapeutic strategy for the treatment of acute myeloid leukemia and for the development of small-molecule inhibitors of protein disulfide isomerase.

Introduction Acute myeloid leukemia (AML), the most prevalent acute leukemia among adults, is a malignancy of myeloid lineage cells characterized by the inhibition of cell differentiation leading to accumulation of abnormal white blood cells.1 The use of differentiation-inducing agents, such as all-trans retinoic acid and arsenic trioxide, for the treatment of acute promyelocytic leukemia has brought remarkable therapeutic effects.2,3 However, not all patients with acute promyelocytic leukemia benefit from differentiation treatment and there has been no such significant progress in the treatment of other types of AML.4 The development of new therapeutic agents exerting anti-leukemic effects by targeting unique cellular mechanisms of differentiation is still, therefore, a pressing need of clinical importance.5 It is particularly desirable to develop differentiation-promoting compounds that induce terminal differentiation of leukemic cells leading to cell cycle arrest followed by cell death, and obviate overt cytotoxicity. A critical transcription factor involved in the development and differentiation of myeloid lineage cells is CCAAT enhancer-binding protein a (C/EBPa). In C/EBPa-deficient mice granulocyte differentiation is blocked, 6 and C/EBPa expression in bipotential myeloid progenitors is sufficient to induce granulocytic development.7 Dysregulation of C/EBPa activity is frequently observed in AML patients. Lack of, aberrant or suboptimal C/EBPa activity can result from genomic mutations in the CEBPA gene,8 transcriptional suppression originating from promoter hypermethylation, or functional inactivation by phosphorylation.9 A translational block that occurs in cells experiencing endoplasmic reticulum stress has also been reported as a mechanism leading to C/EBPa downregulation at the mRNA level.10 Various mechanisms such as loss of Ca2+ homeostasis, inhibition of disulfide bond formation, oxidative stress, or hypoxia, lead to endoplasmic reticulum stress, which triggers the unfolded protein response. The role of the unfolded protein response is to restore protein homeostasis and normal endoplasmic reticulum function. Accordingly, this response has been reported to be upregulated in a significant percentage of patients with AML and to be associated with a more favorable course of the disease.10 We have previously developed SK053, a peptidomimetic inhibitor of thioredoxin that exerts cytostatic/cytotoxic effects and endoplasmic reticulum stressmediated apoptosis in tumor cells.11 Conspicuously, we have observed that AML cells incubated with SK053 undergo growth arrest followed by differentiation into more mature myeloid stages and cell death. We, therefore, employed RNA sequencing and a biotin affinity 1844

probe-labeling approach to identify the molecular mechanism of the differentiation-promoting effects of SK053, revealing protein disulfide isomerase (PDI) as a druggable target for AML treatment.

Methods A detailed description of the methods used can be found in the Online Supplementary Data file.

Culture conditions for the acute myeloid leukemia cell lines NB4, MOLM14, HS-5 and HL-60 cells were cultured in RPMI 1640 (Sigma-Aldrich) supplemented with 10% heat-inactivated fetal bovine serum (Hyclone), 100 mg/mL streptomycin and 100 U/mL penicillin (P/S, Sigma-Aldrich). KG1 cells were cultured in IMDM (Gibco) supplemented with 20% heat-inactivated fetal bovine serum (Hyclone) and P/S. HeLa cells were cultured in DMEM (Gibco) supplemented with 10% heat-inactivated fetal bovine serum (Hyclone) and P/S. All cells were cultured at 37°C in a fully humidified atmosphere of 5% CO2.

Acute myeloid leukemia cell differentiation assays Myeloid differentiation was assessed by staining cytospun cells with May-Grünwald-Giemsa as well as in flow cytometry by examining the surface expression of CD11b. A conventional semi-quantitative microscopic nitroblue tetrazolium (NBT; Sigma-Aldrich) assay was used to determine the production of superoxide anion upon stimulation with phorbol myristate acetate in differentiated AML cells. A modified quantitative NBT assay was established and implemented by dissolving the blue formazan particles from equal cell numbers in sodium hydroxide, followed by measurement of absorbance at 620 nm using an ASYS UVM 340 microplate reader (Biochrom, Cambridge, UK).

RNA extraction and quantitative real-time polymerase chain reaction Total RNA was extracted from HL-60 cells using an RNA purification kit (EurX) and relative gene expression was quantified using a LightCycler II 480 Real-Time PCR System (Roche, Basel, Switzerland), LightCycler Probes Master and hydrolysis probes [Universal Probe Library (UPL), Roche] according to the manufacturer's recommendations.

Western blot HL-60 cells were washed twice with ice-cold phosphatebuffered saline, lysed in RIPA buffer (50 mmol/L Tris base, 150 mmol/L NaCl, 1% NP-40, 0.25% sodium deoxycholate, and 1 mmol/L EDTA) or cytoplasmic/nuclear fractions of proteins were extracted using an NE-PER Kit (Pierce). Immunoblotting was done via standard procedures using the antibodies according to the manufacturer’s instructions. haematologica | 2018; 103(11)

Protein disulfide isomerase as a target in leukemia

Isolation of primary acute myeloid leukemia blasts and evaluation of the anti-leukemic effects of SK053

SK053 induces gene expression profile associated with stress response and differentiation

The study was approved by the Institutional Review Board of the Medical University of Lublin (approval n. KE-0254/71/2010) and the Medical University of Warsaw (approval n. KB 182/2011) and was conducted according to the Declaration of Helsinki. Each patient signed written informed consent to the procedures. Leukemic cells were obtained from the bone marrow of patients with AML and isolated by 1.077 g/mL Histopaque (Sigma-Aldrich) density gradient centrifugation. Cells were seeded in IMDM culture medium, supplemented with 10% heat-inactivated fetal bovine serum at a density 5Ă&#x2014;105 cells/mL and incubated for 72 h with SK053. Cytotoxic effects, growth and differentiation upon incubation with SK053 were evaluated with methods described above. 7-AAD staining was used to discriminate dead cells. Additionally, anti-CD33 antibody (BD Biosciences, cat. n. 551378) was used to identify myeloid cells among all peripheral blood mononuclear cells. The patientsâ&#x20AC;&#x2122; basic characteristics are shown in Online Supplementary Table S11.

Results

To get further insight into the mechanisms of SK053 activity in AML cells, we sequenced and analyzed transcriptomes of controls and SK053-treated HL-60 cells using next-generation sequencing. Analysis of transcriptomes of controls and HL-60 cells incubated with SK053 for 48 h (Online Supplementary Figure S2) revealed differential expression of genes (Online Supplementary Figure S3A, Online Supplementary Table S1) that, in gene ontology analysis, clustered into two major groups: (i) biological processes (Online Supplementary Figure S3B, Online Supplementary Table S2) involving the stress response (e.g. genes encoding calreticulin or proteasome subunits), regulation of apoptosis (e.g. genes encoding Bcl-2-related protein, CFLAR or TNFRSF10B) and myeloid cell differentiation (CBFB, JAK2 or PIR) and (ii) molecule function (Online Supplementary Figure S3C, Online Supplementary Table S3) associated with changes in oxidoreductases activity [e.g. NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, ATP synthase or cytochrome c]. Furthermore, a 5-day incubation of HL-60 cells with SK053 induced the expression of genes regulating myeloid cell differentiation (JUN, CSF1, ID2) and cell proliferation (BCL6, PNMT) (Online Supplementary Figure S4A-C, Online Supplementary Tables S4-S6). Immunoblotting of the lysates of HL-60 cells confirmed the results of the next-generation sequencing analysis showing that SK053 induces endoplasmic reticulum stress and the unfolded protein response (Online Supplementary Figure S5).

SK053 induces differentiation of acute myeloid leukemia cells

SK053 binds to and inhibits protein disulfide isomerase

HL-60 acute promyelocytic leukemia cells were incubated for 24 to 120 h with increasing concentrations of SK053 and cell growth as well as cytotoxic effects were determined by counting viable cells and flow cytometry. SK053 inhibited cell growth in a time- and concentration-dependent manner (Figure 1A). We observed similar cytotoxic effects of SK053 against HL-60 cells grown in cell culture medium alone or on the top of bone marrow stromal cells (HS5) mimicking the interaction between AML blasts and the bone marrow niche (Figure 1B). HL60 cells incubated with SK053 also showed features of more mature myelopoietic stages with a decreased nucleus/cytoplasm ratio and a greater reduction of NBT (Figure 1C), corresponding to NADPH oxidase activity. An NBT reduction assay (Figure 1D) as well as increased surface levels of CD11b measured in flow cytometry (Figure 1E) confirmed the differentiation-promoting effects of SK053. Quantitative real-time polymerase chain reaction analysis revealed that incubation of HL-60 cells with SK053 resulted in increased levels of transcripts for differentiation-associated genes, such as differentiation-driving transcription factors (CEBPB, CEBPE), early markers of neutrophil differentiation (HK3) and genes expressed by mature granulocytes (CSF3R encoding G-CSFR, CSF2R encoding GM-CSFR, ELANE and MPO) (Figure 1F). Cytostatic/cytotoxic and differentiation-promoting effects of SK053 were also observed in NB4, another acute promyelocytic leukemia cell line, as well as in non-acute promyelocytic leukemia AML cell lines, such as MOLM14 and KG1 (Online Supplementary Figure S1).

We then used lentiviral particles encoding short hairpin RNA (shRNA) to target thioredoxin in HL-60 cells. Intriguingly, thioredoxin knock-down with shRNA failed to inhibit the growth of HL-60 cells (Online Supplementary Figure S6A). Subsequently, we used two other shRNA sequences targeting thioredoxin (Online Supplementary Figure S6B) and assessed the effects of thioredoxin knockdown in HL-60 and MOLM14 cells. While both hairpins effectively suppressed the growth of MOLM14 cells, only one shRNA slightly inhibited and the other had no effect on the growth of HL-60 cells (Online Supplementary Figure S6C,D). These observations indicate that inhibition of thioredoxin might not be responsible for the differentiation-inducing effects of SK053 in HL-60 cells. Therefore, using a biotin affinity probe-labeling approach followed by mass spectrometry, we sought to identify other potential intracellular SK053 targets in these cells. Incubation of HL-60 cells for 4 h with SK-BIO, an active, biotinylated SK053 analog (see Online Supplementary Figure S8 for the general structures of compound used herein), followed by pulldown with avidin-coated beads and sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), revealed a prominent band of ~50-70 kDa molecular weight in silver-stained gels (Figure 2A). An inactive, biotinylated analog of SK053 that lacks the electrophilic double bond (referred to as SK-IN, Online Supplementary Figure S8) failed to precipitate any protein. The ~50-70 kDa protein was identified in mass spectrometry to be a protein disulfide isomerase (PDI), a product of the P4HB gene (Online Supplementary Figure S9A-D, Online Supplementary Table S7). SDS-PAGE followed by the

Statistical analysis Data were analyzed using Microsoft Excel 2010 and GraphPad Prism 6.0 for Windows (GraphPad Software Inc., La Jolla, CA, USA) software. The results were analyzed for significance using a two-tailed Student t-test or one-way ANOVA with a Dunnett post-hoc test. Statistical significance was defined as P values <0.05.

haematologica | 2018; 103(11)

1845

J. Chlebowska-Tuz et al. A

1846

Figure 1. SK053 induces differentiation of AML cells. (A) HL-60 cells were incubated with various concentrations of SK053 for up to 5 days. Each day the number of dead cells was evaluated with trypan blue exclusion (mean ± SD from 6 experiments). (B) Co-culture of HL-60 cells with universal bone marrow stromal cells (HS5) did not impair SK053 cytotoxicity. HL-60 cells were seeded in a transwell system on top of HS5 cells and incubated with SK053 for 72 h. Next, HL-60 cells were analyzed for viability using propidium iodide (PI) staining in flow cytometry (data are shown as mean percentages of PI-positive cells ± SD from 3 experiments); *P<0.05 in oneway ANOVA with the Dunnett posthoc test. (C) May-GrünwaldGiemsa (MGG) staining (upper panel) and nitro-blue tetrazolium (NBT) (lower panel) in HL-60 cells incubated with SK053 for 120 h. All-trans retinoic acid (ATRA) served as a positive control. (D) Semi-quantitative colorimetric NBT reduction assay in HL-60 cells incubated for 5 days with 10 mM SK053 (data are shown as mean increases, in %, over controls ± SD for 3 experiments); *P<0.05 in one-way ANOVA with the Dunnett post-hoc test. (E) Mean percentage ± SD of HL-60 cells expressing CD11b myeloid marker was determined in flow cytometry in 3 experiments; *P<0.05 in one-way ANOVA with the Dunnett post hoc test. (F) Real-time quantitative polymerase chain reaction analysis of HL-60 cells incubated for 5 days with 10 mM SK053; results are presented as mean target-toreference ratio ± SD of three experiments; *P<0.05 vs. controls, two-tailed Student t test, # P<0.05 vs. controls, one-tailed Student t test.

haematologica | 2018; 103(11)

Protein disulfide isomerase as a target in leukemia

Figure 2. SK053 binds to protein disulfide isomerase. (A) HL-60 cells were incubated with 100 mM SK-BIO and lysed. Proteins binding SK-BIO were precipitated with avidin-coated beads and eluted with buffers E1 and E2 as described in the Methods section. Next, the samples were separated by SDS-PAGE. A band obtained with buffer E1 (black rectangle) was excised from silver-stained gel and analyzed by mass spectrometry. An inactive biotinylated SK053 analog (SK-IN) was used as a negative control. A representative result of a series of experiments is presented. (B) Recombinant human (rhu) PDI was incubated with a 10× molar excess of SK-BIO for 10-120 min at 37°C (left) or preincubated for 1 h with a 10× molar excess of SK053 followed by 1 h incubation with a 10× molar excess of SK-BIO (right). Next, proteins were separated by SDS-PAGE followed by western blotting (WB) using anti-biotin monoclonal antibody. Membranes stained with Ponceau red served as loading controls. A representative result of a series of experiments is presented. (C) Reduced (red) and oxidized (ox) rhuPDI were incubated with SK-BIO followed by immunoblotting for biotin. Gels stained with Coomasie blue served as loading controls. SK-IN was used as a negative control. A representative result of a series of experiments is presented. (D) The lysates of HL-60 cells were incubated with SK-BIO followed by biotin immunoprecipitation and western blotting. Biotin detection in total cell lysates (input) was used as a loading control. Lysate incubated with protein G agarose beads in the absence of antibodies was used as a negative control (No Ab). A representative result of a series of experiments is presented. IP: immunoprecipitation; TXN: thioredoxin.

detection of biotin revealed covalent binding of SK-BIO to recombinant human (rhu) PDI (Figure 2B). Oxidation of rhuPDI, which decreases the number of sulfhydryl groups, reduced binding of SK-BIO to PDI indicating a thiol-dependent interaction (Figure 2C). Moreover, in a competition assay, pre-incubation of rhuPDI with SK053 prevented binding of SK-BIO, indicating the same binding site for both compounds (Figure 2B). Immunoprecipitation of PDI and thioredoxin from the lysates of HL-60 cells cultured with SK-BIO revealed the presence of biotinylated PDI but not thioredoxin (Figure 2D), indicating that in HL-60 cells PDI is a preferential target for the compound. Further mass spectrometry revealed that SK053 binds to the catalytic cysteines localized in the fourth thioredoxin-like domain of PDI (Online Supplementary Figure S9E). Furthermore, the molecular dynamics simulation followed by covalent docking of SK053 to the fourth thioredoxin-like domain of human PDI (pdb|4ekz) indicated that the binding of the truncated form of SK053 (a molecule without the leaving group, Online Supplementary Figure S8D) is equally likely to occur by forming a covalent bond with either sulfur from Cys397 or Cys400 (Online Supplementary Figure S10, Online Supplementary Movie 1). Next, we sought to determine whether binding of SK053 translates into inhibition of PDI activity. A turbidimetric assay of PDI-mediated insulin reduction revealed that SK053 efficiently blocks the enzymatic activity of rhuPDI with an IC50 value of haematologica | 2018; 103(11)

9.98 mM (Figure 3A,B). To further investigate whether SK053 can inhibit PDI activity in cells, we transfected HeLa cervical cancer cells with a plasmid encoding an endoplasmic reticulum-tuned fluorescent redox-responsive probe (roGFPiE)12 and measured the response of this probe using fluorescence lifetime imaging in a time-correlated single-photon counting mode. The fluorescent lifetime of the probe declined in SK053-treated HeLa cells, reflecting the reductive shift in the dithiol-disulfide steady state and indicating SK053-mediated inhibition of PDI activity in the endoplasmic reticulum of living cells (Figure 3C).

Inhibition of protein disulfide isomerase increases C/EBPa levels and inhibits the growth of acute myeloid leukemia cells Recently, it was shown that PDI interacts with the stem loop region of mRNA for C/EBPa thereby blocking its translation.13 Considering that C/EBPa is a crucial transcription factor driving myeloid cells differentiation14,15 we investigated the effects of SK053 on C/EBPa levels in AML cells. Incubation of HL-60 cells with SK053 increased the nuclear levels of the 42 kDa C/EBPa isoform (Figure 4A) and induced CEBPA mRNA expression (Figure 4C). Additionally, the oncogenic p30 kDa C/EBPa levels were strongly suppressed in cells incubated with SK053 (Figure 4B). Accordingly, SK053 decreased the levels of SOX4 mRNA (Figure 1G), a direct target of C/EBPa, the expres1847

J. Chlebowska-Tuz et al. A

Figure 3. SK053 inhibits enzymatic activity of protein disulfide isomerase. (A) A turbidimetric assay of recombinant human PDI-mediated insulin disulfide reduction (mean Âą SD of 6 experiments). (B) Determination of the IC50 value for SK053 with SigmaPlot software. (C) Fluorescence lifetime imaging (FLIM) of HeLa cells transfected with plasmid encoding an endoplasmic reticulum-tuned fluorescent redox-responsive probe (roGFPiE) tracking the activity of the PDI in cells; 100 mM dithiothreitol (DTT), a reducing agent, was used as a positive control. A representative result of a series of experiments is presented.

sion of which has been shown to correlate inversely with C/EBPa activity, increased self-renewal of leukemic cells as well as a block in blast differentiation.16 shRNA-mediated knock-down of PDI (P4HB) expression significantly suppressed the proliferation of HL-60 cells (Figure 4D,E). Lentiviral transduction of HL-60, MOLM14 and KG1 cells with another shRNA sequence confirmed that PDI knockdown is associated with significant growth suppression (Figure 4F,G) and upregulates C/EBPa protein (Online Supplementary Figure S7A). PDI knock-down also increased the percentage of CD11b+ HL-60 cells as compared with nontargeting controls, but failed to do so in MOLM14 and KG1 cells (Figure 4H). Unexpectedly, SK053 strongly suppressed the growth and induced differentiation of HL-60 cells transduced with two independent shRNA sequences targeting CEBPA (Online Supplementary Figure S7B-D), indicating that the presence of this differentiation-associated transcription factor is not necessary for the antileukemic effects of SK053.

Protein disulfide isomerase inhibition targets leukemia-initiating cells and is effective in primary acute myeloid leukemia cells AML is characterized by a small population of selfrenewing leukemic stem cells referred to as leukemia-initiating cells that give rise to a large population of immature leukemic blasts.17 Since leukemia-initiating cells are intrinsically resistant to chemotherapeutics and AML frequently relapses due to incomplete elimination of leukemia-initiating cells18 we sought to investigate the effects of SK053 on the survival of CD123+CD34+CD381848

leukemia-initiating cell-like cells within the KG1 AML cell line. We observed a concentration-dependent decrease in the percentage of these cells when incubated with SK053 (Figure 5A,B). Moreover, pretreatment of KG1 cells with SK053 decreased the clonogenic potential of these cells in a colony formation assay in vitro (Figure 5C,D). Moreover, SK053 exerted antileukemic effects in primary AML blasts isolated from bone marrow or peripheral blood of AML patients. We observed a decrease in blast survival in all samples from six patients investigated, accompanied by an increased percentage of cells expressing CD11b marker in five out of the six samples (Figure 6).

Discussion In this study we demonstrate that SK053 induces differentiation of AML cells. SK053 has been developed as a thioredoxin inhibitor and was reported to induce endoplasmic reticulum stress and apoptosis of tumor cells.11 Thus, it was to some extent unexpected to observe that SK053 induces cytostatic/cytotoxic effects against HL-60 cells (Figure 1A,B), while thioredoxin knock-down with shRNA turned out to be largely ineffective in suppressing the growth of these cells (Online Supplementary Figure S6C). This observation prompted us to look for an additional cellular target for SK053. Using biotinylated SK053 as a bait in a biotin affinity probe-labeling approach we observed that SK053 binds to PDI (Figure 2, Online Supplementary Figure S9). Molecular docking confirmed haematologica | 2018; 103(11)

Protein disulfide isomerase as a target in leukemia

Figure 4.SK053 increases CEBPA levels in HL-60 cells and protein disulfide isomerase knock-down impairs the growth of acute myeloid leukemia cells and induces differentiation of HL-60 cells. (A) HL-60 cells were incubated for 2 or 5 days with 10 mM SK053. Subcellular fractions were isolated using a NE-PER kit. HDAC2 served as a loading control for the nuclear fraction. A representative result of a series of experiments is presented. (B) HL-60 cells were incubated for 1-120 h with 10 mM SK053, harvested, lysed and total cell lysates were immunoblotted for 30 kDa C/EBPa. a-tubulin levels served as a loading control. A representative result of a series of experiments is presented. For (A) and (B) the controls are HL-60 cells incubated with dimethylsulfoxide at the same concentration as used for SK053-treated group, harvested on day 5. (C) Real-time quantitative polymerase chain reaction of HL-60 cells incubated for the indicated times with 10 mM SK053; results are presented as mean target-to-reference ratio ± SD of three experiments. (D) Western blot showing the efficacy of PDI or thioredoxin (TXN) knock-down 5 days after transduction with shRNA-encoding lentiviruses. PDI(a) and PDI(b) indicate two different shRNA; β-actin (βA) served as a loading control. (E) Growth of HL-60 cells transduced with non-targeting (NTC), TXN [TXN(a)]- or PDI [PDI(b)]-targeting shRNA lentiviral particles. Cells were counted daily in trypan blue from day 1 until day 8. *P<0.05 vs. NTC in a two-tailed Student t-test. For (C) and (D) the controls are non-modified HL-60 cells. (F) Western blot showing the efficacy of PDI knock-down in AML cell lines transduced with PDI-targeting shRNA [PDI(d)]-encoding lentiviruses. GAPDH served as a loading control. (G) Growth of HL-60, MOLM14 and KG1 cells transduced with non-targeting (NTC) or PDI-targeting [PDI(d)] shRNA lentiviral particles. Cells were counted daily in trypan blue from day 1 until day 9. *P<0.05 vs. NTC in a two-tailed Student t-test. (H) Mean percentage ± SD of AML cells modified with nontargeting (NTC) or PDI-targeting [PDI(d)] shRNA, expressing CD11b myeloid marker determined in flow cytometry (n=4 experiments, except for KG1: n=2), ***P<0.0001 vs. NTC, two-tailed Student t-test; NS: not significant.

haematologica | 2018; 103(11)

1849

J. Chlebowska-Tuz et al. A

Figure 5. SK053 decreases the percentage of KG1 leukemia-initiating cells and their clonogenic potential in vitro. (A) KG1 cells were incubated for 72 h with the indicated concentrations of SK053, harvested and stained for flow cytometry. Representative density plots from a series of experiments are presented. (B) Mean percentages of CD123+ KG1 cells among CD34+ and CD38– cells ± SD (n=3 experiments) *P<0.05 vs. controls, one-way ANOVA with Dunnett post-hoc test. (C) Representative pictures of the clonogenic assay plates on day 14 after seeding KG1 cells pretreated for 72 h with SK053. (D) Mean clone size (pixel area) ± SEM (n=3 experiments); **P<0.0001 vs. controls, one-way ANOVA with Dunnett post-hoc test.

the feasibility of this interaction (Online Supplementary Figure S10) and further precipitation, as well as enzymatic and functional analyses, revealed that SK053 binds to and inhibits the activity of human PDI (Figures 2 and 3). PDI is the original member of a family of PDI proteins that contain a characteristic CXXC motif, with two cysteine residues forming a disulfide bond that is cleaved by oxidoreductases or by thiol-disulfide exchange. Although all PDI family members contain a thioredoxin-like domain, they differ considerably in size, domain composition and enzymatic properties. PDI is involved in the formation and isomerization of disulfide bonds between cysteine residues of polypeptides as they fold.19 Such disulfide modifications participate in post-translational protein control and affect the functions of many proteins. PDI also participates in the maintenance of cellular homeostasis by mediating oxidative protein folding and acts as a chaperone.19 A number of studies indicate that various PDI are highly expressed in multiple cancer types as compared with matched normal tissues and play an important role in supporting cancer progression.19 PDI is induced in tumor cells during the unfolded protein response and has been associated with chemoresistance.20,21 Inhibition of PDI activity with a non1850

selective inhibitor, bacitracin, enhanced apoptosis triggered by bortezomib or fenretinide.22 Propynoic acid carbamoyl methyl amide (PACMA31), an irreversible PDI inhibitor, exhibited significant antitumor effects in ovarian cancer models23 and potentiated the antitumor effects of sorafenib in hepatocellular carcinoma.24 A structurally different PDI inhibitor (CCF642) was shown to exert anti-myeloma activity associated with the induction of endoplasmic reticulum stress and apoptosis-inducing calcium release.25 All these observations indicate that PDI is a potential drug target in cancer treatment. Remarkably, Heafliger et al. have shown that PDI binds to the stem loop region of C/EBPa mRNA, thereby blocking its translation.13 Our results indicate that SK053 upregulates C/EBPa levels (Figure 4), downregulates its downstream effector SOX4, and induces differentiation of AML cells (Figure 1F). However, despite C/EBPa upregulation we have observed that SK053 still exerts potent growthinhibitory effects in HL-60 cells with stably suppressed CEBPA expression induced by two different shRNA sequences (Online Supplementary Figure S7). Thus, it seems unlikely that PDI inhibition with SK053 directly leads to stabilization of CEBPA mRNA. Nonetheless, PDI knockhaematologica | 2018; 103(11)

Protein disulfide isomerase as a target in leukemia

Figure 6. SK053 exerts antileukemic effects and induces differentiation of primary leukemic blasts isolated from the bone marrow of patients with acute myeloid leukemia. (Left) Primary AML blasts isolated from six patients (numbered consecutively Pt 1 to Pt 6) were incubated with the indicated concentrations of SK053 for 72 h. Each day the number of dead cells was evaluated with trypan blue exclusion. Graphs present mean percentages of dead cells Âą SD (n=6 experiments). *P<0.05 vs. controls, one-way ANOVA with Dunnett post-hoc test. (Right) Primary AML blasts isolated from six patients were incubated with the indicated concentrations of SK053 for 72 h. Percentages of CD11b+ cells among live (7-AAD-) and CD33+ cells were evaluated in flow cytometry. Graphs present mean percentages of dead cells Âą SD (n=3 experiments). *P<0.05 vs. controls, one-way ANOVA with Dunnett post-hoc test.

haematologica | 2018; 103(11)

1851

J. Chlebowska-Tuz et al.

down with shRNA (Figure 4) exerted cytostatic/cytotoxic effects in HL-60, MOLM14 and KG1 cells and induced differentiation of HL-60 cells. In contrast to HL-60 cells, thioredoxin knock-down in MOLM14 and KG1 cells was also associated with inhibition of cell growth. Thus, it seems that both thioredoxin and PDI are involved in the survival of these cells, and since SK053 targets both enzymes, the antileukemic activity of this compound can be attributed to its dual selectivity. The findings that SK053 significantly decreases the percentage of CD123+CD34+CD38â&#x20AC;&#x201C; leukemia-initiating cells and decreases the clonogenic potential of these cells in the colony formation assay in vitro are particularly important. Leukemia-initiating cells seem to have a pivotal role in the relapse of AML and are considered to be resistant to conventional chemotherapy and targeted therapies.26 Altogether, we show that SK053 targets PDI by thioldependent interactions, modulates C/EBPÎą levels and

References 1. Dohner H, Weisdorf DJ, Bloomfield CD. Acute myeloid leukemia. N Engl J Med. 2015;373(12):1136-52. 2. Mi JQ, Li JM, Shen ZX, Chen SJ, Chen Z. How to manage acute promyelocytic leukemia. Leukemia. 2012;26(8):1743-1751. 3. Lengfelder E, Hofmann WK, Nowak D. Impact of arsenic trioxide in the treatment of acute promyelocytic leukemia. Leukemia. 2012;26(3):433-442. 4. Coombs CC, Tavakkoli M, Tallman MS. Acute promyelocytic leukemia: where did we start, where are we now, and the future. Blood Cancer J. 2015;5:e304. 5. Montalban-Bravo G, Garcia-Manero G. Novel drugs for older patients with acute myeloid leukemia. Leukemia. 2015;29(4): 760-769. 6. Zhang DE, Zhang P, Wang ND, Hetherington CJ, Darlington GJ, Tenen DG. Absence of granulocyte colony-stimulating factor signaling and neutrophil development in CCAAT enhancer binding protein alphadeficient mice. Proc Natl Acad Sci USA. 1997;94(2):569-574. 7. Radomska HS, Huettner CS, Zhang P, Cheng T, Scadden DT, Tenen DG. CCAAT/enhancer binding protein alpha is a regulatory switch sufficient for induction of granulocytic development from bipotential myeloid progenitors. Mol Cell Biol. 1998;18(7):4301-4314. 8. Pabst T, Mueller BU, Zhang P, et al. Dominant-negative mutations of CEBPA, encoding CCAAT/enhancer binding protein-alpha (C/EBPalpha), in acute myeloid leukemia. Nat Genet. 2001;27(3):263-270. 9. Radomska HS, Basseres DS, Zheng R, et al.

1852

10.

11.

12.

13.

14.

15.

16.

17.

18.

induces differentiation of AML cells. These observations indicate that PDI is a druggable target for differentiation treatment in AML. Acknowledgments The authors thank Dr. Magdalena Winiarska from the Department of Immunology, MUW and Prof. Andrzej Dziembowski from the Institute of Biochemistry and Biophysics PAS for their critical and helpful review of this work as well as Dr. Tomasz Stoklosa from the Department of Immunology, MUW for advice concerning the clonogenic assays. This work was supported by the Polish National Science Center, grant numbers: 2013/10/E/NZ5/00778 (DN), 2014/15/B/ST6/05082 (DP), Foundation for Polish Science (TEAM to DP), Ministry of Science and Higher Education, grant number: IP2011 038971 (DN) and European Commission Horizon 2020 Programme 692180STREAMH2020-TWINN-2015 (JG).

Block of C/EBP alpha function by phosphorylation in acute myeloid leukemia with FLT3 activating mutations. J Exp Med. 2006;203(2):371-381. Schardt JA, Weber D, Eyholzer M, Mueller BU, Pabst T. Activation of the unfolded protein response is associated with favorable prognosis in acute myeloid leukemia. Clin Cancer Res. 2009;15(11):3834-3841. Klossowski S, Muchowicz A, Firczuk M, et al. Studies toward novel peptidomimetic inhibitors of thioredoxin-thioredoxin reductase system. J Med Chem. 2012;55 (1):55-67. Avezov E, Cross BC, Kaminski Schierle GS, et al. Lifetime imaging of a fluorescent protein sensor reveals surprising stability of ER thiol redox. J Cell Biol. 2013;201(2):337-349. Haefliger S, Klebig C, Schaubitzer K, et al. Protein disulfide isomerase blocks CEBPA translation and is up-regulated during the unfolded protein response in AML. Blood. 2011;117(22):5931-5940. Ohlsson E, Schuster MB, Hasemann M, Porse BT. The multifaceted functions of C/EBPalpha in normal and malignant haematopoiesis. Leukemia. 2016;30(4):767775. Pulikkan JA, Tenen DG, Behre G. C/EBPalpha deregulation as a paradigm for leukemogenesis. Leukemia. 2017;31(11): 2279-2285. Zhang H, Alberich-Jorda M, Amabile G, et al. Sox4 is a key oncogenic target in C/EBPalpha mutant acute myeloid leukemia. Cancer Cell. 2013;24(5):575-588. Bonnet D, Dick JE. Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nat Med. 1997;3(7):730-737. Jordan CT. Unique molecular and cellular

19.

20.

21.

22.

23.

24.

25.

26.

features of acute myelogenous leukemia stem cells. Leukemia. 2002;16(4):559-562. Xu S, Sankar S, Neamati N. Protein disulfide isomerase: a promising target for cancer therapy. Drug Discov Today. 2014;19(3): 222-240. Higa A, Taouji S, Lhomond S, et al. Endoplasmic reticulum stress-activated transcription factor ATF6alpha requires the disulfide isomerase PDIA5 to modulate chemoresistance. Mol Cell Biol. 2014;34(10):1839-1849. Tufo G, Jones AW, Wang Z, et al. The protein disulfide isomerases PDIA4 and PDIA6 mediate resistance to cisplatin-induced cell death in lung adenocarcinoma. Cell Death Differ. 2014;21(5):685-695. Lovat PE, Corazzari M, Armstrong JL, et al. Increasing melanoma cell death using inhibitors of protein disulfide isomerases to abrogate survival responses to endoplasmic reticulum stress. Cancer Res. 2008;68(13): 5363-5369. Xu S, Butkevich AN, Yamada R, et al. Discovery of an orally active small-molecule irreversible inhibitor of protein disulfide isomerase for ovarian cancer treatment. Proc Natl Acad Sci USA. 2012;109(40):1634816353. Won JK, Yu SJ, Hwang CY, et al. Protein disulfide isomerase inhibition synergistically enhances the efficacy of sorafenib for hepatocellular carcinoma. Hepatology. 2017;66(3): 855-868. Vatolin S, Phillips JG, Jha BK, et al. Novel protein disulfide isomerase inhibitor with anticancer activity in multiple myeloma. Cancer Res. 2016;76(11):3340-3350. Pollyea DA, Jordan CT. Therapeutic targeting of acute myeloid leukemia stem cells. Blood. 2017;129(12):1627-1635.

haematologica | 2018; 103(11)

ARTICLE

Acute Myeloid Leukemia

Genetics of acute myeloid leukemia in the elderly: mutation spectrum and clinical impact in intensively treated patients aged 75 years or older

Victoria V. Prassek,1 Maja Rothenberg-Thurley,1 Maria C. Sauerland,2 Tobias Herold,1,3,4 Hanna Janke,1 Bianka Ksienzyk,1 Nikola P. Konstandin,1 Dennis Goerlich,2 Utz Krug,5 Andreas Faldum,2 Wolfgang E. Berdel,2 Bernhard Wörmann,6 Jan Braess,7 Stephanie Schneider,1 Marion Subklewe,1 Stefan K. Bohlander,8 Wolfgang Hiddemann,1,3,4 Karsten Spiekermann1,3,4 and Klaus H. Metzeler1,3,4

Ferrata Storti Foundation

Haematologica 2018 Volume 103(11):1853-1861

Laboratory for Leukemia Diagnostics, Department of Medicine III, University Hospital, LMU Munich, Germany; 2Institute of Biostatistics and Clinical Research, University of Münster, Germany; 3German Cancer Consortium (DKTK), Partner Site Munich, Germany; 4 German Cancer Research Center (DKFZ), Heidelberg, Germany; 5Hospital Leverkusen, Germany; 6Charité – University Hospital Berlin, Germany; 7Department of Oncology and Hematology, Hospital Barmherzige Brüder, Regensburg, Germany and 8Department of Molecular Medicine and Pathology, University of Auckland, New Zealand 1

ABSTRACT

cute myeloid leukemia is a disease of the elderly (median age at diagnosis, 65-70 years). The prognosis of older acute myeloid leukemia patients is generally poor. While genetic markers have become important tools for risk stratification and treatment selection in young and middle-aged patients, their applicability in very old patients is less clear. We sought to validate existing genetic risk classification systems and identify additional factors associated with outcomes in intensively treated patients aged ≥75 years. In 151 patients who received induction chemotherapy in the AMLCG-1999 trial, we investigated recurrently mutated genes using a targeted sequencing assay covering 64 genes. The median number of mutated genes per patient was four. The most commonly mutated genes were TET2 (42%), DNMT3A (35%), NPM1 (32%), SRSF2 (25%) and ASXL1 (21%). The complete remission rate was 44% and the 3-year survival was 21% for the entire cohort. While adverse-risk cytogenetics (MRC classification) were associated with shorter overall survival (P=0.001), NPM1 and FLT3-ITD mutations (present in 18%) did not have a significant impact on overall survival. Notably, none of the 13 IDH1-mutated patients (9%) reached complete remission. Consequently, the overall survival of this subgroup was significantly shorter than that of IDH1-wildtype patients (P<0.001). In summary, even among very old, intensively treated, acute myeloid leukemia patients, adverse-risk cytogenetics predict inferior survival. The spectrum and relevance of driver gene mutations in elderly patients differs from that in younger patients. Our data implicate IDH1 mutations as a novel marker for chemorefractory disease and inferior prognosis. (AMLCG-1999 trial: clinicaltrials.gov identifier, NCT00266136)

Introduction The incidence of acute myeloid leukemia (AML) is highest among the elderly, with the median age at diagnosis being around 70 years.1 The prognosis of older AML patients is generally considered poor. In this age group, comorbidities, poor performance status and a reluctance of physicians and patients to use treatment regimens perceived as toxic often lead to the decision to avoid induction chemotherapy in favor of less intensive therapies.2-6 On the other hand, data from patient registries and randomized trials suggest that intensive induction therapy haematologica | 2018; 103(11)

Correspondence: klaus.metzeler@med.uni-muenchen.de

Received: February 22, 2018. Accepted: June 11, 2018. Pre-published: June 14, 2018. doi:10.3324/haematol.2018.191536 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1853 ©2018 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.

1853

V.V. Prassek et al.

can result in prolonged overall survival (OS), at least in a subset of elderly patients aged 70-79 years, and may even be beneficial in selected octogenarians.4,7-10 However, it remains unclear which genetic and clinical factors are relevant to identify those elderly patients most likely to benefit from, and least likely to be harmed by, induction chemotherapy.11 Advances in the field of molecular genetics and the development of next-generation sequencing expanded our knowledge of recurrently mutated genes in AML and their role in disease pathophysiology, and led to refined risk classifications in younger patients.1,12,13 However, elderly patients were underrepresented or excluded in these studies, and the question thus arises whether the spectrum and prognostic relevance of gene mutations are similar in very old AML patients. Some studies of established genetic risk factors indicate that there may be important differences between elderly and younger patients.12,14,15 For example, the FLT3-ITD mutation is a well-recognized adverse prognostic factor in young adults, whereas its impact is reduced or absent in elderly patients.1,7,14,16,17 Therefore, comprehensive genetic analyses in cohorts of very old patients are needed to clarify the relevance of distinct gene alterations in this subset of patients. To identify prognostic factors associated with clinical outcomes in elderly AML patients, we studied 151 patients aged ≥75 years who received intensive induction therapy. The mutational spectrum in 64 recurrently mutated AML genes was analyzed by targeted next-generation sequencing. We then studied associations of genetic alterations with other known prognostic factors, patients’ characteristics, and outcomes. Our aim was to define subsets of patients who may benefit from intensive induction therapy. Furthermore, we evaluated the prognostic relevance of the British Medical Research Council (MRC) 2010 and European LeukemiaNet (ELN) 2017 risk classification schemes in this subset of patients.1,18

Methods Patients, treatment, karyotype and molecular analyses

We studied 151 patients aged ≥75 years with newly diagnosed AML or high-risk myelodysplastic syndromes (10% - <20% bone marrow blasts; n=3), diagnosed according to World Health Organization criteria, who had suitable bone marrow or peripheral blood specimens for genetic analysis. All patients received intensive induction treatment during the German AML Cooperative Group AMLCG-1999 randomized, multicenter, phase III trial (clinicaltrials.gov identifier, NCT00266136) between 1999 and 2011. Participants aged ≥60 years were randomized to receive a first induction course with high-dose cytarabine and mitoxantrone (HAM) or with standard-dose cytarabine, daunorubicin and 6thioguanine (TAD-9). A second HAM induction course was administered from day 21 only if ≥5% residual blasts were present in a bone marrow aspirate taken on day 15. Another cycle of TAD-9 was given as consolidation therapy, followed by monthly cytarabine-based maintenance chemotherapy (Online Supplementary Data). Karyotype analyses were performed centrally and results were classified according to the 2010 MRC classification.18 Molecular analysis encompassed sequencing of 64 genes recurrently mutated in AML. We analyzed either known mutational hotspots or the entire coding sequence using an amplicon-based approach (Haloplex, Agilent, Boeblingen, Germany) as described 1854

previously.12 The median sequencing coverage of the target region across all samples was 520-fold, and 98.8% of the target region was covered at >30-fold. Gene alterations with variant allele frequencies of ≥2% were classified as driver mutations, variants of unknown significance or germline polymorphisms.12 The Catalogue Of Somatic Mutations In Cancer (COSMIC), The Cancer Genome Atlas (TCGA) and the Single Nucleotide database (dbSNP) served as databases to compare and classify the mutational results.19-21 NPM1,22 FLT3-ITD23 and CEBPA24 mutations were additionally tested using polymerase chain reaction followed by Sanger sequencing and/or fragment analysis. Written informed consent for inclusion in the clinical trial and genetic analyses was provided by all patients. All study protocols were in accordance with the Declaration of Helsinki and approved by the institutional review boards of each participating center.

Statistical analysis Associations among gene mutations, and between gene mutations and patients’ pretreatment features, were analyzed using the Fisher exact test for categorical variables and the Wilcoxon ranksum test for continuous variables. The Kaplan-Meier method was used to calculate estimated survival probabilities, with the logrank test evaluating differences between survival distributions. A multivariate Cox regression model including known risk factors [MRC risk category and Eastern Cooperative Oncology Group performance status (ECOG PS)], and gene mutations that showed a univariate association with OS (at P<0.10), was used to identify factors associated with survival. Statistical analyses were performed using SPSS version 24.0 (Chicago, IL, USA). All statistical tests are two-sided, and a P value of ≤0.05 was considered statistically significant.

Results Patients’ characteristics

We identified 151 patients aged ≥75 years treated in the AMLCG-1999 trial for whom suitable material for genetic analyses was available. Details of this trial have been published elsewhere.25 The patients’ baseline characteristics are shown in Table 1. The median age of the subjects was 76 years (range, 75-86). Eighty-one percent of patients had a clinical diagnosis of de novo AML, 15% had secondary AML and 3% had therapy-related AML. Three patients (2%) had high-risk myelodysplastic syndromes (10-<20% bone marrow blasts). Most patients (90%) had an ECOG PS of ≤2. Among 137 patients with cytogenetic data, 82% belonged to the intermediate-risk group according to the MRC 2010 classification, including 52% with cytogenetically normal AML. Only three patients (2%) had favorable cytogenetics, and 16% fell in the MRC adverse-risk group.

Treatment outcomes In the overall cohort, the rates of complete remission (CR) and CR with incomplete blood count recovery (CRi) were 44% and 4%, respectively. The median event-free survival (EFS) was 1.7 months. The median relapse-free survival (RFS) for patients achieving a remission was 12 months, and the median OS was 6.0 months (Online Supplementary Figure S1). The 3-year OS rate for the entire study population was 21%, and survival was similar for patients aged 75-79 or 80-86 years (P=0.3) (Online Supplementary Figure S2A). Patients with an ECOG PS of 3 or 4 had shorter OS compared to patients with an ECOG PS of 0-2 (P=0.036) (Online Supplementary Figure S2B), haematologica | 2018; 103(11)

Genetics and outcomes in AML patients ≥75 years

mostly because of an increased risk of early death within 60 days from starting treatment (26% for patients with an ECOG PS of 0-1, 40% for those with a PS of 2, and 75% for patients with an ECOG PS of 3 or 4). Patients randomized to HAM induction tended to have longer OS compared to patients randomized to TAD induction (median OS: TAD, 3.1 months versus HAM, 7.8 months; P=0.09) (Online Supplementary Results and Online Supplementary Figure S3).

according to the MRC classification, the favorable- and intermediate-risk groups were analyzed jointly. Patients in the favorable- or intermediate-risk categories had a nonsignificantly higher remission rate than adverse-risk patients (CR/CRi, 51% versus 27%; P=0.2). The EFS, RFS, and OS of favorable- and intermediate-risk patients were significantly longer than those for adverse-risk patients (EFS: P=0.001; RFS: P=0.006; OS: P=0.001) (Figure 1A and Online Supplementary Figure S4).

Impact of cytogenetics on survival

Gene mutations and patients’ pretreatment characteristics

Since only three patients had favorable-risk cytogenetics Table 1. Patients’ baseline characteristics.

Variable Age [years], median (range) Male sex Disease type De novo AML Secondary AML Therapy-related AML MDS (10%-<20% blasts) ECOG performance statusa 0 1 2 ≥3 WBC [×109/L], median (range) Blast count at diagnosis % bone marrow blasts, median (range) % peripheral blood blasts, median (range) MRC cytogenetic risk categoryb Favorable Intermediate Adverse ELN 2017 genetic groupc Favorable Intermediate Adverse Remission status after induction CR CRi Persistent AML Death during induction Most commonly mutated genes TET2 DNMT3A NPM1 SRSF2 ASXL1 RUNX1 FLT3-ITD NRAS IDH2 TP53 FLT3-TKD IDH1

Total cohort (n=151) 76 (75 – 86) 81 (54%) 122 (81%) 22 (15%) 4 (3%) 3 (2%) 13 (10%) 70 (51%) 42 (31%) 12 (9%) 14.2 (0.1 – 318.1)

80 (10 – 100) 26 (0 – 100) 3 (2%) 112 (82%) 22 (16%) 37 (28%) 32 (24%) 65 (49%) 66 (44%) 6 (4%) 25 (17%) 54 (35%) 63 (42%) 53 (35%) 48 (32%) 38 (25%) 31 (21%) 28 (19%) 27 (18%) 25 (17%) 23 (15%) 21 (14%) 18 (12%) 13 (9%)

ECOG: Eastern Cooperative Oncology Group; WBC: white blood cell count; MRC: British Medical Research Council; ELN: European LeukaemiaNet; CR: complete remission; CRi: complete remission with incomplete blood count recovery; ITD: internal tandem duplication; TKD: tyrosine kinase domain. aNo data available for 14 patients; b No karyotype was available for 14 patients; cClassification was not possible for 17 patients due to lack of karyotype or FLT3-ITD allelic ratio.

haematologica | 2018; 103(11)

We identified a total of 622 driver mutations affecting 64 genes, with a median number of four mutated genes per patient (range, 1-10 mutations/patient) (Online Supplementary Figure S5). Older age (81-86 years versus 7580 years) was not associated with a higher number of driver gene mutations (P=0.5, data not shown). The most frequently mutated genes in this age group were TET2 (63/151, 42%), DNMT3A (53/151, 35%), NPM1 (48/151, 32%), FLT3 (45/151, 30%), SRSF2 (38/151, 25%), ASXL1 (31/151, 21%), and RUNX1 (28/151, 19%) (Table 1, Figure 2A). Compared to previously studied patients aged <60 or 60-74 years12, those aged ≥75 years had a higher frequency of TET2, SRSF2 and ASXL1 mutations. TP53 mutations occurred in 14% of patients (21/151) and were strongly associated with complex karyotypes (P<0.001). Eightythree percent of patients (126/151) harbored one or more

Figure 1. Overall survival according to the MRC and ELN risk classifications. (A) Overall survival for patients in the favorable- and intermediate-risk groups (green) compared to the adverse-risk group (red) according to the MRC cytogenetic risk category. (B) Overall survival for patients in the favorable-(green), intermediate(orange) and adverse-risk groups (red) according to the ELN 2017 genetic category.

1855

V.V. Prassek et al.

mutations in TET2, DNMT3A, SRSF2, ASXL1, TP53 or SF3B1 - genes known to be involved in age-associated clonal hematopoiesis (Figure 2B).26,27 RUNX1 mutations were positively associated with mutations in SRSF2 and ASXL1 (SRSF2, P=0.02; ASXL1, P=0.01), while ASXL1 and DNMT3A mutations were inversely correlated with each other (P=0.003). NPM1 mutations were mutually exclusive with RUNX1 mutations, and only two patients had co-mutations in NPM1 and ASXL1 (P<0.001) (Figure 2B).

Association between gene mutations, therapy response and survival The number of mutated genes per patient was not associated with OS in a comparison of patients with one to three (62/151, 41%), four to seven (80/151, 53%) or eight to ten (9/151, 6%) mutated genes (P=0.6) (Online Supplementary Figure S6). Univariate analyses of the associations between the most common gene mutations and CR rate and OS are shown in Online Supplementary Table S1. NPM1 mutations, which have been shown to be associated with a higher CR rate in younger AML patients, were not associated with response to induction treatment in our cohort. Patients with NPM1 mutations had a significant longer EFS (P=0.027) (Online Supplementary Figure S7A) and tended to have a longer RFS (P=0.081) (Online Supplementary

Figure S7B) and OS (P=0.090) (Figure 3A) than wildtype patients. Among patients who achieved CR, those with mutated NPM1 had a significantly longer OS (calculated from the day of achieving CR) than NPM1-wildtype patients (P=0.037) (Online Supplementary Figure S7C). Within the subgroup of patients with cytogenetically normal AML, there was no difference in OS between NPM1mutated and -wildtype patients (P=0.4, data not shown). Likewise, the OS of NPM1mutated/FLT3-ITDwildtype patients - a known favorable-risk subgroup, at least in younger patients - was similar to that of the rest of the cohort (P=0.59) (Figure 3B). FLT3-ITD mutations had no impact on EFS, RFS, or OS in the entire cohort [EFS: P=0.54 (Online Supplementary Figure S8A); RFS: P=0.78 (Online Supplementary Figure S8B); OS: P=0.32 (Figure 3C)] or in patients with cytogenetically normal AML (P=0.26) (Online Supplementary Figure S8C). Since the FLT3-ITD allelic ratio is an important prognosticator recognized in the ELN 2017 classification, we also explored its impact on OS. A high FLT3-ITD allelic ratio (≥0.5) was not associated with a shorter OS compared to a low allelic ratio (<0.5) or FLT3-ITD wildtype (OS: P=0.53) (Online Supplementary Figure S8D). In the context of co-mutated NPM1, there was also no impact on OS of a high FLT3-ITD allelic ratio compared to a low allelic ratio or to NPM1mutated/FLT3-ITDwildtype patients (OS: P=0.4)

Figure 2. Genetic landscape of old acute myeloid leukemia patients. (A) Driver gene mutations in 151 AML patients ≥75 years of age at primary diagnosis. The bar chart shows the 15 most commonly mutated genes in the 151 AML patients aged ≥75 years compared to 664 patients aged <60 and 60-74 at primary diagnosis (data from Metzeler et al.12). (B) Heatmap showing associations between different driver gene mutations. Each column represents one patient.

1856

haematologica | 2018; 103(11)

Genetics and outcomes in AML patients ≥75 years

(Online Supplementary Figure 7D). Patients with TP53 mutations showed trends towards shorter EFS (P=0.079) (Online Supplementary Figure S9A) and OS (P=0.073) (Figure 3D), and had a significantly shorter RFS (P=0.041) (Online Supplementary Figure S9B), compared to TP53-wildtype patients. TP53-mutated patients who achieved CR had a significantly shorter OS (calculated from the day of achieving CR) compared to that of TP53-wildtype patients in CR (P=0.005) (Online Supplementary Figure S9C).

IDH1 mutations: impact on event-free and overall survival, association with patients’ pretreatment characteristics and other gene mutations

with cytogenetically normal AML (P=0.069): 10/13 IDH1mutated patients had a normal karyotype, whereas one patient had a 7q deletion, one had trisomy 8, and karyotype was unknown for one patient. Further information on clinical characteristics of the 13 IDH1-mutated patients are shown in Online Supplementary Table S3. IDH1 and IDH2 mutations were mutually exclusive (P=0.2), and IDH2 mutations, found in 15% of patients (23/151), had no impact on OS (P=0.5). TET2 mutations were mutually exclusive with IDH1 mutations (P=0.001), and with mutated IDH2 (P=0.002). Five IDH1-mutated patients also carried NPM1 mutations. While patients with NPM1 mutations overall tended to have favorable OS (Figure 3A), IDH1 mutations had a dominant negative prognostic effect even when cooccurring with mutated NPM1 (Figure 4B). Four IDH1mutated patients had simultaneous FLT3-TKD mutations, an association that almost reached statistical significance (P=0.051). There was no association between IDH1 and FLT3-ITD mutations (P=0.5).

In our cohort of old AML patients, the only gene significantly associated with OS in univariate analysis was IDH1 (Online Supplementary Table S1). None of the IDH1 codon R132-mutated patients (13/151, 9%) reached CR/CRi (P<0.001). Consequently, IDH1-mutated patients had a significantly shorter EFS (P<0.001) (Online Supplementary Figure S10A) and OS (P≤0.001) Figure 4A) than IDH1-wildtype patients. Within this elderly cohort of patients, IDH1-mutated patients showed a trend towards older age (P=0.08) (Online Supplementary Table S2). Furthermore, IDH1mutated patients tended to have higher platelet counts (P=0.20) and lower WBC counts (P=0.10) than IDH1-wildtype patients. There was no difference in peripheral blood or bone marrow blast counts between IDH1-wildtype and IDH1-mutated patients. IDH1 mutations were associated

We used a multivariate Cox regression model to identify the most relevant predictors of OS in this cohort of intensively treated AML patients ≥75 years (Table 2). In this multivariate model, using the MRC classification, ECOG PS, and IDH1, NPM1 and TP53 mutations as covariates, only IDH1 mutations and the MRC adverse cytogenetic

Multivariate analysis of prognostic factors in old, intensively treated acute myeloid leukemia patients

Figure 3. Overall survival according to gene mutations. (A) NPM1 mutations: overall survival in NPM1-mutated (green) compared to NPM1-wildtype patients (red). (B) NPM1-mutated/FLT3-ITD-negative patients: overall survival in NPM1-mutated/FLT3-ITD-wildtype patients (green) compared to other patients (red). (C) FLT3-ITD mutations: overall survival in FLT3-ITD-mutated (red) compared to FLT3-ITD-wildtype patients (green). (D) TP53 mutations: overall survival in TP53-mutated patients (red) compared to TP53-wildtype patients (green). “Mut” denotes mutated and “wt” wildtype.

haematologica | 2018; 103(11)

1857

V.V. Prassek et al.

risk category had significant negative impacts on OS. There was a trend towards worse OS for patients with poor ECOG PS. Patients who were characterized by wildtype IDH1, favorable or intermediate cytogenetics and an ECOG PS of 0-2 had a significantly longer median OS than patients who had at least one high-risk feature (mutated IDH1, adverse cytogenetics, and/or performance status ≥3) (3year OS, 25% versus 5%; P<0.001) (Figure 5).

Association between the European LeukemiaNet 2017 and Medical Research Council classifications and outcomes When we applied the novel ELN 2017 genetic risk classification to our cohort of very old patients, more patients were assigned to the favorable- and adverse-risk categories, and fewer patients to the intermediate-risk group, compared to the MRC classification (Table 1). Surprisingly, patients classified as ELN-2017 intermediaterisk had longer OS (median, 10.7 months) compared to both the favorable- and adverse-risk groups (median, 3.4 months and 3.6 months, respectively; P=0.009) (Figure 1B). Furthermore, in the MRC adverse-risk subgroup, all patients died within 2 years, whereas seven ELN-2017 adverse-risk patients survived for 3 years or longer.

Discussion Elderly AML patients are frequently underrepresented in clinical trials evaluating induction chemotherapy schedules, as well as in studies of the genetic basis of the disease.12,28-30 For example, the widely-used MRC cytogenetic risk categories were derived from a cohort of patients aged 16-59 years.18 This selection bias contrasts with the epidemiology of AML, which is mostly a disease of elderly patients. While many study groups have excluded older patients from trials involving induction chemotherapy, and less-intensive regimens, including the hypomethylating agents, decitabine and azacitidine, are now widely used in this age group, some studies suggest that a subgroup of old patients may benefit from induction chemotherapy.7,8,10 There is, therefore, a vital need to identify factors that are associated with outcomes and that could support therapeutic decision-making in this difficult-to-treat patient cohort. We designed our study to focus on a cohort of very old patients (aged 75 years or older) included in a trial of induction chemotherapy. In this age group, the benefits of induction chemotherapy and potential clinical and genetic markers for therapy success or failure are not well defined. We found that, even among very old patients, favorable-

Table 2. Multivariate analysis for overall survival.

Variable A

MRC cytogenetic risk group (adverse vs. favorable/ intermediate) NPM1-mutated TP53-mutated IDH1-mutated ECOG performance status (3-4 vs. 0-2)

HR (95% CI)

2.21 (1.19 – 4.09)

0.012

0.97 (0.62 – 1.51) 1.32 (0.69 – 2.49) 3.68 (1.87 – 7.23) 1.68 (0.89 – 3.18)

0.889 0.402 <0.001 0.113

HR: hazard ratio; CI: confidence interval; MRC: British Medical Research Council; ECOG: Eastern Cooperative Oncology Group.

Figure 4. IDH1 mutations and survival. (A) Overall survival in IDH1-mutated patients (red line) compared to IDH1-wildtype patients (green line). (B) Overall survival in NPM1-mutated/IDH1-mutated patients (red line) compared to NPM1mutated/IDH1-wildtype patients (green line). Mut: mutated; wt: wildtype.

1858

Figure 5. Impact of Eastern Cooperative Oncology Group performance status, Medical Research Council classification and IDH1 mutations on overall survival. Overall survival in patients with an ECOG performance status of 0-2, favorable or intermediate MRC category and IDH1-wildtype compared to patients with an ECOG performance status of 3-4 or MRC adverse category or IDH1 mutation.

haematologica | 2018; 103(11)

Genetics and outcomes in AML patients ≥75 years

and intermediate-risk cytogenetics remain associated with a relatively favorable OS in a comparison with adverserisk cytogenetics. This finding is in agreement with that of a retrospective analysis by Heiblig and colleagues.31 In their study, older age (≥75 years in a cohort with an age range of 70-93 years) was also a strong prognostic factor in terms of OS whereas according to our own data, higher age (80-86 years versus 75-79 years) did not associate with OS, potentially due to the fact that only very fit octagenarians were enrolled in a trial of induction chemotherapy. Patients in the cohort reported by Heiblig and colleagues received three different treatment modalities: intensive chemotherapy, less intensive treatment (i.e. low-dose AraC, azacitidine or decitabine) or best supportive care only.31 The median OS in patients aged ≥75 years, regardless of therapy type, was 4.3 months with a 3-year OS rate of 14%, compared to a median OS of 10 months and a 3year OS rate of 20% in patients aged <75 years. In patients ≥75 years of age who all received intensive chemotherapy, Heiblig and colleagues reported a 3-year OS rate of 24%, which is comparable to the 3-year OS rate of 21% found in our retrospective analysis. Importantly, the analysis by Heiblig and colleagues also showed that long-term survival beyond 3 years was exceedingly rare in patients treated with hypomethylating agents, while survival times exceeding 10 years were observed in intensively treated patients. Our previous analysis of the mutational landscape in 664 intensively treated AML patients included 288 patients ≥60 years of age at primary diagnosis.12 We reported that the mutational spectrum in older AML patients differed from that in younger patients (<60 years). In this study, we extend this finding to very old patients aged ≥75 years. The five most frequently mutated genes were TET2, DNMT3A, NPM1, SRSF2 and ASXL1.12 The high incidence of TET2, DNMT3A, SRSF2, ASXL1, TP53 and SF3B1 mutations in our patients is in agreement with reports showing that these genes are frequently mutated in age-associated clonal hematopoiesis.26,32 Overall, 83% (126/151) of AML patients ≥75 years carried at least one mutation in one of these six genes. Recently, Baldus and colleagues published data on the genetic and epigenetic landscape in 93 elderly AML patients (65-90 years) enrolled in a Study Alliance Leukemia (SAL) registry.33 Similar to our results, they identified a high frequency of mutations in DNMT3A, TET2, SRSF2, ASXL1 and RUNX1. Strikingly, NPM1 mutations were much less common (16.1% of patients) than in our cohort (32%). This difference remains unexplained, since there are no obvious differences regarding, for example, the proportion of patients with de novo versus secondary AML which could affect the frequency of NPM1 mutations. We unexpectedly identified IDH1 mutations as the strongest genetic predictor of shorter survival in this age group. The prognostic value of IDH1 and IDH2 mutations in AML has been studied by multiple groups but is still controversial.34,35 While some analyses found no impact on OS,36 others demonstrated associations with poor37-40 or favorable prognosis.28 A recently published meta-analysis of 33 studies found that IDH1 and IDH2 mutations, analyzed jointly, do not affect OS or EFS.34 IDH1 mutations, when analyzed separately, are associated with lower CR rates, shorter EFS and shorter OS. A negative impact of IDH1 mutations on outcomes has also been described in haematologica | 2018; 103(11)

younger NPM1-mutated/FLT3-ITD-negative patients.37,39 These findings are generally compatible with the results of our study. This meta-analysis also revealed that IDH2mutated patients had a longer OS than IDH2-wildtype patients, while in our study there was no impact of IDH2 mutations on OS (P=0.53). This discrepancy may be explained by our focus on very old AML patients, among whom the prognostic relevance of gene mutations may differ from that among younger patients. It remains unclear why mutations in two related genes with apparently similar functional consequences21,41,42 might have varying or even opposite impacts on OS. The poor outcome of old AML patients with mutated IDH1 needs to be validated in additional cohorts. If the negative prognostic effect of IDH1 mutations in old patients is confirmed in other studies, these patients should not be considered candidates for induction chemotherapy. Targeted inhibitors of the mutant IDH1 enzyme, which are currently being tested in clinical trials, may become a preferred therapeutic approach for these patients in the future. In our multivariate analysis, results of cytogenetic analysis (classified according to the MRC system) were a strong predictor of outcomes. Compared to the MRC risk categories, application of the new ELN-2017 genetic risk classification led to an expansion of the favorable- and adverse-risk categories (Table 1). The inclusion of RUNX1, ASXL1 and TP53 mutations in the ELN adverse-risk category, in particular, affects the risk classification of this cohort of very old patients in whom these mutations are common. Many patients who were classified as intermediate-risk based on cytogenetic results alone according to the MRC recommendations are now re-classified into the adverse-risk group. At least in our cohort, however, the MRC cytogenetic risk classification appears to provide better prognostic stratification compared to the ELN-2017 system. Thus, further validation of the ELN-2017 classification in larger cohorts of elderly patients seems warranted to define its applicability in this age group. Our finding that in patients aged 75 or older, the prognosis of ELN-2017 favorable-risk patients was not better, and may in fact be worse, than that of the intermediaterisk group suggests that favorable molecular markers established in younger patients (e.g., mutated NPM1 without or with a low FLT3-ITD allelic ratio, biallelic CEBPA mutations) have weaker prognostic relevance in elderly patients. Indeed, in our cohort, the NPM1-mutated/FLT3-ITD-negative genotype did not associate with favorable outcomes. NPM1 mutations alone were associated with significantly longer EFS, but there were only trends towards longer RFS and OS. Among patients who achieved CR after induction therapy, those with mutated NPM1 had a significantly longer OS, suggesting that these patients might have benefited from cytarabine-based consolidation and prolonged monthly maintenance therapy as used in the AML-CG 1999 trial. In contrast, the MRC adverse-risk group, defined by cytogenetic alterations only, had significantly shorter OS compared to the cytogenetic favorable- and intermediate-risk groups. In contrast to the ELN-2017 adverse-risk group, no MRC adverse-risk patient survived beyond 2.1 years. Thus, karyotype rather than gene mutations appears to be the major factor defining the prognosis in very old, intensively treated AML patients.43 One limitation of our study is that all patients were treated with the same treatment modality - intensive 1859

V.V. Prassek et al.

chemotherapy. Consequently, we were unable to study interactions between genetic alterations and type of treatment (intensive versus not intensive) with regard to patientsâ&#x20AC;&#x2122; outcomes. However, we found that patients randomized to the more intensive induction regimen (HAM) tended to have longer OS compared to those randomized to TAD. While this non-significant difference may be due to the play of chance, the finding that better outcomes were observed in the more intensive arm supports the conclusion that induction chemotherapy is tolerable for selected patients in this age group. Two randomized phase III trials in newly diagnosed AML patients aged â&#x2030;Ľ65 years compared treatment with the hypomethylating agents, azacitidine and decitabine, to alternative therapies (best supportive care, low-dose cytarabine, or intensive chemotherapy).44,45 The median OS of azacitidine-treated patients was 10.4 months, and that of decitabine-treated patients was 7.7 months, compared to a median OS of 6.0 months in our study. Obviously, these cohorts of patients cannot be compared directly because of their different age ranges and inclusion criteria. While the lower age limit of our analysis was higher than that in the trials of hypmethylating agents (75 versus 65 years), our cohort represents highly selected patients who were, despite their advanced age, deemed fit enough to undergo induction chemotherapy. Notwhitstanding this obvious limitation, it is noteworthy that the 3-year OS rate in our cohort was nominally higher than the rates in both the trials of hypomethylating agents. In our study, 24 patients (21%)

References 1. Dohner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129(4):424-447. 2. Hiddemann W, Kern W, Schoch C, et al. Management of acute myeloid leukemia in elderly patients. J Clin Oncol. 1999;17(11):3569-3576. 3. Appelbaum FR, Gundacker H, Head DR, et al. Age and acute myeloid leukemia. Blood. 2006;107(9):3481-3485. 4. Juliusson G, Billstrom R, Gruber A, et al. Attitude towards remission induction for elderly patients with acute myeloid leukemia influences survival. Leukemia. 2006;20(1):42-47. 5. Burnett AK, Milligan D, Prentice AG, et al. A comparison of low-dose cytarabine and hydroxyurea with or without all-trans retinoic acid for acute myeloid leukemia and high-risk myelodysplastic syndrome in patients not considered fit for intensive treatment. Cancer. 2007;109(6):1114-1124. 6. Rao AV. Fitness in the elderly: how to make decisions regarding acute myeloid leukemia induction. Hematology Am Soc Hematol Educ Program. 2016;2016(1):339-347. 7. Wetzler M, Mrozek K, Kohlschmidt J, et al. Intensive induction is effective in selected octogenarian acute myeloid leukemia patients: prognostic significance of karyotype and selected molecular markers used in the European LeukemiaNet classification. Haematologica. 2014;99(2):308-313. 8. Lowenberg B, Zittoun R, Kerkhofs H, et al.

1860

10.

11. 12.

13.

14.

15.

were still alive 3 years after primary diagnosis, whereas two patients (1%) in the azacitidine and no patient in the decitabine trial were still alive at 3 years. Thus, at least a subset of AML patients may benefit from intensive induction therapy in terms of sustained remissions and longterm survival, even at the age of 75 years or above. Nevertheless, the longer median OS achieved in the trials of hypomethylating agents suggests that these agents may be the preferred choice for many elderly patients. In clinical practice, the choice between intensive, potentially curative chemotherapy or a palliative therapeutic approach in old AML patients should be guided by an evidence-based, individualized assessment of the potential risks (e.g., treatment-related mortality) and benefits of intensive chemotherapy. In our study, favorable- and intermediate-risk cytogenetics, IDH1-wildtype status and good performance status characterized a subset of old AML patients who may achieve long-term survival after induction chemotherapy. These patients might be candidates for intensive therapy in the absence of medical contraindications. On the other hand, patients with adverserisk cytogenetics, IDH1 mutations or poor performance status did not benefit from intensive induction therapy and might fare better with alternative strategies such as hypomethylating agents, or novel targeted agents that have recently been approved or are currently under development.46 In summary, our results show that clinicians should consider intensive chemotherapy as a therapeutic option even for selected, very old AML patients.

On the value of intensive remission-induction chemotherapy in elderly patients of 65+ years with acute myeloid leukemia: a randomized phase III study of the European Organization for Research and Treatment of Cancer Leukemia Group. J Clin Oncol. 1989;7(9):1268-1274. Juliusson G, Swedish AMLG. Most 70- to 79-year-old patients with acute myeloid leukemia do benefit from intensive treatment. Blood. 2011;117(12):3473-3474. Juliusson G, Antunovic P, Derolf A, et al. Age and acute myeloid leukemia: real world data on decision to treat and outcomes from the Swedish Acute Leukemia Registry. Blood. 2009;113(18):4179-4187. Krug U, Gale RP, Berdel WE, et al. Therapy of older persons with acute myeloid leukaemia. Leuk Res. 2017;60:1-10. Metzeler KH, Herold T, Rothenberg-Thurley M, et al. Spectrum and prognostic relevance of driver gene mutations in acute myeloid leukemia. Blood. 2016;128(5):686-698. 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. Ferrara F, Criscuolo C, Riccardi C, et al. FLT3 mutations have no prognostic impact in elderly patients with acute myeloid leukemia and normal karyotype. Am J Hematol. 2009;84(8):532-535. Mrozek K, Marcucci G, Nicolet D, et al. Prognostic significance of the European LeukemiaNet standardized system for reporting cytogenetic and molecular alterations in adults with acute myeloid leukemia. J Clin Oncol. 2012;30(36):4515-

4523. 16. Andersson A, Johansson B, Lassen C, Mitelman F, Billstrom R, Fioretos T. Clinical impact of internal tandem duplications and activating point mutations in FLT3 in acute myeloid leukemia in elderly patients. Eur J Haematol. 2004;72(5):307-313. 17. Whitman SP, Maharry K, Radmacher MD, et al. FLT3 internal tandem duplication associates with adverse outcome and gene- and microRNA-expression signatures in patients 60 years of age or older with primary cytogenetically normal acute myeloid leukemia: a Cancer and Leukemia Group B study. Blood. 2010;116(18):3622-3626. 18. Grimwade D, Hills RK, Moorman AV, et al. Refinement of cytogenetic classification in acute myeloid leukemia: determination of prognostic significance of rare recurring chromosomal abnormalities among 5876 younger adult patients treated in the United Kingdom Medical Research Council trials. Blood. 2010;116(3):354-365. 19. Forbes SA, Beare D, Gunasekaran P, et al. COSMIC: exploring the world's knowledge of somatic mutations in human cancer. Nucleic Acids Res. 2015;43(Database issue):D805-811. 20. Sherry ST, Ward MH, Kholodov M, et al. dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 2001;29(1):308-311. 21. Cancer Genome Atlas Research N. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):2059-2074. 22. Schnittger S, Schoch C, Kern W, et al. Nucleophosmin gene mutations are predictors of favorable prognosis in acute myeloge-

haematologica | 2018; 103(11)

Genetics and outcomes in AML patients â&#x2030;Ľ75 years

23.

24.

25.

26.

27.

28.

29.

30.

31.

nous leukemia with a normal karyotype. Blood. 2005;106(12):3733-3739. Kiyoi H, Naoe T, Yokota S, et al. Internal tandem duplication of FLT3 associated with leukocytosis in acute promyelocytic leukemia. Leukemia Study Group of the Ministry of Health and Welfare (Kohseisho). Leukemia. 1997;11(9):1447-1452. Benthaus T, Schneider F, Mellert G, et al. Rapid and sensitive screening for CEBPA mutations in acute myeloid leukaemia. Br J Haematol. 2008;143(2):230-239. Krug U, Berdel WE, Gale RP, et al. Increasing intensity of therapies assigned at diagnosis does not improve survival of adults with acute myeloid leukemia. Leukemia. 2016;30(6):1230-1236. 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. Jaiswal S, Fontanillas P, Flannick J, et al. Agerelated clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371(26):2488-2498. Patel JP, Gonen M, Figueroa ME, et al. Prognostic relevance of integrated genetic profiling in acute myeloid leukemia. N Engl J Med. 2012;366(12):1079-1089. Kihara R, Nagata Y, Kiyoi H, et al. Comprehensive analysis of genetic alterations and their prognostic impacts in adult acute myeloid leukemia patients. Leukemia. 2014;28(8):1586-1595. Nazha A, Ravandi F. Acute myeloid leukemia in the elderly: do we know who should be treated and how? Leuk Lymphoma. 2014;55(5):979-987. Heiblig M, Le Jeune C, Elhamri M, et al. Treatment patterns and comparative effectiveness in elderly acute myeloid leukemia patients (age 70 years or older): the Lyonuniversity hospital experience. Leuk

haematologica | 2018; 103(11)

Lymphoma. 2017;58(1):110-117. 32. Bullinger L, Dohner K, Dohner H. Genomics of acute myeloid leukemia diagnosis and pathways. J Clin Oncol. 2017;35(9):934-946. 33. Silva P, Neumann M, Schroeder MP, et al. Acute myeloid leukemia in the elderly is characterized by a distinct genetic and epigenetic landscape. Leukemia. 2017;31(7): 1640-1644. 34. Xu Q, Li Y, Lv N, et al. Correlation between isocitrate dehydrogenase gene aberrations and prognosis of patients with acute myeloid leukemia: a systematic review and meta-analysis. Clin Cancer Res. 2017;23(15): 4511-4522. 35. Green CL, Evans CM, Hills RK, Burnett AK, Linch DC, Gale RE. The prognostic significance of IDH1 mutations in younger adult patients with acute myeloid leukemia is dependent on FLT3/ITD status. Blood. 2010;116(15):2779-2782. 36. 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. 37. 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. 38. Yamaguchi S, Iwanaga E, Tokunaga K, et al. IDH1 and IDH2 mutations confer an adverse effect in patients with acute myeloid leukemia lacking the NPM1 mutation. Eur J Haematol. 2014;92(6):471-477. 39. Boissel N, Nibourel O, Renneville A, et al. Prognostic impact of isocitrate dehydrogenase enzyme isoforms 1 and 2 mutations in acute myeloid leukemia: a study by the Acute Leukemia French Association group. J

Clin Oncol. 2010;28(23):3717-3723. 40. Abbas S, Lugthart S, Kavelaars FG, et al. Acquired mutations in the genes encoding IDH1 and IDH2 both are recurrent aberrations in acute myeloid leukemia: prevalence and prognostic value. Blood. 2010;116(12): 2122-2126. 41. Figueroa ME, Abdel-Wahab O, Lu C, et al. Leukemic IDH1 and IDH2 mutations result in a hypermethylation phenotype, disrupt TET2 function, and impair hematopoietic differentiation. Cancer Cell. 2010;18(6):553567. 42. Im AP, Sehgal AR, Carroll MP, et al. DNMT3A and IDH mutations in acute myeloid leukemia and other myeloid malignancies: associations with prognosis and potential treatment strategies. Leukemia. 2014;28(9):1774-1783. 43. Grimwade D, Walker H, Harrison G, et al. The predictive value of hierarchical cytogenetic classification in older adults with acute myeloid leukemia (AML): analysis of 1065 patients entered into the United Kingdom Medical Research Council AML11 trial. Blood. 2001;98(5):1312-1320. 44. 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. 45. 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. 46. Wei AH, Tiong IS. Midostaurin, enasidenib, CPX-351, gemtuzumab ozogamicin, and venetoclax bring new hope to AML. Blood. 2017;130(23):2469-2474.

1861

ARTICLE

Acute Myeloid Leukemia

Ferrata Storti Foundation

MDM2- and FLT3-inhibitors in the treatment of FLT3-ITD acute myeloid leukemia, specificity and efficacy of NVP-HDM201 and midostaurin

Katja Seipel,1,2 Miguel A.T. Marques,1 Corinne Sidler,1 Beatrice U. Mueller2 and Thomas Pabst2

Haematologica 2018 Volume 103(11):1862-1872

1 Department for Biomedical Research, University of Bern and 2Department of Medical Oncology, Inselspital, Bern University Hospital, Switzerland

ABSTRACT

Correspondence: thomas.pabst@insel.ch

Received: February 20, 2018. Accepted: June 29, 2018. Pre-published: July 5, 2018. doi:10.3324/haematol.2018.191650 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1862

rognosis for FLT3-ITD positive acute myeloid leukemia with high allelic ratio (>0.5) is poor, particularly in relapse, refractory to or unfit for intensive treatment, thus highlighting an unmet need for novel therapeutic approaches. The combined use of compounds targeting both the mutated FLT3 receptor and cellular p53 inhibitors might be a promising treatment option for this poor risk leukemia subset. We therefore assessed MDM2 and FLT3 inhibitors as well as cytotoxic compounds used for conventional induction treatment as single agents and in combination for their ability to induce apoptosis and cell death in leukemic cells. Acute myeloid leukemia cells represented all major morphologic and molecular subtypes with normal karyotype, including FLT3-ITD (>0.5) and FLT3 wild type, NPM1 mutant and NPM1 wild type, as well as TP53 mutant and TP53 wild type cell lines. Acute myeloid leukemia cells with mutated or deleted TP53 were resistant to MDM2- and FLT3-inhibitors. FLT3-ITD positive TP53 wild type acute myeloid leukemia cells were significantly more susceptible to FLT3-inhibitors than FLT3-ITD negative TP53 wild type cells. The presence of a NPM1 mutation reduced the susceptibility of TP53 wild type acute myeloid leukemia cells to the MDM2 inhibitor NVP-HDM201. Moreover, the combined use of MDM2- and FLT3-inhibitors was superior to single agent treatment, and the combination of midostaurin and NVP-HDM201 was as specific and effective against FLT3-ITD positive TP53 wild type cells as the combination of midostaurin with conventional induction therapy. In summary, the combined use of the MDM2 inhibitor NVP-HDM201 and the FLT3 inhibitor midostaurin was a most effective and specific treatment to target TP53 and NPM1 wild type acute myeloid leukemia cells with high allelic FLT3-ITD ratio. These data suggest that the combined use of NVP-HDM201 and midostaurin might be a promising treatment option particularly in FLT3-ITD positive acute myeloid leukemia relapsed or refractory to conventional therapy.

ÂŠ2018 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.

1862

Introduction Acute myeloid leukemia (AML) is a clonal hematopoietic disorder characterized by blocked differentiation and deregulated proliferation of hematopoietic precursor cells. At the cellular level, specific genetic and epigenetic alterations lead to changes in cellular signaling pathways including the common inactivation of the p53 tumor suppressor axis, and thereby contribute to blocked differentiation and accumulation of leukemic blasts in the blood and the bone marrow. The past decade has witnessed major advances in our comprehension of the biologic heterogeneity of AML.1 AML genetic variants are assigned into favorable, intermediate and poor risk categories, and a major molecular subgroup within the poor risk AML is characterized by genetic alterations of the FLT3 receptor gene. FLT3 internal tanhaematologica | 2018; 103(11)

Targeting FLT3-ITD in AML

dem duplications (FLT3-ITD) are the most common mutations in the FLT3 receptor gene. FLT3-mutated AML account for 25-35% of all AML, and their prognosis is poor, particularly in unfit, refractory or relapsed patients. Targeting the mutated FLT3 receptor is a promising approach to treat this specific AML subset. Midostaurin (PKC412) is a first generation type III receptor tyrosine kinase inhibitor that has been extensively studied in vitro and in clinical trials as a treatment for AML patients with mutated FLT3.2,3 After successful phase II clinical trials, midostaurin was found to significantly prolong survival of FLT3-mutated AML patients when combined with conventional induction and consolidation therapies in a randomized phase III clinical trial leading to the first new drug approval in AML in over 40 years.4 Midostaurin is a multi-targeted kinase inhibitor able to block FLT3 autophosphorylation and to induce growth arrest and apoptosis in FLT3-dependent leukemia.5 Midostaurin is orally administered and generally well tolerated as a single agent. Quizartinib (ACC220) and gilteritinib (ASP2215) are second and third generation FLT3 inhibitors currently in evaluation for the treatment of FLT3-mutated AML.6-8 Targeting the p53 antagonist MDM2 is a novel approach to restore the crucial p53 tumor suppressor function in AML cells.9 Idasanutlin (RG7833) is a second generation MDM2 inhibitor that has been studied in vitro and in vivo as a treatment for AML patients with wild type TP53.10 NVP-CGM09711 and NVP-HDM20112 are second generation MDM2 inhibitors that are currently evaluated in single-agent phase I studies in patients with advanced tumors with wild type TP53 (clinicaltrials.gov identifiers 01760525 and 02143635). Like midostaurin, NVP-HDM201 is orally administered and expected to be well tolerated as single agent. In this study, we investigated the combined treatment with MDM2- and FLT3- inhibitors, in particular NVPHDM201 and midostaurin, on AML cells in order to identify a potentially effective treatment specifically for FLT3-ITD AML refractory to or unfit for intensive chemotherapy. The study might provide the rationale for initiating a clinical study in FLT3-ITD AML evaluating this combination.

Methods Patient samples Mononuclear cells of AML patients diagnosed and treated at the University Hospital, Bern, Switzerland between 2005 and 2015 were included in this study. Informed consent from all patients was obtained according to the Declaration of Helsinki, and the studies were approved by decisions of the local ethics committee of Bern, Switzerland. Mutational screening for FLT3, NPM1, TP53 and conventional karyotype analysis of at least 20 metaphases were performed for each patient. Peripheral blood mononuclear cells (PBMCs) and bone marrow mononuclear cells (BMMCs) were collected at the time of diagnosis before initiation of treatment.

AML cell lines OCI-AML2 (AML-M4, FLT3wt, DNMT3A R882C, NPM1wt, TP53wt); OCI-AML3 (AML-M4, FLT3wt, DNMT3A R882C, NPM1mut, TP53wt), MOLM-13 (AML-M5, t(9;11), FLT3-ITD, TP53wt), MOLM-16 (AML-M0, FLT3wt, TP53mut), MV4-11 (AML-M5, t(4;11), FLT3-ITD, TP53wt), ML-2 (AML-M4, t(6;11), FLT3wt, TP53mut), PL-21 (AML-M3, FLT3wt, TP53hemi) and HL-60 (AML-M2, FLT3wt, TP53null) cells were supplied by the Leibniz Institute DSMZ, German Collection of Microorganisms and Cell Cultures. AML cells were grown in RPMI 1640 (SIGMA-ALDRICH, St. Louis, MO, USA) supplemented with 20% fetal bovine serum (FBS, Biochrom GmbH, Germany).

Cytotoxicity assays AML cells were treated with the MDM2 inhibitors NVPHDM201, NVP-CGM097, idasanutlin (RG7388), the FLT3 inhibitors midostaurin (PKC412), quizartinib (ACC220), gilteritinib (ASP2215) or with genotoxic compounds cytarabine and idarubicin in equimolar concentrations. NVP-HDM201 and NVP-CGM097 investigational compounds were supplied by Novartis, Switzerland, whereas RG7833, PKC412, ACC220 and ASP2215 were purchased at MCE (MedChemExpress, Monmouth Junction, NJ, USA). Cytarabine and idarubicin were purchased at Sigma-Aldrich (St.Louis, MO, USA) and SelleckChem (Houston, TX, USA). Cell viability was determined using the MTT-based in vitro toxicology assay (TOX1, Sigma-Aldrich) with four repeat measurements per dosage. Data

Table 1. Genetic variants in AML cell lines.

FLT3

TP53

NPM1

mutated genes

HL-60

del

MOLM-13 MOLM-16

ITD wt

wt V173M C238S

wt wt

MV4-11 OCI-AML2

ITD wt A680V wt

wt wt

L287fs

wt P336L

wt P36fs

NRAS Q61L CDKN2A R80X MLL-AF9 (t9;11) MLL V1368L MTOR T571K MLL-ENL (t4;11) DNMT3A R635W MLL K1751* DNMT3A R882C NRAS Q61L KRAS A146V

OCI-AML3 PL-21

wt: wild type; ITD: internal tandem duplication; del: deletion

haematologica | 2018; 103(11)

1863

K. Seipel et al.

are depicted as XY graphs with median and interquartile range, as box plots or scatter plots with mean values. Statistical analysis was done on GraphPad Prism (version 7, GraphPad software, LaJolla, CA, USA) in grouped analysis and significance calculated by Mann-Whitney test. Combination indexes were calculated on CompuSyn software (version 1.0; ComboSyn, Inc. Paramus, NJ,USA).

Measurement of mRNA expression by qPCR RNA was extracted from AML cells and quantified using qPCR.

The RNA extraction kit was supplied by Macherey-Nagel, DĂźren, Germany. Reverse transcription was done with MMLV-RT (Promega, Madison, WI, USA). Real-time PCR was performed on the ABI7500 Real-Time PCR Instrument using ABI universal master mix (Applied Biosystems, Austin, TX, USA) and gene specific probes Hs00355782_m1 (CDKN1A), Hs01050896_m1 (MCL1) and Hs02758991_g1 (GAPDH) (ThermoFischer Scientific, Waltham, MA, USA). Measurements of CDKN1A and MCL1 expression were normalized with GAPDH values (ddCt relative quantitation). Assays were performed in three or more independ-

Figure 1. Variable responses of AML cell lines to FLT3 and MDM2 inhibitors. Dose response curves in AML cell lines treated with FLT3 inhibitors (A,B,C) and MDM2 inhibitors (D,E,F) as single compound treatment with midostaurin (PKC412) (A), quizartinib (AC220) (B), gilteritinib (ASP2215) (C), idasanutlin (RG7833) (D), NVPCGM097 (E) or NVP-HDM201 (F), in a variety of AML cell lines (G) and combination treatments with NVP-HDM201 and PKC412, ACC220 or ASP2215 in MOLM-13 cells (H). Combination indexes were calculated according to Chou Talalay.42

1864

haematologica | 2018; 103(11)

Targeting FLT3-ITD in AML

Results

ent experiments. Statistical analysis was done on GraphPad Prism software using two-tailed t-tests (version 7, GraphPad software, LaJolla, CA, USA). Data are depicted in column bar graphs plotting mean with SD values.

Sensitivity of AML cell lines to MDM2 and FLT3 inhibitors To determine the sensitivity of AML cells to MDM2 and FLT3 inhibitors, AML cell lines were treated with three MDM2- and three FLT3-inhibitors for 24 hours in dose escalation experiments before cell viability assessment. The AML cell lines covered the major morphologic and molecular subtypes including particularly FLT3-ITD and FLT3 wild type, NPM1 mutant and wild type, as well as TP53 wild type, mutant, hemizygous and null cells (Table I). The two FLT3-ITD cell lines MV4-11 and MOLM-13 had high allelic ratios of FLT3-ITD and chromosomal

Antibodies and cytometry Staining for apoptosis was done using AnnexinV-CF488A (Biotium, Germany) in AnnexinV buffer and Hoechst 33342 (10 mg/ml) for 15 min. at 37C, followed by several washes. Propidium iodide was added shortly before imaging on the Nucleocounter NC-3000 (ChemoMetec, Allerod, Denmark). For cell cycle analysis cells were incubated in lysis buffer with DAPI (10 mg/ml) for 5 min. at 37Â°C and analyzed on NC-3000 imager.

Figure 2. Variable responses of AML blast cells to midostaurin and HDM201. Cell viability was determined in AML patient cells treated with midostaurin (PKC412) (A) or NVP-HDM201 (B). AML cells were grouped according to major molecular subtypes (FLT3/TP53/NPM1). Cell viability measurements are depicted in dose response curves (A, B) or with 50nM single compound (C, D), as well as combination treatment of midostaurin with conventional induction therapy (E) or midostaurin with NVP-HDM201 (F). PBM are peripheral blood monocytes of normal controls. AML patient samples were analyzed in groups of at least four individual samples (Online Supplementary Table S1) using GraphPad prism software. Significance is denoted for P<0.05 (*); P<0.005 (**); P<0.0005 (***); P<0.0001 (****); P>0.05 (ns).

haematologica | 2018; 103(11)

1865

K. Seipel et al.

translocations with MLL gene rearrangements. Some cell lines contained additional mutations in driver genes such as DNMT3A (OCI-AML2, OCI-AML3) and RAS genes (OCI-AML3, PL-21). DNMT3a and RAS gene mutations may influence sensitivity to MDM2 or FLT3 inhibitors. The MDM2 inhibitors included idasanutlin (RG7388), NVP-CGM097 and NVP-HDM201. The FLT3 inhibitors included the 1st, 2nd and 3rd generation inhibitors midostaurin (PKC412), quizartinib (ACC220) and gilteritinib (ASP2215). The FLT3-ITD positive and TP53 wild type cell lines MOLM-13 and MV4-11 were most susceptible to all three FLT3- and all three MDM2-inhibitors (Figure 1). The effects on MOLM-13 cell survival induced by the three FLT3 inhibitors were consistent with IC50 values of 200nM (Figure 1A,B,C). Quizartinib had a greater potency with 70% cell viability after treatment with 10nM compound, while midostaurin and gilteritinib had comparable potencies with 90% cell viability at 10nM compound. OCI-AML2 and PL-21 cells showed some response to midostaurin while OCI-AML3, MOLM-16 and HL-60 cells were rather resistant to midostaurin and quizartinib. With respect to gilteritinib, however, OCI-AML3, PL-21 and HL-60 cells showed some response and only MOLM16 cells were resistant. The effects on MOLM-13 cell survival varied for the MDM2 inhibitors with IC50 values of 300nM NVP-HDM201, 1 mM RG7388 and 10 mM NVPCGM097 (Figure 1D,E,F). MOLM-13 cells were most susceptible to MDM2 inhibitor idasanutlin with IC50 at 1 mM RG7388. MV4-11 and OCI-AML2 cells showed some response to idasanutlin with IC50 of 10 mM RG7388 while OCI-AML3, PL-21, HL-60 and MOLM-16 cells were rather resistant to idasanutlin. With respect to the MDM2 inhibitors NVP-CGM097 and NVP-HDM201 MOLM-13, MV4-11 and OCI-AML2 cells showed consistent susceptible responses while OCI-AML3 and PL-21 cells were less susceptible and MOLM-16 and HL60 cells were resistant. In order to define the most effective treatment combination we focused our studies on the latest and most potent MDM2 inhibitor NVP-HDM201 and tested its effects in single agent treatment and together with the three FLT3inhibitor compounds in MOLM-13 cells. The combination of NVP-HDM201 and midostaurin had excellent synergistic effects on cell survival with a combination index of 0.4, while the combination of NVP-HDM201 with quizartinib or gilteritinib had only moderate synergistic effects with combination indexes of 0.7 and 0.8 (Figure 1H). To determine the relevance of the order of addition, midostaurin and NVP-HDM201 were tested as direct combination or sequential treatment and found to be effective independent of sequence of application. NVP-HDM201 pretreatment followed by midostaurin treatment had similar effects on cell viability as midostaurin pretreatment followed by NVP-HDM201 treatment. Moreover, both sequential treatments had comparable effects on cell viability as direct combination treatment (Online Supplementary Figure S1).

Sensitivity of AML patient cells to the MDM2 inhibitor HDM201 and the FLT3 inhibitor midostaurin To determine the sensitivity of NK-AML blast cells to HDM201 and midostaurin, mononuclear cells isolated from peripheral blood or bone marrow of NK-AML patients were subjected to in vitro cytotoxicity assays. The NK-AML cells covered the major morphologic and molecular subtypes including FLT3-ITD and FLT3 wild type, 1866

NPM1 mutant and wild type, as well as TP53 mutant and wild type cells (Online Supplementary Table S1). Most of the FLT3-ITD AML cells had a high allelic ratio of FLT3-ITD (>0.5). Only few of the patient samples contained additional mutations in driver genes, one with DNMT3A, one with RAS mutations. Samples of AML blast cells were grouped according to the major molecular subtypes (FLT3/TP53/ NPM1) and comprised at least four samples per molecular genetic combination, with median 84% blast cells ranging from 25 to 95% (Online Supplementary Table S1). Both midostaurin and NVP-HDM201 used as single agent treatment induced varying levels of loss in cell viability with correlation to certain AML subsets (Figure 2). Data for MOLM-13 and MV4-11 cell lines were includ-

Figure 3. Synergistic responses in FLT3-ITD AML blast cells to midostaurin and HDM201. Cell viability was determined in individual AML patient cells treated with increasing dosage of NVP-HDM201 as single compound and in combination treatment with midostaurin (PKC412) in FLT3-ITD/TP53wt/NPM1wt relapsed AML (patient 13) (A), FLT3-ITD/TP53wt/NPM1wt primary AML (patient 16) (B) and in FLT3-ITD/TP53wt/NPM1mut primary AML (patient 20) (C). Combination indexes were calculated according to Chou Talalay.42

haematologica | 2018; 103(11)

Targeting FLT3-ITD in AML

ed as, compared to primary FLT3-ITD AML cells with 80% blast cells, FLT3-ITD AML cell lines with 100% blast cells are more susceptible to midostaurin (Figure 2A), and less susceptible to NVP-HDM201 (Figure 2B). With respect to midostaurin, FLT3-ITD/TP53wt NKAML cells were distinctly more susceptible than FLT3ITD/TP53wt cells. Patient derived AML blast cells characterized by FLT3-ITD/TP53wt/NPM1wt were susceptible to midostaurin with a median loss of 30% viability after treatment with 500nM PKC412. FLT3-ITD/TP53wt/NPM1wt cells with 11q23/MLL abnormalities, MOLM-13 (t(9;11) and MV4-11 (t(4;11) lost 60% cell viability within 24 hours when treated with 500nM midostaurin. All other AML cells including

FLT3-ITD/TP53wt/ NPM1mut, FLT3wt and TP53mut cells were less susceptible to midostaurin with 0-10% reduced viability at 500nM midostaurin (Figure 2A). With respect to NVP-HDM201, we observed that the same NK-AML blast cells characterized by FLT3ITD/TP53wt/NPM1wt and sensitive to midostaurin were most susceptible to the MDM2 inhibitor NVP-HDM201, with a median loss of 45% viability within 24 hours at 100nM NVP-HDM201. MOLM-13 and MV4-11 cells were less susceptible with a loss of 20% viability at 100nM NVP-HDM201. FLT3-ITD/TP53wt/NPM1wt cells responded with a median 20% loss of viability and FLT3ITD/TP53wt/NPM1wt with median 10% loss of viability at 100nM NVP-HDM201. All other AML cells including

Figure 4. Dose-dependent induction of apoptosis and cell death in FLT3-ITD AML cells. Induction of tumor suppressor protein p53 in MV4-11 (A) and MOLM-13 cells (B) treated for 24 hours with the indicated amounts of NVP-HDM201 and midostaurin. Relative quantitation of CDKN1A mRNA (C) and MCL-1 mRNA (D) in AML cells treated for 24 hours with midostaurin (PKC412) (black bars) and NVP-HDM201 (light grey bars) alone or in combination (dark grey bars). Cytometric assays in MV411 AML cells treated with NVP-HDM201 and midostaurin alone and in combination to measure induction of cell death (subG1 fraction) using DAPI staining (E) and induction of apoptosis using AnnexinV/PI staining (F).

haematologica | 2018; 103(11)

1867

K. Seipel et al.

FLT3-ITD/TP53wt/NPM1wt and TP53mut cells were minimally susceptible to NVP-HDM201 with 1-5% reduced viability when treated with 1 mM compound (Figure 2B). In single agent treatments with 50nM compound, FLT3ITD/TP53wt/NPM1wt cells were significantly more susceptible to midostaurin and NVP-HDM201 than FLT3ITD/TP53wt/NPM1mut and FLT3wt cells (Figure 2C,D). FLT3-ITD/TP53wt/NPM1wt cells lost more than 20% cell viability within 24hours at 50nM PKC412 while all other AML cells and normal peripheral blood monocytes were unaffected by this low dose treatment (Figure 2C). Similarly, FLT3-ITD/TP53wt/NPM1wt cells lost 30% cell viability within 24 hours at 50nM NVP-HDM201 while all other AML cells were significantly less affected by this low dose treatment and TP53mut AML cells as well as normal peripheral blood monocytes were unaffected (Figure 2D). These data propose that the ideal target population for the treatment with midostaurin and NVP-HDM201 are FLT3-ITD NK-AML cells with a high allelic ratio of FLT3ITD that are wild type for TP53 and NPM1.

Specificity and efficacy of combined HDM201 and midostaurin against FLT3-ITD/TP53wt/NPM1wt AML cells The response of AML cells to 50nM midostaurin in combination with conventional induction therapy (CI, 20nM cytarabine and 20nM idarubicin) or in combination with 50nM NVP-HDM201 was determined by in vitro cytotoxicity assays. Similar to the single agent treatments reported above, FLT3-ITD/TP53wt NK-AML cells were most susceptible to the combined treatment whereas FLT3-ITD/TP53mut cells turned out to be resistant and FLT3-ITD/TP53wt cells showed intermediate responses (Figure 2E,F). The combination of midostaurin with conventional induction treatment had significant effects on TP53wt AML cells with 30-40% median loss of cell viability in FLT3-ITD cells and 15-25% reduction in FLT3wt cells exposed to 20nM CI and 50nM midostaurin for 24 hours (Figure 2E). TP53mut AML cells and normal peripheral blood monocytes were affected with 5-10% median losses in cell viability. The effects of conventional induction treatment on cell viability were enhanced by the addition of midostaurin in FLT3-ITD/TP53wt cells independent of their NPM1 status (Figure 2E). The combination of midostaurin with NVP-HDM201 was as effective as the combination of midostaurin with standard induction therapy. As in the single agent treatments, FLT3-ITD/TP53wt/NPM1wt cells were most susceptible to this combination with 40% median loss of cell viability in 24 hours to 50nM NVP-HDM201 and 50nM midostaurin (Figure 2F). FLT3-ITD/TP53wt/NPM1mut and FLT3wt/TP53wt/NPM1wt AML cells were less susceptible with a median reduction of 20% cell viability in this combination treatment. FLT3wt/TP53wt/NPM1mut and TP53mut AML cells as well as normal peripheral blood monocytes were least affected with 3% median losses in cell viability. The combination of NVP-HDM201 and midostaurin had synergistic effects on cell survival of FLT3-ITD positive AML cells with a combination index of 0.63 in the relapsed AML patient #13 (Figure 3A), and moderate synergistic effects in FLT3-ITD positive primary AML cells (Figure 3B,C). FLT3-ITD/TP53wt/NPM1wt pri1868

mary AML cells (patient #16) were most susceptible to this combination treatment (Figure 3B) with reduced responses in FLT3-ITD/TP53wt/NPM1wt (patient#20) primary AML (Figure 3C). To confirm p53 activation in the presence of MDM2 inhibitors we determined the expression levels of the tumor suppressor protein p53 and of the p53 target genes CDKN1A and MCL1 in AML cells treated for 24 hours with single compounds and with combined treatment. Protein p53 was stabilized and p53 levels were increased in AML cells treated with 200nM NVP-HDM201, with three- to eightfold induction in MV4-11, MOLM-13 and OCI-AML3 cells, while OCI-AML2 cells had a high p53 level with a maximal 20% increase (Figure 4 A, B, C, D). CDKN1A gene expression was significantly induced in FLT3-ITD/TP53wt/NPM1wt AML cells (MOLM-13, MV4-11, patient#13) and in FLT3wt/TP53wt/NPM1wt cells (OCI-AML2) treated with 50nM NVP-HDM201 (Figure 4E), and in FLT3wt/TP53wt/NPM1mut cells (OCIAML3) treated with 500nM NVP-HDM201, but not in FLT3wt/TP53mut cells (MOLM-16). To reach the same level of p53 target gene expression induced by 50nM NVPHDM201 in FLT3-ITD/NPM1wt (MOLM-13, MV4-11, pat#13) and FLT3wt/NPM1wt (OCI-AML2) cells, ten times more MDM2 inhibitor was required in FLT3wt/NPM1mut (OCI-AML3) cells. Yoshimoto et al. 2009 showed that FLT3-ITD up-regulates the apoptosis inhibitor MCL-1 to promote survival of stem cells in acute myeloid leukemia. They analyzed the function of MCL-1 in FLT3-ITD AML and showed that the enforced expression of MCL-1 prevented MV4-11 cells from apoptosis in the presence of 100nM midostaurin. Inhibition of MCL-1 by shRNA resulted in apoptosis of MV4-11 cells. To elucidate the mechanism of apoptosis induction by NVPHDM201 and midostaurin we analyzed MCL-1 expression in a variety of AML cells (Figure 4F). MCL-1 gene expression was repressed in the presence of 50nM NVP-HDM201 or 50nM midostaurin in FLT3-ITD/TP53wt/NPM1wt AML cells (MOLM-13, MV4-11, patient#13), with enhanced effects in the combination treatments. MCL-1 gene expression was repressed in the presence of 50nM NVP-HDM201 in FLT3wt/TP53wt/NPM1wt cells (OCI-AML2) and by 500nM NVP-HDM201 in FLT3wt/TP53wt/NPM1mut cells (OCI-AML3), but not by midostaurin. There was no repression of MCL-1 gene expression in FLT3wt/TP53mut cells (MOLM-16). In the susceptible FLT3-ITD cell lines MV4-11 and MOLM-13 as well as in the relapsed FLT3ITD AML sample (patient#13) both compounds led to a significant reduction in MCL-1 gene expression with enhanced reduction in the combination treatments (Figure 4F). The effect of NVP-HDM201 and midostaurin treatment on MCL-1 gene repression appeared to be strongly synergistic with a combination index of 0.25. To further assess pro-apoptotic effects in AML cells treated with midostaurin and with the MDM2 inhibitor NVPHDM201, cells were stained with AnnexinV and DAPI and analyzed on a cell imager. Apoptosis and cell death were induced in FLT3-ITD/TP53wt/NPM1wt cells in a dose dependent manner by both inhibitors in single compound and combination treatments. There was a significant increase in dead cells with subG1 DNA content, and a concomitant loss of cells in defined cell cycle stages, most prominently a reduction of cells with G0/G1 phase DNA content, but also of cells with S-phase and G2 phase haematologica | 2018; 103(11)

Targeting FLT3-ITD in AML

content in MV4-11 cells treated with 20nM PKC412 and 50nM NVP-HDM201 (Figure 4G). Moreover, there was a significant increase in the number of AnnexinV positive apoptotic cells and a concomitant reduction in AnnexinV negative non-apoptotic cells in MV4-11 cells treated with 20nM PKC412 and 50nM NVP-HDM201 (Figure 4H). The pro-apoptotic and lethal effects of the single compound treatments were enhanced in the combined treatments with a combination index of 0.44 indicating synergistic pro-apoptotic and lethal effects with NVP-HDM201 and midostaurin. A similar induction of apoptosis and cell death was also detected in MOLM-13 cells treated with 20nM PKC412 and 50nM NVP-HDM201 (Online Supplementary Figure S2). While there was no pro-apoptotic effect in FLT3wt/TP53wt/NPM1mut (OCI-AML3) and FLT3-ITD/TP53mut (PL-21) cells at 100nM compounds, a low-level induction of apoptosis and cell death was detected in OCI-AML3 and PL-21 cells after 24 hours of treatment with 1 mM compounds (data not shown). In summary, our data indicate that NVP-HDM201 and midostaurin can induce apoptosis and cell death effectively and specifically in FLT3-ITD/TP53wt/NPM1wt AML cells. FLT3-ITD is a constitutively active growth factor

receptor signaling via PI3K-AKT,13 via RAS-MEK-ERK13 and via STAT514,15 leading to cell growth and proliferation via p53 inhibition and MCL1 induction (Figure 5). We have shown that MDM2 inhibition by NVP-HDM201 can reactivate p53 function leading to induction of CDKN1A and inhibition of MCL1 gene expression. Inhibition of FLT3-ITD by midostaurin, however, did not restore p53 function, but led to reduced MCL1 gene expression via RAS-MEK-ERK and/or STAT5 signaling (Figure 5). These data suggest that the combined use of NVP-HDM201 and midostaurin might be a promising treatment option particularly in FLT3-ITD AML relapsed or refractory to conventional therapy.

Discussion Acute myeloid leukemia (AML) characterized by normal karyotype (NK) and presence of the mutated FLT3 growth factor receptor gene variant FLT3-ITD comprises 27-34% of newly diagnosed AML. The subset of NK-AML patients with high allelic ratio of FLT3-ITD (>0.5) and NPM1 wild type is associated with adverse risk and low-

Figure 5. Schematic representation of the FLT3-ITD signaling pathways and downstream effects. FLT3-ITD is a constitutively active growth factor receptor signaling via PI3K-AKT, via RAS-MEKERK and via STAT5 leading to cell growth and proliferation via p53 inhibition and MCL1 induction. p53 function can be reactivated by NVP-HDM201 treatment leading to induction of CDKN1A and inhibition of MCL1 gene expression. MCL1 gene expression can be inhibited by NVP-HDM201 via p53 induction and by midostaurin (PKC412) via RAS-MEK-ERK and/or STAT5 signaling.

haematologica | 2018; 103(11)

1869

K. Seipel et al.

est survival rates. Survival rates are higher in NK-AML patients with high allelic ratio of FLT3-ITD and NPM1mut or low allelic ratio of FLT3-ITD and NPM1 wild type or FLT3wt and NPM1mut which have all been classified as intermediate risk.16 The leukemic cells of all these AML subsets have substantially elevated levels of cellular p53 antagonists and reduced p53 activity17 which identifies them as targets for treatments aiming to restore p53 function including conventional chemotherapy with genotoxic compounds and non-genotoxic treatments with p53 reactivating compounds. MDM2 is an established cellular p53 antagonist frequently overexpressed in AML cells. A variety of MDM2 inhibitors have been developed and tested in AML cell lines. Compounds currently in clinical trials for the treatment of AML include idasanutlin (RG7388), NVPCGM097 and NVP-HDM201.12,18,19 We have studied the MDM2 inhibitor idasanutlin in combination with the MEK inhibitor cobimetinib in AML cell lines and patient samples and found this combination to be effective only in AML cells expressing high levels of FLT3 and MDM2 protein.20 The activity of NVP-CGM097 was investigated in AML cell lines and primary AML cells expressing wild type and mutant p53, alone and in combination with the FLT3 inhibitor PKC412 (midostaurin) or the MEK inhibitor AZD 6244.19 Synergy was observed when NVPCGM097 was combined with FLT3 inhibition against oncogenic FLT3 expressing cells, as well as when combined with MEK inhibition in cells with activated MAPK signalling. In addition to reactivating p53 in AML cells by specific MDM2 inhibition the FLT3 receptor can be directly targeted by (more or less) specific tyrosine kinase inhibitors.5,21 In the present study, we tested a variety of FLT3 and MDM2 inhibitors. The effects on cell survival of FLT3ITD AML cells were consistent for the FLT3 inhibitors midostaurin (PKC412), quizartinib (ACC220) and gilteritinib (ASP2215), but varied for the MDM2 inhibitors idasanutlin (RG7388), NVP-CGM097 and NVP-HDM201. The most potent MDM2 inhibitor NVP-HDM201 exhibited superior combinatory effects on cell viability of FLT3ITD AML cells together with midostaurin, and moderate combinatory effects together with quizartinib and gilteritinib. The different combinatory potentials may be related to the target specificity of the three FLT3 inhibitors. While quizartinib (ACC220) inhibits FLT3 and PDGFR kinases,22 gilteritinib (ASP2215) inhibits FLT3, LTK, ALK, and AXL kinases,8 and midostaurin (PKC412) inhibits FLT3, KIT, PKC, PPK, VEGFR-2, PDGFR, and SYK kinases.23 PDGFR and VEGFR-2 are expressed in the bone marrow of AML patients24,25 and, like FLT3, signal via PI3K/AKT and MDM2 to inhibit p53.26,27 The stem cell growth factor receptor KIT is expressed in the bone marrow and, like FLT3-ITD, signals via PI3K/AKT and MDM2 to inhibit p53, and via JAK2 and STAT5.28 MOLM-13 and MV4-11 AML cells were most susceptible to the FLT3 inhibitor midostaurin. Both cell lines have a high allelic ratio of FLT3-ITD and harbor MLL rearrangements, created by t(9;11) and t(4;11), encoding MLL-AF9 and MLL-ENL, respectively. FLT3 and MLL cooperate in AML29 and leukemic cells with mutations in FLT3 and MLL are known to be susceptible to midostaurin.30,31 In the absence of MLL mutations, FLT3-ITD AML cells were less susceptible to midostaurin, but more susceptible to NVPHDM201 indicating that the presence of MLL fusion pro1870

teins may change the susceptibility of AML cells to FLT3 and MDM2 inhibitors. NK-AML cells with FLT3-ITD/TP53wt/NPM1wt were particularly susceptible to the combination of midostaurin with conventional induction therapy or with NVPHDM201. In contrast, NK-AML cells with mutated TP53 were rather resistant to midostaurin and NVP-HDM201, and they have previously been shown to be resistant to chemotherapy with genotoxic compounds32 and to other MDM2 inhibitors.33 As both the conventional induction therapy and MDM2 inhibition induce cell cycle exit and apoptosis via p53 activation, these treatments can only be effective in a TP53 wild type context.9 The presence of one mutated TP53 allele may be sufficient to suppress the function of the remaining p53 wild type protein. Hence, restoring wild-type p53 activity may be a promising option for treatment of AML with mutated TP53.34 HL-60 cells with deleted TP53 were resistant to midostaurin and HDM201, indicating that these two compounds specifically target AML cells with functional p53 protein. This is in apparent contrast to the FLT3- inhibitor sorafenib and the MDM2 inhibitor nutlin-3 which promoted synergistic cytotoxicity irrespectively of FLT3 and p53 status via induction of the pro-apoptotic Bcl-2 family members Bax and Bak in p53 wild type and p53 deleted cells.35 It seems remarkable that FLT3wt/TP53wt/NPM1mut AML cells turned out to be as refractory to midostaurin and NVP-HDM201 as TP53mut cells. This suggests that the presence of mutated NPM1 protein is sufficient to prevent induction of p53 target genes by midostaurin or NPVHDM201 in these cells. The mutated NPM1 protein (NPM1c) reduces the susceptibility of FLT3-ITD AML cells to both midostaurin and NVP-HDM201, indicating that the same molecular mechanisms may be involved in FLT3wt and FLT3-ITD cells. NPM1c is exported to the cytoplasm36 where it inhibits the tumor suppressor protein p53 by cytoplasmic retention. This leads to reduction of p53 protein in the nucleus and decreases its stability and transcriptional activation function.37 It appears that this loss of p53 activity by NPM1c cannot be compensated by inhibition of MDM2 or FLT3, at least not with NVPHDM201 or midostaurin. In contrast, nutlin-3 can induce apoptosis and cell death in FLT3wt/TP53wt/NPM1mut cells,17 indicating a fundamental difference in the activities of the two MDM2 inhibitors. The molecular mechanisms involved in p53 activation appear to be different in the conventional induction treatment. Here, FLT3wt/TP53wt/NPM1mut AML cells were more susceptible to conventional induction treatment than TP53mut cells, indicating that the presence of NPM1c is not sufficient to prevent p53 activation by genotoxic substances. This may be reflected by the usually favorable outcome of FLT3wt/TP53wt/NPM1mut AML patients after standard induction treatment.38,39 Midostaurin can enhance effects of conventional induction therapy in FLT3-ITD cells in vitro and in vivo. As evaluated in a randomized, double-blind, phase III international study (clinicaltrials.gov identifier 00651261) in highrisk patients with newly diagnosed, FLT3-mutant AML (CALGB10603/RATIFY trial), the multikinase inhibitor midostaurin plus standard chemotherapy improved survival compared with placebo plus chemotherapy.40,41 On April 28, 2017, the U.S. Food and Drug Administration approved midostaurin (RYDAPT, Novartis Pharmaceuticals Corp.) for the treatment of adult patients haematologica | 2018; 103(11)

Targeting FLT3-ITD in AML

with newly diagnosed acute myeloid leukemia (AML) who are FLT3 mutation-positive (FLT3+), in combination with standard cytarabine and daunorubicin induction and cytarabine consolidation. This combination therapy is, however, not suitable for NK AML patients with FLT3-ITD in relapse or refractory to conventional induction treatment or unfit for intensive treatment. For the subset of AML patients with a high ratio of FLT3-ITD and adverse prognosis, the combined use of non-genotoxic targeted compounds, such as the combination of midostaurin and NVP-HDM201, may represent a promising treatment option. Synergistic effects on cell viability with midostaurin and NVP-HDM201 were observed independent of sequence of application, indicating that the order of target inhibition for FLT3 and MDM2 was not important. Sequential application of NVP-HDM201 and midostaurin

References 1. Sweet K, Lancet J. State of the art update and next questions: acute myeloid leukemia. Clin Lymphoma Myeloma Leuk. 2017;17(11):703-709. 2. Fischer T, Stone RM, Deangelo DJ, et al. Phase IIB trial of oral Midostaurin (PKC412), the FMS-like tyrosine kinase 3 receptor (FLT3) and multi-targeted kinase inhibitor, in patients with acute myeloid leukemia and high-risk myelodysplastic syndrome with either wild-type or mutated FLT3. J Clin Oncol. 2010; 28(28):43394345. 3. Gallogly MM, Lazarus HM, Cooper BW. Midostaurin: a novel therapeutic agent for patients with FLT3-mutated acute myeloid leukemia and systemic mastocytosis. Ther Adv Hematol. 2017;8(9):245-261. 4. Levis M. Midostaurin approved for FLT3mutated AML. Blood. 2017; 129(26):34033406. 5. Weisberg E, Boulton C, Kelly LM, et al. Inhibition of mutant FLT3 receptors in leukemia cells by the small molecule tyrosine kinase inhibitor PKC412. Cancer Cell. 2002;1(5):433-443. 6. Levis M. Quizartinib in acute myeloid leukemia. Clin Adv Hematol Oncol. 2013; 11(9):586-588. 7. Hills RK, Gammon G, Trone D, Burnett AK. Quizartinib significantly improves overall survival in FLT3-ITD positive AML patients relapsed after stem cell transplantation or after failure of salvage chemotherapy: a comparison with Historical AML Database (UK NCRI data). Blood. 2015; 126(23):2557-2557. 8. Ueno Y, Mori M, Kamiyama Y, Kaneko N, Isshiki E, Takeuchi M. Gilteritinib (ASP2215), a novel FLT3/AXL inhibitor: preclinical evaluation in combination with azacitidine in acute myeloid leukemia. Blood. 2016;128(22):2830-2830. 9. Kojima K, Konopleva M, Samudio IJ, et al. MDM2 antagonists induce p53-dependent apoptosis in AML: implications for leukemia therapy. Blood. 2005; 106(9):3150-3159. 10. Reis B, Jukofsky L, Chen G, et al. Acute myeloid leukemia patients’ clinical response to idasanutlin (RG7388) is associated with pre-treatment MDM2 protein expression in leukemic blasts.

haematologica | 2018; 103(11)

had the same effects on cell viability as direct combination treatment. Pretreatment with one inhibitor did not enhance the susceptibility of AML cells to the second inhibitor. This leaves several options for treatment regimens in a prospective clinical trial. The best therapy regimen in combination treatments with midostaurin and NVP-HDM201 for AML patients will have to be defined empirically. Acknowledgments We thank Novartis for providing the investigational compounds NVP-HDM201 and NVP-CGM097. Funding This work was supported by a grant from the Swiss National Science Foundation (SNF) #310030_127509 to TP.

Haematologica. 2016;101(5):e185-188. 11. Holzer P, Masuya K, Furet P, et al. Discovery of a dihydroisoquinolinone derivative (NVP-CGM097): a highly potent and selective MDM2 inhibitor undergoing phase 1 clinical trials in p53wt tumors. J Med Chem. 2015;58(16):6348-6358. 12. Furet P, Masuya K, Kallen J, et al. Discovery of a novel class of highly potent inhibitors of the p53–MDM2 interaction by structurebased design starting from a conformational argument. Bioorg Med Chem Lett. 2016; 26(19):4837-4841. 13. Takahashi S. Downstream molecular pathways of FLT3 in the pathogenesis of acute myeloid leukemia: biology and therapeutic implications. J Hematol Oncol. 2011;4:13. 14. Spiekermann K, Bagrintseva K, Schwab R, Schmieja K, Hiddemann W. Overexpression and constitutive activation of FLT3 induces STAT5 activation in primary acute myeloid leukemia blast cells. Clin Cancer Res. 2003; 9(6):2140-2150. 15. Yoshimoto G, Miyamoto T, JabbarzadehTabrizi S, et al. FLT3-ITD up-regulates MCL-1 to promote survival of stem cells in acute myeloid leukemia via FLT3-ITD–specific STAT5 activation. Blood. 2009; 114(24):5034-5043. 16. 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. 17. Seipel K, Marques MT, Bozzini M-A, Meinken C, Mueller BU, Pabst T. Inactivation of the p53-KLF4-CEBPA axis in acute myeloid leukemia. Clin Cancer Res. 2016;22(3):746-756. 18. Lehmann C, Friess T, Birzele F, Kiialainen A, Dangl M. Superior anti-tumor activity of the MDM2 antagonist idasanutlin and the Bcl-2 inhibitor venetoclax in p53 wild-type acute myeloid leukemia models. J Hematol Oncol. 2016;9(1):50. 19. Weisberg E, Halilovic E, Cooke VG, et al. Inhibition of wild-type p53-expressing AML by the novel small molecule HDM2 inhibitor CGM097. Mol Cancer Ther. 2015; 14(10):2249-2259. 20. Seipel K, Marques MAT, Sidler C, Mueller BU, Pabst T. The cellular p53 inhibitor MDM2 and the growth factor receptor FLT3 as biomarkers for treatment responses to the MDM2-inhibitor idasanutlin and the MEK1 inhibitor cobimetinib in acute myeloid leukemia. Cancers. 2018;10(6).

21. Swords R, Freeman C, Giles F. Targeting the FMS-like tyrosine kinase 3 in acute myeloid leukemia. Leukemia. 2012; 26(10):21762185. 22. Kampa-Schittenhelm KM, Heinrich MC, Akmut F, Döhner H, Döhner K, Schittenhelm MM. Quizartinib (AC220) is a potent second generation class III tyrosine kinase inhibitor that displays a distinct inhibition profile against mutant-FLT3, PDGFRA and -KIT isoforms. Mol Cancer. 2013;12:19. 23. Weisberg E, Sattler M, Manley PW, Griffin JD. Spotlight on midostaurin in the treatment of FLT3-mutated acute myeloid leukemia and systemic mastocytosis: design, development, and potential place in therapy. OncoTargets Ther. 2017; 11:175182. 24. Foss B, Ulvestad E, Bruserud Ø. Plateletderived growth factor (PDGF) in human acute myelogenous leukemia: PDGF receptor expression, endogenous PDGF release and responsiveness to exogenous PDGF isoforms by in vitro cultured acute myelogenous leukemia blasts. Eur J Haematol. 2001;67(4):267-278. 25. Padró T, Bieker R, Ruiz S, et al. Overexpression of vascular endothelial growth factor (VEGF) and its cellular receptor KDR (VEGFR-2) in the bone marrow of patients with acute myeloid leukemia. Leukemia. 2002;16(7):1302-1310. 26. Dos Santos C, McDonald T, Ho YW, et al. The Src and c-Kit kinase inhibitor dasatinib enhances p53-mediated targeting of human acute myeloid leukemia stem cells by chemotherapeutic agents. Blood. 2013; 122(11):1900-1913. 27. Lei H, Velez G, Kazlauskas A. Pathological signaling via platelet-derived growth factor receptor {alpha} involves chronic activation of Akt and suppression of p53. Mol Cell Biol. 2011;31(9):1788-1799. 28. Kent D, Copley M, Benz C, Dykstra B, Bowie M, Eaves C. Regulation of hematopoietic stem cells by the steel factor/KIT signaling pathway. Clin Cancer Res. 2008;14(7):1926-1930. 29. Stubbs MC, Kim YM, Krivtsov AV, et al. MLL-AF9 and FLT3 cooperation in acute myelogenous leukemia: development of a model for rapid therapeutic assessment. Leukemia. 2008;22(1):66-77. 30. Armstrong SA, Kung AL, Mabon ME, et al. Inhibition of FLT3 in MLL. Validation of a

1871

K. Seipel et al.

31.

32.

33.

34.

35.

1872

therapeutic target identified by gene expression based classification. Cancer Cell. 2003;3(2):173-183. Stam RW, den Boer ML, Schneider P, et al. Targeting FLT3 in primary MLL-generearranged infant acute lymphoblastic leukemia. Blood. 2005;106(7):2484-2490. Wattel E, Preudhomme C, Hecquet B, et al. p53 mutations are associated with resistance to chemotherapy and short survival in hematologic malignancies. Blood. 1994; 84(9):3148-3157. Long J, Parkin B, Ouillette P, et al. Multiple distinct molecular mechanisms influence sensitivity and resistance to MDM2 inhibitors in adult acute myelogenous leukemia. Blood. 2010;116(1):71-80. Muller PAJ, Vousden KH. Mutant p53 in cancer: new functions and therapeutic opportunities. Cancer Cell. 2014; 25(3):304317. Zauli G, Celeghini C, Melloni E, et al. The

36. 37.

38.

39.

sorafenib plus nutlin-3 combination promotes synergistic cytotoxicity in acute myeloid leukemic cells irrespectively of FLT3 and p53 status. Haematologica. 2012; 97(11):1722-1730. Grisendi S, Mecucci C, Falini B, Pandolfi PP. Nucleophosmin and cancer. Nat Rev Cancer. 2006;6(7):493-505. Colombo E, Marine J-C, Danovi D, Falini B, Pelicci PG. Nucleophosmin regulates the stability and transcriptional activity of p53. Nat Cell Biol. 2002;4(7):529-533. Schlenk RF, Döhner K, Krauter J, et al. Mutations and treatment outcome in cytogenetically normal acute myeloid leukemia. N Engl J Med. 2008;358(18):1909-1918. Sperr WR, Zach O, Pöll I, et al. Karyotype plus NPM1 mutation status defines a group of elderly patients with AML (≥60 years) who benefit from intensive post-induction consolidation therapy. Am J Hematol. 2016;91(12):1239-1245.

40. Stone RM, Mandrekar S, Sanford BL, et al. The multi-kinase inhibitor midostaurin (M) prolongs survival compared with placebo (P) in combination with daunorubicin (D)/cytarabine (C) induction (ind), highdose C consolidation (consol), and as maintenance (maint) therapy in newly diagnosed acute myeloid leukemia (AML) patients (pts) age 18-60 with FLT3 mutations (muts): an international prospective randomized (rand) P-controlled doubleblind trial (CALGB 10603/RATIFY [Alliance]). Blood. 2015;126(23):6-6. 41. Starr P. Midostaurin the first targeted therapy to improve survival in AML: potentially practice-changing. Am Health Drug Benefits. 2016;9(Spec Issue):1-21. 42. Chou T-C. Drug combination studies and their synergy quantification using the Chou-Talalay method. Cancer Res. 2010; 70(2):440-446.

haematologica | 2018; 103(11)

ARTICLE

Acute Lymphoblastic Leukemia

Radiation exposure from computerized tomography and risk of childhood leukemia: Finnish register-based case-control study of childhood leukemia (FRECCLE)

Ferrata Storti Foundation

Atte Nikkilä,1 Jani Raitanen,2,3 Olli Lohi1,4 and Anssi Auvinen2,3,5

1 Faculty of Medicine and Biosciences, University of Tampere; 2Faculty of Social Sciences, University of Tampere; 3UKK Institute for Health Promotion Research, Tampere; 4Tampere Center for Child Health Research, University of Tampere and Tampere University Hospital and 5STUK – Radiation and Nuclear Safety Authority, Helsinki, Finland

Haematologica 2018 Volume 103(11):1873-1880

ABSTRACT

he only well-established risk factors for childhood leukemia are high-dose ionizing radiation and Down syndrome. Computerized tomography is a common source of low-dose radiation. In this study, we examined the magnitude of the risk of childhood leukemia after pediatric computed tomography examinations. We evaluated the association of computed tomography scans with risk of childhood leukemia in a nationwide register-based casecontrol study. Cases (n=1,093) were identified from the populationbased Finnish Cancer Registry and three controls, matched by gender and age, were randomly selected for each case from the Population Registry. Information was also obtained on birth weight, maternal smoking, parental socioeconomic status and background gamma radiation. Data on computed tomography scans were collected from the ten largest hospitals in Finland, covering approximately 87% of all pediatric computed tomography scans. Red bone marrow doses were estimated with NCICT dose calculation software. The data were analyzed using exact conditional logistic regression analysis. A total of 15 cases (1.4%) and ten controls (0.3%) had undergone one or more computed tomography scans, excluding a 2-year latency period. For one or more computed tomography scans, we observed an odds ratio of 2.82 (95% confidence interval: 1.05 – 7.56). Cumulative red bone marrow dose from computed tomography scans showed an excess odds ratio of 0.13 (95% confidence interval: 0.02 – 0.26) per mGy. Our results are consistent with the notion that even low doses of ionizing radiation observably increase the risk of childhood leukemia. However, the observed risk estimates are somewhat higher than those in earlier studies, probably due to random error, although unknown predisposing factors cannot be ruled out.

Correspondence: atte.nikkila@uta.fi

Received: January 6 2018. Accepted: June 26, 2018. Pre-published: July 5, 2018. doi:10.3324/haematol.2018.187716 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1873 ©2018 Ferrata Storti Foundation

Introduction Leukemia is the most common childhood malignancy.1 The incidence rates of childhood leukemia in Finland are comparable to those in other European countries and show a slight increasing trend up to the 1990s.2 Acute lymphoblastic leukemia accounts for approximately 85% of all childhood leukemias. The major histological subtype of acute lymphoblastic leukemia is precursor B-cell acute lymphoblastic leukemia (~85%).1 Well-established risk factors for childhood leukemia include high doses of ionizing radiation, alkylating chemotherapy agents, as well as Down syndrome and some rare congenital syndromes such as Fanconi anemia, Bloom syndrome and ataxia telangiectasia.1,3-6 A number of genetic variants have also been associated haematologica | 2018; 103(11)

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.

1873

A. Nikkilä et al.

with increased risk of leukemia.7,8 Furthermore, there is reasonably consistent evidence of a slightly increased risk associated with large birth weight relative to gestational time.9 A higher risk has also been suggested for older parental age, delivery by Cesarean section, and paternal smoking.10–13 However, daycare attendance, allergic diseases, maternal folic acid supplementation before birth, and early immune stimulation have been suggested to reduce the risk of leukemia.14–17 Although high doses of ionizing radiation increase the risk of childhood leukemia, the magnitude of any effect from low doses remains uncertain. Some studies have suggested increased risks associated with background radiation and following x-ray examinations in utero and post-natally.18–22 Computed tomography (CT) imaging has been used for almost four decades and its frequency of utilization increased greatly during the 1980s-1990s. The annual number of scans peaked around year 2002; more recently CT scans have been partly replaced by magnetic resonance imaging in pediatric imaging, partly because of the risk of cancer from ionizing radiation.23 In 2015, 5,311 pediatric CT scans were performed in the Finnish population of 1,024,000 children under 17 years old, which is a low rate compared to that in many other countries.23,24

Four high-quality studies have investigated the association of pediatric CT scans and childhood leukemia.25–28 The interpretation of the findings must include an evaluation of confounding by indication, i.e. underlying conditions predisposing children to both CT scans and leukemia.28–30 Nevertheless, the evidence is still limited and the magnitude of the risk needs to be characterized further. In this study, we examined the magnitude of the risk of childhood leukemia after pediatric CT examinations using a nationwide case-control design with efforts to avoid reverse causation.

Methods We used a register-based, case-control study with individually matched controls. The key characteristics of the material have been presented previously.10 Briefly, all cases of childhood leukemia (M9800–M9948 in ICD-O-3) diagnosed in Finland during 1990–2011 (n=1,100) before the age of 15 years were identified from the Finnish Cancer Registry (Figure 1). Three controls were individually matched, by sex and year of birth, for each case from the Population Register Center. In all analyses, a 2-

Figure 1. Flow chart depicting the selection of study subjects. The flow of cases is represented on a white background and the controls on a light gray background. The necessary exclusions are shown in red boxes. Dashed lines represent the linking of the study subjects with the CT scans collected.

1874

haematologica | 2018; 103(11)

Pediatric CT scans and childhood leukemia

year latency period was used, in part to deal with reverse causation due to confounding by indication.31 Also, multiple predisposing factors (Online Supplementary Table S1) were accounted for with outpatient register data. The methods are described in more detail in the Online Supplementary Material. We obtained data on all CT scans performed on pediatric patients (<15 years) from all five university hospitals and the five

largest central hospitals in Finland (Table 1, Figure 2). The period of data availability varied between hospitals, because radiological databases with information on each CT scan for individual patients were introduced at different times. We estimated that the data from the study hospitals covered 87% of all pediatric CT scans performed in Finland during 1975–2011 (see the Online Supplementary Material for details). For each CT scan, we

Table 1. The collection and availability of electronically stored computed tomography scans.

Hospital Helsinki University Hospital Tampere University Hospital Oulu University Hospital Turku University Hospital Kuopio University Hospital Central Finland Central Hospital Satakunta Central Hospital Seinäjoki Central Hospital Päijänne Tavastia Central Hospital North Karelia Central Hospital TOTAL

City

Data availability

Number of CT scans

Helsinki Tampere Oulu Turku Kuopio Jyväskylä Pori Seinäjoki Lahti Joensuu

1990–2011 1978–2011 1993–2011 1996–2011 1996–2011 2002–2011 1995–2011 1999–2011 2000–2011 1993–2011

31,825 9,236 7,513 7,360 5,408 2,571 1,948 1,759 1,597 1,191 80,783

All Finnish university hospitals are listed first followed by the five chosen central hospitals.

Figure 2. Flow chart linking the collected computed tomography scans to the study subjects The flow of the CT scans is represented on a white background and the CT scans to different body parts on a light gray background. The necessary exclusions are shown in red boxes.

haematologica | 2018; 103(11)

1875

A. Nikkilä et al.

obtained the parameters used for dose assessment including year, body part, use of contrast medium and the number of sequences. Manufacturers and models of CT scanners in each hospital were acquired from the Radiation and Nuclear Safety Authority (STUK). For dose calculations, we assumed in the main analysis that each CT scan was performed using the latest CT scanner available at the hospital. Data on a total of 80,783 pediatric CT scans were obtained and of those, 49 CT scans were performed on the study subjects, excluding the 2-year latency period (Table 1). Half (n=25) were head scans, and 19 were lung scans. Of the CT scans, 36 were performed on 15 (1.4%) cases and 13 scans on 10 (0.3%) controls. The CT scan parameters were obtained based on expert opinion of an experienced hospital physicist (Online Supplementary Table S2). The doses were estimated using the NCICT software (v1.2).32 Age- and sex-specific pediatric software phantoms (for neonates, and children aged 1, 5, 10, and 15 years) were used. The input for dose calculation also included the scanner manufacturer and model. If data were available only on the manufacturer, a manufacturer-specific average was used. It was assumed that a head or body filter was used, based on the target body part. The cumulative absorbed red bone marrow (RBM) doses were obtained as the sums of absorbed RBM doses from all CT scans for each study subject. The dose from a scan was multiplied by 1.5 if contrast medium was used, consistent with tissuespecific coefficients suggested for other tissues.33 Alternative dose estimates were obtained based on values reported in the literature.34 We identified subjects with Down syndrome (40 cases and 2 controls) from the Congenital Malformation Register and Care Register for Health Care, and those with a previous malignancy (2 cases) from the Finnish Cancer Registry. They were excluded to avoid confounding by indication (reverse causation). We also collected information on birth weight (large for gestational age) and maternal smoking during pregnancy from the Medical Birth Register, as well as socioeconomic status and education of the parents from Statistics Finland. Residential exposure to background gamma radiation, including natural terrestrial radiation and Chernobyl fallout, was estimated as described previously.8 Due to small frequencies, exact conditional logistic regression in SAS 9.4 was used for estimating odds ratios (OR), excess odds ratios and their confidence intervals (CI).35 Statistical power calculations indicated that the sample size is sufficient for detecting a linear dose-response with an OR of 1.05 or greater per 1 mGy increase in cumulative RBM dose with a statistical power of 80% using asymptotic conditional logistic regression.36 The ethical committee of Pirkanmaa Hospital district reviewed the study protocol (tracking number R14074) and, in accordance with Finnish regulations, no informed consent was required for this register-based study. In addition, each hospital approved our study protocol before delivering the data on CT scans. We obtained permission to use data from the Finnish Cancer Registry, the Medical Birth Register, Care Register for Health Care and Congenital Malformation Register from the National Institute of Health and Welfare (1774/5.05.00/2014), as well as census data on socioeconomic status from Statistics Finland (TK-52-306-16).

Results In our nationwide register-based study, after excluding cases with an incorrect personal identification number or prohibition to use their data, we identified 1,093 cases of childhood leukemia diagnosed in 1990-2011. Most of the 1876

cases were acute lymphoblastic leukemia (81.1%) or acute myeloid leukemia (13.0%). The median age at diagnosis among cases was 4.52 years (interquartile range, IQR 2.72 – 8.23). Of the cases and controls, 52% were male (Table 2). The criteria for large for gestational age were met by 121 (13.3%) of the cases and 275 (9.9%) of the controls. Table 2. The characteristics of cases and controls before any exclusions.

Cases (n=1,093) Controls (n=3,279) Gender Female Male Large for gestational age No Ys Missing Mother’s smoked during pregnancy No Yes Missing Down syndrome No Yes Parents’ education Mother Upper secondary Bachelor’s degree Master’s or doctor’s degree Missing Father Upper secondary Bachelor’s degree Master’s or doctor’s degree Missing Parents’ socioeconomic status Mother Self-employed Upper level employee Lower level employee Manual worker Other Missing Father Self-employed Upper level employee Lower level employee Manual worker Other Missing Age at leukemia diagnosis, years 0–2 2–7 7 – 15 Leukemia type Pre-B-ALL Pre-T-ALL Unclassified ALL Acute myeloid leukemia Other

48.0% (525) 52.0% (568)

48.0% (1575) 52.0% (1704)

86.7% (788) 13.3% (121) 184

90.1% (2493) 9.9% (275) 511

0.001

83.1% (742) 16.9% (151) 200

84.5% (2296) 15.5% (420) 563

0.096

96.3% (1053) 3.7% (40)

99.9% (3277) 0.1% (2)

48.5% (530) 22.3% (244) 10.2% (112) 18.9% (207)

50.6% (1659) 23.1% (756) 9.8% (321) 16.6% (543)

ref 0.869 0.406

52.0% (568) 15.2% (166) 10.0% (110) 22.8% (249)

51.4% (1685) 16.2% (532) 10.2% (334) 22.2% (728)

ref 0.423 0.880

7.7% (84) 16.1% (176) 34.8% (380) 21.4% (231) 18.2% (199) 2.1% (23)

8.3% (273) 15.7% (514) 34.5% (1130) 20.6% (674) 20.3% (664) 0.7% (24)

ref 0.477 0.521 0.490 0.859

13.9% (152) 17.6% (192) 18.3% (197) 34.0% (372) 12.4% (135) 4.1% (45)

12.0% (395) 18.2% (596) 17.9% (587) 35.0% (1148) 14.3% (469) 2.6% (84)

ref 0.178 0.273 0.170 0.036

<0.001

14.3% (156) 55.5% (605) 33.4% (332) 75.6% (826) 5.9% (64) 1.8% (20) 13.6% (149) 3.1% (34)

The reported P-values are from an univariate conditional logistic regression model. The nonbinary variables were treated as factors and the reference categories are marked with “ref”. ALL: acute lymphoblastic leukemia.

haematologica | 2018; 103(11)

Pediatric CT scans and childhood leukemia

After exclusions, eight cases (0.7%) and nine controls (0.3%) had undergone at least one CT scan. The median RBM dose was 10.1 mGy (IQR 4.79 – 13.6) for the exposed cases and 6.29 mGy (IQR 5.69 – 7.14) for the exposed controls (Figure 3). The corresponding literaturebased values were 26.5 mGy and 17.6 mGy. The RBM doses calculated with NCICT from thoracic CT scans varied between 1.8 and 6.8 mGy (median 4.0 mGy) and similarly, the doses for head CT scans varied between 1.6 and 10.7 mGy (median 6.6 mGy). The OR for any versus no CT was 2.82 (95% CI: 1.05 – 7.56). For two or more pediatric CT scans, the OR was 5.22 (95% CI: 0.89 – 69.9). For any head CT scans, an OR of 4.00 (95% CI: 1.39 – 11.5) was obtained. The overall excess OR of childhood leukemia was 0.13 (95% CI: 0.02 – 0.26) per mGy of absorbed RBM dose calculated with the NCICT software (Table 3). Using the cumulative RBM dose estimates from the literature, an excess OR of 0.05 (95% CI: 0.01 – 0.10) per mGy was obtained. In an analysis by dose tertile calculated with

NCICT, the excess OR relative to zero dose were 1.26 (95% CI: -0.50 – 10.1) for the first group, 0.09 (95% CI: -0.89 – 10.5) for the second, and 5.00 (95% CI: 0.10 – 31.7) for the last (Figure 4). For the most common subtype, precursor B-cell acute lymphoblastic leukemia, the excess OR per mGy was 0.14 (95% CI: 0.02 – 0.29) using estimates from NCICT and 0.06 (0.01 – 0.11) for literature-based estimates. The excess OR for any versus no CT scans was 2.25 (95% CI: 0.08 – 8.75) for acute lymphoblastic leukemia and 2.88 (95% CI: 0.22 – 11.4) for precursor B-cell acute lymphoblastic leukemia. In the analysis by age at diagnosis/reference date, the excess OR for any versus no CT scans was 3.50 (95% CI: -0.25 – 25.9) for children aged 2 – <7 years and 1.27 (95% CI: -0.32 – 6.54) for those aged 7 – <15 years. Covariate (confounder) adjustments (large for gestational age, maternal smoking during pregnancy, parental education and parental socioeconomic status) did not alter the OR for CT exposure by more than 0.05 units, with the exception of maternal smoking, which increased the OR related to the number of pediatric CT scans (0 versus 1 or more) (approximately 0.10 units). Nevertheless, we preferred the unadjusted model, as missing data on maternal smoking resulted in wider confidence intervals for the main variables. The OR were higher when the subjects with Down syndrome were not excluded (for 1 or more CT scans OR=5.21, 95% CI: 2.19 – 12.4 and for cumulative RBM dose excess OR=0.19 per mGy, 95% CI: 0.07 – 0.32). No evidence of a different effect of the RBM doses on leukemia risk for subjects with or without Down synTable 3. The frequencies of computed tomography scans for subjects >2 years old at the reference date and odds ratios calculated with exact conditional logistic regression.

Cases 911 CT scans 0 903 1 4 2 or more 4 by type (1 or more) ALL 7 pre-B-ALL 7 by age-group (1 or more) 2 - <7 years 3 7 - <15 years 5 by dose index (NCICT/literature) low, 4.79a/11.6b mGy 3 medium, 6.72a/19. 8b mGy 1 high, 13.8a/33.2b mGy 4 per 1 mGy (NCICT) TOTAL pre-B-ALL per 1 mGy (literature) TOTAL pre-B-ALL

Figure 3. Histograms of (A) the ages of the subjects at the time of computed tomography scan and (B) the cumulative doses received by the subjects, calculated with NCICT.

haematologica | 2018; 103(11)

Controls

OR (95% CI)

2730 2721 7 2

1.85 (0.39, 7.36) 6.22 (0.89, 68.9)

7 6

3.25 (1.08, 9.75) 3.88 (1.22, 12.4)

2 7

4.50 (0.75, 26.9) 2.27 (0.68, 7.54)

4 3 2

2.26 (0.50, 10.1) 1.09 (0.11, 10.5) 6.00 (1.10, 32.8) 1.13 (1.02, 1.26) 1.14 (1.02, 1.29) 1.05 (1.01, 1.10) 1.06 (1.01, 1.11)

The reference group for all calculated odds ratio (OR) is zero CT scans for categorical variables. Study subjects with Down syndrome or cancer diagnoses were excluded. All reported OR are from an unadjusted model. The median doses for dose-index classes calculated with NCICT are marked with a. The respective class medians based on literature are marked with b. ALL: acute lymphoblastic leukemia; pre-B-ALL: precursor B-cell acute lymphoblastic leukemia.

1877

A. Nikkilä et al.

drome was found to suggest effect modification (interaction P=0.99). When the oldest possible CT scanner (at maximum, 10 years old) at the hospital was used in dose estimation instead of the most modern CT scanner, the median cumulative RBM dose for cases was 9.71 mGy (IQR 7.09 – 18.7) and for controls 7.14 mGy (IQR 5.71 – 12.6), with an excess OR of 0.11 (95% CI: 0.02 – 0.22) per mGy. When the cumulative RBM dose from terrestrial gamma radiation and Chernobyl fallout was included in the model, the OR for cumulative RBM dose from pediatric CT scans remained unchanged. The median cumulative dose from residential gamma radiation was 1.96 mSv for cases and 1.90 mSv for controls. The distributions of cities of the last addresses of cases and controls were analyzed to evaluate whether cases and controls belonged to catchment populations of different hospitals, which might have caused differential misclassification due to contrasting data availability. No difference in the distributions was noted (chi-squared test, P=0.30). The age and CT scan years of the subjects are reported in Online Supplementary Table S3.

Discussion We estimated the impact of radiation exposure from pediatric CT scans on risk of childhood leukemia in a nationwide register-based case-control study in Finland. Overall, a statistically significant increase in risk per mGy of RBM absorbed dose was found. The central estimate is larger than in previous studies, but the confidence intervals overlap with earlier results, and the effect size is compatible with extrapolation from high-dose studies. The higher main point estimate is likely influenced by random error, as the dose estimates were imprecise due to lack of detail in dosimetric data, including parameter values used for the scanner. It is also possible that the typical values based on expert opinion are representative of current procedures, but may underestimate doses from older examinations, which could inflate the risk estimates per unit dose. However, our site-specific dose estimates calculated with NCICT were quite comparable with those reported in the British study.25 We minimized the potential for systematic error by adjusting for several confounders and used consistent procedures for the cases and controls. The risk estimates were slightly higher for precursor B-cell acute lymphoblastic leukemia than for other leukemias, but the difference was not statistically significant. Two large studies have been published on the subject prior to ours. The cohort studies from the United Kingdom and Australia reported a significant risk of childhood leukemia associated with RBM dose from pediatric CT scans.25,26 Pearce et al. found an excess relative risk of 0.04 per mGy and Mathews et al. reported a relative risk of 1.2 for one or more CT scans with an excess relative risk of 0.04 per mGy. The Australian cohort had 211 exposed leukemia cases and the UK study 74. A smaller German cohort study reported an increased leukemia incidence following two or more CT scans, but a non-significant dose-response based on 12 exposed cases.27 Based on the Life Span Study in Japan, the extrapolated excess relative risk for childhood exposure would be approximately 0.05 per mGy.37 1878

Other major sources of ionizing radiation were taken into consideration by including cumulative RBM doses from terrestrial gamma radiation and Chernobyl fallout, and this did not affect the results. In our data, the average cumulative RBM dose from CT for the controls was only 0.002 mGy, which is approximately 0.1% of the average annual RBM dose in Finland.38 We accounted for medical use of radiation, to which tomography scans make the largest contribution, and terrestrial gamma radiation, which accounts for nearly two-thirds of average annual radiation to the RBM in Finland.23,39 In addition, there is little evidence to assume that other sources of ionizing radiation, such as cosmic radiation or internal exposure to natural radioisotopes, would distribute unequally among the cases and controls. The coefficient 1.5 for incremental dose due to CT imaging with contrast medium was chosen pragmatically based on the coefficients for other body parts, as the effects on RBM dose were not reported separately.33 Based on limited population statistics available from the Radiation and Nuclear Safety Authority,23 roughly 30 CT scans were expected for the controls. However, only 13 scans were recorded among them. This might partly reflect incomplete availability of data, but the estimate of the expected numbers is highly uncertain because of lack of data on pediatric CT scans prior to 2008. It is also worth noting that pediatric CT scans are performed less frequently in Finland than in several other countries.24 Our material consists of a comprehensive set of childhood leukemia cases and representative controls, which should eliminate selection bias by virtue of a registerbased approach, which required no consent or information from the study subjects or their families. The study period covers the years in which the use of pediatric CT scans was most frequent, as the annual number of pediatric CT scans has been decreasing in Finland since the year 2000.23 The data on CT scans were obtained from

Figure 4. Dose-response curve of cumulative red bone marrow dose from pediatric computed tomography scans and childhood leukemia. The point estimates with 95% confidence intervals are for the three dose index levels and the fitted curve is for the cumulative RBM dose calculated with NCICT. The shaded area represents the 95% confidence interval for the continuous dose-response. The vertical axis is on a binary logarithm scale.

haematologica | 2018; 103(11)

Pediatric CT scans and childhood leukemia

hospital databases to avoid recall bias, and also included the scanner model and use of contrast medium. As in other studies, the most common single CT scan in our analysis was a head scan.23 Radiation doses to RBM from the CT scans were calculated using the best available methods, employing NCICT software, with age- and sex-specific phantoms and taking into account the scanner model. The scanning parameters entered into the software were based on the settings and procedures commonly used in Finland, although data were not available for each scan. We also evaluated the effects of choosing the most modern CT scanner at each imaging site and the OR showed robust behavior. We also had data on several important risk factors including Down syndrome, parental socioeconomic status, large for gestational age and maternal smoking. We were able to incorporate data on cancer predisposing factors, which have been shown to be of importance recently.28,30 Inclusion of cases with Down syndrome would have increased the risk estimates, possibly because Down syndrome is associated with increased risks of both leukemia and infections.4,40 We also explored the joint effect of Down syndrome and cumulative RBM dose and found no interaction. Subgroup analyses of exploratory nature were carried out by subtype of childhood leukemia and age at diagnosis, although these were underpowered. Our study has some shortcomings. We were able to obtain data from all ten hospitals only after 2002, thus exposure assessment is not uniformly complete for subjects born prior to that year. Only a minor improvement in statistical power would have been reached by collecting pediatric CT scans from the rest of the imaging centers in Finland. In addition, there is no reason to assume that the missed CT scans would have been unequally distributed for the cases and controls, i.e. result in differen-

References 1. Eden T. Aetiology of childhood leukaemia. Cancer Treat Rev. 2010;36(4):286–297. 2. Cancer Incidence, Mortality, Prevalence and Survival in the Nordic Countries, Version 8.0 (20.12.2017). 3. Hsu W-L, 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. 4. Hasle H. Pattern of malignant disorders in individuals with Down’s syndrome. Lancet Oncol. 2001;2(7):429–436. 5. Haddy N, Cécile Le Deley M, Samand A, et al. Role of radiotherapy and chemotherapy in the risk of secondary leukaemia after a solid tumour in childhood. Eur J Cancer. 2006;42(16):2757–2764. 6. Kamiya K, Ozasa K, Akiba S, et al. Longterm effects of radiation exposure on health. Lancet. 2015;386(9992):469–478. 7. Wiemels JL, Walsh KM, de Smith AJ, et al. GWAS in childhood acute lymphoblastic leukemia reveals novel genetic associations at chromosomes 17q12 and 8q24.21. Nat Commun. 2018;9(1):286. 8. Ripperger T, Bielack SS, Borkhardt A, et al.

haematologica | 2018; 103(11)

10.

11.

12.

13.

tial misclassification. For dose estimation, complete information on the scanning parameters is included in the modern picture archiving systems, but was not available before the year 2000. Use of parameters for each individual scan would have provided more accurate dose estimates. The unexpectedly lower median dose of cases for older scanners found in our sensitivity analysis may be due to random error. The number of different CT scanners in our analysis was limited and thus the estimates of average dose were imprecise. Our results support the notion that even small doses of radiation from pediatric CT scans produce a small, but detectable increase in leukemia risk. In the subgroup analyses, we observed no substantial differences by age or leukemia subtype, although slightly higher risks were found for precursor B-cell acute lymphoblastic leukemia. Acknowledgments The authors would like to thank Isabelle Thierry-Chef (IARC), Hannele Niiniviita (Turku University Hospital), Juha Suutari (STUK) and Rebecca Smith-Bindman (University of California, San Francisco) for their valuable suggestions and comments regarding modeling of CT scan doses with the data available in Finland. We are also grateful to Dr Choonsik Lee (National Cancer Institute) for his insightful comments related to modeling contrast media and for providing us with his state-ofthe-art dose calculation software (NCICT) and Anniriikka Rantala (STUK) for collecting the data on CT scanners used in Finland. Päivi Laarne’s (Tampere University Hospital) crucial input regarding the scanning parameters enabled us to use NCICT software. Funding for the study was obtained from the Finnish Foundation for Pediatric Research, Väre Foundation for Pediatric Cancer Research and Competitive State Research Financing of the Expert Responsibility area of Tampere University Hospital (9T030 and 9U030).

Childhood cancer predisposition syndromes-A concise review and recommendations by the Cancer Predisposition Working Group of the Society for Pediatric Oncology and Hematology. Am J Med Genet A. 2017;173(4):1017–1037. Milne E, Greenop KR, Metayer C, et al. Fetal growth and childhood acute lymphoblastic leukemia: findings from the Childhood Leukemia International Consortium. Int J Cancer. 2013;133(12):2968-2979. Nikkilä A, Erme S, Arvela H, et al. Background radiation and childhood leukemia: a nationwide register-based casecontrol study. Int J Cancer. 2016;139(9): 1975–1982. Marcotte EL, Thomopoulos TP, InfanteRivard C, et al. Caesarean delivery and risk of childhood leukaemia: a pooled analysis from the Childhood Leukemia International Consortium (CLIC). Lancet Haematol. 2016;3(4):e176–e185. Kendall GM, Little MP, Wakeford R, et al. A record-based case-control study of natural background radiation and the incidence of childhood leukaemia and other cancers in Great Britain during 1980 – 2006. Leukemia. 2013;27(1):3–9. Metayer C, Petridou E, Aranguré JMM, et al. Parental tobacco smoking and acute myeloid

14.

15.

16.

17.

18.

19.

leukemia: the Childhood Leukemia International Consortium. Am J Epidemiol. 2016;184(4):261–273. Metayer C, Milne E, Dockerty JD, et al. Maternal supplementation with folic acid and other vitamins and risk of leukemia in offspring: a Childhood Leukemia International Consortium study. Epidemiology. 2014;25(6):811–822. Rudant J, Lightfoot T, Urayama KY, et al. Childhood acute lymphoblastic leukemia and indicators of early immune stimulation: a Childhood Leukemia International Consortium study. Am J Epidemiol. 2015; 181(8):549–562. Ma X, Buffler PA, Selvin S, et al. Daycare attendance and risk of childhood acute lymphoblastic leukaemia. Br J Cancer. 2002; 86(9):1419–1424. Linabery AM, Jurek AM, Duval S, Ross JA. The association between atopy and childhood/adolescent leukemia: a meta-analysis. Am J Epidemiol. 2010;171(7):749–764. Shu XO, Potter JD, Linet MS, et al. Diagnostic X-rays and ultrasound exposure and risk of childhood acute lymphoblastic leukemia by immunophenotype. Cancer Epidemiol Biomarkers Prev. 2002;11(2):177–185. Meinert R, Kaletsch U, Kaatsch P, Schüz J, Michaelis J. Associations between child-

1879

A. Nikkilä et al.

20.

21.

22.

23. 24. 25.

26.

1880

hood cancer and ionizing radiation: results of a population-based case-control study in Germany. Cancer Epidemiol Biomarkers Prev. 1999;8(9):793–799. Infante-Rivard C, Mathonnet G, Sinnett D. Risk of childhood leukemia associated with diagnostic irradiation and polymorphisms in DNA repair genes. Environ Health Perspect. 2000;108(6):495. Bartley K, Metayer C, Selvin S, Ducore J, Buffler P. Diagnostic X-rays and risk of childhood leukaemia. Int J Epidemiol. 2010;39(6): 1628–1637. Harbron RW, Chapple C-L, O’Sullivan JJ, et al. Cancer incidence among children and young adults who have undergone x-ray guided cardiac catheterization procedures. Eur J Epidemiol. 2018;33(4):393-401. Suutari J. Radiologisten tutkimusten ja toimenpiteiden määrät vuonna 2015. STUK; 2015. UNSCEAR. Sources, Effects and Risks of Ionizing Radiation, UNSCEAR 2017 Report. United Nations. Pearce MS, Salotti JA, Little MP, et al. Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retrospective cohort study. Lancet. 2012;380(9840):499–505. Mathews JD, Forsythe AV, Brady Z, et al. Cancer risk in 680,000 people exposed to

27.

28.

29.

30.

31. 32.

33.

computed tomography scans in childhood or adolescence: data linkage study of 11 million Australians. BMJ. 2013;346:f2360. Krille L, Dreger S, Schindel R, et al. Risk of cancer incidence before the age of 15 years after exposure to ionising radiation from computed tomography: results from a German cohort study. Radiat Environ Biophys. 2015;54(1):1–12. Journy N, Rehel J-L, Ducou Le Pointe H, et al. Are the studies on cancer risk from CT scans biased by indication? Elements of answer from a large-scale cohort study in France. Br J Cancer. 2015;112(1):185–193. Berrington De Gonzalez A, Salotti JA, Mchugh K, et al. Relationship between paediatric CT scans and subsequent risk of leukaemia and brain tumours: assessment of the impact of underlying conditions. Br J Cancer. 2016;114(4):388-394. Journy N, Roué T, Cardis E, et al. Childhood CT scans and cancer risk: impact of predisposing factors for cancer on the risk estimates. J Radiol Prot. 2016;36(1):N1–N7. ANNEx BScienti. UNSCEAR 2013 Report. Lee C, Kim KP, Bolch WE, Moroz BE, Folio L. NCICT: a computational solution to estimate organ doses for pediatric and adult patients undergoing CT scans. J Radiol Prot. 2015;35(4):891-909. Amato E, Salamone I, Naso S, Bottari A,

34.

35. 36.

37.

38. 39.

40.

Gaeta M, Blandino A. Can contrast media increase organ doses in CT examinations? A clinical study. Am J Roentgenol. 2013;200(6): 1288–1293. Kim KP, Berrington de González A, Pearce MS, et al. Development of a database of organ doses for paediatric and young adult CT scans in the United Kingdom. Radiat Prot Dosimetry. 2012;150(4):415–426. Mehta CR, Patel NR. Exact logistic regression: theory and examples. Stat Med. 1995;14(19):2143–2160. Lachin JM. Sample size evaluation for a multiply matched case-control study using the score test from a conditional logistic (discrete Cox PH) regression model. Stat Med. 2008;(27):2509–2523. Wakeford R, Little MP, Kendall GM. Risk of childhood leukemia after low-level exposure to ionizing radiation. Expert Rev Hematol. 2010;3(3):251–254. Muikku M, Arvela H, Järvinen H, et al. Annoskakku 2004 Suomalaisten keskimääräinen efektiivinen annos. 2005. Muikku M, Bly R, Lahtinen J, et al. Suomalaisten keskimääräinen efektiivinen annos Suomalaisten keskimääräinen efektiivinen annos. 2014. Ram G, Chinen J. Infections and immunodeficiency in Down syndrome. Clin Exp Immunol. 2011;164(1):9–16.

haematologica | 2018; 103(11)

ARTICLE

Chronic Lymphocytic Leukemia

Adherence to the Western, Prudent, and Mediterranean dietary patterns and chronic lymphocytic leukemia in the MCC-Spain study Marta Solans,1,2,3* Adela Castelló,1,4,5* Yolanda Benavente,1,6 Rafael Marcos-Gragera,2,3 Pilar Amiano,1,7 Esther Gracia-Lavedan,1,8,9 Laura Costas,6 Claudia Robles,10 Eva Gonzalez-Barca,11 Esmeralda de la Banda,12 Esther Alonso,12 Marta Aymerich,13 Elias Campo,13 Trinidad Dierssen-Sotos,1,14 Guillermo Fernández-Tardón,1,15 Rocio Olmedo-Requena,1,16,17 Eva Gimeno,18 Gemma Castaño-Vinyals,1,8,9,19, Nuria Aragonés,1,20 Manolis Kogevinas,1,8,9,19 Silvia de Sanjose,1,6,21 Marina Pollán1,4** and Delphine Casabonne1,6**

Centro de Investigación Biomédica en Red: Epidemiología y Salud Pública (CIBERESP), Madrid, Spain; 2Research Group on Statistics, Econometrics and Health (GRECS), University of Girona, Spain; 3Epidemiology Unit and Girona Cancer Registry, Catalan Institute of Oncology, Girona, Spain; 4Cancer Epidemiology Unit, National Centre for Epidemiology, Carlos III Institute of Health, Madrid, Spain; 5Faculty of Medicine, University of Alcalá, Alcalá de Henares, Madrid, Spain; 6Unit of Molecular Epidemiology and Genetic in Infections and Cancer (UNIC-Molecular), Cancer Epidemiology Research Programme (IDIBELL), Catalan Institute of Oncology, L' Hospitalet De Llobregat, Spain; 7 Public Health Division of Gipuzkoa, BioDonostia Research Institute, San Sebastian, Spain; 8ISGlobal, Barcelona, Spain; 9Universitat Pompeu Fabra (UPF), Barcelona, Spain; 10 Unit of Information and Interventions in Infections and Cancer (UNIC-I&I), Cancer Epidemiology Research Programme, (IDIBELL), Catalan Institute of Oncology, L' Hospitalet De Llobregat, Spain; 11Hematology, Bellvitge Biomedical Research Institute (IDIBELL), Catalan Institute of Oncology, L'Hospitalet de Llobregat, Spain; 12Hematology Laboratory, Department of Pathology, Hospital Universitari de Bellvitge, L'Hospitalet de Llobregat, Spain; 13Hospital Clinic de Barcelona, University of Barcelona, CIBERONC, Barcelona Spain; 14University of Cantabria - Marqués de Valdecilla Research Institute (IDIVAL), Santander, Spain; 15University Institute of Oncology (IUOPA), University of Oviedo, Spain; 16Department of Preventive Medicine and Public Health, University of Granada, Spain; 17Instituto de Investigación Biosanitaria de Granada, Hospitales Universitarios de Granada, Spain; 18Hematology Department, Hospital del Mar, Barcelona, Spain; 19Hospital del Mar Medical Research Institute (IMIM), Barcelona, Spain; 20Epidemiology Section, Public Health Division, Department of Health of Madrid, Spain and 21PATH, Reproductive Health, Seattle, WA, USA 1

*MS and AC contributed equally to this work. **MP and DC contributed equally to this work.

Ferrata Storti Foundation

Haematologica 2018 Volume 103(11):1881-1888

Correspondence: dcasabonne@iconcologia.net

ABSTRACT

iet is a modifiable risk factor for several neoplasms but evidence for chronic lymphocytic leukemia (CLL) is sparse. Previous studies examining the association between single-food items and CLL risk have yielded mixed results, while few studies have been conducted on overall diet, reporting inconclusive findings. This study aimed to evaluate the association between adherence to three dietary patterns and CLL in the multicase-control study (MCC-Spain) study. Anthropometric, sociodemographic, medical and dietary information was collected for 369 CLL cases and 1605 controls. Three validated dietary patterns, Western, Prudent and Mediterranean, were reconstructed in the MCCSpain data. The association between adherence to each dietary pattern and CLL was assessed, overall and by Rai stage, using mixed logistic regression models adjusted for potential confounders. High adherence to a Western dietary pattern (i.e. high intake of high-fat dairy products, processed meat, refined grains, sweets, caloric drinks, and convenience food) was associated with CLL [ORQ4 vs. Q1=1.63 (95%CI 1.11; 2.39); P-trend=0.02; OR 1-SD increase=1.19 (95%CI: 1.03; 1.37)], independently of Rai stages. No differences in the association were observed according to sex, Body Mass Index, energy intake, tobacco, physical activity, working on a farm, or family history of hematologic malignancies. No associations were observed for Mediterranean and Prudent dietary patterns and CLL. This study provides the first evidence for an association between a Western dietary pattern and CLL, suggesting that a proportion of CLL cases could be prevented by modifying dietary habits. Further research, especially with a prospective design, is warranted to confirm these findings. haematologica | 2018; 103(11)

Received: March 6, 2018. Accepted: June 25, 2018. Pre-published: June 28, 2018. doi:10.3324/haematol.2018.192526 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1881 ©2018 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.

1881

M. Solans et al.

Introduction Chronic lymphocytic leukemia (CLL) is the commonest leukemia among the adult population in Western countries, with an annual incidence rate of around 5 per 100,000 person-years in Europe,1 but its etiology is still poorly understood. A pooled analysis of 2440 CLL cases and 15,186 controls from the InterLymph consortium showed significant inverse associations with atopic conditions, smoking, blood transfusion history, and recreational sun exposure, and positive associations with height, hepatitis C virus seropositivity, living or working on a farm, working as a hairdresser, and family history of hematologic malignancies.2 Diet is a modifiable risk factor for several neoplasms,3 but evidence for CLL is inconclusive. Epidemiological data on the association of diet and CLL are heterogeneous, and mainly arise from studies on nutrients or single food items. While most prospective studies4-12 did not find any association with a wide range of dietary factors, case-control studies13-25 have yielded contradictory results for items such as meat, dairy or vegetable intake. Some authors argue that focusing on overall dietary patterns instead of on individual foods or nutrients may better capture dietary variability in the populationâ&#x20AC;&#x2122;s diet while allowing the evaluation of interactions between dietary factors.26 However, the few studies that have been conducted on overall diet and CLL25,27,28 reported inconclusive findings, mainly due to small sample size. A population-based multicase-control study (MCCSpain) was launched to evaluate the influence of environmental exposures and their interaction with genetic factors in CLL, among other cancers.29 The aim of the present study was to evaluate the association between adherence to three validated dietary patterns,30 Western, Prudent and Mediterranean, and CLL in the MCC-Spain study.

Methods Study population MCC-Spain is a multicentric case-control study with population controls and cases with common tumors (prostate, breast, colorectal, gastroesophageal and CLL) in Spain. Between 2010 and 2013, CLL cases aged 20-85 years were recruited in 11 Spanish hospitals from 5 Spanish provinces (Asturias, Barcelona, Cantabria, Girona and Granada). Simultaneously, population-based controls frequency-matched to cases according to age (5-year intervals), sex, and province of recruitment were randomly selected from primary care centers within the hospitalsâ&#x20AC;&#x2122; catchment areas. Participation rates were 87% in cases and 53% in controls, with variability among geographical regions. After applying specific diet exclusion criteria (excluding participants with no dietary data or with missing or implausible energy intakes under 750 or over 4500 kcal/day), a total of 1605 controls and 369 CLL cases were included in the study. All participants gave informed consent. Approval for the study was obtained from the ethical review boards of all recruiting centers. Additional information regarding the study design has been provided elsewhere.29

Outcome definition Chronic lymphocytic leukemia cases were diagnosed according to the International Workshop on CLL criteria: presence of an absolute count â&#x2030;Ľ5 x109 B cells/L for three or more months in peripheral blood and a clonal population of CD5+, CD19+, and 1882

CD23+ B cells.31 All diagnoses were morphologically and immunologically confirmed using flow cytometry immunophenotype and complete blood cell count. CLL and small lymphocytic lymphoma were considered the same underlying disease. Given the indolent course of the disease, CLL cases were recruited and interviewed within three years from diagnosis. Disease severity at interview was evaluated using the Rai staging system obtained from medical records and verified by local hematologists. CLL subjects were then categorized into two groups based on Rai stage: a) low-risk category including asymptomatic patients with lymphocytosis only (Rai 0); and b) intermediate/high-risk category including patients with either lymphadenopathy, hepatomegaly, splenomegaly, anemia and/or thrombocytopenia (Rai I-IV).

Data collection Data on socio-demographic factors, lifestyle and personal/family medical history were collected through face-to-face interviews performed by trained personnel. Height and weight at different ages were self-reported. The questionnaire in Spanish is available at www.mccspain.org. In addition, subjects were provided a semi-quantitative Food Frequency Questionnaire (FFQ), which was a modified version from a previous tool validated in Spain to include regional products.32 The FFQ was self-administered and returned by mail or filled out face-to-face. It included 140 food items with portion sizes specified for each item, and assessed usual dietary intake during the previous year. Cross-check questions on aggregated food group consumption were used to adjust the frequency of food consumption and reduce misreporting of food groups with large numbers of items.33 Nutrient intakes were estimated using food composition tables published for Spain, and other sources.34 The response rate of the FFQ was slightly lower in cases (82%) than in controls (87%). Overall, responsiveness was not associated with age, and individuals from Granada were less likely to answer the FFQ than those from Barcelona. Those individuals who did not answer the diet questionnaire had a lower level of education and, in controls, were also more likely to be women.

Dietary patterns Three validated dietary patterns identified in a Spanish casecontrol study (EpiGEICAM)30 were reconstructed in the MCCstudy: a) a Western dietary pattern characterized by high intake of high-fat dairy products, processed meat, refined grains, sweets, caloric drinks, convenience food and sauces; b) a Prudent pattern, with high intake of low-fat dairy products, vegetables, fruits, whole grains and juices; and c) a Mediterranean pattern, defined by a high intake of fish, vegetables, legumes, boiled potatoes, fruits, olives, and vegetable oil. Further information on identification of the dietary patterns can be found elsewhere.30 In brief, dietary information extracted from a semi-quantitative questionnaire in the EpiGEICAM study was converted to mean daily intake in grams and grouped into 26 food categories. Major existing dietary patterns were identified in the control population by applying principal components analysis (PCA) without rotation of the variance-covariance matrix over the 26 inter-correlated food groups. The set of loadings obtained represent the correlation between the consumption of each food group and the component/pattern score, and can be used to apply such patterns to other populations.35 In the MCC-study, we grouped the FFQ items into the same 26 food groups (Online Supplementary Table S1), and calculated the score of adherence to the Western, Prudent and Mediterranean dietary patterns as a linear combination of the loads described in the EpiGEICAM study and the log-transformed centered food group consumption reported by the participants of MCC-Spain study. haematologica | 2018; 103(11)

Dietary patterns and chronic lymphocytic leukemia

Figure 1. Association between adherence to dietary patterns and chronic lymphocytic leukemia in the multicase-control (MCCSpain) study. OR: Odds Ratio; 95%CI: 95% confidence interval; Q: quartile; SD: standard deviation. Black squares indicate OR and horizontal lines repre1 Logistic sent 95%CI. regression models adjusted for age, sex, education, energy intake (kcal/day) with province of residence as random effect.

Statistical analysis As descriptive analyses, we compared anthropometric, sociodemographic and lifestyle characteristics between cases and controls. χ2 test was used to evaluate the level of significance of the differences observed in categorical variables, Student t-test for normally distributed continuous variables, and Wilcoxon rank-sum test for non-normally distributed continuous variables. In addition, we analyzed the distribution of each dietary pattern (continuous) across categories of descriptive variables. Student t-test was used to assess differences observed in variables with two categories and ANOVA for those with more than two categories. The association between the dietary patterns and CLL was evaluated using mixed logistic regression models with random province-specific intercepts. The exposure variables (adherence to Western, Prudent or Mediterranean patterns) were included in the model both as continuous variables [1-standard deviation (SD) increase in the controls’ scores] and as categorical variables (according to the quartile distribution in all controls). All models were adjusted for age (years, continuous), sex, education (no formal education, primary school, secondary school, university), and energy intake (kcal/day, continuous) as fixed effects and province of residence as a random effect term. Height (cm, continuous), waist-to-hip ratio (continuous), Body Mass Index (BMI in kg/m2, continuous), experience working on a farm (yes, no), family history of hematologic malignancies (yes, no), alcohol consumption (g/day, continuous), smoking status (never, past, current), and physical activity [in the last 10 years, measured in Metabolic Equivalent of Task (METs)/week: inactive (0), low (0.1-8), moderate (8-15.9) and very active (≥16)] were examined as potential confounders, but were not included in the final models as they were not found alone, or in combination, to affect the estimates. Interaction terms were modeled between each of these separate variables and the dietary score (continuous), and tested using loglikelihood ratio tests. A possible effect modification of sex, BMI, energy intake, tobacco, physical activity, working on a farm, and family history of hematologic malignancies was tested including an interaction term between each of the patterns and such varihaematologica | 2018; 103(11)

ables. The estimation of the effects according to Rai stage (0 vs. I-IV) was calculated with multinomial logistic regression models adjusted by the set of variables described above plus province of residence as random effect term. Finally, sensitivity analyses were performed to examine how the inclusion of: i) cases with longer period of time from diagnosis to recruitment (<1 year vs. ≥ 1 year); and ii) cases treated before the interview affected the overall estimates. Odds Ratios (OR) and 95% confidence intervals (CI) were also obtained with multinomial logistic regression models. The P-value for heterogeneity of effects across Rai stage and for sensitivity analyses was obtained with the Wald test. All analyses were performed using STATA/MP (v.14.1, 2015, StataCorp LP) and statistical significance was set at two-sided P<0.05.

Results Distribution of baseline characteristics between cases and controls is shown in Table 1. Compared with controls, cases were more adherent to the Western pattern, while no differences in level of adherence to the Prudent and Mediterranean patterns were observed in bivariate analyses. CLL cases were also slightly older, had a higher waistto-hip ratio, and were more likely to have a family history of hematologic malignancy and to have worked on a farm. No other differences were observed for any of the other pre-selected variables. The distribution of key characteristics of controls according to level of adherence to each dietary pattern is shown in Online Supplementary Figure S1. Controls with greater adherence to a Western pattern were more likely to be men, younger, taller, current smokers, less prone to have worked in farming or agriculture, had a lower BMI and waist-to-hip ratio, and a higher level of education, energy and alcohol intake. Those with a higher adherence to a Prudent pattern were more likely to be women, younger, taller, physically active, never/former smokers, 1883

M. Solans et al. Table 1. Baseline characteristics of chronic lymphocytic leukemia for cases and controls in the multicase-control (MCC-Spain) study.

Western, mean (SD) Prudent, mean (SD) Mediterranean, mean (SD) Province, n(%) Barcelona Asturias Cantabria Granada Girona Age (years), mean (SD) Sex, n(%) Male Female Energy intake (kcal/day), mean (SD) Current alcohol intake (g/day), median (IQI)2 BMI (kg/m2), mean (SD)2 Height (cm), mean(SD)2 Waist-to-hip ratio3, n(%) Low Moderate High Unknown Smoking status, n(%) Never Former Current Unknown Education, n(%) No formal education Primary Secondary University Physical activity4, n(%) Inactive Low Moderate Very active Unknown Ever worked in farming or agriculture, n(%) No Yes Unknown Family history of hematologic malignancy, n(%) No Yes Rai stage 0 I-IV Unknown

Controls (n=1605)

Cases (n=369)

5.88 (1.46) 6.55 (1.12) 7.08 (1.00)

6.06 (1.40) 6.66 (1.03) 7.16 (0.88)

0.03 0.07 0.20 <0.001

900 (56) 211 (13) 281 (18) 144 (9) 69 (4) 64.30 (10.54)

242 (66) 51 (14) 21 (6) 27 (7) 28 (8) 66.19 (10.12)

936 (58) 669 (42) 1901.15 (585.88) 8.80 (0.58;27.32) 26.99 (4.50) 165.63 (8.51)

217 (59) 152 (41) 1937.91 (612.06) 8.82 (0.83;24.47) 27.32 (4.43) 165.97 (9.11)

460 (29) 449 (28) 682 (42) 14 (1)

77 (21) 98 (27) 192 (52) 2 (1)

696 (43) 602 (38) 303 (19) 4 (0)

165 (45) 134 (36) 67 (18) 3 (1)

357 (22) 502 (31) 461 (29) 285 (18)

94 (25) 106 (29) 107 (29) 62 (17)

656 (41) 219 (14) 190 (12) 502 (31) 38 (2)

136 (37) 55 (15) 47 (13) 118 (32) 13 (4)

1257 (78) 323 (20) 25 (2)

258 (70) 108 (29) 3 (1)

1551 (97) 54 (3)

333 (90) 36 (10)

199 (54) 150 (41) 20 (5)

0.002 0.95

0.28 0.75 0.21 0.50 0.004

0.39

0.54

0.50

<0.001

SD: standard deviation; IQI: interquartile interval; BMI: Body Mass Index. 1P-value for heterogeneity calculated with the Student t-test for comparison of normally distributed continuous variables, with the Wilcoxon rank-sum test for comparison of non-normally distributed continuous variables (alcohol intake), and with the Ď&#x2021;2 test for categorical variables. 2% of missing values in continuous variables: alcohol intake (2%), BMI (4%), height (3%). 3Waist-to-hip ratio risk categories according to WHO criteria. 4Physical activity, in the last ten years, measured in METs/week: inactive (0), low (0.1-8), moderate (8-15.9), and very active (â&#x2030;Ľ16). In bold: P<0.05.

1884

haematologica | 2018; 103(11)

Dietary patterns and chronic lymphocytic leukemia

Table 2. Association between adherence to dietary patterns and chronic lymphocytic leukemia by severity of the disease, in the multicase-control (MCC-Spain) study.

Controls N(%) (n=1605) Western Q1 Q2 Q3 Q4 P-trend 1-SD increase Prudent Q1 Q2 Q3 Q4 P-trend 1-SD increase Mediterranean Q1 Q2 Q3 Q4 P-trend 1-SD increase

402 (25) 401 (25) 401 (25) 401 (25)

Rai 0 Cases N(%) OR1 (95% CI) (n=199) 45 (23) 47 (24) 47 (24) 60 (30)

42 (21) 53 (27) 52 (26) 52 (26)

48 (24) 51 (26) 55 (28) 45 (23)

Rai I-IV Cases (n=150)

OR1 (95% CI)

P-het2

1 1.11 (0.71;1.74) 1.17 (0.73;1.87) 1.60 (0.97;2.65) 0.07 1.15 (0.95;1.39)

26 (17) 39 (26) 37 (25) 48 (32)

1 1.40 (0.82;2.38) 1.32 (0.76;2.30) 1.71 (0.95;3.06) 0.11 1.26 (1.02;1.56)

0.50

1 1.09 (0.70;1.70) 1.07 (0.68;1.69) 1.01 (0.62;1.64) 0.98 0.99 (0.83;1.18)

30 (20) 37 (25) 32 (21) 51 (34)

1 1.16 (0.69;1.93) 0.97 (0.56;1.66) 1.47 (0.86;2.51) 0.23 1.18 (0.96;1.45)

0.17

1 0.88 (0.57;1.36) 0.88 (0.57;1.37) 0.65 (0.40;1.06) 0.11 0.88 (0.74;1.04)

29 (19) 29 (19) 46 (31) 46 (31)

1 0.94 (0.55;1.62) 1.39 (0.83;2.30) 1.32 (0.76;2.27) 0.17 1.15 (0.93;1.41)

0.04

OR: Odds Ratio; 95% CI: 95% Confidence Interval; Q: quartile; SD: Standard Deviation. 1Multinomial logistical regression models adjusted for age, sex, education, energy intake (kcal/day) with province of residence as a random effect. 2P-value for the heterogeneity of effects. In bold: P<0.05.

more highly educated, less prone to have worked in farming or agriculture, and with a higher energy intake and lower alcohol consumption. Finally, controls with a greater adherence to a Mediterranean pattern were more likely to be men, physically active, showing a lower proportion of ever smokers and having worked in farming or agriculture, and a higher energy intake. Figure 1 summarizes the adjusted ORs for the association between CLL and level of adherence to the Western, Prudent and Mediterranean dietary patterns. Individuals in the highest quartile of the Western score had an OR for CLL of 1.63 (95%CI: 1.11; 2.39) compared with individuals with low adherence (P for trend 0.02). Each SD increment in the score was associated with a 19% higher OR of having CLL (95%CI: 1.03; 1.37). No associations were observed for Mediterranean and Prudent diet patterns. The impact of each individual covariate (region, age, sex, education, and energy intake) in the association of the three dietary patterns and CLL is provided in Online Supplementary Figure S2. Since CLL is more prevalent in men, who are also more likely to adhere to a Western dietary pattern (Table 1 and Online Supplementary Figure S1), all analyses were stratified according to sex. No differences across sexes were observed for any of the dietary patterns [P-heterogeneity (P-het): Western (0.79), Prudent (0.11) and Mediterranean (0.17); data not shown]. In addition, no differences were observed according to BMI, energy intake, tobacco, physical activity, working on a farm, and family history of haematologica | 2018; 103(11)

hematologic malignancies (all P for interaction >0.05; data not shown). Analyses according to Rai-stage did not show significant heterogeneity of effects for the Western or Prudent dietary patterns (P-het=0.50 and 0.17, respectively). However, weak opposite trends in relation to a Mediterranean diet pattern were observed; it was inversely associated (although not statistically significant) with Rai 0 CLL [OR 1-SD increase= 0.88 (95%CI: 0.74; 1.04)] and positively related with Rai I-IV CLL [OR 1-SD increase= 1.15 (95%CI: 0.93; 1.41)] (P-het=0.04) (Table 2). Sensitivity analyses according to time from diagnosis to recruitment yielded similar results for the three dietary patterns (Online Supplementary Table S2). Similarly, excluding cases treated prior to consent (n=79) did not materially modify the results [P-het in trends: Western (0.25), Prudent (0.32) and Mediterranean (0.33)], but higher ORs for a Western dietary pattern were observed in cases treated prior to consent in comparison to those not treated (Online Supplementary Table S3).

Discussion This study provides, for the first time, evidence of an association between adherence to a Western dietary pattern and CLL. By contrast, no associations were found for a Prudent or Mediterranean pattern. There is limited evidence linking extrinsic-risk factors, 1885

M. Solans et al.

and particularly diet, with non-Hodgkin lymphoma (NHL). In the 2007 report by the World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR),3 the Panel decided not to make any judgements regarding the causality of associations between specific dietary factors and NHL, but pointed out a suggestive inverse association with vegetables, fruit, and alcoholic beverages, and a positive association with meat, total fat, body fatness, and dairy. Recent meta-analyses further support these associations,36,37 but there is still not sufficient evidence to establish a causal role. Similarly, data on the association of diet and CLL are inconclusive and mainly arise from studies on nutrients or single food items. To our knowledge, 9 prospective studies4-12 and 13 case-control studies13-25 have been published on this topic. With the exception of a few studies that found positive associations with consumption of processed meat and poultry,4 total carbohydrate8 or fat (in women)11 intake, and inverse associations with isoflavones consumption,10 generally large prospective studies found no associations between a wide range of dietary factors and CLL. By contrast, case-control studies have yielded contradictory results for meat,13,14,16,24,25 dairy products,13-17 fish15,18 or vegetables and fruit14,16,19,22,23,25 intake. Inconsistencies in previous epidemiological nutritional studies in part reflect the difficulty in disentangling the influence of single food items that, when consumed in combination, may be highly correlated and exert synergistic or antagonistic effects on CLL risk. The examination of dietary patterns, which better reflect the complexity of dietary intake, has been used to address such limitations.26 So far, only a few studies have examined associations of dietary patterns and risk of CLL,25,27,28 reporting inconclusive findings, mainly due to small sample size. Ollberding et al.28 pointed out that a high adherence to a ‘Meat, Fat and Sweets’ dietary pattern, characterized by a high intake of French fries, red meat, processed meat, pizza, salty snacks, sweets and desserts, was associated with an increased risk of overall NHL (ORQ4 vs. Q1=3.6; 95%CI: 1.9, 6.8) in a Nebraska case-control study. This association was maintained when stratifying according to lymphoma subtypes but sub-analyses did not include CLL cases due to sample size (n=25). By contrast, a large prospective cohort in the US did not find associations with ‘Fat and Meat’ pattern and CLL etiology.27 However, this pattern did not include sweets and deserts, sweetened beverages, or convenience foods, which may be important contributing factors of these associations. Thus, not only differences in the study design and setting, but also in food groups loaded in these data-driven analyses, should be carefully considered when comparing results. In line with our findings, no associations with overall NHL25,27,28 or CLL25,27 were detected for a ‘healthy’ dietary pattern characterized by high intake of fruit and vegetables. We observed opposite trends in relation to a Mediterranean diet pattern and Rai stages, with stronger adherence among cases with higher disease severity (P-het=0.04). We hypothesize that reverse causality could partly explain these results. While Rai 0 patients are usually diagnosed in a routine blood test and present an indolent course, Rai I-IV are more prone to be symptomatic (e.g. night sweat, fatigue, weight loss or fever) and to receive active treatment. Thus, those patients with a more severe disease (and probably more concerned about their illness) would be more prone to shift towards a healthier 1886

dietary pattern. However, these results have to be taken with caution since none of the trends showed statistically significant associations. Chronic lymphocytic leukemia is the most common leukemia in Western countries while its incidence is much lower in Eastern countries, where it accounts for only 13% of NHL in most series.38 While genetic backgrounds may be responsible for some of the differences in the CLL incidence, some studies have suggested that environmental factors also play an important role. A dramatic increase in CLL incidence in Taiwan in recent years was associated with a strong birth-cohort effect, that corresponded to the Westernization of lifestyle in Taiwan since 1960.39 In addition, a higher incidence of CLL has been reported among US-born Asians compared to foreign-born Asians, pointing out the influence of environmental factors that change with immigration and acculturation to a Westernized lifestyle.40 Our results further support the view that adopting a Western diet could partly explain these incidence patterns. A Western diet has been associated with obesity phenotypes,41 including a higher waist-to-hip ratio, which has in turn been recently linked to higher OR of CLL, particularly in women, in the MCC-Spain study.42 Despite the fact that CLL cases showed a higher waist-to-hip ratio than controls in our study, waist-to-hip ratio and BMI were not included as covariates in the final multiple-adjusted model since they did not change risk estimates. Hence, an independent effect of the Western dietary pattern may be contributing to CLL lymphomagenesis, which seems plausible from a mechanistic point of view. On one hand, it has been well-established that dietary changes, and particularly switching from a low-fat, plant polysaccharide-rich diet to a high-fat, high-sugar Western diet, can induce alterations in microbiota composition.43 Beyond its role in the biosynthesis of key components (e.g. vitamins, essential amino acids or short chain fatty acid byproducts), several studies using germ-free mice suggest that microbiota also plays a fundamental role on the induction, training, and function of the host immune system.44 Exposure to a Western diet may have selected for a microbiota that lack the resilience and diversity required to establish balanced immune responses, and this phenomenon is proposed to account for some of the dramatic rise in autoimmune and chronic inflammatory disorders found in high-income countries. On the other hand, a diet high in fat, refined grains, red and processed meats, and sweets has been largely associated with higher levels of inflammatory markers45 and with inflammation-related chronic diseases.46 In particular in CLL, the strong production of inflammatory cytokines and chemokines accompanied by activation of intra-cellular pro-inflammatory pathways, and the presence of somatic mutations that activate proinflammatory signaling pathways, suggest that chronic inflammation plays a pathophysiological role in this disease.47 Thus, an inflammation-related mechanism may in part underlie the observed associations with CLL, although no research on the inflammatory potential of diet and CLL risk has yet been conducted. The dietary patterns used in this study were identified using the control population of a multicentric case-control study on female breast cancer in Spain.30 By contrast, the MCC-Spain study included male participants, who may have different dietary habits. However, this difference does not preclude the application of the original scoring haematologica | 2018; 103(11)

Dietary patterns and chronic lymphocytic leukemia

system over the current sample. Scores of adherence to dietary patterns can be calculated following the exact same rules over different populations, resulting in different levels of adherence while still being valid, as has been recently proved.35 As a matter of fact, the current dietary patterns had previously been constructed in the MCCstudy and a Western dietary pattern was positively associated with gastric,48 breast49 and colorectal50 cancers. One of the main limitations is the study design since case-control studies are prone to selection and recall biases. Measurement errors in the estimation of food intake due to the use of self-reported FFQ are also of some concern. However, the FFQ was validated in the Spanish population and included regional products.32 Moreover, some questions about general dietary habits were included in the questionnaire and were used to adjust the responses to the FFQ following the methodology described in Calvert et al.33 The inclusion of prevalent cases might be another cause for concern since patients who survived might have a very different etiology than those who died soon after diagnosis. In addition, diet can be influenced by many external factors and patients who survive longer might have substantially modified their diet. However, results of the sensitivity analysis suggested that the use of prevalent cases might not have introduced selective survival bias or reverse causation. We may have been limited by the small sample size and lack of statistical power to detect significant associations when evaluating certain subgroups. Finally, although we adjusted for a range of potential confounders, residual confounding factors cannot be totally ruled out. The strengths of the study include the substantial sample size of CLL cases, with specific information on clinical presentation. We were able to collect detailed information on demographics and disease stage, and statistically adjust for a number of potential confounding factors. This allowed the evaluation of potential interactions of diet

References 1. Sant M, Allemani C, Tereanu C, et al. Incidence of hematologic malignancies in Europe by morphologic subtype: results of the HAEMACARE project. Blood. 2010;116(19):3724-3734. 2. Slager SL, Benavente Y, Blair A, et al. Medical History, Lifestyle, Family History, and Occupational Risk Factors for Chronic Lymphocytic Leukemia/Small Lymphocytic Lymphoma: The InterLymph Non-Hodgkin Lymphoma Subtypes Project. JNCI Monogr. 2014; 2014(48):4151. 3. WCRF. World Cancer Research Fund. Food, Nutrition, Physical Activity, and the Prevention of Cancer: a Global Perspective. 2nd ed. Washington, USA: American Institute for Cancer Research, 2007. 4. Rohrmann S, Linseisen J, Jakobsen MU, et al. Consumption of meat and dairy and lymphoma risk in the European Prospective Investigation into Cancer and Nutrition. Int J Cancer. 2011;128(3):623-634. 5. Daniel CR, Sinha R, Park Y, et al. Meat

haematologica | 2018; 103(11)

10.

with numerous covariates and the exploration of the associations by stage. Finally, the multi-centric nature of the study, including both rural and urban areas, provided a wide geographic variability of dietary intake data. In conclusion, in this Spanish population-based casecontrol study, greater adherence to a Western dietary pattern was associated with CLL. These novel results suggest that a proportion of CLL cases could be prevented by modifying dietary patterns. Further research, especially with a prospective design, is warranted to confirm these findings. Acknowledgments The authors would like to thank all the subjects who participated in the study and all CLL MCC-Spain collaborators (the list can be found the Online Supplementary Appendix, List S1). Funding Predoctoral contract to MS (CIBERESP), Spanish Ministry of Economy and Competitiveness Juan de la Cierva de Incorporación grant IJCI-2014-20900. Spanish Ministry of Economy and Competitiveness - Carlos III Institute of Health cofunded by FEDER funds/European Regional Develpment Fund (ERDF) - a way to build Europe [(grants PI17/01280, PI11/01810, PI14/01219, PI11/02213, PI09/1662, PI15/00966, RCESP C03/09, RTICESP C03/10, RTIC RD06/0020/0095, RD12/0036/0056, Rio Hortega CM13/00232, SV-09-CLINIC-1 and Centro de Investigación Biomédica en Red: Epidemiología y Salud Pública (CIBERESP))] and the Agència de Gestió d'Ajuts Universitaris i de Recerca AGAUR (2017SGR1085, 2014SGR756). The ICGC CLL-Genome Project was funded by Spanish Ministerio de Economía y Competitividad (MINECO) through the Instituto de Salud Carlos III (ISCIII), PMP15/00007 and Centro de Investigación Biomédica en Red: Oncología (CIBERONC). ISGlobal is a member of the CERCA Programme, Generalitat de Catalunya.

Intake Is Not Associated with Risk of NonHodgkin Lymphoma in a Large Prospective Cohort of U.S. Men and Women. J Nutr. 2012;142(6):1074-1080. Tsai H-T, Cross AJ, Graubard BI, Oken M, Schatzkin A, Caporaso NE. Dietary factors and risk of chronic lymphocytic leukemia and small lymphocytic lymphoma: a pooled analysis of two prospective studies. Cancer Epidemiol Biomarkers Prev. 2010;19(10):2680-2684. McCullough ML, Teras LR, Shah R, Diver WR, Gaudet MM, Gapstur SM. Artificially and sugar-sweetened carbonated beverage consumption is not associated with risk of lymphoid neoplasms in older men and women. J Nutr. 2014;144(12):2041-2049. Ross JA, Kasum CM, Davies SM, Jacobs DR, Folsom AR, Potter JD. Diet and risk of leukemia in the Iowa Women’s Health Study. Cancer Epidemiol Biomarkers Prev. 2002;11(8):777-781. Erber E, Maskarinec G, Lim U, Kolonel LN. Dietary vitamin D and risk of non-Hodgkin lymphoma: the multiethnic cohort. Br J Nutr. 2010;103(4):581-584. Chang ET, Canchola AJ, Clarke CA, et al.

11.

12.

13.

14.

15.

Dietary phytocompounds and risk of lymphoid malignancies in the California Teachers Study cohort. Cancer Causes Control. 2011;22(2):237-249. Bertrand KA, Giovannucci E, Rosner BA, Zhang SM, Laden F, Birmann BM. Dietary fat intake and risk of non-Hodgkin lymphoma in 2 large prospective cohorts. Am J Clin Nutr. 2017;106(2):650-656. Kabat GC, Kim MY, Wactawski-Wende J, Shikany JM, Vitolins MZ, Rohan TE. Intake of antioxidant nutrients and risk of nonHodgkin’s Lymphoma in the Women’s Health Initiative. Nutr Cancer. 2012; 64(2):245-254. Charbonneau B, O’Connor HM, Wang AH, et al. Trans fatty acid intake is associated with increased risk and n3 fatty acid intake with reduced risk of non-hodgkin lymphoma. J Nutr. 2013;143(5):672-681. Chang ET, Smedby KE, Zhang SM, et al. Dietary factors and risk of non-hodgkin lymphoma in men and women. Cancer Epidemiol Biomarkers Prev. 2005; 14(2):512-520. Chang ET, Balter KM, Torrang A, et al. Nutrient intake and risk of non-Hodgkin’s

1887

M. Solans et al.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26. 27.

28.

1888

lymphoma. Am J Epidemiol. 2006; 164(12): 1222-1232. Purdue MP, Bassani DG, Klar NS, Sloan M, Kreiger N. Dietary factors and risk of nonHodgkin lymphoma by histologic subtype: a case-control analysis. Cancer Epidemiol Biomarkers Prev. 2004;13(10):1665-1676. Mikhak B, Bracci PM, Gong Z. Intake of vitamins D and A and calcium and risk of non-Hodgkin lymphoma: San Francisco Bay Area population-based case-control study. Nutr Cancer. 2012;64(5):674-684. Fritschi L, Ambrosini GL, Kliewer E V, Johnson KC. Dietary fish intake and risk of leukaemia, multiple myeloma, and nonHodgkin lymphoma. Cancer Epidemiol Biomarkers Prev. 2004;13(4):532-537. Holtan SG, O’Connor HM, Fredericksen ZS, et al. Food-frequency questionnairebased estimates of total antioxidant capacity and risk of non-Hodgkin lymphoma. Int J cancer. 2012;131(5):1158-1168. Koutros S, Zhang Y, Zhu Y, et al. Nutrients contributing to one-carbon metabolism and risk of non-Hodgkin lymphoma subtypes. Am J Epidemiol. 2008;167(3):287-294. Luczynska A, Kaaks R, Rohrmann S, et al. Plasma 25-hydroxyvitamin D concentration and lymphoma risk: results of the European Prospective Investigation into Cancer and Nutrition. Am J Clin Nutr. 2013;98(3):827-838. Chiu BC-H, Kwon S, Evens AM, Surawicz T, Smith SM, Weisenburger DD. Dietary intake of fruit and vegetables and risk of non-Hodgkin lymphoma. Cancer Causes Control. 2011;22(8):1183-1195. Casabonne D, Gracia E, Espinosa A, et al. Fruit and vegetable intake and vitamin C transporter gene (SLC23A2) polymorphisms in chronic lymphocytic leukaemia. Eur J Nutr. 2017;56(3):1123-1133. Aschebrook-Kilfoy B, Ollberding NJ, Kolar C, et al. Meat intake and risk of nonHodgkin lymphoma. Cancer Causes Control. 2012;23(10):1681-1692. Campagna M, Cocco P, Zucca M, et al. Risk of lymphoma subtypes and dietary habits in a Mediterranean area. Cancer Epidemiol. 2015;39(6):1093-1098. Hu FB. Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin Lipidol. 2002;13(1):3-9. Erber E, Maskarinec G, Gill JK, Park S-Y, Kolonel LN. Dietary patterns and the risk of non-Hodgkin lymphoma: the Multiethnic Cohort. Leuk Lymphoma. 2009; 50(8):1269-1275. Ollberding NJ, Aschebrook-Kilfoy B, Caces DBD, Smith SM, Weisenburger DD, Chiu

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

BC-H. Dietary patterns and the risk of nonHodgkin lymphoma. Public Health Nutr. 2014;17(7):1531-1537. Castano-Vinyals G, Aragones N, PerezGomez B, et al. Population-based multicase-control study in common tumors in Spain (MCC-Spain): rationale and study design. Gac Sanit. 2015;29(4):308-315. Castello A, Pollan M, Buijsse B, et al. Spanish Mediterranean diet and other dietary patterns and breast cancer risk: case-control EpiGEICAM study. Br J Cancer. 2014;111(7):1454-1462. Hallek M, Cheson BD, Catovsky D, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood. 2008; 111(12):5446-5456. Garcia-Closas R, Garcia-Closas M, Kogevinas M, et al. Food, nutrient and heterocyclic amine intake and the risk of bladder cancer. Eur J Cancer. 2007; 43(11):17311740. Calvert C, Cade J, Barrett JH, Woodhouse A. Using cross-check questions to address the problem of mis-reporting of specific food groups on Food Frequency Questionnaires. UKWCS Steering Group. United Kingdom Women’s Cohort Study Steering Group. Eur J Clin Nutr. 1997;51(10):708-712. Tablas de composición de alimentos del CESNID [Internet]. [cited 2017 Nov 9]. Available from: http://www.sennutricion.org/es/2013/ 05/13/tablas-de-composicin-de-alimentosdel-cesnid Castello A, Buijsse B, Martin M, et al. Evaluating the Applicability of Data-Driven Dietary Patterns to Independent Samples with a Focus on Measurement Tools for Pattern Similarity. J Acad Nutr Diet. 2016; 116(12):1914-1924.e6. Chen GC, Lv DB, Pang Z, Liu QF. Fruits and vegetables consumption and risk of nonHodgkin’s lymphoma: a meta-analysis of observational studies. Int J Cancer. 2013; 133(1):190-200. Caini S, Masala G, Gnagnarella P, et al. Food of animal origin and risk of nonHodgkin lymphoma and multiple myeloma: A review of the literature and metaanalysis. Crit Rev Oncol Hematol. 2016; 100:16-24. Anderson JR, Armitage JO, Weisenburger DD. Epidemiology of the non-Hodgkin’s lymphomas: distributions of the major sub-

39.

40.

41.

42.

43.

44. 45.

46.

47.

48.

49.

50.

types differ by geographic locations. NonHodgkin’s Lymphoma Classification Project. Ann Oncol. 1998;9(7):717-720. Wu S-J, Huang S-Y, Lin C-T, Lin Y-J, Chang C-J, Tien H-F. The incidence of chronic lymphocytic leukemia in Taiwan, 1986-2005: a distinct increasing trend with birth-cohort effect. Blood. 2010; 116(22):4430-4435. Clarke CA, Glaser SL, Gomez SL, et al. Lymphoid malignancies in U.S. Asians: incidence rate differences by birthplace and acculturation. Cancer Epidemiol Biomarkers Prev. 2011;20(6):1064-1077. Paradis AM, Godin G, Perusse L, Vohl MC. Associations between dietary patterns and obesity phenotypes. Int J Obes (Lond). 2009;33(12):1419-1426. Benavente Y, Casabonne D, Costas L, et al. Established and suggested exposures on CLL/SLL etiology: Results from the CLLMCC-Spain study. Cancer Epidemiol. 2018;52:106–111. Turnbaugh PJ, Ridaura VK, Faith JJ, Rey FE, Knight R, Gordon JI. The effect of diet on the human gut microbiome: a metagenomic analysis in humanized gnotobiotic mice. Sci Transl Med. 2009;1(6):6ra14. Belkaid Y, Hand TW. Role of the microbiota in immunity and inflammation. Cell. 2014;157(1):121-141. Barbaresko J, Koch M, Schulze MB, Nothlings U. Dietary pattern analysis and biomarkers of low-grade inflammation: a systematic literature review. Nutr Rev. 2013;71(8):511–527. Thorburn AN, Macia L, Mackay CR. Diet, metabolites, and “western-lifestyle” inflammatory diseases. Immunity. 2014 Jun;40(6):833–842. Caligaris-Cappio F. Inflammation, the microenvironment and chronic lymphocytic leukemia. Haematologica. 2011; 96(3):353–355. Castello A, Fernandez de Larrea N, Martin V, et al. High adherence to the Western, Prudent, and Mediterranean dietary patterns and risk of gastric adenocarcinoma: MCC-Spain study. Gastric Cancer. 2018;21(3):372-382. Castello A, Boldo E, Perez-Gomez B, et al. Adherence to the Western, Prudent and Mediterranean dietary patterns and breast cancer risk: MCC-Spain study. Maturitas 2017;103:8-15. Castelló A, Amiano P, Fernández de Larrea N, al. Low adherence to the Western and high adherence to the Mediterranean dietary patterns could prevent colorectal cancer. Eur J Nutr. 2018 Mar 26. [Epub ahead of print]

haematologica | 2018; 103(11)

ARTICLE

Chronic Lymphocytic Leukemia

Safety of obinutuzumab alone or combined with chemotherapy for previously untreated or relapsed/refractory chronic lymphocytic leukemia in the phase IIIb GREEN study Véronique Leblond,1 Melih Aktan,2 Christelle M. Ferra Coll,3 Caroline Dartigeas,4 Jens Kisro,5 Marco Montillo,6 João Raposo,7 Jean-Louis Merot,8 Susan Robson,9 Ekaterina Gresko,9 Francesc Bosch,10 Stephan Stilgenbauer11 and Robin Foà12

UPMC GRC11-GRECHY, AP-HP Hôpital Pitié-Salpêtrière, Paris, France; 2Istanbul Üniversitesi, Turkey; 3Institut Català d'Oncologia (ICO), Hospital Germans Trias i Pujol, Josep Carreras Leukaemia Research Institute, Barcelona, Spain; 4Hôpital Bretonneau CHU de Tours, France; 5Onkologische Schwerpunktpraxis Lübeck, Germany; 6Niguarda Cancer Center, ASST Grande Ospedale Metropolitano Niguarda, Milan, Italy; 7Hospital de Santa Maria, Lisbon, Portugal; 8IQVIA, St Ouen, France; 9F. Hoffmann-La Roche Ltd., Basel, Switzerland; 10University Hospital Vall d’Hebron, Barcelona, Spain; 11Internal Medicine III, Ulm University, Germany and 12Hematology, ‘Sapienza’ University, Rome, Italy 1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(11):1889-1898

ABSTRACT

he safety of obinutuzumab, alone or with chemotherapy, was studied in a non-randomized, open-label, non-comparative, phase IIIb study (GREEN) in previously untreated or relapsed/refractory chronic lymphocytic leukemia. Patients received obinutuzumab 1000 mg alone or with chemotherapy (investigator’s choice of fludarabinecyclophosphamide for fit patients, chlorambucil for unfit patients, or bendamustine for any patient) on days 1, 8 and 15 of cycle 1, and day 1 of cycles 2-6 (28-day cycles), with the cycle 1/day 1 dose administered over two days. The primary end point was safety/tolerability. Between October 2013 and March 2016, 972 patients were enrolled and 971 treated (126 with obinutuzumab monotherapy, 193 with obinutuzumab-fludarabine-cyclophosphamide, 114 with obinutuzumab-chlorambucil, and 538 with obinutuzumab-bendamustine). Grade ≥3 adverse events occurred in 80.3% of patients, and included neutropenia (49.9%), thrombocytopenia (16.4%), anemia (9.6%), and pneumonia (9.0%); rates were similar in first-line and relapsed/refractory patients, and in first-line fit and unfit patients. Using expanded definitions, infusion-related reactions were observed in 65.4% of patients (grade ≥3, 19.9%; mainly seen during the first obinutuzumab infusion), tumor lysis syndrome in 6.4% [clinical and laboratory; highest incidence with obinutuzumab-bendamustine (9.3%)], and infections in 53.7% (grade ≥3, 20.1%). Serious and fatal adverse events were seen in 53.1% and 7.3% of patients, respectively. In first-line patients, overall response rates at three months post treatment exceeded 80% for all obinutuzumab-chemotherapy combinations. In the largest trial of obinutuzumab to date, toxicities were generally manageable in this broad patient population. Safety data were consistent with previous reports, and response rates were high. (clinicaltrials.gov identifier: 01905943). Introduction Obinutuzumab (GA101) is a glycoengineered, type II anti-CD20 antibody, which has demonstrated significant activity and adequate tolerability in chronic lymphocytic leukemia (CLL), including studies where the drug was administered as monotherapy or combined with chemotherapy.1-7 Based on the results of the pivotal phase III CLL11 trial,2,3 in which the combination of obinutuzumab and chlorambucil (G-Clb) was shown to be clinically superior (in terms of progression-free survival and treatment response) to rituximab plus chlorambucil in adult patients with previously untreated CLL and comorbidities, obinutuzumab was approved haematologica | 2018; 103(11)

Correspondence: veronique.leblond@aphp.fr

Received: December 21, 2017. Accepted: July 4, 2018. Pre-published: July 5, 2018. doi:10.3324/haematol.2017.186387 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1889 ©2018 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.

1889

V. Leblond et al.

(as G-Clb; in November 2013 in the US and May 2014 in Europe) for this indication.8,9 Following its approval in CLL, a large study (GREEN) was undertaken to further inform the risk-benefit profile of obinutuzumab in a broad population of patients that is representative of that encountered in everyday practice. GREEN (clinicaltrials.gov identifier: 01905943) is an ongoing phase IIIb safety study of obinutuzumab, as a single agent or in combination with chemotherapy, in fit [defined as a Cumulative Illness Rating Scale (CIRS) score of ≤6 and creatinine clearance (CrCl) ≥70 mL/minute (min)] and unfit (defined as a CIRS score of >6 and/or CrCl <70 mL/min) patients with previously untreated (first-line) or relapsed/refractory (R/R) CLL. This study is collecting safety data for the largest cohort of CLL patients treated with obinutuzumab to date. This paper reports data for the primary objective of the GREEN study (a primary analysis snapshot), which was to assess the overall safety and tolerability of obinutuzumabbased treatment. An exploratory objective was to investigate the effectiveness of different approaches (including a modified initial obinutuzumab dosage, slower infusion rate and additional steroid pre-medication) to reduce infusion-related reactions (IRRs), which were observed during obinutuzumab infusion in the CLL11 trial, particularly during the first administration.2

Methods Design GREEN is a non-randomized, non-comparative, open-label study. Patients received intravenous obinutuzumab 1000 mg, alone or with chemotherapy [investigator’s choice of fludarabinecyclophosphamide (FC), Clb or bendamustine (benda), based primarily on fitness; see Online Supplementary Appendix], on days 1 (split over 2 consecutive days), 8 and 15 of cycle 1, and on day 1 of cycles 2-6 (six 28-day cycles). Alternative administration approaches for the first obinutuzumab infusion were studied in three first-line cohorts to assess their effect on IRR mitigation (Online Supplementary Appendix). Patients received intravenous prednisolone (or equivalent) 1 hour (h) pre-dose on day 1/day 2 of cycle 1. Risk minimization measures, including prophylaxis and investigator training (Online Supplementary Appendix), were instigated for patients considered at risk of tumor lysis syndrome (TLS; defined initially as nodes ≥10 cm, or ≥5-<10 cm with lymphocytes ≥25×109/L; definition later expanded for G-benda-treated patients following 2 fatal TLS cases) (Online Supplementary Appendix). Neutropenia prophylaxis was also recommended (Online Supplementary Appendix). GREEN was conducted according to the Declaration of Helsinki, Good Clinical Practice guidelines, and local laws/regulations. Study documentation was approved by institutional review boards/ethics committees at each site. Patients gave written informed consent.

Patients Patients were aged 18 years or over with CLL requiring treatment [National Cancer Institute/International Workshop on Chronic Lymphocytic Leukemia (NCI/iwCLL) criteria10], an Eastern Cooperative Oncology Group (ECOG) performance status of 0-2 and adequate hematologic function (see Online Supplementary Appendix for eligibility criteria).

1890

Study procedures Adverse events (AEs) were graded by NCI Common Terminology Criteria for AEs version 4.0. Response was assessed by investigators according to NCI/iwCLL criteria10 at the final response assessment, scheduled 84 days after the last dose of study medication.

Statistical analysis The primary end point was safety/tolerability. Safety outcomes included AEs, grade ≥3 AEs (primary outcome of interest), serious AEs (SAEs), and AEs of special/particular interest (AESIs/AEPIs). Overall response rate (ORR) and complete response [(CR; including CR with incomplete marrow recovery (CRi)] at the final response assessment were among the secondary efficacy end points (Online Supplementary Appendix). Time-to-event end points are not presented due to insufficient follow up (median, 20.8-28.8 months, depending on treatment); post-treatment follow up is still ongoing for patients who have not discontinued the study. IRRs were defined as any AE occurring during/within 24 h of obinutuzumab infusion and considered related to obinutuzumab. IRR incidence in first-line patients was an exploratory end point. Safety was evaluated in patients treated with at least one dose of study medication. Response was assessed in the intent-to-treat (ITT) population comprising all enrolled patients. A sample size of 950 patients [630 first-line (approximately equal proportions of fit and unfit) and 320 R/R patients] was planned [based on adequate precision, by 95% Clopper-Pearson confidence intervals (CIs), to estimate incidence rate of grade ≥3 AEs if the observed rate was 1-25%], with no formal statistical hypothesis testing. As a non-randomized study, treatment comparability was not applicable. Data are presented using descriptive statistics. Incidence rates and two-sided 95% Clopper-Pearson CI were calculated for grade ≥3 AEs and ORRs. Additional AEs and response amendments were reported late by some sites. While not captured here, updates are reported in the Online Supplementary Appendix.

Results Patients Patients were enrolled between October 2013 and March 2016 at 169 centers in 31 countries in Africa, North and South America, Asia and Europe. This primary analysis took place after all treated patients had finished study treatment and undergone a final response assessment (data cut-off for primary snapshot analysis, December 29, 2016). The ITT population comprised 972 patients and the safety population included 971 patients [630 (64.8%) firstline, including 339 (34.9%) fit (CIRS ≤6 and CrCl ≥70 mL/min) and 291 (29.9%) unfit (CIRS >6 and/or CrCl <70 mL/min) patients; and 341 (35.1%) R/R patients]; one first-line patient was enrolled but not treated and therefore not included in the safety population. Seven patients from one site in Romania were excluded from the analyses due to non-compliance with Good Clinical Practice guidelines. At the data cut-off, 195 (20.1%; 80 first-line and 115 R/R) ITT patients had discontinued the study and 777 (79.9%; 551 first-line and 226 R/R) were still on study (all in follow up). Primary reasons for study discontinuation included death [n=105 (10.8%); 40 first-line and 65 R/R], withdrawal of consent (n=63, 6.5%), AE (n=10, 1.0%), haematologica | 2018; 103(11)

Safety of obinutuzumab with/without chemotherapy in CLL

loss to follow up (n=9, 0.9%), investigator decision (n=4, 0.4%) and other (n=4, 0.4%). In the ITT population, median age was 66.0 (range 33-90) years, 63.5% of patients were male, 59.6%/36.9%/3.5% had an ECOG performance status of

0/1/2, 79.2% had a CIRS score of ≤6 and 61.0% had CrCl ≥70 mL/min (Table 1). Binet stage distribution at screening was 25.3% stage A, 41.2% stage B, 32.9% stage C and 0.6% missing. Five hundred and thirty-three (54.8%) patients had a high tumor burden (with nodes ≥10 cm, or

Table 1. Demographics and baseline disease characteristics according to line of therapy and fitness of patients (intent-to-treat population).

Characteristic

First-line fit (n=339)

First-line unfit (n=292)

First-line all (n=631)

R/R (n=341)

Total (n=972)

Median age, (range) years Age ≥65 years, n (%) Age ≥75 years, n (%) Male, n (%) ECOG performance status, n(%) 0 1 2 Binet stage, n(%) A B C Missing B symptoms, n (%)* Bulky disease (≥5 cm), n(%)

59.0 (33-82) 102 (30.1) 15 (4.4) 233 (68.7)

72.0 (45-87) 223 (76.4) 114 (39.0) 166 (56.8)

65.0 (33-87) 325 (51.5) 129 (20.4) 399 (63.2)

68.0 (33-90) 213 (62.5) 98 (28.7) 218 (63.9)

66.0 (33-90) 538 (55.3) 277 (28.5) 617 (63.5)

237 (69.9) 98 (28.9) 4 (1.2)

154 (52.7) 127 (43.5) 11 (3.8)

391 (62.0) 225 (35.7) 15 (2.4)

188 (55.1) 134 (39.3) 19 (5.6)

579 (59.6) 359 (36.9) 34 (3.5)

97 (28.6) 161 (47.5) 81 (23.9) 0 120 (35.4)

72 (24.7) 112 (38.4) 108 (37.0) 0 87 (29.8)

169 (26.8) 273 (43.3) 189 (30.0) 0 207 (32.8)

77 (22.6) 127 (37.2) 131 (38.4) 6 (1.8) 114 (33.4)

246 (25.3) 400 (41.2) 320 (32.9) 6 (0.6) 321 (33.0)

240 (70.8) 259 (76.4)

149 (51.0) 230 (78.8)

389 (61.6) 489 (77.5)

210 (61.6) 214 (62.8)

599 (61.6) 703 (72.3)

339 (100) 0

172 (58.9) 120 (41.1)

511 (81.0) 120 (19.0)

259 (76.0) 82 (24.0)

770 (79.2) 202 (20.8)

0 339 (100)

230 (78.8) 62 (21.2)

230 (36.5) 401 (63.5)

149 (43.7) 192 (56.3)

379 (39.0) 593 (61.0)

87 (25.7) 189 (55.8) 63 (18.6)

72 (24.7) 146 (50.0) 74 (25.3)

159 (25.2) 335 (53.1) 137 (21.7)

84 (24.6) 160 (46.9) 97 (28.4)

243 (25.0) 495 (50.9) 234 (24.1)

140 (41.3) 134 (39.5) 65 (19.2)

104 (35.6) 115 (39.4) 73 (25.0)

244 (38.7) 249 (39.5) 138 (21.9)

93 (27.3) 153 (44.9) 95 (27.9)

337 (34.7) 402 (41.4) 233 (24.0)

14 (4.1) 55 (16.2) 45 (13.3) 106 (31.3) 18 (5.3) 58 (17.1) 43 (12.7)

20 (6.8) 33 (11.3) 48 (16.4) 97 (33.2) 7 (2.4) 43 (14.7) 44 (15.1)

34 (5.4) 88 (13.9) 93 (14.7) 203 (32.2) 25 (4.0) 101 (16.0) 87 (13.8)

46 (13.5) 67 (19.6) 33 (9.7) 79 (23.2) 16 (4.7) 33 (9.7) 67 (19.6)

80 (8.2) 155 (15.9) 126 (13.0) 282 (29.0) 41 (4.2) 134 (13.8) 154 (15.8)

101 (29.8) 181 (53.4) 57 (16.8)

90 (30.8) 146 (50.0) 56 (19.2)

191 (30.3) 327 (51.8) 113 (17.9)

64 (18.8) 188 (55.1) 89 (26.1)

255 (26.2) 515 (53.0) 202 (20.8)

Lymphocytes ≥25x10 /L, n(%) Total CIRS score, n(%) ≤6 >6 CrCl at screening, n(%) <70 mL/min ≥70 mL/min ZAP-70 expression, n(%) Negative Positive Missing CD38 expression, n(%) Negative Positive Missing Cytogenetics, n(%) 17p deletion 11q deletion 12q trisomy 13q deletion Other No abnormality Missing IgVH mutation status, n (%) Mutated Unmutated Missing 9

R/R: relapsed/refractory; ECOG: Eastern Cooperative Oncology Group; CIRS: Cumulative Illness Rating Scale; CrCl: creatinine clearance; n: number; min: minute. *Patients with at least one B symptom (unexplained fever >38°C, drenching night sweats >1 month or weight loss >10% of body mass in preceding 6 months).

haematologica | 2018; 103(11)

1891

V. Leblond et al. Table 2. Summary of adverse events according to line of therapy and fitness of patients (safety population).

N (%)

First-line fit (n=339)

First-line unfit (n=291)

First-line all (n=630)

AEs of any grade (in ≥10% of patients in the safety population by preferred term) Any 328 (96.8) 288 (99.0) 616 (97.8) Neutropenia 216 (63.7) 153 (52.6) 369 (58.6) Pyrexia 112 (33.0) 98 (33.7) 210 (33.3) Thrombocytopenia 99 (29.2) 89 (30.6) 188 (29.8) Nausea 99 (29.2) 77 (26.5) 176 (27.9) Anemia 68 (20.1) 79 (27.1) 147 (23.3) Chills 47 (13.9) 53 (18.2) 100 (15.9) Diarrhea 47 (13.9) 44 (15.1) 91 (14.4) Vomiting 57 (16.8) 36 (12.4) 93 (14.8) Fatigue 31 (9.1) 41 (14.1) 72 (11.4) Pneumonia 28 (8.3) 33 (11.3) 61 (9.7) Constipation 35 (10.3) 38 (13.1) 73 (11.6) Cough 27 (8.0) 32 (11.0) 59 (9.4) Leukopenia 43 (12.7) 20 (6.9) 63 (10.0) Hypotension 26 (7.7) 36 (12.4) 62 (9.8) Dyspnea 28 (8.3) 25 (8.6) 53 (8.4) Grade ≥3 AEs (in ≥5% of patients in the safety population by preferred term) Any 266 (78.5) 233 (80.1) 499 (79.2) Neutropenia 177 (52.2) 129 (44.3) 306 (48.6) Thrombocytopenia 46 (13.6) 46 (15.8) 92 (14.6) Anemia 23 (6.8) 30 (10.3) 53 (8.4) Pneumonia 17 (5.0) 24 (8.2) 41 (6.5) Febrile neutropenia 29 (8.6) 18 (6.2) 47 (7.5) Leukopenia 26 (7.7) 11 (3.8) 37 (5.9) TLS 14 (4.1) 32 (11.0) 46 (7.3) Lymphopenia 20 (5.9) 9 (3.1) 29 (4.6) Grade 5 (fatal) AEs (in ≥2 patients in the safety population by preferred term) Any 13 (3.8) 18 (6.2) 31 (4.9) Pneumonia 1 (0.3) 4 (1.4) 5 (0.8) Sepsis 1 (0.3) 2 (0.7) 3 (0.5) Death 1 (0.3) 1 (0.3) 2 (0.3) Richter syndrome 0 0 0 AML 1 (0.3) 0 1 (0.2) Febrile neutropenia 0 2 (0.7) 2 (0.3) Septic shock 0 0 0 TLS 1 (0.3) 1 (0.3) 2 (0.3) SAEs (in ≥2% of patients in the safety population by preferred term) Any 148 (43.7) 171 (58.8) 319 (50.6) Neutropenia 39 (11.5) 23 (7.9) 62 (9.8) Pneumonia 20 (5.9) 24 (8.2) 44 (7.0) Febrile neutropenia 25 (7.4) 18 (6.2) 43 (6.8) Pyrexia 15 (4.4) 13 (4.5) 28 (4.4) TLS 4 (1.2) 21 (7.2) 25 (4.0) Thrombocytopenia 8 (2.4) 11 (3.8) 19 (3.0) Grade ≥3 AESIs/AEPIs (basket terms)* Neutropenia 194 (57.2) 138 (47.4) 332 (52.7) Infections 48 (14.2) 57 (19.6) 105 (16.7) IRRs 60 (17.7) 65 (22.3) 125 (19.8)

R/R (n=341)

Total (n=971)

334 (97.9) 198 (58.1) 101 (29.6) 115 (33.7) 94 (27.6) 83 (24.3) 57 (16.7) 44 (12.9) 42 (12.3) 46 (13.5) 55 (16.1) 42 (12.3) 55 (16.1) 46 (13.5) 42 (12.3) 46 (13.5)

950 (97.8) 567 (58.4) 311 (32.0) 303 (31.2) 270 (27.8) 230 (23.7) 157 (16.2) 135 (13.9) 135 (13.9) 118 (12.2) 116 (11.9) 115 (11.8) 114 (11.7) 109 (11.2) 104 (10.7) 99 (10.2)

281 (82.4) 179 (52.5) 67 (19.6) 40 (11.7) 46 (13.5) 27 (7.9) 29 (8.5) 16 (4.7) 20 (5.9)

780 (80.3) 485 (49.9) 159 (16.4) 93 (9.6) 87 (9.0) 74 (7.6) 66 (6.8) 62 (6.4) 49 (5.0)

40 (11.7) 7 (2.1) 2 (0.6) 2 (0.6) 3 (0.9) 0 0 2 (0.6) 0

71 (7.3) 12 (1.2) 5 (0.5) 4 (0.4) 3 (0.3) 2 (0.2) 2 (0.2) 2 (0.2) 2 (0.2)

197 (57.8) 43 (12.6) 48 (14.1) 25 (7.3) 8 (2.3) 11 (3.2) 12 (3.5)

516 (53.1) 105 (10.8) 92 (9.5) 68 (7.0) 36 (3.7) 36 (3.7) 31 (3.2)

189 (55.4) 90 (26.4) 68 (19.9)

521 (53.7) 195 (20.1) 193 (19.9) continued on the next page

1892

haematologica | 2018; 103(11)

Safety of obinutuzumab with/without chemotherapy in CLL continued from previous page.

Thrombocytopenia TLS Second malignancies Second malignancies† Hemorrhagic events HBV reactivation Cardiac events PML

47 (13.9) 14 (4.1) 12 (3.5) 12 (3.5) 2 (0.6) 0 9 (2.7) 0

48 (16.5) 32 (11.0) 23 (7.9) 20 (6.9) 2 (0.7) 1 (0.3) 14 (4.8) 0

95 (15.1) 46 (7.3) 35 (5.6) 32 (5.1) 4 (0.6) 1 (0.2) 23 (3.7) 0

68 (19.9) 16 (4.7) 26 (7.6) 24 (7.0) 5 (1.5) 0 9 (2.6) 1 (0.3)

163 (16.8) 62 (6.4) 61 (6.3) 56 (5.8) 9 (0.9) 1 (0.1) 32 (3.3) 1 (0.1)

R/R: relapsed/refractory; AE: adverse event; TLS: tumor lysis syndrome; AML: acute myeloid leukemia; SAE: serious adverse event; AESI: adverse event of special interest; AEPI: adverse event of particular interest; IRR: infusion-related reaction; HBV: hepatitis B virus; PML: progressive multifocal leukoencephalopathy; MedDRA: Medical Dictionary for Regulatory Activities; n: number; h: hour. *Neutropenia and thrombocytopenia selection was via their MedDRA basket dataset subgroups; infection selection was via the MedDRA system order class ‘Infections and Infestations’; IRRs were defined as any AE occurring during or within 24 h of obinutuzumab infusion and considered related to obinutuzumab; TLS and PML were defined by their preferred terms; second malignancy selection was via the MedDRA system organ class ‘Neoplasms Benign, Malignant, and Unspecified’ starting six months after the first study drug intake; hemorrhagic event selection was via the MedDRA basket dataset subgroup; HBV reactivation was defined as any AE with the preferred term containing ‘hepatitis B’ or ‘hepatitis acute’ that was additionally assessed as HBV reactivation via medical review; and cardiac event selection was via the MedDRA system order class ‘Cardiac Disorders’. †Second malignancy selection [standardized MedDRA query (SMQ)], including malignant and unspecified tumors (wide) starting six months after the first study drug intake.

≥5 cm but <10 cm with lymphocytes ≥25x109/L) and were classified as being at increased risk of TLS. Other criteria also used to determine TLS risk are specified in the Online Supplementary Appendix. Median number of prior lines of therapy received by R/R patients was 1.0 (range 1.0-3.0) (Online Supplementary Table S1).

tuzumab occurred in 14.6% of patients (first-line, 14.4%; R/R, 15.0%), with 5.4% discontinuing obinutuzumab due to IRRs, 3.9% due to neutropenia and 1.8% due to thrombocytopenia. Treatment-emergent AEs by treatment are shown in Table 3.

Severe and serious AEs Treatment exposure Treatment received was [G-mono; n=126 (12.9%); firstline n=62, R/R n=64), G-FC (n=193 (19.9%); first-line n=153, R/R n=40), G-Clb (n=114 (11.7%); first-line n=68, R/R n=46) and G-benda (n=538 (55.3%); first-line n=347, R/R n=191]. Seven hundred and eighty-nine (81.2%) patients completed all six cycles of protocol-specified treatment and 182 (18.7%) discontinued treatment prematurely. For all chemotherapy regimens, most patients received all six treatment cycles, i.e. 79.0% for benda, 84.5% for fludarabine, 85.0% for cyclophosphamide, and 76.3% for Clb. The main reasons for not completing study treatment were tolerability [including AEs; n=146 (15.0%)] and withdrawal of consent [n=15 (1.5%)]. Patients received a median of 9 (range 1-13) administrations of obinutuzumab, with 94.5% of patients receiving ≥90% of the planned dose. Median exposure time to obinutuzumab was 20.4 (range 0.1-30.1) weeks.

Safety Median observation time was 24.5 (range 0.3-37.8) months. In the safety analysis, the most frequent treatment-emergent AEs (any grade, by preferred term), occurring in ≥20% patients, were neutropenia (58.4%), pyrexia (32.0%), thrombocytopenia (31.2%), nausea (27.8%), and anemia (23.7%), with no notable differences between the first-line and R/R, or fit and unfit subgroups (Table 2). Overall, 23.4% of patients (n=227) had at least one prolonged cytopenia (any grade, occurring during the treatment period and still present >24 days after end of treatment) and 2.4% (n=23) had at least one late-onset cytopenia (any grade, occurring during the post-treatment follow-up period, ≥24 days after end of treatment). AEs were considered related to obinutuzumab in 85.8% of patients, most commonly neutropenia (40.2%), thrombocytopenia (22.6%), nausea (16.8%), pyrexia (23.2%) and anemia (11.1%). AEs leading to discontinuation of obinuhaematologica | 2018; 103(11)

Grade ≥3 AEs (the primary safety outcome of interest) occurred in 79.2% (95% CI: 75.8-82.3%) of first-line patients 78.5% (95% CI: 73.7-82.7%) in fit and 80.1% (95% CI:75.0-84.5%) in unfit patients and 82.4% (95% CI:77.9-86.3%) of R/R patients (Table 2). Among the 80.3% of patients overall who experienced grade ≥3 AEs (Table 2), the most frequent events were neutropenia (49.9%), thrombocytopenia (16.4%), anemia (9.6%) and pneumonia (9.0%). The most common SAEs were neutropenia (10.8%), pneumonia (9.5%) and febrile neutropenia (7.0%); the overall rate of SAEs in the safety population was 53.1% (Table 2). Grade ≥3 AEs and SAEs generally occurred at a similar frequency in first-line and R/R patients, and in first-line fit and unfit patients (Table 2), although the overall rate of SAEs was higher in firstline unfit (58.8%) than first-line fit (43.7%) patients.

Deaths A total of 112 (11.5%) patients died during the study (12 within 28 days of their last dose of study treatment and 100 during the post-treatment follow-up period), primarily due to AEs [n=71 (7.3%)]. AEs leading to death were numerically more common in R/R patients [n=40 (11.7%)] than in first-line patients [n=31 (4.9%)]; fit n=13, unfit n=18). Pneumonia [n=12 (1.2%)]) and sepsis [n=5 (0.5%)]) were the most common AEs leading to death (Table 2). By treatment received, the lowest rate of death due to AEs was observed in the G-FC group [4.7% (9/193)] vs. 7.9% (9/114) in the G-Clb group, 7.8% (42/538) in the G-benda group and 8.7% (11/126) in the G-mono group)] (Table 3). Two patients died due to TLS (both in the first-line G-benda subgroup). Disease progression was listed as the primary cause of death in 43 (4.4%) patients.

Adverse events of special or particular interest AESIs/AEPIs (any grade, as defined in the footnote to Table 2 and Table 3) reported in the overall safety popula1893

V. Leblond et al. Table 3. Summary of adverse events according to treatment (safety population).

N (%)

G-mono (n=126)

G-FC (n=193)

AEs of any grade (in ≥10% of patients in the safety population by preferred term) Any 123 (97.6) 191 (99.0) Neutropenia 50 (39.7) 143 (74.1) Pyrexia 29 (23.0) 69 (35.8) Thrombocytopenia 28 (22.2) 69 (35.8) Nausea 22 (17.5) 76 (39.4) Anemia 20 (15.9) 52 (26.9) Chills 17 (13.5) 30 (15.5) Diarrhea 6 (4.8) 39 (20.2) Vomiting 8 (6.3) 44 (22.8) Fatigue 7 (5.6) 19 (9.8) Pneumonia 14 (11.1) 17 (8.8) Constipation 6 (4.8) 24 (12.4) Cough 15 (11.9) 28 (14.5) Leukopenia 5 (4.0) 25 (13.0) Hypotension 14 (11.1) 17 (8.8) Dyspnea 10 (7.9) 24 (12.4) Grade ≥3 AEs (in ≥5% of patients in the safety population by preferred term) Any 95 (75.4) 169 (87.6) Neutropenia 42 (33.3) 129 (66.8) Thrombocytopenia 15 (11.9) 38 (19.7) Pneumonia 11 (8.7) 12 (6.2) Febrile neutropenia 8 (6.3) 21 (10.9) Anemia 7 (5.6) 21 (10.9) Hypotension 6 (4.8) 3 (1.6) Leukopenia 3 (2.4) 20 (10.4) TLS 2 (1.6) 5 (2.6) Lymphopenia 0 10 (5.2) Grade 5 (fatal) AEs (in ≥2 patients in the safety population by preferred term) Any 11 (8.7) 9 (4.7) Pneumonia 1 (0.8) 2 (1.0) Sepsis 1 (0.8) 2 (1.0) Death 1 (0.8) 1 (0.5) Richter syndrome 1 (0.8) 1 (0.5) AML 0 1 (0.5) Febrile neutropenia 1 (0.8) 0 Septic shock 1 (0.8) 0 TLS 0 0 SAEs (in ≥5% of patients in the safety population by preferred term) Any 67 (53.2) 87 (45.1) Pneumonia 11 (8.7) 13 (6.7) Neutropenia 9 (7.1) 28 (14.5) Febrile neutropenia 6 (4.8) 20 (10.4) Pyrexia 3 (2.4) 11 (5.7) TLS 1 (0.8) 3 (1.6) Grade ≥3 AESIs/AEPIs (basket terms)* Neutropenia 49 (38.9) 136 (70.5) IRRs 31 (24.6) 35 (18.1) Infections 27 (21.4) 30 (15.5) Thrombocytopenia 16 (12.7) 39 (20.2)

G-Clb (n=114)

G-benda (n=538)

113 (99.1) 60 (52.6) 28 (24.6) 36 (31.6) 26 (22.8) 26 (22.8) 17 (14.9) 10 (8.8) 14 (12.3) 22 (19.3) 19 (16.7) 14 (12.3) 15 (13.2) 8 (7.0) 17 (14.9) 11 (9.6)

523 (97.2) 314 (58.4) 185 (34.4) 170 (31.6) 146 (27.1) 132 (24.5) 93 (17.3) 80 (14.9) 69 (12.8) 70 (13.0) 66 (12.3) 71 (13.2) 56 (10.4) 71 (13.2) 56 (10.4) 54 (10.0)

87 (76.3) 51 (44.7) 22 (19.3) 15 (13.2) 2 (1.8) 10 (8.8) 7 (6.1) 3 (2.6) 5 (4.4) 0

429 (79.7) 263 (48.9) 84 (15.6) 49 (9.1) 43 (8.0) 55 (10.2) 7 (1.3) 40 (7.4) 50 (9.3) 39 (7.2)

9 (7.9) 3 (2.6) 0 1 (0.9) 0 1 (0.9) 0 0 0

42 (7.8) 6 (1.1) 2 (0.4) 1 (0.2) 1 (0.2) 0 1 (0.2) 1 (0.2) 2 (0.4)

57 (50.0) 16 (14.0) 6 (5.3) 2 (1.8) 1 (0.9) 3 (2.6)

305 (56.7) 52 (9.7) 62 (11.5) 40 (7.4) 21 (3.9) 29 (5.4)

53 (46.5) 24 (21.1) 23 (20.2) 24 (21.1)

283 (52.6) 103 (19.1) 115 (21.4) 84 (15.6) continued on the next page

1894

haematologica | 2018; 103(11)

Safety of obinutuzumab with/without chemotherapy in CLL continued from previous page.

Second malignancies Second malignancies† TLS Hemorrhagic events HBV reactivation Cardiac events PML

6 (4.8) 5 (4.0) 2 (1.6) 1 (0.8) 0 3 (2.4) 1 (0.8)

9 (4.7) 8 (4.1) 5 (2.6) 0 0 6 (3.1) 0

7 (6.1) 6 (5.3) 5 (4.4) 0 0 3 (2.6) 0

39 (7.2) 37 (6.9) 50 (9.3) 8 (1.5) 1 (0.2)‡ 20 (3.7) 0

G: obinutuzumab; mono: monotherapy; FC: fludarabine-cyclophosphamide; Clb: chlorambucil; benda: bendamustine; AE: adverse event; TLS: tumor lysis syndrome; AML: acute myeloid leukemia; SAE: serious adverse event; AESI: adverse event of special interest; AEPI: adverse event of particular interest; IRR: infusion-related reaction; HBV: hepatitis B virus; PML: progressive multifocal leukoencephalopathy; MedDRA: Medical Dictionary for Regulatory Activities; n: number; h: hour. *Neutropenia and thrombocytopenia selection was via their MedDRA basket dataset subgroups; infection selection was via the MedDRA system order class ‘Infections and Infestations’; IRRs were defined as any AE occurring during or within 24 h of obinutuzumab infusion and considered related to obinutuzumab; TLS and PML were defined by their preferred terms; second malignancy selection was via the MedDRA system organ class ‘Neoplasms Benign, Malignant, and Unspecified’ starting six months after the first study drug intake; hemorrhagic event selection was via the MedDRA basket dataset subgroup; HBV reactivation was defined as any AE with the preferred term containing ‘hepatitis B’ or ‘hepatitis acute’ that was additionally assessed as HBV reactivation via medical review; and cardiac event selection was via the MedDRA system order class ‘Cardiac Disorders’. †Second malignancy selection: standardized MedDRA query (SMQ), including malignant and unspecified tumors (wide) starting 6 months after the first study drug intake, ‡Three patients had HBV reactivation in total, 2 of which were grade <3.

tion were IRRs (65.4%; grade ≥3, 19.9%), neutropenia (61.7%; grade ≥3, 53.7%), infections (53.7%; grade ≥3, 20.1%), thrombocytopenia (32.3%; grade ≥3, 16.8%), cardiac events (11.2%; grade ≥3, 3.3%), second malignancies [(8.4% by MedDRA system organ class, including grade ≥3, 6.3% (listed in full in Online Supplementary Table S2); 7.7% by standardized MedDRA query, including grade ≥3, 5.8%)], hemorrhagic events (7.1%; grade ≥3, 0.9%), TLS (6.4%; all grade ≥3 by definition), hepatitis B virus reactivation (0.3%; grade ≥3, 0.1%) and progressive multifocal leukoencephalopathy (0.1%; grade ≥3, 0.1%). The most commonly reported infections by preferred term were pneumonia (11.9%), bronchitis (6.9%), upper respiratory tract infection (6.9%), nasopharyngitis (5.4%) and urinary tract infection (5.4%). Grade ≥3 AESIs/AEPIs were reported at a similar frequency in first-line and R/R patients, and in first-line fit and unfit patients (Table 2); however, grade ≥3 TLS as an AESI was more common in first-line (7.3%) than in R/R patients (4.7%), and in firstline unfit (11.0%) than in first-line fit (4.1%) patients, and grade ≥3 infections as AESIs were more common in R/R patients (26.4%) than in first-line patients (16.7%).

TLS and IRRs In the 62 patients with TLS events, 32 cases had laboratory TLS and 30 had clinical TLS. Except for the 2 fatal cases described below, all TLS events resolved, none with sequelae, and there was no recurrence in any patient. In 41 of the 62 patients with TLS, no change in drug dosing was needed; treatment was interrupted or delayed in 17 patients and discontinued in 4 patients. A higher rate of TLS was observed in patients who received G-benda (9.3% overall; 6.6% in first-line fit, 14.4% in first-line unfit, and 6.8% in R/R patients) compared with the other regimens. Of the 2 patients with fatal TLS, one had bulky disease (age 79 years) and the other lymphadenopathy (age 45 years); the older patient also had chronic renal failure at baseline. Both patients died in hospital after cardiovascular events (sudden cardiac arrest and acute cardiac failure, respectively). The frequency of IRRs was similar among the three dosing cohorts regardless of the IRR mitigation strategy used, although grade ≥3 IRRs, serious IRRs and IRRs leading to obinutuzumab discontinuation were more common in Cohort 3, along with TLS (as a preferred term) (Table 4). haematologica | 2018; 103(11)

Treatment response Among first-line patients, the ORR in the ITT population at the final response assessment was 89.5% with GFC, 82.4% with G-Clb, 81.8% with G-benda, and 63.5% with G-mono (Table 5); respective CR/CRi rates were 46.4%, 16.2%, 35.7% and 20.6%. In R/R patients, the ORR was 82.5% with G-FC, 54.3% with G-Clb, 72.8% with G-benda and 42.2% with G-mono; CR/CRi rates were 22.5%, 6.5%, 19.9% and 4.7%, respectively. Response rates for the 80 patients with 17p deletion are also shown in Table 5.

Discussion GREEN evaluated the safety and tolerability of obinutuzumab, alone or combined with chemotherapy, in a broad CLL patient population, including first-line (fit and unfit) and R/R patients. The chemotherapy partner options that were available to GREEN investigators mirror those used in standard practice with anti-CD20 antibodies in CLL.11,12 Notably, GREEN represents the first large-scale report of safety data for obinutuzumab in CLL patients following its approval. While GREEN was subject to certain limitations, the study provides valuable information on the overall safety profile of obinutuzumab, alone or combined with chemotherapy, in a broad CLL population. Importantly, obinutuzumab-based treatment demonstrated a generally manageable toxicity profile. Because of the non-comparative/non-randomized study design and potential investigator bias on patient allocation to cohorts/treatment, specific treatment regimens could not be compared directly. Furthermore, as treatment allocation was based on investigator’s choice, some subgroups were under-represented [e.g. first-line unfit and fit patients treated with G-mono (n=32 and n=31, respectively), and R/R patients treated with G-FC (n=40) or G-Clb (n=46)], making it difficult to draw conclusions from these small patient cohorts. However, this under-representation was not surprising given that most investigators followed current guideline recommendations for treatment.11,12 All patients were also analyzed as treated; for example, the G-mono group included patients who discontinued treatment after their first obinutuzumab administration due to AEs before receiving their planned chemotherapy regimen (n=23), as 1895

V. Leblond et al. Table 4. Summary of infusion-related reactions according to the approach used to prevent or mitigate these events in first-line patients.

N (%)*

Cohort 1 (n=237)†

Any IRR 141 (59.5) Grade ≥3 IRRs 45 (19.0) Serious IRRs 26 (11) IRRs leading to any obinutuzumab discontinuation 4 (1.7) IRRs (reported by ≥2% patients in any cohort, any grade by preferred term) Chills 49 (20.7) Pyrexia 45 (19.0) Nausea 24 (10.1) Vomiting 16 (6.8) TLS 14 (5.9) Hypertension 12 (5.1) Hypotension 12 (5.1) Thrombocytopenia 8 (3.4) Dyspnea 7 (3.0) Hypersensitivity 7 (3.0) Chest discomfort 6 (2.5) Flushing 5 (2.1) Hot flush 5 (2.1) Anemia 5 (2.1) Oxygen saturation decreased 5 (2.1) Hyperhidrosis 5 (2.1) Headache 3 (1.3) Tremor 3 (1.3) Rash 2 (0.8) AST increased 2 (0.8) ALT increased 2 (0.8) Feeling hot 2 (0.8) Dizziness 1 (0.4)

Cohort 2 (n=228)‡

Cohort 3 (n=151)§

153 (67.1) 41 (18.0) 24 (10.5) 5 (2.2)

96 (63.6) 37 (24.5) 23 (15.2) 12 (7.9)

26 (11.4) 42 (18.4) 34 (14.9) 14 (6.1) 6 (2.6) 10 (4.4) 31 (13.6) 16 (7.0) 17 (7.5) 1 (0.4) 10 (4.4) 8 (3.5) 1 (0.4) 3 (1.3) 3 (1.3) 7 (3.1) 12 (5.3) 5 (2.2) 6 (2.6) 8 (3.5) 6 (2.6) 5 (2.2) 5 (2.2)

13 (8.6) 28 (18.5) 7 (4.6) 4 (2.6) 15 (9.9) 5 (3.3) 4 (2.6) 9 (6.0) 9 (6.0) 1 (0.7) 3 (2.0) 2 (1.3) 1 (0.7) 5 (3.3) N/R 1 (0.7) 5 (3.3) N/R 4 (2.6) 6 (4.0) 6 (4.0) 1 (0.7) N/R

IRR: infusion-related reaction; TLS: tumor lysis syndrome; N/R: not reported; AST: aspartate aminotransferase; ALT: alanine aminotransferase; G: obinutuzumab; mono: monotherapy; FC: fludarabine-cyclophosphamide; Clb: chlorambucil; benda: bendamustine; n: number; h: hour. Patients within each cohort could have received any of the permitted immunochemotherapy regimens: G-mono (Cohort 1, n=15; Cohort 2, n=25; Cohort 3, n=20), G-FC (Cohort 1, n=61; Cohort 2, n=57; Cohort 3, n=33), G-Clb (Cohort 1, n=12; Cohort 2, n=40; Cohort 3, n=16) or G-benda (Cohort 1, n=149; Cohort 2, n=106; Cohort 3, n=82). *Sixteen previously untreated patients from the safety population were excluded from the cohort analysis as they did not receive treatment as planned (n=15) or were not assigned to any cohort (n=1). †Cycle 1 day 1 dose of obinutuzumab over two days: 25 mg (12.5 mg/h) + 975 mg (50-400 mg/h). ‡Cycle 1 day 1 dose of obinutuzumab over two 2 days: 100 mg (25 mg/h) + 900 mg (50-400 mg/h) with oral dexamethasone 20 mg (or equivalent) 12 h pre-dose. §Cycle 1 day 1 dose of obinutuzumab over two days: doses and infusion rates as in Cohort 1 with pre-medication scheme as in Cohort 2.

well as patients who were only ever planned to receive single-agent obinutuzumab. As such, patients in this subgroup had a higher rate of AEs and discontinuations due to AEs than would be expected for patients treated with G-mono, based on previous single-agent studies.4,5 The safety data from GREEN were generally in line with the safety profile for obinutuzumab-based immunochemotherapy previously observed in patients treated for CLL1-7 and non-Hodgkin lymphoma.13-18 Common AEs included IRRs (typically mild or moderate events observed almost exclusively during the first obinutuzumab infusion), infections and hematologic toxicities. The higher frequency of grade ≥3 AEs (including infections) and SAEs compared with the pivotal CLL11 study (which enrolled first-line patients with co-existing conditions2,3) likely reflects the broader patient population, and inclusion of more heavily pre-treated R/R patients. As expected, R/R patients in GREEN experienced more AEs and more deaths due to AEs or disease progression com1896

pared with first-line patients. While the rate of deaths due to AEs, particularly infections/sepsis, in first-line patients was higher than expected, it is reflective of that seen in clinical practice (rather than in classical clinical trials), where a broad range of patients and difficult-to-treat infections are also encountered. Predictably, there was a higher rate of SAEs and fatal AEs in first-line unfit versus fit patients; an observation that may have been due to the general health of the patients rather than the treatment regimen(s) received. The high reported rates of AESIs/AEPIs, including neutropenia, thrombocytopenia, IRRs, infections and TLS, may have resulted from the inclusion of R/R and unfit patients who may be more vulnerable to the adverse effects of treatment, although this did not appear to markedly affect grade ≥3 AESI/AEPI rates. Furthermore, despite the additional risk minimization measures, the rate of IRRs, including TLS, remained relatively high, particularly in Cohort 3. During the initial stages of recruithaematologica | 2018; 103(11)

Safety of obinutuzumab with/without chemotherapy in CLL Table 5. Summary of response at the final response assessment according to treatment (intent-to-treat population).

All patients (N=972) First-line, % (95% CI) ORR CR (including CRi) R/R, % (95% CI) ORR CR (including CRi) Patients with 17p deletion* (n=80) First-line, n/N (%) ORR CR (including CRi) R/R, n/N (%) ORR CR (including CRi)

G-mono (n=127)

G-FC (n=193)

G-Clb (n=114)

G-benda (n=538)

63.5 (50.4-75.3) 20.6

89.5 (83.6-93.9) 46.4

82.4 (71.2-90.5) 16.2

81.8 (77.4-85.8) 35.7

42.2 (29.9-55.2) 4.7

82.5 (67.2-92.7) 22.5

54.3 (39.0-69.1) 6.5

72.8 (65.9-79.0) 19.9

1/2 (50.0) 0/2

1/5 (20.0) 1/5 (20.0)

5/7 (71.4) 1/7 (14.3)

12/20 (60.0) 5/20 (25.0)

2/6 (33.3) 0/6

5/6 (83.3) 0/6

5/7 (71.4) 0/7

12/27 (44.4) 3/27 (11.1)

n/N: number; G: obinutuzumab; mono: monotherapy; FC: fludarabine-cyclophosphamide; Clb: chlorambucil; benda: bendamustine; CI: confidence interval; ORR: overall response rate; CR: complete response; CRi: complete response with incomplete marrow recovery; R/R: relapsed refractory. *17p deletion status was determined by fluorescence in situ hybridization.

ment into Cohort 3, up-dated and expanded definitions of patients at risk of TLS and additional TLS risk mitigation measures (for patients treated with G-benda) were implemented. Nonetheless, the TLS rate in GREEN, including 2 fatal cases, highlights the need for careful risk assessment, prophylaxis and monitoring, particularly in unfit patients [with a CIRS score of >6 and/or reduced renal function (CrCl <70 mL/min)] treated with the Gbenda regimen, in whom a high incidence of TLS (14.4%) was observed. It should be noted that, because of the nonrandomized study design, it is impossible to conclude whether the increase in TLS seen in G-benda-treated patients in this trial was due to the chemotherapy partner or to differences in patients’ characteristics compared with the other treatment cohorts. The current labeling states that any patients with a high tumor burden, high circulating lymphocyte count (>25x109/L) or renal impairment, who are considered at greater risk for TLS, should receive appropriate TLS prophylaxis with anti-hyperuricemics (e.g. allopurinol or rasburicase) and hydration prior to obinutuzumab infusion.8,9 Pre-treatment should then be followed by intensive monitoring of clinical signs/symptoms and laboratory parameters during the first few days of treatment. For IRRs, it is recommended that patients are pre-medicated with an intravenous corticosteroid, acetaminophen and antihistamine, and then monitored closely during obinutuzumab infusion.8,9 Antimicrobial prophylaxis is advised for patients with prolonged severe neutropenia to prevent infection; granulocyte colonystimulating factors should be considered in case of grade ≥3 neutropenia. All four obinutuzumab-based immunochemotherapy regimens appeared manageable in both first-line (fit or unfit) and R/R patients with CLL. G-FC, which was the most intensive regimen, was associated with a high rate of grade ≥3 neutropenia, but this did not translate into an elevated incidence of infection; an observation that may be explained by the underlying fitness of patients who received G-FC. Fitness may also explain the low rate of haematologica | 2018; 103(11)

deaths due to AEs in G-FC-treated patients. Investigation of strategies to prevent or mitigate IRRs during the first infusion of obinutuzumab was inconclusive, with rates comparable to those reported for G-Clb in CLL11 (grade ≥3, 21%).2 Despite efforts to minimize IRRs using approaches whereby the dosage of obinutuzumab was modified, the infusion rate slowed and/or additional corticosteroid was given as pre-medication, no one strategy appeared better than another. A recent nursing review of all IRR data from GREEN and CLL11 concluded that IRRs observed with obinutuzumab during the first infusion are generally manageable in CLL patients through treatment interruptions, but management could be improved considerably with extra vigilance during the first infusion.19 Analysis of anti-leukemic activity revealed high response rates across all settings and regimens, thus supporting findings from previous studies, including CLL11 and phase I/II trials, which have evaluated the G-Clb, G-FC, G-benda and G-mono regimens.1-7 Response rates tended to be higher in first-line versus R/R patients, and in patients who received combination versus single-agent obinutuzumab therapy. The response rates also compared favorably with those reported for rituximab-containing immunochemotherapy (rituximab plus Clb, benda or FC) in CLL.2,3,20-24 While longer-term data are required to confirm the efficacy of obinutuzumab-based therapy in GREEN, they do suggest that these regimens are clinically active and associated with a generally manageable toxicity profile. In conclusion, in the largest obinutuzumab patient cohort analyzed to date, the GREEN primary safety data were in line with the safety and tolerability profile previously observed in patients receiving obinutuzumab-based treatment for CLL. Toxicities were generally manageable and response rates were encouraging in this broad population of CLL patients, including previously untreated, fit and unfit patients and those with R/R disease. Based on these data, future trials are warranted. 1897

V. Leblond et al.

Acknowledgments The authors would like to thank the patients and their families, and the GREEN study investigators and site staff. They would also like to acknowledge Miriam Amor for her valuable contribution to the statistical analyses. Third-party medical

References 1. Goede V, Fischer K, Busch R, et al. Chemoimmunotherapy with GA101 plus chlorambucil in patients with chronic lymphocytic leukemia and comorbidity: results of the CLL11 (BO21004) safety run-in. Leukemia. 2013;27(5):1172-1174. 2. Goede V, Fischer K, Busch R, et al. Obinutuzumab plus chlorambucil in patients with CLL and coexisting conditions. N Engl J Med. 2014;370(12):11011110. 3. Goede V, Fischer K, Engelke A, et al. Obinutuzumab as frontline treatment of chronic lymphocytic leukemia: updated results of the CLL11 study. Leukemia. 2015;29(7):1602-1604. 4. Cartron G, de Guibert S, Dilhuydy MS, et al. Obinutuzumab (GA101) in relapsed/refractory chronic lymphocytic leukemia: final data from the phase 1/2 GAUGUIN study. Blood. 2014; 124(14):2196-2202. 5. Byrd JC, Flynn JM, Kipps TJ, et al. Randomized phase 2 study of obinutuzumab monotherapy in symptomatic, previously untreated chronic lymphocytic leukemia. Blood. 2016;127(1):79-86. 6. Brown JR, O’Brien S, Kingsley CD, et al. Obinutuzumab plus fludarabine/ cyclophosphamide or bendamustine in the initial therapy of CLL patients: the phase 1b GALTON trial. Blood. 2015; 125(18):2779-2785. 7. Danilov A, Yimer H, Boxer M, et al. Results of a phase II multicenter study of obinutuzumab plus bendamustine in pts with previously untreated chronic lymphocytic leukemia (CLL). Haematologica. 2017;102:71 [abstract P249]. ® 8. Gazyva (obinutuzumab) injection, for intravenous infusion. Full prescribing information. Revised 2/2016. 9. Gazyvaro® (obinutuzumab) 1,000 mg concentrate for solution for infusion. Summary of product characteristics. Last updated 28/07/2016. 10. Hallek M, Cheson BD, Catovsky D, et al;

1898

11.

12.

13.

14.

15.

16.

17.

writing assistance, under the direction of the authors, was provided by Cheryl Wright, PhD of Gardiner-Caldwell Communications (Macclesfield, UK), and was funded by F. Hoffmann-La Roche Ltd. GREEN was sponsored by F. Hoffmann-La Roche Ltd.

International Workshop on Chronic Lymphocytic Leukemia. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood. 2008;111(12):54465456. Eichhorst B, Robak T, Montserrat E, et al; ESMO Guidelines Committee: Chronic lymphocytic leukaemia. ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2015;26 Suppl 5:v78-v84. Wierda WG, Zelenetz AD, Gordon LI, et al. NCCN guidelines insights: chronic lymphocytic leukemia/small lymphocytic leukemia, version 1.2017. J Natl Compr Canc Netw. 2017;15(3):293-311. Sehn LH, Chua N, Mayer J, et al. Obinutuzumab plus bendamustine versus bendamustine monotherapy in patients with rituximab-refractory indolent nonHodgkin lymphoma (GADOLIN): a randomised, controlled, open-label, multicentre, phase 3 trial. Lancet Oncol. 2016;17(8):1081-1093. Sehn LH, Goy A, Offner FC, et al. Randomized phase II trial comparing obinutuzumab (GA101) with rituximab in patients with relapsed CD20+ indolent Bcell non-Hodgkin lymphoma: final analysis of the GAUSS study. J Clin Oncol. 2015;33(30):3467-3474. Salles GA, Morschhauser F, Solal-Céligny P, et al. Obinutuzumab (GA101) in patients with relapsed/refractory indolent nonHodgkin lymphoma: results from the phase II GAUGUIN study. J Clin Oncol. 2013; 31(23):2920-2926. Morschhauser FA, Cartron G, Thieblemont C, et al. Obinutuzumab (GA101) monotherapy in relapsed/refractory diffuse large B-cell lymphoma or mantle-cell lymphoma: results from the phase II GAUGUIN study. J Clin Oncol. 2013; 31(23):2912-2919. Radford J, Davies A, Cartron G, et al.

18.

19.

20.

21.

22.

23.

24.

Obinutuzumab (GA101) plus CHOP or FC in relapsed/refractory follicular lymphoma: results of the GAUDI study (BO21000). Blood. 2013;122(7):1137-1143. Marcus R, Davies A, Ando K, et al. Obinutuzumab for the first-line treatment of follicular lymphoma. N Engl J Med. 2017;377(14):1331-1334. Dawson K, Moran M, Guindon K, Wan H. Managing infusion-related reactions for patients with chronic lymphocytic leukemia receiving obinutuzumab. Clin J Oncol Nurs. 2016;20(2):E41-E48. Eichhorst B, Fink AM, Bahlo J, et al. International group of investigators; German CLL Study Group (GCLLSG). First-line chemoimmunotherapy with bendamustine and rituximab versus fludarabine, cyclophosphamide, and rituximab in patients with advanced chronic lymphocytic leukaemia (CLL10): an international, open-label, randomised, phase 3, non-inferiority trial. Lancet Oncol. 2016;17(7):928-942. Fischer K, Bahlo J, Fink AM, et al. Longterm remissions after FCR chemoimmunotherapy in previously untreated patients with CLL: updated results of the CLL8 trial. Blood. 2016;127(2):208-215. Skarbnik AP, Faderl S. The role of combined fludarabine, cyclophosphamide and rituximab chemoimmunotherapy in chronic lymphocytic leukemia: current evidence and controversies. Ther Adv Hematol. 2017;8(3):99-105. Hallek M, Fischer K, Fingerle-Rowson G, et al. Addition of rituximab to fludarabine and cyclophosphamide in patients with chronic lymphocytic leukaemia: a randomized, open-label, phase 3 trial. Lancet. 2010; 376(9747):1164-1174. Nunes AA, da Silva AS, Souza KM, Koury Cde N, de Mello LM. Rituximab, fludarabine, and cyclophosphamide versus fludarabine and cyclophosphamide for treatment of chronic lymphocytic leukemia: a systematic review with metaanalysis. Crit Rev Oncol Hematol. 2015; 94(3):261-269.

haematologica | 2018; 103(11)

ARTICLE

Non-Hodgkin Lymphoma

Inferior survival in high-grade B-cell lymphoma with MYC and BCL2 and/or BCL6 rearrangements is not associated with MYC/IG gene rearrangements

Ellen D. McPhail,1 Matthew J. Maurer,2 William R. Macon,1 Andrew L. Feldman,1 Paul J. Kurtin,1 Rhett P. Ketterling,1 Rakhee Vaidya,3 James R. Cerhan,4 Stephen M. Ansell,5 Luis F. Porrata,5 Grzegorz S. Nowakowski,5 Thomas E. Witzig1,5 and Thomas M. Habermann5

Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, MN; 2 Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, MN; 3 Department of Hematology and Oncology, Wake Forest Baptist Health, Winston-Salem, NC; 4Department of Health Sciences Research, Mayo Clinic, Rochester, MN and 5 Division of Hematology, Mayo Clinic, Rochester, MN, USA 1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(11):1899-1907

ABSTRACT

igh-grade B-cell lymphomas with MYC and BCL2 and/or BCL6 rearrangements (double-/triple-hit lymphoma) have an aggressive clinical course. We investigated the prognostic value of transformation from low-grade lymphoma, cytological features (high grade versus large cell), MYC rearrangement partners (immunoglobulin versus nonimmunoglobulin gene), and treatment. We evaluated 100 adults with double-/triple-hit lymphoma, reviewing cytological features; cell of origin; and rearrangements of MYC, BCL2, and BCL6 using MYC, BCL2, and BCL6 break-apart and IGH/MYC, IGL/MYC, IGK/MYC, and IGH/BCL2 dual-fusion interphase fluorescence in situ hybridization probes. Outcome analysis was restricted to patients with lymphoma, de novo or at transformation, who received anthracycline-based chemotherapy. Among them, 60% had high-grade cytological features; 91% had a germinal center B-cell phenotype, and 60% had a MYC/IG rearrangement. Germinal center B-cell phenotype was associated with BCL2 rearrangements (P<0.001). Mean (95% confidence interval) 5-year overall survival was 49% (37%-64%). Transformation from previously treated and untreated low-grade lymphoma was associated with inferior overall survival (hazard ratio, 2.99; P=0.008). Patients with high-grade cytological features showed a non-significant tendency to inferior outcome (hazard ratio, 2.32; P=0.09). No association was observed between MYC rearrangement partner and overall survival (hazard ratio, 1.00; P=0.99). Compared with patients receiving rituximab, cyclophosphamide, doxorubicin, and vincristine (R-CHOP) and dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin, and rituximab (EPOCHR), patients receiving rituximab, cyclophosphamide, vincristine, doxorubicin, methotrexate/ifosfamide, etoposide, and cytarabine (RCODOX-M/IVAC) had a non-significant tendency to better overall survival (hazard ratio, 0.37; P=0.10). In conclusion, high-grade B-cell lymphomas with MYC and BCL2 and/or BCL6 rearrangements had heterogeneous outcomes and MYC/IG rearrangements were not associated with inferior overall survival. Introduction The diagnosis of ‘high-grade B-cell lymphoma with MYC and BCL2 and/or BCL6 rearrangements’ (double-/triple-hit lymphoma, DH/THL) was established in the 2016 revision of the World Health Organization (WHO) classification of lymphoid neoplasms.1 This category includes all large B-cell lymphomas with haematologica | 2018; 103(11)

Presented as an abstract and poster at the American Society of Hematology 58th Annual Meeting and Exposition, San Diego, California, USA, December 3-6, 2016.

Correspondence: mcphail.ellen@mayo.edu

Received: February 7, 2018. Accepted: June 12, 2018. Pre-published: June 14, 2018. doi:10.3324/haematol.2018.190157 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1899 ©2018 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.

1899

E.D. McPhail et al.

rearrangements of MYC and BCL2 or BCL6, or both, besides those that meet criteria for follicular or lymphoblastic lymphoma. It encompasses some cases previously called ‘B-cell lymphoma, unclassifiable, with features intermediate between diffuse large B-cell lymphoma (DLBCL) and Burkitt lymphoma (BL) (BCLU)’, in accordance with the 2008 WHO classification.2 These cases were considered clinically aggressive but were difficult to diagnose because of vague diagnostic criteria. By morphological evaluation, DH/THL may show a cytological spectrum. The spectrum can range from (i) monotonous medium-sized cells with round nuclei, finely dispersed chromatin, and starry sky appearance, resembling BL, to (ii) intermediate-sized cells with slight pleomorphism and slightly irregular nuclear contours, resembling BCLU,2 to (iii) large lymphoid cells with round to irregular nuclear contours, variably sized nucleoli, and varying amounts of cytoplasm, resembling DLBCL. Experts have recommended that during evaluation, details of the morphological appearance be added in a comment to the record because of the potential prognostic importance of morphological characteristics.1 Yet, few studies have specifically addressed morphological appearance as a prognostic indicator. MYC is a powerful transcriptional factor that helps to drive the cell from G0/1 phase to S phase and promotes cell proliferation and growth, DNA replication, and protein biosynthesis. It was identified initially as the molecular target of the 8q24 rearrangement characteristic of BL but was subsequently identified in various B-cell lymphomas, including 5% to 15% of DLBCL and 30% to 60% of highgrade B-cell lymphomas.3-5 The MYC rearrangement partner in BL is almost invariably an immunoglobulin (IG) gene (IGH, IGK, or IGL), whereas in DLBCL and highgrade B-cell lymphoma it is a non-IG gene in about 40% of cases.5-9 Common non-IG MYC partners include lymphomagenesis-related genes, such as BCL6, BCL11A, PAX5, and IKAROS.10 In IG/MYC rearrangements, MYC is juxtaposed to an IG enhancer, usually resulting in pronounced amplification of MYC protein expression, whereas MYC expression and MYC transcript levels are often less robust in the clinical setting of non-IG/MYC rearrangements.5,9 The prognostic significance of the MYC partner gene is controversial. Some groups found that a non-IG MYC partner was a survival advantage,5,6,8 while other groups observed no significant difference between IG and non-IG partner cases.9,11 DH/THL was established as a new diagnostic category in part because of its aggressive clinical behavior. However, most DH/THL cases have a BCL2 rearrangement (i.e., MYC/BCL2 or MYC/BCL2/BCL6). The clinical behavior of those lacking the BCL2 rearrangement (i.e., MYC/BCL6 cases) is not well understood because a limited number are available for analysis. At present, the prognostic significance of MYC/BCL6 in this context is controversial, with different groups identifying superior outcome,5,12 no difference in outcome,13 or inferior outcome.9,14,15 However, fewer than 100 cases have been described in the literature. The reported median overall survival (OS) for DH/THL in different series range from 4.5 to 34 months.6,10,13,16-26 Patients were treated primarily with rituximab, cyclophosphamide, doxorubicin, and vincristine (RCHOP)27; dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin, and rituximab 1900

(EPOCH-R)28; rituximab, cyclophosphamide, vincristine, doxorubicin, and dexamethasone (R-hyper CVAD); methotrexate; cytarabine25; and rituximab, cyclophosphamide, vincristine, doxorubicin, and methotrexate/rituximab, ifosfamide, etoposide, and cytarabine (RCODOX-M/IVAC)29 with or without an autologous stem cell transplant in first complete remission. The median progression-free survival and OS were not improved in some series,13,23,25 but were improved in one series.30 Autologous stem cell transplantation in the relapse setting is associated with poor outcomes.31-33 Few studies have addressed the prognostic significance of transformation of low-grade lymphoma to DH/THL.34 In the light of the controversy surrounding these issues, the present study investigated the prognostic significance of several of these parameters, including morphological evaluation, IG/MYC versus non-IG rearrangement partner, presence or absence of a BCL2 rearrangement, transformation from low-grade lymphoma, and therapeutic regimens in diagnostic cases of DH/THL at the Mayo Clinic in Rochester, Minnesota. To our knowledge, this study represents the largest single-institution study of these characteristics among contemporary DH/THL patients.

Methods The Mayo Clinic Institutional Review Board approved this study and all patients provided consent. Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines were followed. Cases were identified through review of Mayo Clinic patients in the Mayo Clinic Lymphoma Database (1998-2015) and the Lymphoma Specialized Program of Research Excellence Molecular Epidemiology Resource (20022015). Five cases were identified from the Molecular Epidemiology Resource through fluorescence in situ hybridization (FISH) performed for other studies.35 The therapeutic regimens were R-CHOP, dose-adjusted EPOCH-R, R-CODOXM/IVAC, R-hyper-CVAD, methotrexate, cytarabine, and non– anthracycline-based treatment. Additional information regarding case identification and case criteria is detailed in the Online Supplementary Appendix. All cases were diagnosed initially by a Mayo Clinic hematopathologist as either DLBCL or BCLU according to the 2008 WHO criteria;2 all were reclassified as DH/THL according to the 2016 WHO criteria.1 Morphological re-review to assess high-grade versus large-cell histological characteristics was performed by four Mayo Clinic hematopathologists (ALF, PJK, WRM, and EDM). Definitions of high-grade and large-cell cytological features are detailed in the Online Supplementary Appendix. For outcome analysis, only cases with a consensus rereview diagnosis were used. Cell of origin was determined according to the Hans classifier.36 Immunohistochemical methods and criteria are detailed in the Online Supplementary Appendix. Interphase FISH was performed on either paraffin sections of tissue specimens or smears of bone marrow aspirate specimens according to previously described methods37,38 using break-apart probes for MYC and BCL6; dual-fusion FISH probes for IGH/MYC, IGL/MYC, and IGK/MYC; and either a BCL2 break-apart probe or an IGH/BCL2 dual-fusion FISH probe. Further details are provided in the Online Supplementary Appendix. Because of tissue limitations, not all probe sets were performed in all cases. Specifically, in some cases, the MYC rearrangement partner could not be haematologica | 2018; 103(11)

Prognostic markers of double-/triple-hit lymphomas

identified, and in other cases, the BCL6 rearrangement status was unknown. Cases with concurrent MYC and BCL2 rearrangements, concurrent MYC and BCL6 rearrangements, and concurrent MYC, BCL2, and BCL6 rearrangements are referred to as MYC/BCL2, MYC/BCL6, and MYC/BCL2/BCL6, respectively. Clinical outcome analysis was limited to patients with DH/THL characteristics identified at initial diagnosis or at transformation from previously diagnosed low-grade lymphoma and who received an anthracycline-based chemotherapy regimen at DH/THL diagnosis. Previous therapies for the 11 patients with prior low-grade lymphoma are as follows: none (n=2); radiation therapy only (n=2); bendamustine and rituximab (B-R) only (n=2); single-agent rituximab (n=1); B-R followed by ibritumomab consolidation (n=1); R-CHOP with maintenance rituximab (n=1); seven prior therapies, including rituximab, cyclophosphamide, vincristine, and prednisone (R-CVP), CHOP, ifosfamide, carboplatin, and etoposide (ICE), and B-R (n=1); and four prior therapies, including R-CVP and B-R (n=1). Patients with recurrent B-cell lymphoma with high-grade or large-cell histological features were excluded from this analysis because the MYC/BCL2/BCL6 rearrangement status of the initial biopsy was unknown. OS was defined as the time from DH/THL diagnosis to death of any cause or to last follow-up. Event-free survival (EFS) was defined as the time from diagnosis to progression, relapse, retreatment after initial chemotherapy, or death of any cause. EFS12 was defined as event-free status at 12 months after diagnosis. EFS12 was used as an endpoint because of limited followup and because progression or relapse occurred primarily in the first 12 months after diagnosis.39 OS and EFS were evaluated with Kaplan-Meier curves and Cox proportional hazards models. All analyses were performed using statistical software (SAS version 9.4; SAS Institute Inc.) and R 3.3.1 (R Project for Statistical Computing). Statistical significance was defined by P values less than 0.05.

Results Morphological and tumor characteristics The study involved 100 patients (male to female ratio, 64:36) with the median (range) age at DH/THL diagnosis of 61 (29-87) years. Sixty-seven patients had DH/THL identified at initial diagnosis; 22, at the time of transformation of previously diagnosed low-grade lymphoma; and 11, from a recurrent specimen in previously diagnosed lymphoma with large-cell or high-grade morphological features for which original diagnostic material was not studied (Table 1). Slides were available for consensus review in 72 cases, and in 65 cases a consensus diagnosis was reached. Of these, 39 patients (60%) had high-grade morphological features (Figure 1A) and 26 (40%) had large-cell morphological features (Figure 1B). According to the Hans classifier, the phenotype of 91 cases (91%) was germinal center B-cell (GCB); six (6%), non-GCB; and three (3%), unknown. In immunohistochemistry analysis, 37 of 43 cases (86%) met criteria for MYC positivity, 76 of 84 (90%) expressed BCL2, and 30 of 37 (81%) expressed both (called double expressers).

MYC rearrangement partner (IG versus non-IG gene) and BCL2/BCL6 rearrangement status The MYC rearrangement partner was an IG gene in 52 cases (39 IGH, 6 IGK, and 7 IGL), a non-IG gene in 35 haematologica | 2018; 103(11)

cases, and unknown in 13 cases (Figure 2A). Fifty-nine cases were MYC/BCL2; 13, MYC/BCL6; 20, MYC/BCL2/BCL6; and 8, MYC/BCL2 (unknown BCL6) (Figure 2B). The MYC rearrangement partner (IG versus non-IG gene) was not associated with morphological features (P=0.96), cell of origin (P=0.18), or BCL2 rearrangement status (P=0.27) (Table 2). However, MYC expression by immunohistochemistry was significantly more common in IG/MYC than in non-IG/MYC cases (95% versus 74%; P=0.049). GCB phenotype was present in 100% of the 85 cases with BCL2 rearrangements [MYC/BCL2, MYC/BCL2/BCL6, and MYC/BCL2 (unknown BCL6); collectively referred to as DHBCL2/THL], and in six (50%) of the 12 MYC/BCL6 cases

Table 1. Clinicopathological features, genetic characteristics, and therapeutic regimens of all patients compared with those of the outcome analysis cohort.

Variable

All patientsa (n=100)

Age, years <60 47 (47) â&#x2030;Ľ60 53 (53) Sex Male 64 (64) Female 36 (36) Timing of diagnosis De novo 67 (67) Transformation 22 (22) Recurrence 11 (11) Morphologic review, central (n=65) Large cell 26 (40) High grade 39 (60) COO per Hans classifier (n=97) GCB 91 (94) Non-GCB 6 (6) MYC immunohistochemistry (n=43) Positive 37 (86) Negative 6 (14) % MYC+, range 20-90 % MYC+, median 70 MYC FISH: rearrangement partner(n=87) IG gene 52 (60) Non-IG gene 35 (40) BCL2 and BCL6 FISH: rearrangement status MYC/BCL2 59 (59) MYC/BCL6 13 (13) MYC/BCL2/BCL6 20 (20) MYC/BCL2 (BCL6 unknown) 8 (8) Therapy R-CHOP 36 (36) R-EPOCH 17 (17) R-CODOX-M/IVAC 17 (17) R-hyper-CVAD 6 (6) Platinum-based salvage 10 (10) Other/none/unavailable 14 (14)

Outcome analysis cohorta (n=70) 34 (49) 36 (51) 41 (59) 29 (41) 59 (84) 11 (16) 0 (0) (n=43) 19 (43) 24 (56) (n=69) 64 (93) 5 (7) (n=29) 26 (90) 3 (10) 0-90 50 (n=61) 35 (57) 26 (43) 39 (56) 11 (16) 14 (20) 6 (9) 32 (46) 17 (24) 15 (21) 6 (9) 0 (0) 0 (0)

COO: cell of origin; FISH: fluorescence in situ hybridization; GCB: germinal center B cell; IG: immunoglobulin; hyper-CVAD: cyclophosphamide, vincristine: doxorubicin, and dexamethasone; R-CHOP: rituximab, cyclophosphamide, doxorubicin, and vincristine; R-CODOX-M/IVAC: rituximab, cyclophosphamide, vincristine, doxorubicin, and high-dose methotrexate alternating with rituximab, ifosfamide, etoposide, and high-dose cytarabine; R-EPOCH: rituximab, etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin. aValues are presented as number (%) of patients unless specified otherwise.

1901

E.D. McPhail et al.

(P<0.001). All 13 MYC/BCL6 cases were de novo diagnoses, whereas 38% of DH-BCL2/THL cases were either transformed (n=22) or relapsed (n=11) (P=0.03). No association was observed between morphological features and BCL2 rearrangement status (P=0.16).

Therapy, clinical characteristics, and outcome The therapeutic regimens were R-CHOP (n=36); doseadjusted EPOCH-R (n=17); R-CODOX-M/IVAC (n=17); R-hyper-CVAD, methotrexate, and cytarabine (n=6); platinum-based salvage (n=10); and non-anthracycline-based treatment (n=14). Survival was analyzed for 70 patients receiving anthracycline-based curative-intent therapy at diagnosis of DH/THL (Table 1). The median (interquartile range) age of these patients was 61 (29-82) years; 41 (59%) were males. Fifty-nine patients had de novo DH/THL and 11 had transformation of a previously diagnosed low-grade lymphoma. For treatment, 32 patients

received R-CHOP; 17, dose-adjusted EPOCH-R; 15, RCODOX-M/IVAC; and 6, R-hyper-CVAD with methotrexate and cytarabine. Twelve patients (17%) received consolidation with autologous stem cell transplantation. At a median (range) follow-up of 21 (1-87) months, EFS12 was 46% (Figure 3A) and median OS was 22 months (95% CI: 13 months - unreached) (Figure 3B). At a 5-year follow-up, the OS rate was 49% (95% CI: 37%-64%). Patients with DH/THL at transformation of previously diagnosed low-grade lymphoma (n=11) had poor outcomes [median OS, 10.8 months; hazard ratio (HR), 2.99; 95% CI: 1.33-6.71; EFS12, 10%] - inferior to the outcomes of patients with de novo DH/THL (median OS, 22 months; P=0.008; EFS12, 52%) (Figure 4A). However, seven of the 11 patients had received prior immunochemotherapy for their low-grade lymphoma. Patients with high-grade morphological features on a con-

Figure 1. Comparison of highgrade and large-cell cytologic characteristics. (A) Doublehit/triple-hit lymphoma (DH/THL) with a high-grade morphological pattern (hematoxylin-eosin, original magnification Ă&#x2014;400). (B) DH/THL with large-cell morphological characteristics (hematoxylineosin, original magnification Ă&#x2014;400).

Figure 2. MYC rearrangement partners and BCL2 and BCL6 rearrangement status. (A) MYC rearrangement partners in double-hit/triple-hit lymphoma (DH/THL) by interphase dualfusion fluorescence in situ hybridization (FISH). Cases with an IG/MYC rearrangement (IGH/MYC, IGK/MYC, or IGL/MYC) are contiguous. (B) BCL2 and BCL6 rearrangement status in DH/THL by interphase FISH. All cases had a MYC rearrangement. BCL2-rearranged cases are contiguous whereas MYC/BCL6 cases are not contiguous.

1902

haematologica | 2018; 103(11)

Prognostic markers of double-/triple-hit lymphomas

sensus pathology review (n=24) showed a nonsignificant tendency to inferior outcome (median OS, 13.5 months; HR, 2.32; 95% CI: 0.88-6.12; EFS12, 37%) compared with that of patients with large-cell morphological features (n=19) (median OS, unreached; P=0.09; EFS12, 57%) (Figure 4B). Patients with MYC/BCL6 (BCL2 rearrangement-negative) tumors showed a nonsignificant tendency to better outcomes (n=11) (median OS, unreached; HR, 0.42; 95% CI: 0.13-1.40; EFS12, 64%) than those of patients with BCL2 rearrangement (DH-BCL2/THL; n=59; median OS, 21.7 months; P=0.16; EFS12, 42%) (Figure 4C). There was no association between MYC partner (IG gene versus non-IG gene) and OS (for IG, HR, 1.00; 95% CI: 0.48-2.09; P=0.99) (Figure 4D). Compared with patients receiving all other therapies, patients treated with R-CODOX-M/IVAC had superior EFS12 (72% versus 39%, P=0.04) and showed a nonsignificant tendency to improved OS (HR, 0.37; 95% CI, 0.111.23; P=0.10) (Figure 5). However, patient selection bias may confound this outcome because the patients were significantly younger (P<0.001) than the patients who received other anthracycline-based regimens. Nevertheless, outcomes stayed consistent although not significant for EFS12 (HR, 0.27; 95% CI: 0.07-1.16; P=0.08) and for OS (HR, 0.40; 95% CI: 0.11-1.51; P=0.18), after adjustment for age in multivariable logistic and Cox models, respectively.

Discussion The diagnosis of ‘high-grade B-cell lymphoma, with MYC and BCL2 and/or BCL6 rearrangements’ (DH/THL) was established in the 2016 revision of the WHO classification of lymphoid neoplasms, in part to acknowledge the prognostic significance of the MYC and BCL2 or MYC and BCL6 rearrangements, or both, in B-cell lymphomas with large-cell or high-grade morphological features. However, our study shows that within this umbrel-

la category, several subgroups have distinctive features and varying clinical outcomes. Strengths of this singleinstitution study include thorough pathology review, comprehensive analyses of genetic features and cell of origin, and availability of key clinical, treatment, and outcome data. The main limitations of this retrospective study include missing MYC rearrangement data for some cases, potential for selection bias in treatment choice given to the patient, inclusion of cases over an 11-year time frame during which clinical management evolved rapidly, and small sample size for some analyses. Although there is likely some bias toward selection of cases with high-grade cytological features in the early specimens, this bias should not have been present in the 57 cases that were identified after our institution began performing interphase FISH to identify DH/THL in all Bcell lymphomas with either large-cell or high-grade histological features. Of the specimens with histological consensus re-review, 16/20 (80%) of the early cases but only 23/45 (51%) of the later cases had a morphological diagnosis of high-grade B-cell lymphoma, suggesting that the selection bias present in the early cases had been mitigated in the later group. Our data suggest several important observations. First, patients with DH/THL at transformation of previously diagnosed low-grade lymphoma had a dismal outcome with a median OS of 10.8 months and EFS12 of 10%. In light of the small number of cases and potential confounding effect of the prior immunochemotherapy for some patients, this observation should be interpreted with caution. However, it raises an interesting biological question that may warrant investigation in future studies. Few previous studies have addressed this, although a report of two cases of transformation of low-grade follicular lymphoma to DH/THL described an aggressive clinical course.34 In DH/THL, induction failure occurs early and is worse among patients aged 60 years or older. Nevertheless, the OS for the entire cohort at 5 years was 49%, suggestive of long-term cure in a subset of patients.

Table 2. Correlations between clinicopathological and genetic characteristics in de novo, transformed, and recurrent double-hit/triple-hit lymphoma.a

Characteristic All patients Timing of diagnosis De novo Transformed Recurrent Morphological evaluation High grade Large cell Cell of origin GCB Non-GCB MYC partner IG gene Non-IG gene BCL2 Rearranged Not rearranged

All patients

67 (67) 22 (22) 11 (11)

Morphological evaluation High Large grade cell

Cell of origin GCB

Non-GCB

MYC partner Non-IG IG gene gene

Double expresserb

BCL2 Not Rearranged Rearranged

Yes

39 (60)

26 (40)

91 (94)

6 (6)

52 (60)

35 (40)

87 (87)

13 (13)

30 (81)

7 (19)

20 (51) 10 (63) 9 (90)

19 (49) 6 (38) 1 (10)d

59 (91) 21 (100) 11 (100)

6 (9) 0 (0)c 0 (0)

34 (60) 12 (60) 6 (60)

23 (40) 8 (40) 4 (40)

54 (81) 22 (100) 11 (100)

13 (19) 0 (0)c 0 (0)

20 (87) 8 (73) 2 (67)

3 (13) 3 (27) 1 (33)

36 (95) 22 (88)

2 (5) 3 (12)

20 (53) 13 (52)

18 (48) 12 (48)

35 (90) 20 (77)

4 (10) 6 (23)

16 (80) 10 (83)

4 (20) 2 (17)

48 (62) 2 (33)

30 (38) 4 (67)

85 (93) 0 (0)

6 (7) 6 (100)e

28 (82) 2 (67)

6 (18) 1 (33)

46 (88) 28 (80)

4 (12) 7 (20)

18 (84) 12 (75)

3 (16) 4 (25)

28 (85) 2 (50)

5 (15) 2 (50)

39 (60) 26 (40) 91 (94) 6 (6) 52 (60) 35 (40)

GCB: germinal center B cell; IG: immunoglobulin. aValues are presented as number (%) of patients. bExpression of both MYC and BCL2. cP=0.01 to <0.05. dP=0.05 to 0.10. eP≤0.001.

haematologica | 2018; 103(11)

1903

E.D. McPhail et al.

Of note, progression or relapse occurred almost exclusively within 12 months of diagnosis, suggesting that EFS12 is worth further evaluation as an endpoint for studies of DH/THL.39 Treatment with R-CODOX-M/IVAC may result in superior outcomes for patients younger than 60 years, despite the preponderance of aggressive (high-grade) histological findings within this group. However, selection bias may also have an important role in the superior outcomes for this cohort. A multicenter retrospective analysis evaluated effect of induction regimen and stem cell transplantation on outcomes in DHL, showing that intensive induction therapy was not associated with improved OS.13 In a meta-analysis of 11 studies and 394 patients, OS was not different across the treatment approaches used in our series.40 Sun et al. reported a 2-year progres-

sion-free survival rate of 60% and a 2-year OS rate of 82% for 16 patients treated with R-CODOX-M/IVAC followed by hematopoietic stem cell transplantation.30 No randomized clinical trial for this histological subset is currently available. Second, DH/THL patients with high-grade histological characteristics showed a tendency, although not statistically significant, toward a more aggressive clinical course. Published literature has provided mixed results on this issue. Johnson et al., in their retrospective study of BCLU and DLBCL with concurrent MYC and BCL2 rearrangements identified through karyotypic analysis, showed that BCLU morphological appearance was associated with poorer outcome (P<0.001).6 Other investigators have obtained similar results.40-42 Blastoid morphological characteristics have also been associated with adverse clinical

Figure 3. Event-free survival and overall survival of patients receiving anthracycline-based therapy. (A) Event-free survival of the 70 patients receiving anthracyclinebased therapy at the time of diagnosis of double-hit/triple-hit lymphoma. (B) Overall survival of the 70 patients.

1904

Figure 4. Overall survival according to transformation of previously diagnosed low-grade lymphoma, morphological characteristics, BCL2 rearrangement status, and MYC partner (IG gene versus non-IG gene). (A) Patients with doublehit/triple-hit lymphoma (DH/THL) at transformation of previously diagnosed low-grade lymphoma had inferior overall survival (OS) compared with patients who had de novo DH/THL [hazard ratio (HR) 2.99, P=0.007)]. (B) DH/THL patients with high-grade morphological characteristics showed a tendency toward inferior OS compared with patients with large-cell morphological characteristics, but it was not statistically significant (HR 2.32, P=0.09). (C) DH/THL patients with BCL2 rearrangements [MYC/BCL2, MYC/BCL2/BCL6, and MYC/BCL2 (BCL6 unknown)], socalled DH-BCL2/TH, showed a nonsignificant tendency toward inferior OS compared with those who did not have BCL2 rearrangements (MYC/BCL6) (HR 2.44, P=0.16). (D) No association was observed between MYC partner (IG versus non-IG) and OS (P=0.99).

haematologica | 2018; 103(11)

Prognostic markers of double-/triple-hit lymphomas

outcome.43 In contrast, Petrich et al., in their retrospective study of 311 DH/THL patients, reported that histological appearance had no effect on OS; however, theirs was a multicenter study across 23 medical centers without central pathological review.13 Our cases were derived from a single institution, and outcome analysis was restricted to cases with a consensus re-review diagnosis among four expert hematopathologists, lending support to the validity of this finding. The clinical significance of high-grade histological characteristics underscores the importance of the revised WHO recommendation to note morphological appearance in all DH/THL cases.1 Furthermore, because DH/THL cases often show morphological characteristics of DLBCL (40% in this series), a complete interphase FISH analysis to exclude DH/THL is important for all cases with large-cell or high-grade morphological appearance. Also, low MYC expression by immunohistochemistry does not reliably predict absence of a MYC rearrangement. In the present series, 14% of evaluable cases, all with a MYC rearrangement by definition, had fewer than 40% MYC-positive cells by immunohistochemistry. However, these cases were too few to analyze outcome, and it is not known whether absence of MYC expression in DH/THL has a favorable impact on survival. Third, clinical and pathological differences were reported and depended on whether a BCL2 rearrangement was present. BCL2-rearranged cases [i.e., MYC/BCL2, MYC/BCL2/BCL6, or MYC/BCL2 (BCL6 unknown)] showed heterogeneous presentation as de novo, recurrent, or transformation events but were uniformly of GCB type. The latter finding has been previously described.4,5,7,9,26 Conversely, MYC/BCL6 cases uniformly developed de novo but were equally likely to be GCB or non-GCB type. Thus, all patients with non-GCB DH/THL in our cohort had MYC/BCL6. Although the association between non-GCB subtype and MYC/BCL6 has been reported previously,5,9 this is the largest cohort to date to investigate this issue. Clinically, our 12 MYC/BCL6 cases with outcome analysis showed a nonsignificant tendency to superior OS and EFS12, regardless of whether only the de novo cases were evaluated. Previous studies have yielded con-

flicting results. Two studies showed an association between MYC/BCL6 and poor prognosis.9,14 In the first study, only six FISH-confirmed cases were evaluated.14 In the second study, the MYC/BCL6 group, which had older patients with uniformly large-cell histological features, was compared with a MYC/BCL2 group that was younger and had various histological patterns, including low-grade lymphoma.9 Two other studies reported improved survival of patients with MYC/BCL6 compared to those with MYC/BCL2 DH/THL, but only seven and four cases, respectively, were studied.5,44 Another group showed better survival in de novo MYC/BCL6 with largecell morphological features compared with MYC/BCL2 DH/THL combined with MYC-rearranged single-hit DLBCL.12 Four additional studies â&#x20AC;&#x201C; three included patients with large-cell lymphomas and high-grade histological characteristics13,15,25 and one comprised only subjects with large-cell histological features26 - showed no difference in survival between MYC/BCL2 and MYC/BCL6 DH/THL. However, the former two studies also included cases of follicular lymphoma,13,15 and the third study included cases with extra intact signals, as well as rearrangements of MYC, BCL2, and BCL6.25 This lack of uniformity may have affected the outcome data. More studies are needed to resolve the issue of the prognostic significance of MYC/BCL6 in DH/THL. Some groups have suggested that it may not be necessary to perform interphase FISH studies to exclude DH/THL in large-cell or high-grade B-cell lymphomas of non-GCB phenotype. However, if this approach were followed, about one-half of MYC/BCL6 DH/THL would fail to be identified. The MYC/BCL6 cases in our study showed a tendency toward superior OS and EFS12. Nevertheless, in the light of conflicting outcome data in the literature, potential for poor clinical outcome, and suggested benefit from more aggressive therapy in this cohort, we advocate the performance of interphase FISH to exclude DH/THL in all large-cell and high-grade B-cell lymphomas, regardless of their cell of origin. Fourth, we found no association between MYC partner (IG gene versus non-IG gene) and clinical outcome. To our knowledge, our study represents the largest (n=87) and most comprehensively studied cohort to date in which

Figure 5. Overall survival according to treatment regimen. Patients treated with R-CODOX-M/RIVAC had superior overall survival, although the difference was not statistically significant (hazard ratio 0.37, P=0.10). DA-EPOCH-R: dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin, and rituximab; R-CHOP: rituximab, cyclophosphamide, doxorubicin, and vincristine; R-CODOX-M/IVAC: rituximab, cyclophosphamide, vincristine, doxorubicin, and high-dose methotrexate alternating with rituximab, ifosfamide, etoposide, and high-dose cytarabine; Rhyper-CVAD: rituximab, cyclophosphamide, vincristine, doxorubicin, and dexamethasone.

haematologica | 2018; 103(11)

1905

E.D. McPhail et al.

this issue has been addressed. Previous studies on this matter have yielded conflicting results. Johnson et al., in a retrospective study of 54 DLBCL and BCLU cases with MYC/BCL2 identified by abnormal karyotype, showed that a non-IG/MYC partner was associated with a favorable clinical outcome.6 However, they also found an association between non-IG/MYC partner and DLBCL histological appearance (P<0.001). In contrast, we found no such association in our cohort (P=0.96). This difference may, therefore, in part explain the divergent findings of clinical outcome between the two studies. In addition, case identification by abnormal karyotype may have selected for more aggressive disease. Pedersen et al. prospectively studied a cohort of 237 DLBCL and BCLU cases, including primary, transformed, and relapsed cases.8 They observed that IG/MYC rearrangements (n=9) were associated with a worse OS than were nonIG/MYC rearrangements (n=10). However, only 19 cases were studied, and this analysis included both MYC/BCL2 and MYC single-hit cases but lacked MYC/BCL6 cases. In addition, the patients received various treatment regimens, although most included rituximab plus intensive chemotherapy. An additional controversial element was the FISH protocol, because cases that were positive for a MYC rearrangement on the basis of a split MYC breakapart probe but were negative for IGH/MYC fusion were then studied with IGK and IGL break-apart probes rather than IGK/MYC and IGL/MYC dual-fusion FISH probes. Cases with concurrent MYC and IGK or IGL split breakapart probes were presumed to represent fusions, likely overestimating the number of IG/MYC cases. Copie-Bergman et al. performed interphase FISH using break-apart probes for MYC, BCL2, and BCL6 and dualfusion FISH probes for IGH/MYC/CEP8, IGK/MYC, and IGL/MYC as needed in 574 de novo DLBCL cases treated with rituximab–anthracycline-based chemotherapy.5 They found that the DH/THL cohort as a whole (n=32) had poorer outcome than the DLBCL cohort that lacked DH/THL rearrangements (P=0.046). However, on evaluating the IG/MYC and non-IG/MYC partner cases separately, Copie-Bergman et al. found that the 12 IG/MYC DH/THL cases had a poorer prognosis than the 19 DH/THL cases without non-IG/MYC rearrangement. They also observed poorer OS and progression-free survival for the 25 MYC/BCL2 cases than for the seven MYC/BCL6 cases, but they did not report the IG gene versus non-IG gene partners for these cases. Therefore, the IG versus non-IG gene results may have been confounded by the number of MYC/BCL2 versus MYC/BCL6 cases in each group. More than 90% of their DH/THL cases had large-cell histological characteristics (3 were reclassified

References 1. Swerdlow SH, Campo E, Pileri SA, et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375–2390. 2. Swerdlow SH, Campo E, Harris NL, et al, eds. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. 4th ed. Lyon, France: International Agency for Research on Cancer; 2008.

1906

as high grade on re-review); by comparison, 60% of our cases had high-grade histological characteristics, which may have contributed to the discrepant findings between the two groups. Other studies have shown no difference in survival between IG/MYC and non-IG/MYC groups. Aukema et al. studied 80 MYC rearrangement-positive B-cell lymphomas, with exclusion of pediatric cases as well as adult cases with a gene expression profile of molecular BL.9 Most cases had the histological appearance of DLBCL, although some had other histological patterns, such as BCLU and follicular lymphoma. Similar to our study, Aukema et al. found no difference in survival between the IG/MYC and non-IG/MYC groups, both overall and within the MYC/BCL2 (n=26) and MYC/BCL6 (n=14) subgroups. However, various treatment regimens were used, and only some patients received immunotherapy (rituximab). Li et al. also found no significant difference in survival in MYC/BCL2 DH/THL cases with IG (n=23) versus non-IG (n=5) MYC partners.11 More studies are needed to resolve the issue of the prognostic significance of MYC rearrangement partner in DH/THL. Our study, as well as those of Aukema et al.9 and Li et al.,11 suggests that it may not be necessary to identify whether the MYC gene rearrangement partner is an IG gene. Our current approach is to perform FISH on all aggressive B-cell lymphomas. However, in the light of these findings coupled with the association of DH/THL with such features as GCB phenotype, high-grade B-cell lymphoma morphological characteristics, and high MYC expression in immunohistochemistry, further risk-benefit analyses of alternative triage strategies are warranted. In conclusion, in this large, retrospective, single-institution study of DH/THL, although no differences in survival were seen between the IG/MYC and non-IG/MYC groups, transformation from previously treated and untreated low-grade lymphoma was associated with inferior OS, and there was a trend toward inferior OS in patients with high-grade morphological patterns and the presence of a BCL2 rearrangement. Acknowledgments We thank Kay M. Ristow for her assistance with database management. This work was supported in part by National Institutes of Health (NIH) Grant no. P50 CA97274 to the University of Iowa and Mayo Clinic Lymphoma Specialized Program of Research Excellence, National Cancer Institute (NCI) Grant n. R01 CA200703, NCI Grant n. U01 CA195568, and the Henry J. Predolin Foundation, Inc. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

3. Valera A, Lopez-Guillermo A, CardesaSalzmann T, et al; Grup per l’Estudi dels Limfomes de Catalunya i Balears. MYC protein expression and genetic alterations have prognostic impact in patients with diffuse large B-cell lymphoma treated with immunochemotherapy. Haematologica. 2013;98 (10):1554–1562. 4. Bellas C, Garcia D, Vicente Y, et al. Immunohistochemical and molecular characteristics with prognostic significance in

diffuse large B-cell lymphoma. PLoS One. 2014;9(6):e98169. 5. Copie-Bergman C, Cuilliere-Dartigues P, Baia M, et al. MYC-IG rearrangements are negative predictors of survival in DLBCL patients treated with immunochemotherapy: a GELA/LYSA study. Blood. 2015; 126(22):2466–2474. 6. Johnson NA, Savage KJ, Ludkovski O, et al. Lymphomas with concurrent BCL2 and MYC translocations: the critical factors asso-

haematologica | 2018; 103(11)

Prognostic markers of double-/triple-hit lymphomas

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

ciated with survival. Blood. 2009;114(11): 2273–2279. Pedersen MO, Gang AO, Poulsen TS, et al. Double-hit BCL2/MYC translocations in a consecutive cohort of patients with large Bcell lymphoma: a single centre’s experience. Eur J Haematol. 2012;89(1):63–71. Pedersen MO, Gang AO, Poulsen TS, et al. MYC translocation partner gene determines survival of patients with large B-cell lymphoma with MYC- or double-hit MYC/BCL2 translocations. Eur J Haematol. 2014;92(1):42–48. Aukema SM, Kreuz M, Kohler CW, et al; Molecular Mechanisms in Malignant Lymphomas Network Project. Biological characterization of adult MYC-translocation-positive mature B-cell lymphomas other than molecular Burkitt lymphoma. Haematologica. 2014;99(4):726-735. Bertrand P, Bastard C, Maingonnat C, et al. Mapping of MYC breakpoints in 8q24 rearrangements involving non-immunoglobulin partners in B-cell lymphomas. Leukemia. 2007;21(3):515–523. Li S, Saksena A, Desai P, et al. Prognostic impact of history of follicular lymphoma, induction regimen and stem cell transplant in patients with MYC/BCL2 double hit lymphoma. Oncotarget. 2016;7(25):38122– 38132. Clipson A, Barrans S, Zeng N, et al. The prognosis of MYC translocation positive diffuse large B-cell lymphoma depends on the second hit. J Pathol Clin Res. 2015;1(3):125–133. Petrich AM, Gandhi M, Jovanovic B, et al. Impact of induction regimen and stem cell transplantation on outcomes in double-hit lymphoma: a multicenter retrospective analysis. Blood. 2014;124(15):2354–2361. Pillai RK, Sathanoori M, Van Oss SB, Swerdlow SH. Double-hit B-cell lymphomas with BCL6 and MYC translocations are aggressive, frequently extranodal lymphomas distinct from BCL2 double-hit B-cell lymphomas. Am J Surg Pathol. 2013;37(3): 323–332. Li S, Desai P, Lin P, et al. MYC/BCL6 doublehit lymphoma (DHL): a tumour associated with an aggressive clinical course and poor prognosis. Histopathology. 2016;68(7): 1090–1098. Kanungo A, Medeiros LJ, Abruzzo LV, Lin P. Lymphoid neoplasms associated with concurrent t(14;18) and 8q24/c-MYC translocation generally have a poor prognosis. Mod Pathol. 2006;19(1):25–33. Le Gouill S, Talmant P, Touzeau C, et al. The clinical presentation and prognosis of diffuse large B-cell lymphoma with t(14;18) and 8q24/c-MYC rearrangement. Haematologica. 2007;92(10):1335–1342. Lin P, Medeiros LJ. High-grade B-cell lymphoma/leukemia associated with t(14;18) and 8q24/MYC rearrangement: a neoplasm of germinal center immunophenotype with poor prognosis. Haematologica. 2007;92 (10):1297–1301. Niitsu N, Okamoto M, Miura I, Hirano M. Clinical features and prognosis of de novo diffuse large B-cell lymphoma with t(14;18) and 8q24/c-MYC translocations. Leukemia. 2009;23(4):777–783. Tomita N, Tokunaka M, Nakamura N, et al.

haematologica | 2018; 103(11)

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

Clinicopathological features of lymphoma/leukemia patients carrying both BCL2 and MYC translocations. Haematologica. 2009;94(7):935–943. Snuderl M, Kolman OK, Chen YB, et al. Bcell lymphomas with concurrent IGH-BCL2 and MYC rearrangements are aggressive neoplasms with clinical and pathologic features distinct from Burkitt lymphoma and diffuse large B-cell lymphoma. Am J Surg Pathol. 2010;34(3):327–340. Akyurek N, Uner A, Benekli M, Barista I. Prognostic significance of MYC, BCL2, and BCL6 rearrangements in patients with diffuse large B-cell lymphoma treated with cyclophosphamide, doxorubicin, vincristine, and prednisone plus rituximab. Cancer. 2012;118(17):4173–4183. Li S, Lin P, Fayad LE, et al. B-cell lymphomas with MYC/8q24 rearrangements and IGH@BCL2/t(14;18)(q32;q21): an aggressive disease with heterogeneous histology, germinal center B-cell immunophenotype and poor outcome. Mod Pathol. 2012;25(1):145–156. Cohen JB, Geyer SM, Lozanski G, et al. Complete response to induction therapy in patients with MYC-positive and double-hit non-Hodgkin lymphoma is associated with prolonged progression-free survival. Cancer. 2014;120(11):1677–1685. Oki Y, Noorani M, Lin P, et al. Double hit lymphoma: the MD Anderson Cancer Center clinical experience. Br J Haematol. 2014;166(6):891–901. Ye Q, Xu-Monette ZY, Tzankov A, et al. Prognostic impact of concurrent MYC and BCL6 rearrangements and expression in de novo diffuse large B-cell lymphoma. Oncotarget. 2016;7(3):2401–2416. Coiffier B, Lepage E, Briere J, et al. CHOP chemotherapy plus rituximab compared with CHOP alone in elderly patients with diffuse large-B-cell lymphoma. N Engl J Med. 2002;346(4):235–242. Wilson WH, Dunleavy K, Pittaluga S, et al. Phase II study of dose-adjusted EPOCH and rituximab in untreated diffuse large B-cell lymphoma with analysis of germinal center and post-germinal center biomarkers. J Clin Oncol. 2008;26(16):2717–2724. Mead GM, Barrans SL, Qian W, et al; UK National Cancer Research Institute Lymphoma Clinical Studies Group; Australasian Leukaemia Lymphoma Group. A prospective clinicopathologic study of dose-modified CODOX-M/IVAC in patients with sporadic Burkitt lymphoma defined using cytogenetic and immunophenotypic criteria (MRC/NCRI LY10 trial). Blood. 2008;112(6):2248–2260. Sun H, Savage KJ, Karsan A, et al. Outcome of patients with non-Hodgkin lymphomas with concurrent MYC and BCL2 rearrangements treated with CODOX-M/IVAC with rituximab followed by hematopoietic stem cell transplantation. Clin Lymphoma Myeloma Leuk. 2015;15(6):341–348. Cuccuini W, Briere J, Mounier N, et al. MYC+ diffuse large B-cell lymphoma is not salvaged by classical R-ICE or R-DHAP followed by BEAM plus autologous stem cell transplantation. Blood. 2012;119(20):4619–4624. Van Den Neste E, Schmitz N, Mounier N, et al. Outcomes of diffuse large B-cell lym-

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

phoma patients relapsing after autologous stem cell transplantation: an analysis of patients included in the CORAL study. Bone Marrow Transplant. 2017;52(2):216–221. Herrera AF, Mei M, Low L, et al. Relapsed or refractory double-expressor and doublehit lymphomas have inferior progressionfree survival after autologous stem-cell transplantation. J Clin Oncol. 2017;35(1): 24–31. Xu X, Zhang L, Wang Y, et al. Double-hit and triple-hit lymphomas arising from follicular lymphoma following acquisition of MYC: report of two cases and literature review. Int J Clin Exp Pathol. 2013;6(4):788– 794. Novak AJ, Asmann YW, Maurer MJ, et al. Whole-exome analysis reveals novel somatic genomic alterations associated with outcome in immunochemotherapy-treated diffuse large B-cell lymphoma. Blood Cancer J. 2015;5:e346. Hans CP, Weisenburger DD, Greiner TC, et al. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004;103(1):275–282. Remstein ED, Dogan A, Einerson RR, et al. The incidence and anatomic site specificity of chromosomal translocations in primary extranodal marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue (MALT lymphoma) in North America. Am J Surg Pathol. 2006;30(12):1546–1553. Cataldo KA, Jalal SM, Law ME, et al. Detection of t(2;5) in anaplastic large cell lymphoma: comparison of immunohistochemical studies, FISH, and RT-PCR in paraffin-embedded tissue. Am J Surg Pathol. 1999;23(11):1386–1392. Maurer MJ, Ghesquieres H, Jais JP, et al. Event-free survival at 24 months is a robust end point for disease-related outcome in diffuse large B-cell lymphoma treated with immunochemotherapy. J Clin Oncol. 2014;32(10):1066–1073. Landsburg DJ, Falkiewicz MK, Maly J, et al. Outcomes of patients with double-hit lymphoma who achieve first complete remission. J Clin Oncol. 2017;35(20):2260–2267. McClure RF, Remstein ED, Macon WR, et al. Adult B-cell lymphomas with Burkitt-like morphology are phenotypically and genotypically heterogeneous with aggressive clinical behavior. Am J Surg Pathol. 2005;29(12):1652–1660. Landsburg DJ, Nasta SD, Svoboda J, Morrissette JJ, Schuster SJ. ‘Double-hit’ cytogenetic status may not be predicted by baseline clinicopathological characteristics and is highly associated with overall survival in B cell lymphoma patients. Br J Haematol. 2014;166(3):369–374. Moore EM, Aggarwal N, Surti U, Swerdlow SH. Further exploration of the complexities of large B-cell lymphomas with MYC abnormalities and the importance of a blastoid morphology. Am J Surg Pathol. 2017;41(9): 1155–1166. Tzankov A, Xu-Monette ZY, Gerhard M, et al. Rearrangements of MYC gene facilitate risk stratification in diffuse large B-cell lymphoma patients treated with rituximabCHOP. Mod Pathol. 2014;27(7):958–971.

1907

ARTICLE

Non-Hodgkin Lymphoma

Ferrata Storti Foundation

PD-L1+ tumor-associated macrophages and PD-1+ tumor-infiltrating lymphocytes predict survival in primary testicular lymphoma

Marjukka Pollari,1,2 Oscar Brück,3 Teijo Pellinen,4 Pauli Vähämurto,1,5 Marja-Liisa Karjalainen-Lindsberg,6 Susanna Mannisto,1,5 Olli Kallioniemi,4,7 Pirkko-Liisa Kellokumpu-Lehtinen,2,8 Satu Mustjoki,3,9 Suvi-Katri Leivonen1,5 and Sirpa Leppä1,5

Haematologica 2018 Volume 103(11):1908-1914

Research Program Unit, Faculty of Medicine, University of Helsinki, Finland; Department of Oncology, Tampere University Hospital, Finland; 3Hematology Research Unit Helsinki, Department of Clinical Chemistry and Hematology, University of Helsinki, Finland; 4Institute for Molecular Medicine Finland (FIMM), Helsinki, Finland; 5 Department of Oncology, Comprehensive Cancer Center, Helsinki University Hospital, Finland; 6Department of Pathology, Helsinki University Hospital, Finland; 7Science for Life Laboratory, Karolinska Institutet, Department of Oncology and Pathology, Solna, Sweden; 8Faculty of Medicine and Life Sciences, University of Tampere, Finland; 9 Department of Hematology, Comprehensive Cancer Center, Helsinki University Hospital, Finland 1 2

ABSTRACT

Correspondence: sirpa.leppa@helsinki.fi

Received: May 6, 2018. Accepted: July 16, 2018. Pre-published: July 19, 2018. doi:10.3324/haematol.2018.197194 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1908

rimary testicular lymphoma is a rare and aggressive lymphoid malignancy, most often representing diffuse large B-cell lymphoma histologically. Tumor-associated macrophages and tumor-infiltrating lymphocytes have been associated with survival in diffuse large Bcell lymphoma, but their prognostic impact in primary testicular lymphoma is unknown. Here, we aimed to identify macrophages, their immunophenotypes and association with lymphocytes, and translate the findings into survival of patients with primary testicular lymphoma. We collected clinical data and tumor tissue from 74 primary testicular lymphoma patients, and used multiplex immunohistochemistry and digital image analysis to examine macrophage markers (CD68, CD163, and c-Maf), T-cell markers (CD3, CD4, and CD8), B-cell marker (CD20), and three checkpoint molecules (PD-L1, PD-L2, and PD-1). We demonstrate that a large proportion of macrophages (median 41%, range 0.0899%) and lymphoma cells (median 34%, range 0.1-100%) express PDL1. The quantity of PD-L1+CD68+ macrophages correlates positively with the amount of PD-1+ lymphocytes, and a high proportion of either PD-L1+CD68+ macrophages or PD-1+CD4+ and PD-1+CD8+ T cells translates into favorable survival. In contrast, the number of PD-L1+ lymphoma cells or PD-L1– macrophages do not associate with outcome. In multivariate analyses with IPI, PD-L1+CD68+ macrophage and PD-1+ lymphocyte contents remain as independent prognostic factors for survival. In conclusion, high PD-L1+CD68+ macrophage and PD-1+ lymphocyte contents predict favorable survival in patients with primary testicular lymphoma. The findings implicate that the tumor microenvironment and PD-1 – PD-L1 pathway have a significant role in regulating treatment outcome. They also bring new insights to the targeted therapy of primary testicular lymphoma.

1908

Introduction Primary testicular lymphoma (PTL) is a rare and aggressive lymphoid malignancy affecting mainly elderly men. The biology of PTL is beginning to emerge,1-7 and the outcome has improved with the addition of anthracycline-based chemotherapy, central nervous system (CNS) targeted therapy and irradiation of the contralateral testis.8-10 The majority of PTLs represent diffuse large B-cell lymphoma (DLBCL) displaying more often non-germinal center B-cell (GCB) than GCB-like signatures.11 Somatic mutations in NF-κ-B pathway genes, such as MYD88 and CD79B, as well as rearrangements of programmed cell death ligand (PD-L) -1 and -2 genes, have been shown to be enriched in PTL.2,4 In addition, two stromal signatures associated haematologica | 2018; 103(11)

Tumor-associated macrophages in PTL

with outcome have been described in primary, mainly nodal DLBCL patients treated with immunochemotherapy, forming the backbone for our study.12 We have recently demonstrated that tumor-associated macrophages (TAMs) have a favorable prognostic impact on survival in DLBCL patients after immunochemotherapy,13 whereas other groups have investigated the role of programmed cell death-1 (PD-1) pathway in DLBCL.14-18 While PD-1 protein is expressed predominantly by activated tumorinfiltrating lymphocytes (TILs), its ligands (PD-L1 and PDL2) have been shown to be expressed both by the tumor cells and the tumor microenvironment.15,19-21 An unexpected feature has been that PD-L1 expression by the tumorinfiltrating myeloid and other immune cells can be more prevalent than PD-L1 expression by the tumor cells.15,19,20 Recently, it was also shown that the expression of PD-L1, not only by the tumor cells but also by the host cells, plays a critical role in mediating the immunosuppressive function of the PD-1 pathway.21 In DLBCL, expression of PD-L1 by lymphoma cells has been associated with poor outcome.14 Interestingly, 9p24.1/PD-L1/PD-L2 copy number alterations and additional translocations of these loci are frequent in PTLs (>50%), leading to increased expression of the PD-Ls,4 and possibly also to immune escape. Whether the expression of PD-1 and PD-Ls predict survival in PTL, and in which compartments, is unknown. With the aim of resolving the relative expression of checkpoint molecules by the tumor and host immune cells in patients with PTL, we examined B cells, TAMs, TILs, and checkpoint molecules by using multiplex immunohistochemistry (mIHC),22 allowing simultaneous detection of CD68+ TAMs, CD163+ or c-Maf+ M2-polarized TAMs, CD4+ and CD8+ T cells, CD20+ B cells, and the checkpoint molecules PD-L1, PD-L2 and PD-1. The findings were correlated with clinical parameters and survival.

Methods Patients We identified 74 PTL patients with DLBCL histology diagnosed between the years 1987 and 2013 from the pathology databases of the University Hospitals in Southern Finland. Histological diagnosis was established from surgical pretreatment tumor tissue according to current criteria of the World Health Organization (WHO) classification.23 The majority of the patients were treated with anthracycline-based chemotherapy. About half of the patients received rituximab as a part of their treatment. Contralateral testis was treated with surgical excision or irradiation for a minority of the patients. Patients were divided into three equal tertiles, based on the content of different immune cell subtypes (high, intermediate, low). The patient characteristics are described in more detail in Table 1. The protocol and sampling were approved by the Institutional Review Boards, Ethics Committees and the Finnish National Supervisory Authority for Welfare and Health.

Multiplex immunohistochemistry (mIHC) Formalin-fixed, paraffin-embedded (FFPE) primary tumor tissues were collected from the local biobanks and reviewed to match the latest WHO classification.23 Selection of the cores on the tissue microarray (TMA) was based on the evaluation of a hematopathologist. TMA was constructed and the sections (3.5 µm) stained with 4-plex primary antibody panels (PD-L1, PD-L2, haematologica | 2018; 103(11)

CD68, c-MAF; CD3, CD4, CD8, PD-1; CD20, CD163, PD1, PDL1; Online Supplementary Table S1), followed by fluorescently labelled secondary antibodies and DAPI counterstain (nuclear stain). A more detailed description of the stainings is provided in the Online Supplementary Methods. Fluorescent images were acquired with AxioImager.Z2 (Zeiss, Germany). Machine-learning platform CellProfiler24 2.1.2 was used for cell segmentation, intensity measurements (upper quartile intensity) and immune cell classification. Different cell types were quantified as proportion to all cells (e.g., PD-L1⁺CD68⁺ implying the number of PD-L1⁺CD68⁺ TAMs from all cells in a TMA spot) or as a proportion to a specific cell subtype (e.g., PD-L1⁺CD68⁺/CD68⁺ implying the number of PD-L1⁺CD68⁺ cells from all CD68⁺ TAMs). Spots with less than 5000 cells were excluded from the analysis, and data from duplicate spots from the same patient were merged.

Gene expression analysis CD68, CD163, MAF, MS4A1 (CD20), CD274 (PD-L1), PDCD1LG2 (PD-L2), and PDCD1 (PD-1) mRNA levels were measured from 60 PTL samples using digital gene expression analysis with NanoString nCounter (Nanostring Technologies, Seattle, WA, USA).25

Survival definitions and statistical analyses Overall survival (OS) was defined as time between diagnosis and death from any cause, disease specific survival (DSS) as time between diagnosis and lymphoma related death, and progression free survival (PFS) as time between diagnosis and lymphoma progression or death from any cause. Statistical analyses were performed with IBM SPSS v.24.0 (IBM, Armonk, NY, USA). Differences in the frequency of prognostic factors between three patient groups were analyzed by KruskalWallis test. Correlations between gene expression values and cell counts as well as between different immune cell subpopulations were tested with Spearman's rank correlation. Survival rates were estimated using the Kaplan–Meier method. Univariate and multivariate analyses were performed according to the Cox proportional hazards regression model. The potential bias due to duration of follow up was assessed by Schoenfeld residual. Probability values below 0.05 were considered statistically significant. All comparisons and all comparative tests were two-tailed.

Results Patient characteristics Patient and treatment characteristics of the study cohort are shown in Table 1. The majority of the patients represented non-GCB phenotype, low stage, and had low/intermediate International Prognostic Index (IPI). Altogether, 34 deaths, 24 relapses and 24 lymphoma-associated deaths occurred during the median follow up of 67 months (range from 6.7 to 120 months). Five-year OS, DSS and PFS rates were 56%, 68%, and 53%, respectively.

Association of CD68, PD-L1 and PD-L2 encoding gene expression with survival First, we determined the gene expression of the macrophage markers (CD68, CD163 and MAF), checkpoint molecules CD274 (PD-L1), PDCD1LG2 (PD-L2) and PDCD1 (PD-1), and the B-cell marker MS4A1 (CD20). CD68 expression correlated positively with CD274 (rs=0.654, P<0.001), PDCD1LG2 (rs=0.636, P<0.001), CD163 (rs=0.602, P<0.001), and MAF (rs=0.425, P=0.001) levels, and to a lesser extent with PDCD1 (rs=0.300, P=0.020), whereas no correlation between CD68 and 1909

M. Pollari et al.

Table 1. Patient and treatment characteristics.

PD-L1+ CD68+ low

PD-L1+ CD68+ intermed.

PD-L1+ CD68+ high

74 70 (36-92)

25 (34) 68 (38-86)

24 (32) 73 (37-92)

25 (34) 66 (46-90)

17 (23) 57 (77)

6 (24) 19 (76)

4 (17) 20 (83)

7 (28) 18 (72)

0.638

17 (23) 56 (76) 1 (1)

8 (32) 17 (68)

4 (17) 20 (83)

5 (20) 19 (76) 1 (4)

0.426

47 (64) 24 (32) 3 (4)

10 (40) 15 (60)

16 (67) 6 (25) 2 (8)

21 (84) 3 (12) 1 (4)

0.002

50 (68) 20 (27) 4 (5) 36 (49) 34 (46) 7 (9) 23 (31) 12 (16) 11 (15) 60 (81) 35 (47) 1 (1) 9 (12)

13 (52) 11 (44) 1 (4) 9 (36) 8 (32) 2 (8) 6 (24) 2 (8) 4 (16) 18 (72) 9 (36) 0 (0) 4 (16)

17 (71) 5 (21) 2 (8) 11 (46) 10 (42) 3 (13) 7 (29) 3 (13) 4 (17) 21 (88) 11 (46) 1 (4) 4 (17)

20 (80) 4 (16) 1 (4) 16 (64) 16 (64) 2 (8) 10 (40) 7 (28) 3 (12) 21 (84) 15 (60) 0 (0) 1 (4)

0.065

All n (%)

Number of patients Median age (range) Age <60, years â&#x2030;Ľ 60, years Molecular subgroup GCBa Non-GCB NA Stage I-II III-IV NA IPI score 0-2 3-5 NA CNS prophylaxis IV prophylaxis IT prophylaxis Contralateral testis treated Irradiation Surgical excision Anthracycline-based chemotherapy Treated with rituximab Relapse of contralateral testis CNS progression

0.137 0.057 0.856 0.464 0.136 0.884 0.305 0.237 0.377 0.312

a GCB: germinal center B-cell like; NA: not applicable; IPI: International prognostic Index; CNS: central nervous system; IV: intravenous; IT: intrathecal; P: P-value determined by Kruskal-Wallis test.

MS4A1 expression was found. Furthermore, the expression of CD68, CD274 and PDCD1LG2 genes analyzed as continuous variables, but not PDCD1, CD163 or MAF, translated into favorable survival (Table 2).

High PD-L1+ TAM content predicts favorable survival To explore the expression of the checkpoint molecules in the tumor cells and in the microenvironment in more detail, we analyzed the cell immunophenotypes with mIHC from a PTL TMA using four primary antibodies and DAPI (nuclear stain) simultaneously (Figure 1A-C; see also Table 1 for the TMA cohort used and Online Supplementary Table S1 for the antibody panels). The marker CD68 was used to identify all TAMs. Subpopulations of TAMs were defined by the presence and absence of CD163, c-MAF, PD-L1 and PD-L2 (Figure 1A-B, D). In addition, CD20 marker was used to identify lymphoma cells (Figure 1B). For detecting TILs, a panel with CD3, CD4, CD8, and PD1 antibodies was used (Figure 1C). As proof of concept, we found high agreement with the gene expression and the mIHC data when analyzing the quantities of CD68+ macrophages (rs=0.637, P<0.001), 1910

Table 2. Cox regression analysis at the univariate level showing association of gene expression levels with overall survival.

Gene symbol

HRa

95% CI

CD68 CD274 PDCD1LG2 PDCD1 CD163 MAF

0.505 0.737 0.688 0.846 0.914 0.899

0.290-0.881 0.592-0.919 0.505-0.936 0.659-1.088 0.636-1.313 0.551-1.466

0.016 0.007 0.017 0.192 0.627 0.668

a HR: hazard ratio; CI: confidence interval. Boldface font indicates statistical significance (P<0.05).

lymphoma cells (rs=0.704, P<0.001) and PD-L1+ cells (rs=0.710, P<0.001) (Online Supplementary Figure S1). The proportions of the different cell types in the tumor tissue are shown in Figure 1D. The most prominent non-malignant cell type was CD3+ T-lymphocyte (median 45%, range 5-97%). TAM and PD-L1+ cell contents showed a great variation between the samples (CD68+ TAMs, median 23%, range 3-81%; PD-L1+ cells, median 15%, range 0.01-100%), and a large proportion of lymphoma cells (median 34%, range 0.1-100%) and TAMs (median 41%, haematologica | 2018; 103(11)

Tumor-associated macrophages in PTL

Figure 1. Characterization of cell immunophenotypes with mIHC. (A-C) Representative images from 4-plex mIHC stainings. Panels (low, intermediate, and high) show representative images from the corresponding tertiles, based on the content of different immune cell subtypes. The insets highlight cells with higher magnification. PD-L1=blue, PD-L2=red, CD68=white, c-Maf=green (A); PD-L1=blue, CD163=red, CD20=white, PD1=green (B); CD3=blue, CD8=red, CD4=white, PD1=green (C). Scale bar 40 mm. (D) Proportions of distinct immune cell subpopulations from all cells. PD-L1+CD68+ indicating the content of PD-L1+ TAMs, PD-L1+CD163+ and PDL1+CD68+c-Maf+ the content of PD-L1+ M2-polarized TAMs, PD-1+CD3+CD4+ and PD-1+CD3+CD8+ the content of PD-1+ TILs, and PD-L1+CD20+ the content of PD-L1+ lymphoma cells.

range 0.1-99%) expressed PD-L1. Due to a low proportion of PD-L2+ cells (0.06%) (data not shown), PD-L2 was excluded from further analyses. We further observed that a high number of PD-L1+ cells, high proportion of PD-L1+CD68+ macrophages from all cells, as well as a high proportion of PD-L1+CD68+ macrophages from all CD68+ macrophages (PDL1+CD68+/CD68+), associated with favorable OS when analyzed as continuous variables (Table 3). In order to use an objective cutoff, we stratified the patients into three equal subgroups based on tertiles of the PD-L1⁺CD68⁺ macrophage counts (high, intermediate, low). The 5-year OS and DSS rates were clearly worse for the patients with a low number of PD-L1⁺CD68⁺ macrophages (≤4.75% corresponding to the lowest tertile of the patients) in comparison to the patients with intermediate or high numbers (>4.75%, 5-y OS, 39% vs. 66%, P=0.014; 5-y DSS, 53% vs. 76%, P=0.056; Figure 2A). When PD-L1⁺CD68⁺ macrophage count was included in a multivariate analysis with IPI, both factors had independent prognostic value for OS (Table 4). In contrast, neither PD-L1+ lymphoma cells, PD-L1+CD68– cells nor any other TAM phenotypes were significantly associated with survival (Table 3). When comparing the three PD-L1+CD68+ TAM subgroups (high, intermediate and low), no significant differences in age, molecular subtype, IPI score or treatments were observed (Table 1). However, high PD-L1+CD68+ macrophage count was associated with limited disease stage. When the patients treated in the pre-rituximab era were removed from the analyses, a trend towards worse survival was maintained for the patients with low number of PD-L1+CD68+ macrophages (≤5.97%, the lowest tertile; OS, P=0.093, Online Supplementary Figure S2A). These haematologica | 2018; 103(11)

Table 3. Cox regression analysis at the univariate level showing association of cell immunophenotypes with overall survival.

Cell immunophenotype +

PD-L1 CD20+ PD-L1+CD20+ PD-L1+CD68CD68+ PD-L1+CD68+ PD-L1+CD68+/CD68+ PD-L1–CD68+ CD68+c-Maf+ PD-L1+CD68+c-Maf+ CD163+ PD-L1+CD163+ CD3+ PD-1+CD3+CD4+ PD-1+CD3+CD8+

HRa

95% CI

0.983 1.009 0.993 0.981 0.986 0.965 0.987 1.012 0.835 0.734 0.996 0.989 0.194 0.089 0.042

0.967-0.999 0.995-1.023 0.978-1.008 0.955-1.007 0.964-1.008 0.933-0.999 0.975-0.998 0.982-1.043 0.668-1.044 0.518-1.041 0.978-1.014 0.969-1.010 0.053-0.712 0.008-0.999 0.003-0.537

0.038 0.209 0.376 0.146 0.196 0.042 0.027 0.437 0.113 0.083 0.666 0.298 0.013 0.050 0.015

a HR: hazard ratio; CI: confidence interval; PD-L1+CD68+ implies the number of PDL1+CD68+ TAMs from all cells; PD-L1+CD68+/CD68+ implies the number of PD-L1+CD68+ TAMs from all CD68+ TAMs. Boldface font indicates statistical significance (P<0.05).

results highlight the clinical relevance and possible functional connection of PD-L1+ TAMs for PTL progression.

Association of PD1+ TILs with survival Given the prognostic value of PD-L1+ TAMs, we then determined their association with T cells by mIHC. The marker CD3 was used to identify all T cells. Subpopulations of T cells were then defined by the pres1911

M. Pollari et al. A

Figure 2. Association of the immune cell subtypes with survival. (A-C) Cell immunophenotypes were determined by mIHC from 74 PTL patients. Patients were stratified into three equal subgroups (high, intermediate and low) based on tertiles of PD-L1+CD68+ TAM, PD-1+CD3+CD4+ T cell, and PD-1+CD3+CD8+ T-cell counts. KaplanMeier plots depict survival differences between the PD-L1+CD68+ (A), PD-1+CD3+CD4+ (B), and PD-1+CD3+CD8+ (C) groups. P-values were determined by univariate Cox regression analysis (HR, hazard ratio with 95% confidence interval).

Table 4. Cox regression analysis at multivariate level showing independent association of low cell immunophenotypes and IPI high (IPI 3-5) with overall survival.

Cell immunophenotype +

PD-L1 CD68 IPI PD-L1+CD68+/CD68+ IPI PD-1+CD3+CD4+ IPI PD-1+CD3+CD8+ IPI

HRa

95% CI

2.214 4.325 2.275 3.608 2.654 4.907 2.259 4.971

1.054-4.650 2.008-9.312 1.054-4.909 1.643-7.923 1.261-5.586 2.275-10.585 1.075-4.748 2.314-10.678

0.036 <0.001 0.036 0.001 0.010 <0.001 0.031 <0.001

HR, hazard ratio; CI: confidence interval; IPI: International Prognostic Index.

ence and absence of CD4, CD8 and PD-1 (Figure 1C-D). As with CD4+ T-helper and CD8+ cytotoxic cells in general, PD-1+CD3+CD4+ and PD-1+CD3+CD8+ T-cell counts correlated with the PD-L1+ TAM counts (Online Supplementary Table S2). Furthermore, as overall with T-cells,25 a high and intermediate number of PD-1+ CD4+ and CD8+ T-cells associated with superior survival (PD1+CD3+CD4+ cells ≤5.7% corresponding to the lowest tertile vs. other patients; 5-y OS, 34% vs. 68%, P=0.002; 5-y DSS, 43% vs. 81%, P<0.001; PD-1+CD3+CD8+ cells, ≤7.2% 1912

corresponding to the lowest tertile vs. other patients; 5-y OS, 39% vs. 65%, P=0.008; 5-y DSS, 43% vs. 81%, P<0.001; Figures 2B-C). In multivariate analyses with IPI, both PD-1+CD3+CD4+ and PD-1+CD3+CD8+ T-cell counts maintained an independent association with OS (Table 4). When the patients treated in the pre-rituximab era were removed from the analyses, a low number of PD-1+ T cells maintained their adverse impact on survival (PD1+CD3+CD4+ cells, ≤8.50%, the lowest tertile; OS, P=0.001 and PD-1+CD3+CD8+ cells, ≤11.02%, the lowest tertile; OS, P=0.034; Online Supplementary Figure S2B-C).

Discussion In this study, we applied mIHC and digital image analysis to a TMA comprised of PTL tissue from 74 patients. We show that PTL microenvironment contains a heterogeneous TAM population. Among these, PD-L1+ TAMs were the predominant subpopulation, and high infiltration of PD-L1+CD68+ TAMs was associated with favorable survival. Additionally, PD-1+ CD4+ and CD8+ TIL contents correlated with PD-L1+ TAM infiltration and survival, and both PD-L1+ TAMs and PD-1+ TILs emerged as independent indicators of survival for the patients with PTL. In contrast, neither PD-L1+ lymphoma cells, other PD-L1+ cells haematologica | 2018; 103(11)

Tumor-associated macrophages in PTL

than TAMs, nor other TAM phenotypes correlated with survival. The findings highlight the specific roles of TAMs, TILs and PD-1-PD-L1 axis in regulating survival and therapy resistance in PTL. mIHC is a novel technology enabling multi-parametric readout from a single tissue section. In our study, the simultaneous use of multiple markers is important in many ways. Firstly, while PD-L1 was found to be expressed both in TAMs and B cells including lymphoma cells, the prognostic impact of PD-L1 positivity was restricted to TAMs. Thus, the use of just one marker would not be able to detect the survival association. Secondly, the spatial relationships between TILs, TAMs and lymphoma cells are retained in our experimental strategy, allowing for a more precise appreciation of their biological interactions. Thirdly, since mIHC was performed on all evaluable PTL tissue areas on the TMA, thereby providing an overall snapshot of the PTL microenvironment, we can avoid a bias of earlier observations focusing only on hot spot areas of immune cell counts using single marker immunohistochemistry. However, it should be noted that while the overall infiltration of PD-L1+ TAMs and PD-1+ TILs had a significant impact on survival, their functional statuses remain to be explored. Combining our panel with other multiplex panels for immunoregulatory molecules, such as FoxP3, LAG-3 or IDO-1 and IDO-2, may be useful in the evaluation of response to immunotherapy. As described in a recent review article by Xu-Monette et al., the PD-L1 expression in the tumor microenvironment has not been previously well defined in B-cell lymphomas, and association with survival has not been demonstrated.18 PD-1 is a protein, which is classically upregulated upon activation of T lymphocytes. Interaction between PD-1 and PD-L1 was previously thought to induce immune tolerance by leading T lymphocytes to apoptosis.26 Further studies have, however, revealed that the expression of PD-L1 on tumor cells can lead to immune escape, to T-cell exhaustion and a state of nonresponsiveness, further enabling immune escape of the tumor cells.27-29 Moreover, in addition to binding to PD-1,

References 5. 1. Deng L, Xu-Monette ZY, Loghavi S, et al. Primary testicular diffuse large B-cell lymphoma displays distinct clinical and biological features for treatment failure in rituximab era: a report from the International PTL Consortium. Leukemia. 2016; 30(2):361-372. 2. Twa DDW, Mottok A, Savage KJ, Steidl C. The pathobiology of primary testicular diffuse large B-cell lymphoma: Implications for novel therapies. Blood Rev. 2018; 32(3):249-255. 3. Frick M, Bettstetter M, Bertz S, et al. Mutational frequencies of CD79B and MYD88 vary greatly between primary testicular DLBCL and gastrointestinal DLBCL. Leuk Lymphoma. 2018;59(5):1260-1263. 4. Chapuy B, Roemer MG, Stewart C, et al. Targetable genetic features of primary tes-

haematologica | 2018; 103(11)

PD-L1 and PD-L2 can also bind to CD80/B7-1 (PD-L1)30,31 and RGMb (PD-L2),32 indicating that the PD-1 – PD-L1 pathway is much more complex than previously anticipated.18 In addition to PD-L1, macrophages express PD-1.33,34 Recently, Gordon and coworkers showed that PD-1 expression by TAMs inhibits phagocytosis and tumor immunity.35 In addition, they demonstrated that blockade of PD-1 – PD-L1 interaction increases macrophage phagocytosis, reduces tumor growth and lengthens survival in mouse models of colon cancer, suggesting the PD-1 – PDL1 pathway has a significant role in TAM function and tumor survival. Based on our findings, we suggest that the PD-1 - PD-L1 signaling between TAMs and TILs has clinical relevance in PTL. As PD-1 engagement on T cells to its ligands has been linked to decreased anti-tumor immunity, and early experience on PD-1 blockade in PTL has shown promising results,36 the association of high PD-L1+ TAM and PD-1+ Tcell count with favorable outcome in response to immunochemotherapy seems paradoxical. Yet, the interaction of PD-L1+ TAMs and PD-1+ T cells might modify the tumor microenvironment in PTL, or otherwise promote an anti-tumor immune response following immunochemotherapy. In conclusion, we argue that high PD-L1+ TAM and PD-1+ T-cell counts correlate with each other and with favorable outcome in patients with PTL. Higher PDL1+CD68+ TAM scores seem to protect the patients from progression and death, and identify a group of patients with favorable prognosis. Interestingly, apart from PD-L1+CD68+ TAMs, no association was found between other PD-L1+ cells or PD-L1– TAMs and survival. Together, the data demonstrate that the PD-1 - PD-L1 axis in PTL affects the survival of patients with PTL. Acknowledgments We thank Drs. Petri Auvinen and Lars Paulin (Institute of Biotechnology, University of Helsinki), Finland for the Nanostring analyses. Anne Aarnio and Marika Tuukkanen are acknowledged for technical assistance.

ticular and primary central nervous system lymphomas. Blood. 2016;127(7):869-881. Menter T, Ernst M, Drachneris J, et al. Phenotype profiling of primary testicular diffuse large B-cell lymphomas. Hematol Oncol. 2014;32(2):72-81. Twa DD, Mottok A, Chan FC, et al. Recurrent genomic rearrangements in primary testicular lymphoma. J Pathol. 2015; 236(2):136-141. Kridel R, Telio D, Villa D, et al. Diffuse large B-cell lymphoma with testicular involvement: outcome and risk of CNS relapse in the rituximab era. Br J Haematol. 2017; 176(2):210-221. Vitolo U, Chiappella A, Ferreri AJ, et al. First-line treatment for primary testicular diffuse large B-cell lymphoma with rituximab-CHOP, CNS prophylaxis, and contralateral testis irradiation: final results of an international phase II trial. J Clin Oncol. 2011;29(20):2766-2772.

9. Tokiya R, Yoden E, Konishi K, et al. Efficacy of prophylactic irradiation to the contralateral testis for patients with advanced-stage primary testicular lymphoma: an analysis of outcomes at a single institution. Int J Hematol. 2017;106(4):533-540. 10. Zucca E, Conconi A, Mughal TI, et al. Patterns of outcome and prognostic factors in primary large-cell lymphoma of the testis in a survey by the International Extranodal Lymphoma Study Group. J Clin Oncol. 2003;21(1):20-27. 11. Cheah CY, Wirth A, Seymour JF. Primary testicular lymphoma. Blood. 2014; 123(4):486-493. 12. Lenz G, Wright G, Dave SS, et al. Stromal gene signatures in large-B-cell lymphomas. N Engl J Med. 2008;359(22):2313-2323. 13. Riihijarvi S, Fiskvik I, Taskinen M, et al. Prognostic influence of macrophages in patients with diffuse large B-cell lymphoma: a correlative study from a Nordic

1913

M. Pollari et al.

14.

15.

16.

17. 18.

19.

20.

1914

phase II trial. Haematologica. 2015; 100(2):238-245. Kiyasu J, Miyoshi H, Hirata A, et al. Expression of programmed cell death ligand 1 is associated with poor overall survival in patients with diffuse large B-cell lymphoma. Blood. 2015;126(19):21932201. Kwon D, Kim S, Kim PJ, et al. Clinicopathological analysis of programmed cell death 1 and programmed cell death ligand 1 expression in the tumour microenvironments of diffuse large B cell lymphomas. Histopathology. 2016; 68(7):1079-1089. Goodman A, Patel SP, Kurzrock R. PD-1PD-L1 immune-checkpoint blockade in Bcell lymphomas. Nature reviews Clinical oncology. 2017;14(4):203-220. Eyre TA, Collins GP. Immune checkpoint inhibition in lymphoid disease. Br J Haematol. 2015;170(3):291-304. Xu-Monette ZY, Zhou J, Young KH. PD-1 expression and clinical PD-1 blockade in B-cell lymphomas. Blood. 2018;131(1):6883. Carey CD, Gusenleitner D, Lipschitz M, et al. Topological analysis reveals a PD-L1 associated microenvironmental niche for Reed-Sternberg cells in Hodgkin lymphoma. Blood. 2017;130(22):2420-2430. Powles T, Eder JP, Fine GD, et al. MPDL3280A (anti-PD-L1) treatment leads to clinical activity in metastatic bladder cancer. Nature. 2014;515(7528):558-562.

21. Lau J, Cheung J, Navarro A, et al. Tumour and host cell PD-L1 is required to mediate suppression of anti-tumour immunity in mice. Nat Commun. 2017;8:14572. 22. Blom S, Paavolainen L, Bychkov D, et al. Systems pathology by multiplexed immunohistochemistry and whole-slide digital image analysis. Sci Rep. 2017; 7(1):15580. 23. Swerdlow SH, Campo E, Harris NL, et al. WHO classification of tumours of haematopoietic and lymphoid tissues, Revised Fourth Edition. 2017; Volume 2. 24. Carpenter AE, Jones TR, Lamprecht MR, et al. CellProfiler: image analysis software for identifying and quantifying cell phenotypes. Genome Biol. 2006;7(10):R100. 25. Leivonen S-K, Pollari M, BrĂźck O, et al. Clinical impact of the T/NK-cell signature predicts poor survival in patients with primary testicular and diffuse large B-cell Lymphomas. Blood. 2017;130 (Suppl 1):#4027. 26. Dong H, Strome SE, Salomao DR, et al. Tumor-associated B7-H1 promotes T-cell apoptosis: a potential mechanism of immune evasion. Nat Med. 2002;8(8):793800. 27. Wherry EJ, Kurachi M. Molecular and cellular insights into T cell exhaustion. Nat Rev Immunol. 2015;15(8):486-499. 28. Pardoll DM. The blockade of immune checkpoints in cancer immunotherapy. Nat Rev Cancer. 2012;12(4):252-264. 29. Pauken KE, Wherry EJ. Overcoming T cell

30.

31.

32.

33.

34. 35.

36.

exhaustion in infection and cancer. Trends Immunol. 2015;36(4):265-276. Butte MJ, Keir ME, Phamduy TB, Sharpe AH, Freeman GJ. Programmed death-1 ligand 1 interacts specifically with the B7-1 costimulatory molecule to inhibit T cell responses. Immunity. 2007;27(1):111-122. Park JJ, Omiya R, Matsumura Y, et al. B7H1/CD80 interaction is required for the induction and maintenance of peripheral Tcell tolerance. Blood. 2010;116(8):12911298. Xiao Y, Yu S, Zhu B, et al. RGMb is a novel binding partner for PD-L2 and its engagement with PD-L2 promotes respiratory tolerance. J Exp Med. 2014;211(5):943-959. Huang X, Venet F, Wang YL, et al. PD-1 expression by macrophages plays a pathologic role in altering microbial clearance and the innate inflammatory response to sepsis. Proc Natl Acad Sci USA. 2009; 106(15):6303-6308. Bally AP, Lu P, Tang Y, et al. NF-kappaB regulates PD-1 expression in macrophages. J Immunol. 2015;194(9):4545-4554. Gordon SR, Maute RL, Dulken BW, et al. PD-1 expression by tumour-associated macrophages inhibits phagocytosis and tumour immunity. Nature. 2017; 545(7655):495-499. Nayak L, Iwamoto FM, LaCasce A, et al. PD-1 blockade with nivolumab in relapsed/refractory primary central nervous system and testicular lymphoma. Blood. 2017;129(23):3071-3073.

haematologica | 2018; 103(11)

ARTICLE

Stem Cell Transplantation

A phase II/III randomized, multicenter trial of prednisone/sirolimus versus prednisone/ sirolimus/calcineurin inhibitor for the treatment of chronic graft-versus-host disease: BMT CTN 0801 Paul A. Carpenter,1 Brent R. Logan,2 Stephanie J. Lee,1 Daniel J. Weisdorf,3 Laura Johnston,4 Luciano J. Costa,5 Carrie L. Kitko,6 Javier Bolaños-Meade,7 Stefanie Sarantopoulos,8 Amin M. Alousi,9 Sunil Abhyankar,10 Edmund K. Waller,11 Adam Mendizabal,12 Jiaxi Zhu,12 Kelly A. O’Brien,12 Aleksandr Lazaryan,3 Juan Wu,12 Eneida R. Nemecek,13 Steven Z. Pavletic,14 Corey S. Cutler,15 Mary M. Horowitz2 and Mukta Arora3 on behalf of the BMT CTN.

Fred Hutchinson Cancer Research Center, Seattle, WA; 2Medical College of Wisconsin, Milwaukee, WI; 3University of Minnesota, Minneapolis, MN; 4Stanford Hospital and Clinics, Stanford, CA; 5University of Alabama at Birmingham, AL; 6Vanderbilt University School of Medicine, Nashville, TN; 7Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD; 8Duke University Medical Center, Durham, NC; 9 University of Texas MD Anderson Cancer Center, Houston, TX; 10The University of Kansas Medical Center, Kansas City, KS; 11Emory University School of Medicine, Atlanta, GA; 12The Emmes Corporation, Rockville, MD; 13Oregon Health and Sciences University, Portland, OR; 14Experimental Transplantation and Immunology Branch, National Cancer Institute, Bethesda, MD and 15Dana Farber Cancer Institute, Boston, MA, USA

Ferrata Storti Foundation

Haematologica 2018 Volume 103(11):1915-1924

ABSTRACT

nitial therapy of chronic graft-versus-host disease is prednisone ± a calcineurin-inhibitor, but most patients respond inadequately. In a randomized, adaptive, phase II/III, multicenter trial we studied whether prednisone/sirolimus or prednisone/sirolimus/photopheresis was more effective than prednisone/sirolimus/calcineurin-inhibitor for treating chronic graft-versus-host disease in treatment-naïve or early inadequate responders. Primary endpoints of this study were proportions of subjects alive without relapse or secondary therapy with 6-month complete or partial response in phase II, or with 2-year complete response in phase III. The prednisone/sirolimus/photopheresis arm closed prematurely because of slow accrual and the remaining two-drug versus three-drug study ended in phase II due to statistical futility with 138 evaluable subjects. The two-drug and three-drug arms did not differ in rates of 6-month complete or partial response (48.6% versus 50.0%, P=0.87), or 2-year complete response (14.7% versus 15.5%, P=0.90). Serum creatinine values >1.5 times baseline were less frequent in the calcineurin-inhibitor-free arm at 2 months (1.5% versus 11.7%, P=0.025) and 6 months (7.8% versus 24.0%, P=0.016). Higher adjusted Short Form-36 Physical Component Summary and Physical Functioning scores were seen in the two-drug arm at both 2 months (P=0.02 and P=0.04, respectively) and 6 months (P=0.007 and P=0.001, respectively). Failure-free survival and overall survival rates at 2 years were similar for patients in the the two-drug and three-drug arms (48.6% versus 46.2%, P=0.78; 81.5% versus 74%, P=0.28). Based on similar long-term outcomes, prednisone/sirolimus is a therapeutic alternative to prednisone/sirolimus/calcineurin-inhibitor for chronic graft-versus-host disease, being easier to administer and better tolerated. Clinicaltrials.gov identifier: NCT01106833. haematologica | 2018; 103(11)

Correspondence: pcarpent@fredhutch.org

Received: April 7, 2018. Accepted: June 26, 2018. Pre-published: June 28, 2018. doi:10.3324/haematol.2018.195123 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1915 ©2018 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.

1915

P.A. Carpenter et al.

Introduction Chronic graft-versus-host disease (GvHD) is the major treatment-related complication among patients who survive after allogeneic hematopoietic cell transplantation. The protracted duration of chronic GvHD and its protean and sometimes irreversible organ manifestations makes it the leading cause of impaired immunity, compromised functional status, and late treatment-related deaths.1,2 Standard immunosuppressive therapy with prednisone ± a calcineurin inhibitor (CNI) has not changed for three decades, but most patients respond inadequately with slow and most often incomplete control of their disease. The results of six phase III chronic GvHD trials that tested prednisone or prednisone/CNI backbones with or without an experimental immunosuppressive therapy have not changed clinical practice, because the treatments tested in the experimental arms did not improve outcomes due to lack of efficacy and/or more toxicity.3-8 Between 20052007, reports of phase II trials from single centers showed promising results for prednisone combined with sirolimus, rituximab, mycophenolate mofetil, pentostatin or extracorporeal photopheresis (ECP).9-13 At the same time, renewed understanding of regulatory T cells (Treg) led to the hypothesis that experimental therapies permissive of Treg expansion would abrogate GvHD better than CNI-containing (control) immunosuppressive therapy. Sirolimus and ECP are permissive of Treg expansion.14-18 Because of their acceptance in clinical practice sirolimus and ECP were considered good candidates to add to prednisone in clinical trials investigating chronic GvHD. This background, together with the 2005 National Institutes of Health (NIH) chronic GvHD criteria,19 plus other efforts designed to propel the field forward,20,21 motivated the Blood and Marrow Transplant Clinical Trials Network (BMT CTN) to conduct an intervention trial for chronic GvHD. The intent was an adaptive phase II/III design to minimize between-phase downtime. The primary purpose of phase II was to select the more promising of two CNI-free approaches (prednisone/sirolimus and prednisone/sirolimus/ECP) in order to proceed seamlessly into a definitive phase III study against a CNI-containing comparator arm (prednisone/sirolimus/CNI). The study also provided an opportunity to use the NIH diagnostic and response criteria prospectively and to validate them.

Methods Patients Adult and pediatric allogeneic hematopoietic cell transplant recipients were eligible if they had classic chronic GvHD ± acute GvHD (overlap subtype) that met 2005 NIH diagnostic consensus criteria.19 Eligibility criteria were broad and allowed a period of steroid exposure prior to enrollment to ensure congruence with standard practice. Thus, eligible patients were either: newly diagnosed patients, defined as individuals who had received <14 days of prednisone (or equivalent) before randomization to the study therapy, or previously treated, but responding inadequately after ≤16 weeks of initial therapy with prednisone and/or a CNI ± an additional non-sirolimus agent started at the time that the chronic GvHD was diagnosed. Major reasons for exclusion were patients with late persistent acute GvHD or recurrent acute GvHD only, patients unable to begin prednisone at a dose of 0.5 mg/kg day (or equivalent), patients already receiving sirolimus 1916

for treatment of chronic GvHD, and patients already receiving sirolimus (for prophylaxis or treatment of acute GvHD) along with prednisone at ≥0.25 mg/kg/day (or equivalent) ± additional agents. Patients were also ineligible if they had an invasive fungal or viral infection not responding to appropriate therapies, a creatinine clearance <50 mL/min/1.73 m2 based on the CockcroftGault (adults) or Schwartz (age ≤12 years) formula, an absolute neutrophil count <1.5x109/L, a requirement for platelet transfusion, or a progressive or recurrent malignancy defined other than by quantitative molecular assays. Institutional review boards at all participating centers provided ethics approval. All patients or their parents signed informed consent to participation in the trial in accordance with the Declaration of Helsinki.

Treatment plan The starting dose of prednisone (or prednisone-equivalent) was 1 mg/kg once daily, unless contraindicated, in which case the prednisone dose began at 0.5-1 mg/kg once daily. It was recommended that the dose of prednisone was tapered down, over 4-8 weeks to reach a dose of 0.5-1 mg/kg every other day, with the tapering starting within 2 weeks after the first evidence of GvHD improvement. Once an every-other-day prednisone (or equivalent) regimen was achieved, this dose remained constant for 1012 weeks until all reversible chronic GvHD manifestations resolved. A second taper was then attempted and could follow individual institutional guidelines, but it was recommended that the extent of the tapering be approximately calibrated to the magnitude of an individual patient’s every-other-day prednisone dose. CNI therapy was continued by targeting trough serum levels of 5-10 ng/mL for tacrolimus (120-200 ng/mL for cyclosporine). The sirolimus therapy began at a dose of 2 mg orally once daily (1 mg/m2 per day if the patient weighed <40 kg) to target trough serum levels of 3-12 ng/mL. Supportive care was provided in accordance with institutional guidelines reflecting standard practices appropriate for chronic GvHD.22

Study design The trial was designed as an adaptive phase II/III randomized, open-label, prospective study of three treatments for chronic GvHD (Figure 1, Online Supplementary Figures S1 and S2). Phase II included two parallel, 100-patient, randomized trials, comparing prednisone/sirolimus or prednisone/sirolimus/ECP versus identical (prednisone/sirolimus/CNI) comparator arms. The primary objective of phase II was to estimate the proportion of study subjects at 6 months after randomization with complete or partial response, who were alive without relapse or receipt of secondary immunosuppressive therapy. A sufficiently promising phase II result would determine whether the trial would proceed into phase III with additional accrual, would continue as phase II, but be followed for longer phase III endpoints without additional accrual, or would end in failure without further follow-up (Online Supplementary Figure S2). Full details of the study design, endpoint definitions, statistical analysis and study timeline are contained in the Online Supplementary Material.

Results Patients One hundred patients were evaluated for the phase II primary endpoint after all had completed 6 months of follow-up. The Z-statistic comparing complete/partial response rates (51% versus 50%, Z=0.11, stopping boundary Z6≤0.9) did not support proceeding to phase III and, together with pre-specified outcome scenarios, guided the haematologica | 2018; 103(11)

Sirolimus-calcineurin inhibitor for chronic GvHD

Data and Safety Monitoring Board recommendation to suspend further phase III accrual; however, all 151 enrolled subjects were followed for phase III endpoints (Online Supplementary Figure S2B). The endpoint review committee response adjudication determined that 13 (10%) subjects had not satisfied the NIH criteria for the diagnosis of chronic GvHD at enrollment and were ineligible (Online Supplementary Figure S1C); a screening diagnostic checklist for NIH-defined chronic GvHD was required after July 24, 2013. One-hundred thirty-eight remaining subjects were randomized: 72 were assigned to two drugs and 66 to three drugs. Endpoint review committee-adjudicated complete/partial response rates and provider-reported complete/partial response rates were concordant at 6 months (Cohen κ = 0.78) and at 2 years (κ = 0.88). Participants lost to follow-up, or who relapsed, died, or began secondary immunosuppressive therapy were excluded from these agreement tests. Patients’ demographic and transplant characteristics were mostly similar between treatment groups (Table 1). The median age was 50.2 years in the two-drug arm and 54.7 years in the three-drug arm. Characteristics that were more frequent in the two-drug arm were male gender (66.7% versus 50%), Hispanic ethnicity (11.1% versus 0%), an underlying diagnosis of acute leukemia (65.3% versus 37.9%), transplants done for early stage disease (63.9% versus 40.9%), myeloablative conditioning (65.3% versus 43.9%), and CNI/methotrexate-based GvHD prophylaxis (61% versus 33.3%). Lymphoma was less frequent in the two-drug arm (6.9% versus 25.8%). Donor or stem cell sources did not differ between arms.

Chronic GvHD characteristics were similar between the two-drug and three-drug arms (Table 2), including the proportion of subjects with high-risk chronic GvHD (51.4% versus 48.5%), median time from hematopoietic cell transplant to enrollment (7.3 versus 7.7 months), and median time from chronic GvHD diagnosis to enrollment (7 versus 10 days). Sixty-two percent of the study subjects were enrolled within 14 days of a new diagnosis of chronic GvHD and 38% were responding inadequately to initial therapy; classic and overlap subtypes were equally represented. Global severity scores were mainly moderate (67%), then mild (21%) and severe (12%); 43% of the patients had progressive onset chronic GvHD. The organs involved were the skin (76%), mouth (74%), liver (71%), eyes (54%), gastro-intestinal tract (46%), joints/fascia (24%), genital tract (17%), and lungs (15%) with the pattern of involvement being similar between the treatment groups except for genital tract involvement, which was more common in the group treated with three drugs (10% versus 26%, but only 63% of the study population were scored).

Response and survival The treatment success rate at 6 months was 48.6% [95% confidence interval (CI): 36.7%-60.7%] with two drugs versus 50.0% (95% CI: 37.4%-62.6%) with three drugs P=0.87). The rate of more stringently defined treatment success at 2 years (complete responses only) was 14.7% (95% CI: 7.3%-25.4%) with two drugs versus 15.5% (95% CI: 7.4%-27.4%) with three drugs (P=0.90) (Table 3). Since very good partial response is clinically rel-

Figure 1. BMT CTN 0801 Consort flow diagram. ECP: extracorporeal photopheresis; PDN: prednisone; SRL: sirolimus; CNI: calcineurin inhibitor.

haematologica | 2018; 103(11)

1917

P.A. Carpenter et al.

evant, we compared the proportion of patients with complete response/very good partial response who otherwise satisfied the 2-year treatment success definition. By this measure, the 29.4% (95% CI: 19.0-41.7%) in the twodrug arm did not differ from the 22.4% (95% CI: 12.535.3%) in the three-drug arm (P=0.37). To address whether certain categories of chronic GvHD responded differently in the study arms overall, in univariate analysis, we compared subjects with mild, moderate and severe chronic GvHD, and de novo, quiescent onset and progressive chronic GvHD. There were no statistically significant differences between these groups in terms of the odds of treatment success at 6 months or 2 years. Similarly, no particular individual organ involvement was more likely to be associated with treatment success using either treatment regimen. The overall survival rate at 2 years was 81.5% in the group treated with two drugs versus 74% in the group treated with three drugs (P=0.28) (Figure 2A). Progressionfree survival rates were 78.6% with two drugs versus 67.3% with three drugs (P=0.14) (Figure 2B) and the 2year failure-free survival rates were 48.6% with two drugs versus 46.2% with three drugs (P=0.78) (Figure 2C,D). The cumulative incidence of relapse (10.1% versus 14.9%, Table 1. Demographic and hematopoietic cell transplant characteristics.

Gender Female Male Ethnicity Hispanic or Latino Not Hispanic or Latino Unknown/not answered Race White Non-white Unknown/not answered Age, years Median <20 ≥20 Primary disease Acute leukemia Chronic leukemia MDS/MPS Lymphoma Other Disease stage Early Intermediate High risk Stem cell type Bone marrow (BM) Peripheral blood stem cells (PBSC) Cord blood Donor source (for BM/PBSC) Related Unrelated Type of conditioning regimen Myeloablative Non-myeloablative or reduced intensity 1918

P=0.40), non-relapse mortality (5.6% versus 11.1%, P=0.26), and the cumulative incidence of secondary immunosuppressive therapy (38.5% versus 29.4%, P=0.22) were not different between the groups treated with two or three drugs. By 2 years, 13 deaths had occurred in the two-drug arm and 16 in the three-drug arm (Online Supplementary Table S1). Almost one-third of deaths were due to GvHD (30.8% in the prednisone/sirolimus group versus 31.3% in the prednisone/sirolimus/CNI group). Recurrent or progressive malignancy was the primary cause of death in 30.8% of the prednisone/sirolimus group versus 12.5% in the prednisone/sirolimus/CNI group (P=0.36). There were two deaths (15.4%) primarily from infection in the two-drug arm and three (18.8%) in the three-drug arm. Curves depicting cumulative incidence of discontinuation of systemic immunosuppressive therapy by 2 years overlapped, with the values at 2 years being 23.2% in the twodrug arm and 20% in the three-drug arm (P=0.71) (Figure 3). Mean (standard deviation) daily glucocorticoid doses (mg/kg) at baseline were 0.9 (0.4) for the group given prednisone/sirolimus versus 0.8 (0.2) for the group given prednisone/sirolimus/CNI (P=0.13). From baseline to 1 year the mean (standard deviation) daily dose reduction was 0.7 mg/kg (0.3) for the prednisone/sirolimus group and 0.5 mg/kg (0.5) for the prednisone/sirolimus/CNI group.

Two drugs N = 72 (%)

Three drugs N = 66 (%)

24 (33.3) 48 (66.7)

33 (50.0) 33 (50.0)

8 (11.1) 63 (87.5) 1 (1.4)

0 (0.0) 59 (89.4) 7 (10.6)

54 (75.0) 13 (18.1) 5 (6.9)

57 (86.4) 9 (13.6) 0 (0.0)

50.2 5 (6.9) 67 (92.1)

54.7 2 (3.0) 64 (97.0)

47 (65.3) 7 (9.7) 12 (16.7) 5 (6.9) 1 (1.4)

25 (37.9) 8 (12.1) 9 (13.6) 17 (25.8) 7 (10.6)

46 (63.9) 11 (15.3) 15 (20.8)

27 (40.9) 21 (31.8) 18 (27.3)

5 (6.9) 64 (88.9) 3 (4.2)

5 (7.6) 60 (90.9) 1 (1.5)

29 (42.0) 40 (58.0)

38 (58.5) 27 (41.5)

GvHD prophylaxis after transplant CNI + methotrexate ± other CNI + MMF CNI + MMF ± Other CNI + MMF + ATG CNI + sirolimus CNI + corticosteroids CNI only Other regimen Donor age, years Mean (SD) Median Donor age, years <20 20-40 41-60 >60 Donor gender Female Male Unknown Donor recipient gender mismatch Female-female Female-male Male-female Male-male

29 (43.9) 37 (56.1)

Two drugs: prednisone/sirolimus, Three drug: prednisone/sirolimus/calcineurin-inhibitor; MDS/MPS: myelodysplastic syndrome/myeloproliferative syndrome; BM: bone marrow; PBSC: peripheral blood stem cells; CNI: calcineurin inhibitor; MMF: mycophenolate mofetil; ATG: antithymocyte globulin; GvHD: graft-versus-host disease; SD: standard deviation.

47 (65.3) 25 (34.7)

continued from the previous column

Two drugs N = 72 (%)

Three drugs N = 66 (%)

44 (61.0) 9 (12.5) 5 (6.9) 0 (0.0) 3 (4.2) 2 (2.8) 5 (6.9) 4 (5.6)

22 (33.3) 15 (22.7) 7 (10.6) 1 (1.5) 5 (7.6) 0 (0.0) 5 (7.6) 11 (16.7)

38.5 (12.9) 38.8

44.4 (15.2) 47.0

3 (4.8) 31 (50.0) 24 (38.7) 4 (6.5)

4 (6.6) 21 (34.4) 27 (44.3) 9 (14.8)

35 (48.6) 37 (51.4) 0 (0.0)

26 (39.4) 38 (57.6) 2 (3.0)

14 (19.4) 21 (29.2) 10 (13.9) 27 (37.5)

10 (15.6) 16 (25.0) 21 (32.8) 17 (26.6)

haematologica | 2018; 103(11)

Sirolimus-calcineurin inhibitor for chronic GvHD

Biomarkers

T-cell (Tcon) ratios. Treatment success at day 180, irrespectively of treatment arm, was associated with significantly higher median Treg levels at day 60, 21 (range, 278) versus 10 (range, 1-53) among those with treatment failure (P=0.006). This trend attenuated at day 180 (20 versus 16, P=0.14). Similarly, Treg:Tcon ratios at baseline and day 60 were higher among patients in whom treatment was a success than among those in whom it failed (data not shown). B-cell numbers were not significantly different between arms, except at day 180 when the median B-cell number (cells per microliter) was higher in the two-drug arm, 148 (range, 0-1547) than in the three-drgu arm, 68 (range, 01035) (P=0.045). Overall, median plasma BAFF levels were low (1.0 â&#x20AC;&#x201C; 1.9 ng/mL) at all time-points, potentially

Correlative biology studies were attempted using a prespecified analysis of plasma B-cell activating factor (BAFF) levels determined by enyme-linked immunosorbent assay, and CD3+CD4+CD35+CD127â&#x20AC;&#x201C; Treg and CD19+ Bcell enumeration in blood by flow cytometry. Data were available for 99% of patients at baseline, 86% at 60 days and 76% at 180 days after beginning study therapy. Median Treg levels (cells per microliter) in both arms were identical, 26, at baseline (P=0.96), and were lower, 18, at day 60 (P=0.87). At day 180 after starting study therapy, the median Treg number was higher in the group treated with two drugs, 21 (range, 0-68) than in the group treated with three drugs 14.5 (range, 1-239) (P=0.04), but there was no significant difference in Treg: conventional

Figure 2. Overall survival, progression-free survival, and failure-free survival. (A) Probability of overall survival by treatment arm (P=0.281). (B) Probability of progression-free survival by treatment arm (P=0.142). Progression-free survival was defined as no clinical evidence of progression or relapsed disease, or any therapy used to treat persistent, progressive, or relapsed disease including withdrawal of immunosuppressive therapy or donor lymphocyte infusion. (C) Probability of failure-free survival in the (C) sirolimus + prednisone (2-drug) arm and (D) control (3-drug) arm. Failure-free survival was defined by the absence of secondary immunosuppressive therapy for chronic graft-versus-host disease, non-relapse mortality, and recurrent or progressive malignancy during treatment. Note: the numbers shown are estimates at each time point for each endpoint. Siro/Pred: sirolimus + prednisone; NRM: non-relapse mortality; FFS: failure-free survival.

haematologica | 2018; 103(11)

1919

P.A. Carpenter et al. Table 2. Chronic graft-versus-host disease characteristics.

Time from transplant to enrollment in months, median (range) Time from diagnosis of chronic GvHD to enrollment in days, median (range) Chronic GvHD risk statusa High Standard Category of chronic GvHD diagnosis relative to enrollment Treatment naïveb Early inadequate respondersc Chronic GvHD presentation Overlap acute and chronic GvHD Classic chronic GvHD Chronic GvHDs severityd Mild Moderate Severe Chronic GvHD onset De-novo Quiescent Progressive Skin score 0 1 2 3 Mouth score 0 1 2 3 Gastro-intestinal tract score 0 1 2 3 Eye score 0 1 2 3 Joints and fascia score 0 1 2 3 Genital tract score 0 1 2 3 Lung score 0 1 2 3 Liver score 0 1

continued from the previous column

Two drugsa N = 72 (%)

Three drugs N = 66 (%)

7.3 (3.4, 28.6)

7.7 (2.5, 42.4)

7.0 (0.0, 202.0)

10.0 (0.0, 310.0)

37 (51.4) 35 (48.6)

32 (48.5) 34 (51.5)

49 (68) 23 (32)

36 (54.5) 30 (46.5)

36 (50.0) 36 (50.0)

34 (53.1) 30 (46.9)

16 (22.2) 45 (62.5) 11 (15.3)

13 (20.6) 45 (71.4) 5 (7.9)

30 (41.7) 10 (13.9) 32 (44.4)

24 (36.4) 14 (21.2) 28 (42.4)

14 (19.7) 16 (22.5) 22 (31.0) 19 (26.8)

19 (29.7) 11 (17.2) 20 (31.3) 14 (21.9)

20 (27.8) 31 (43.1) 18 (25.0) 3 (4.2)

16 (25.0) 26 (40.6) 17 (26.6) 5 (7.8)

40 (56.3) 17 (23.9) 12 (16.9) 2 (2.8)

33 (51.6) 21 (32.8) 8 (12.5) 2 (3.1)

33 (46.5) 25 (35.2) 9 (12.7) 4 (5.6)

29 (45.3) 24 (37.5) 10 (15.6) 1 (1.6)

53 (74.6) 10 (14.1) 7 (9.9) 1 (1.4)

49 (76.6) 11 (17.2) 4 (6.3) 0 (0.0)

43 (89.6) 3 (6.3) 2 (4.2) 0 (0.0)

28 (73.7) 8 (21.1) 0 (0.0) 2 (5.3)

62 (86.1) 8 (11.1) 1 (1.4) 1 (1.4)

54 (84.4) 6 (9.4) 3 (4.7) 1 (1.6)

19 (26.4) 14 (19.4)

21 (31.8) 23 (34.9)

continued in the next column

1920

2 3 Acute GvHD prior to enrollment Yes No

Two drugsa N = 72 (%)

Three drugs N = 66 (%)

22 (30.6) 17 (23.6)

14 (21.2) 8 (12.1)

42 (58.3) 30 (41.7)

42 (63.6) 24 (36.4)

Two drugs: prednisone/sirolimus; three drugs: prednisone/sirolimus/calcineurin-inhibitor; GvHD: graft-versus-host disease. aChronic GvHD risk status was defined as high-risk in patients with platelets <100,000, >50% skin involvement, bronchiolitis obliterans or those receiving prednisone ≥0.5 mg/kg/day (or equivalent) at the time of chronic GvHD diagnosis. bTreatment-naïve: newly diagnosed for <14 days. cEarly inadequate responders: previously treated and not responding ≤16 weeks from time of first diagnosis of chronic GvHD. dChronic GvHD severity was defined as per the NIH Consensus criteria.31

because of steroid use,23 and did not differ significantly at day 180 between patients in whom treatment had or had not been successful by day 180. Among patients in whom treatment was successful, the median baseline BAFF levels were slightly higher, 1.6 ng/mL (range, 0.2-17.7) versus 1.2 ng/mL (range, 0.1-19.4; P=0.046). BAFF/B-cell ratios were not significantly different between study arms or between patients in whom treatment was or was not successful.

Quality of life Health-related quality of life was measured by selfreport instruments, including the Functional Assessment of Cancer Therapy Bone Marrow Transplantation (FACTBMT), Short Form-36 (SF-36). and Lee symptom scale. Figure 4 shows significantly better scores, at 2 and 6 months, with two drugs versus three drugs for the SF-36 Physical Component Summary, adjusted for baseline scores (P=0.02 and P=0.04), and demonstrated further by SF-36 Physical Functioning sub-scores (P=0.007 and P=0.001). Unadjusted, seven-item FACT-BMT Physical Well-Being scores were higher at 2 months with two drugs, but this observation did not hold when adjusted for baseline scores (P=0.27, graph not shown). Lee symptom scale scores did not differ between arms (data not shown).

Toxicity and adverse events The proportion of patients with grade 3-5 toxicities was similar, 63.9% for those in the two-drug arm versus 56.1% in the three-drug arm. Thrombotic microangiopathy was observed in one patient (1.4%) in the two-drug arm and in three (4.5%) in the three-drug arm. The mean (standard deviation) increase in serum creatinine from baseline to 2 months was significantly higher with three drugs, 0.1 (0.3) mg/dL than with two drugs, 0 (0.2) mg/dL (P=0.002). The proportion of patients with serum creatinine >1.5 times baseline was significantly higher among patients treated with three drugs than among those treated with two drugs at both 2 months (11.7% versus 1.5%, P=0.025), and at 6 months (24.0% versus 7.8%, P=0.016). Twenty-seven subjects in the two-drug arm versus 27 in the three-drug arm (40.9% versus 37.5%, P=0.682) experienced severe to life-threatening/fatal infection episodes with slightly more infection episodes occurring in the three-drug arm (80 versus 52). One patient in the two-drug arm and another three in the three-drug arm developed non-infectious pneumonitis. haematologica | 2018; 103(11)

Sirolimus-calcineurin inhibitor for chronic GvHD

Discussion In this phase II, multicenter, randomized trial comparing two CNI-free approaches against a CNI-containing comparator arm, the comparison of two versus three drugs showed similar outcomes with the CNI-free chronic GvHD therapy. Subjects who received prednisone/sirolimus had better renal and physical function at 2 and 6 months, but these improvements did not result in long-term advantages for recipients of the two-drug immunosuppressive therapy. However, between 6 and 12 months, 20%-30% of subjects treated with two drugs and 11%-24% treated with three drugs had already switched to secondary immunosuppressive therapy, potentially attenuating longer-term associations. Earlier studies suggested that narrower targeting of CNI and sirolimus blood levels might mitigate nephrotoxicity.9,25 Data to evaluate this were not collected, but it is conceivable that side effects, particularly with three drugs, were partially mitigated by detailed study guidance for managing CNI and sirolimus blood levels.

Table 3. Treatment success.

Treatment success at 6 months Two drugs Three drugs (N=72) (N=66) Treatment successa Yes

P-value 0.871

CR PR No Relapse Secondary therapy Death Not in CR/PR

35 (48.6%) [95% CI: 36.7%-60.7%] 3 32 37 (51.4%) [95% CI: 39.3%-63.4%] 1 14 4 18

33 (50.0%) [95% CI: 37.4%-62.6%] 6 27 33 (50.0%) [95% CI: 37.4%-62.6%] 3 7 6 17

Our hypothesis was that CNI-free two-drug immunosuppressive therapy would not impede Treg expansion and would thereby abrogate chronic GvHD better than would three-drug immunosuppressive therapy. The few statistically significant associations that we observed in Treg number and BAFF serum levels were inconsistent and/or difficult to understand suggesting that chronic GvHD biomarker response associations are more complex than Treg alone. Rates of secondary therapy for lack of efficacy between day 90 and day 180 were almost double with two-drug immunosuppressive therapy compared to three-drug therapy, which might have attenuated predicted Treg response associations. We learned important lessons to inform and potentially improve future study design in chronic GvHD. Had phase II been positive, the 51 additionally accrued subjects beyond the phase II target would have jump-started phase III accrual, thereby confirming “built-in” adaptive phase II/III design utility. Second, to begin endpoint review committee adjudications 3 years after enrollment and then discover that 10% of subjects had been ineligible was problematic. For a complex clinical syndrome like chronic GvHD, our fundamental eligibility criterion that required patients to be “diagnosed according to NIH guidelines” was too open to misinterpretation. This problem was resolved after introducing a screening checklist to confirm that patients met the NIH diagnostic criteria for chronic GvHD (Online Supplementary Material). Lastly, a similar problem existed with respect to complete/partial response evaluations. While considerable efforts have been made to standardize and develop more objective chronic GvHD response instruments,26 case report forms for this trial had multiple sections with categorical check boxes and required measures in many organs. The completion of case report forms was inconsistent and has been considered burdensome by others.27 Months of iterative communications between the endpoint review committee, site investigators, and the Data Coordinating Center to clean

Treatment success at 2 years b

Treatment success Yes CR off IST CR on IST No

VGPRc off IST VGPRc on IST PR off IST PR on IST Progression Relapse Secondary therapy Deathd Not evaluablee

0.899 10 (14.7%) [95% CI: 7.3%-25.4%] 3 7 58 (85.3%) [95% CI: 74.6%-92.7%] 6 4 1 9 1 5 28 4 4

9 (15.5%) [95% CI: 7.4%-27.4%] 7 2 49 (84.5%) [95% CI: 72.6%-92.7%] 1 3 3 7 2 8 18 7 8

Two drugs: prednisone/sirolimus, three drugs: prednisone/sirolimus/calcineurin-inhibitor; a Treatment success at 6 months is defined as a complete (CR) or partial response (PR) without secondary systemic immunosuppressive therapy (IST) and no recurrent malignancy or death through 6 months after randomization. bPartial response is excluded from the definition of treatment success at 2 years. cVery good partial response (VGPR) is defined as having trivial residual and asymptomatic graft-versus-host disease features, while on a prednisone dose that is physiological or less (defined as ≤5 mg daily or ≤10 mg every other day). This category splits further into truly off all IST (meaning also off sirolimus and/or calcineurin inhibitor or not. dDeath, relapse, and secondary therapy are mutually exclusive, based on which event occurred first. e Participant withdrew study consent.

haematologica | 2018; 103(11)

Figure 3. Cumulative incidence of discontinuation of all systemic immunosuppressive therapy without the need to add additional therapy by 2 years.

1921

P.A. Carpenter et al.

data were necessary to enable meaningful analysis. Near real-time monitoring (8-10 weeks after diagnosis) was invoked later by the BMT CTN in an effort to improve the quality of clinically annotated data in a more recent prospective correlative study of serum biomarkers, GvHD and other clinical outcomes.28 These experiences from the 0801 CTN trial also directly informed simplifications implemented in the 2014 revision of the NIH diagnostic and response criteria.29.30

Regardless of treatment arm, success rates at 6 months were only modest; approximately half of all subjects failed to achieve a complete or partial response and remain alive without relapse or receipt of secondary immunosuppressive therapy. There were fewer relapses in the two-drug arm, which contained more patients with early stage disease, more with acute leukemia (likely in first remission), and more recipients of myeloablative conditioning. Conceivably these chance imbalances might have resulted

Figure 4. Quality of life. Unadjusted graphs are shown for SF-36 quality of life scores:23 (left) Physical Component Summary scores; (right) Physical Functioning subscale scores.

Figure 5. Six-month landmark analysis. (A) Overall survival after the 6-month landmark with a median follow up for survivors of 30 months. The mortality rates after the 6-month landmark were similar in the complete or partial response group and stable or progressive disease group (hazard ratio, 0.71; 95% confidence interval: 0.17-2.96; P=0.63) or secondary treatment groups (hazard ratio, 0.54; 95% confidence interval: 0.13-2.26; P=0.39), respectively. (B) Cumulative incidence of discontinuation of immunosuppressive therapy after the 6-month landmark. Patients who died or were lost to follow up or experienced relapse before the landmark were excluded (n=19) and no patients ended immunosuppressive therapy before the landmark. CR/PR: patients with complete or partial response without relapse and without secondary therapy at the time of assessment (n=68). SD/PD/Rx: patients not in complete or partial response and alive without relapse and without secondary therapy at the time of assessment (n=31), and patients who had secondary therapy without relapse regardless of response before the landmark (n=16); IST: immunosuppressive therapy.

1922

haematologica | 2018; 103(11)

Sirolimus-calcineurin inhibitor for chronic GvHD

in fewer relapses than in the potentially higher-risk patients who received three drugs. However, our sample size provided insufficient power to be able to detect a difference in relapse. The proportion of subjects who received secondary immunosuppressive therapy in either arm of the trial was not significantly different and it made no difference whether the patient had been enrolled with “newly diagnosed” versus “inadequately responding” chronic GvHD (data not shown). Two large studies assessed failure-free survival among patients who received non-uniform initial chronic GvHD therapy and showed that the rates of failure-free survival were either similar31 or lower32 to those in our study in the context of overall generally higher rates of moderate and severe chronic GvHD. The Consortium recently evaluated 202 subjects in a landmark analyses to determine whether failure-free survival plus complete/partial responses at either 6 months or 1 year predicted downstream clinical benefit. At 1 year, the <20% of study subjects who satisfied the failure-free survival plus complete/partial response end-point versus all other states (failure-free survival with stable/progressive GvHD, or received secondary immunosuppressive therapy) were associated over the subsequent 5 years with significantly fewer GvHD disease manifestations, lower mortality, and earlier discontinuation of immunosuppressive therapy.33 Because a similar 6-month failure-free survival plus complete/partial response landmark analysis revealed less striking associations with downstream clinical benefit, the authors proposed 1-year failure-free survival plus complete/partial responses as the primary endpoint for future pivotal clinical trials of initial therapy. Although 1year endpoint data were not available, at 6 months our two groups without complete/partial response (i.e. failure-free survival with stable/progressive GvHD, or received secondary immunosuppressive therapy) behaved similarly when analyzed separately. They were, therefore, collapsed into one “stable disease/progressive disease/secondary immunosuppressive therapy” group of 47 patients for comparison to 68 patients with complete/partial response who satisfied the failure-free survival endpoint, resulting in just over half the size of the cohort used for the Consortium landmark analysis. Our 6-month failure-free survival plus complete/partial response endpoint predicted earlier time to discontinue immunosuppressive therapy compared to that in the combined group with stable or progressive disease or secondary immunosuppressive therapy (hazard ratio,

References 1. Socie G, Stone JV, Wingard JR, et al. Longterm survival and late deaths after allogeneic bone marrow transplantation. Late Effects Working Committee of the International Bone Marrow Transplant Registry. N Engl J Med. 1999;341(1):14-21. 2. Pidala J, Kurland B, Chai X, et al. Patientreported quality of life is associated with severity of chronic graft-versus-host disease as measured by NIH criteria: report on baseline data from the Chronic GVHD Consortium. Blood. 2011;117(17):4651-4657. 3. Sullivan KM, Witherspoon RP, Storb R, et al.

haematologica | 2018; 103(11)

2.05; 95% CI: 1.15-3.68; P=0.02) (Figure 5), but similar to the findings of Martin et al.,33 there was no striking survival benefit at 6 months, perhaps because of the shorter follow up and/or the small sample size. Larger prospective studies are needed to verify the utility of this endpoint in predicting better survival and shorter duration of immunosuppressive therapy. In summary, this randomized trial showed no difference in response rates between the two treatment arms. Analyses of nephrotoxicity and quality of life demonstrate that initial therapy of chronic GvHD with prednisone/sirolimus is an acceptable alternative and better tolerated than three-drug therapy including a CNI. Our study could not address the relative merits of prednisone/sirolimus versus prednisone ± CNI. We were not able to include a prednisone ± CNI treatment arm (current standard for initial therapy), because the study included patients with high-risk chronic GvHD or early treatment failure. Success rates for prednisone/sirolimus as initial therapy in treatment-naïve or early inadequate responders are insufficient to warrant a randomized controlled trial versus prednisone with or without CNI. For early chronic GvHD therapy, novel approaches that improve rates of complete/parital responses and failure-free survival are required. Given the inherent complexity of chronic GvHD trials, we advise real-time diagnostic checklists to ensure patients’ eligibility, and real-time data auditing to protect data integrity. Acknowledgments The authors would like to thank the National Heart, Lung, and Blood Institute and the National Cancer Institute for supporting this study (grant #U10HL069294). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. We thank the laboratory of Jerome Ritz for performing all of the immunophenotyping and measuring serum BAFF levels. We thank Saurabh Chhabra, Gregory Yanik, Richard Maziarz, Suhag Parikh, Mark Litzow, Hillard Lazarus, Marcelo Pasquini, Andrew Artz, Krishna Gundabolu, Mark Juckett, Peter Westervelt, Pablo Parker, George Selby, George Chen, John Wingard, Scott Rowley, Scott Solomon, David Porter, Carlos Bachier, Paul Shaughnessy, James Essell, Marcie Riches, Thomas Shea, Michael Pulsipher, Edward Ball, John McCarty, Samantha Jaglowski, Guenther Koehne, and John Lister for enrolling patients on this trial. We thank the members of the Blood and Marrow Transplant Clinical Trials Network, the research nurses, and the patients who participated in this trial.

Prednisone and azathioprine compared with prednisone and placebo for treatment of chronic graft-versus-host disease: prognostic influence of prolonged thrombocytopenia after allogeneic marrow transplantation. Blood. 1988;72(2):546-554. 4. Koc S, Leisenring W, Flowers ME, et al. Therapy for chronic graft-versus-host disease: a randomized trial comparing cyclosporine plus prednisone versus prednisone alone. Blood. 2002;100(1):48-51. 5. Arora M, Wagner JE, Davies SM, et al. Randomized clinical trial of thalidomide, cyclosporine, and prednisone versus cyclosporine and prednisone as initial therapy for chronic graft-versus-host disease. Biol

Blood Marrow Transplant. 2001;7(5):265273. 6. Koc S, Leisenring W, Flowers ME, et al. Thalidomide for treatment of patients with chronic graft-versus-host disease. Blood. 2000;96(12):3995-3996. 7. Martin PJ, Storer BE, Rowley SD, et al. Evaluation of mycophenolate mofetil for initial treatment of chronic graft-versus-host disease. Blood 2009;113(21):5074-5082. 8. Gilman AL, Schultz KR, Goldman FD, et al. Randomized trial of hydroxychloroquine for newly diagnosed chronic graft-versus-host disease in children: a Children’s Oncology Group study. Biol Blood Marrow Transplant. 2012;18(1):84-91.

1923

P.A. Carpenter et al. 9. Couriel DR, Saliba R, Escalon MP, et al. Sirolimus in combination with tacrolimus and corticosteroids for the treatment of resistant chronic graft-versus-host disease. Br J Haematol. 2005;130(3):409-417. 10. Couriel DR, Hosing C, Saliba R, et al. Extracorporeal photochemotherapy for the treatment of steroid-resistant chronic GVHD. Blood. 2006;107(8):3074-3080. 11. Cutler C, Miklos D, Kim HT, et al. Rituximab for steroid-refractory chronic graft-versus-host disease. Blood. 2006;108 (2):756-762 12. Lopez F, Parker P, Nademanee A, et al. Efficacy of mycophenolate mofetil in the treatment of chronic graft-versus-host disease. Biol Blood Marrow Transplant. 2005;11(4):307-313. 13. Jacobsohn DA, Chen AR, Zahurak M, et al. Phase II study of pentostatin in patients with corticosteroid-refractory chronic graft-versus-host disease. J Clin Oncol. 2007;25(27): 4255-4261. 14. Zeiser R, Nguyen VH, Beilhack A, et al. Inhibition of CD4+CD25+ regulatory T-cell function by calcineurin-dependent interleukin-2 production. Blood. 2006;108(1): 390-399. 15. Zeiser R, Leverson-Gower DB, Zambricki EA, et al. Differential impact of mammalian target of rapamycin inhibition on CD4+CD25+Foxp3+ regulatory T cells compared with conventional CD4+ T cells. Blood. 2008;111(1):453-462. 16. Coenen JJ, Koenen HJ, van Rijsen E, et al. Rapamycin, not cyclosporine, permits thymic generation and peripheral preservation of CD4+Cd25+FoxP3+ T cells. Bone Marrow Transplant. 2007;39(9):537-545. 17. Gatza E, Rogers CE, Clouthier SG, et al. Extracorporeal photopheresis reverses experimental graft-versus-host disease through regulatory T cells. Blood.

1924

2008;112(4):1515-1521. 18. Biagi E, Di Biaso I, Leoni V, et al. Extracorporeal photochemotherapy is accompanied by increasing levels of circulating CD4+CD25+GITR+Foxp3+CD62L+ functional regulatory T-cells in patients with graft-versus-host disease. Transplantation. 2007;84(1):31-39. 19. Filipovich AH, Weisdorf D, Pavletic S, et al. National Institutes of Health consensus development project on criteria for clinical trials in chronic graft-versus-host disease: I. Diagnosis and staging working group report. Biol Blood Marrow Transplant. 2005;11(12):945-956. 20. Pavletic SZ, Martin PJ, Lee SJ, et al. National Institutes of Health consensus development project on criteria for clinical trials in chronic graft-versus-host disease: IV. Response criteria working group report. Biol Blood Marrow Transplant. 2006;12(3):252-266. 21. Martin PJ, Weisdorf D, Przepiorka D, et al. National Institutes of Health consensus development project on criteria for clinical trials in chronic graft-versus-host disease: VI. Design of clinical trials working group report. Biol Blood Marrow Transplant. 2006;12(5):491-505. 22. Carpenter PA, Kitko CL, Elad S, et al. National Institutes of Health consensus development project on criteria for clinical trials in chronic graft-versus-host disease: V. The 2014 ancillary therapy and supportive care working group report. Biol Blood Marrow Transplant. 2015;21(7):1167-1187. 23. Sarantopoulos S, Stevenson KE, Kim HT, et al. Altered B-cell homeostasis and excess BAFF in human chronic graft-versus-host disease. Blood. 2009;113(6):3865-3874. 24. Ware JE, Kosinski M, Bjorner JB, et al. User's Manual for the SF-36v2 Health Survey. 2nd edn. QualityMetric Incorporated; Lincoln, RI: 2007.

25. Johnston LJ, Brown J, Shizuru JA, et al. Rapamycin (sirolimus) for treatment of chronic graft-versus-host disease. Biol Blood Marrow Transplant. 2005;11(1):47-55. 26. Lee SJ. Classification systems for chronic graft-versus-host disease. Blood. 2017;129 (1):30-37. 27. Duarte RF, Greinix H, Rabin B, et al. Uptake and use of recommendations for the diagnosis, severity scoring and management of chronic GVHD: an international survey of the EBMT-NCI Chronic GVHD Task Force. Bone Marrow Transplant. 2014;49(1):49-54. 28. h t t p s : / / c l i n i c a l t r i a l s . g o v / c t 2 / s h o w / NCT01879072 29. Jagasia MH, Greinix HT, Arora M, et al. National Institutes of Health consensus development project on criteria for clinical trials in chronic graft-versus-host disease: I. The 2014 diagnosis and staging working group report. Biol Blood Marrow Transplant. 2015;21(3):389-401. 30. Lee SJ, Wolff D, Kitko C, et al. Measuring therapeutic response in chronic graft-versushost disease. National Institutes of Health consensus development project on criteria for clinical trials in chronic graft-versus-host disease: IV. The 2014 response criteria working group report. Biol Blood Marrow Transplant. 2015;21(6):984-99. 31. Inamoto Y, Flowers ME, Sandmaier BM, et al. Failure-free survival after initial systemic treatment of chronic graft-versus-host disease. Blood. 2014;124(8):1363-1371. 32. Palmer J, Chai X, Martin PJ, et al. Failure-free survival in a prospective cohort of patients with chronic graft-versus-host disease. Haematologica. 2015;100(5):690-695. 33. Martin PJ, Storer BE, Inamoto Y, et al. An endpoint associated with clinical benefit after initial treatment of chronic graft-versus-host disease. Blood. 2017;130(3):360367.

haematologica | 2018; 103(11)

ARTICLE

Coagulation & its Disorders

N-linked glycosylation modulates the immunogenicity of recombinant human factor VIII in hemophilia A mice

Ferrata Storti Foundation

Jesse D. Lai,1 Laura L. Swystun,1 Dominique Cartier,1 Kate Nesbitt,1 Cunjie Zhang,2 Christine Hough,1 James W. Dennis2 and David Lillicrap1 1 2

Department of Pathology & Molecular Medicine, Queenâ&#x20AC;&#x2122;s University, Kingston and Lunenfeld Tanenbaum Research Institute, Mount Sinai Hospital, ON, Canada

ABSTRACT

Haematologica 2018 Volume 103(11):1925-1936

mmune responses to factor VIII remain the greatest complication in the treatment of severe hemophilia A. Recent epidemiological evidence has highlighted that recombinant factor VIII produced in baby hamster kidney cells is more immunogenic than factor VIII produced in Chinese hamster ovary cells. Glycosylation differences have been hypothesized to influence the immunogenicity of these synthetic concentrates. In two hemophilia A mouse models, baby hamster kidney cell-derived factor VIII elicited a stronger immune response compared to Chinese hamster ovary cell-derived factor VIII. Furthermore, factor VIII produced in baby hamster kidney cells exhibited accelerated clearance from circulation independent of von Willebrand factor. Lectin and mass spectrometry analysis of total N-linked glycans revealed differences in high-mannose glycans, sialylation, and the occupancy of glycan sites. Factor VIII desialylation did not influence binding to murine splenocytes or dendritic cells, nor surface co-stimulatory molecule expression. We did, however, observe increased levels of immunoglobulin M specific to baby hamster kidney-derived factor VIII in naĂŻve hemophilia A mice. De-N-glycosylation enhanced immunoglobulin M binding, suggesting that N-glycan occupancy masks epitopes. Elevated levels of immunoglobulin M and immunoglobulin G specific to baby hamster kidney-derived factor VIII were also observed in healthy individuals, and de-N-glycosylation increased immunoglobulin G binding. Collectively, our data suggest that factor VIII produced in baby hamster kidney cells is more immunogenic than that produced in Chinese hamster ovary cells, and that incomplete occupancy of N-linked glycosylation sites leads to the formation of immunoglobulin M- and immunoglobulin Gfactor VIII immune complexes that contribute to the enhanced clearance and immunogenicity in these mouse models of hemophilia A.

Introduction The immune response that develops in ~30% of severe hemophilia A (HA) patients remains the most serious complication in factor VIII (FVIII) replacement therapy. Why FVIII-neutralizing antibodies, known as inhibitors, develop in only some patients remains unclear. The type of FVIII concentrate has been proposed as one of the factors that influences the risk of FVIII immunogenicity. Three independent cohort studies have described differences among different recombinant (r) FVIII concentrates, where a 2nd generation full-length rFVIII was associated with a 1.6-fold increase in inhibitor risk compared to 3rd generation products.1-3 Most recently, a fourth retrospective analysis further reported hazard ratios of 2.81 and 1.64 for inhibitor incidence with 2nd and 3rd generation rFVIII, respectively, compared to plasma-derived FVIII.4 While these observational studies provide compelling evidence, the mechanistic basis for such findings has only been hypothesized, and not systematically examined.5 The transition from 2nd to 3rd generation rFVIII is based on the removal of human and animal proteins in the production process and final product formulation. haematologica | 2018; 103(11)

Correspondence: david.lillicrap@queensu.ca

Received: January 11, 2018. Accepted: July 9, 2018. Pre-published: July 12, 2018. doi:10.3324/haematol.2018.188219 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/11/1925 ÂŠ2018 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.

1925

J.D. Lai et al.

In the aforementioned studies, 2nd and 3rd generation products refer specifically to Kogenate FS© (Bayer) produced in baby hamster kidney cells (BHK; BHK-rFVIII), and Advate© (Shire) produced in Chinese hamster ovary cells (CHO; CHO-rFVIII). In addition, these products differ by a single amino acid (AA) at position 1241 in the FVIII B domain, aspartic acid and glutamic acid, respectively. However, the immunologic relevance of this substitution appears insignificant as AA 1241 is poorly represented in both the major histocompatibility class II-restricted peptidome of human monocyte-derived dendritic cells (DCs), and the repertoire of DR15-restricted CD4+ T-cell epitopes.6,7 Commercial formulations of BHK-rFVIII have very recently been reported to contain a higher presence of protein aggregates, which have previously been shown to be immunogenic.8,9 Another hypothesis proposes that the differential immunogenicity is attributed to post-translational modification of rFVIII, and specifically to differential glycosylation patterns between BHK- and CHO-rFVIII at the 25 potential N-linked sites.5,10 Glycans have been implicated in FVIII intracellular trafficking and folding, as well as clearance by the asialoglycoprotein receptor and Siglec5.11-14 High-mannose glycans have been hypothesized to facilitate the uptake of FVIII via the mannose receptor on DCs and macrophages, however the data have been conflicting, and the in vivo significance of this interaction is unclear.15,16 Previous non-clinical studies have reported similar, or increased, immunogenicity of BHK-rFVIII compared to CHO-rFVIII, as well as a decreased, but statistically insignificant, inhibitory antibody response to deglycosylated FVIII.17-19 No mechanistic explanation for these differences has been provided. Here, we used the previously described BHK- and CHO-rFVIII concentrates, Kogenate FS® and Advate®, respectively, and assessed their relative immunogenicities in two complementary murine models of HA. We further characterized the glycosylation profiles of each product, and evaluated their role in the development of the antiFVIII immune response in these murine models.

Methods FVIII concentrates The following human rFVIII concentrates were used: Kogenate FS® (full-length BHK-rFVIII; Bayer, Leverkusen, Germany), Advate® (full-length CHO-rFVIII; Shire, Dublin, Ireland), and Xyntha® (CHO-B-domain deleted (BDD)-rFVIII; Pfizer, New York City, NY, USA). Information on FVIII clearance, antigen/activity assays, deglycosylation, and von Willebrand factor (VWF) binding is available in the Online Supplementary Methods.

Mice Sex and littermate-matched 8-12 week old C57Bl/6 F8 exon 16 knockout (KO) mice with a human full-length F8 transgene containing an R593C point mutation (HA-R593C mice) that is transcribed, but for which FVIII protein is undetectable in plasma, were used for preliminary experiments.20 Results were extended using similarly controlled “conventional” C57Bl/6 F8 exon 16 KO mice (HA mice).21 Mice were treated by subcutaneous or tail vein intravenous injection of 6 IU (240 IU/kg; as per manufacturer’s label) of rFVIII biweekly for two weeks. Lipopolysacharride (LPS; 1 mg) was used as an adjuvant with the first FVIII infusion where indicated. HA mice were challenged with 2 IU (80 IU/kg) of rFVIII 1926

using the same regimen. Blood was collected by cardiac puncture in one-tenth volume of 3.2% sodium citrate 28 days after the first administration of FVIII. Mouse experiments were approved by the Queen’s University Animal Care Committee.

Anti-FVIII antibodies and FVIII inhibitor assays FVIII-specific immunoglobulin G (IgG) titres were quantified by enzyme-linked immunosorbent assay (ELISA) and FVIII inhibitors were measured by a 1-stage FVIII clotting assay using an automated coagulometer (Siemens BCS XP, Berlin, Germany), as previously described.22,23 Where indicated, anti-FVIII IgG was quantified using a standard curve generated using the human anti-FVIII monoclonal antibody, EL14 (provided by Dr. Jan Voorberg, Sanquin Research, Amsterdam, The Netherlands).24 Information on human sample collection is available in the Online Supplementary Methods. FVIII-specific IgM was assessed by indirect ELISA. rFVIII (1 µg/mL) was adsorbed to Nunc Maxisorp 96-well plates overnight. Samples were diluted 1:20 and incubated for 2 hrs. IgM was detected using horseradish peroxidase (HRP)-conjugated goat antimouse or anti-human IgM (Southern Biotech, Birmingham, AL, USA). Bovine serum albumin (BSA)-coated wells were used as controls. Plates were developed for 15 minutes using o-phenylenediamine (Sigma, St. Louis, MO, USA) and read at 492 nm.

Lectin binding assays rFVIII products were adsorbed to Maxisorp microtitre plates at 1 mg/mL overnight at 4°C. All products saturated binding at this concentration (Online Supplementary Figure S1). Plates were blocked with 1% BSA in phosphate buffered saline (PBS) + 0.01% Tween-20 for 1 hr, and subsequently incubated with biotinylated lectins (Vector Laboratories, Burlingame, CA, USA) for 30 min. Detection was facilitated using streptavidin-poly-HRP (ThermoFisher Scientific, Waltham, MA, USA) and developed for 5 min. Statistical analysis was performed using the Student t test.

FVIII preparation for mass spectrometry FVIII samples were desalted on ViVaspin, 50kDa MWCO (Sartorius, Goettingen, Germany) spin columns. 20 mg of protein in 400 ml of 50 mM ammonium bicarbonate was reduced in 10 mM dithiothreitol (DTT) at 60°C for 40 min then alkylated in 25 mM iodoacetamide in darkness for 30 min. The reaction was stopped by the addition of 20 mM DTT in darkness for 40 min. Trypsin at 1:50 ratio was added to the sample and incubated at 37°C overnight.

Glycopeptide analysis by liquid chromatography and tandem mass spectrometry (LC–MS/MS) Peptides were applied to a nano-HPLC Chip using an Agilent 1200 series microwell-plate autosampler interfaced with a Agilent 6550 Q-TOF MS (Agilent Technologies, Santa Clara, CA, USA). The reverse-phase nano-HPLC Chip (G4240-62002) had a 40 nL enrichment column and a 75 mm x 150 mm separation column packed with 5 mm Zorbax 300SB-C18. The mobile phase was 0.1% formic acid in water (v/v) as solvent A, and 0.1% formic acid in ACN (v/v) as solvent B. The flow rate was 0.3 mL/min with gradient schedule; 3% B (0-1 min); 3-40% B (1-90 min); 40-80% B(9095 min); 80% B (95-100 min) and 80-3% B (100-105 min). Mascot search was used to identify proteins and sequence coverage. Extracted glycopeptides were identified by Agilent Masshunter Quantitative Analysis software by the presence of hexose and N-acetylhexosamine. Glycan structures were predicted for extracted glycopeptides by GlycoMod. Glycan structure by MS/MS and occupancy of consensus N_X_S/T N-glycosylation sites were determined manually. haematologica | 2018; 103(11)

Glycosylation modulates FVIII immunogenicity

Results BHK-rFVIII is more immunogenic than CHO-rFVIII in mouse models of hemophilia A While an International Standard exists for determination of the FVIII:C potency of FVIII concentrates (and products are labeled in International Units), there is no standardized method to accurately quantify FVIII:Ag between products.25 Although there are differences in specific coagulant activity between commercial preparations (and between different lots of concentrate) (Online Supplementary Figure S2), we dosed rFVIII by the coagulant activity reported by the manufacturer so as to recapitulate clinical practice and the data reported by previous FVIII inhibitor epidemiological studies. To assess the immunogenic differences between CHO- and BHK-rFVIII, we administered rFVIII subcutaneously or intravenously to HA-R593C mice biweekly (6 IUs; 240 IU/kg per administration) for two weeks and analyzed plasma on day 28 (Figure 1A). After subcutaneous immunization, we observed a significant increase in the incidence of antiFVIII IgG antibodies in mice treated with BHK-rFVIII compared to CHO-rFVIII 28 days after the initial injection (93.75% vs. 47.3% respectively, P=0.0042; Figure 1B). Further serological analysis showed that the aFVIII IgG titre was also greater when mice were immunized with BHK-rFVIII (Figure 1C). A similar difference was observed in the incidence of all FVIII inhibitory antibodies (93.75% vs. 36.8%, P=0.0003; Figure 1D). However no significant differences were observed in the inhibitory antibody titres (Figure 1E). To account for the typical biodistribution of FVIII in the bloodstream, we administered rFVIII intravenously with the first injection containing 1 mg of lipopolysaccharide (LPS) as an adjuvant (Figure 1F). We did not observe any significant difference in the incidence of aFVIII IgG between products in these experiments (Figure 1G). However, BHK-rFVIII elicited higher titres of aFVIII IgG (Figure 1H) among FVIII-responders. Analysis of FVIII inhibitory antibodies showed no differences in incidence (Figure 1I) or inhibitor titre (Figure 1J) using this testing protocol. All mice developed FVIII-specific antibodies by day 28 (Figure 2B), and there were no differences in the titre of FVIII-specific antibodies (Figure 2C). However, mice immunized with BHK-rFVIII exhibited a higher incidence of all FVIII antibodies with an inhibitory titer >1 bethesda unit (BU; 100% vs. 54.5%; Figure 2D), but no difference in the magnitude among FVIII-responders (Figure 2E).

that there is no significant difference in the binding to murine VWF between CHO- and BHK-derived rFVIII, and that the differences observed in clearance and immunogenicity are independent of VWF and may suggest structural moieties, such as post-translational modifications, that facilitate or inhibit cellular uptake.

rFVIII produced in BHK or CHO cell lines contains significantly different glycosylation profiles Given the differences in clearance and immunogenicity observed in vivo, we next assessed rFVIII glycosylation by using a panel of lectins to detect specific, exposed, carbohydrate linkages on CHO- and BHK-rFVIII. CHO-BDDrFVIII was used as a control, as it lacks all but six of the potential N-linked sites. Multiple lectins specific for similar structures were used to confirm findings, and the median-centered optical absorbance readings were compared between products using the Student t test (Online Supplementary Table S1). These data suggest that BHKrFVIII has a higher degree of sialylation and fucosylation, and a lower presence of high-mannose glycans compared to CHO-rFVIII (Figure 4A; Principal component analysis Online Supplementary Figure S3). We further confirmed that these glycan differences were conserved across three different lots of rFVIII products (Figure 4B). Table 1. N-linked glycan structures detected across BHK-, CHO-, and CHO-BDD-rFVIII products.

BHK-rFVIII exhibits accelerated clearance that is independent of binding to murine VWF We next assessed the clearance of rFVIII from the mouse circulation. Following an intravenous infusion of either BHK-rFVIII or CHO-rFVIII at 200 IU/kg in HA mice, plasma FVIII:C was measured by chromogenic assay at the indicated time points and normalized to a 5 min post-infusion sample. Our results show a significant increase in the clearance rate of BHK-rFVIII compared to CHO-rFVIII (6.22 hrs vs. 9.53 hrs; P=0.02) (Figure 3A). Given the dominant influence of VWF on FVIII half-life, we assessed whether the rFVIII products exhibit differential binding to endogenous mouse VWF (Figure 3B). FVIIIbinding was reported as a function of the amount of FVIII:Ag at each dilution in the assay. These data suggest haematologica | 2018; 103(11)

1927

J.D. Lai et al.

By LC-MS/MS, we detected 22 of the 25 potential Nlinked Asn-X-Ser/Thr consensus sequences, and identified a total of 21 unique glycans (Table 1; for peptide coverage and glycan construction see Online Supplementary Figure S4 and Figure S5). The intensity of glyco-peptides was normalized to the area under the curve of non-glycosylated peptides relative to CHO-rFVIII. The predicted N-linked

glycan sites at Asn1001, Asn1005, and Asn1512 were not detected by our methods. At each occupied site, we observed significant heterogeneity in the glycan structures (Figure 5A). We next grouped glycan stuctures into either high-mannose, asialylated, partially sialylated, or fully sialylated. In agreement with our lectin binding data, we observed high-mannose glycans at Asn757 and Asn1300

Figure 1. Immunogenic differences between BHK-rFVIII and CHO-rFVIII in HA-R593C mice. (A) Mice were immunized subcutaneously (SC) with 6 IU (0.6 Âľg; 240 IU/kg) of either BHK- or CHO-rFVIII biweekly for 2 weeks. Blood was collected by cardiac puncture 28 days after the first infusion. Plasma was assessed for (B) the incidence and (C) titre of FVIII-specific IgG, as well as (D) the incidence of inhibitors above 1 and 5 BU and (E) the magnitude of inhibitory activity. (F) Mice were immunized intravenously (IV) with 6 IU (0.6 mg; 240 IU/kg) of either BHK- or CHO-rFVIII product twice weekly for 2 weeks. The first infusion contained 1 mg of lipopolysacharride. Plasma samples 28 days after the first infusion were assessed for (G) the incidence and (H) titre of FVIII-specific IgG. (I) Comparison of the incidence of inhibitors above 1 BU and the (J) magnitude of inhibitory activity. Horizontal lines and error bars represent the mean and SEM, respectively. Mann-Whitney U and Fisherâ&#x20AC;&#x2122;s exact test were used where appropriate. *P<0.05; **P<0.01; ***P<0.001. BHK: baby hamster kidney cells; CHO: Chinese hamster ovary cells: rFVIII: recombinant factor VIII; Ig: immunoglobulin; n.s. not significant; BU: bethesda unit.

1928

haematologica | 2018; 103(11)

Glycosylation modulates FVIII immunogenicity

CHO-rFVIII, we removed the terminal sialic acids from BHK-rFVIII using a2-3,6,8 neuraminidase for 18 hr at 37°C. Desialylation was confirmed by lectin binding assay (Online Supplementary Figure S7A). Desialylated FVIII:C procoagulant function was 62.3% of the control, however, FVIII antigenicity was conserved (Online Supplementary Figure S7B,C). The removal of sialic acid from BHK-rFVIII did not influence the binding to splenocytes or CD11c+ dendritic cells (Figure 6A,B). Considering the potential immunomodulatory role of sialic acid, we next investigated whether different glycoforms of rFVIII could influence the expression of the co-stimulatory molecules, CD80 and CD86, on naïve or LPS-stimulated splenocytes and DCs. While we observed significant upregulation of both CD80 and CD86 in LPS-stimulated conditions, the removal of sialic acid did not influence surface co-stimulatory molecule expression in either splenocytes or DCs (Figure 6C, D). Similarly, HA-R593C mice treated subcutaneously with BHK-rFVIII or its desialylated glycoform, as described above, did not exhibit significant differences in the incidence or the titre of FVIIIspecific IgG (Figure 6E, F).

in CHO-rFVIII but not in BHK-rFVIII (Figure 5B). Similarly, the sialylation of sub-terminal galactose residues is more complete in BHK-rFVIII as evidenced by a high proportion of fully-sialylated glycans (Figure 5C-E). We did not observe differences in the frequency of fucosylated Nlinked glycans between products (data not shown).

Glycosylation of rFVIII does not influence binding to or induction of IFNγ production by splenocytes or splenic dendritic cells in vitro Glycosylation differences between CHO- and BHKrFVIII may alter the association of FVIII with different cell types or receptors. To address this, we employed a FVIII binding assay to unsorted splenocytes from rFVIII naïve HA or HA-R593C mice. However, we did not observe significant differences in the binding of the different rFVIII preparations to these cells (Online Supplementary Figure S6A,B). We further assessed downstream immune responses to rFVIII in naïve mixed lymphocyte populations using an interferon (IFN)γ enzyme-linked immunospot (ELISPOT). Splenocytes were isolated from naïve HA mice or HA-R593C mice and stimulated with the two rFVIII proteins for 48 hr. The number of cells secreting IFNγ did not differ between the two rFVIII concentrates (Online Supplementary Figure S6C). Since these data suggest that the early rFVIII innate immune responses are similar between HA mice and HA-R593C mice, we used HA-R593C mice for subsequent experiments.

N-linked glycans prevent binding of non-neutralizing IgM and IgG to rFVIII We next evaluated the interaction of non-neutralizing IgM and IgG on rFVIII, and the ability of N-linked glycans to regulate this interaction. We found that incubation of naïve HA, or HA-R593C mouse plasma on a FVIII-coated microtitre plate resulted in the increased binding of BHKrFVIII-specific IgM (Figure 7A,B) relative to CHO-rFVIII. FVIII-specific IgG was not detected (data not shown). Of note, competition through preincubation with other forms of FVIII showed that these IgM molecules are spe-

Differential FVIII immunogenicity in hemophilia A mouse models is not explained by differences in sialic acid content Given that the degree of N-linked sialylation was the greatest source of glycan variation between BHK- and

Figure 2. Immunogenic differences between BHK- and CHO-rFVIII in HA mice. (A) F8 exon 16 KO HA mice were immunized intravenously (IV) with 6 IU (0.6 mg; 240 IU/kg) of either BHK- or CHO-rFVIII biweekly for 2 weeks. Blood was collected by cardiac puncture 28 days after the first infusion, and plasma was isolated by centrifugation. Samples were assessed for (B) incidence and (C) titres of FVIII-specific IgG. FVIII-responders were analyzed for (D) the incidence (P=0.015) and (E) concentration of FVIII inhibitors. The horizontal lines and error bars represent the mean and SEM. Mann-Whitney U test and Fisher’s exact test were used where appropriate. *P<0.05. BHK: baby hamster kidney cells; CHO: Chinese hamster ovary cells: rFVIII: recombinant factor VIII; Ig: immunoglobulin; n.s. not significant; BU: bethesda unit.

haematologica | 2018; 103(11)

1929

J.D. Lai et al.

cific to BHK-rFVIII, and that the interaction is not inhibited by human VWF (Figure 7C). We next evaluated whether N-linked glycans sterically hinder access of IgM to certain epitopes in the FVIII protein backbone. From our MS data, we observed that there were consistently higher proportions of unoccupied Nglycan sites at Asn900, Asn1255, Asn1259, Asn1282, Asn1300, and Asn1810 in BHK-rFVIII compared to CHOrFVIII (Figure 7D). We subsequently removed all the Nlinked glycans from BHK- and CHO-rFVIII using Peptide: N-Glycosidase F (PNGase F), and observed a greater increase in IgM binding to deglycosylated CHO-rFVIII (Figure 7E) when compared to the native rFVIII glycoform. To extend these findings to humans, we collected plasma from healthy human volunteers and quantified levels of FVIII-specific IgM. We observed an increased binding of IgM to BHK-rFVIII compared to CHO-rFVIII and CHOBDD-rFVIII (Figure 7F). The presence of FVIII-specific IgG antibodies has previously been reported in healthy individuals.26,27 We generated a standard curve using the recombinant antibody, EL14, which bound equally to rFVIII products (Online Supplementary Figure S8). Upon incubation with plasma from healthy human subjects, we observed that a higher proportion of human IgG bound to BHK-rFVIII compared to CHO-rFVIII (Figure 7G). We further determined that this binding was enhanced when the N-glycans were removed from rFVIII (Figure 7H).

P=0.02

Discussion Factor VIII immunogenicity remains a significant concern among hemophilia treaters, and mounting evidence suggests that rFVIII products differ from pdFVIII concentrates as well as among themselves. Non-human cell lines have been shown to add immunogenic non-human glycan structures, Gal(a1-3)GalÎ˛1-GlcNAc-R (aGal) and 5-glycolylneuraminic acid (Neu5Gc) to rFVIII.28,29 In two murine models of HA, in which both non-human glycans are present, we found that BHK-rFVIII was more immunogenic than CHO-rFVIII. These data suggest that although Neu5Gc and aGal have the potential to induce FVIII immunity in humans, as seen with the immune responses against cetuximab, there are additional mechanisms that contribute to this response.30,31 As per its routine clinical use, mice were dosed by the procoagulant activity of rFVIII. However, the level of inactive FVIII protein in the commercial concentrates likely plays a role in immunogenicity. BHK-rFVIII has been reported to have higher FVIII:C and FVIII:Ag than advertised.32 We were unable to observe a similar trend across four lots of rFVIII, perhaps due to different methodologies used. High-dose intensive FVIII treatment has been implicated as a risk factor for inhibitor development, however this correlation is likely facilitated by the inflammatory milieu of concurrent surgery or bleeding.33 Whether expo-

Figure 3. Characterization of FVIII half-life and VWF binding in hemophilia A mice. (A) HA mice were intravenously infused with 200 IU/kg of either BHK-rFVIII or CHO cell line-derived rFVIII. FVIII activity was measured by chromogenic assay, and normalized to a 5 min time point post-injection. Data are representative of the aggregate data of 3 independent experiments totalling at least 5 biological replicates per time point and 2 lots of each rFVIII product. Error bars represent SEM. (B) Murine VWF from HA mouse pooled plasma was captured on a microtitre plate coated with a polyclonal rabbit anti-human VWF antibody. Increasing concentrations of either BHK-derived, or CHO-derived rFVIII were incubated with the captured VWF and detected using a polyclonal sheep anti-human FVIII antibody. Data shown is one technical replicate representative of 3 independent experiments. BHK: baby hamster kidney cells; CHO: Chinese hamster ovary cells; rFVIII: recombinant factor VIII.

1930

haematologica | 2018; 103(11)

Glycosylation modulates FVIII immunogenicity

sure to greater amounts of FVIII protein accounts for the increase in immunogenicity in a non-inflammatory steady state is unclear. Consistent with our data, Delignat et al. standardized FVIII:Ag between products using human plasma, and observed enhanced immunogenicity of BHKrFVIII compared to CHO-rFVIII in HA mice.18 The immunogenic disparity between rFVIII concentrates was greatest when administered subcutaneously in HA-R593C mice. In F8 exon 16 KO HA mice, central tolerance for human FVIII is limited, which may explain why differences between very similar proteins are not as apparent in this mouse model.34 In humanized F8 exon 16 KO HA-R593C mice, immunological tolerance to human FVIII necessitates an adjuvant to elicit an immune response via intravenous administration, which may result in a general heightened immune reactivity against all antigens, thus preventing the resolution of subtle immunogenic differences.20 Of note, there are major differences in the biodistribution of FVIII when administered intravenously, where it complexes with murine VWF and predominantly localizes in the liver and spleen, versus subcutaneous delivery, where VWF association is less likely, and FVIII localizes to the draining lymph node.35 This likely directs FVIII to different populations of phagocytic cells, particularly DCs and macrophages.36 The removal of FVIII from circulation can lead to clearance and/or antigen presentation. Even in the dominating

presence of endogenous murine VWF, BHK-rFVIII was cleared at an accelerated rate, independent of differences in VWF-binding under static conditions. This difference in half-life has not been described in human patients, and may only be apparent in this mouse model due to interspecies differences in the mechanisms of FVIII and VWF clearance. In this scenario, assuming that FVIII does not influence VWF clearance to a significant extent, it is possible that the ~5% of BHK-rFVIII that circulates without VWF is cleared faster, thus shifting the equilibrium to promote further FVIII dissociation from VWF.37 We hypothesized that post-translational structures could contribute to the different clearance kinetics of these proteins in mice. Our analysis of rFVIII glycans confirms the findings of previous studies: that rFVIII possesses predominantly core-fucosylated biantennary complex glycans, a heavily sialylated B domain, high mannose glycans at Asn239 and Asn2118, and unoccupied sites at Asn582, Asn943, Asn1384, and Asn1685.28,38,39 Contrary to previous studies, we did not detect tetraantennary glycans or Neu5Gc glycans (aGal cannot be detected directly using our methods). Lectin binding analysis of exposed N- and O-linked glycans and mass spectrometry analysis of total N-linked glycans collectively showed differences in sialic acid and high-mannose glycan content. A similar analysis showed an absence of high mannose glycans as well as greater sialylation in CHO-derived rFVII compared to that produced

Figure 4. Lectin array analysis of exposed glycans on rFVIII products. rFVIII products were adsorbed on microtitre plates at 1 mg/mL and assessed using a panel of biotinylated lectins. BSA was adsorbed as a control. Binding was detected using a streptavidin poly-HRP and read at 492 nm. (A) Heat plot demonstrating the median-centred binding values of lectins with varying carbohydrate specificites to different rFVIII products. Values shown are representative of the mean of at least 3 independent experiments on a single lot of each rFVIII product. Statistical summary available in Online Supplementary Table S1. (B) Lectin binding analysis of different production lots of full-length rFVIII from CHO or BHK cells. Data are representative of at least 3 independent experiments. BHK: baby hamster kidney cells; CHO: Chinese hamster ovary cells: rFVIII: recombinant factor VIII; BSA: bovine serum albumin; BDD: B-domain deleted.

haematologica | 2018; 103(11)

1931

J.D. Lai et al.

plex glycan formation has been shown to increase the specific activity of FVIII, suggesting that the enhanced negative charge due to increased sialic acid may alter the affinities between BHK-rFVIII and its interacting coagulant partners.43 While the influence of the FVIII procoagulant activity on immunogenicity remains a subject of debate, the increased negative charge associated with enhanced sialylation may modulate clearance receptor binding.44,45 Sialic acid can signal through inhibitory receptors such as Siglec-5, or the activating homolog Siglec-14, which have nearly identical binding motifs.14,46 In our study, we

in BHK cells, suggesting that differences in glycan profiles may be protein-specific.40 Given the conflicting evidence relating to the role of mannosylation on FVIII immunity, and the high abundance of endogenous high-mannose glycans on other plasma proteins that do not exhibit similar immunogenic responses, we assessed sialic acid as a regulator of FVIII immunogenicity.15,16,41 N-glycans on BHK-rFVIII exhibited greater sialylation, which considering our clearance data, is in contrast to previously reported influences of sialic acid on FVIII and VWF half-life.42 The inhibition of com-

Figure 5. Comparative N-linked glycosylation of different rFVIII products by LC-MS/MS. (A) Heat plot representing the proportion of each of the detected glycans at each N-linked consensus sequence across different rFVIII products. Sites portrayed with an X were not detected in the analyses. Glycan structures are denoted as additions to the core (Man3GlcNAc2-Asn) where H=hexose, N=GlcNAc, F=fucose, S=sialic acid. (B) High-mannose, (C) asialylated, (D) partially sialylated, and (E) fully sialylated N-linked glycans presented as a proportion of the total occupied glycans. Amino acid numbering for N-linked Asn glycosylation sites does not include the 19 amino acid signal peptide. Each rFVIII product was analyzed at least twice, including an initial assessment of sample quality. BHK: baby hamster kidney cells; CHO: Chinese hamster ovary cells; rFVIII: recombinant factor VIII; BDD: B-domain deleted.

1932

haematologica | 2018; 103(11)

Glycosylation modulates FVIII immunogenicity

observed no difference in the binding of BHK- or CHOrFVIII to naïve splenocytes or DCs. Moreover, the removal of a2-3, and a2-6 sialic acids did not significantly influence cellular binding or modulation of surface co-stimulatory molecule responses in vitro, and did not influence the immunogenicity of BHK-rFVIII in mice. In fact, the FVIII immune response was attenuated when exposed to desialylated BHK-rFVIII compared to exposure to unmodified rFVIII, potentially due to desialylated BHK-rFVIII behaving as a unique antigen.47 Although the antigenicity of desialylated rFVIII was not altered as determined by

ELISA, the decrease in FVIII:C activity suggests an alteration of global tertiary structure that may influence FVIII immunogenicity. Interactions between FVIII and its plasma binding partners may regulate its association with endocytic cells in the liver and spleen. We observed increased binding to BHK-rFVIII by non-neutralizing IgM in the plasma of naïve HA and HA-R593C mice as well as by non-neutralizing IgM and IgG from healthy human subjects. The presence of mouse and human VWF in these experiments suggests that these immune complexes can form in the cir-

haematologica | 2018; 103(11)

Figure 6. Sialic acid does not alter FVIII binding to, and maturation of splenocytes and dendritic cells in vitro, and does not modulate the FVIII immune response in vivo. Desialylated and control rFVIII were incubated with FVIII naïve splenocytes for 1 hr at 4°C and assessed for binding by flow cytometry. Cells were gated as total (A) splenocytes or (B) CD11c+ dendritic cells (DCs). rFVIII association with DCs was quantified as the proportion of FVIII+CD11c+ out of all CD11c+ cells. Flow cytometry characterization of the surface co-stimulatory molecules, CD80 and CD86, were performed on (C) splenocytes and (D) CD11c+ DCs incubated for 24 hr under the indicated conditions (1 mg/mL LPS or rFVIII; n=4). HA-R593C mice were administered 1 mg of rFVIII and its desialylated glycoform subcutaneously biweekly for 2 weeks. Plasma samples 28 days after the initial infusion were assessed for (E) the incidence of FVIII-specific IgG and (F) the titre of the anti-FVIII IgG response. Line and error bars represent mean and SEM respectively. Data representative of at least 3 independent deglycosylation reactions. **P<0.01; ***P<0.001. MFI: mean fluorescence intensity; rFVIII: recombinant factor VIII; Ig: immunoglobulin; n.s. not significant.

1933

J.D. Lai et al. A

Figure 7. FVIII-specific IgM and IgG in naïve HA mice and healthy humans bind preferentially to BHK-rFVIII. FVIII-specific IgM was detected by indirect ELISA using plasma from naïve (A) HA-R593C mice and (B) HA mice. (C) Competition assay in which the specificity of these IgM antibodies for BHK-rFVIII was assessed by a 30 min preincubation with either other rFVIII concentrates, or human pdVWF (n=3). (D) Glycan occupancy at each N-linked Asn consensus sequence derived from LCMS/MS analysis. (E) Murine IgM binding to deglycosylated rFVIII products (n=7). (F) FVIII-binding of IgM isolated from healthy humans. (G) Binding of IgG from healthy human plasma to different rFVIII products. (H) Binding of IgG from healthy human plasma to de-N-glycosylated BHK-rFVIII. Line and error bars represent mean and SEM respectively. The Wilcoxon test was used for paired statistical analysis. *P<0.05; **P<0.01; ***P<0.001. BHK: baby hamster kidney cells; CHO: Chinese hamster ovary cells; rFVIII: recombinant factor VIII; BDD: B-domain deleted; Ig: immunoglobulin; PNGaseF: Peptide: N-Glycosidase F.

culation in the presence of the protective effect of VWF.48 N-linked glycans may result in steric hindrance to prevent Ig binding. Indeed, removal of N-linked glycans from BHK- and CHO-rFVIII increased IgM binding to a greater extent in the latter, suggesting that glycans on CHO-rFVIII better mask underlying epitopes. These rFVIII immune complexes may provide an explanation for the enhanced immunogenicity documented with BHK-rFVIII through fragment crystallizable (Fc)mediated uptake by DCs.49 The binding of IgM to antigens further facilitates the binding of mannose-binding lectin, and both can trigger the deposition of complement that greatly increases the binding potential of FVIII to endocytic receptors.50 These data therefore support the observation of increased clearance and immunogenicity of BHKrFVIII. The presence of FVIII-specific IgM in rFVIII naïve HA mice, and both IgM and IgG in healthy individuals, suggests an innately immunogenic property of FVIII. 1934

Previous studies have reported anti-FVIII antibodies in up to 19% of healthy subjects.26,27,51 These autoantibodies have been mapped to several regions in the FVIII heavy chain, and a single region in the A3 domain of the FVIII light chain.51 Complete glycosylation of the partially occupied sites described herein may inhibit these initial immune complexes from forming. Although these antirFVIII Igs likely possess low binding affinities, the avidity of IgM binding, in addition to IgG binding, may compensate and contribute to cellular uptake of FVIII immune complexes leading to either clearance or antigen presentation.52,53 In this study, our data suggest that N-linked glycans shield underlying FVIII epitopes. We propose that the increased immunogenicity of BHK-rFVIII shown in two murine models of hemophilia A and four separate epidemiological studies is, in part, related to incomplete Nlinked glycosylation that exposes immunogenic epitopes haematologica | 2018; 103(11)

Glycosylation modulates FVIII immunogenicity

to FVIII-specific IgM and IgG, that may, in turn, facilitate the formation of immune complexes in the circulation. Collectively, our studies provide an additional biological complement to evidence presented in recent epidemiological investigations showing that 2nd generation BHKrFVIII is more immunogenic than 3rd generation CHOrFVIII.

References 1. Gouw SC, van der Bom JG, Ljung R, et al. Factor VIII products and inhibitor development in severe hemophilia A. N Engl J Med. 2013;368(3):231-239. 2. Collins PW, Palmer BP, Chalmers E a, et al. Factor VIII brand and the incidence of factor VIII inhibitors in previously untreated UK children with severe hemophilia A, 2000-2011. Blood. 2014;1 24 (23): 33893397. 3. Calvez T, Virginie H, Milien V, Rothschild C. Recombinant factor VIII products and inhibitor development in previously untreated boys with severe hemophilia A. Blood. 2014;124(23):3398-3409. 4. Calvez T, Chambost H, D’Oiron R, et al. Analyses of the FranceCoag cohort support differences in immunogenicity among one plasma-derived and two recombinant factor VIII brands in boys with severe hemophilia A. Haematologica. 2018; 103(1):179189. 5. Lai J, Hough C, Tarrant J, Lillicrap D. Biological considerations of plasma-derived and recombinant factor VIII immunogenicity. Blood. 2017;129(24):3147-3154. 6. van Haren SD, Herczenik E, ten Brinke A, Mertens K, Voorberg J, Meijer AB. HLADR-presented peptide repertoires derived from human monocyte-derived dendritic cells pulsed with blood coagulation factor VIII. Mol Cell Proteomics. 2011; 10(6):M110.002246. 7. Steinitz KN, van Helden PM, Binder B, et al. CD4+ T-cell epitopes associated with antibody responses after intravenously and subcutaneously applied human FVIII in humanized hemophilic E17 HLADRB1*1501 mice. Blood. 2012; 119(17):4073-4082. 8. Healey JF, Parker ET, Lollar P. Identification of aggregates in therapeutic formulations of recombinant full-length factor VIII products by sedimentation velocity analytical ultracentrifugation. J Thromb Haemost. 2018; 16(2):303-315. 9. Pisal DS, Kosloski MP, Middaugh CR, Bankert RB, Balu-iyer S V. Native-like aggregates of factor VIII are immunogenic in von Willebrand Factor deficient and hemophilia A mice.. J Pharm Sci. 2012; 101(6):2055-2065. 10. Vehar G, Keyt B, Eaton D, Rodriguez H. Structure of human factor VIII. Nature. 1984;312(5992):337-342. 11. Moussalli M, Pipe SW, Hauri HP, Nichols WC, Ginsburg D, Kaufman RJ. Mannosedependent endoplasmic reticulum (ER)Golgi intermediate compartment-53-mediated ER to Golgi trafficking of coagulation factors V and VIII. J Biol Chem. 1999; 274(46):32539-32542. 12. Pipe SW, Morris J a, Shah J, Kaufman RJ. Differential Interaction of Coagulation Factor VIII and Factor V with Protein Chaperones Calnexin and Calreticulin. J

haematologica | 2018; 103(11)

Funding This work was supported by an Operating and Foundation Grant from the Canadian Institutes of Health Research (MOP-10912, FDN-154285), GlycoNet, and from the Canadian Hemophilia Society. JDL is supported in part by an Ontario Graduate Scholarship and the Franklin Bracken Fellowship. and DL is the recipient of a Canada Research Chair in Molecular Hemostasis.

Biol Chem. 1998;273(14):8537-8544. 13. Bovenschen N, Rijken DC, Havekes LM, van Vlijmen BJM, Mertens K. The B domain of coagulation factor VIII interacts with the asialoglycoprotein receptor. J Thromb Haemost. 2005;3(6):1257-1265. 14. Pegon JN, Kurdi M, Casari C, et al. Factor VIII and von Willebrand factor are ligands for the carbohydrate-receptor Siglec-5. Haematologica. 2012;97(12):1855-1863. 15. Dasgupta S, Navarrete A-M, Bayry J, et al. A role for exposed mannosylations in presentation of human therapeutic self-proteins to CD4+ T lymphocytes. Proc Natl Acad Sci USA. 2007;104(21):8965-8970. 16. Herczenik E, van Haren SD, Wroblewska A, et al. Uptake of blood coagulation factor VIII by dendritic cells is mediated via its C1 domain. J Allergy Clin Immunol. 2012; 129(2):501-509.e5. 17. Delignat S, Dasgupta S, Andre S, et al. Comparison of the immunogenicity of different therapeutic preparations of human factor VIII in the murine model of hemophilia A. Haematologica. 2007;92(10):14231426. 18. Delignat S, Repessé Y, Gilardin L, et al. Predictive immunogenicity of Refacto AF. Haemophilia. 2014;20(4):486-492. 19. Kosloski MP, Miclea RD, Balu-Iyer S V. Role of glycosylation in conformational stability, activity, macromolecular interaction and immunogenicity of recombinant human factor VIII. AAPS J. 2009;11(3):424-431. 20. Bril WS, van Helden PMW, Hausl C, et al. Tolerance to factor VIII in a transgenic mouse expressing human factor VIII cDNA carrying an Arg 593 to Cys substitution. Thromb Haemost. 2006;95(2):341-347. 21. Bi L, Lawler AM, Antonarakis SE, High KA, Gearhart JD, Kazazian HH. Targeted disruption of the mouse factor VIII gene produces a model of haemophilia A. Nat Genet. 1995;10(1):119-121. 22. Hausl C, Maier E, Schwarz HP, et al. Longterm persistence of anti-factor VIII antibody-secreting cells in hemophilic mice after treatment with human factor VIII. Thromb Haemost. 2002;87(5):840-845. 23. Lai JD, Moorehead PC, Sponagle K, et al. Concurrent influenza vaccination reduces anti-FVIII antibody responses in murine hemophilia A. Blood. 2016;1 27(26):34393449. 24. van Den Brink EN, Turenhout E a, Davies J, et al. Human antibodies with specificity for the C2 domain of factor VIII are derived from VH1 germline genes. Blood. 2000; 95(2):558-563. 25. Raut S, Bevan S, Hubbard AR, Sands D, Barrowcliffe TW. A collaborative study to establish the 7th International Standard for Factor VIII Concentrate. J Thromb Haemost. 2005;3(1):119-126. 26. Algiman M, Dietrich G, Nydegger UE, Boieldieu D, Sultan Y, Kazatchkine MD. Natural antibodies to factor VIII (antihemophilic factor) in healthy individuals.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36. 37.

38.

39.

Proc Natl Acad Sci USA. 1992; 89(9):37953799. Whelan SFJ, Hofbauer CJ, Horling FM, et al. Distinct characteristics of antibody responses against factor VIII in healthy individuals and in different cohorts of hemophilia A patients. Blood. 2013; 121(6):1039-1048. Hironaka T, Furukawa K, Esmon PC, et al. Comparative study of the sugar chains of factor VIII purified from human plasma and from the culture media of recombinant baby hamster kidney cells. J Biol Chem. 1992;267(12):8012-8020. Kannicht C, Ramstrom M, Kohla G, et al. Characterisation of the post-translational modifications of a novel, human cell linederived recombinant human factor VIII. Thromb Res. 2013;131(1):78-88. Ghaderi D, Taylor RE, Padler-Karavani V, Diaz S, Varki A. Implications of the presence of N-glycolylneuraminic acid in recombinant therapeutic glycoproteins. Nat Biotechnol. 2010;28(8):863-867. Chung CH, Mirakhur B, Chan E, et al. Cetuximab-induced anaphylaxis and IgE specific for galactose-alpha-1,3-galactose. N Engl J Med. 2008;358(11):1109-1117. Pahl S, Pavlova A, Driesen J, Müller J, Pötzsch B, Oldenburg J. In vitro characterization of recombinant factor VIII concentrates reveals significant differences in protein content, activity and thrombin activation profile. Haemophilia. 2013; 19(3):392398. Gouw SC, van den Berg HM, Fischer K, et al. Intensity of factor VIII treatment and inhibitor development in children with severe hemophilia A: the RODIN study. Blood. 2013;121(20):4046-4055. Qadura M, Waters B, Burnett E, et al. Recombinant and plasma-derived factor VIII products induce distinct splenic cytokine microenvironments in hemophilia A mice. Blood. 2009;114(4):871-880. Navarrete A, Dasgupta S, Delignat S, et al. Splenic marginal zone antigen-presenting cells are critical for the primary alloimmune response to therapeutic factor VIII in hemophilia A. J Thromb Haemost. 2009; 7(11):1816-1823. Henri S, Vremec D, Kamath A, et al. The dendritic cell populations of mouse lymph nodes. J Immunol. 2001;167(2):741-748. Noe DA. A Mathematical Model of Coagulation Factor VIII Kinetics. Pathophysiol Haemost Thromb 1996; 26(6):289-303. Medzihradszky KF, Besman MJ, Burlingame a L. Structural characterization of site-specific N-glycosylation of recombinant human factor VIII by reversed-phase high-performance liquid chromatographyelectrospray ionization mass spectrometry. Anal Chem. 1997;69(19):3986-3994. Lenting PJ, Pegon JN, Christophe OD, Denis C V. Factor VIII and von Willebrand factor--too sweet for their own good.

1935

J.D. Lai et al. Haemophilia. 2010;16 Suppl 5194-5199. 40. Bรถhm E, Seyfried BK, Dockal M, et al. Differences in N-glycosylation of recombinant human coagulation factor VII derived from BHK, CHO, and HEK293 cells. BMC Biotechnol. 2015;15:87. 41. Clerc F, Reiding KR, Jansen BC, Kammeijer GSM, Bondt A, Wuhrer M. Human plasma protein N-glycosylation. Glycoconj J. 2016;33(3):309-343. 42. Sodetz JM, Pizzo SV, McKee PA. Relationship of sialic acid to function and in vivo survival of human factor VIII/von Willebrand factor protein. J Biol Chem. 1977;252(15):5538-5546. 43. Pittman DD, Tomkinson KN, Kaufman RJ. Post-translational requirements for functional factor V and factor VIII secretion in mammalian cells. J Biol Chem. 1994;269(25):17329-17337. 44. Skupsky J, Zhang AH, Su Y, Scott DW. A role for thrombin in the initiation of the immune response to therapeutic factor VIII.

1936

Blood. 2009;114(21):4741-4748. 45. Meeks SL, Cox CL, Healey JF, et al. A major determinant of the immunogenicity of factor VIII in a murine model is independent of its procoagulant function. Blood. 2012; 120(12):2512-2520. 46. Ali SR, Fong JJ, Carlin AF, et al. Siglec-5 and Siglec-14 are polymorphic paired receptors that modulate neutrophil and amnion signaling responses to group B Streptococcus. J Exp Med. 2014;211(6):1231-1242. 47. Purohit VS, Middaugh CR, Balasubramanian S V. Influence of aggregation on immunogenicity of recombinant human Factor VIII in hemophilia A mice. J Pharm Sci. 2006;95(2):358-371. 48. Dasgupta S, Repesse Y, Bayry J, et al. VWF protects FVIII from endocytosis by dendritic cells and subsequent presentation to immune effectors. Blood. 2007; 109(2):610612. 49. Hartholt RB, Wroblewska A, Herczenik E, et al. Enhanced uptake of blood coagulation

50.

51.

52.

53.

factor VIII containing immune complexes by antigen presenting cells. J Thromb Haemost. 2017;15(2):329-340. McMullen ME, Hart ML, Walsh MC, Buras J, Takahashi K, Stahl GL. Mannose-binding lectin binds IgM to activate the lectin complement pathway in vitro and in vivo. Immunobiology. 2006;211(10):759-766. Gilles JG, Saint-Remy JMR. Healthy subjects produce both anti-Factor VIII and specific anti-idiotypic antibodies. J Clin Invest. 1994;94(4):1496-1505. Hofbauer CJ, Whelan SFJ, Hirschler M, et al. Affinity of FVIII-specific antibodies reveals major differences between neutralizing and nonneutralizing antibodies in humans. Blood. 2014;125(7):1180-1188. Hofbauer CJ, Kepa S, Schemper M, et al. FVIII-binding IgG modulates FVIII half-life in patients with severe and moderate hemophilia A without inhibitors. Blood. 2016;128(2):293-296.

haematologica | 2018; 103(11)