Page 1


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

Editor-in-Chief Jan Cools (Leuven)

Deputy Editor Luca Malcovati (Pavia)

Managing Director Antonio Majocchi (Pavia)

Associate Editors Hélène Cavé (Paris), Ross Levine (New York), Claire Harrison (London), Pavan Reddy (Ann Arbor), Andreas Rosenwald (Wuerzburg), Juerg Schwaller (Basel), Monika Engelhardt (Freiburg), Wyndham Wilson (Bethesda), Paul Kyrle (Vienna), Paolo Ghia (Milan), Swee Lay Thein (Bethesda), Pieter Sonneveld (Rotterdam)

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

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

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

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


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

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

Institutional Euro 500

Personal Euro 150

Advertisements. Contact the Advertising Manager, Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, e-mail: marketing@haematologica.org). Disclaimer. Whilst every effort is made by the publishers and the editorial board to see that no inaccurate or misleading data, opinion or statement appears in this journal, they wish to make it clear that the data and opinions appearing in the articles or advertisements herein are the responsibility of the contributor or advisor concerned. Accordingly, the publisher, the editorial board and their respective employees, officers and agents accept no liability whatsoever for the consequences of any inaccurate or misleading data, opinion or statement. Whilst all due care is taken to ensure that drug doses and other quantities are presented accurately, readers are advised that new methods and techniques involving drug usage, and described within this journal, should only be followed in conjunction with the drug manufacturer’s own published literature. Direttore responsabile: Prof. Edoardo Ascari; Autorizzazione del Tribunale di Pavia n. 63 del 5 marzo 1955. Printing: Tipografia PI-ME, via Vigentina 136, Pavia, Italy. Printed in Agosto 2017.


haematologica calendar of events

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

2nd EBMT International Transplant Course The European Group for Blood and Marrow Transplantation (EBMT) Chairs: M Mohty, J Kuball, R Duarte September 8-10, 2017 Barcelona, Spain

Highlights of Past EHA - HOPE Cairo 2017 EHA Chairs: A Kamel, A El-Beshlawy, J Gribben, M Qari September 14-15, 2017 Cairo, Egypt

3 ESH International Conference on New Concepts in Lymphoid Malignancies: Focus on CLL and Indolent Lymphoma European School of Haematology (ESH) Chairs: M Hallek, L Staudt, S Stilgenbauer, A ThomasTikhonenko September 15-17, 2017 Mandelieu, France rd

13th Educational Course of the Lymphoma Working Party on "Treatment of Malignant Lymphoma: State-of-the-Art and Role of Stem Cell Transplantation" The European Group for Blood and Marrow Transplantation (EBMT) Chairs: S Montoto, A Sureda, M Trneny September 21-22, 2017 Prague, Czech Republic

9th International Nurses Study Day EBMT - Nurses Group October 5, 2017 Manchester, UK

ESH 4th International Conference on Acute Myeloid Leukemia "Molecular and Translational": Advances in Biology and Treatment European School of Haematology (ESH) Chairs: B Löwenberg, H Döhner, M Tallman October 5-7, 2017 Estoril, Portugal

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

Crash course on diagnosis and treatment of non-infectious complications after HCT EMBT – Complications and Quality of Life Working Party October 19-20, 2017 Granada, Spain

EHA Tutorial on Biology and Management of Myeloid Malignancies October 20-21, 2017 Yerevan, Armenia

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

The 4th International Congress on Controversies in Stem Cell Transplantation and Cellular Therapies COSTEM Chairs: N Kröger, A Nagler October 27-29, 2017 Berlin, Germany

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

28th Congress of the Hellenic Society of Haematology Hellenic Society of Haematology Chairs: P Panayotidis, E Terpos November 2-4, 2017 Athina, Greece

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

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

Calendar of Events updated on July 21, 2017


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

Table of Contents Volume 102, Issue 9: September 2017 Cover Figure Image generated by www.somersault1824.com.

Editorials 1467

Modeling mixed-lineage-rearranged leukemia initiation in CD34+ cells: a "CRISPR" solution RaĂşl Torres-Ruiz et al.

Review Article 1469

Chronic lymphocytic leukemia cells are active participants in microenvironmental cross-talk Martijn HA van Attekum et al.

Articles Red Cell Biology & its Disorders

1477

Hemoglobin concentration, total hemoglobin mass and plasma volume in patients: implications for anemia James M. Otto et al.

Coagulation & its Disorders

1486

Long-term impact of joint bleeds in von Willebrand disease: a nested case-control study Karin P.M. van Galen et al.

1494

Comparison of risk prediction scores for venous thromboembolism in cancer patients: a prospective cohort study Nick van Es et al.

Myelodysplastic syndrome

1502

Molecular analysis of myelodysplastic syndrome with isolated deletion of the long arm of chromosome 5 reveals a specific spectrum of molecular mutations with prognostic impact: a study on 123 patients and 27 genes Manja Meggendorfer et al.

Myeloproliferative Disorders

1511

Characteristics and clinical significance of cytogenetic abnormalities in polycythemia vera Guilin Tang et al.

Chronic Myeloid Leukemia

1518

Combined targeting of STAT3 and STAT5: a novel approach to overcome drug resistance in chronic myeloid leukemia Karoline V. Gleixner et al.

1530

Incidence of second primary malignancies and related mortality in patients with imatinib-treated chronic myeloid leukemia Gabriele Gugliotta et al.

Acute Myeloid Leukemia

1537

High-throughput profiling of signaling networks identifies mechanism-based combination therapy to eliminate microenvironmental resistance in acute myeloid leukemia Zhihong Zeng et al.

Haematologica 2017; vol. 102 no. 9 - September 2017 http://www.haematologica.org/


haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation 1549

Ultrasensitive detection of acute myeloid leukemia minimal residual disease using single molecule molecular inversion probes Adam Waalkes et al.

1558

CRISPR-Cas9-induced t(11;19)/MLL-ENL translocations initiate leukemia in human hematopoietic progenitor cells in vivo Jana Reimer et al.

1567

Reduced hematopoietic stem cell frequency predicts outcome in acute myeloid leukemia Wenwen Wang et al.

Acute Lymphoblastic Leukemia

1578

HLA-DRB1*07:01–HLA-DQA1*02:01–HLA-DQB1*02:02 haplotype is associated with a high risk of asparaginase hypersensitivity in acute lymphoblastic leukemia Nóra Kutszegi et al.

Chronic Lymphocytic Leukemia

1587

The prohibitin-binding compound fluorizoline induces apoptosis in chronic lymphocytic leukemia cells through the upregulation of NOXA and synergizes with ibrutinib, 5-aminoimidazole-4-carboxamide riboside or venetoclax Ana M. Cosialls et al.

1594

Extracellular vesicles of bone marrow stromal cells rescue chronic lymphocytic leukemia B cells from apoptosis, enhance their migration and induce gene expression modifications Emerence Crompot et al.

Non-Hodgkin Lymphoma

1605

Anaplastic lymphoma kinase-positive anaplastic large cell lymphoma with the variant RNF213-, ATIC- and TPM3-ALK fusions is characterized by copy number gain of the rearranged ALK gene Jo-Anne van der Krogt et al.

Plasma Cell Disorders

1617

The spectrum of somatic mutations in monoclonal gammopathy of undetermined significance indicates a less complex genomic landscape than that in multiple myeloma Aneta Mikulasova et al.

Letters to the Editor Letters are available online only at www.haematologica.org/content/102/9.toc

e332

Mutated ASXL1 and number of somatic mutations as possible indicators of progression to chronic myelomonocytic leukemia of myelodysplastic syndromes with single or multilineage dysplasia Ana Valencia-Martinez et al. http://www.haematologica.org/content/102/9/e332

e336

Low-dose methotrexate in myeloproliferative neoplasm models Kavitha Chinnaiya et al. http://www.haematologica.org/content/102/9/e336

e340

Center-level variation in accuracy of adverse event reporting in a clinical trial for pediatric acute myeloid leukemia: a report from the Children’s Oncology Group Tamara P. Miller et al. http://www.haematologica.org/content/102/9/e340

e344

Acute myeloid leukemia stem cell function is preserved in the absence of autophagy Amy H. Porter et al. http://www.haematologica.org/content/102/9/e344

Haematologica 2017; vol. 102 no. 9 - September 2017 http://www.haematologica.org/


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

e348

Combining flow cytometry and WT1 assessment improves the prognostic value of pre-transplant minimal residual disease in acute myeloid leukemia Fabio Guolo et al. http://www.haematologica.org/content/102/9/e348

e352

Chlorambucil plus rituximab as front-line therapy for elderly and/or unfit chronic lymphocytic leukemia patients: correlation with biologically-based risk stratification Luca Laurenti et al. http://www.haematologica.org/content/102/9/e352

e356

Peripheral T-cell lymphoma cell line T8ML-1 highlights conspicuous targeting of PVRL2 by t(14;19)(q11.2;q13.3) Stefan Ehrentraut et al. http://www.haematologica.org/content/102/9/e356

e360

Improved classification of leukemic B-cell lymphoproliferative disorders using a transcriptional and genetic classifier Alba Navarro et al. http://www.haematologica.org/content/102/9/e360

e364

The level of deletion 17p and bi-allelic inactivation of TP53 has a significant impact on clinical outcome in multiple myeloma Sharmilan Thanendrarajan et al. http://www.haematologica.org/content/102/9/e364

e368

Long-term CD38 saturation by daratumumab interferes with diagnostic myeloma cell detection Anna Oberle et al. http://www.haematologica.org/content/102/9/e368

Case Reports Case Reports are available online only at www.haematologica.org/content/102/9.toc

e371

GATA1 erythroid-specific regulation of SEC23B expression and its implication in the pathogenesis of congenital dyserythropoietic anemia type II Roberta Russo et al. http://www.haematologica.org/content/102/9/e371

e375

Recessive grey platelet-like syndrome with unaffected erythropoiesis in the absence of the splice isoform GFI1B-p37 Harald Schulze et al. http://www.haematologica.org/content/102/9/e375

Haematologica 2017; vol. 102 no. 9 - September 2017 http://www.haematologica.org/


EDITORIALS Modeling mixed-lineage-rearranged leukemia initiation in CD34+ cells: a "CRISPR" solution Raúl Torres-Ruiz,1,2 Sandra Rodriguez-Perales,1 Clara Bueno2,3 and Pablo Menendez2,3,4 1

Molecular Cytogenetics and Genome Editing Unit, Human Cancer Genetics Program, Centro Nacional de Investigaciones Oncológicas (CNIO), Madrid; 2Josep Carreras Leukemia Research Institute and Department of Biomedicine, School of Medicine, University of Barcelona; 3Centro de Investigación Biomédica en Red de Cáncer (CIBER-ONC), ISCIII, Barcelona and 4Instituciò Catalana de Recerca i Estudis Avançats (ICREA), Passeig Lluis Companys, Barcelona, Spain. E-mail: rtorres@carrerasresearch.org or pmenendez@carrerasresearch.org doi:10.3324/haematol.2017.173740

I

n this issue of haematologica, Reimer et al.1 present an improved strategy based on genome engineering, viral vector transduction, and the use of CD34+ human hematopoietic stem and progenitor cells (HPSCs) to recreate a human leukemic chromosomal rearrangement, t(11;19)/MLL-ENL, in its natural genomic environment. This model provides new clues as to the complex molecular mechanisms of mixed-lineage-rearranged (MLLr) leukemia and opens up new avenues for the genomic reconstruction to study leukemia initiation and evolution. A common and disease/lineage-specific molecular signature of leukemia involves the generation of recurrent reciprocal chromosomal translocations, which are considered to be the oncogenic initiating drivers.2 Chromosomal genomic rearrangements are complex and implicate illegitimate recombination or juxtaposition of normally separated genes during DNA replication, and results in oncogene activation or, more commonly in leukemia, the generation of novel fusion genes.3 Our current understanding on how the nature of the target cell and the spatial organization of chromosomes in the nucleus contribute to chromosomal rearrangements (i.e. translocations) is very limited. Questions about the nature of the target cell in which the translocation arises and initiates leukemia can not really be studied with primary patient samples because all molecular insults are in place at the time of disease presentation.4-6 As an alternative, patientderived cell lines have been widely exploited to study the contribution of translocations to leukemogenesis; however, associated problems can arise when using this material. Human cancer cell lines are generated from primary cells once the full transformation events have taken place, and this can present challenges for distinguishing between driver and passenger events. Moreover, serial passage of cell lines can cause genotypic variation, and even heterogeneity in cultures, resulting in a loss of information on the leukemia initiation and the different steps of progression.7 These caveats aside, cell lines are powerful tools for ascertaining and characterizing the cancer gene, and have over recent decades increased our general understanding of the molecular pathophysiology of chromosomal rearranged leukemia.8 Beyond in vitro studies, genetically modified animal and cellular models constitute invaluable tools for cancer investigation, but they also have limitations, due in part to the manner in which they are generated. Most of the extensively used leukemia models are generated using viral vector-based approaches (primarily recombinant retroviruses and lentiviruses), randomly integrated plasmid DNA or, in a more refined manner, by generating “knock-in” fusion genes.5,6,9,10 Some of the major concerns regarding these methodologies are: i) the high potential mutagenic rate associated with uncontrolled haematologica | 2017; 102(8)

cassette integration that could lead to a growth advantage and variegated cell lines (for plasmid DNA and integrative recombinant virus approaches); ii) an exogenous strong transgene expression controlled by promoters that function in a broad range of cells; or iii) the expression of just one of the fusion genes generated by a chromosomal translocation and the retention of the two wild-type alleles (for the “knock-in” strategy) which is not seen in patients.11 These concerns guide the search for more faithful models capable of recapitulating the initial genetic events associated with the leukemogenic process with the least alteration in the genomic architecture. Until recently, the most effective way to replicate these events was based on the generation of chromosomal translocations using translocator technology. This approach involves the use of the Cre/loxP site-specific system via prior engineering in the mouse genome.12 While some leukemia animal and cellular models have been successfully developed using this strategy, its use in human cells is extremely inefficient. More recently, the explosion of new genome editing technologies, particularly the CRISPR/Cas9 system, has permitted the efficient recreation of de novo cancer-associated translocations in vitro in mouse and human cells,13 and in vivo in mouse models.14 This is important because it has been demonstrated that, although conditional models or viral vector expression systems can efficiently generate chromosomal translocations, the engineered cells do not always initiate a leukemic-like phenotype in mouse xenograft models. In the present article, Reimer et al.1 combine the undemanding aspects of lentiviral generation and delivery with the advantages of chromosomal translocation generation by the CRISPR system. The authors engineered an advanced lentiviral CRISPR/Cas9 vector for efficient transduction of human CD34+ HPSCs. This new lentiviral vector permits the induction of double strand breaks (DSBs) in the MLL and ENL intronic regions with very high efficiency (78-83%), favoring the generation of the t(11;19) rearrangement. As a consequence of the translocation, two derivative chromosomes are generated, der(11) and der(19), leading to the expression of both fusion transcripts and the concomitant loss of one copy of the wild-type MLL and ENL genes. The authors describe a transient outgrowth advantage in longterm cultures of the primary human CD34+ HSPCs t(11;19)+, but more significantly, they demonstrated that when these cells were injected into primary immunodeficient recipients, the in vivo environment favored oncogenic transformation, initiating a monocytic leukemia-like disease. It is important to emphasize that whereas this transformative process does 1467


Editorials

not fully recapitulate the human leukemic phenotype in primary recipients, secondary recipients developed acute lymphoblastic leukemia, albeit with incomplete penetrance. Based on these findings, the authors conclude that “environmental cues not only contribute to the disease phenotype, but also to t(11;19)/MLL-ENL-mediated oncogenic transformation”.1 This result resembles the effect of chemotherapy in patients, but it will need to be confirmed in future studies. While the results described in this study represent a great improvement in the field, some questions remain and some issues have still not been fully clarified. For example, although the authors demonstrated an extremely high DSB induction efficiency, the translocation rate remained very low (0.2% or 1.6*10-3), even when compared with other publications.13,15,16 This could be due to the genome architecture of the leukemia-initiating cell, the proximity between involved loci or the presence of repetitive elements.17 In addition, although it is a more advantageous model due to the conservation of all endogenous regulatory elements (promoters, enhancers, miRNA binding sites and rearranged genome architecture) and the possibility to study the initial steps of the leukemic process, the approach is grounded on the use of random integrative lentiviruses, which could lead to mutagenic effects and growth advantages associated with their integration pattern. Moreover, the continuous expression of the CRISPR/Cas9 components may increase the likelihood of undesirable off-target effects over time. Regarding this latter issue, two alternative approaches have recently been described that could further improve the use of the CRISPR system in the generation of leukemic models. The first one relies on the use of a “hit-and-run” protein-based Cas9 system that circumvents the lentiviral integration concerns and lessens the chances of off-target effects while permitting the generation of human chromosomal translocations in a wide variety of primary stem cells with higher efficiencies.16 The second approach takes advantage of the classical “knock-in” model, but with application to human cells with conditional allele expression and resistance selection cassettes in combination with the CRISPR/Cas9 system.15 Because of their inducible nature, both approaches could widen our knowledge of the first steps leading to the establishment of cancer and to its progression. The CRISPR/Cas9 system has revolutionized functional biology, biotechnology, and genomic medicine. The present article by Reimer et al.1 illustrates how the use of more accurate models generated by genome engineering techniques in human CD34+ HPSCs can transform the field of basic leukemia biology. A deeper knowledge of the CRISPR approach and the development of new applications should open new horizons for the study of the molecular and cellular processes of cancer, and will make it easier to reproduce the complex cancer genome and epigenome, allowing a more rigorous molecular analysis of the molecular mechanisms involved in tumor progression and the identification of oncogenes and tumor suppressor genes. Specifically, further developments to isolate the minor fraction of bona fide genome-edited CD34+

1468

clones (via antibiotic selection, reporter expression, etc.) would open up fascinating new avenues in the study of leukemia and cancer modeling. Acknowledgments This work was supported by the European Research Council (CoG-2014-646903) and the Spanish Ministry of EconomyCompetitiveness (SAF-2016-80481-R and RTC-2016-46031) to PM, the Asociación Española Contra el Cáncer to CB, and the ISCIII/FEDER (PI14/01191 & PI14/01884) to CB and SR-P. PM acknowledges the financial support from the Obra Social La Caixa-Fundaciò Josep Carreras, the Inocente-Inocente Foundation and Generalitat de Catalunya. PM is investigator of the Spanish Cell Therapy co-operative network (TERCEL). RTR was supported by an international fellowship from Lady TATA Memorial Trust.

References 1. Reimer J, Knoess S, Labuhn M, et al. CRISPR-Cas9-induced t(11;19)/MLL-ENL translocations initiate leukemia in human hematopoietic progenitor cells in vivo. Haematologica. 1 June 2017 [Epub ahead of print]. 2. Mitelman F, Johansson B, Mertens F. The impact of translocations and gene fusions on cancer causation. Nat Rev Cancer. 2007;7(4):233-245. 3. Richardson C, Jasin M. Frequent chromosomal translocations induced by DNA double-strand breaks. Nature. 2000;405(6787):697-700. 4. Bueno C, Montes R, Catalina P, Rodríguez R, Menendez P. Insights into the cellular origin and etiology of the infant pro-B acute lymphoblastic leukemia with MLL-AF4 rearrangement. Leukemia. 2011;25(3):400-410. 5. Rodríguez R, Tornin J, Suarez C, et al. Expression of FUS-CHOP fusion protein in immortalized/transformed human mesenchymal stem cells drives mixoid liposarcoma formation. Stem Cells. 2013;31(10):2061-2072. 6. Rodríguez R, Rubio R, Gutierrez-Aranda I, et al. FUS-CHOP fusion protein expression coupled to p53 deficiency induces liposarcoma in mouse but not in human adipose-derived mesenchymal stem/stromal cells. Stem Cells. 2011;29(2):179-192. 7. Feng H, Zhang Y, Liu K, et al. Intrinsic gene changes determine the successful establishment of stable renal cancer cell lines from tumor tissue. Int J Cancer. 2017;140(11):2526-2534. 8. McCormack E, Bruserud O, Gjertsen BT. Review: genetic models of acute myeloid leukaemia. Oncogene. 2008;27(27):3765-3779. 9. Cook GJ, Pardee TS. Animal models of leukemia: any closer to the real thing? Cancer Metastasis Rev. 2013;32(1-2):63-76. 10. Montes R, Ayllón V, Gutierrez-Aranda I, et al. Enforced expression of MLL-AF4 fusion in cord blood CD34+ cells enhances the hematopoietic repopulating cell function and clonogenic potential but is not sufficient to initiate leukemia. Blood. 2011;117(18):4746-4758. 11. Sadelain M, Papapetrou EP, Bushman FD. Safe harbours for the integration of new DNA in the human genome. Nat Rev Cancer. 2011;12(1):51-58. 12. Rodriguez-Perales S, Cano F, Lobato MN, Rabbitts TH. MLL gene fusions in human leukaemias: in vivo modelling to recapitulate these primary tumourigenic events. Int J Hematol. 2008;87(1):3-9. 13. Torres R, Martin MC, Garcia A, Cigudosa JC, Ramirez JC, RodriguezPerales S. Engineering human tumour-associated chromosomal translocations with the RNA-guided CRISPR–Cas9 system. Nat Commun. 2014;5:3964. 14. Maddalo D, Manchado E, Concepcion CP, et al. In vivo engineering of oncogenic chromosomal rearrangements with the CRISPR/Cas9 system. Nature. 2014;516(7531):423-427. 15. Spraggon L, Martelotto LG, Hmeljak J, et al. Generation of conditional oncogenic chromosomal translocations using CRISPR-Cas9 genomic editing and homology-directed repair. J Pathol. 2017;242(1):102-112. 16. Torres-Ruiz R, Martinez-Lage M, Martin MC, et al. Efficient Recreation of t(11;22) EWSR1-FLI1(+) in Human Stem Cells Using CRISPR/Cas9. Stem Cell Reports. 2017;8(5):1408-1420. 17. Roukos V, Voss TC, Schmidt CK, Lee S, Wangsa D, Misteli T. Spatial dynamics of chromosome translocations in living cells. Science. 2013;341(6146):660-664.

haematologica | 2017; 102(8)


REVIEW ARTICLE

Chronic lymphocytic leukemia cells are active participants in microenvironmental cross-talk Martijn HA van Attekum,1,2 Eric Eldering1,3 and Arnon P Kater2,3

Department of Experimental Immunology, Academic Medical Center, University of Amsterdam; 2Department of Hematology, Academic Medical Center, University of Amsterdam; 3Lymphoma and Myeloma Center Amsterdam (LYMMCARE), Academic Medical Center, University of Amsterdam, the Netherlands

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

1

Haematologica 2017 Volume 102(9):1469-1476

ABSTRACT

T

he importance of the tumor microenvironment in chronic lymphocytic leukemia is widely accepted. Nevertheless, the understanding of the complex interplay between the various types of bystander cells and chronic lymphocytic leukemia cells is incomplete. Numerous studies have indicated that bystander cells provide chronic lymphocytic leukemia-supportive functions, but it has also become clear that chronic lymphocytic leukemia cells actively engage in the formation of a supportive tumor microenvironment through several cross-talk mechanisms. In this review, we describe how chronic lymphocytic leukemia cells participate in this interplay by inducing migration and tumor-supportive differentiation of bystander cells. Furthermore, chronic lymphocytic leukemia-mediated alterations in the interactions between bystander cells are discussed. Upon bystander cell interaction, chronic lymphocytic leukemia cells secrete cytokines and chemokines such as migratory factors [chemokine (C-C motif) ligand 22 and chemokine (CC motif) ligand 2], which result in further recruitment of T cells but also of monocyte-derived cells. Within the tumor microenvironment, chronic lymphocytic leukemia cells induce differentiation towards a tumorsupportive M2 phenotype of monocyte-derived cells and suppress phagocytosis, but also induce increased numbers of supportive regulatory T cells. Like other tumor types, the differentiation of stromal cells towards supportive cancer-associated fibroblasts is critically dependent on chronic lymphocytic leukemia-derived factors such as exosomes and platelet-derived growth factor. Lastly, both chronic lymphocytic leukemia and bystander cells induce a tolerogenic tumor microenvironment; chronic lymphocytic leukemia-secreted cytokines, such as interleukin-10, suppress cytotoxic T-cell functions, while chronic lymphocytic leukemia-associated monocyte-derived cells contribute to suppression of T-cell function by producing the immune checkpoint factor, programmed cell death-ligand 1. Deeper understanding of the active involvement and cross-talk of chronic lymphocytic leukemia cells in shaping the tumor microenvironment may offer novel clues for designing therapeutic strategies.

Introduction Chronic lymphocytic leukemia (CLL) is a prototypic malignancy that not only depends on intrinsic genetic defects, but is maintained by interactions with bystander cells in microenvironmental niches such as the lymph node. Bystander cells involved include T cells, monocyte-derived cells (MDC), and stromal cells (such as endothelial cells, fibroblastic reticular cells, and pericytes). Signals emanating from these cells critically affect several key features of malignancy of CLL cells, haematologica | 2017; 102(9)

Correspondence: a.p.kater@amc.nl

Received: October 27, 2016. Accepted: June 8, 2017. Pre-published: August 3, 2017. doi:10.3324/haematol.2016.142679 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1469 Š2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1469


M.HA van Attekum et al.

such as cell survival, chemo-resistance, cell proliferation, and migration.1 Moreover, these signals result in an immunotolerant milieu in the CLL lymph node, in which the response to both pathogens2 and neo-antigen-expressing malignant cells3 is dampened. Multiple types of regulators are involved in these communication processes: first, interleukins, such as interleukin (IL)-4 and IL-21, are involved in cell survival and proliferation4,5 and IL-10 in immunosuppression.6 Second, chemokines, including C-C motif chemokine (CCL)2, 3, 4, and 22, have an important role in chemo-attraction of cells towards the tumor microenvironment (TME).7,8 In addition, CCL2 might play a role in tumor cell survival by indirect support via the microenvironment.9 Third, growth factors, such as insulin-like growth factor 1, can promote survival.10 Fourth, membrane-bound factors from bystander cells, such as CD40L and integrins, can induce cell survival.11 Fifth, small vesicles, such as microvesicles and exosomes containing RNA, proteins, lipids or metabolites that are produced by either bystander cells12 or CLL cells,13,14 could transmit signals. Sixth, nucleoside adenosine is involved in dampening the local immune response and causing chemoresistance in CLL cells.15 Although it is by now well established that the factors secreted by bystander cells are essential for sustaining CLL (summarized in a recent review by Ten Hacken & Burger1), it has also become clear that these interactions are reciprocal in nature. As shown in other tumor types, upon contact with tumor cells, bystander cells can undergo changes that drive tumor progression.7 Considering that CLL bystander cells include immune cells normally involved in highly adaptable immune responses, they are highly susceptible to (malignant) B-cell-derived signals. Alongside local changes leading to tumor progression, bystander cell alterations lead to systemic changes that can orchestrate recruitment of peripheral cells towards the TME.7 Although various studies have suggested that bystander cell changes can take place at the genetic level,7 recent evidence has shown unaltered stromal genomes, suggesting that microenvironmental signals are not mediated via genetic events.7 These findings indicate that the stromal alterations are reversible, and that identification of the factors driving stromal cell changes may yield new therapeutic options. In this review we analyze contemporary literature and our own recent findings to provide an overview of current evidence that signals emanating from CLL cells are crucial in creating a tumor-supportive TME. Second, as several reports show interdependency of bystander cells, we address how communication among bystander cells can contribute, in the context of CLL, to supportive TME interactions. We focus on T cells, MDC and stromal cells which together with CLL cells can form a tetrad exchanging reciprocal signals. For each of these, the functional effects of CLL cells towards the bystander cells are discussed followed by the relevant mechanisms. Lastly, we discuss effects between bystander cells.

T-cell interactions Although it has been described that CD4+ T helper type 1 (Th1) cells recognize CLL antigens,3 activated Th1 cells also induce CLL-cell proliferation and survival.16 Furthermore, T cells activate mitochondrial metabolism in CLL cells, which renders CLL cells more resistant to chemotherapy and contributes to cell proliferation.17 Pro1470

tumor signals from T cells include both antigen-independent proliferation factors (CD40L in combination with IL215) as well as survival inducing factors [interferon (IFN)γ,18 IL-4,4 and CD40L19] (Figure 1). These pro-survival signals result in a nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB)-dependent upregulation of Bcell lymphoma 2 (BCL2) family members BCL2-related protein A1 (BFL-1) and B-cell lymphoma-extra large (BCLXL),20 and protein kinase B (AKT)-dependent upregulation of induced myeloid leukemia cell differentiation protein (MCL-1).21 With respect to CLL proliferation, mitogenactivated protein kinase (MAPK) and signal transducer and activator of transcription (STAT)3 pathways play additional roles.22 The interaction of CLL cells with T cells sensitizes CLL cells to additional TME-derived pro-tumor signals; first, Bcell receptor (BCR) signaling is enhanced by a microRNA155-dependent mechanism after CD40L stimulation.23 Second, CLL cells upregulate adhesion protein CD44 after CD40L stimulation, leading to hyaluronic acid binding, which increases retention in the lymph node.24 Third, alongside a direct survival-inducing effect of T-cell-secreted IFN-γ on CLL cells, CD38 is upregulated on CLL cells after IFN-γ stimulation. CD38 can subsequently relay MDC-derived CD31 survival signals,25 although this has been difficult to confirm in vitro.26 These findings indicate that the pro-tumor effects of T cells might be partially mediated via other TME elements. Various groups have described aberrations in the T-cell population in CLL patients. The total number of both CD4+ and CD8+ T cells is increased27 and a skewing of their ratio towards CD8+ cells occurs in both mice28 and humans.29 This skewing does not precede the occurrence of CLL, as it is not present during monoclonal B-cell lymphocytosis,30 but even at an early stage of CLL, expansion of the CD8+ T-cell population is correlated with an adverse outcome.29 These findings indicate that CLL cells are the causative agent in this correlation. Furthermore, with respect to T-cell developmental stages, an increase in effector cells at the expense of naïve cells is observed.31 The functional consequences of this skewing are currently unknown, but it could be speculated that a decreased naïve T-cell pool reduces the number of potential cytotoxic T cells directed towards CLL neo-antigens. Alongside the effects of CLL cells on T-cell skewing, CLL cells induce an exhausted T-cell phenotype.32 This phenotype is characterized by increased expression of exhaustion markers CD160, CD244, BLIMP-1, and programmed cell death protein (PD)-132 and an inability to produce adequate levels of immune-activating cytokines upon stimulation,33 similar to the phenotype of T cells directed towards chronic virus infections. Concurrently, effective synapse formation of T cells is suppressed by causing non-polarized release of lytic granules.34 These mechanisms likely contribute to T-cell dysfunction. Lastly, CLL cells are involved in the induction of migration of T cells towards the lymph node.8 Several mechanisms have been linked to the observed suppression of T-cell function by CLL cells; it has been observed that CLL cells overexpress immune inhibitory factors such as programmed death ligand (PD-L)1 and PDL235 and T cells from CLL patients have increased levels of the PD-1 receptor.29 As PD-1 expression also increases with age, these observations should be interpreted with caution. To study causality, the Em-TCL1 mouse model, in haematologica | 2017; 102(9)


Role of CLL cells in TME cross-talk

Figure 1. Interactions between chronic lymphocytic leukemia cells and bystander cells that contribute to the formation of a tumor-supportive microenvironment. Within the tetrad of CLL cells, T cells, MDC, and stromal cell, relevant effects (in bold) and signaling molecules involved in the interaction of CLL cells with T cells, MDC, and stromal cells are indicated. IL: interleukin; CCL: chemokine (C-C motif) ligand; IFN: interferon; APRIL: a proliferation inducing ligand; BAFF: B-cell activating factor; NAMPT: nicotinamide phosphoribosyltransferase; HMGB-1: high mobility group box 1; IDO: indoleamine 2,3-dioxygenase; MIF: migration inhibitory factor; CAF: cancer associated fibroblasts; PDGF: platelet derived growth factor.

which oncogene T-cell leukemia/lymphoma protein 1 (TCL1) is overexpressed under control of the B-cell-specific immunoglobulin heavy enhancer, has been used.36 Although CLL in this mouse model is mainly driven by TCL1 in contrast to heterogeneous drivers in human disease, findings in this model have been valuable in explaining at least some of the observed immune disturbances in human CLL.36 Using this model, aging bias was excluded by showing that adoptive transfer of CD19+ cells of either wild-type or TCL-1 donor mice towards young wild-type recipients also induces PD-1 on T cells.35 Alongside PD-1mediated signaling, CLL cells produce the immune inhibitory cytokine IL-10.6 Also, unknown contact-dependent factors produced by CLL cells actively impair T-cell synapse formation.37 In addition, adenosine, which is produced in the hypoxic CLL TME, can also contribute to decreased T-cell proliferation.15 Very recently, a link between CLL-mediated T-cell dysfunction and altered immune metabolism was made by showing CLL-mediated suppression of T-cell glucose metabolism.38 Whether impaired metabolism is a direct consequence of competition for fuels between the tumor cells and T cells, as has been shown in experimental models in other tumors,39 or is solely due to CLL-mediated decreased AKT/mTOR signaling38 has still to be resolved. It is important to note that the mechanistic causes of T-cell haematologica | 2017; 102(9)

expansion and skewing remain largely obscure, but the defects in T-cell function might underlie the compensatory expansion seen in CLL patients.29 Several factors secreted by CLL cells can induce migration of T cells towards the CLL lymph node. CCL22 for instance, is secreted by CLL cells in the lymph node, which results in the recruitment of T cells.8 Interestingly, as CCL22 preferentially induces migration of T helper type 2 (Th2) and T regulatory (Treg) CD4+ cells,40 secretion of this chemokine could lead to skewing in the lymph node towards CLL-supporting and immunosuppressive T cells at the expense of cytotoxic T cells. Together with Tcell recruitment via CCL22, CLL cells secrete CCL3 and CCL4 upon interaction with MDC41 and levels of CCL3 correlate with increased T-cell numbers in CLL lymph nodes.42 Finally, the fact that T cells show reduced motility upon direct contact with CLL cells43 could indicate that T cells are retained in the lymph node once recruited. It has recently become evident that CLL cells also affect the phenotype of non-conventional T cells. A small population (1-10%) of the total T-cell pool carries the highly conserved γδ T-cell receptor (TCR) instead of the more prevalent αβ TCR.44 Within this population, Vγ9Vδ2 T cells are the predominant subset present in the peripheral blood. In contrast to the recognition of peptide antigens by αβ T cells, Vγ9Vδ2 T cells respond to stress molecules 1471


M.HA van Attekum et al.

in malignant cells, in a TCR-dependent yet major histocompatibility complex (MHC)-independent process. As a consequence, these γδ T cells could suppress CLL cells acting independently of MHC antigen presentation.44 Compared to cells from healthy donors, however, these γδ cells show a dysfunctional phenotype in CLL.45 Interestingly, we found that these defects are spontaneously reverted when patient-derived γδ T cells are cultured in the absence of CLL cells,46 in support of continuous, active subversion by CLL cells. In their role as immunosuppressive cells, Treg cells, on the other hand, secrete several immunosuppressive cytokines such as IL-10 and their number correlates with worse prognosis in several tumors.47 In CLL, the frequency of forkhead box protein (FOXP)3+ Treg cells is increased in advanced disease.48 IL-10 production by Treg cells is higher in the CLL lymph node than in peripheral blood,49 in accordance with microenvironmental signals engaging in immunosuppressive skewing.

Monocyte-derived cell interactions MDC include monocytes, macrophages, and dendritic cells. These cells can, on the one hand, secrete essential survival factors for CLL cells, while, on the other hand, they can potentially mount an immune response against malignant cells as co-stimulators of B- or T-cell-mediated responses.50 According to the dichotomized view of macrophage differentiation proposed in normal biology, M1 differentiated immunogenic macrophages mainly convey anti-tumor signals, while M2 wound healing macrophages are pro-tumorigenic overall.51 The delayed disease development associated with MDC depletion in the TCL1 mouse model52,53 suggests that MDC have a crucial, tumor-supportive function in CLL. Their supportive role is further indicated by the observation that a higher number of MDC correlates with worse prognosis in CLL patients.54 Whereas MDC play important roles in inducing CLL-cell survival55 and have migratory effects on CLL cells55 (Figure 1), their role in inducing proliferation is subordinate; stimulation of CLL cells by macrophages does not induce proliferation (unpublished observation) and furthermore no spatial correlation between the MDC marker CD68 and the proliferation marker Ki67 exists in lymph nodes from CLL patients.56 We recently found that MDCmediated survival depended on chemokine signaling via CCR1.21 Nurse-like cells are monocyte-derived cells which develop following prolonged in vitro culture with CLL cells55 and have been identified in both the spleen and lymph nodes of CLL patients.57 Nurse-like cells are thought to induce CLL survival effects via factors such as A proliferation inducing ligand (APRIL), B-cell activating factor (BAFF) or C-X-C motif chemokine (CXCL)12 (reviewed by Ten Hacken & Burger1). In line with this, transgenic APRIL overexpression in the TCL1 mouse led to faster disease progression.58 By contrast, using a novel APRIL-overexpression system and an APRIL decoy receptor, we recently found in vitro that direct effects of APRIL produced by macrophages on CLL cells are negligible.56 This discrepancy could be reconciled by postulating that in vivo effects of APRIL may be indirect, as exemplified by the recent finding that immunosuppressive IL-10 is produced by non-malignant B cells upon stimulation of APRIL receptor TACI.59-61 In line with the overall pro-tumor effect of CLL-associated MDC, pro-tumor M2 differentiation of macrophages 1472

in the presence of CLL cells has been found ex vivo and in vitro.62-64 Functionally, these cells show impaired immunocompetence, as antigen presentation and immune response initiation are decreased.65 In addition, CLL-associated monocytes are defective in their phagocytic function.66 Moreover, dendritic cells in mice that have undergone adoptive transfer of TCL1 CLL cells show a decrease of MHC class II expression and an increase of the immunosuppressive molecule PD-L1,52 and in CLL patients they have a suppressed immature phenotype showing decreased proliferation and cytokine production after stimulation.67 Several groups,53,68 including our group,56 have found that the CLL lymph node is interspersed with macrophages. As recruitment of these supportive macrophages depends on chemokine gradients emanating from the lymph node, it is postulated that CLL cells can provide these migratory signals. Indeed, it has recently been shown that in the TCL1 mouse model, CLL-infiltrated tissues harbor an increased number of monocytes compared to non-transgenic mice.52 Several CLL-secreted factors have been suggested to contribute to the pro-tumorigenic M2 differentiation of monocytes, which include nicotinamide phosphoribosyltransferase (NAMPT)63 and high mobility group box 1 (HMGB1).68 As NAMPT is also secreted by CLL-differentiated MDC, it could form a positive feedback loop keeping MDC in a CLL-supportive state.63 Furthermore, by generating a hypoxic TME, CLL cells indirectly induce M2 differentiation as hypoxia increases adenosine production by MDC, which is known for its M2 differentiating capacity.15 Besides the direct effects of these factors in inducing M2 differentiation, CLL-associated monocytes are primed for M2 differentiation as they show increased phosphorylation of downstream STAT molecules in response to M2differentiating cytokines IL-4 and IL-10.52 The persistent M2-differentiating signals emanating from CLL cells residing in the lymph node, in combination with PD-1 stimulation,69 could explain the immune dysfunction of MDC. These effects are further enhanced by autocrine stimulation via PD-L1 or IL-10 expressed by the MDC themselves.52 Furthermore, IL-10 is responsible for immune dysfunction seen in dendritic cells,67 as here it leads to STAT6 suppression via suppressor of cytokine signaling 5.70 Interestingly, the tumor supportive phenotype in MDC is reversible, as IFN-γ stimulation results in transdifferentiation of pro-tumorigenic (M2) CLL-associated monocytes towards M1 macrophages.71 Similarly, inhibiting PD1 signals could restore macrophage function,69 suggesting there is potential for therapeutic intervention in these pathways. Although several chemokines could account for the recruitment of monocytes towards the CLL lymph node, a critical role for CCR2 has recently been proposed. Adoptive transfer of CLL cells from TCL1 mice to CCR2 knockout mice led to a decrease in monocyte numbers in the spleen.52 We recently found that primary CLL cells are able to secrete several monocyte-attracting chemokines such as CCL2, 3, 4, 5, 7, and 24, and CXCL5 and 10 after stimulation with the T-cell factor CD40L, resulting in monocyte migration. In line with data from Hanna et al.,52 specific inhibition of CCR2 by small molecules could completely abrogate the migration towards CLL cells.62 Others have found that knockout of macrophage migrahaematologica | 2017; 102(9)


Role of CLL cells in TME cross-talk

tion inhibitory factor (MIF) reduced the number of macrophages in the spleen of TCL1 mice, suggesting an additional role for this chemokine in MDC migration.72

Stromal-cell interactions Stromal cells constitute the connective tissue of organs and supply them with structure, anchoring and supportive signals. By definition, they are of non-hematopoietic origin. Different types of stromal cells include fibroblasts, reticular cells, and endothelial cells. Stromal cells can play a supportive role in various tumor environments, including the CLL lymph node73 and stromal cell numbers generally correlate with tumor progression and worse prognosis.74 Via several mechanisms, stromal cells can directly support CLL cells, for example, by inducing chemoresistance, promoting migration, and increasing cell survival via factors such as NOTCH1 (reviewed by Ten Hacken & Burger1). In addition, they induce CLL-cell proliferation75 and change CLL-cell metabolism76 (Figure 1). At the level of metabolism, stromal cells can supplement the defective CLL cystine transport by secreting large amounts of cystein into the TME.77 Alongside these direct effects, stromal cells can govern changes in CLL cells that make them more receptive to other microenvironmental signals. Upon co-culture with stromal cells, CLL cells upregulate transcription factor hypoxia-inducible factor-1Îą, which can induce changes in chemokine receptor expression in CLL cells that consequently retain them in the TME.78 As is the case for T cells and MDC, the stromal-cell secretome depends on the extracellular signals it receives. In order to provide tumor support, different stromal cell types have been shown to transdifferentiate into so-called cancer-associated fibroblasts in different malignancies.74 In CLL, it has been suggested that this differentiation takes place via specific microRNA delivered through exosomes.14 To support CLL cells, cancer-associated fibroblasts require AKT signaling.79 A bidirectional cross-talk in which CLL cells induce AKT and extracellular signal-regulated kinase (ERK) signaling has been described80 and platelet-derived growth factor is one secreted factor that can cause this activation.81 In summary, these mechanisms underpin the dependence on CLL-secreted factors for tumor-supportive differentiation of stromal cells.

Interactions between bystander cells: monocyte-derived cells and T cells We have so far discussed several direct reciprocal interactions between CLL cells and bystander cells. Considering that all cells within an ecosystem partake in reciprocal signaling, interactions between bystander cells can likewise contribute to the formation of a supportive TME in CLL. Based on their role in the normal immune response, it is to be expected that MDC can also affect the phenotype of T cells in the context of CLL. Indeed, MDC contribute to T-cell skewing in CLL as skewing was reverted after depletion of MDC via clodronate treatment in the TCL1 mouse model.52 Furthermore, MDC are involved at several levels of Tcell suppression; first, in the context of CLL, MDC induce expression of PD-1 on T cells,63 while PD-L1 is upregulated on CLL-differentiated monocytes,52 both contributing to T-cell suppression. Second, CLL-differentiated monocytes inhibit T-cell proliferation63 and third, they can inhibit Thaematologica | 2017; 102(9)

cell activation and promote the differentiation towards Treg cells via secretion of immunosuppressive indoleamine 2,3dioxygenase (IDO).64 Like CLL cells, CLL-differentiated MDC can secrete chemokines that can attract T cells towards the lymph node, such as CXCL12. Furthermore, this chemokine enhances the expression of CLL-cell survival stimuli such as IFN-Îł by T cells.82 Similarly, in mouse studies, splenic monocytes show increased levels of T-cell-attracting chemokines such as CXCL9 and 10 after adoptive transfer of TCL1 CLL cells.52 Concurrently, expression of the receptor for these chemokines (CXCR3), increases on T cells.52 This indicates that supporting cells are not only recruited to the TME via induction of attracting chemokines in the lymph node, but also by an increased susceptibility to recruitment via chemokine receptor upregulation. A subset of MDC, myeloid-derived suppressor cells (MDSC; expressing CD11b and CD33 and low levels of human leukocyte antigen-DR), provides important tumorsupportive factors in several other malignancies due to its immunosuppressive nature.83 In CLL, it has been shown that the MDSC population is expanded64,84 and that T cells are suppressed by MDSC.64 Furthermore, the number of MDSC correlated with the number of CLL cells in patients.84 These data indicate that MDSC might also suppress the T-cell response in the context of CLL.

Conclusion and therapeutic consequences Cellular cross-talk is the driving force in establishing supportive interactions between elements within the TME. In this review, we have described several CLL-supportive mechanisms by bystander cells and the contribution of CLL cells. It is, however, important to keep in mind that CLL is an intra- and inter-tumoral genetically heterogeneous disease and that several of the described supportive TME mechanisms might depend on a specific genetic background. As an example, CLL cells harboring a NOTCH1 mutation might be more sensitive to NOTCH1 ligands present in the TME.85 Likewise, a subset of CLL, specifically cases that harbor the trisomy 12 aberration, overexpresses CD49d, which might make them more sensitive to lymph node homing.86 In the same vein, it has been shown that CD38-overexpressing CLL cells are more reactive to (microenvironmental) CXCL12 signaling and BCR signals.87 Lastly, IGHV mutation status and expression levels and mutations of intracellular BCR signaling proteins such as zeta-chain-associated protein kinase 70, spleen tyrosine kinase, and Bruton tyrosine kinase (BTK) can dictate CLL-cell responses to TME BCR signals.1 This shows that the receptiveness of CLL cells to TME support and subsequent disease outcome might depend on genetic alterations specific to CLL patients or to particular clones. With these caveats in mind, we here discuss potential consequences for CLL therapy. With the advent of new treatment modalities for CLL, the potential side-effects that novel therapies have on bystander cells should be considered. For instance, because MDC-mediated antibody responses depend on BTK,88 ibrutinib treatment reduces FcÎłR-mediated cytokine production,88 inhibits activation,69 and changes metabolism69 in monocytes, which can inhibit their immune function, as has been shown for antibody-dependent cell-mediated cytotoxicity.89 The outgrowth of adoptively transferred CLL cells 1473


M.HA van Attekum et al.

was, however, impaired in Btk knockout recipient mice, and macrophages deficient for its upstream tyrosine-protein kinase, Lyn, showed diminished CLL-supportive capacity ex vivo.90 This suggests that the effects of ibrutinib on macrophages would be clinically beneficial. The depletion of immunosuppressive MDSC by ibrutinib91 could furthermore support its beneficial clinical effects. Lastly, ibrutinib targets the T-cell-expressed BTK homolog, interleukin-2-inducible kinase (ITK), which is an important modulator of T-cell signaling and function.92 Interestingly, as ITK inhibition preferentially affects Th2 cells because Th1 cells express a compensatory kinase, a potentially beneficial Th1 anti-tumor skewing occurs,92 as was recently observed in vivo in pancreas carcinoma-engrafted mice.93 In the context of chimeric antigen receptor T-cell therapy, T-cell expansion and increased tumor clearance were found when this therapy was used concurrently with ibrutinib,94 indicating that ibrutinib treatment can overcome the suppressive effects of CLL cells on T cells. The effects of the kinase inhibitor idelalisib on bystander cells are generally CLL-supportive, as idelalisib reduces cytotoxic cytokine production of T cells95 and in macrophages it reduces antibody-dependent cell-mediated cytotoxicity89 and migration,96 although specific inhibi-

References 1. Ten Hacken E, Burger JA. Microenvironment interactions and B-cell receptor signaling in chronic lymphocytic leukemia:implications for disease pathogenesis and treatment. Biochim Biophys Acta. 2016;1863(3):401413. 2. Jurado-Camino T, Cordoba R, EstebanBurgos L, et al. Chronic lymphocytic leukemia: a paradigm of innate immune cross-tolerance. J Immunol. 2015;194(2):719727. 3. Rajasagi M, Shukla SA, Fritsch EF, et al. Systematic identification of personal tumorspecific neoantigens in chronic lymphocytic leukemia. Blood. 2014;124(3):453-462. 4. Dancescu M, Rubio-Trujillo M, Biron G, Bron D, Delespesse G, Sarfati M. Interleukin 4 protects chronic lymphocytic leukemic B cells from death by apoptosis and upregulates Bcl-2 expression. J Exp Med. 1992;176(5):1319-1326. 5. Pascutti MF, Jak M, Tromp JM, et al. IL-21 and CD40L signals from autologous T cells can induce antigen-independent proliferation of CLL cells. Blood. 2013;122(17):30103019. 6. DiLillo DJ, Weinberg JB, Yoshizaki A, et al. Chronic lymphocytic leukemia and regulatory B cells share IL-10 competence and immunosuppressive function. Leukemia. 2013;27(1):170-182. 7. Polyak K, Haviv I, Campbell IG. Co-evolution of tumor cells and their microenvironment. Trends Genet. 2009;25(1):30-38. 8. Ghia P, Strola G, Granziero L, et al. Chronic lymphocytic leukemia B cells are endowed with the capacity to attract CD4+, CD40L+ T cells by producing CCL22. Eur J Immunol. 2002;32(5):1403-1413. 9. Burgess M, Cheung C, Chambers L, et al. CCL2 and CXCL2 enhance survival of primary chronic lymphocytic leukemia cells in vitro. Leuk Lymphoma. 2012;53(10):19881998.

1474

tion of phosphoinositide 3-kinase Îł leads to an immunostimulatory macrophage differentiation.97 Given the critical pro-tumor effects of bystander cells, these findings suggest that complete tumor eradication after debulking treatment with chemotherapeutics can only be achieved after restoration of T-cell function by ibrutinib94 or lenalidomide,98 which can be complemented with CLLdirected chimeric antigen receptor T cells and PD-L1 inhibition.99 In addition, as ibrutinib treatment results in migration of CLL cells out of the lymph node, subsequent CLL-attracting chemokine inhibition could avoid (re)formation of a tumor-supportive microenvironment and increase the effectiveness of cytotoxic therapies. The effectiveness of this strategy of migration inhibition has been shown, for instance, in vivo in prostate cancer, in which metastases were reduced after CXCR4 inhibition.100 In conclusion, future insights into the dynamics of cellular interactions and the effects of (existing) therapies on these dynamics would substantially aid in designing optimal treatment strategies. Acknowledgments This work was supported by Dutch Cancer Foundation grant number UVA 2011-5097 (APK).

10. Yaktapour N, Ubelhart R, Schuler J, et al. Insulin-like growth factor-1 receptor (IGF1R) as a novel target in chronic lymphocytic leukemia. Blood. 2013;122(9):1621-1633. 11. de la Fuente MT, Casanova B, Garcia-Gila M, Silva A, Garcia-Pardo A. Fibronectin interaction with alpha4beta1 integrin prevents apoptosis in B cell chronic lymphocytic leukemia: correlation with Bcl-2 and Bax. Leukemia. 1999;13(2):266-274. 12. Webber J, Yeung V, Clayton A. Extracellular vesicles as modulators of the cancer microenvironment. Semin Cell Dev Biol. 2015;40:27-34. 13. Yeh YY, Ozer HG, Lehman AM, et al. Characterization of CLL exosomes reveals a distinct microRNA signature and enhanced secretion by activation of BCR signaling. Blood. 2015;125(21):3297-3305. 14. Paggetti J, Haderk F, Seiffert M, et al. Exosomes released by chronic lymphocytic leukemia cells induce the transition of stromal cells into cancer-associated fibroblasts. Blood. 2015;126(9):1106-1117. 15. Serra S, Vaisitti T, Audrito V, et al. Adenosine signaling mediates hypoxic responses in the chronic lymphocytic leukemia microenvironment. Blood advances. 2016;1(1):47-61. 16. Os A, Burgler S, Ribes AP, et al. Chronic lymphocytic leukemia cells are activated and proliferate in response to specific T helper cells. Cell Rep. 2013;4(3):566-577. 17. van der Windt GJ, van Bruggen JA, Gessel S, Terpstra S, Eldering E, Kater AP, Impact of signals from the tumor microenvironment on chronic lymphocytic leukemia metabolism. Keystone Cancer Metabolism; 2016. 18. Buschle M, Campana D, Carding SR, Richard C, Hoffbrand AV, Brenner MK. Interferon gamma inhibits apoptotic cell death in B cell chronic lymphocytic leukemia. J Exp Med. 1993;177(1):213-218. 19. Kitada S, Zapata JM, Andreeff M, Reed JC. Bryostatin and CD40-ligand enhance apoptosis resistance and induce expression of cell

20.

21.

22. 23.

24.

25.

26.

27.

28.

survival genes in B-cell chronic lymphocytic leukaemia. Br J Haematol. 1999;106(4):9951004. Tromp JM, Tonino SH, Elias JA, et al. Dichotomy in NF-kappaB signaling and chemoresistance in immunoglobulin variable heavy-chain-mutated versus unmutated CLL cells upon CD40/TLR9 triggering. Oncogene. 2010;29(36):5071-5082. van Attekum MH, Terpstra S, Slinger E, et al. Macrophages confer survival signals via CCR1-dependent translational MCL-1 induction in chronic lymphocytic leukemia. Oncogene. 2017 Feb 13. [Epub ahead of print] Slinger E, Kater AP, Thijssen R, Eldering E. Targeting BCR-independent proliferation of CLL cells. Blood. 2015;126(23):2916. Cui B, Chen L, Zhang S, et al. MicroRNA155 influences B-cell receptor signaling and associates with aggressive disease in chronic lymphocytic leukemia. Blood. 2014;124(4): 546-554. Girbl T, Hinterseer E, Grossinger EM, et al. CD40-mediated activation of chronic lymphocytic leukemia cells promotes their CD44-dependent adhesion to hyaluronan and restricts CCL21-induced motility. Cancer Res. 2013;73(2):561-570. Burgler S, Gimeno A, Parente-Ribes A, et al. Chronic lymphocytic leukemia cells express CD38 in response to Th1 cell-derived IFNgamma by a T-bet-dependent mechanism. J Immunol. 2015;194(2):827-835. Tonino SH, Spijker R, Luijks DM, van Oers MH, Kater AP. No convincing evidence for a role of CD31-CD38 interactions in the pathogenesis of chronic lymphocytic leukemia. Blood. 2008;112(3):840-843. Mackus WJ, Frakking FN, Grummels A, et al. Expansion of CMV-specific CD8+CD45RA+CD27- T cells in B-cell chronic lymphocytic leukemia. Blood. 2003;102(3):1057-1063. Hofbauer JP, Heyder C, Denk U, et al. Development of CLL in the TCL1 transgenic

haematologica | 2017; 102(9)


Role of CLL cells in TME cross-talk

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

mouse model is associated with severe skewing of the T-cell compartment homologous to human CLL. Leukemia. 2011;25(9):1452-1458. Nunes C, Wong R, Mason M, Fegan C, Man S, Pepper C. Expansion of a CD8(+)PD-1(+) replicative senescence phenotype in early stage CLL patients is associated with inverted CD4:CD8 ratios and disease progression. Clin Cancer Res. 2012;18(3):678-687. te Raa GD, Tonino SH, Remmerswaal EB, et al. Chronic lymphocytic leukemia specific T-cell subset alterations are clone-size dependent and not present in monoclonal B lymphocytosis. Leuk Lymphoma. 2012;53(11):2321-2325. Tonino SH, van de Berg PJ, Yong SL, et al. Expansion of effector T cells associated with decreased PD-1 expression in patients with indolent B cell lymphomas and chronic lymphocytic leukemia. Leuk Lymphoma. 2012;53(9):1785-1794. Riches JC, Davies JK, McClanahan F, et al. T cells from CLL patients exhibit features of Tcell exhaustion but retain capacity for cytokine production. Blood. 2013;121(9): 1612-1621. Christopoulos P, Pfeifer D, Bartholome K, et al. Definition and characterization of the systemic T-cell dysregulation in untreated indolent B-cell lymphoma and very early CLL. Blood. 2011;117(14):3836-3846. Kabanova A, Sanseviero F, Veronica C, et al. Human cytotoxic T lymphocytes form dysfunctional immune synapses with B cells characterized by non-polarized lytic granule release. Cell Rep. 2016;15(1):9-18. McClanahan F, Riches JC, Miller S, et al. Mechanisms of PD-L1/PD-1-mediated CD8 T-cell dysfunction in the context of agingrelated immune defects in the Emicro-TCL1 CLL mouse model. Blood. 2015;126(2):212221. Simonetti G, Bertilaccio MT, Ghia P, Klein U. Mouse models in the study of chronic lymphocytic leukemia pathogenesis and therapy. Blood. 2014;124(7):1010-1019. Ramsay AG, Johnson AJ, Lee AM, et al. Chronic lymphocytic leukemia T cells show impaired immunological synapse formation that can be reversed with an immunomodulating drug. J Clin Invest. 2008;118(7):24272437. Siska PJ, van der Windt GJ, Kishton RJ, et al. Suppression of glut1 and glucose metabolism by decreased Akt/mTORC1 signaling drives T cell impairment in B cell leukemia. J Immunol. 2016;197(6):2532-2540. Chang CH, Qiu J, O'Sullivan D, et al. Metabolic competition in the tumor microenvironment is a driver of cancer progression. Cell. 2015;162(6):1229-1241. White GE, Iqbal AJ, Greaves DR. CC chemokine receptors and chronic inflammation--therapeutic opportunities and pharmacological challenges. Pharmacol Rev. 2013;65(1):47-89. Burger JA, Quiroga MP, Hartmann E, et al. High-level expression of the T-cell chemokines CCL3 and CCL4 by chronic lymphocytic leukemia B cells in nurselike cell cocultures and after BCR stimulation. Blood. 2009;113(13):3050-3058. Hartmann EM, Rudelius M, Burger JA, Rosenwald A. CCL3 chemokine expression by chronic lymphocytic leukemia cells orchestrates the composition of the microenvironment in lymph node infiltrates. Leuk Lymphoma. 2016;57(3):563-571. Ramsay AG, Evans R, Kiaii S, Svensson L,

haematologica | 2017; 102(9)

44.

45.

46.

47.

48.

49.

50. 51. 52.

53.

54.

55.

56.

57.

58.

Hogg N, Gribben JG. Chronic lymphocytic leukemia cells induce defective LFA-1-directed T-cell motility by altering Rho GTPase signaling that is reversible with lenalidomide. Blood. 2013;121(14):2704-2714. Born WK, Reardon CL, O'Brien RL. The function of gammadelta T cells in innate immunity. Curr Opin Immunol. 2006;18(1): 31-38. Coscia M, Vitale C, Peola S, et al. Dysfunctional Vgamma9Vdelta2 T cells are negative prognosticators and markers of dysregulated mevalonate pathway activity in chronic lymphocytic leukemia cells. Blood. 2012;120(16):3271-3279. de Weerdt I, Terpstra S, Hofland T, et al. Chronic lymphocytic leukemia (CLL) cells are susceptible to γδ-T cell mediated killing, provided CLL-derived γδ-T cell dysfunction Can be Reversed. Blood. 2015;126(23):2914. Halvorsen EC, Mahmoud SM, Bennewith KL. Emerging roles of regulatory T cells in tumour progression and metastasis. Cancer Metastasis Rev. 2014;33(4):1025-1041. Piper KP, Karanth M, McLarnon A, et al. Chronic lymphocytic leukaemia cells drive the global CD4+ T cell repertoire towards a regulatory phenotype and leads to the accumulation of CD4+ forkhead box P3+ T cells. Clin Exp Immunol. 2011;166(2):154-163. Lad DP, Varma S, Varma N, Sachdeva MU, Bose P, Malhotra P. Regulatory T-cell and Thelper 17 balance in chronic lymphocytic leukemia progression and autoimmune cytopenias. Leuk Lymphoma. 2015;56(8):2424-2428. Erb P, Feldmann M. Role of macrophages in in vitro induction of T-helper cells. Nature. 1975;254(5498):352-354. Lewis CE, Pollard JW. Distinct role of macrophages in different tumor microenvironments. Cancer Res. 2006;66(2):605-612. Hanna BS, McClanahan F, Yazdanparast H, et al. Depletion of CLL-associated patrolling monocytes and macrophages controls disease development and repairs immune dysfunction in vivo. Leukemia. 2016;30(3):570579. Galletti G, Scielzo C, Barbaglio F, et al. Targeting macrophages sensitizes chronic lymphocytic leukemia to apoptosis and inhibits disease progression. Cell Rep. 2016;14(7):1748-1760. Gustafson MP, Abraham RS, Lin Y, et al. Association of an increased frequency of CD14+ HLA-DR lo/neg monocytes with decreased time to progression in chronic lymphocytic leukaemia (CLL). Br J Haematol. 2012;156(5):674-676. Burger JA, Tsukada N, Burger M, Zvaifler NJ, Dell'Aquila M, Kipps TJ. Blood-derived nurse-like cells protect chronic lymphocytic leukemia B cells from spontaneous apoptosis through stromal cell-derived factor-1. Blood. 2000;96(8):2655-2663. van Attekum MHA, Terpstra S, Reinen E, Kater AP, Eldering E. Macrophage-mediated chronic lymphocytic leukemia cell survival is independent of APRIL signaling. Cell Death Discov. 2016;2:16020. Tsukada N, Burger JA, Zvaifler NJ, Kipps TJ. Distinctive features of "nurselike" cells that differentiate in the context of chronic lymphocytic leukemia. Blood. 2002;99(3):10301037. Lascano V, Guadagnoli M, Schot JG, et al. Chronic lymphocytic leukemia disease progression is accelerated by APRIL-TACI interaction in the TCL1 transgenic mouse model. Blood. 2013;122(24):3960-3963.

59. Saulep-Easton D, Vincent FB, Quah PS, et al. The BAFF receptor TACI controls IL-10 production by regulatory B cells and CLL B cells. Leukemia. 2016;30(1):163-172. 60. Hua C, Audo R, Yeremenko N, et al. A proliferation inducing ligand (APRIL) promotes IL-10 production and regulatory functions of human B cells. J Autoimmun. 2016;73:64-72. 61. van Attekum MH, Kater AP, Eldering E. The APRIL paradox in normal versus malignant B cell biology. Cell Death Dis. 2016;7(6):e2276. 62. van Attekum MHA, Terpstra S, Reinen E, et al. The T-cell/CLL/macrophage triad shapes a supportive tumor microenvironment in CLL. Blood. 2015;126(23):1715. 63. Audrito V, Serra S, Brusa D, et al. Extracellular nicotinamide phosphoribosyltransferase (NAMPT) promotes M2 macrophage polarization in chronic lymphocytic leukemia. Blood. 2015;125(1):111123. 64. Jitschin R, Braun M, Buttner M, et al. CLLcells induce IDOhi CD14+HLA-DRlo myeloid-derived suppressor cells that inhibit T-cell responses and promote TRegs. Blood. 2014;124(5):750-760. 65. Bhattacharya N, Diener S, Idler IS, et al. Nurse-like cells show deregulated expression of genes involved in immunocompetence. Br J Haematol. 2011;154(3):349-356. 66. Maffei R, Bulgarelli J, Fiorcari S, et al. The monocytic population in chronic lymphocytic leukemia shows altered composition and deregulation of genes involved in phagocytosis and inflammation. Haematologica. 2013;98(7):1115-1123. 67. Orsini E, Guarini A, Chiaretti S, Mauro FR, Foa R. The circulating dendritic cell compartment in patients with chronic lymphocytic leukemia is severely defective and unable to stimulate an effective T-cell response. Cancer Res. 2003;63(15):44974506. 68. Jia L, Clear A, Liu FT, et al. Extracellular HMGB1 promotes differentiation of nurselike cells in chronic lymphocytic leukemia. Blood. 2014;123(11):1709-1719. 69. Qorraj M, Bruns H, Bottcher M, et al. The PD-1/PD-L1 axis contributes to immune metabolic dysfunctions of monocytes in chronic lymphocytic leukemia. Leukemia. 2017;31(2):470-478. 70. Toniolo PA, Liu S, Yeh JE, Ye DQ, Barbuto JA, Frank DA. Deregulation of SOCS5 suppresses dendritic cell function in chronic lymphocytic leukemia. Oncotarget. 2016;7(29):46301-46314. 71. Gautam S, Fatehchand K, Elavazhagan S, et al. Reprogramming nurse-like cells with Interferon-gamma to interrupt chronic lymphocytic leukemia cell survival. J Biol Chem. 2016;291(27):14356-14362. 72. Reinart N, Nguyen PH, Boucas J, et al. Delayed development of chronic lymphocytic leukemia in the absence of macrophage migration inhibitory factor. Blood. 2013;121(5):812-821. 73. Malhotra D, Fletcher AL, Turley SJ. Stromal and hematopoietic cells in secondary lymphoid organs: partners in immunity. Immunol Rev. 2013;251(1):160-176. 74. Raffaghello L, Vacca A, Pistoia V, Ribatti D. Cancer associated fibroblasts in hematological malignancies. Oncotarget. 2015;6(5): 2589-2603. 75. Mittal AK, Chaturvedi NK, Rai KJ, et al. Chronic lymphocytic leukemia cells in a lymph node microenvironment depict molecular signature associated with an

1475


M.HA van Attekum et al.

76.

77.

78.

79.

80.

81.

82.

83.

1476

aggressive disease. Mol Med. 2014;20:290301. Jitschin R, Braun M, Qorraj M, et al. Stromal cell-mediated glycolytic switch in CLL cells involves Notch-c-Myc signaling. Blood. 2015;125(22):3432-3436. Zhang W, Trachootham D, Liu J, et al. Stromal control of cystine metabolism promotes cancer cell survival in chronic lymphocytic leukaemia. Nat Cell Biol. 2012;14(3):276-286. Valsecchi R, Coltella N, Belloni D, et al. HIF1alpha regulates the interaction of chronic lymphocytic leukemia cells with the tumor microenvironment. Blood. 2016;127(16): 1987-1997. Lutzny G, Kocher T, Schmidt-Supprian M, et al. Protein kinase c-beta-dependent activation of NF-kappaB in stromal cells is indispensable for the survival of chronic lymphocytic leukemia B cells in vivo. Cancer Cell. 2013;23(1):77-92. Ding W, Nowakowski GS, Knox TR, et al. Bi-directional activation between mesenchymal stem cells and CLL B-cells: implication for CLL disease progression. Br J Haematol. 2009;147(4):471-483. Ding W, Knox TR, Tschumper RC, et al. Platelet-derived growth factor (PDGF)PDGF receptor interaction activates bone marrow-derived mesenchymal stromal cells derived from chronic lymphocytic leukemia: implications for an angiogenic switch. Blood. 2010;116(16):2984-2993. Borge M, Nannini PR, Morande PE, et al. CXCL12 is a costimulator for CD4+ T cell activation and proliferation in chronic lymphocytic leukemia patients. Cancer Immunol Immunother. 2013;62(1):113-124. Gabrilovich DI, Ostrand-Rosenberg S, Bronte V. Coordinated regulation of myeloid cells by tumours. Nat Rev Immunol. 2012;12(4):253-268.

84. Ferrer G, Yan X-J, Franca B, et al. Chronic Lymphocytic leukemia patients and EÂľTCL1 mice share a phenotype of functional granulocyte-like and dysfunctional monocyte-like myeloid derived suppressor cells. Blood. 2015;126(23):614. 85. Seke Etet PF, Vecchio L, Nwabo Kamdje AH. Interactions between bone marrow stromal microenvironment and B-chronic lymphocytic leukemia cells: any role for Notch, Wnt and Hh signaling pathways? Cell Signal. 2012;24(7):1433-1443. 86. Zucchetto A, Caldana C, Benedetti D, et al. CD49d is overexpressed by trisomy 12 chronic lymphocytic leukemia cells: evidence for a methylation-dependent regulation mechanism. Blood. 2013;122(19):3317-3321. 87. Malavasi F, Deaglio S, Damle R, Cutrona G, Ferrarini M, Chiorazzi N. CD38 and chronic lymphocytic leukemia: a decade later. Blood. 2011;118(13):3470-3478. 88. Ren L, Campbell A, Fang H, et al. Analysis of the effects of the Bruton's tyrosine kinase (Btk) inhibitor ibrutinib on monocyte Fcgamma receptor (FcgammaR) function. J Biol Chem. 2016;291(6):3043-3052. 89. Da Roit F, Engelberts PJ, Taylor RP, et al. Ibrutinib interferes with the cell-mediated anti-tumor activities of therapeutic CD20 antibodies: implications for combination therapy. Haematologica. 2015;100(1):77-86. 90. Nguyen PH, Fedorchenko O, Rosen N, et al. LYN Kinase in the tumor microenvironment is essential for the progression of chronic lymphocytic leukemia. Cancer Cell. 2016;30(4):610-622. 91. Stiff A, Trikha P, Wesolowski R, et al. Myeloid-derived suppressor cells express Bruton's tyrosine kinase and can be depleted in tumor-bearing hosts by ibrutinib treatment. Cancer Res. 2016;76(8):2125-2136. 92. Dubovsky JA, Beckwith KA, Natarajan G, et al. Ibrutinib is an irreversible molecular

inhibitor of ITK driving a Th1-selective pressure in T lymphocytes. Blood. 2013;122 (15):2539-2549. 93. Gunderson AJ, Kaneda MM, Tsujikawa T, et al. Bruton tyrosine kinase-dependent immune cell cross-talk drives pancreas cancer. Cancer Discov. 2016;6(3):270-285. 94. Fraietta JA, Beckwith KA, Patel PR, et al. Ibrutinib enhances chimeric antigen receptor T-cell engraftment and efficacy in leukemia. Blood. 2016;127(9):1117-1127. 95. Herman SE, Gordon AL, Wagner AJ, et al. Phosphatidylinositol 3-kinase-delta inhibitor CAL-101 shows promising preclinical activity in chronic lymphocytic leukemia by antagonizing intrinsic and extrinsic cellular survival signals. Blood. 2010;116(12):2078-2088. 96. Mouchemore KA, Sampaio NG, Murrey MW, Stanley ER, Lannutti BJ, Pixley FJ. Specific inhibition of PI3K p110delta inhibits CSF-1-induced macrophage spreading and invasive capacity. FEBS J. 2013;280(21):52285236. 97. Kaneda MM, Messer KS, Ralainirina N, et al. PI3Kgamma is a molecular switch that controls immune suppression. Nature. 2016;539(7629):437-442. 98. Gandhi AK, Kang J, Havens CG, et al. Immunomodulatory agents lenalidomide and pomalidomide co-stimulate T cells by inducing degradation of T cell repressors Ikaros and Aiolos via modulation of the E3 ubiquitin ligase complex CRL4(CRBN.). Br J Haematol. 2014;164(6):811-821. 99. McClanahan F, Hanna B, Miller S, et al. PDL1 checkpoint blockade prevents immune dysfunction and leukemia development in a mouse model of chronic lymphocytic leukemia. Blood. 2015;126(2):203-211. 100. Wong D, Kandagatla P, Korz W, Chinni SR. Targeting CXCR4 with CTCE-9908 inhibits prostate tumor metastasis. BMC Urol. 2014;14:12.

haematologica | 2017; 102(9)


ARTICLE

Red Cell Biology & its Disorders

Hemoglobin concentration, total hemoglobin mass and plasma volume in patients: implications for anemia

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

James M. Otto,1 James O.M. Plumb,2,3,4 Eleri Clissold,2,3,4 Shriya B. Kumar,2,3,4 Denis J. Wakeham,5 Walter Schmidt,6 Michael P.W. Grocott,2,3,4 Toby Richards1 and Hugh E. Montgomery7

Division of Surgery and Interventional Science, University College London, UK; Anaesthesia and Critical Care Research Unit, University Hospital Southampton NHS Foundation Trust, Southampton, UK; 3Integrative Physiology and Critical Illness Group, Clinical and Experimental Sciences, Faculty of Medicine, University of Southampton, UK; 4 Critical Care Research Area, Southampton NIHR Respiratory Biomedical Research Unit, Southampton, UK; 5School of Sport, Physiology and Health Group, Cardiff Metropolitan University, UK; 6Department of Sports Medicine/Sports Physiology, University of Bayreuth, Germany and 7Centre for Human Health and Performance/ Institute of Sport, Exercise and Health, University College London, and NIHR University College London Hospitals Biomedical Research Centre, UK 1 2

Haematologica 2017 Volume 102(9):1477-1485

ABSTRACT

I

n practice, clinicians generally consider anemia (circulating hemoglobin concentration < 120 g.l-1 in non-pregnant females and < 130 g.l-1 in males) as due to impaired hemoglobin synthesis or increased erythrocyte loss or destruction. Rarely is a rise in plasma volume relative to circulating total hemoglobin mass considered as a cause. But does this matter? We explored this issue in patients, measuring hemoglobin concentration, total hemoglobin mass (optimized carbon monoxide rebreathing method) and thereby calculating plasma volume in healthy volunteers, surgical patients, and those with inflammatory bowel disease, chronic liver disease or heart failure. We studied 109 participants. Hemoglobin mass correlated well with its concentration in the healthy, surgical and inflammatory bowel disease groups (r=0.687-0.871, P<0.001). However, they were poorly related in liver disease (r=0.410, P=0.11) and heart failure patients (r=0.312, P=0.16). Here, hemoglobin mass explained little of the variance in its concentration (adjusted R2=0.109 and 0.052; P=0.11 and 0.16), whilst plasma volume did (R2 change 0.724 and 0.805 in heart and liver disease respectively, P<0.0001). Exemplar patients with identical (normal or raised) total hemoglobin masses were diagnosed as profoundly anemic (or not) depending on differences in plasma volume that had not been measured or even considered as a cause. The traditional inference that anemia generally reflects hemoglobin deficiency may be misleading, potentially resulting in inappropriate tests and therapeutic interventions to address ‘hemoglobin deficiency’ not ‘plasma volume excess’. Measurement of total hemoglobin mass and plasma volume is now simple, cheap and safe, and its more routine use is advocated. Introduction Anemia is defined as a reduction in the circulating concentration of hemoglobin ([Hb]) to < 120 g.l-1 in non-pregnant females and < 130 g.l-1 in males.1 Such reductions can result from the destruction or loss of erythrocytes or a failure of their production, or from impaired hemoglobin synthesis. Given its diverse etiology, anemia is common, affecting over 1.6 billion people worldwide,2 and is associated with impaired functional capacity, reduced quality of life,3 and poorer outcome in diverse disease states.4-9 haematologica | 2017; 102(9)

Correspondence: toby.richards@ucl.ac.uk

Received: March 24, 2017. Accepted: June 6, 2017 Pre-published: June 8, 2017 doi:10.3324/haematol.2017.169680 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1477 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1477


J.M. Otto et al.

For these reasons, circulating hemoglobin concentrations are routinely measured in clinical practice. Once anemia is identified, the cause of impaired hemoglobin synthesis or erythrocytosis, or of increased red cell loss or destruction, is often sought and treatment (either of the underlying cause, or through the administration of packed donor red cells) initiated. However, it is now becoming clear that this approach may be somewhat simplistic. The concentration of circulating hemoglobin will depend not just on the total circulating quantity of hemoglobin (total hemoglobin mass, (tHb-mass), but also on the volume of plasma (plasma volume, PV) in which it is suspended. In clinical practice these factors are not routinely considered separately, or measured, and even in experienced hands their estimation is not trivial. However, such assessment may be important if the drivers of an altered [Hb], and the appropriate therapeutic response, are to be truly understood. Thus, the circulating hemoglobin concentration in athletes matches that in sedentary individuals, meaning that a contribution of increased red cell oxygen carriage to elite performance was largely dismissed. Then, in 2001, Heinicke and colleagues demonstrated that tHb-mass was increased by some 35% in elite endurance athletes, [Hb] matching that in the untrained only because PV was expanded to a similar degree.10 Likewise, alterations in the relationship between plasma volume and tHb-mass might strongly influence [Hb] in disease states. Anemia is common in diseases such as cancer,11 inflammatory bowel disease (IBD),12 chronic heart failure (CHF),8 chronic kidney disease (CKD)13 and chronic liver disease (CLD).14 Traditionally, this has been considered the result of a reduced tHb-mass. But a low [Hb] might also be found when hemoglobin synthesis, erythrocytosis and tHb-mass are all entirely normal (or even high), if PV is disproportionately expanded. This can occur through disease-related changes in global water balance or distribution of body water. Thus, patients with IBD might face enteric blood loss, suppressed hemoglobin synthesis or anemia of chronic inflammation; CLD patients may likewise lose blood, have an expanded circulating PV due to hyperaldosteronism, or face fluid shifts as a result of raised portal venous pressure or hypoalbuminemia; and patients with chronic heart failure may suffer an increase in circulating volume due to factors including increased renin-angiotensin-aldosterone axis activity.15 In contrast, contractions in PV caused by pharmacotherapy may mask a fall in tHb-mass by maintaining [Hb].16 The extent to which this is true has not been addressed, largely due to historical methodological limitations: circulating red cell volume (RCV, ml) or PV has generally been determined through radiolabeling of red blood cells (RBCs) or albumin, respectively.17 Such techniques are costly and time consuming, and not without risk, and are thus not routinely deployed unless in special circumstances (this method is recommended for disease diagnosis by the Polycythemia Vera Study Group).18 Such barriers may be overcome through the use of an ‘optimized carbon monoxide (CO) rebreathing’ (oCOR) method:19 inhaled CO binds avidly to circulating hemoglobin, and the concentration of the resultant carboxyhemoglobin (COHb) complex can be readily measured. Knowing the quantity of absorbed CO, tHb-mass can be measured and, knowing [Hb], PV calculated. The method is cheap, simple and safe to use, but has only rarely been 1478

applied in the clinical setting. Thus, the relative contributions of PV and tHb-mass to measured [Hb] across disease states have not been described. Herein we sought to address this issue.

Methods This was a prospective, observational clinical study. We studied five groups: healthy volunteers (HV), preoperative patients awaiting major surgery and those suffering IBD, as well as patients in whom alterations in PV might more commonly occur (those with CLD or CHF). Each participant was studied on 1 occasion between February 2015 and May 2016.

Optimized Carbon Monoxide Rebreathing Method (oCOR) The use of CO to determine tHb-mass was first proposed in the late 1800s, with refined techniques being published 100 years later.20 In 2005, Schmidt and Prommer reported a simpler and faster technique (described in detail below) which also required less blood sampling.19 It was applied almost exclusively, however, in the fields of athletic physiology, and thus failed to come to the attention of the bulk of the broader clinical/medical community. tHb-mass was determined using the validated oCOR method described in detail by Schmidt and Prommer.19 In brief, COHb concentration in blood was measured before and after 2 min rebreathing a known CO volume (0.5 to 1.0 ml.kg-1 in this study depending on sex). Each participant was seated for 15 min to allow stabilization of PV, after which a mouthpiece connected them via a container of ‘soda lime’~10g (carbon dioxide scrubber) to a spirometer (Spico-CO Respirations-Applikator, Blood Tec, Germany) and a 3 liter anesthetic bag pre-filled with 100% oxygen. The patient exhaled to residual volume, breathed in the CO dose via the spirometer, held their breath for 10s, then continued normal breathing into the closed circuit via the spirometer for 1 min 50s. The participant then exhaled to residual volume, this exhaled volume being collected and analyzed to quantify the CO not absorbed into the bloodstream. Disconnected from the mouthpiece, participants finally fully exhaled to residual volume into a CO gas analyzer (Dräger Pac 7000, Drägerwerk AG & Co. KGaA, Germany) before and at 4 min after CO rebreathing, in order to determine the CO concentration exhaled after disconnecting the patient from the spirometer, that will also have not been absorbed into the blood.

Statistical Analysis Statistical analysis was performed using SPSS Statistics (Version 23.0 for Apple Macintosh, Chicago, IL, USA). Values are presented as mean ± standard deviation (SD), unless otherwise stated. Median and interquartile range (IQR) are reported when variables were not normally distributed. Categorical variables are presented as frequency (%). Pearson’s correlation coefficient assessed the relationship between [Hb] and tHb-mass, allowing adjustment for PV (ml). Linear regression assessed the proportion of variance in tHb-mass explained by [Hb], allowing adjustment for PV (ml). In both correlation and regression analyses [Hb] and tHb-mass are expressed in g.l-1 and grams, respectively, and PV in mls. This is also the case elsewhere, unless otherwise stated. Differences across sub-groups were assessed by one-way analysis of variance (ANOVA), prior to which the assumption of normality was tested by the Levene’s test for homogeneity of variances. Where homogeneity of variance was verified, being the case for [Hb], hematocrit (Hct), tHb-mass (g), PV (%) and weight, a one-way ANOVA was performed, with post hoc comparisons by haematologica | 2017; 102(9)


[Hb], tHb-mass and PV in anemia

A

B

C

D

Figure 1. Hematological variables in healthy volunteers and patient sub-groups. Hct (%), hematocrit percentage (A); [Hb], hemoglobin concentration (B); tHb-mass, total hemoglobin mass (g & g.kg-1) (C & D). *P<0.0001 for Hct and [Hb] in LD patients compared to all other groups. No differences in tHb-mass (g or g.kg-1) between groups. Data expressed as mean (± standard deviation error bars). HV: healthy volunteers; IBD: inflammatory bowel disease; HF: heart failure; LD: liver disease.

Gabriel’s test due to the slightly different sample sizes across subgroups. When homogeneity of variance was violated, as was the case for PV (ml & ml.kg-1), age and tHb-mass (g.kg-1), a Welch ANOVA21 was used with post hoc comparisons made by the Games-Howell test, as this does not rely on the assumption of equal variances.22 All tests were two-sided with statistical significance being accepted as a P-value of <0.05.

Power calculation The power calculation was based on the study by Hinrichs et al.23 and was performed using G*3 Power version 3.1.9.2.24 According to Hinrichs and colleagues, the relationship (expressed as Pearson’s correlation coefficient) between [Hb] and tHb-mass was r=0.59, P<0.05. Based on this, using a two-tailed correlation: bivariate normal model, we calculated that 21 patients would provide 80% power at the 5% significance level to detect a correlation of at least r=0.59 between [Hb] and tHb-mass. Given that 5 groups were studied, a total of 105 participants were required.

and sub-groups. Sixteen patients were tested at Southampton General Hospital, 90 at University College London Hospital (UCLH; including HV) and 3 at the Royal Free Hospital (RFH). Surgical specialties are shown in the Online Supplementary Table S1, with patient characteristics and etiology of disease for IBD patients in the Online Supplementary Table S2. Patient characteristics, medications and etiology of CLD and CHF are shown in the Online Supplementary Table S3 and Online Supplementary Table S4, respectively. There were no differences in weight between sub-groups. HV (n=21) were younger compared to all patient groups (P<0.0001) with CHF patients (n=22) being older than IBD (n=21, P=0.001), surgical (n=28, P=0.008) and CLD patients (n=16, P=0.002). Figure 1A-D shows hematological variables and Figure 2A,B displays PVs across different sub-groups.

Hemoglobin concentration

Ethical approval was granted by the London - Camden and Kings Cross Research Ethics Committee (REC reference: 13/LO/1902). Written informed consent was obtained from all participants.

Hemoglobin concentration was 128.4 ± 18.1 g.l-1 in the subjects overall. Across sub-groups, CLD patients had lower [Hb] when compared to HV and other disease groups (P<0.0001, Figure 1B). Anemia prevalence (n=34 [39%]) varied across disease sub-groups (CLD 15 [94%], CHF 12 [55%], IBD 2 [9%], surgical 5 [18%]). Amongst HV, 5 (24%, all female) were anemic.

Results

Total hemoglobin mass

One hundred and nine participants (61% male: mean (IQR) age 52 (36-64) years) consented to take part in the study. The Online Supplementary Figure S1 shows a Consolidated Standards of Reporting Trials (CONSORT) flow diagram indicating included and excluded patients

Mean ± SD tHb-mass was 669 ± 181 grams (8.5 ± 1.9 g.kg-1 body mass) in subjects overall, with no statistically significant differences across sub-groups (Figure 1C,D). tHb-mass was higher in males than females: (758 ± 152 vs. 533 ± 132 g, P<0.0001; and 8.9 ± 1.8 vs. 7.9 ± 2.0 g.kg-1, P=0.006, respectively).

Ethical approval

haematologica | 2017; 102(9)

1479


J.M. Otto et al.

Plasma volume -1

Mean ± SD PV was 3667 ± 1020 ml (47.1 ± 11.3 ml.kg ) in subjects overall. PV was higher in males than females (4040 ± 1038 vs. 3093 ± 672 ml, respectively, P<0.0001), but not when weight adjusted (47.6 ± 11.3 vs. 46.2 ± 11.5 ml.kg-1, P=0.556). It also differed across disease groups, being expanded and more varied in CLD (4965 ± 1447 ml) compared to HV (3429 ± 538 ml, P=0.006), IBD (3202 ± 653 ml, P=0.002) and surgical patients (3297 ± 590 ml, P=0.003), but not CHF (3883 ± 953 ml, P=0.100), see Figure 2A. Adjusted for body weight, PV was similarly expanded in CLD (59.1 ± 16.0 ml.kg-1) when compared to IBD (42.5 ± 8.3 ml.kg-1, P=0.008) and surgical patients (41.3

A

± 7.8 ml.kg-1, P=0.004), but again was similar to that in CHF patients (48.6 ± 9.2 ml.kg-1, P=0.172) or HV (49.1 ± 7.9 ml.kgl-1, P=0.191), see Figure 2B. Hemoglobin concentration was influenced by the degree of PV expansion (Figure 3A), being lower in patients with severe PV expansion (n= 46, 119.0 ± 17.7 g.l-1) than mild to moderate PV expansion (n=36, 131.1 ± 13.8 g.l-1, P=0.005), and normal PV (n=24, 140.7 ± 15.5 g.l-1, P<0.0001), but not PV contraction (n=3, 143.8 ± 11.8 g.l-1, P=0.139).

Relationships between hemoglobin concentration and total hemoglobin mass In the study cohort as a whole, [Hb] (g.l-1) correlated

B

Figure 2. Plasma volume in healthy volunteers and patient sub-groups. Plasma volume (ml & ml.kg-1) (A & B). (A) PV (ml) between LD and HV, *P=0.006, PV (ml) between LD and IBD, **P=0.002 and PV (ml) between LD and surgical patients, †P=0.003; (B) PV (ml kg-1) between LD and IBD, *P=0.008, PV between LD and surgical patients, **P=0.004. Data expressed as mean (± standard deviation error bars). HV: healthy volunteers; IBD: inflammatory bowel disease; HF: heart failure; LD: liver disease; PV: plasma volume.

Table 1. Hematological variables in anemic and non-anemic participants.

Variable -1

[Hb] (g.l ) Hct (%) tHb-mass (g) tHb-mass (g.kg-1) BV (ml) BV (ml.kg-1) PV (ml) PV (ml.kg-1) PV (%) RCV (ml) RCV (ml.kg-1) MCV (fl) MCH (pg) MCHC (g.l-1) RDW (%) Creatinine ( mol.l-1) Albumin (g.l-1)

Anemic (n=39)

Non-anemic (n=70)

P

109.6 ± 12.5 35.0 ± 5.9 594 ± 192 7.7 ± 1.9 5978 ± 1837 77.3 ± 16.7 4083 ± 1351 52.6 ± 12.1 68% 1894 ± 692 24.7 ± 7.1 88.0 ± 6.2 29.3 ± 2.9 333.1 ± 18.0 14.5 ± 1.7 114.9 ± 117.4 37.4 ± 8.7

138.9 ± 10.6 42.4 ± 3.0 711 ± 162 9.0 ± 1.8 5606 ± 1146 71.6 ± 14.8 3434 ± 686 44.0 ± 9.6 61% 2171 ± 501 27.5 ± 5.7 91.0 ± 6.6 30.1 ± 2.2 330.5 ± 12.2 13.7 ± 1.1 84.3 ± 58.2 43.7 ± 4.9

< 0.0001 < 0.0001 0.001 0.001 0.257 0.066 0.007 < 0.0001 < 0.0001 0.018 0.023 0.038 0.218 0.477 0.024 0.166 0.001

Anemia defined according to World Health Organization criteria ([Hb] <130 g.l-1 in men and <120 g.l-1 in women. Data are presented as mean ± SD, or frequency (%). [Hb]; Hemoglobin concentration; Hct; hematocrit; tHb-mass; total hemoglobin mass; BV: blood volume; PV: plasma volume; RCV: red cell volume; MCV: mean corpuscular volume; MCH: mean corpuscular hemoglobin; MCHC: mean corpuscular hemoglobin concentration; RDW: red cell distribution width.

1480

haematologica | 2017; 102(9)


[Hb], tHb-mass and PV in anemia

Linear regression models

with tHb-mass (g) (r=0.500, P<0.0001, n=109), this being true in both males (r=0.452, P<0.0001) and females (r=0.462, P<0.0001). Whilst true in HV, IBD and surgical patients (r=0.871, P<0.0001; r=0.687, P<0.0001; and r=0.763, P<0.0001, respectively; Figure 4A-E) this was not the case in patients with CLD or CHF (r=0.410, P=0.114 and r=0.312, P=0.157, respectively). Whilst consistently statistically significant, the strength of the relationship between [Hb] and tHb-mass weakened as PV rose, with r-values falling from 0.94 to 0.91, 0.86 and 0.57 for those with low (n=3), normal (n=24), mild-moderately expanded (n=36) and severely expanded (n=46) PV, respectively, (P=0.216 for the 3 with low PV, and P<0.0001 for the others). Sub-group analysis is hampered by small numbers, but data are presented in the Online Supplementary Table S5 for completeness.

In the whole population, tHb-mass explained 25% of the variance in [Hb] (adjusted R2=0.250, P<0.0001). However, tHb-mass explained different amounts of the variance in [Hb] across patient groups, adjusted R2 for HV, surgical and IBD patients being 0.746, 0.565 and 0.446, respectively, (P<0.0001 in all cases). Of particular note, tHb-mass did not significantly explain variance in [Hb] in the 2 patient groups most likely to suffer expanded PV and shifts in fluid, CLD (adjusted R2=0.109, P=0.114) or CHF patients (adjusted R2=0.052, P=0.157). In keeping, PV independently accounted for a greater proportion of the variance in [Hb] in these groups (R2 change 0.724 in CHF and 0.805 in CLD) than in HV (0.192), surgical patients (0.374) or IBD patients (0.479; P<0.0001 in all cases).

Hematological variables by anemia status Total hemoglobin mass, plasma volume and hemoglobin concentration on an individual level

In the 39 anemic subjects (mean ± SD [Hb] 109.6 ± 12.5 g.l-1), when compared to the 70 non-anemic participants ([Hb] 138.9 ± 10.6 g.l-1), tHb-mass (g & g.kg-1) was significantly lower (594 ± 192 vs. 711 ± 162 g, P=0.001; 7.7 ± 1.9 vs. 9.0 ± 1.8 g.kg-1, P=0.001). However, PV (ml, and ml.kg-1) was also significantly higher in anemic subjects [(4083 ± 1351 vs. 3434 ± 686 ml, P=0.007; 52.6 ± 12.2 vs. 44.0 ± 9.6 ml.kg-1, P<0.0001 (Table 1)].

The data presented thus suggest that, at an individual level, [Hb] was not a good guide to tHb-mass, due to its being strongly influenced by PV. This is illustrated in Figure 5A-E, which shows data from individual participants ranked by weight-adjusted tHb-mass from smallest to largest with corresponding PV (ml.kg-1) and [Hb] (g.l-1) in HV (A), IBD (B), surgical (C), CLD (D) and CHF (E) patients, respectively. These show that patients who share a very similar tHb-mass may exhibit markedly different [Hb] due to differences in PV. For example, in patients with IBD (Figure 5B), patient numbers 17 and 18 have a very similar tHb-mass (9.2 g.kg-1 and 9.3 g.kg-1, respectively), yet 1 is defined as having a high normal [Hb] (161 g.l1 ) and the other as being anemic ([Hb] 107 g.l-1), due to substantial differences in PV (37.2 ml.kg-1 vs. 65.7 ml.kg-1). Similarly, in CLD patients (Figure 5E), tHb-mass in patient numbers 2 and 3 are the same (5.2 g.kg-1) but the first is considered to be mildly anemic ([Hb] 110 g.l-1) while the second is deemed markedly anemic ([Hb] 69 g.l-1) due to a relatively raised PV in the latter (36.5 vs. 67.8 ml.kg-1, respectively).

A

Discussion We have studied the contribution of tHb-mass and PV to differences in the concentration of circulating hemoglobin. We have done so in HV and across a variety of disease states, selected to be more or less likely influenced by intravascular fluid status. By doing so, we have shown that: (i) variation in PV contributes significantly to variation in hemoglobin concentration, (ii) this contribution is greater in cases of CLD or heart failure, in other words diseases in which changes in total body water and in its distribution are more likely to occur, and (iii) perhaps most importantly, this may lead to some individuals being diag-

B

Figure 3. Hemoglobin concentration and hematocrit in all patients categorized by plasma volume status. Plasma volume contraction (n=3) was classified as > minus 8% from expected norms. Normal PV (n=24) was classified as derived PV within ± 8% of the expected normal volume on an individual level. Mild to moderate volume expansion (n=36) was considered >8% to <25% deviation from expected norms, and severe PV expansion (n= 46) as >25% of the expected normal volume. *P=0.005 for [Hb] in severe PV expansion vs. mild to moderate, †P<0.0001 for [Hb] in severe PV expansion vs. normal PV. Data expressed as mean (± standard deviation error bars). PV: plasma volume; [Hb]: hemoglobin concentration (g.l-1); Hct (%): hematocrit percentage.

haematologica | 2017; 102(9)

1481


J.M. Otto et al. Table 2. Anemia classification based on hemoglobin concentration, red cell volume and plasma volume in all patients (n=109).

Anemia status Non anemic

Anemic

Normal PV Normal RCV RCV deficit RCV excess Total Normal RCV RCV deficit RCV excess Total

Mild to moderate PV expansion

Severe PV expansion

PV contraction

Total

8 1 17 26 5 4 1 10

1 0 19 20 5 8 13 26

1 2 0 3 0 0 0 0

22 7 41 70 11 15 14 39

12 4 5 21 1 2 0 3

Normal PV and RCV were classified as derived volumes from measured tHb-mass within ± 8% of the expected normal volumes on an individual level. Mild to moderate volume expansion was considered >8% to <25% deviation from expected norms, and severe as >25% of the expected normal volume. Anemia defined according to World Health Organization criteria ([Hb] <130 g.l-1 in men and <120 g.l-1 in women. Values are expressed as counts. PV: plasma volume: RCV: red cell volume.

nosed as anemic whilst having a tHb-mass which is normal or even elevated. These findings are of direct clinical relevance. A basic medical education teaches that PV may be expanded in some disease states. But it is rarely, if ever, mentioned that this might lead to hemodilution of such a degree as to cause anemia. Thus, a diagnosis of anemia generally leads to investigation of a cause for reduced hemoglobin synthesis or increased erythrocyte destruction or loss. When such features are not identified, a diagnosis of ‘anemia of chronic disease’ is likely made. Rarely, if ever, is hemodilution considered as a cause, and measurement of tHb-mass or plasma volume performed. This is exemplified by the fact that no such measurements had been performed electively in the patients we studied. Such deficits in consideration or action may in part relate to the difficulty and expense of measuring such variables through traditional methods (radiolabeling of RBCs, for instance). Thus, tHb-mass was generally strongly related to [Hb] (r≥0.687 and P<0.0001 in all cases); tHb-mass explained a good deal of the variance in [Hb] (adjusted R2 in HV, surgical and IBD patients being 0.746, 0.565 and 0.446, respectively, P<0.0001 in all cases); and PV independently accounted for only a small proportion of the variance in [Hb] over that due to tHb-mass (R2 change 0.192, 0.374 or 0.479 in HV, surgical and IBD patients, respectively). By contrast, in the 2 patient groups most likely to suffer expanded PV (CLD and CHF), tHb-mass did not correlate with [Hb] (r=0.410, P=0.114; r=0.312, P=0.157, respectively). Likewise, tHb-mass did not significantly explain variance in [Hb] (adjusted R2=0.109, P=0.114; adjusted R2=0.052, P=0.157, respectively), whilst PV independently accounted for a greater proportion of the variance in [Hb] over and above tHb-mass in these groups (R2 change 0.724 in CHF and 0.805 in CLD). Thus, [Hb] is strongly influenced by disease-related changes in PV. The relationship between [Hb] and tHb-mass weakened as PV rose (r-values falling from 0.94 and 0.91 in those with low or normal PV, to 0.57 amongst those in whom PV was severely expanded). As a consequence, even amongst those in our small sample, we identified patients with identical and normal tHb-mass, in some of whom severe anemia would be diagnosed, likely triggering investigations focused upon failed erythrogenesis or increased red cell destruction. As specific exemplars of the phenomenon, 2 IBD patients had similar tHb-masses (9.2 and 9.3 g.kg-1), 1 having a high 1482

normal [Hb] (161 g.l-1) and the other being anemic ([Hb] 107 g.l-1), due to substantial differences in PV (37.2 ml.kg1 vs. 65.7 ml.kg-1). Similarly, 2 CLD patients had the same tHb-mass (5.2 g.kg-1), 1 having mild anemia [Hb] (110 g.l1 ), but the second being markedly anemic ([Hb] 69 g.l-1) due to a relatively raised PV in the latter (36.5 vs. 67.8 ml.kg-1, respectively). Our findings are thus of real clinical importance, as a significantly low [Hb] can trigger a raft of (unwarranted) investigations (such as the assay of circulating hematinic factors) or treatments (such as the administration of packed RBCs), whilst denying the administration of agents to reduce PV, which might sometimes be required. Blood transfusion itself carries risks25 as well as a price in terms of healthcare costs. Meanwhile, in other circumstances, contraction in PV might offer false reassurance by maintaining [Hb] when tHb-mass is low. Whilst tHb-mass is generally used as an index of oxygen carrying capacity and of circulating red cell mass, others have previously reported total circulating red blood cell volume (RCV) in this regard. In hematologically normal control subjects, Hct in the 20-50% range reportedly correlated well with RCV (determined by 51Chromuim (51Cr) labeling of RBCs: r=0.880, P<0.001).26 However, this relationship was disturbed when Hct fell outside this range, owing to wider variability in PV. Such data support those from other radiolabeling studies in suggesting that direct measurement of RCV (rather than the use of [Hb] or Hct) is required for the accurate diagnosis of polycythemia.27 Likewise, data derived from the same technique which we applied (oCOR) show PV to be expanded (variably, but along with increased RCV) in polycythemia rubra vera.28 The focus of such studies differed from ours: namely, they sought to address the degree to which variation in PV altered the accuracy of the diagnosis of polycythemia, whilst we assessed the influence of variation in tHb-mass and PV on [Hb] per se and on the diagnosis of anemia. Nor have any studies in this field been comprehensive across disease states, or assessed tHb-mass (rather than RCV). Nonetheless RCV (by 51Cr labeling) has been shown to be similar in anemic and non-anemic CHF patients, suggesting that PV expansion accounted for this diagnosis.15 Indeed, this has been shown to occur (using I131-labeled albumin) in CHF due to systolic dysfunction, with poor correlation between [Hb] and RCV.29 Likewise, and using the same technique, Miller showed 19 of 32 patients hospitalized with decompensated CHF to be anemic, with haematologica | 2017; 102(9)


[Hb], tHb-mass and PV in anemia

A

C

B

D

E

Figure 4. Unadjusted relationship between hemoglobin concentration and total hemoglobin mass. Healthy controls (A, n=21), patients with IBD (B, n=22), surgical patients (C, n=28), liver disease (D, n=16) and HF (E, n=22). tHb-mass (g): total hemoglobin mass; [Hb] (g.l-1): hemoglobin concentration.

only 4 of these having a true reduction in RCV.30 Using the oCOR method, we extend such observations (Table 2). Overall, 39 of the 109 participants were anemic, in only 2 (13.3%) of whom was this due to a reduced tHb-mass [352 g and 449 g] in the context of a normal PV. In the remaining 86.7%, reduced [Hb] was accounted for by PV expansion (n=4 mild to moderate, tHb-mass 494 Âą 76 g, and n=8 severe, tHb-mass 539 Âą 105 g). Interestingly, 14 of the 39 anemic patients (93%) had a relatively raised tHb-mass, with PV elevated to a greater degree (n=1 mild to moderate PV expansion (tHb-mass 610 g), n=13 severe PV expansion (tHb-mass 758 Âą 220 g). The overall prevalence of anemia for our study participants (36%) is similar to that reported in previous studies in non-cardiac surgical patients (30.4%).9 Nine percent of IBD patients suffered anemia, which is less than has been reported across European Countries (24%).12 Some 54% of CHF patients suffered anemia in the study herein, somewhat more than has been previously reported in the Study of Anemia in a Heart Failure Population (STAMINA-HFP) Registry (34%),31 or in patients with advanced HF (30%),32 but in keeping with the data of others (55.6%33 to 61%).34 Of the CLD patients, 94% suffered anemia, a figure somewhat higher that that previously reported by some (50-75%),35,36 but in keeping with data in decompensated CLD (86%)37 or hepatitis C infection (75%).38 haematologica | 2017; 102(9)

Amongst HV, 5 (24% of healthy volunteers) were anemic, a figure in keeping with global data relating to nonpregnant females (30%),2 and only slightly higher than the 16% reported in non-pregnant women aged 15-49 years from high income regions and 22% in menstruating women from central and eastern Europe.39 This may be related to volunteering bias in the current study whereby those who thought they might be anemic preferentially applied to participate. The study herein utilized the optimized oCOR method, validated against 51Cr radiolabeling methodologies.40 An advantage of our study was its assessment of diverse patient groups. Sample sizes, whilst small, were appropriately powered to explore the relationship between tHbmass and [Hb], albeit that, for administrative reasons, the sample size for patients with CLD (n=16) was below the 21 we originally sought, based on the study by Hinrichs and colleagues.23 Nonetheless, we were still able to demonstrate that PV changes in this group do indeed influence assessed [Hb] and anemia diagnoses. Differing blood sample methods (capillary and venous) yield identical D%COHb (and thus tHb-mass) values.41 However, capillary [Hb] can be marginally higher than that in venous blood.42-45 The use of differing sampling techniques across testing sites may thus have contributed a little to variation in measured [Hb], although capillary blood [Hb] values were all corrected to venous conditions, 1483


J.M. Otto et al. A

D

B

E

C

Figure 5. Individual participant data for total hemoglobin mass, plasma volume and hemoglobin concentration. Participants ranked by tHb-mass (g.kg-1) (dark gray bars) from smallest to largest with corresponding PV (ml.kg-1) (light gray bars) and [Hb] (g.l-1) (black circles) in healthy volunteers (A), inflammatory bowel disease (B), surgical (C), chronic heart failure (D), and chronic liver disease (E). PV: plasma volume; [Hb]: hemoglobin concentration.

and all blood samples were collected from the same anatomical site, and with patients in the same posture (seated). This factor does not therefore weaken the significance of our findings. Whilst the use of different blood gas machines and testing staff may have introduced error in the measurement of tHb-mass, we found a typical error (TE) of repeat tHb-mass measurements of 1.93% (95% confidence interval (CI) 1.3-3.4%: data not shown), values in keeping with other institutions using the oCOR method.19,46 Finally, the oCOR method is quick and simple, avoids the technical difficulties of working with radiolabeled compounds,40 offers an inconsequential burden for patients, is minimally invasive, and is safe even in patients with serious medical conditions and comorbidities such as stable coronary artery disease.47 This greatly widens the applicability of the oCOR test to measure tHb-mass and plasma volume in the clinical setting. To date, its clinical experimental application has been sparse, e.g., to demonstrate that low tHb-mass may account for impaired exertional performance in otherwise healthy diabetics.48 Whilst (rarely as of yet) applied to the measurement of red cell mass in patients with polycythemia rubra vera,28 it has yet to be utilized in routine clinical practice but might find great value, for example in the estimation of red cell mass and PV in cases of presumed excessive erythrocytosis. Our data might support wider use. In conclusion, measured [Hb], and the diagnosis of 1484

anemia, can be strongly influenced by (or can largely depend upon) changes in PV. The scale of this impact may be greater in some diseases than others. Constraining investigation of anemia to the identification of causes of reduced Hb synthesis or of erythrocyte loss or destruction may be inappropriate for many. The concept of â&#x20AC;&#x2DC;anemiaâ&#x20AC;&#x2122; may thus need refining in clinical practice, and the oCOR method may support better and more appropriate assessments of the factors influencing circulating [Hb]. Acknowledgments The authors would like to thank Siemens for supplying the RAPIDPoint 500 blood gas machine and associated consumables used during this study at the Southampton General Hospital. In addition, we would like to thank the NIHR/Wellcome Clinical Research Facility for allowing patients to be tested in this facility at University College London Hospital. Finally, we would like to thank Dr Nadine Wachsmuth for her assistance in training JO and JOMP in the optimized carbon monoxide rebreathing method at the Department of Sports Medicine/Sports Physiology, University of Bayreuth, Germany. Funding HM is funded in part by the NIHR University College London Hospitals Biomedical Research Centre, to whom we express our thanks. MPWG is funded in part by the NIHR Southampton Respiratory Biomedical Research Unit. haematologica | 2017; 102(9)


[Hb], tHb-mass and PV in anemia

References 1. World Health Organisation. Haemoglobin concentration for the diagnosis of anaemia and assessment of severity 2011. http://www.who.int/vmnis/indicators/hae moglobin/en/. Accessed 28th September 2016. 2. World Health Organization. Worldwide prevalence of anaemia 1993-2005. 2008. http://apps.who.int/iris/bitstream/10665/43 894/1/9789241596657_eng.pdf. Accessed 5th October 2016. 3. Farag YM, Keithi-Reddy SR, Mittal BV, et al. Anemia, inflammation and health-related quality of life in chronic kidney disease patients. Clin Nephrol. 2011;75(6):524533. 4. Chambellan A, Chailleux E, Similowski T, Group AO. Prognostic value of the hematocrit in patients with severe COPD receiving long-term oxygen therapy. Chest. 2005; 128(3):1201-1208. 5. Lipsic E, Asselbergs FW, van der Meer P, et al. Anaemia predicts cardiovascular events in patients with stable coronary artery disease. Neth Heart J. 2005;13(7-8):254-258. 6. Ezekowitz JA, McAlister FA, Armstrong PW. Anemia is common in heart failure and is associated with poor outcomes: insights from a cohort of 12 065 patients with newonset heart failure. Circulation. 2003; 107(2):223-225. 7. Caro JJ, Salas M, Ward A, Goss G. Anemia as an independent prognostic factor for survival in patients with cancer: a systemic, quantitative review. Cancer. 2001; 91(12):2214-2221. 8. Miceli A, Romeo F, Glauber M, de Siena PM, Caputo M, Angelini GD. Preoperative anemia increases mortality and postoperative morbidity after cardiac surgery. J Cardiothorac Surg. 2014;9:137. 9. Musallam KM, Tamim HM, Richards T, et al. Preoperative anaemia and postoperative outcomes in non-cardiac surgery: a retrospective cohort study. Lancet. 2011; 378(9800):1396-1407. 10. Heinicke K, Wolfarth B, Winchenbach P, et al. Blood volume and hemoglobin mass in elite athletes of different disciplines. Int J Sports Med. 2001;22(7):504-512. 11. Knight K, Wade S, Balducci L. Prevalence and outcomes of anemia in cancer: a systematic review of the literature. Am J Med. 2004;116 Suppl 7A:11S-26S. 12. Filmann N, Rey J, Schneeweiss S, et al. Prevalence of anemia in inflammatory bowel diseases in european countries: a systematic review and individual patient data meta-analysis. Inflamm Bowel Dis. 2014;20(5):936-945. 13. McClellan W, Aronoff SL, Bolton WK, et al. The prevalence of anemia in patients with chronic kidney disease. Curr Med Res Opin. 2004;20(9):1501-1510. 14. Gonzalez-Casas R, Jones EA, Moreno-Otero R. Spectrum of anemia associated with chronic liver disease. World J Gastroenterol. 2009;15(37):4653-4658. 15. Adlbrecht C, Kommata S, Hulsmann M, et al. Chronic heart failure leads to an expanded plasma volume and pseudoanaemia, but does not lead to a reduction in the body's red cell volume. Eur Heart J. 2008;29(19): 2343-2350. 16. Feigenbaum MS, Welsch MA, Mitchell M, Vincent K, Braith RW, Pepine CJ. Contracted plasma and blood volume in chronic heart

haematologica | 2017; 102(9)

failure. J Am Coll Cardiol. 2000;35(1):51-55. 17. The International Committee for Standardisation in Haematology (ICSH). Recommended methods for measurement of red-cell and plasma volume. J Nucl Med. 1980;21(8):793-800. 18. Pearson TC. Evaluation of diagnostic criteria in polycythemia vera. Semin Hematol. 2001;38(1 Suppl 2):21-24. 19. Schmidt W, Prommer N. The optimized CO-rebreathing method: a new tool to determine total haemoglobin mass routinely. Eur J Appl Physiol. 2005;95(5-6):486-495. 20. Thomsen JK. Blood and plasma volumes determined by carbon monoxide gas, 99mTc-labelled erythrocytes, 125I-albumin and the T 1824 technique. 1991;51(2):185190. 21. Welch BL. On the comparison of several mean values: an alternative approach. 1951;38(3/4):330-336. 22. Field A. Discovering statistics using SPSS. London: Sage Publications, 2005. 23. Hinrichs T, Franke J, Voss S, Bloch W, Schanzer W, Platen P. Total haemoglobin mass, iron status, and endurance capacity in elite field hockey players. J Strength Cond Res. 2010;24(3):629-638. 24. Faul F, Erdfelder E, Buchner A, Lang AG. Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behav Res Methods. 2009;41:1149-1160. 25. Gilliss BM, Looney MR, Gropper MA. Reducing noninfectious risks of blood transfusion. Anesthesiology. 2011; 115(3):635649. 26. Huber H, Lewis SM, Szur L. The influence of anaemia, polycythaemia and splenomegaly on the relationship between venous haematocrit and red-cell volume. Br J Haematol. 1964; 10:567-575. 27. Lorberboym M, Rahimi-Levene N, Lipszyc H, Kim CK. Analysis of red cell mass and plasma volume in patients with polycythemia. Arch Pathol Lab Med. 2005; 129(1):89-91. 28. Ahlgrim C, Schumacher YO, Wrobel N, Waller CF, Pottgiesser T. Application of the optimized CO-rebreathing method for determination of hemoglobin mass in patients with polycythemia vera. Ann Hematol. 2014;93(7):1159-1165. 29. Abramov D, Cohen RS, Katz SD, Mancini D, Maurer MS. Comparison of blood volume characteristics in anemic patients with low versus preserved left ventricular ejection fractions. Am J Cardiol. 2008;102(8):10691072. 30. Miller WL, Mullan BP. Peripheral venous hemoglobin and red blood cell mass mismatch in volume overload systolic heart failure: implications for patient management. J Cardiovasc Transl Res. 2015; 8(7):404-410. 31. Adams KF, Jr., Patterson JH, Oren RM, et al. Prospective assessment of the occurrence of anemia in patients with heart failure: results from the Study of Anemia in a Heart Failure Population (STAMINA-HFP) Registry. Am Heart J. 2009;157(5):926-932. 32. Horwich TB, Fonarow GC, Hamilton MA, MacLellan WR, Borenstein J. Anemia is associated with worse symptoms, greater impairment in functional capacity and a significant increase in mortality in patients with advanced heart failure. J Am Coll Cardiol. 2002;39(11):1780-1786. 33. Silverberg DS, Wexler D, Blum M, et al. The use of subcutaneous erythropoietin and intravenous iron for the treatment of the

34.

35.

36.

37. 38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

anemia of severe, resistant congestive heart failure improves cardiac and renal function and functional cardiac class, and markedly reduces hospitalizations. J Am Coll Cardiol. 2000;35(7):1737-1744. Androne AS, Katz SD, Lund L, et al. Hemodilution is common in patients with advanced heart failure. Circulation. 2003;107(2):226-229. Senzolo M, Burroughs AK. Haematological abnormalities in liver disease. In: Rodes J, Benhamou JP, Blei AT, Reichen J, Rizzetto M, eds. Textbook of Hepatology: From Basic Science to Clinical Practice. Oxford, UK: Blackwell Publishing Ltd; 2007. Gonzalez-Casas R, Jones EA, Moreno-Otero R. Spectrum of anemia associated with chronic liver disease. World J Gastroenterol. 2009;15(37):4653-4658. Kumar EH, Radhakrishnan A. Prevalence of anaemia in decompensated chronic liver disease. World J Med Sci. 2014;10(1):56-60. McHutchison JG, Manns MP, Longo DL. Definition and management of anemia in patients infected with hepatitis C virus. Liver Int. 2006;26(4):389-398. Stevens GA, Finucane MM, De-Regil LM, et al. Global, regional, and national trends in haemoglobin concentration and prevalence of total and severe anaemia in children and pregnant and non-pregnant women for 1995-2011: a systematic analysis of population-representative data. Lancet Glob Health. 2013;1(1):e16-25. Gore CJ, Hopkins WG, Burge CM. Errors of measurement for blood volume parameters: a meta-analysis. J Appl Physiol (1985). 2005;99(5):1745-1758. Garvican LA, Burge CM, Cox AJ, Clark SA, Martin DT, Gore CJ. Carbon monoxide uptake kinetics of arterial, venous and capillary blood during CO rebreathing. Exp Physiol. 2010;95(12):1156-1166. Patel, A.J., Wesley, R., Leitman, S.F. & Bryant, B.J. Capillary versus venous haemoglobin determination in the assessment of healthy blood donors. Vox Sang. 2013;104(4):317-323. Shahshahani, H.J., Meraat, N. & Mansouri, F. Evaluation of the validity of a rapid method for measuring high and low haemoglobin levels in whole blood donors. Blood Transfus. 2013;11(3):385-390. Ziemann, M., Lizardo, B., Geusendam, G. & Schlenke, P. Reliability of capillary hemoglobin screening under routine conditions. Transfusion. 2011;51(12):2714-2719. Baart, A.M., de Kort, W.L., van den Hurk, K. & Pasker-de Jong, P.C. Hemoglobin assessment: precision and practicability evaluated in the Netherlands-the HAPPEN study. Transfusion. 2016;56(8):1984-1993. Turner G, Richardson AJ, Maxwell NS, Pringle JS. Comparison of total haemoglobin mass measured with the optimized carbon monoxide rebreathing method across different Radiometer ABL-80 and OSM-3 hemoximeters. Physiol Meas. 2014;35(12): N41-9. Karlsen, T., Leinan, I. M., Aamot, I.-L., Dalen, H. & Stoylen, A. Safety of the COrebreathing method in patients with coronary artery disease. Med Sci Sports Exerc. 2016;48(1):33-38. Koponen AS, Peltonen JE, Paivinen MK, Aho JM, Hagglund HJ, Uusitalo AL, et al. Low total haemoglobin mass, blood volume and aerobic capacity in men with type 1 diabetes. Eur J Appl Physiol. 2013; 113(5):11811188.

1485


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Coagulation & its Disorders

Ferrata Storti Foundation

Long-term impact of joint bleeds in von Willebrand disease: a nested case-control study Karin P.M. van Galen,1 Piet de Kleijn,2 Wouter Foppen,3 Jeroen Eikenboom,4 Karina Meijer,5 Roger E.G. Schutgens,6 Kathelijn Fischer,7 Marjon H. Cnossen,8 Joke de Meris,9 Karin Fijnvandraat,10 Johanna G. van der Bom,11 Britta A.P. Laros-van Gorkom,12 Frank W.G. Leebeek13 and Eveline P. Mauser-Bunschoten14 for the Win study group

Van Creveldkliniek, University Medical Center Utrecht; 2Van Creveldkliniek and Department of Rehabilitation, Physical Therapy Science and Sports, University Medical Centre Utrecht; 3Department of Radiology, University Medical Center Utrecht; 4 Department of Thrombosis and Hemostasis and Einthoven Laboratory for Vascular and Regenerative Medicine, Leiden University Medical Center; 5Department of Hematology, University of Groningen, University Medical Center Groningen; 6Van Creveldkliniek, University Medical Center Utrecht; 7Van Creveldkliniek and Julius Center Department of Epidemiology, University Medical Center Utrecht; 8Department of Pediatric Hematology, Erasmus University Medical Center-Sophia Childrenâ&#x20AC;&#x2122;s Hospital, Rotterdam; 9Dutch Society of Haemophilia Patients, Leiden; 10Department of Pediatric Hematology, Academisch Medisch Centrum, Emma Children's Hospital, Amsterdam; 11Jon J van Rood Center for Clinical Transfusion Medicine, Sanquin Research, Leiden, and Department of Clinical Epidemiology, Leiden University Medical Center; 12Department of Hematology, Radboud University Medical Center, Nijmegen; 13Department of Hematology, Erasmus University Medical Center, Rotterdam and 14Van Creveldkliniek, University Medical Center Utrecht, the Netherlands 1

Haematologica 2017 Volume 102(9):1486-1493

ABSTRACT

P

Correspondence: k.p.m.vangalen@umcutrecht.nl

Received: March 10, 2017. Accepted: May 30, 2017. Pre-published: June 1, 2017. doi:10.3324/haematol.2017.168617 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1486 Š2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1486

atients with severe von Willebrand disease (VWD) may develop arthropathy after joint bleeds. Information on its prevalence and severity is limited. We aimed to assess the occurrence and severity of arthropathy in VWD and its impact on daily life. VWD patients with and without verified joint bleeds were matched for age, sex and Factor VIII level or von Willebrand Factor activity in a nested case-control study within the Willebrand in the Netherlands study. Assessments included the Hemophilia Joint Health Score (0-124), Pettersson score (0-13 per joint X-ray), Hemophilia Activity List score (0-100), joint pain (Visual Analog Scale 0-10), and the Impact on Participation and Autonomy questionnaire (0-20). Arthropathy was defined as a Hemophilia Joint Health Score of 10 or higher, or a Pettersson score over 3 of at least one joint. We included 48 patients with verified joint bleeds (cases) and 48 controls: 60% males, mean age 46 years (range 18-80), median von Willebrand Factor activity 5 versus 8 IU/dL and Factor VIII 24 versus 36 IU/dL. Arthropathy occurred in 40% of the cases versus 10% of the controls (P<0.01). The cases reported more functional limitations compared to the controls (median Hemophilia Activity List score: 88 vs. 100, P<0.01). Arthropathy was related to joint pain and less social participation (Visual Analog Scale>3: 13 of 19 vs. 3 of 28, P<0.01, and median score on the participation questionnaire 6.1 vs. 0.9, P<0.01). In conclusion, arthropathy occurs in 40% of VWD patients after joint bleeds and is associated with pain, radiological abnormalities, functional limitations, and less social participation (Dutch trial register: NTR4548).

Introduction Von Willebrand disease (VWD) is a congenital bleeding disorder with a population prevalence of 0.6-1.3%.1 VWD is caused by a deficiency (type 1), dysfunction (type 2) or absence (type 3) of von Willebrand factor (VWF) and is mainly associated with mucocutaneous bleeding and menorrhagia. Joint bleeds also occur in haematologica | 2017; 102(9)


Long-term impact of joint bleeds in VWD

VWD, predominantly in severely affected patients with concomitant low Factor VIII levels (FVIII).2 Low FVIII occurs because its chaperone protein VWF, which protects FVIII from degradation, is (partly) missing. In hemophilia patients with FVIII or FIX deficiency, recurrent joint bleeds are the main cause of pain and functional limitations, which is known as hemophilic arthropathy.3 Joint bleeding occurs in half of the patients with type 3 VWD and in 510% of moderate and severe type 1 and type 2 VWD patients.4 However, hardly any information is available on the existence and severity of blood-induced arthropathy in VWD.5 There are no data on joint health and the influence of arthropathy on daily life activities in VWD.5 In the Willebrand in the Netherlands (WiN) study, almost a quarter of the patients reported joint bleeds. These joint bleeds appeared to have a negative impact on health-related quality of life and joint integrity, according to self-reported and retrospective medical file data.4 To describe and measure health and disability, the World Health Organization's (WHO) International Classification of Functioning, Disability and Health (ICF) standard has been endorsed. This standard focuses on impact rather than cause of disability, including functioning and participation, as well as environmental and personal factors.6 It is, therefore, useful to study the consequences of joint bleeds in VWD within this broad ICF perspective.

It is important to assess joint outcome after joint bleeds in order to find and implement optimal treatment strategies that prevent or diminish arthropathy.7 The primary objective of this nested case-control study is to assess both occurrence and severity of arthropathy in patients with VWD after joint bleeds (ICF domain on body structure and function), compared to VWD patients without clinically overt joint bleeding, and to compare self-perceived functional abilities and social participation (ICF domains on activity and participation) between these two groups. The secondary objective is to explore a possible association between arthropathy and environmental and personal factors (ICF contextual factors).

Methods Ethical approval was obtained from the medical ethical committee of the University Medical Center Utrecht, the Netherlands (Dutch trial register: NTR4548).

Study population A nested case-control study was performed within the national cohort WiN study between August 2013 and July 2015. Cases were selected from this WiN study cohort, based on verified joint bleeds (VWD-JB patients).2,4 Adult patients with VWD according to the definitions of the WiN study (historically lowest VWF activ-

Table 1. Baseline characteristics of 48 von Willebrand disease patients treated for joint bleeds and 48 controls.

Sex Females (n, %) 19 (40%) 19 (40%) Age (y) Type VWD*

Levels (IU/dL; med, IQR)†

Cumulative n. of joint bleeds#

Joint bleed treatment Relevant comorbidity¶

VWD-JB patients n=48

VWD controls n=48

45 (18-78) 48 (18-74) 8 (17%) 21 (44%) 15 5 0 1 19 (39%) 13 (1-36) 5 (0-14) 24 (3-46)

48 (20-80) 47 (18-73) 20 (42%) 25 (52%) 17 5 2 1 3 (6%) 22 (12-28) 8 (3-30) 36 (24-54) 34 (71%) 9 (19%) 5 (10%) 5 (10%)

Males (n, %) 29 (60%) 29 (60%) Males (median, range) Females (median, range) 1 (n, %) 2 (n, %) 2A (n) 2B (n) 2M (n) 2N (n) 3 (n, %) VWF:Ag VWF:Act FVIII:C No. JB (n, %) Once 2-5 6-10 >10 VWF concentrate Desmopressin n (%)

8 (17%) 13 (27%) 6 (12%) 21 (44%) 44 (92%) 10 (21%) 3 (6%)

VWD: von Willebrand disease; JB: joint bleed; n.: number; y: years; med: median; IQR: 25-75% interquartile range; IU/dL: units per deciliter; VWF:Ag: von Willebrand factor antigen level;VWF:Act:VWF activity level; FVIII:C: Factor VIII level. *Based on centrally determined [n=94, data from the Willebrand in the Netherlands (WiN) study] or historic VWD type (n=4, as known in the hemophilia treatment center). †Centrally measured: based on 72 patients for whom plasma was available at times of WiN inclusion and after exclusion of pregnant patients and those who had received clotting factor concentrate (CFC) or desmopressin less than 72 hours before the laboratory assessment. #Based on medical file data and information provided by the participants during the study visit. ¶5 controls (2 gout, 1 psoriatric arthritis, 1 congenital flexion contraction elbows, 1 elbow fracture) and 3 patients (1 septic arthritis knee, 1 psoriatric arthritis, 1 ankle fracture).

haematologica | 2017; 102(9)

1487


K.P.M. van Galen et al. ity <30 IU/dL8) from all hemophilia treatment centers in the Netherlands were eligible. Joint bleeds were verified if we found medical file documentation of treatment with clotting factor or desmopressin for at least one joint bleed. VWD patients without prior treated joint bleeds were included as controls (VWD controls). These VWD controls were primarily selected from the WiN cohort and matched 1:1 to the VWD-JB patients for age, sex and, if possible, on FVIII or otherwise on VWF activity (Online Supplementary Table S1).4 The goal of the case-control design was to distinguish joint bleed-related arthropathy from other causes of joint deterioration, such as age-related osteoarthritis or subclinical joint bleeding. The absence of documented treatment for joint bleeds in the medical files at inclusion was regarded as “no clinically relevant joint bleeds”. Exclusion criteria were inability to give informed consent, recent joint bleeding without complete recovery, and lack of available medical files. Restricted motion due to another musculoskeletal disorder [e.g. gout or rheumatoid arthritis (RA), excluding age-related osteoarthritis] was regarded as relevant musculoskeletal comorbidity.

Primary outcome parameters for joint health assessment Participants were invited for a half-day visit to the hemophilia treatment center, and underwent physical examination and measurements. Joint X-rays were performed. One experienced physio-

therapist conducted the Hemophilia Joint Health score (HJHS, 030/joint, total range 0-124) in all participants. The HJHS has been developed to analyze joint outcome in hemophilia.9 It is an 11item scoring tool to assess joint health of elbows, knees and ankles that includes range of motion, crepitus on motion, swelling, muscle atrophy, pain, joint strength, and global gait. Range of motion reference values to calculate the HJHS were derived from Soucie et al.10 For the analyses, we dichotomized the HJHS≥10 as ‘arthropathy’ based on our interobserver reliability results in VWD and because this cut-off value has also been used in hemophilia to define arthropathy.11,12 X-rays were taken from all ankles, knees and elbows with prior bleeds, the contralateral joints of the VWD-JB patients, and from the ipsilateral joints of the matched VWD controls. X-rays were scored by one radiologist according to Pettersson (PS, 0-13 points per joint), using a consensus atlas.13 We defined a PS>3 of one or more joints as ‘radiological arthropathy’, based on the Limits of Agreement of the PS joint in hemophilia patients.14 All participants were asked to complete the Hemophilia Activity List (HAL, normalized score: 0-100). This questionnaire has been developed to analyze self-perceived functional abilities in hemophilia.15,16 The HAL total and three component sub scores [‘upper extremity activities’ (HAL upper), ‘basic lower extremity activities’ (HAL lowbas) and ‘complex lower extremity activities’ (HAL lowcom)] were calculated. A score of 100 means no func-

Table 2. Arthropathy, functional abilities and contextual factors compared between the von Willebrand disease-joint bleeds (VWD-JB) patients and VWD controls.

Mean CFC use* CFC prophylaxis¶ because of JB Surgery in large joints‡ BMI HJHS total HJHS ≥ 10 PS >3 PS >3 ankle PS >3 knee PS >3 elbow HAL total HAL upper HAL lowbas HAL lowcom HAL total <95 Figure 8 (sec) Preferred speed Maximum speed Joint pain§ VAS mean ** VAS >3 joint D-AIMS2affect IPA

VWD-JB patients n=48

VWD controls n= 48

P

Median IU FVIII/kg/y (IQR) n (%) n (%) Overall n (%) Because of JB n (%) Mean (range) Median (IQR) n (%) n (%) n (%) n (%) n (%) Median (IQR) Median (IQR) Median (IQR) Median (IQR) n/total (%)

51 (1.3-188) 13 (27%) 11 (23%) 22 (46%) 9 (19%) 26 (18-49) 5 (1-15) 19 (40%) 12 (25%) 9 (19%) 5 (10%) 2 (4%) 88 (69-98) 93 (82-100) 87 (53-100) 80 (44-100) 32/48 (67%)

0 (0-0) 1 (2%) 10 (21%) 25 (17-37) 1.5 (0-5) 5 (10%) 2 (4%) 1 (2%) 1 (2%) 0 100 (87-100) 100 (91-100) 100 (86-100) 100 (77-100) 16/46 (35%)

<0.01 <0.01

Median (range) Median (range) Overall n (%) Median (IQR) n (%) Anxiety median (range) Mood median (range) Sum score median (range)

15 (10-25) 11 (8-20) 29 (60%) 3.5 (1-5.7) 17 (35%) 13 (7-24) 12 (5-22) 1.93 (0-11)

14 (11-24) 10 (8-20) 23 (48%) 2.1 (1.4-4.7) 9 (19%) 12 (5-22) 9 (5-17) 0.97 (0-14)

0.24 0.71 0.22 0.39 0.07 0.33 0.87 0.14

<0.01 0.28 <0.01 <0.01 <0.01 <0.01 0.09 0.15 <0.01 0.01 <0.01 <0.01 <0.01

VWD: von Willebrand disease; JB: joint bleeds; CFC: clotting factor concentrate; IU: units; IQR: 25-75% interquartile range; BMI: Body Mass Index; HJHS: Hemophilia Joint Health Score; VAS: Visual Analog Scale pain scale (total range 0-10 cm); PS: Pettersson score; sec: seconds; IPA: Impact on Participation and Autonomy questionnaire. HAL: Hemophilia Activity List questionnaire; DAIMS2affect: Dutch Arthritis Impact Measurement Scales-2.*In units Factor VIII/kg/year calculated over five years (2005-2009), based on medical file data (n=90). ¶Currently or in the past; defined as at least 1 regular CFC infusion per week for at least 45 consecutive weeks. §Chronic joint pain, not related to a recent joint bleed. ‡ Knee, ankle, elbow, shoulder or hip surgery. **N=27 patients and n=21 controls.

1488

haematologica | 2017; 102(9)


Long-term impact of joint bleeds in VWD

tional limitations. For statistical analyses, the total HAL score was dichotomized into ‘no functional limitations’ (>95) and ‘some functional limitations’ (<95). This threshold was chosen as it was the median HAL score in patients with severe hemophilia.7 We recently validated both the HJHS and HAL in VWD.12

Secondary outcome parameters All participants were asked to perform the modified Figure 8 walk test17 and to complete the Impact on Participation and Autonomy questionnaire (IPA, 31 items with 0-4 points per item on five domains: 0=very good, 4=bad; IPA sum score of mean scores per item range 0-20).16,18 Participants were also requested to

complete the affect component of the Dutch Arthritis Impact Measurement Scales-2 (D-AIMS2affect), asking questions on anxiety and depression ranging from 5 (never) to 25 (always) for both components, as well as the McGill Pain Questionnaire-Dutch Language Version.19,20 The mean Visual Analog Score (VAS 0-10 cm) for pain was calculated from the minimum and maximum VAS reported in this questionnaire. A mean VAS of more than 3 in joints (chronic joint pain, not related to a recent joint bleed) was considered as clinically-relevant joint pain. Body Mass Index (BMI) was calculated after measuring body weight and height during the study visit. Data on the cumulative number of joint bleeds, history of joint surgery, use of prophylaxis (defined as at

Table 3. Characteristics, functional abilities and contextual factors compared within the von Willebrand disease-joint bleed (VWD-JB) patients with versus without arthropathy (HJHS ≥10 or PS >3 of one or more joints).

Sex Age (y) VWD subtype

Levels (IU/dL; med, IQR)†

JB sites & cum n. Elbow Elbow >5 JB Knee Knee >5 JB Ankle Ankle >5 JB Mean CFC use* CFC prophylaxis¶ because of JB Surgery in large joints‡ HJHS total HAL total HAL upper HAL lowbas HAL lowcom HAL total <95 Figure 8 (sec) Preferred speed Maximum speed Joint pain§ VAS mean ** VAS >3 joint D-AIMS2affect IPA¥

Males (n,%) Females (n,%) (median, range) Type 1 Type 2 Type 3 VWF:Ag VWF:Act FVIII:C n (%)

Median IU FVIII/kg/y (IQR) n (%) n (%) Overall n (%) Because of JB n (%) Median (IQR) Median (IQR) Median (IQR) Median (IQR) Median (IQR) n/total (%) Median (range) Median (range) Overall n (%) Median (IQR) n (%) Anxiety median (range) Mood median (range) Sum score median (range)

VWD-JB and arthropathy n=19

VWD-JB no arthropathy n=29

P

7 (37%) 12 (63%) 50 (28-64) 3 4 12 1 (0-50) 0 (0-19) 5 (1-51)

22 (76%) 7 (24%) 44 (18-78) 5 17 7 13 (7-30) 6 (0-12) 26 (10-42)

<0.01

6 (32%) 2 (11%) 13 (68%) 4 (21%) 16 (84%) 13 (68%) 343 (79-821)

12 (41%) 0 25 (86%) 3 (10%) 21 (72%) 6 (21%) 18 (0-73)

12 (63%) 11 (58%) 12 (63%) 9 (47%) 16 (14-26) 70 (49-78) 88 (71-93) 50 (37-83) 53 (24-71) 32/48 (67%)

1 (3%) 0 19 (66%) 0 2 (0.3-4) 95 (88-100) 100 (91-100) 97 (87-100) 89 (73-100) 16/46 (35%)

0.05 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01

16 (15-20) 12 (11-13) 17 (89%) 3.5 (3.1-6.3) 13 (68%) 13 (10-17) 10 (8-12) 6.1 (2.5-8.6)

14 (13-15) 9.7 (9.2-11) 12 (41%) 0.3 (0.0-1.1) 4 (14%) 13 (10-16) 9 (7-11) 0.9 (0-2.4)

<0.01 <0.01 <0.01 <0.01 <0.01 0.90 0.14 <0.01

0.38 0.90 0.01 0.01 0.75 0.90 0.36

0.07 0.30 <0.01 <0.01 <0.01

VWD: von Willebrand disease; JB: joint bleeds; y: year; IQR: interquartile range; cum: cumulative; CFC: clotting factor concentrate; IU: units; n: number: HJHS: Hemophilia Joint Health Score; HAL: Hemophilia Activity List questionnaire; VAS:Visual Analog Scale pain scale (total range 0-10 cm). *In units factor VIII/kg/year calculated over five years (20052009), based on medical file data (n=44). ¶or in the past; defined as at least 1 regular CFC infusion per week for at least 45 consecutive weeks. †Centrally measured: based on 34 patients of whom plasma was available at time of Willebrand in the Netherlands (WiN) study inclusion and after exclusion of pregnant patients and those who had received CFC or desmopressin less than 72 hours before the laboratory assessment. §Chronic joint pain, not related to a recent joint bleed. ‡Knee, ankle, elbow, shoulder or hip surgery. **Based on n=17 arthropathy patients and n=26 patients without arthropathy. ¥N=12 and n=27, respectively; med: median; sec: seconds; DAIMS2affect: Dutch Arthritis Impact Measurement Scales-2.

haematologica | 2017; 102(9)

1489


K.P.M. van Galen et al.

least 1 regular clotting factor infusion per week for at least 45 consecutive weeks, currently or in the past), and total clotting factor use (in units factor VIII/kg/year calculated over 5 years), were retrieved from the medical files and verified by the principal investigator (KG) by asking the participants during the study visit. The cumulative lifetime number of joint bleeds was categorized as a history of 0, 1, 2-5, 6-10 or >10 joint bleeds, based on medical file data and verified by asking the participants during the study visit.

Sample size A sample size of 50 VWD-JB patients and 50 VWD controls would be sufficient to detect a difference in joint function of 8 points in the HJHS, with a power of 80%, and significance level of 5% (2-sided). When the inter-rater variability is taken into account, differences of more than 7-10 points in the total HJHS are likely to represent clinically significant differences.12,21,22 With a sample size of 50 in each group we would be able to detect a difference of 17% in the proportion of patients with ‘radiological arthropathy’ and with ‘some functional limitations’, with a power of 80% and significance level of 5% (1-sided) based on an estimated prevalence of 5% in the control group.

Statistical analysis For statistical analysis, we used IBM SPSS v.23. To evaluate normal distribution of continuous scores, we used Q-Q plots. Mann-Whitney U and χ2 tests were used to compare continuous values and proportions, respectively. To explore whether BMI, relevant musculoskeletal comorbidity, and anxiety or mood (DAIMS2affect) could explain a difference in HJHS between the VWD-JB patients and VWD controls, we used multivariable negative binomial regression analysis because of the skewed distribution and excess of zeros of the HJHS on its continuous scale.23 Multivariable logistic regression analysis was performed to explore whether these determinants could explain possible differences in the occurrence of ‘radiological arthropathy’ and functional limitations (PS>3 and HAL total <95 as dependent variable, respectively). We used logistic regression analysis to explore whether FVIII, type 3 VWD, higher age, age at the first joint bleed, or the cumulative number of joint bleeds were associated with arthropathy (HJHS≥10 or PS>3) within the VWD-JB group and whether arthropathy was associated with clinically relevant joint pain (VAS mean >3) within the whole study cohort.

Table 4. Associations between the Hemophilia Joint Health Score (HJHS), Pettersson Score (PS) and Hemophilia Activity List (HAL) and co-variables. A. Among 48 von Willebrand disease-joint bleed (VWD-JB) patients compared to 48 VWD controls (independent variable).

Rate ratio

95% CI

P

2.5 2.4 2.9 2.8

1.6-3.9 1.6-3.8 1.9-4.7 1.8-4.3

<0.01 <0.01 <0.01 <0.01

7.7 BMI 7.7 D-AIMS2-affect 8.8 Relcom* 8.5 HAL <95 3.8 BMI 3.6 D-AIMS2-affect 4.6 Relcom* 4.0 B. Among 48 VWD-JB patients: exploration of possible predictors of arthropathy.

1.6-36 1.6-37 1.8-43 1.7-42 1.6-8.8 1.5-8.6 1.9-12 1.7-9.4

0.01 0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01

Dependent variable

Odds ratio

95% CI

P

4.6 5.1 6.9 1.0 2.2 2.2 2.1 7.9 10 7.8 6.8 1.0 3.8 3.9 3.4 17

1.3-16 1.5-18 0.8-57 1.0-1.1 0.7-7.4 1.2-4.0 1.0-4.1 2.0-31 1.9-53 1.8-35 0.6-84 1.0-1.1 1.0-15 1.4-11 1.2-10 2-149

0.02 0.01 0.07 0.28 0.19 0.01 0.04 <0.01 <0.01 <0.01 0.14 0.34 0.06 <0.01 0.02 0.01

Dependent variable HJHS

Co-variable

BMI D-AIMS2-affect Relcom*

Odds ratio

PS >3

HJHS ≥10

PS >3

Co-variable ¥

FVIII <10 IU/dL Type 3 VWD Type 3 VWD & FVIII¥ Age Age 1th JB<10 y No. joint bleeds# No. joint bleeds# & FVIII¥ > 5 joint bleeds in at least one joint FVIII <10 IU/dL¥ Type 3 VWD Type 3 VWD & FVIII¥ Age Age 1th JB <10 y No. joint bleeds# No. joint bleeds# & FVIII¥ > 5 joint bleeds in at least one joint

CI: Confidence Interval; HJHS: Hemophilia Joint Health Score; no.: number; PS: Pettersson score; HAL: Hemophilia Activity List questionnaire; FVIII: historically lowest Factor VIII level; D-AIMS2: Dutch Arthritis Impact; JB: joint bleed; Relcom: relevant comborbidity. BMI: Body Mass Index. *Possible restricted motion due to a musculoskeletal disorder for other medical reasons (excluding age-related osteoarthritis). ¥Historically lowest FVIII level. #Cumulative number of joint bleeds categorized (see Table 1).

1490

haematologica | 2017; 102(9)


Long-term impact of joint bleeds in VWD

Results Baseline characteristics In total, 119 patients were screened for participation: 19 patients did not want to participate and 4 were excluded on the basis of the exclusion criteria (n=1) or because they did not meet the inclusion criteria (n=3). A total of 48 VWD-JB patients and 48 VWD controls were included with a mean age of 46 years (range 18-80). Most patients had participated actively in the WiN study (95%: 44 of 48 VWD-JB patients and 47 of 48 VWD controls).2 Baseline characteristics are presented in Table 1. Despite matching, the median FVIII and VWF levels were lower in the VWDJB group, compared to the VWD control group (Table 1 and Online Supplementary Table S1). This was due to the wide range in age and clotting factor levels within the original WiN cohort, the low correlation between VWF and FVIII levels, and the relatively limited size of 804 subjects.2 Furthermore, due to the matching of clotting factor levels instead of subtype, more VWD-JB patients had type 3 VWD (19 vs. 3 VWD controls, P<0.01) (Table 1).

Body structure and function: occurrence and severity of arthropathy The VWD-JB patients were four times more likely to have developed arthropathy compared to the VWD controls (HJHS≥10 or PS>3 of one or more joints: OR 3.8, 95%CI: 1.4-10). This difference became somewhat stronger after correcting for the unverified joint bleeds in the control group (OR 6.8, 95%CI: 1.8-25). Arthropathy, as detected by HJHS≥10, occurred in 40% of the VWD-JB patients compared to 10% of the VWD controls (OR 5.8 95%CI: 2.0-17, P<0.01). VWD-JB patients had more severe and more often radiological joint abnormalities compared to the controls, especially in the ankles (Online Supplementary Table S2). Arthropathy based on X-rays (PS>3 of one or more joints) was found in 25% of the VWD-JB patients compared to 4% of the VWD controls (OR 7.7 95%CI: 1.6-37, P=0.01). None of the VWD-JB patients had a PS>3 in a contralateral joint without prior bleeds.

Self-perceived functional abilities and Figure 8 walk test VWD-JB patients reported more functional limitations in the HAL, compared to the VWD controls (median HAL total score 88 vs. 100) (Table 2). Some functional limitations (HAL total score <95) were reported by 67% of the VWD-JB patients compared to 35% of the controls (HAL total <95, OR 3.8; 95%CI: 1.6-8.8, P<0.01). No significant difference in performance according to the modified Figure 8 walk tests was found between the VWD-JB patients and VWD controls (Table 2). However, the VWDJB patients with arthropathy did perform worse on the Figure 8 test compared to those without arthropathy (Table 3).

IPA, environmental and personal factors, joint pain No significant differences in the IPA score (participation questionnaire) and level of anxiety or mood were found between the VWD-JB patients and VWD controls (Table 2). However, the VWD-JB patients with arthropathy had a lower score on the IPA compared those without arthropathy, corresponding to less social participation (Table 3). haematologica | 2017; 102(9)

Clotting factor prophylaxis because of joint bleeds was used by 58% (11 of 19) of the VWD-JB patients with arthropathy (HJHS≥10). BMI, relevant musculoskeletal comorbidity or mood/anxiety did not influence the difference in arthropathy between the VWD-JB patients and VWD controls in multivariable analysis (Table 4A). Clinically relevant joint pain (VAS mean >3 of ≥1 joints) was reported by 36% of the VWD-JB patients compared to 19% of the VWD controls (OR 2.4 95%CI:0.9-6.1, P=0.07) (Table 2). Within the VWD-JB patients, arthropathy HJHS≥10 was strongly associated with clinically relevant joint pain (OR 18 95%CI: 3.9-84, P<0.01). Arthropathy PS>3 showed a weaker association with clinically relevant joint pain in the VWD-JB patients (OR 3.6 95%CI: 0.93-14, P=0.06). Fifty-nine percent (10 of 17) of the VWD-JB patients with clinically relevant joint pain used or had used pain medication for joint pain.

Predictors of arthropathy Within the VWD controls, the patients with arthropathy HJHS≥10 were significantly older compared to the VWD controls with HJHS<10 (median age 63 vs. 46 years, P=0.03). In contrast, this age difference was not found within the VWD-JB patients (median age 50 in the VWDJB patients with arthropathy vs. 44 in those without arthropathy, P=0.38) (Table 3). More females than males had arthropathy within the VWD-JB patients (Table 3). The occurrence of arthropathy within the VWD-JB patients was associated with the cumulative number of joint bleeds. A low FVIII level less than 10 IU/dL at VWD diagnosis (in all 19 type 3, one type 2, and 3 type 1 VWDJB patients) was also associated with the occurrence of arthropathy. Type 3 VWD was associated with arthropathy, but not independent from FVIII levels (Table 4B). Patients who had their first joint bleed under the age of 11 years showed a trend towards the occurrence of ‘radiological arthropathy’ more often compared to the VWD-JB patients who had their first joint bleed over ten years of age (OR 3.8; 95%CI: 1.0-15, P=0.06). The VWD-JB patients who had a history of more than 5 joint bleeds in at least one joint (n=25 of 48) were also significantly more likely to have developed arthropathy (Table 4B). In a logistic regression model, the difference in the occurrence of arthropathy between the patients and controls was dependent on bleeding and not on FVIII levels: after adjustment for a history of more than 5 joint bleeds in at least one joint the OR changed from 3.8 to 1.8 (95%CI: 112, P=0.36), but it remained stable after adjustment for historically lowest FVIII level at 3.5 (95%CI: 1.3-9.5, P=0.02).

Discussion This nested case-control study demonstrates that 40% of patients with VWD and documented treatment for joint bleeds (VWD-JB patients) have arthropathy at physical examination (HJHS ≥10) and 25% ‘radiological arthropathy’ of ankles, knees or elbows (PS>3). Arthropathy occurred significantly less often in matched patients with VWD who had never received treatment for a joint bleed (10% HJHS≥10 and 4% PS>3). Arthropathy is related to clinically relevant joint pain and less social participation. The VWD-JB patients also reported significantly more functional limitations in the HAL question1491


K.P.M. van Galen et al.

naire, compared to the VWD controls. Within the VWDJB patient group, the most important predictors of the development of arthropathy were a FVIII level <10 IU/dL and the cumulative number of joint bleeds. Comparing VWD-JB patients to matched VWD controls without a relevant history of joint bleeds allowed us to get a good impression of the consequences of joint bleeds in VWD, independently of the burden of VWD itself and age-related osteoarthritis. More X-rays were taken from the VWD-JB patients compared to VWD controls, which could have led to a potential overestimation of the difference in the occurrence of ‘radiological arthropathy’ between the two groups. However, only one of the VWD controls had a PS>3 not related to joint bleeding. None of the VWD-JB patients had a PS>3 based on X-rays of a contralateral joint without a history of joint bleeds. For the first time, we used the HJHS to measure arthropathy in VWD. The HJHS has been validated to assess joint outcome in hemophilia.22 The cut off of HJHS≥10 to define arthropathy has been used in hemophilia patients with a lower mean age of 25 years.11 We found a median HJHS of 10 in VWD patients with a history of more than 5 joint bleeds within our validation study (on the same VWD cohort) which supports the rationale of a similar cut off for arthropathy in VWD.12 Nevertheless, we also observed 5 cases of arthropathy HJHS≥10 within the VWD controls, associated with older age. A higher cut off to define arthropathy would lead to loss of sensitivity. However, the incidence of osteoarthritis increases with age, especially after 50 years of age.24 An age-specific HJHS cut off to define arthropathy could be the subject of further study. Recall and information bias, including misclassification of joint symptoms for joint bleeds, probably hampered the reliability of the data on the cumulative number of joint bleeds.25 The occurrence of unverified joint bleeds in 14 VWD controls led to an underestimation of the difference in the occurrence of arthropathy between the VWD-JB patients and VWD controls. Further study is needed to find a better threshold for severity of joint bleeds to cause arthropathy than verification by documentation on treatment for joint bleeds. The strong association between a HJHS≥10 and clinically relevant joint pain could partly be explained by assigning points for joint pain in the HJHS assessment (max. 1 point for each of 6 joints). This study is the first to report functional consequences of joint bleeds in VWD. It appears that the prevalence of arthropathy due to joint bleeds in the total population of patients with moderate to severe VWD is less than 6%, since 44 of the VWD-JB patients within the current study were selected from the original WiN cohort (n=804).

References 1. Leebeek FW and Eikenboom JC. Von Willebrand's Disease. N Engl J Med. 2016;375(21):2067-2080. 2. de Wee EM, Sanders YV, MauserBunschoten EP, et al. Determinants of bleeding phenotype in adult patients with moderate or severe von Willebrand disease.

1492

However, not all eligible WiN patients participated in the current study. Literature on joint damage due to joint bleeds is scarce in VWD and arthropathy has not been well defined.5 More studies on arthropathy should be conducted in other VWD cohorts to confirm the validity of our results. In hemophilia, progressive arthropathy predominantly occurs when patients report more than 5 joint bleeds before the start of prophylaxis.26 A history of more than 5 treated joint bleeds was also associated with arthropathy in the current study, which is in accordance with the findings of our cross-sectional study.4 As in hemophilia, we found VWD arthropathy mostly in the ankle joints (19% of VWD JB patients had a PS ankle >3).22 In mild and moderate hemophilia, ankle arthropathy, less strictly defined as a PS>0, occurred in 48% of the patients and was also associated with FVIII levels less than 10 IU/dL and with ankle bleeding at a young age.27 The HJHS is suitable to detect arthropathy in VWD and physiotherapists should be involved in patient care after joint bleeds, as in hemophilia, to obtain complete recovery.28 To get a better impression of the prevalence of arthropathy in VWD, upfront joint assessment is needed in future population studies on VWD, especially those VWD patients with type 3 VWD, low FVIII levels, and those who received prior treatment for joint bleeds. The VWD patients with the poorest joint outcome were also the most heavily treated patients. Fifty-eight percent of the patients with an HJHS≥10 had used clotting factor prophylaxis. However, the association of arthropathy with clinically relevant joint pain, lower health-related quality of life,4 functional limitations and less social participation, suggests that there might be room for more intensive treatment of joint bleeds in VWD. In accordance with the ICF model, this should include a rehabilitation program aimed at full functional recovery. It remains to be determined whether more intensive prophylaxis to prevent arthropathy and improve participation in VWD would be cost-effective.29 In conclusion, arthropathy as detected by the Hemophilia Joint Health Score occurs in almost half of VWD patients treated for joint bleeds; this is consistent with more radiological joint abnormalities compared to matched VWD controls without a relevant history of joint bleeds. VWD patients with arthropathy reported more functional limitations and less social participation compared to those without arthropathy, and clinically relevant joint pain in the majority of cases. Funding This research was supported by an unrestricted grant from CLS Behring and the Dutch Hemophilia Foundation

Thromb Haemost. 2012;108(4):683-692. 3. Jansen NW, Roosendaal G, Lafeber FP. Understanding haemophilic arthropathy: an exploration of current open issues. Br J Haematol. 2008;143(5):632-640. 4. van Galen KP, Sanders YV, Vojinovic U, et al. Joint bleeds in von Willebrand disease patients have significant impact on quality of life and joint integrity: a cross-sectional study. Haemophilia. 2015;21(3):e185-e192.

5. van Galen KP, Mauser-Bunschoten EP, Leebeek FW. Hemophilic arthropathy in patients with von Willebrand disease. Blood Rev. 2012;26(6):261-266. 6. Jette AM and Keysor JJ. Disability models: implications for arthritis exercise and physical activity interventions. Arthritis Rheum. 2003;49(1):114-120. 7. Fischer K, Nijdam A, Holmstrom M, et al. Evaluating outcome of prophylaxis in

haematologica | 2017; 102(9)


Long-term impact of joint bleeds in VWD

8.

9. 10.

11.

12.

13.

14.

haemophilia: objective and self-reported instruments should be combined. Haemophilia. 2016 Feb 8 [Epub ahead of print] de Wee EM, Leebeek FWG, Eikenboom JCJ. Diagnosis and Management of von Willebrand Disease in The Netherlands. Semin Thromb Hemost. 2011;37(5):480487. Hilliard P, Funk S, Zourikian N, et al. Hemophilia joint health score reliability study. Haemophilia. 2006;12(5):518-525. Soucie JM, Wang C, Forsyth A, et al. Range of motion measurements: reference values and a database for comparison studies. Haemophilia. 2011;17(3):500-507. Fischer K, Steen CK, Petrini P, et al. Intermediate-dose versus high-dose prophylaxis for severe hemophilia: comparing outcome and costs since the 1970s. Blood. 2013;122(7):1129-1136. van Galen KP, Timmer M, de Kleijn P, et al. Joint Assessment in Von Willebrand Disease: Validation of the Haemophilia Joint Health Score and Haemophilia Activities List. Thromb Haemost. 2017 May 11 [Epub ahead of print] Foppen W, van der Schaaf IC, Beek FJ, Verkooijen HM, Fischer K. Scoring haemophilic arthropathy on X-rays: improving inter- and intra-observer reliability and agreement using a consensus atlas. Eur Radiol. 2016;26(6):1963-1970. Foppen W, van der Schaaf IC, Beek FJ, Verkooijen HM, Fischer K. Scoring haemophilic arthropathy on X-rays: improving inter- and intra-observer reliability and agreement using a consensus atlas. Eur Radiol. 2016;26(6):1963-1970.

haematologica | 2017; 102(9)

15. van Genderen FR, van Meeteren NL, van der Bom JG, et al. Functional consequences of haemophilia in adults: the development of the Haemophilia Activities List. Haemophilia. 2004;10(5):565-571. 16. van Genderen FR, Westers P, Heijnen L, et al. Measuring patients' perceptions on their functional abilities: validation of the Haemophilia Activities List. Haemophilia. 2006;12(1):36-46. 17. Nijdam A, Foppen W, de Kleijn P, et al. Discontinuing early prophylaxis in severe haemophilia leads to deterioration of joint status despite low bleeding rates. Thromb Haemost. 2016;115(5):931-938. 18. Cardol M, de Haan RJ, van den Bos GA, de Jong BA, de Groot IJ. The development of a handicap assessment questionnaire: the Impact on Participation and Autonomy (IPA). Clin Rehabil. 1999;13(5):411-419. 19. van der Kloot WA, Oostendorp RA, van der Meij J, van den Heuvel J. [The Dutch version of the McGill pain questionnaire: a reliable pain questionnaire]. Ned Tijdschr Geneeskd. 1995;139(13):669-673. 20. de Joode EW, van Meeteren NL, van den Berg HM, de Kleijn P, Helders PJ. Validity of health status measurement with the Dutch Arthritis Impact Measurement Scale 2 in individuals with severe haemophilia. Haemophilia. 2001;7(2):190-197. 21. Feldman BM and Pullaneyagum E. Response to 'Limits of agreement between raters are required for use of HJHS 2.1 in clinical studies'. Haemophilia. 2015; 21(1):e71. 22. Fischer K and de Kleijn P. Using the Haemophilia Joint Health Score for assessment of teenagers and young adults:

23.

24.

25.

26.

27.

28.

29.

exploring reliability and validity. Haemophilia. 2013;19(6):944-950. den Uijl IE, Fischer K, van der Bom JG, Grobbee DE, Rosendaal FR, Plug I. Analysis of low frequency bleeding data: the association of joint bleeds according to baseline FVIII activity levels. Haemophilia. 2011;17 (1):41-44. Oliveria SA, Felson DT, Reed JI, Cirillo PA, Walker AM. Incidence of symptomatic hand, hip, and knee osteoarthritis among patients in a health maintenance organization. Arthritis Rheum. 1995;38(8):11341141. Ceponis A, Wong-Sefidan I, Glass CS, von Drygalski A. Rapid musculoskeletal ultrasound for painful episodes in adult haemophilia patients. Haemophilia. 2013;19(5):790-798. Kreuz W, Escuriola-Ettingshausen C, Funk M, Schmidt H, Kornhuber B. When should prophylactic treatment in patients with haemophilia A and B start?--The German experience. Haemophilia. 1998;4(4):413-417. Ling M, Heysen JP, Duncan EM, Rodgers SE, Lloyd JV. High incidence of ankle arthropathy in mild and moderate haemophilia A. Thromb Haemost. 2011;105(2):261-268. de Kleijn P, Gilbert M, Roosendaal G, Poonnose PM, Narayan PM, Tahir N. Functional recovery after bleeding episodes in haemophilia. Haemophilia. 2004;10 Suppl 4:157-160. Abshire TC, Federici AB, Alvarez MT, et al. Prophylaxis in severe forms of von Willebrand's disease: results from the von Willebrand Disease Prophylaxis Network (VWD PN). Haemophilia. 2013;19(1):76-81.

1493


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Coagulation & its Disorders

Ferrata Storti Foundation

Comparison of risk prediction scores for venous thromboembolism in cancer patients: a prospective cohort study Nick van Es,1 Marcello Di Nisio,2 Gabriela Cesarman,3 Ankie Kleinjan,1 Hans-Martin Otten,4 Isabelle Mahé,5 Ineke T. Wilts,6 Desirée C. Twint,7 Ettore Porreca,8 Oscar Arrieta,3 Alain Stépanian,9 Kirsten Smit,7 Michele De Tursi,8 Suzanne M. Bleker,1 Patrick M. Bossuyt,10 Rienk Nieuwland,11 Pieter W. Kamphuisen6,12 and Harry R. Büller1

Department of Vascular Medicine, Academic Medical Center, Amsterdam, the Netherlands; 2Department of Medicine and Ageing Sciences, G. D’Annunzio University, Chieti, Italy; 3Department of Hematology, National Cancer Institute Mexico, Mexico City, Mexico; 4Department of Internal Medicine, Slotervaart hospital, Amsterdam, the Netherlands; 5Department of Internal Medicine, Hôpital Louis Mourier, Paris, France; 6 Department of Internal Medicine, University Medical Center Groningen, the Netherlands; 7Department of Internal Medicine, VU Medical Center, Amsterdam, the Netherlands; 8Department of Medical, Oral and Biotechnological Sciences, G. D’Annunzio University, Chieti, Italy; 9Department of Hematology, Hôpital Lariboisière, Paris, France; 10Department of Clinical Epidemiology, Biostatistics, and Bioinformatics, Academic Medical Center, Amsterdam, the Netherlands; 11Department of Experimental Clinical Chemistry, Academic Medical Center, Amsterdam, the Netherlands and 12 Department of Internal Medicine, Tergooi Hospital, Hilversum, the Netherlands 1

Haematologica 2017 Volume 102(9):1494-1501

ABSTRACT

I

Correspondence: n.vanes@amc.nl

Received: March 17, 2017. Accepted: May 25, 2017. Pre-published: May 26, 2017. doi:10.3324/haematol.2017.169060 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1494 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1494

n ambulatory patients with solid cancer, routine thromboprophylaxis to prevent venous thromboembolism is not recommended. Several risk prediction scores to identify cancer patients at high risk of venous thromboembolism have been proposed, but their clinical usefulness remains a matter of debate. We evaluated and directly compared the performance of the Khorana, Vienna, PROTECHT, and CONKO scores in a multinational, prospective cohort study. Patients with advanced cancer were eligible if they were due to undergo chemotherapy or had started chemotherapy in the previous three months. The primary outcome was objectively confirmed symptomatic or incidental deep vein thrombosis or pulmonary embolism during a 6-month followup period. A total of 876 patients were enrolled, of whom 260 (30%) had not yet received chemotherapy. Fifty-three patients (6.1%) developed venous thromboembolism. The c-statistics of the scores ranged from 0.50 to 0.57. At the conventional positivity threshold of 3 points, the scores classified 13-34% of patients as high-risk; the 6-month incidence of venous thromboembolism in these patients ranged from 6.5% (95%CI: 2.8-12) for the Khorana score to 9.6% (95%CI: 6.6-13) for the PROTECHT score. High-risk patients had a significantly increased risk of venous thromboembolism when using the Vienna (subhazard ratio 1.7; 95%CI: 1.0-3.1) or PROTECHT (subhazard ratio 2.1; 95%CI: 1.23.6) scores. In conclusion, the prediction scores performed poorly in predicting venous thromboembolism in cancer patients. The Vienna CATS and PROTECHT scores appear to discriminate better between low- and high-risk patients, but further improvements are needed before they can be considered for introduction into clinical practice.

Introduction Venous thromboembolism (VTE) complicates the clinical course in 4-5% of cancer patients1,2 and is a major cause of morbidity and mortality.3 The management of VTE in cancer patients is particularly challenging since both the risk of recurrent VTE and major bleeding are high during anticoagulant treatment. Current international guidelines do not recommend routine pharmacological thromboprophylaxis in ambulatory cancer patients.4-6 Low-molecular weight heparin (LMWH) in prohaematologica | 2017; 102(9)


Predicting cancer-associated venous thromboembolism

phylactic doses halves the risk of VTE1 and is associated with an absolute risk reduction of 2-2.5%. However, the corresponding number needed to treat (40-50) using thromboprophylaxis is considered too low to justify the potential increased risk of bleeding and the burden of daily subcutaneous injections for a prolonged period of time.4 Risk stratification tools have been developed to identify a subset of cancer patients in whom the risk of developing VTE is high enough to justify thromboprophylaxis. The best validated tool is a score proposed by Khorana and colleagues7 which aims to identify cancer patients receiving chemotherapy at high risk of VTE based on the tumor type, hemoglobin concentration or use of erythropoietin stimulating agents, white blood cell count, platelet count, and Body Mass Index (BMI) (Table 1). To improve the discriminatory performance of the Khorana score, others have proposed modifications by adding biomarker measurements8 or type of chemotherapy,9 or by replacing BMI with performance status (Table 1).10 Although these scores performed well in the initial derivation studies, there have either been no subsequent external validation studies or those that have been carried out have reported conflicting results.11-13 Yet, it is important that performance is maintained across different patient populations and settings. In addition, scores have not been directly compared in a large study of representative

patients. To fill this gap, we evaluated and directly compared four clinical prediction scores for VTE in patients with advanced cancer receiving chemotherapy in a multinational cohort study.

Methods Study design and patients Data were collected in a multinational, prospective cohort study performed in seven hospitals in The Netherlands, Italy, France, and Mexico, designed to evaluate clinical and laboratory predictors for cancer-associated VTE. Here we report on the performance of four published clinical prediction scores for cancer-associated VTE. Outpatients with lung, esophageal, colorectal, pancreatic, breast, prostate, gastric, ovarian, or bladder cancer classified as stage III or IV according to the American Joint Committee on Cancer criteria were eligible if they were scheduled for chemotherapy within seven days or had started chemotherapy in the previous three months. Exclusion criteria included current prophylactic or therapeutic anticoagulation or adjuvant chemotherapy. None of the patients included had received routine thromboprophylaxis in accordance with current guidelines. Patients were recruited between July 2008 and February 2016. The study was approved by the institutional review boards of all participating hospitals. Patients included in the study provided written informed consent. The study was registered at clinicaltrials.gov

B Cumulative incidence

Cumulative incidence

A

Days since inclusion

Days since inclusion

D Cumulative incidence

Cumulative incidence

C

Days since inclusion

Days since inclusion

Figure 1. Cumulative incidence of venous thromboembolism in low- and high-risk patients. Cumulative incidence of venous thromboembolism in patients enrolled prior to chemotherapy (n=260) who were classified as being at low or high risk of venous thromboembolism by the (A) Khorana score, (B) Vienna CATS score, (C) PROTECHT score, and (D) CONKO score.

haematologica | 2017; 102(9)

1495


N. van Es et al.

(identifier: 02095925) after the enrollment of the first patient. The present report adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement.14

Study procedures At baseline, healthcare professionals interviewed patients and collected baseline clinical and laboratory information from the medical charts using a standardized case report form. Patients were enrolled at oncology or vascular medicine departments. Blood was drawn in 0.109 M citrated tubes via antecubital venepuncture or through a peripheral catheter shortly after placement. Within one hour after blood collection, platelet poor plasma was prepared by centrifugation at 1560 g for 20 minutes and stored at -80°C after snap freezing. A follow-up visit was scheduled at 180 days from the clinic visit by telephone and/or chart review. The primary outcome was the composite of objectively confirmed symptomatic or incidental pulmonary embolism (PE), distal or proximal leg deep vein thrombosis (DVT), or non-catheter-related upper extremity DVT (UEDVT), or symptomatic catheter-related UEDVT. VTE was considered incidental if diagnosed on imaging performed for reasons other than suspicion of VTE. Incidental events were included in the primary outcome since current guidelines suggest a treatment similar to that for symptomatic events. Patients did not undergo screening for VTE. Imaging reports of all potential outcomes were verified by 2 of the authors who were blinded to possible predictors of VTE. Deaths were not adjudicated routinely for fatal PE, but PE was considered to be fatal only if autopsy confirmed PE or in the case of an objective test positive for PE prior to death.

Evaluation of prediction scores We evaluated the following clinical prediction scores for cancerassociated VTE: the Khorana,7 Vienna CATS,8 PROTECHT,9 and CONKO scores10 (Table 1). Characteristics of the derivation studies are provided in the Online Supplementary Appendix text 1. Evaluation of the Khorana score was included in the initial study protocol, while the other scores were published during the course of the study and were subsequently included in the analysis. We were unable to evaluate a recently proposed genetic risk score.15 In order to calculate the Vienna CATS score, D-dimer (INNO-

VANCE, Siemens) and soluble P-selectin (ELISA, R&D Systems, Minneapolis, MN, USA) concentrations were measured centrally in baseline samples. Analysis of samples of 98 patients from one center provided implausible results due to incorrect shipment, forcing us to omit these measurements, while the patients were retained in the dataset and among the imputed data. For continuous scores, the overall discriminatory performance was evaluated. For dichotomized scores, we calculated the proportion of high-risk patients, the cumulative VTE incidence among high-risk patients, the cumulative VTE incidence among low-risk patients, and the difference in VTE incidence between low- and high-risk patients. Since the scores will be used in a dichotomous fashion in clinical practice by conflating the low and intermediate groups, we evaluated them at the conventional positivity threshold of 3 points, and at exploratory positivity thresholds of 2 and 4 points. Since the derivation of the Khorana, PROTECHT, Vienna CATS, and CONKO scores was almost entirely based on symptomatic VTE, a sensitivity analysis restricted to symptomatic events was performed. Another sensitivity analysis, restricted to the first 90 days of follow up, was performed for comparison with Khorana’s derivation study which had a median follow up of 2.5 months.7 Our study was designed to include cancer patients prior to or within three months of the start of chemotherapy. Since prechemotherapy blood counts are incorporated into all evaluated clinical prediction scores, we restricted the main analysis to the group of patients who had not yet received chemotherapy. The analyses were then repeated in the complete study group, calculating each score based on pre-chemotherapy blood counts, which were collected retrospectively at inclusion in patients who had already started chemotherapy. Results are thus presented for the group that mirrors cancer patients in whom the decision about thromboprophylaxis is made prior to chemotherapy and for all patients, including those for whom the question as to whether to provide thromboprophylaxis is discussed during the first months of chemotherapy. Assuming a 6-month VTE incidence of 5-6%, the aim was to enroll approximately 800-1000 patients in order to observe about 50 events, which was considered to be sufficient for multivariable regression modeling with the five items in the Khorana score.

Table 1. Risk prediction scores for venous thromboembolism in cancer patients.

Item

Pancreatic or gastric cancer (very high-risk tumors) Lung, gynecological, lymphoma, bladder, or testicular (high-risk tumors) Pre-chemotherapy hemoglobin <10 g/dL or use of erythropoietin stimulating agents Pre-chemotherapy white blood cell count >11 x 109/L Pre-chemotherapy platelet count ≥350 x 109/L Body Mass Index >35 kg/m2 D-dimer >1.44 mg/L Soluble P-selectin >53.1 ng/L Gemcitabine chemotherapy Platinum-based chemotherapy WHO performance status ≥2

Khorana score (points)

Vienna CATS score (points)

PROTECHT score (points)

CONKO score (points)

+2 +1 +1 +1 +1 +1 -

+2 +1 +1 +1 +1 +1 +1 +1 -

+2 +1 +1 +1 +1 +1 +1 +1 -

+2 +1 +1 +1 +1 +1

WHO: World Health Organization.

1496

haematologica | 2017; 102(9)


Predicting cancer-associated venous thromboembolism

Statistical analysis The cumulative VTE incidence from enrollment to six months in low- and high-risk patients was estimated using a competing risk time-to-event analysis in which death was treated as a competing event for VTE. Unlike naïve Kaplan-Meier and Cox regression analysis, a competing risk analysis does not treat death as a censored observation, but rather takes into account that VTE cannot occur after death, thereby providing less biased estimates.16 Time to VTE was considered censored when patients underwent cancer surgery with curative intent, started therapeutic anticoagulation for other reasons than VTE, were lost to follow up, or at the end of the 6-month follow-up period. Confidence intervals (CI) at specific time points were calculated using Choudhury’s method.17 To evaluate the discriminatory performance of the scores, the time-dependent concordance index (c-index) was calculated while accounting for death as a competing risk.18 The 95% confidence intervals were calculated by repeating the analyses in 250 bootstrap samples. To assess the difference in VTE risk between low- and high-risk patients, subdistribution hazard ratios (SHR) and 95% CIs for the dichotomized scores were estimated using the competing risks regression model of Fine and Gray.19 The predictive value of the separate items in each score was assessed by multivariable competing risks regression models. The proportionality assumption

was checked by adding an interaction term between each variable and time to the model. Multiple imputation was used to minimize the bias associated with missing data.20 We assumed a 'missing' at random pattern in which missingness depends on other observed variables. Center, inclusion year, all baseline characteristics, and outcome data were included in the imputation model to create twenty imputed datasets. Analyses were performed separately in each imputed dataset; estimates with standard errors were combined across the datasets using Rubin’s rule.21 The complete case analyses were provided for comparison. A significance level of 0.05 was used in statistical testing. All analyses were performed in R, v.3.3.2 (R Foundation for Statistical Computing, Vienna, Austria; www.R-project.org), in particular using the “mice” v.2.25 package for multiple imputation, the “cmprsk” v.2.2-7 for the competing risk analyses, and the “pec” package v.2.4.9 for the time-dependent c-indices.

Results During the 7.5-year study period, 876 patients with stage III or IV solid cancer were enrolled, of whom 260 (30%) had not yet started chemotherapy. The mean age was 64 years; 59% of patients were male. Baseline charac-

Table 2. Baseline characteristics.

Characteristic

Prior to chemotherapy (N=260)

All patients (N=876)

63 (10) 156 (60)

64 (11) 516 (59)

25 (4.3) 7 (2.7) 25 (9.6) 3 (1.2) 26 (10)

25 (4) 27 (3.1) 79 (8) 13 (1.5) 109 (12)

73 (28) 47 (18) 50 (19) 36 (14) 18 (6.9) 11 (4.2) 16 (6.2) 7 (2.7) 2 (0.8) 179 (69) 34 (13) 157 (60) 21 (8.1) 15 (5.8) 45 (17) 67 (26) 76 (29) 33 (13)

224 (26) 170 (19) 158 (18) 109 (12) 82 (9.4) 41 (4.7) 40 (4.6) 39 (4.5) 13 (1.5) 581 (66) 132 (15) 518 (59) 58 (6.6) 64 (7.3) 141 (16) 229 (26) 274 (31) 87 (9.9)

Age, y, mean (SD) Male, n (%) Body Mass Index Mean (SD), kg/m2 >35 kg/m2, n (%) WHO performance status ≥2, n (%) Previous VTE, n (%) Antiplatelet therapy, n (%) Tumor type, n (%) Lung Esophagus Colorectal Pancreas Breast Prostate Gastric Ovarian Bladder Distant metastasis, n (%) Gemcitabine, n (%) Platinum-based chemotherapy, n (%) Erythropoietin stimulating agents, n (%) Pre-chemotherapy hemoglobin <10 g/dL, n (%) Pre-chemotherapy white blood cell count >11 x 109/L, n (%) Pre-chemotherapy platelet count ≥350 x 109/L, n (%) D-dimer >1.44 mg/L Soluble P-selectin >53.1 ng/L y: years; SD: Standard Deviation; VTE: venous thromboembolism; N, n: number; WHO: World Health Organization.

haematologica | 2017; 102(9)

1497


N. van Es et al.

Table 3. Performance of different scores in patients enrolled prior to chemotherapy (n=260) and all patients (n=876).

Patients enrolled prior to chemotherapy (n=260) Time-dependent c-index at 180 days (95% CI) High-risk patients (≥3 points), % (95% CI) 6-month VTE risk in low-risk patients (≤2 points), % (95% CI) 6-month VTE risk in high-risk patients (≥3 points), % (95% CI) Subhazard ratio for high- (≥3 points) vs. low-risk patients (≤2 points) All patients (n=876) Time-dependent c-index at 180 days (95% CI) High-risk patients (≥3 points), % (95% CI) 6-month VTE risk in low-risk patients (≤2 points), % (95% CI) 6-month VTE risk in high-risk patients (≥3 points), % (95% CI) SHR for high- (≥3 points) vs. low-risk patients (≤2 points)

Khorana score

Vienna CATS score

PROTECHT score

CONKO score

0.50 (0.42-0.57) 13 (9.5-18) 8.4 (5.1-13) 6.0 (1.0-18) 0.69 (0.17-2.9)

0.57 (0.48-0.66) 31 (26-37) 7.9 (4.4-13) 8.4 (3.5-16) 1.1 (0.41-2.7)

0.54 (0.45-0.63) 34 (28-40) 7.4 (4.0-12) 9.5 (4.4-17) 1.3 (0.53-3.15)

0.50 (0.44-0.57) 15 (11-20) 8.6 (5.3-13) 5.2 (0.9-16) 0.58 (0.14-2.4)

0.52 (0.47-0.58) 13 (11-15) 6.5 (4.8-8.4) 6.5 (2.8-12) 1.0 (0.46-2.3)

0.58 (0.51-0.66) 27 (24-30) 5.5 (3.8-7.6) 9.1 (5.8-13) 1.7 (1.0-3.1)

0.59 (0.52-0.66) 35 (31-38) 4.7 (3.1-6.8) 9.6 (6.6-13) 2.1 (1.2-3.6)

0.53 (0.47-0.59) 15 (13-18) 6.4 (4.7-8.4) 7.1 (3.5-12) 1.1 (0.56-2.4)

CI: Confidence Interval; SHR: subdistribution Hazard Ratio; VTE: venous thromboembolism; vs.: versus; n: number.

teristics of the patients enrolled prior to chemotherapy and the complete study group are summarized in Table 2. No significant differences were found between patients enrolled prior to chemotherapy and those enrolled during chemotherapy, except for higher P-selectin levels in the former group. The distribution of the scores are summarized in Online Supplementary Appendix text 2. Multiple imputation was used to calculate the Khorana in 33 patients (4%), the Vienna CATS score in 134 patients (15%), the PROTECHT score in 38 patients (4%), and the CONKO score in 47 patients (5%), in whom data on one or more of the score items were missing. Overall, 53 patients (6.1%) developed VTE during the 6month follow-up period, of whom 27 only had PE (3.1%), 15 had proximal DVT (1.7%), 7 had UEDVT (0.8%), 3 had PE and DVT (0.3%), and one had isolated distal DVT (0.1%). The corresponding cumulative incidence in the competing risks analysis was 6.5% (95%CI: 4.9-8.3%) at 180 days. VTE was symptomatic in 33 cases (62%). Two events (3.8%) were fatal. The median time to VTE was 57 days (interquartile range 33-116). Ten patients (1.1%) were lost to follow up and 163 patients (19%) died.

Performance of scores Of the 260 patients who were enrolled prior to chemotherapy, 20 (7.7%) developed VTE, including 13 symptomatic events. In this group, the area under the ROC curves for the different scores (reflecting discriminatory performance) ranged from 0.50 (95%CI: 0.44-0.57) for the CONKO score and 0.50 (95%CI: 0.42-0.57) for the 1498

Khorana score to 0.57 (95%CI: 0.48-0.66) for the Vienna CATS score (Table 3). At the conventional positivity threshold of 3 points, the proportion of patients classified as high-risk ranged from 13% (95%CI: 9.5-18%) for the Khorana score to 34% (95%CI: 28-40%) for the PROTECHT score (Table 3). The 6-month VTE incidence in these high-risk patients ranged from 5.2% (95%CI: 0.916%) for the CONKO score to 9.5% (95%CI: 4.4-17%) for the PROTECHT score (Table 3; see Figure 1 for timeto-event curves). At the evaluated positivity thresholds of 2, 3, or 4 points, none of the scores was able to discriminate between low- and high-risk patients (Online Supplementary Appendix text 3). The analyses were repeated in all 876 patients, including the 616 patients enrolled after the start of chemotherapy (median 28 days; interquartile range 16-56) for whom blood counts collected before chemotherapy were used to calculate the scores. Overall, results were comparable to the analyses restricted to patients enrolled prior to chemotherapy. The area under the ROC curves ranged from 0.52 (95%CI: 0.47-0.58) for the Khorana score to 0.59 (95%CI: 0.52-0.66) for the PROTECHT score (Table 3). At the conventional positivity threshold of 3 points, the Khorana score again classified 13% (95%CI: 11-15%) as high-risk, whereas the PROTECHT score classified 35% of patients (95%CI: 31-38%) as high-risk (Table 3). Among high-risk patients, the cumulative VTE incidence at six months ranged from 6.5% (95%CI: 2.8-12%) when using the Khorana score to 9.6% (95%CI: 6.6-13%) when using the PROTECHT score (Table 3). Subhazard ratios haematologica | 2017; 102(9)


Predicting cancer-associated venous thromboembolism

for VTE in high- versus low-risk patients ranged from 1.0 (95%CI: 0.46-2.2) for the Khorana score to 1.7 (95%CI: 1.0-3.1) for the Vienna CATS score and 2.1 (95%CI: 1.23.7) for the PROTECHT score (Table 3; see Online Supplementary Appendix text 4 for time-to-event curves). At positivity thresholds of 2 or 4 points, the difference in VTE risk between high- and low-risk patients was not significant for any of the scores (Online Supplementary Appendix text 3).

Sensitivity analyses The sensitivity analysis restricted to symptomatic VTE in the total study group yielded comparable results (Online Supplementary Appendix text 5). The 6-month risk of symptomatic VTE in high-risk patients ranged from 3.7% (95%CI: 1.2-8.6%) for the Khorana score to 6.2% (95%CI: 3.8-9.4%) for the PROTECHT score. In the sensitivity analysis restricted to the first 90 days of follow up, in which 34 of 876 patients (3.9%) developed VTE, the discriminatory performance of all scores was slightly better (Online Supplementary Appendix text 6). The 90-day VTE incidence in high-risk patients (â&#x2030;Ľ3 points) ranged from 3.6% (95%CI: 1.2-8.4%) for the Khorana score to 6.4% (95%CI: 3.7-10%) for the Vienna CATS score and 6.4% (95%CI: 4.0-9.6%) for the PROTECHT score. The results from the complete case analysis did not substantially differ from the analysis of the imputed datasets (Online Supplementary Appendix text 7). Results of the multivariable analyses are shown in Online Supplementary Appendix text 8. None of the Khorana score items were significantly associated with VTE in any of the scores. In the Vienna CATS score, the dichotomized D-dimer result was significantly associated with VTE (SHR 2.4; 95%CI: 1.3-4.4), while in the PROTECHT score, both gemcitabine (SHR 3.7; 95%CI: 1.8-7.6) and platinum-based chemotherapy (SHR 2.8; 95%CI: 1.4-5.6) were associated with VTE conditional on the other items.

Discussion This multinational, prospective cohort study provides a direct comparison of the performance of four clinical and biomarker-based prediction scores for VTE in patients with advanced solid cancer receiving chemotherapy. All scores had a poor discriminatory performance, although the Vienna CATS and PROTECHT scores were able to discriminate between high- and low-risk patients when used dichotomously in the complete study group. The 6month VTE incidence among patients classified as highrisk by these two scores was approximately 2-fold higher than in low-risk patients. The poor overall discriminatory performance of the scores could partly be explained by the findings of the multivariable analysis. Hemoglobin levels, white blood cell counts, and platelet counts were not significantly associated with VTE. In a large prospective cohort study, Posch and colleagues also showed that the predictive performance of these items was limited,22 thereby questioning the relevance of these items for VTE risk prediction in cancer patients. In addition, in the present study, only 3% of patients had a BMI over 35 kg/m2, of whom none developed VTE. The low prevalence of obesity among cancer patients led Pelzer and colleagues to propose the CONKO score, in which BMI was replaced with World Health haematologica | 2017; 102(9)

Organization performance status.10 However, we did not observe any significant improvement in discriminatory performance of this modified score compared to the Khorana score. The Vienna CATS and PROTECHT score were the only two scores that could identify high-risk patients when used dichotomously. However, this significant association was observed in the complete study group, but not in patients enrolled prior to chemotherapy. Since the characteristics of patients enrolled prior to and during chemotherapy were comparable, possible explanations include differences in unobserved confounders or imprecision in the estimates. The predictive performance of the two scores in the complete study group appeared to be predominantly driven by the predictive performance of Ddimer levels in the Vienna CATS score and type of chemotherapy in the PROTECHT score, which could explain why they performed better than the Khorana and CONKO scores. Drawbacks of the Vienna CATS score include measuring D-dimer and soluble P-selectin, which may be difficult in clinical practice, while the PROTECHT score is more complex than the Khorana score because another two items are added. Moreover, the risk of VTE patients classified as low-risk by these scores was still appreciable, with 6-month rates of 5-6%, compared to 810% in high-risk patients. The Khorana score has already been extensively evaluated in various studies with different patient populations and duration of follow up. In studies restricting enrollment to single tumor types, such as pancreatic13,23 and lung cancer,12 the Khorana score was often not able to discriminate between low- and high-risk patients, while other studies did confirm the scoreâ&#x20AC;&#x2122;s predictive ability.8,24,25 In addition, in the present study, differences in case-mix compared to the derivation study may partly explain the poor performance. For example, in the present study, the proportion of patients with esophageal cancer was higher and enrollment was restricted to patients with advanced cancer. However, evaluating clinical prediction score in other settings and populations is the very essence of external validation studies; a score should maintain its predictive performance across a broad spectrum of patients, including those more recently treated, as well as different geographical locations, before it can be adopted into clinical practice. It has to be noted that the original Khorana score was derived in a study with a median follow up of only 2.5 months and that not all cancer types were represented. Whether results should be extrapolated to other tumor types, and whether clinical and laboratory information obtained prior to chemotherapy remains predictive beyond the initial cancer treatment period remains uncertain. This notion was supported by additional analyses, in which we observed that the area under the time-dependent ROC-curves decreased rapidly during the first weeks of follow up (data not shown). Similarly, in the sensitivity analysis restricted to the first three months of follow up, the discriminatory performance of the dichotomized Vienna CATS and PROTECHT scores appeared to improve somewhat compared to the complete follow-up period. The present prospective study is one of the largest in which multiple prediction scores for cancer-associated VTE were evaluated in a representative sample of patients with various tumor types. By restricting enrollment to 1499


N. van Es et al.

patients with locally advanced or metastatic solid cancers, the majority of the patients received long-term palliative rather than short-term neoadjuvant chemotherapy, resembling the populations enrolled in trials evaluating thromboprophylaxis in ambulatory cancer patients.26,27 Loss to follow up was minimal and outcome events were centrally verified blinded to the score results to prevent outcome bias. The multinational design of the study prevented potential single-center bias and strengthens the external validity of the findings.28 Our study has several limitations that merit consideration. We realize that the evaluated prediction scores all include pre-chemotherapy blood counts, while our study was designed to also include patients who had recently started chemotherapy. The main analysis was therefore restricted to 260 patients who were about to start their anticancer treatment. For patients who were receiving chemotherapy at enrollment, retrospectively collected pre-chemotherapy blood counts were used to calculate the scores to avoid potential effects of chemotherapy. The number of events observed in patients enrolled prior to chemotherapy (n=20) was sufficient to detect potential differences in VTE incidence between low- and high-risk patients, although the power may have been somewhat lower. Given the small difference in VTE risk between low- and high-risk patients, it is unlikely that a larger sample of patients enrolled prior to chemotherapy would have altered our conclusions. Along the same line, the enrollment of patients who had already started chemotherapy may have resulted in immortal time bias. Since the risk of VTE is usually highest in the first months of chemotherapy, the estimate of the absolute VTE risk may have been conservative. In addition, if the scores were to have been truly predictive, the performance of the scores may have been underestimated. However, the findings in the complete study group were similar (and even slightly better) to those in the group enrolled prior to chemotherapy. Sudden deaths were not routinely adjudicated for fatal PE, although PE was considered to be fatal if autopsy confirmed the diagnosis or when PE was objectively confirmed prior to death. Still, this approach may have resulted in a conservative estimate of the VTE incidence. As the proportion of such lethal events among all events is often small, relative risk estimates are unlikely to be affected. We realize that the evaluated scores were predominantly derived from observations of symptomatic events. Since treatment recommendations are similar for incidental and for symptomatic events, thromboprophylaxis is just as important to prevent incidental VTE. These events were, therefore, included in the primary outcome. The findings of the sensitivity analysis restricted to symptomatic events was comparable to those from the overall analysis, suggesting that the examined scores have similar performance in predicting incidental events. A different D-dimer assay was used in the present study

References 1. Di Nisio M, Porreca E, Candeloro M, De Tursi M, Russi I, Rutjes AW. Primary prophylaxis for venous thromboembolism in ambulatory cancer patients receiving

1500

than in the derivation study of the Vienna CATS score, which could have impacted on the discriminatory performance of the dichotomous D-dimer result. Nevertheless, D-dimer was one of the strongest predictors of VTE in the multivariable analysis, suggesting that the evaluated threshold of 1.44 mg/L may be a reasonable one, also for other assays. What are the clinical implications of the present findings and the future steps for using prediction tools for cancerassociated VTE? A recent Cochrane systematic review demonstrated that LMWH in prophylactic doses reduces the relative risk of symptomatic VTE by 46% when compared to no thromboprophylaxis.1 Assuming that these findings also apply to high-risk patients, the use of the Vienna CATS or PROTECHT scores to select patients for thromboprophylaxis could theoretically result in an absolute risk reduction of approximately 4-5% with a score of 20-25 needed in order to decide to start treatment. Subcutaneous injections for six months can be burdensome, perhaps even more for oncological patients with a limited life expectancy. The same Cochrane review also showed that LMWH thromboprophylaxis was associated with a non-significant 44% increase in the risk of major bleeding with a score of 125 needed before treatment is considered harmful; this should be carefully balanced against the potential benefits. Future studies should aim to refine current risk prediction tools or develop new models in order to further improve the risk-benefit ratio of thromboprophylaxis in ambulatory cancer patients. Risk models that take into account the differences in baseline risk and prognostic factors across various types of cancer would allow personalized risks to be calculated more precisely by only using predictors that are relevant for a specific cancer type. Whether thromboprophylaxis with direct oral anticoagulants is a safe, effective, and less burdensome alternative in ambulatory cancer patients is currently being investigated in several ongoing trials that use the Khorana score for risk stratification (e.g. clinicaltrials.gov identifiers: 02555878 and 02048865). Interestingly, one of the trials applies the Khorana score at a lower positivity threshold of 2 points, which classifies a greater proportion of patients in the high-risk group, but in parallel may decrease the positive predictive value of the score. The present findings do not support the use of any of the examined scores to select patients for thromboprophylaxis. The discriminatory performance of the dichotomized Vienna CATS and PROTECHT scores is somewhat encouraging, but confirmation in subsequent observational and intervention studies providing thromboprophylaxis to high-risk patients is needed before they can be used in clinical practice. Funding This work was supported by unrestricted grants from the participating hospitals.

chemotherapy. Cochrane Database Syst Rev. 2016;12(2):CD008500. 2. Horsted F, West J, Grainge MJ. Risk of venous thromboembolism in patients with cancer: a systematic review and meta-analysis. PLoS Med. 2012;9(7):e1001275. 3. Khorana AA, Francis CW, Culakova E,

Kuderer NM, Lyman GH. Thromboembolism is a leading cause of death in cancer patients receiving outpatient chemotherapy. J Thromb Haemost. 2007;5(3):632â&#x20AC;&#x201C;634. 4. Lyman GH, Bohlke K, Khorana AA, et al. Venous Thromboembolism Prophylaxis and

haematologica | 2017; 102(9)


Predicting cancer-associated venous thromboembolism

5.

6.

7.

8. 9.

10.

11.

12.

Treatment in Patients With Cancer: American Society of Clinical Oncology Clinical Practice Guideline Update 2014. J Clin Oncol. 2013;31(17):654-656. Mandala M, Falanga A, Roila F, Mandalà M, Falanga A, Roila F. Management of venous thromboembolism (VTE) in cancer patients: ESMO Clinical Practice Guidelines. Ann Oncol. 2011;22 Suppl 6:vi85-92. Farge D, Debourdeau P, Beckers M, et al. International clinical practice guidelines for the treatment and prophylaxis of venous thromboembolism in patients with cancer. J Thromb Haemost. 2013;11(1):56-70. Khorana AA, Kuderer NM, Culakova E, Lyman GH, Francis CW. Development and validation of a predictive model for chemotherapy-associated thrombosis. Blood. 2008;111(10):4902-4907. Ay C, Dunkler D, Marosi C, et al. Prediction of venous thromboembolism in cancer patients. Blood. 2010;116(24):5377-5382. Verso M, Agnelli G, Barni S, Gasparini G, LaBianca R. A modified Khorana risk assessment score for venous thromboembolism in cancer patients receiving chemotherapy: the Protecht score. Intern Emerg Med. 2012;7(3):291-292. Pelzer U, Sinn M, Stieler J, Riess H. [Primary pharmacological prevention of thromboembolic events in ambulatory patients with advanced pancreatic cancer treated with chemotherapy?]. Dtsch Med Wochenschr. 2013;138(41):2084-2088. Simanek R, Vormittag R, Ay C, et al. High platelet count associated with venous thromboembolism in cancer patients: results from the Vienna Cancer and Thrombosis Study (CATS). J Thromb Haemost. 2010;8(1):114-120. Mansfield AS, Tafur AJ, Wang CE, Kourelis T V., Wysokinska EM, Yang P. Predictors of

haematologica | 2017; 102(9)

13.

14.

15.

16.

17.

18.

19.

20.

active cancer thromboembolic outcomes: validation of the Khorana score among patients with lung cancer. J Thromb Haemost. 2016;14(9):1773-1778. Muñoz Martín AJ, García Alfonso P, Rupérez Blanco AB, Pérez Ramírez S, Blanco Codesido M, Martín Jiménez M. Incidence of venous thromboembolism (VTE) in ambulatory pancreatic cancer patients receiving chemotherapy and analysis of Khorana’s predictive model. Clin Transl Oncol. 2014;16(10):927-930. von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007; 370(9596):1453-1457. Martín AJM, Ziyatdinov A, Rubio VC, et al. PO-04 - A new genetic risk score for predicting venous thromboembolic events in cancer patients receiving chemotherapy. Thromb Res. 2016;140 Suppl:S177-178. Putter H, Fiocco M, Geskus RB. Tutorial in biostatistics: competing risks and multistate models. Stat Med. 2007;26(11):23892430. Choudhury JB. Non-parametric confidence interval estimation for competing risks analysis: application to contraceptive data. Stat Med. 2002;21(8):1129-1144. Wolbers M, Blanche P, Koller MT, Witteman JCM, Gerds TA. Concordance for prognostic models with competing risks. Biostatistics. 2014;15(3):526-539. Fine JP, Gray RJ. A Proportional Hazards Model for the Subdistribution of a Competing Risk. J Am Stat Assoc. 1999; 94(446):496. Donders ART, van der Heijden GJMG, Stijnen T, Moons KGM. Review: a gentle introduction to imputation of missing val-

21. 22.

23.

24.

25.

26.

27.

28.

ues. J Clin Epidemiol. 2006;59(10):10871091. Rubin DB. Inference and missing data. Biometrika. 1976;63(3):581-592. Posch F, Riedl J, Reitter E-M, et al. Hypercoagulabilty, venous thromboembolism, and death in patients with cancer. A Multi-State Model. Thromb Haemost. 2016;115(4):1-10. van Es N, Franke VF, Middeldorp S, Wilmink JW, Büller HR. The Khorana score for the prediction of venous thromboembolism in patients with pancreatic cancer. Thromb Res. 2017;150:30-32. Srikanthan A, Tran B, Beausoleil M, et al. Large retroperitoneal lymphadenopathy as a predictor of venous thromboembolism in patients with disseminated germ cell tumors treated with chemotherapy. J Clin Oncol. 2015;33(6):582-587. Mandala M, Clerici M, Corradino I, et al. Incidence, risk factors and clinical implications of venous thromboembolism in cancer patients treated within the context of phase I studies: the “SENDO experience”. Ann Oncol. 2012;23(6):1416-1421. Agnelli G, George DJ, Kakkar AK, et al. Semuloparin for thromboprophylaxis in patients receiving chemotherapy for cancer. N Engl J Med. 2012;366(7):601-609. Agnelli G, Gussoni G, Bianchini C, et al. Nadroparin for the prevention of thromboembolic events in ambulatory patients with metastatic or locally advanced solid cancer receiving chemotherapy: a randomised, placebo-controlled, double-blind study. Lancet Oncol. 2009;10(10):943-949. Unverzagt S, Prondzinsky R, Peinemann F. Single-center trials tend to provide larger treatment effects than multicenter trials: A systematic review. J Clin Epidemiol. 2013; 66(11):1271-1280.

1501


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Myelodysplastic syndrome

Ferrata Storti Foundation

Haematologica 2017 Volume 102(9):1502-1510

Molecular analysis of myelodysplastic syndrome with isolated deletion of the long arm of chromosome 5 reveals a specific spectrum of molecular mutations with prognostic impact: a study on 123 patients and 27 genes Manja Meggendorfer, Claudia Haferlach, Wolfgang Kern and Torsten Haferlach MLL Munich Leukemia Laboratory, Germany

ABSTRACT

T

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1502

he only cytogenetic aberration defining a myelodysplastic syndrome subtype is the deletion of the long arm of chromosome 5, which, along with morphological features, leads to the diagnosis of myelodysplastic syndrome with isolated deletion of the long arm of chromosome 5. These patients show a good prognosis and respond to treatment such as lenalidomide, but some cases progress to acute myeloid leukemia; however, the molecular mutation pattern is rarely characterized. Therefore, we investigated a large cohort of 123 myelodysplastic syndrome patients with isolated deletion of the long arm of chromosome 5, diagnosed following the World Health Organization classifications 2008 and 2016, by sequencing 27 genes. A great proportion of patients showed no or only one mutation. Only seven genes showed mutation frequencies >5% (SF3B1, DNMT3A, TP53, TET2, CSNK1A1, ASXL1, JAK2). However, the pattern of recurrently mutated genes was comparable to other myelodysplastic syndrome subtypes by comparison to a reference cohort, except that of TP53 which was significantly more often mutated in myelodysplastic syndrome with isolated deletion of the long arm of chromosome 5. As expected, SF3B1 was frequently mutated and correlated with ring sideroblasts, while JAK2 mutations correlated with elevated platelet counts. Surprisingly, SF3B1 mutations led to significantly worse prognosis within cases with isolated deletion of the long arm of chromosome 5, but showed a comparable outcome to other myelodysplastic syndrome subtypes with SF3B1 mutation. However, addressing genetic stability in follow-up cases might suggest different genetic mechanisms for progression to secondary acute myeloid leukemia compared to overall myelodysplastic syndrome patients.

Š2017 Ferrata Storti Foundation

Introduction

Correspondence: manja.meggendorfer@mll.com

Received: February 8, 2017. Accepted: June 16, 2017. Pre-published: June 22, 2017. doi:10.3324/haematol.2017.166173

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.

1502

Myelodysplastic syndromes (MDS) are a heterogeneous group of clonal bone marrow neoplasms characterized by ineffective hematopoiesis, morphologic dysplasia and peripheral cytopenias. While the degree of dysplasia and blast percentage are disease classifying, the specific types of cytopenias have a minor impact on MDS classification. The diagnosis of MDS is driven by the number of dysplastic cell lineages as well as blast counts <1% in the peripheral blood (PB) or <5% in the bone marrow (BM), or <19% in PB or BM for subtypes with excess blasts and at least one cytopenia.1 The risk stratification for MDS patients is categorized according to the Revised International Prognostic Scoring System (IPSS-R), whereby, along with morphological features, cytogenetics also plays a crucial role. Some cytogenetic aberrations can also define an MDS even in the absence of morphologic dysplasia. haematologica | 2017; 102(9)


Mutation pattern of MDS with isolated del(5q)

However, also according to the new World Health Organization (WHO) 2016 classification, the only cytogenetic aberration defining a MDS subtype is the deletion of the long arm of chromosome 5 (del (5q)), giving the diagnosis of MDS with isolated del(5q). Overall, less than 50% of MDS patients show an aberrant karyotype, of which del(5q) is the most common aberration presenting in 1020% of MDS patients.2,3 In about 55% of patients with del(5q) this aberration appears as a sole abnormality, in 17% it presents with one additional aberration, while in another 28% del(5q) appears with two or more additional cytogenetic lesions, leading by definition to a complex karyotype.4 This is shifted towards sole del(5q) (82%) when addressing only low- and intermediate-1-risk MDS according to the IPSS.5 However, the WHO classification emphasizes that the main impact of cytogenetics is the prognostic rather than the classifying information. It was demonstrated that MDS with isolated del(5q) shows a good prognosis, while MDS with del(5q) within a complex karyotype shows a poor prognosis.2,4 In recent years the prognostic impact of additional aberrations was addressed, showing that the presence of only one aberration in addition to del(5) (excluding the abnormality -7 or del(7q)) had no adverse effect on prognosis.6 Therefore, in the new WHO classification 2016 a diagnosis of MDS with isolated del(5q) allows the presence of one additional aberration excluding aberrations affecting chromosome 7.1 In addition to the good prognosis of these patients, a sensitivity to specific treatments, such as lenalidomide, was demonstrated.7,8 Morphologically, MDS with isolated del(5q) is defined by blast counts <1% in the PB and <5% in the BM, severe macrocytic anemia and frequent thrombocytosis. Patients who have MDS with isolated del(5q) show a lower risk for progression to acute myeloid leukemia (AML) than those with other MDS. However, about 10% of these patients evolve to secondary (s)-AML.4,9,10 The underlying pathobiological mechanisms are still being debated, while recent studies indicate that TP53 mutation as well as karyotype evolution predict disease progression.11,12 Furthermore, the separation of MDS from reactive causes remains a challenge. In recent years large data sets became available showing that a limited number of genes were mutated in patients with MDS. Eighty to ninety percent of patients show at least one mutation in one of the >100 addressed genes, supporting the clonal hematopoiesis of the disease and with that the diagnosis.13,14 Moreover, it was demonstrated that the increasing number of gene mutations correlates with the disease outcome in MDS patients; the addition of this data improves the existing risk stratifications.13,14 Mutations in TP53 are generally associated with adverse outcome and an aggressive disease course in MDS. Furthermore, in MDS with isolated del(5q), a TP53 mutation seems to predict a poorer response to lenalidomide and a higher risk of transformation to AML.11 Thus, the respective mutation status should be addressed not only at the time of diagnosis, but also before treatment decisions are made, as recommended in the new WHO 2016 classification.1 Therefore, the aim of the study herein was to determine the frequency of mutations in a large cohort of 123 patients with MDS and isolated del(5q) with a 27 gene panel, and to combine clinical data and prognostic information. haematologica | 2017; 102(9)

Methods Patients cohort We investigated 123 patients (35 male, 88 female, ratio: 1:2.5) diagnosed as having MDS with isolated del(5q), strictly classified according to the WHO classification of 2008 and including the added guidelines of 2016 with respect to cytomorphology and cytogenetics (blast counts below 5% in the BM and below 1% in the PB, and isolated 5q deletion or one additional aberration that does not affect chromosome 7).1 All patientsâ&#x20AC;&#x2122; samples, taken between 2005 and 2013, were sent from different hematological centers to the MLL Munich leukemia laboratory for diagnostic purposes. The median age of the patients was 75 years (range: 3593 years). One hundred and nineteen patients showed an isolated del(5q) while four cases appeared with only one additional cytogenetic aberration (two patients with del(13q), one with del(9q) and one with a trisomy 8). Follow-up data was available in 111/123 cases with a median follow up of 41 months. One hundred and twelve cases from the cohort presented herein were also included in the study by Heuser et al.15 The mutation patterns and prognostic impact of MDS with del(5q) were compared to a MDS cohort of 944 patients representing all MDS subtypes.13 Of the 123 patients diagnosed with MDS with isolated del(5q), 39 were already included in this recent study, resulting in a reference cohort of 905 patients after excluding these 39 cases.13 This reference cohort contained the following MDS subtypes: 41 refractory anemia (RA), 81 refractory anemia with ringed sideroblasts (RARS), 27 RARS with thrombocytosis (RARS-T), 194 refractory cytopenia with multilineage dysplasia (RCMD), 183 refractory cytopenia with multilineage dysplasia and ringed sideroblasts (RCMD-RS), 191 refractory anemia with excess blasts-1 (RAEB1), and 188 refractory anemia with excess blasts-2 (RAEB-2), diagnosed according to the WHO 2008 classification guidelines.16 The median age was 73 years (range: 23-91 years), the male to female ratio was 1:7. Follow-up data in the reference cohort was available in 869 cases with a median follow up of 62 months. All patients gave their consent for genetic analyses and the use of laboratory results for research purposes. The study adhered to the tenets of the Declaration of Helsinki and was approved by the institutional review board of the laboratory.

Cytomorphology Cytomorphology of BM and/or PB samples was performed in all cases following May-GrĂźnwald Giemsa staining, cytochemistry with myeloperoxidase (MPO), non-specific esterase (NSE) and iron staining (Fe).17

Cytogenetics Chromosome preparations and banding analysis of BM and/or PB samples were performed for all 123 cases as previously described according to standard methods.18 For classification of abnormalities and karyotypes, the 2016 International System for Human Cytogenetic Nomenclature (ISCN) guidelines were used.19

Next-generation sequencing (NGS) All patients were analyzed via a myeloid gene panel containing ASXL1, BCOR, BRAF, CSNK1A1, CBL, DNMT3A, ETV6, EZH2, FLT3-TKD, GATA1, GATA2, IDH1, IDH2, JAK2, KIT, NRAS, KRAS, MPL, NPM1, PHF6, RUNX1, SF3B1, SRSF2, TET2, TP53, U2AF1 and WT1. The library of 26 genes was generated with the ThunderStorm (RainDance Technologies, Billerica, MA, USA), and CSNK1A1 with the Access Array System (Fluidigm, San Francisco, CA, USA). Both libraries were sequenced and demultiplexed on a MiSeq instrument (Illumina, San Diego, CA, USA) as described previously.20 The FASTQ files were further processed 1503


M. Meggendorfer et al.

A

B

Figure 1. Mutation frequencies and distributions. (A) The frequencies of the seven most often mutated genes are given, as well as their appearance as a sole abnormality. (B) Illustration of the distribution of the number of mutations per patient.

using the Sequence Pilot software version 4.1.1 Build 510 (JSI Medical Systems, Ettenheim, Germany) for alignment and variant calling. Analysis parameters were set according to the manufacturer’s default recommendation. The validity of the somatic mutations was checked against the publicly accessible Catalogue Of Somatic Mutations In Cancer (COSMIC) v69 database, and functional interpretation was performed using SIFT 1.03, PolyPhen 2.0 and MutationTaster 1.0 algorithms.21 Additionally, TP53 variants were verified using the International Agency for Research on Cancer (IARC) repository.22 Single-nucleotide polymorphisms (SNP) were annotated according to the National Center for Biotechnology Information Single Nucleotide Polymorphism Database (NCBI dbSNP). The detection limit for small nuclear variants was 3% variant allele frequency (VAF). Variants (n=9) not yet described in any public database were excluded from statistical analyses.

criteria leading to a diagnosis of MDS with isolated del(5q).13 Of 944 MDS patients, 84 (9%) carried a del(5q) either as a sole abnormality or co-occurring with other lesions. Of 84 cases, 53 (63%) showed a sole del(5q), while only five cases (6%) appeared with one additional aberration and 26 (31%) showed a complex karyotype. Of the 53 patients with isolated del(5q), 44 fulfilled the diagnostic criteria of <1% blasts in the PB and <5% in the BM. Furthermore, 4/5 cases with one additional lesion were also diagnosed as MDS with isolated del(5q), while the fifth case harbored a -7 as an additional aberration. Therefore, 48 cases were diagnosed as MDS with isolated del(5q), whereof only 4/48 cases (8%) were grouped to these patients based on the new WHO classification and its additional criteria, while the main group was diagnosed using the criteria of the WHO 2008 classification, without allowing for additional lesions.1

Statistical analysis Dichotomous variables were compared between different groups using the χ2-test and continuous variables by the Student’s t-test; results were considered significant at P<0.05. Adjustment for multiple testing was not done. Statistical analyses were performed using SPSS version 19.0 (IBM Corporation, Armonk, NY, USA); the reported P-values are two-sided. Survival curves were calculated for overall survival (OS) according to Kaplan-Meier and compared using the two-sided log-rank test. OS was considered as being the time from diagnosis to death or last follow up.

Results Incidence of MDS with isolated del(5q) following the new WHO classification Following the recently revised 2016 version of the WHO classification for myeloid neoplasms, one additional chromosomal aberration (other than aberration of chromosome 7), in addition to the deletion of the long arm of chromosome 5, allows for the diagnosis of MDS with isolated del(5q).1 Therefore, we analyzed a well characterized MDS cohort for incidences of del(5q) and other diagnostic 1504

Molecular genetic characterization of MDS with isolated del(5q) For a comprehensive molecular characterization we assembled a cohort of 123 patients diagnosed with MDS with isolated del(5q) and investigated 27 genes by mutation analyses. The most frequently mutated gene was SF3B1 (23/123; 19%), followed by DNMT3A (22/123; 18%), TP53 (22/123; 18%), TET2 (14/121; 12%), CSNK1A1 (12/123; 10%), ASXL1 and JAK2 (both 7/123; 6%) (Figure 1A). All other analyzed genes showed mutation frequencies below 5%, with BRAF, ETV6, FLT3-TKD, GATA1, GATA2, IDH1, IDH2, NPM1, NRAS, PHF6 and WT1 showing no mutations at all. Nearly all mutations were heterozygous with a median VAF for six of the most frequently mutated genes of 15-30% (range: 2-50% VAF). Solely CSNK1A1 showed VAF of 3-78%, mimicking a homozygous mutation status, caused by the location of CSNK1A1 on chromosome 5q in the commonly deleted region. Addressing the number of mutations per patient revealed a large number of patients with no mutation (38/123; 31%) or one mutation (53/123; 43%) in any of haematologica | 2017; 102(9)


Mutation pattern of MDS with isolated del(5q)

Figure 2. Molecular and cytogenetic characterization of patients with MDS with isolated del(5q). Illustration of all 123 samples, each column represents one patient. All 27 analyzed genes as well as the occurrence of del(5q) as a sole aberration or with one additional cytogenetic aberration are given for each patient. Light gray: wild-type; red: mutated; orange: variant; light blue: del(5q) sole; dark blue: del(5q) and one additional lesion.

the 27 analyzed genes. Two genes were mutated in 23% (28/123) of patients, while three and four mutations were detected in only 2% (3/123) and 1% (1/123) of patients, respectively, (Figure 1B). Looking at the co-occurrence of gene mutations resulted in a single association of CSNK1A1 and SF3B1 mutations. Only 16% of CSNK1A1 wild-type (wt) cases showed a SF3B1 mutation (18/111), while 42% of CSNK1A1-mutated patients displayed a cooccurring SF3B1 mutation (5/12; P=0.047). Albeit the most frequent gene mutations rarely overlapped and occurred frequently as sole mutations, they were not completely mutually exclusive (Figure 1A and Figure 2).

Comparison of mutation frequencies in MDS with isolated del(5q) and other MDS subtypes In order to analyze the differences in the mutation patterns of MDS with isolated del(5q) and all other MDS subtypes, we investigated the most frequently mutated genes in comparison to a reference cohort, represented by the previously published MDS cohort after excluding the cases diagnosed as MDS with isolated del(5q).13 Comparing the mutation frequencies of the seven most frequently mutated genes in MDS with isolated del(5q) with all other MDS subtypes revealed that the respective six mutated genes, SF3B1, DNMT3A, TP53, TET2, ASXL1 and JAK2, were also ranked within the 15 most frequently mutated genes in all MDS subtypes (Figure 3; CSNK1A1 was not addressed due to missing data in the MDS reference cohort). Of note, TP53 mutations were found significantly more often in MDS with isolated del(5q) in comparison to all other MDS subtypes, with a mutation frequency of 18% (22/123) compared to 6% in the MDS reference cohort (52/905; P<0.001).

Clinical and morphological correlations Upon dividing the patients into groups defined by a BM blast count of <2% and 2-5%, as described in the IPSS-R, haematologica | 2017; 102(9)

we could not detect any correlation with the number of mutations per patient. However, patients who had no mutation (n=38) were younger compared to patients with at least one mutation (n=85; 70 vs. 76 years, P=0.018). There was no difference between these two patient groups as regards to sex, white blood cell count, hemoglobin level or platelet count. Taking single gene mutations into account revealed that TP53 mutations correlated with older age (78 vs. 73 years, P=0.038). All 22 patients with a TP53 mutation were â&#x2030;Ľ67 years of age, with only two patients under 70; 11 patients were between 70 and 79 years of age, eight patients were between 80 and 89 years of age, and one patient in the group was â&#x2030;Ľ90 years of age. Addressing the correlation between percentages of ring sideroblasts (RS) and SF3B1 mutations also showed that in MDS with isolated del(5q) these two parameters significantly correlated with a mean of 17% RS (range: 0-80%) in SF3B1-mutated patients and only 1% RS in SF3B1 wt patients (range: 0-12%, P=0.004). Furthermore, JAK2 mutations correlated with a platelet count of >450,000/ml (80% in JAK2 mutated vs. 12% in JAK2 wt, P=0.002).

Prognostic impact of gene mutations Looking at the prognostic relevance of gene mutations showed, surprisingly, that SF3B1-mutated patients (n=20 with follow-up data) had a significantly inferior outcome than that of SF3B1 wt patients (n=91; median OS: 50 months vs. not reached, P=0.010; Figure 4A). Furthermore, we analyzed the OS compared to the MDS reference cohort representing all other than isolated del(5q) MDS subtypes (n=869 with follow-up data). Overall the median OS was significantly favorable for patients with MDS with isolated del(5q) (62 months vs. not reached, P<0.001; Figure 4B). To better characterize the unexpected worse impact of SF3B1 mutations we analyzed both cohorts with respect to their SF3B1 mutation status as well as 1505


M. Meggendorfer et al.

cytogenetics in terms of normal or aberrant karyotypes (Figure 4C). SF3B1 mutations within a normal karyotype showed favorable outcomes in MDS patients (median OS: 93 months), however, this effect was reduced by an additional aberrant karyotype (median OS: 64 months). Therefore, SF3B1-mutated patients possessing aberrant karyotypes showed comparable outcomes (median OS: 50 vs. 64 months), irrespective of diagnoses of MDS with isolated del(5q) or any other subtype of MDS, while SF3B1 wt patients with isolated del(5q) had a favorable outcome compared to all other SF3B1 wt MDS patients (median OS: not reached vs. 49 months). However, we did not find a prognostic impact of TP53 mutations in our cohort (Online Supplementary Figure S1; median OS was not reached in TP53 wt (n=92) and TP53 mutated (n=19) patients; P=0.094).

Molecular genetics during follow up and progression to AML Follow-up samples were available in 13 patients with a median time between two investigations of 22 months (range: 5-47 months, cases with less than four months between two investigations were excluded). In a further 6/123 (5%) cases we could follow a progression to AML with a median transformation time to s-AML of 24 months (range: 10-53 months). In these 19 cases with a follow-up sample mutation analyses were performed, in 18/19 cases cytogenetics were also available. Overall, in 5/13 (38%) patients showing no transformation to AML no clonal evolution was detectable, neither in karyotype nor in the mutation pattern, leaving 8/13 (62%) patients with clonal evolution. In 4/13 cases (31%) a change in cytogenetics occurred and in 7/13 patients (54%) additional gene mutations appeared; in 3/13 patients clonal evolution occurred in both instances. Studying patients progressing to AML after MDS revealed a similar pattern regarding stability in genetics with 3/5 cases (60%) showing no genetic evolution, 1/5 (20%) a clonal evolution in cytogenetics and 2/6 (33%) exhibiting a change in the mutation pattern. Of note, none of the patients gained a mutation in TP53 or FLT3 during progression to AML. Addressing the gains of genetic lesions in the disease course showed no specific pattern. Cytogenetic gains included â&#x20AC;&#x201C;Y, del(11q), del(13q) and inversion 3 (inv(3)), and gained mutations affected ASXL1, BCOR, CSNK1A1, DNMT3A, IDH1, JAK2, RUNX1 and TET2. However, recurrent gains were limited to JAK2 (n=2) and RUNX1 (n=3), with two cases showing a gain of RUNX1 mutation progressing to AML and one progressing to MDS RAEB-1. Furthermore, the VAF of all gained mutations were <10% (median 4%, range: 3-7%), except for those of RUNX1 during transformation to AML, which showed VAF of 14% and 25%, respectively. An illustration of the different types of clonal evolution is given in Figure 5.

Discussion In the study herein, we investigated the molecular mutation pattern of a large cohort of 123 MDS patients with isolated del(5q) classified according to the WHO classification of 2008 and including the added guidelines of 2016. A large proportion of patients showed no or only one mutation in 27 analyzed genes, with only seven genes 1506

Figure 3. Comparison of mutation frequencies between MDS with isolated del(5q) and all MDS subtypes. Mutation frequencies for MDS were taken from the previously published reference cohort.13 The six most frequently mutated genes in MDS with isolated del(5q) are compared to the 14 most frequently mutated genes in the MDS reference cohort. CSNK1A1 and STAG2 (typed in gray) were not analyzed in the MDS reference cohort or MDS del(5q) cohort, respectively. MDS: myelodysplastic syndromes; MDS del(5q): MDS with isolated del(5q).

showing mutation frequencies >5% (SF3B1, DNMT3A, TP53, TET2, CSNK1A1, ASXL1, JAK2). Albeit in other studies more genes were investigated for mutation analyses, in overall MDS the median number of mutations is higher with three mutations per patient13 or nearly half of the patients showing two or three mutations14 in contrast to MDS with isolated del(5q). This indicates the specific and narrow spectrum of genetic lesions in this specific MDS subtype. However, the most frequently mutated genes were comparable to all other MDS subtypes investigated in a large MDS reference cohort.13 The 15 most frequently mutated genes have also been identified by another large MDS study and by meta-analysis showing consistent data for the MDS mutation landscape.14,23 Furthermore, SF3B1 was frequently mutated and correlated with the presence of RS, which is in line with previously published data on other MDS entities.24-26 In contrast, TP53 was mutated significantly more often in MDS with isolated del(5q). In a number of studies mutations of TP53 were shown to occur in high-risk or therapy-related MDS, MDS-derived leukemia or within a complex karyotype.27-29 Incidences of TP53 mutations in MDS with isolated del(5q) were found to be 5%, much lower than in other MDS subtypes.12 However, in another previous study the incidence of TP53 mutations, investigated by back tracking TP53-mutated samples in low-risk MDS with del(5q), was 18%, suggesting a previous underestimation of TP53 subclonal mutations.11 It was shown that the TP53 mutation was already present in the early haematologica | 2017; 102(9)


Mutation pattern of MDS with isolated del(5q)

B

A

P=0.010

P<0.001

C

Figure 4. Overall survival (OS) analyses. Case numbers, median OS and P-values are given. (A) Prognostic impact of SF3B1 mutations in MDS with isolated del(5q). (B) OS of MDS with isolated del(5q) in comparison to the MDS reference cohort. (C) OS of SF3B1 mutation with additional cytogenetic aberrations in MDS with isolated del(5q) in comparison to the MDS reference cohort. MDS: myelodysplastic syndromes; MDS del(5q): MDS with isolated del(5q); NK: normal karyotype; AK: aberrant karyotype; wt: wild-type; n.r.: not reached; mo: months; mut: mutated.

stages of the disease and increased during the course of the disease.11 This is in line with the present data showing a TP53 mutation incidence of 18%, with 7/22 cases having VAF of <10%. Surprisingly, SF3B1 mutations led to significantly worse OS, but ultimately showed a comparable outcome, still favorable, to all other MDS subtypes with a SF3B1 mutation and an aberrant karyotype. Since a high frequency of MDS patients show a normal karyotype, the need to analyze the impact of karyotype information on SF3B1 mutations in overall MDS in our reference cohort seemed the next logical step. This demonstrated that an accompanying altered karyotype also reduces the favorable impact of SF3B1 mutations. Therefore, the highly favorable impact of SF3B1 mutation was limited to sole SF3B1 mutations without any additional lesions, and might be an explanation as to why the favorable impact is reduced and the overall good prognosis is not driven by SF3B1 mutations alone in MDS with isolated del(5q). However, this finding is in contrast to previous studies which demonstrate that haematologica | 2017; 102(9)

the prognostic effect of SF3B1 mutations is independent of variables that could coexist, such as age, sex and cytogenetics.24,26 In other studies, however, SF3B1 mutations did not keep independent prognostic significance in multivariate analyses or showed no prognostic impact whatsoever, indicating that perhaps MDS represents a heterogeneous disease and as such should be analyzed in respective subgroups.30-33 In MDS it was evidenced that transformation to s-AML is frequently accompanied by mutations in the FLT3 and RAS genes and karyotype evolution.34-38 In addition, RUNX1, GATA2 and CEBPA were identified by progression to s-AML, abrogating normal differentiation.39 Furthermore, some genes have been identified as being affected by mutations occurring as late events in MDS patients, thus giving rise to a potential progression to AML, such as mutations in ASXL1, RUNX1, SRSF2, IDH2 or NRAS.34,35,38 However, most of these genes were not identified to be affected in MDS with isolated del(5q), arguing for an alternative progression mechanism to that 1507


M. Meggendorfer et al.

found in all other MDS patients. Nevertheless, the recurrent gain of RUNX1 mutations in the disease course might indicate transformation to AML. In a comprehensive study of the mutation patterns in AML, it became obvious that some genes were specific for s-AML, covering the spliceosomal genes as well as chromatin modifiers.38,39 The high incidence of SF3B1 mutations in MDS with isolated del(5q) would therefore reflect the MDS origin of the disease rather than a high SF3B1 mutation frequency indicating disease progression. In previous studies, mutations in TP53 and gain of cytogenetic aberrations were speculated

to indicate disease progression and transformation to AML.11,12 However, the clonal evolution was comparable in MDS with isolated del(5q) to that described for overall MDS, with a karyotype evolution in 34% and mutation pattern in 53% of MDS cases transforming to AML.34 Moreover, the clonal evolution appeared in the same manner in cases without progression to AML. Additionally, TP53 mutations were never gained during the disease course, but the mutation burden increased (from 11% to 75%) in the one case which transformed to AML, while in follow-up cases without progression (n=2) the burden did

A

B

C

D

1508

Figure 5. The dynamic of variant allele frequency (VAF) and acquisition and loss of mutated genes as well as cytogenetic lesions at diagnosis and follow up. VAF and clone size are indicated by the height of the shapes. Cases evolving to s-AML are indicated by the red dashed time line, while follow up without progression to s-AML is indicated by a black dashed time line. Disease course showing (A) gain of gene mutations, (B) loss of a gene mutation, (C) cytogenetic evolution, and (D) no significant change in del(5q) clone size as well as gene mutation (VAF). In (D) both illustrations represent three patients. For the third patient in the first row no cytogenetics was available at sAML stage. s-AML: secondary acute myeloid leukemia; del(5q): deltion of the long arm of chromosome 5.

haematologica | 2017; 102(9)


Mutation pattern of MDS with isolated del(5q)

not change. Therefore, one might suppose that in patients with clonal evolution clinical progress will be encountered, while stable patients will also remain clinically stable. It was shown that clinical stability accompanies mutational stability, while developing new mutations resulted in AML progression in patients with MDS with isolated del(5q).40 The previously published worse impact prognosis of TP53 mutations was not observed in the present cohort. Our patients were unselected and the median follow up was 41 months. Thus, our cohort may include a larger proportion of patients earlier in their clinical course compared to cohorts enrolled in treatment studies. Therefore, the negative impact of TP53 mutations may become obvious in later stages of the disease course or even subsequent to the need for starting treatment, a phenomenon that is not represented by the collection of patients presented herein. Of note, 60% (61/101) of our patients were under observation alone, or only received red blood cells or erythropoietin. However, the presence, in advance, of TP53 mutations in the early stages of the disease would therefore be in line with our data, supported by the observed increase of clone size rather than the gain of TP53 mutation during the disease course.11 However, this needs to be further proven in larger cohorts as we had a small number of cases for follow up. In summary, MDS with isolated del(5q) shows a very

References 1. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization (WHO) classification of myeloid neoplasms and acute leukemia. Blood. 2016;127(20):2391-2405. 2. Haase D, Germing U, Schanz J, et al. New insights into the prognostic impact of the karyotype in MDS and correlation with subtypes: evidence from a core dataset of 2124 patients. Blood. 2007;110(13):43854395. 3. Bernasconi P, Klersy C, Boni M, et al. World Health Organization classification in combination with cytogenetic markers improves the prognostic stratification of patients with de novo primary myelodysplastic syndromes. Br J Haematol. 2007;137(3):193-205. 4. Mallo M, Cervera J, Schanz J, et al. Impact of adjunct cytogenetic abnormalities for prognostic stratification in patients with myelodysplastic syndrome and deletion 5q. Leukemia. 2011;25(1):110-120. 5. Germing U, Lauseker M, Hildebrandt B, et al. Survival, prognostic factors and rates of leukemic transformation in 381 untreated patients with MDS and del(5q): A multicenter study. Leukemia. 2012;26(6):12861292. 6. Schanz J, Tuchler H, Sole F, et al. New Comprehensive cytogenetic scoring system for primary myelodysplastic syndromes (MDS) and oligoblastic acute myeloid leukemia after MDS derived from an international database merge. J Clin Oncol. 2012;30(8):820-829. 7. Mallo M, del Rey M, Ibanez M, et al. Response to lenalidomide in myelodysplastic syndromes with del(5q): influence of cytogenetics and mutations. Br J Haematol.

haematologica | 2017; 102(9)

limited mutation spectrum of the recurrently mutated genes in comparison to all other MDS subtypes. However, the seven frequently mutated genes resemble the MDS mutation landscape, with the exception of TP53 mutations, which are more often found in MDS with isolated del(5q). Therefore, progression to s-AML is driven by the worse prognostic impact of TP53 mutation in combination with del(5q) and potentially the gain of RUNX1 mutations, rather than the otherwise identified accompanying mutations during MDS progression. The diagnostic and prognostic approach of MDS with isolated del(5q) should therefore include the mutation status of the seven most frequently mutated genes, including SF3B1, and TP53 and RUNX1 as potential progression markers, investigated with deep sequencing to catch all appearing mutations including those of a subclonal nature with low VAF. Acknowledgments We thank all clinicians for sending samples to our laboratory for diagnostic purposes and for providing clinical information and follow-up data. In addition, we would like to thank all our coworkers at the MLL Munich Leukemia Laboratory for jointly approaching many aspects in the field of leukemia diagnostics and research. We also thank Karolína Perglerová for analyzing FLT3-ITD in the follow-up samples at MLL2 s.r.o., Prague, Czech Republic.

2013;162(1):74-86. 8. List AF, Dewald GW, Bennett J, et al. Lenalidomide in the myelodysplastic syndrome with chromosome 5q deletion. N Engl J Med. 2006;355(14):1456-1465. 9. Giagounidis AA, Germing U, Haase S, et al. Clinical, morphological, cytogenetic, and prognostic features of patients with myelodysplastic syndromes and del(5q) including band q31. Leukemia. 2004;18(1):113-119. 10. Side LE, Curtiss NP, Teel K, et al. RAS, FLT3, and TP53 mutations in therapy-related myeloid malignancies with abnormalities of chromosomes 5 and 7. Genes Chromosomes Cancer. 2004;39(3):217-223. 11. Jadersten M, Saft L, Smith A, et al. TP53 mutations in low-risk myelodysplastic syndromes with del(5q) predict disease progression. J Clin Oncol. 2011;29(15):19711979. 12. Fernandez-Mercado M, Burns A, Pellagatti A, et al. Targeted re-sequencing analysis of 25 genes commonly mutated in myeloid disorders in del(5q) myelodysplastic syndromes. Haematologica. 2013;98(12):18561864. 13. Haferlach T, Nagata Y, Grossmann V, et al. Landscape of genetic lesions in 944 patients with myelodysplastic syndromes. Leukemia. 2014;28(2):241-247. 14. Papaemmanuil E, Gerstung M, Malcovati L, et al. Clinical and biological implications of driver mutations in myelodysplastic syndromes. Blood. 2013;122(22):3616-3627. 15. Heuser M, Meggendorfer M, Cruz MM, et al. Frequency and prognostic impact of casein kinase 1A1 mutations in MDS patients with deletion of chromosome 5q. Leukemia. 2015;29(9):1942-1945. 16. Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H et al. WHO classifica-

17. 18.

19.

20.

21.

22.

23.

tion of tumours of haematopoietic and lymphoid tissues, 4th ed. Lyon: International Agency for Research on Cancer (IARC), 2008. Löffler H, Rastetter J, Haferlach T. Atlas of clinical hematology, 6th ed. Heidelberg: Springer, 2010. Schoch C, Schnittger S, Bursch S, et al. Comparison of chromosome banding analysis, interphase- and hypermetaphaseFISH, qualitative and quantitative PCR for diagnosis and for follow-up in chronic myeloid leukemia: a study on 350 cases. Leukemia. 2002;16(1):53-59. McGowan-Jordan J, Simons A, Schmid M. ISCN 2016: An International System for Human Cytogenomic Nomenclature. Basel, New York: Karger, 2016. Delic S, Rose D, Kern W, et al. Application of an NGS-based 28-gene panel in myeloproliferative neoplasms reveals distinct mutation patterns in essential thrombocythaemia, primary myelofibrosis and polycythaemia vera. Br J Haematol. 2016;175(3):419-426. Schwarz JM, Rodelsperger C, Schuelke M, Seelow D. MutationTaster evaluates disease-causing potential of sequence alterations. Nat Methods. 2010;7(8):575-576. Petitjean A, Mathe E, Kato S, et al. Impact of mutant p53 functional properties on TP53 mutation patterns and tumor phenotype: lessons from recent developments in the IARC TP53 database. Hum Mutat. 2007;28(6):622-629. Rose D, Kohlmann A, Nagata Y, et al. A robust molecular pattern for myelodysplastic syndromes in two independent cohorts investigated by next-generation sequencing can be revealed by comparative bioinformatic analyses. Br J Haematol. 2014;167(2): 278-281.

1509


M. Meggendorfer et al. 24. Papaemmanuil E, Cazzola M, Boultwood J, et al. Somatic SF3B1 mutation in myelodysplasia with ring sideroblasts. N Engl J Med. 2011;365(15):1384-1395. 25. Yoshida K, Sanada M, Shiraishi Y, et al. Frequent pathway mutations of splicing machinery in myelodysplasia. Nature. 2011;478(7367):64-69. 26. Malcovati L, Papaemmanuil E, Bowen DT, et al. Clinical significance of SF3B1 mutations in myelodysplastic syndromes and myelodysplastic/myeloproliferative neoplasms. Blood. 2011;118(24):6239-6246. 27. Haferlach C, Dicker F, Herholz H, et al. Mutations of the TP53 gene in acute myeloid leukemia are strongly associated with a complex aberrant karyotype. Leukemia. 2008:1539-1541. 28. Wong TN, Ramsingh G, Young AL, et al. Role of TP53 mutations in the origin and evolution of therapy-related acute myeloid leukaemia. Nature. 2015;518(7540):552555. 29. Bejar R, Stevenson K, Abdel-Wahab O, et al. Clinical effect of point mutations in myelodysplastic syndromes. N Engl J Med. 2011;364(26):2496-2506.

1510

30. Patnaik MM, Lasho TL, Hodnefield JM, et al. SF3B1 mutations are prevalent in myelodysplastic syndromes with ring sideroblasts but do not hold independent prognostic value. Blood. 2012;119(2):569572. 31. Damm F, Thol F, Kosmider O, et al. SF3B1 mutations in myelodysplastic syndromes: clinical associations and prognostic implications. Leukemia. 2012;26(5):1137-1140. 32. Thol F, Kade S, Schlarmann C, et al. Frequency and prognostic impact of mutations in SRSF2, U2AF1, and ZRSR2 in patients with myelodysplastic syndromes. Blood. 2012;119(15):3578-3584. 33. Cazzola M, Rossi M, Malcovati L. Biologic and clinical significance of somatic mutations of SF3B1 in myeloid and lymphoid neoplasms. Blood. 2013;121(2):260-269. 34. Meggendorfer M, de AA, Nadarajah N, et al. Karyotype evolution and acquisition of FLT3 or RAS pathway alterations drive progression of myelodysplastic syndrome to acute myeloid leukemia. Haematologica. 2015;100(12):e487-e490. 35. Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic classification and prognosis

36.

37.

38.

39.

40.

in acute myeloid leukemia. N Engl J Med. 2016;374(23):2209-2221. Takahashi K, Jabbour E, Wang X, et al. Dynamic acquisition of FLT3 or RAS alterations drive a subset of patients with lower risk MDS to secondary AML. Leukemia. 2013;27(10):2081-2083. Murphy DM, Bejar R, Stevenson K, et al. NRAS mutations with low allele burden have independent prognostic significance for patients with lower risk myelodysplastic syndromes. Leukemia. 2013;27(10):2077-2081. Sperling AS, Gibson CJ, Ebert BL. The genetics of myelodysplastic syndrome: from clonal haematopoiesis to secondary leukaemia. Nat Rev Cancer. 2017;17(1):519. Lindsley RC, Mar BG, Mazzola E, et al. Acute myeloid leukemia ontogeny is defined by distinct somatic mutations. Blood. 2015;125(9):1367-1376. Woll PS, Kjallquist U, Chowdhury O, et al. Myelodysplastic syndromes are propagated by rare and distinct human cancer stem cells in vivo. Cancer Cell. 2014;25(6):794808.

haematologica | 2017; 102(9)


ARTICLE

Myeloproliferative Disorders

Characteristics and clinical significance of cytogenetic abnormalities in polycythemia vera Guilin Tang,1 Juliana E. Hidalgo Lopez,1 Sa A. Wang,1 Shimin Hu,1 Junsheng Ma,2 Sherry Pierce,3 Wenli Zuo,1 Adrian Alejandro Carballo-Zarate,1 C. Cameron Yin,1 Zhenya Tang,1 Shaoying Li,1 L. Jeffrey Medeiros,1 Srdan Verstovsek3 and Carlos E. Bueso-Ramos1

Department of Hematopathology, The University of Texas MD Anderson Cancer Center; Department of Biostatistics, The University of Texas MD Anderson Cancer Center, and 3Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

1 2

Haematologica 2017 Volume 102(9):1511-1518

ABSTRACT

U

p to 20% of patients with polycythemia vera have karyotypic abnormalities at the time of the initial diagnosis. However, the cytogenetic abnormalities in polycythemia vera have not been well characterized and their prognostic impact is largely unknown. In this study, we aimed to address these issues using a large cohort of polycythemia vera patients with cytogenetic information available. The study included 422 patients, 271 in polycythemic phase, 112 with postpolycythemic myelofibrosis, 11 in accelerated phase, and 28 in blast phase. Abnormal karyotypes were detected in 139 (33%) patients, ranging from 20% in those in the polycythemic phase to 90% among patients in accelerated/blast phase. Different phases harbored different abnormalities: isolated del(20q), +8 and +9 were the most common abnormalities in the polycythemic phase; del(20q) and +1q were the most common abnormalities in post-polycythemic myelofibrosis; and complex karyotypes were the most common karyotypes in accelerated and blast phases. Patients with an abnormal karyotype showed a higher frequency of disease progression, a shorter transformation-free survival and an inferior overall survival compared with patients with a normal karyotype in the same disease phase. Cytogenetics could be effectively stratified into three risk groups, low- (normal karyotype, sole +8, +9 and other single abnormality), intermediate- (sole del20q, +1q and other two abnormalities), and high-risk (complex karyotype) groups. We conclude that cytogenetic changes in polycythemia vera vary in different phases of disease, and carry different prognostic impacts. Introduction Polycythemia vera (PV) is a myeloproliferative neoplasm characterized by increased red blood cell production, a somatic gain-of-function mutation of JAK2, and panmyelosis in bone marrow (BM).1,2 The natural course of PV usually includes three phases: the pre-polycythemic phase, polycythemic phase (PP), and post-polycythemic myelofibrosis (post-PV MF). The disease in a small subset of patients may transform into an accelerated phase (AP), with 10-19% blasts in the peripheral blood and/or BM, or a blast phase (BP) with â&#x2030;Ľ20% blasts in peripheral blood/BM. Patients with PV generally have relatively long survival (median, 14-19 years). Potentially fatal complications include thrombosis, progression into myelofibrosis (post-PV MF) or transformation to BP.3 The median survival for patients with postPV MF is 5-6 years4 and patients with blastic transformation often have a dismal prognosis with a median survival of <6 months.5 The frequency of post-PV MF is 4.9-6% at 10 years and 6-14% at 15 years;3,6 and the risk of BP is 2.3-14.4% at 10 years and 5.5-18.7% at 15 years.3,7,8 Advanced age, leukocytosis, BM reticulin fibrosis, and splenomegaly have been reported to be risk factors for post-PV MF and BP;7,9-12 while leukocytosis, advanced age, and history of thrombosis have been haematologica | 2017; 102(9)

Correspondence: gtang@mdanderson.org

Received: January 31, 2017. Accepted: May 3, 2017. Pre-published: May 4, 2017. doi:10.3324/haematol.2017.165795 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1511 Š2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1511


G. Tang et al.

found to be independent risk factors for overall survival (OS).3,7,11 Cytogenetic abnormalities can be detected in 14-20% of patients at the time of the initial diagnosis of PV,13-16 with del(20q), +8, +9 and +1q being the most commonly reported.3,17,18 The low frequency of abnormal karyotypes has made prognostication of PV patients using cytogenetic data challenging and some studies have not shown a prognostic difference between patients with a normal or abnormal karyotype.13 Recently other studies,7,14,19 including one by the International Working Group for Myeloproliferative Neoplasms Research and Treatment (IWG-MRT),7 have found that patients with an abnormal karyotype have a higher risk of disease progression and an inferior outcome. However, the prognostic impact of individual cytogenetic abnormalities was not further classified, and the three most common abnormalities, +8, +9, and del(20q) have not been shown conclusively to have prognostic value.13,20 Here we reviewed 422 patients with PV for whom we had detailed clinicopathological and cytogenetic information. We examined the characteristics of the abnormal karyotypes during different stages of PV; the correlation of acquisition of cytogenetic abnormalities (ACA) and disease progression; and the prognostic impact of different specific cytogenetic abnormalities during different stages of PV.

Methods Patients We searched the archives of The University of Texas MD Anderson Cancer Center (MDACC) for patients with PV who were diagnosed and/or managed at MDACC from January 2005 through June 2016. For patients whose initial diagnosis of PV was established at other hospitals, the pathological material was reviewed in our department to confirm the diagnosis. The clinical presentation, laboratory data, and pathological findings were collected at the time of diagnosis, and during the follow-up. Blasts were counted on peripheral blood smears based on a 200-cell differential count and on BM smears based on 500 cells. The degree of BM myelofibrosis was based on the European Consensus on grading of bone marrow fibrosis.21 The diagnoses of PV and postPV MF were based on the World Health Organization criteria;1 AP was defined as ≥10% blasts and BP as ≥20% blasts in peripheral blood or BM or both. Disease progression (or transformation) was defined as disease that progressed from PP to post-PV MF, AP, or BP; or from post-PV MF to AP or BP. The study was approved by the institutional review board at MDACC.

Conventional cytogenetic analyses Conventional chromosomal analyses were performed on Gbanded metaphase cells prepared from unstimulated 24-h and 48h BM aspirate cultures using standard techniques. The median number of metaphases analyzed was 20 (range, 12 to 30). The karyotypes were documented according to the International System for Human Cytogenetic Nomenclature (ISCN 2016).22 In accordance with standard practice, if cytogenetic testing was performed within 4 months of initial diagnosis, it was considered to be “at initial diagnosis”; all other tests were considered as “beyond initial diagnosis”. A complex karyotype was defined as three or more chromosomal abnormalities. Specific cytogenetic abnormalities identified in four or more patients were grouped separately and the rest were grouped as “other single” or “other double” 1512

abnormalities. ACA was defined as the acquisition of an abnormal clone(s) from a previously normal karyotype, or the acquisition of additional chromosomal abnormalities or abnormal clone(s) from a previously abnormal karyotype.

Statistical analyses An unpaired t-test was used for numerical comparisons between groups. Chi-square and Fisher exact tests were applied for categorical variables. The date of diagnosis was calculated from the date that a BM was performed to establish the diagnosis. OS was estimated by the Kaplan–Meier method from the date of diagnosis until death from any cause (censored at last follow-up for patients who were alive). Transformation-free survival was estimated by the Kaplan-Meier method from the date of diagnosis to the date of disease progression to a higher stage or until death or the last follow-up. P values ≤0.05 were considered to be statistically significant.

Results Clinical and pathological findings A total of 477 patients were diagnosed and/or treated at our institute during the study period, 55 patients were excluded from this study because of lack of cytogenetic information at the time of diagnosis when the first BM evaluation was performed. Of the 422 patients included in the study, 114 patients had BM evaluation and cytogenetic analysis at the initial diagnosis, and the other 308 patients had BM evaluation at a median interval of 58 months after the initial diagnosis. Patients were diagnosed at a median age of 54 years (range, 11-84 years) and the male to female ratio was 227/195 (1.2:1) (Table 1). Prior to the first BM evaluation, 119 patients had not received any treatment, 76 had only been treated with phlebotomy, 10 had received aspirin only, 89 had received hydroxyurea only, 75 patients had been managed with two or three treatments including phlebotomy, hydroxyurea and/or aspirin and 53 patients were also treated with anagrelide (n=39), interferon (n=10), or imatinib (n=4). At the time of diagnosis, 271 (64%) patients were in PP, 112 (26.5%) in post-PV MF, 11 (3%) in AP, and 28 (6.6%) in BP. Since patients in AP shared very similar clinical features and disease course as patients in BP (data not shown), we combined these two groups of patients into one group (AP/BP) in this study. Patients with a normal karyotype and those with an abnormal karyotype had a comparable age, a similar gender distribution and leukocyte counts at all PV stages. However, at the stage of PP, patients with an abnormal karyotype had a significantly lower hemoglobin level and platelet count, a higher frequency of splenomegaly, and a higher grade of BM myelofibrosis. In the stage of post-PV MF, patients with an abnormal karyotype had lower platelet counts and in the AP/BP, patients with normal and abnormal karyotypes showed similar clinical and pathological features (Table 1).

Cytogenetic features The cytogenetic data are summarized in Table 2. The most common chromosomal abnormalities included del(20q) (n=31), +9 (n=10), and +8 (n=8) as a sole abnormality; +1q (n=15) as a component in double abnormalities, eight resulting from aberrations of +1, der(1;7)(q10;p10), and seven from other abnormalities. The most common chromosomal abnormalities detected haematologica | 2017; 102(9)


Cytogenetics in PV

Table 1. Demographic, clinical and pathological features of the study patients.

Stages Karyotype Gender (male/female) Age* (years) Splenomegaly (no/yes) Hemoglobin* (g/dL) Leukocytes* (x109/L) Platelets* (x109/L) Myelofibrosis • MF-0 • MF-1 • MF-2/3 Overall survival* (months)

Polycythemic phase (n=271) Normal Abnormal (80%) karyotype (20%)

P

Post-PV myelofibrosis (n=112) Normal Abnormal P (55%) karyotype (45%)

Accelerated/blast phase (n=39) Normal Abnormal P (10%) karyotype (90%)

119/98

26/28

0.4463

32/30

27/23

0.8504

2/2

21/14

1.0000

54

54

0.4886

51

50

0.8608

65

54

0.1357

154/63

25/29

0.0012

15/47

10/40

0.6531

1/3

21/14

0.2998

14.6

13.3

0.0124

10.0

9.7

0.2998

9.5

9.2

0.9618

11.3

12.6

0.7415

15.7

12.3

0.8764

9.9

9.4

0.3912

421.5

329

0.0025

323

212

0.0034

198

75

0.1315

64 122 23

10 30 13

0.0127

0 0 59

0 0 50

0 1 1

5 6 19

137

116

0.0322

73

47

46

9

0.0460

0.1624

*Presented as median of the values.

in complex karyotypes were -5/del(5q) (n=18), -7/del(7q) (n=18), -17/del(17p)/add(17p) ( n=14), and -18 (n=11). Of note, three cases had sole –Y, and no patient had del5q/-5 or del(7q)/-7 as a sole abnormality. The distribution of chromosomal abnormalities varied in different stages of PV. In the PP, the most common cytogenetic abnormalities were sole del(20q), +8 or +9; in the phase of post-PV MF, the most common ones were sole del(20q) and +1q; and in the AP/BP, the most common one was a complex karyotype. The higher the disease stage, the higher the frequency of abnormal karyotypes: 20% in PP, 45% in post-PV MF, 90% in AP/BP. Additionally, the frequency of a complex karyotype increased as the disease stage advanced, 1.5% in PP; 10.7% in post-PV MF; 61.5% in AP/BP. Among the 114 patients who had cytogenetic analyses performed at the time of initial diagnosis, 107 were in PP and seven in the post-PV MF phase. Among the 107 patients in PP, an abnormal karyotype was detected in 17 (15%) patients, a single abnormality, including +9 (n=6), +8 (n=3), del(20q) (n=3), -Y (n=1), and del(11q) (n=1) in 14, double abnormalities (1 with +1q) in two, and complex karyotype in one. Among the seven patients in post-PV MF, six had a normal karyotype and one (14.3%) had isolated add(21p).

Clinical follow-up and disease progression The median follow-up was 36 months (range, 0-168 months) from diagnosis (the first BM biopsy). Of the 372 patients for whom there was follow-up information, 66 were under observation or received phlebotomy only, 88 were treated with JAK2 inhibitors, 48 with interferon, 18 with imatinib, and 129 received single or combined therapies that included hydroxyurea, anagrelide, revlimid, haematologica | 2017; 102(9)

Table 2. Cytogenetic abnormalities detected at the diagnosis (first bone marrow evaluation).

Polycythemic Post-PV MF AP/BP phase Total phase (n=271) (n=112) (n=39) (n=422) Normal karyotype 217 (80%) Abnormal karyotype 54 (20%) Single abnormalities 41 (76%) - del20q 18 - +9 10 - +8 6 - other single 7 Double abnormalities 9 (17%) - +1q 4 - other two 5 Complex 4 (7%) - del5q/-5 0 - del7q/-7 1 - del17p/-17/i(17q) 1

62 (55%) 50 (45%) 29 (58%) 12 0 1 16 9 (18%) 7 2 12 (24%) 4 2 4

4 (10%) 35 (90%) 5 (14%) 1 0 1 3 6 (17%) 4 2 24 (69%) 14 15 9

283 (67%) 139(33%) 75 (54%) 31 10 8 26 24 (17%) 15 9 40 (29%) 18 18 14

AP/BP: accelerated/blast phase; Post-PV MF: post-polycythemic myelofibrosis.

azacitidine, or induction chemotherapies; ten patients also underwent allogeneic stem cell transplantation. Information on therapies was not available for 23 patients. At last follow-up, 123 (29%) patients had died, including 49 from PP, 44 from post-PV MF, and 30 from AP/BP. The median OS was 137, 60, and 9 months for patients in PP, post-PV MF, AP/BP stages, respectively; the OS was significantly different among patients in different stages (Figure 1). Disease progression was assessed in patients who had two or more (up to 28) BM evaluations during the follow-up, and was evaluated in PP and post-PV MF stages separately. 1513


G. Tang et al.

As shown in Table 3, 146 (54%) patients in PP had follow-up BM evaluations, 45 (31%) patients showed disease progression, 31 patients progressed to post-PV MF and 14 progressed to AP/BP. Patients with an abnormal karyotype showed a higher risk of disease progression and a significantly shorter transformation-free survival compared with patients who had a normal karyotype (Figure 2A). As also shown in Table 3, 76 (68%) patients in the postPV MF phase had follow-up BM evaluations and 21 (28%) showed disease progression to AP/BP. Patients with an abnormal karyotype showed a significantly shorter transformation-free survival (Figure 2B), but a comparable risk of transformation, compared with patients who had a normal karyotype. Of the 107 patients who had BM evaluation at the time of initial diagnosis and were in PP, 59 (55%) had at least one follow-up BM specimen for evaluation. Patients with an abnormal karyotype showed a significantly higher frequency of transformation (60% versus 10%, P<0.0001) and a shorter transformation-free survival (101 months versus undefined, P=0.0004) compared with patients who had a normal karyotype (Table 3). We compared the disease progression among patients with different degrees of myelofibrosis (MF-0, MF-1, and MF-2/3) in PP. Among the 74 patients with MF-0, 32 patients had follow-up BM evaluations: 4/32 (12.5%)

patients showed disease progression. Among the 152 patients with MF-1, 84 patients had follow-up BM evaluations: 26/84 (31%) patients showed disease progression. Among 36 patients with MF-2/3, 29 patients had followup BM evaluations: 15/29 (52%) patients showed disease progression. These data reveal that the higher the grade of MF, the higher the frequency of disease progression (P=0.0570 for MF-0 versus MF-1; P=0.0090 for the three groups).

A

B

Figure 1. Overall survival of patients in different stages. Patients in a higher stage had a significantly inferior overall survival. AP/BP: accelerated/blast phase; MF: post-polycythemic myelofibrosis; PP: polycythemic phase.

Figure 2. Transformation-free survival of patients with normal and abnormal karyotypes. Patients with an abnormal karyotype had a significantly shorter transformation-free survival. (A) Patients in polycythemic phase; (B) patients with post-polycythemic myelofibrosis.

Table 3. Disease progression of patients with normal and abnormal karyotypes.

Stages

Karyotype

*Polycythemic phase (n=146)

Normal (n=111) Abnormal (n=36)

+

*Polycythemic phase (n=59)

Normal (n=49) Abnormal (n=10)

*Post-PV MF (n=76)

Normal (n=41) Abnormal (n=35)

Disease progression No Yes 86 (77%) 16 (44%)

25 (23%) 20 (56%)

Progressed to Post-PV MF AP/BP 18 (16%) 13 (34%)

P=0.0003 44 (90%) 4 (40%) P<0.0001 32 (78%) 23 (66%)

2 (4%) 5 (50%)

163 77 P<0.0001

3 (6%) 1 (10%)

Undefined 101 P=0.0004

9 (22%) 12 (34%)

Undefined Undefined P=0.0343

Total: 19% 9 (22%) 12 (34%)

P=0.3050

7 (6%) 7 (22%) Total: 31%

5 (10%) 6 (60%)

Total: 28%

TFS (median, months)

*Only patients who had two or more bone marrow evaluations were included. +Patients had bone marrow evaluation and karyotyping analysis at initial diagnosis. AP/BP: accelerated/blast phase; post-PV MF: post-polycythemic myelofibrosis; TFS: transformation-free survival.

1514

haematologica | 2017; 102(9)


Cytogenetics in PV

We also compared the OS among patients with MF-0, MF-1, and MF-2/3 in PP. The median OS for patients with MF-0, MF-1, and MF-2/3 were 169, 137, 126 months, respectively (P=0.0498 for MF-0 versus MF-1; P=0.0490 for the three groups).

Acquisition of cytogenetic abnormalities during the course of disease Of the 383 patients in the PP and post-PV MF phase, 224 (136 in PP and 74 with post-PV MF) had at least two (and up to 25) cytogenetic analyses at different time points. After a median interval of 35 months (range, 1-163), 20 of 136 (14.7%) patients in PP had ACA, including 13 (of 110, 11.8%) with a normal karyotype and seven (of 32, 21.8%) with an abnormal karyotype (P=0.1580). Of the 74 patients with post-PV MF, 23 (31%) had ACA, including 10 (of 37, 27%) with a normal karyotype and 13 (of 37, 35%) with an abnormal karyotype (P=0.6160). Of the 114 patients who had BM evaluation at the time of initial diagnosis, 59 (51.8%) had at least one follow-up cytogenetic analysis, and eight (13.6%) of them had ACA: six (12%) of 50 patients with a previously normal karyotype gained an abnormal clone, and two (22%) of nine patients with a previously abnormal karyotype showed clonal evolution. The commonly acquired chromosomal abnormalities included del(7q)/-7 (n=8), +1q (n=7), del(5q)/-5 (n=6), del(20q) (n=5), del(17p)/-17 (n=4), and complex karyotypes (n=14); whereas +8 (n=2), and +9 (n=1) were infrequently acquired during the course of disease.

del(20q), double abnormalities and a complex karyotype had a significantly shorter OS than those with a normal karyotype, whereas there was not a significant difference in OS for patients with sole +8, +9, or other single abnormalities. For patients in the stage of post-PV MF, a complex karyotype correlated with a significantly inferior OS, while sole del(20q), other single abnormalities, or double abnormalities failed to show a significant effect. Based on the results from the above analyses, we grouped the karyotypes into three risk groups: low-risk included a normal karyotype, sole +8, sole +9, and other single abnormalities; intermediate-risk included sole del(20q), double abnormalities (including +1q); and highrisk included complex karyotypes. As shown in Figure 4, patients with low-, intermediate -, and high-risk cytogenetics had significantly different OS, with a median OS of 169, 86, and 9 months in patients in PP (P<0.0001) (Figure 4A), and 83, 46, and 24 months in patients in the post-MF PV stage (P=0.0015) (Figure 4B).

Discussion Historically, the diagnosis of PV has relied mainly on high hemoglobin level (>18.5 g/dL in men and >16.5 g/dL in women), presence of a JAK2 mutation, and panmyelosis in the BM, and conventional cytogenetic testing is not routinely performed at the time of the initial diagnosis of PV, especially in the community hospital setting.

Correlation of abnormal karyotype, acquisition of cytogenetic abnormalities and disease progression A total of 195 patients, 122 in PP and 68 in the phase of post-PV MF, had at least one follow-up analysis on both BM morphology and cytogenetics. Among the 122 patients (92 with normal and 30 with abnormal karyotype) in PP, 18 (14.8%) patients gained ACA, including 11 (12%) patients with normal karyotype and seven (23%) with abnormal karyotype (P=0.1439). A total of 39 (35%) patients, 13 (72%) with ACA and 26 (25%) without ACA (P=0.0002), showed disease progression. Among the 68 patients (35 with normal and 33 with abnormal karyotype) in the phase of post-PV MF, 21 (31%) patients gained ACA, including nine (26%) patients with normal karyotype and 12 (36%) with abnormal karyotype (P=0.4335). A total of 19 (28%) patients, ten (48%) with ACA and nine (19%) without ACA, showed disease progression (P=0.0213). The frequency of ACA was comparable among patients with a normal versus an abnormal karyotype, however, patients with ACA showed a significantly higher risk of disease progression compared to patients without ACA.

A

B

Prognostic significance of cytogenetic abnormalities The median OS for patients with a normal versus an abnormal karyotype was 137 versus 116 months (P=0.0322) for patients in PP (Figure 3A); 73 versus 47 months (P=0.0460) for patients with post-PV MF (Figure 3B); and 46 versus 9 months (P=0.1624) for patients in AP/BP (Table 1). We evaluated the prognostic impact of specific chromosomal abnormalities on patientsâ&#x20AC;&#x2122; survival by stage (PP and post-PV MF, separately). As shown in Table 4 and in the Online Supplementary Material, in PP, patients with sole haematologica | 2017; 102(9)

Figure 3. Overall survival of patients with normal and abnormal karyotypes. Patients with an abnormal karyotype had a significantly inferior overall survival. (A) Patients in polycythemic phase; (B) patients with post-polycythemic myelofibrosis.

1515


G. Tang et al.

Furthermore, cytogenetic testing may not be performed during follow-up, unless there is a suspicion of disease progression (e.g. substantial change in blood cell counts). In our own cohort of patients, karyotype information at the time of initial clinical diagnosis was not available for approximately two-thirds of patients, and about 40% of patients did not have follow-up cytogenetic analyses. These factors plus an overall low frequency of abnormal karyotypes detected in PP have greatly hindered the characterization and risk stratification of cytogenetics in patients with PV. Here we performed a retrospective study on 422 patients who had cytogenetic information at the time of diagnosis (with BM evaluation), and characterized the significance of cytogenetics at different stages of PV. We made a number of significant findings in this large cohort of patients, and the data provide clear guidelines for the application of cytogenetic information in PV patients. We show, for the first time, that cytogenetic changes are dynamic in PV patients and correlate with the course of disease. The dynamic changes are first reflected in the frequency of abnormal karyotype, which increases with increasing disease stage (from 20% in the PP to 90% at BP). The dynamic changes are also shown in the distribution of specific cytogenetic abnormalities: +8 and +9 mainly in PP, +1q in the post-PV MF stage, del(5q)/-5, del(7q)/-7, del(17p)/-17/i(17q) and complex karyotypes in AP and BP. These dynamic changes in conjunction with the findings of ACA suggest that +8 and +9 are likely to be early genetic events during the pathogenesis of PV, while +1q, del(5q)/-5, del(7q)/-7 and complex karyotype are likely to be acquired during the course of disease and associated with advanced stage of disease. Similar findings have been suggested by others in earlier studies.18,23 We confirmed the prognostic relevance of abnormal karyotypes in PV,7,14 but most importantly, we characterized for the first time the prognostic impact of specific recurrent cytogenetic abnormalities. Abnormalities of +8 or +9, predominantly detected in the PP, did not appear to have a significant prognostic effect, which is in line with the findings of a previous study.13 Del(20q), the most common single abnormality in PV, showed a significant adverse effect in patients in the PP but not those with post-PV MF. Other single individual abnormalities when analyzed as a group did not show a significant effect on survival, either for patients in PP or the post-PV MF stage. It is noteworthy that these single individual abnormalities

did not include -5/-5q or -7/-7q. The most common abnormality found in PV patients with double abnormalities was +1q, which showed a significant effect on OS in patients in PP, but not in those in the post-PV MF phase. A complex karyotype showed a significant adverse effect on OS in both the PP and post-PV MF stage, similar to what has been shown in acute myeloid leukemia (AML),24 myelodysplastic syndromes (MDS),25 and chronic myelomonocytic leukemia (CMML).26 Another clear conclusion drawn from this study was the association of acquisition of additional cytogenetic abnor-

A

B

Figure 4. Overall survival of patients with low-, intermediate-, and high-risk cytogenetics. (A) Patients in polycythemic phase; (B) patients with post-polycythemic myelofibrosis.

Table 4. Impact of cytogenetic abnormalities on overall survival of patients in polycythemic phase and with post-polycythaemic myelofibrosis.

Normal karyotype Abnormal karyotype Single abnormality - del20q - +9 - +8 - other single Double abnormalities Complex

Cases N. (%)

Polycythemic phase (n=271) Median OS (months)

217 (80%) 54 (20%)

137 116

0.0322

18 10 6 7 9 4

91 129 Undefined Undefined 86 9

0.0011 0.3742 0.3781 0.6102 0.0483 <0.0001

P*

Cases N. (%)

Post-PV MF (n=112) Median OS (months)

62 (55%) 50 (45%)

73 47

0.0460

12

34

0.2214

1 16 9 12

87 41 18

0.9920 0.1524 0.0017

P*

*Compared to patients with a normal karyotype. Only groups with four or more patients were analyzed.

1516

haematologica | 2017; 102(9)


Cytogenetics in PV

malities and disease progression. By studying patients who had both BM evaluation and chromosomal analysis during follow-up, we found that patients who started with an abnormal karyotype had a higher risk of disease progression and shorter transformation-free survival; ACA was associated with a higher risk of disease transformation. These results are in line with the prognostic impact of ACA in patients with MDS27 and CMML.28 The prognostic effects of specific cytogenetic abnormalities are highly likely to be related to molecular changes that are induced by the corresponding chromosomal abnormalities. +8 is a common cytogenetic abnormality detected in various types of myeloid neoplasms and its prognostic effects are heterogeneous, but commonly assigned to the intermediate-risk group.25,26,29 +9 is commonly detected in Philadelphia-negative myeloproliferative neoplasms but is uncommon in other types of myeloid neoplasms (e.g. AML, MDS), and it has been assigned to favorable risk in patients with primary myelofibrosis.30 In our cohort, +8 and +9 were likely to be early genetic events and were not prognostically significant. +1q, commonly derived from aberrations of +1, and der(1;7)(q10;p10) (which also results in -7q), are relatively common abnormalities found in MDS and are often associated with low-risk MDS.31,32 +1q results in the gain of cyclin-dependent kinases regulatory subunit 1B (CKS1B, located on 1q21), which can override the DNA damage response barrier, promote tumor development,33 and is associated with an adverse prognosis in multiple myeloma.34,35 In our cohort, +1q was detected mainly in post-PV MF, and was one of the most common acquired abnormalities during the course of disease. In multivariate analysis it showed a significant adverse effect on OS. Del(20q) is reported to be associated with a favorable prognosis in primary myelofibrosis and MDS;25,30,36 however, in patients with de novo AML, del(20q) has been associated with a poor response to chemotherapy and is classified as intermediate-II risk.29 In our cohort, del(20q) was the most common sole abnormality detected in both PP and postPV MF and was significantly associated with a poorer OS in patients with PV in the PP. Interestingly, abnormalities involving chromosomes 5, 7 and/or 17 were rarely detected as a sole abnormality in PV, but were the most common abnormalities detected in complex karyotypes.

References 5. 1. Swerdlow SH, Campo C, Harris NL, et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Edited by International Agency for Research on Cancer: Lyon, 2008. 2. Barbui T, Thiele J, Gisslinger H, Finazzi G, Vannucchi AM, Tefferi A. The 2016 revision of WHO classification of myeloproliferative neoplasms: clinical and molecular advances. Blood Rev. 2016;30(6):453-459. 3. Cerquozzi S, Tefferi A. Blast transformation and fibrotic progression in polycythemia vera and essential thrombocythemia: a literature review of incidence and risk factors. Blood Cancer J. 2015;5: e366. 4. Passamonti F, Rumi E, Caramella M, et al. A dynamic prognostic model to predict sur-

haematologica | 2017; 102(9)

6.

7.

8.

These abnormalities were detected in about 44% (7/16) of patients in PP and with post-PV MF, and in 83% (20/24) of patients in AP and BP who had a complex karyotype. A complex karyotype has been associated with a higher risk of disease progression and inferior survival, as has also been shown in patients with AML,24 MDS,25 and CMML.26 One of the commonly detected abnormalities in a complex karyotype was del(17p)/-17/i(17q), which results in deletion of TP53, a gene that plays a critical role in regulating cell-cycle arrest and apoptosis.37 Loss of TP53 is associated with a poorer prognosis in patients with MDS38 and AML.39 Evaluation of myelofibrosis in patients in PP at initial diagnosis showed a slightly higher proportion of patients with a higher grade of myelofibrosis in our study compared with the proportion in the study conducted by Barbui et al.40 The exact reason for this was not clear, but it might be partly due to differences in the referred populations of patients. Nonetheless, both our study and that by Barbui et al. showed that a higher degree of myelofibrosis was often associated with a higher risk of disease progression and a shorter OS. This supports the importance of evaluating BM fibrosis during the initial diagnosis. In summary, we have reported the cytogenetic findings in a large series of patients with PV. The results show that the frequency of an abnormal karyotype and the distribution of cytogenetic abnormalities vary in different stages of PV. About 20% of patients may acquire cytogenetic abnormalities during the course of the disease, an event which is strongly associated with disease progression. Patients with an abnormal karyotype were at a higher risk of disease progression and shorter transformation-free survival and OS. Different cytogenetic abnormalities carried different prognoses and could be effectively stratified into three risk groups. These findings highlight the value of obtaining cytogenetic information in PV patients, which may be useful to guide clinical management and assess the prognosis of patients. Acknowledgments The authors would like to thank Kate Newberry for the review and grammar checking and our colleagues for their helpful discussions and support throughout this study.

vival in post-polycythemia vera myelofibrosis. Blood. 2008;111(7):3383-3387. Kennedy JA, Atenafu EG, Messner HA, et al. Treatment outcomes following leukemic transformation in Philadelphianegative myeloproliferative neoplasms. Blood. 2013;121(14):2725-2733. Passamonti F, Rumi E, Pungolino E, et al. Life expectancy and prognostic factors for survival in patients with polycythemia vera and essential thrombocythemia. Am J Med. 2004;117(10):755-761. Tefferi A, Rumi E, Finazzi G, et al. Survival and prognosis among 1545 patients with contemporary polycythemia vera: an international study. Leukemia. 2013;27(9):18741881. Tefferi A, Guglielmelli P, Larson DR, et al. Long-term survival and blast transformation in molecularly annotated essential thrombocythemia, polycythemia vera, and myelofibrosis. Blood. 2014;124(16):2507-2513.

9. Bonicelli G, Abdulkarim K, Mounier M, et al. Leucocytosis and thrombosis at diagnosis are associated with poor survival in polycythaemia vera: a population-based study of 327 patients. Br J Haematol. 2013;160(2):251-254. 10. Finazzi G, Caruso V, Marchioli R, et al. Acute leukemia in polycythemia vera: an analysis of 1638 patients enrolled in a prospective observational study. Blood. 2005;105(7):2664-2670. 11. Gangat N, Strand J, Li CY, Wu W, Pardanani A, Tefferi A. Leucocytosis in polycythaemia vera predicts both inferior survival and leukaemic transformation. Br J Haematol. 2007;138(3):354-358. 12. Andrieux JL, Demory JL. Karyotype and molecular cytogenetic studies in polycythemia vera. Curr Hematol Rep. 2005;4(3):224-229. 13. Sever M, Quintas-Cardama A, Pierce S, Zhou L, Kantarjian H, Verstovsek S.

1517


G. Tang et al.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

1518

Significance of cytogenetic abnormalities in patients with polycythemia vera. Leuk Lymphoma. 2013;54(12):2667-2670. Diez-Martin JL, Graham DL, Petitt RM, Dewald GW. Chromosome studies in 104 patients with polycythemia vera. Mayo Clin Proc. 1991;66(3):287-299. Gangat N, Strand J, Lasho TL, et al. Cytogenetic studies at diagnosis in polycythemia vera: clinical and JAK2V617F allele burden correlates. Eur J Haematol. 2008;80(3):197-200. Swolin B, Weinfeld A, Westin J. A prospective long-term cytogenetic study in polycythemia vera in relation to treatment and clinical course. Blood. 1988;72(2):386-395. Reilly JT. Pathogenetic insight and prognostic information from standard and molecular cytogenetic studies in the BCR-ABLnegative myeloproliferative neoplasms (MPNs). Leukemia. 2008;22(10):1818-1827. Swolin B, Rodjer S, Westin J. Therapy-related patterns of cytogenetic abnormalities in acute myeloid leukemia and myelodysplastic syndrome post polycythemia vera: single center experience and review of literature. Ann Hematol. 2008;87(6):467-474. Dingli D, Schwager SM, Mesa RA, Li CY, Dewald GW, Tefferi A. Presence of unfavorable cytogenetic abnormalities is the strongest predictor of poor survival in secondary myelofibrosis. Cancer. 2006;106(9): 1985-1989. Lawler SD. Cytogenetic studies in Philadelphia chromosome-negative myeloproliferative disorders, particularly polycythaemia rubra vera. Clin Haematol. 1980;9(1):159-174. Thiele J, Kvasnicka HM, Facchetti F, Franco V, van der Walt J, Orazi A. European consensus on grading bone marrow fibrosis and assessment of cellularity. Haematologica. 2005;90(8):1128-1132. Shaffer LG, McGowan-Jordan J, Schmid M. An International System for Human Cytogenetic Nomenclature. S. Karger, Basel 2013. Passamonti F, Rumi E, Arcaini L, et al.

24.

25.

26.

27.

28.

29.

30.

31.

Leukemic transformation of polycythemia vera: a single center study of 23 patients. Cancer. 2005;104(5):1032-1036. 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. Greenberg PL, Tuechler H, Schanz J, et al. Revised international prognostic scoring system for myelodysplastic syndromes. Blood. 2012;120(12):2454-2465. Tang G, Zhang L, Fu B, et al. Cytogenetic risk stratification of 417 patients with chronic myelomonocytic leukemia from a single institution. Am J Hematol. 2014;89(8):813-818. Jabbour E, Takahashi K, Wang X, et al. Acquisition of cytogenetic abnormalities in patients with IPSS defined lower-risk myelodysplastic syndrome is associated with poor prognosis and transformation to acute myelogenous leukemia. Am J Hematol. 2013;88(10):831-837. Tang G, Fu B, Hu S, et al. Prognostic impact of acquisition of cytogenetic abnormalities during the course of chronic myelomonocytic leukemia. Am J Hematol. 2015;90(10):882-887. Dohner H, Estey EH, Amadori S, et al. Diagnosis and management of acute myeloid leukemia in adults: recommendations from an international expert panel, on behalf of the European LeukemiaNet. Blood. 2010;115(3):453-474. Hussein K, Pardanani AD, Van Dyke DL, Hanson CA, Tefferi A. International Prognostic Scoring System-independent cytogenetic risk categorization in primary myelofibrosis. Blood. 2010;115(3):496499. Slovak ML, O'Donnell M, Smith DD, Gaal K. Does MDS with der(1;7)(q10;p10) constitute a distinct risk group? A retrospective single institutional analysis of

clinical/pathologic features compared to 7/del(7q) MDS. Cancer Genet Cytogenet. 2009;193(2):78-85. 32. Sanada M, Uike N, Ohyashiki K, et al. Unbalanced translocation der(1;7)(q10;p10) defines a unique clinicopathological subgroup of myeloid neoplasms. Leukemia. 2007;21(5):992-997. 33. Martinsson-Ahlzen HS, Liberal V, Grunenfelder B, Chaves SR, Spruck CH, Reed SI. Cyclin-dependent kinase-associated proteins Cks1 and Cks2 are essential during early embryogenesis and for cell cycle progression in somatic cells. Mol Cell Biol. 2008;28(18):5698-5709. 34. Shi L, Wang S, Zangari M, et al. Overexpression of CKS1B activates both MEK/ERK and JAK/STAT3 signaling pathways and promotes myeloma cell drugresistance. Oncotarget. 2010;1(1):22-33. 35. Zhan F, Colla S, Wu X, et al. CKS1B, overexpressed in aggressive disease, regulates multiple myeloma growth and survival through SKP2- and p27Kip1-dependent and -independent mechanisms. Blood. 2007; 109(11):4995-5001. 36. Caramazza D, Begna KH, Gangat N, et al. Refined cytogenetic-risk categorization for overall and leukemia-free survival in primary myelofibrosis: a single center study of 433 patients. Leukemia. 2011;25 (1):82-88. 37. Vousden KH, Lu X. Live or let die: the cell's response to p53. Nat Rev Cancer. 2002; 2(8):594-604. 38. Horiike S, Kita-Sasai Y, Nakao M, Taniwaki M. Configuration of the TP53 gene as an independent prognostic parameter of myelodysplastic syndrome. Leuk Lymphoma. 2003;44(6):915-922. 39. Seifert H, Mohr B, Thiede C, et al. The prognostic impact of 17p (p53) deletion in 2272 adults with acute myeloid leukemia. Leukemia. 2009;23(4):656-663. 40. Barbui T, Thiele J, Passamonti F, et al. Initial bone marrow reticulin fibrosis in polycythemia vera exerts an impact on clinical outcome. Blood. 2012;119(10):2239-2241.

haematologica | 2017; 102(9)


ARTICLE

Chronic Myeloid Leukemia

Combined targeting of STAT3 and STAT5: a novel approach to overcome drug resistance in chronic myeloid leukemia Karoline V. Gleixner,1,2 Mathias Schneeweiss,2 Gregor Eisenwort,2 Daniela Berger,1 Harald Herrmann,2,3 Katharina Blatt,1 Georg Greiner,4 Konstantin Byrgazov,5 Gregor Hoermann,2,4 Marina Konopleva,6 Islam Waliul,7 Abbarna A. Cumaraswamy,8 Patrick T. Gunning,8 Hiroshi Maeda,7 Richard Moriggl,9,10 Michael Deininger,11 Thomas Lion,5,12 Michael Andreeff6 and Peter Valent1,2

Department of Internal Medicine I, Division of Hematology and Hemostaseology, Medical University of Vienna, Austria; 2Ludwig Boltzmann Cluster Oncology, Medical University of Vienna, Austria; 3Department of Radiation Therapy, Medical University of Vienna, Austria; 4Department of Laboratory Medicine, Medical University of Vienna, Austria; 5Children’s Cancer Research Institute (CCRI), Vienna, Austria; 6Department of Leukemia, University of Texas, MD Anderson Cancer Center, Houston, TX, USA; 7Institute of Drug Delivery Sciences, Sojo University, Kumamoto and BioDynamics Research Laboratory, Kumamoto, Japan; 8Department of Chemistry, University of Toronto, Canada; 9 Ludwig Boltzmann Institute for Cancer Research, Vienna, Austria; 10Institute of Animal Breeding and Genetics, University of Veterinary Medicine, Vienna, Austria; 11Division of Hematology and Hematologic Malignancies, Huntsman Cancer Institute, University of Utah, Salt Lake City, UT, USA and 12Department of Pediatrics, Medical University of Vienna, Austria 1

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(9):1519-1529

ABSTRACT

I

n chronic myeloid leukemia, resistance against BCR-ABL1 tyrosine kinase inhibitors can develop because of BCR-ABL1 mutations, activation of additional pro-oncogenic pathways, and stem cell resistance. Drug combinations covering a broad range of targets may overcome resistance. CDDO-Me (bardoxolone methyl) is a drug that inhibits the survival of leukemic cells by targeting different pro-survival molecules, including STAT3. We found that CDDO-Me inhibits proliferation and survival of tyrosine kinase inhibitor-resistant BCR-ABL1+ cell lines and primary leukemic cells, including cells harboring BCR-ABL1T315I or T315I+ compound mutations. Furthermore, CDDO-Me was found to block growth and survival of CD34+/CD38− leukemic stem cells (LSC). Moreover, CDDO-Me was found to produce synergistic growthinhibitory effects when combined with BCR-ABL1 tyrosine kinase inhibitors. These drug-combinations were found to block multiple signaling cascades and molecules, including STAT3 and STAT5. Furthermore, combined targeting of STAT3 and STAT5 by shRNA and STAT5-targeting drugs also resulted in synergistic growth-inhibition, pointing to a new efficient concept of combinatorial STAT3 and STAT5 inhibition. However, CDDO-Me was also found to increase the expression of heme-oxygenase-1, a heat-shock-protein that triggers drug resistance and cell survival. We therefore combined CDDO-Me with the heme-oxygenase-1 inhibitor SMA-ZnPP, which also resulted in synergistic growth-inhibitory effects. Moreover, SMA-ZnPP was found to sensitize BCR-ABL1+ cells against the combination ‘CDDO-Me+ tyrosine kinase inhibitor’. Together, combined targeting of STAT3, STAT5, and heme-oxygenase-1 overcomes resistance in BCR-ABL1+ cells, including stem cells and highly resistant sub-clones expressing BCR-ABL1T315I or T315I-compound mutations. Whether such drug-combinations are effective in tyrosine kinase inhibitor-resistant patients with chronic myeloid leukemia remains to be elucidated. haematologica | 2017; 102(9)

Correspondence: karoline.gleixner@meduniwien.ac.at

Received: January 2, 2017. Accepted: June 7, 2017. Pre-published: June 8, 2017. doi:10.3324/haematol.2016.163436 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1519 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1519


K.V. Gleixner et al.

Introduction Chronic myeloid leukemia (CML) is a stem cell disease characterized by the reciprocal translocation t(9;22) that creates the BCR-ABL1 oncoprotein, a major driver of disease evolution.1-3 Most patients with chronic phase (CP) CML achieve long-lasting cytogenetic and molecular responses when treated with the BCR-ABL1 tyrosine kinase inhibitor (TKI) imatinib.4-6 However, resistance against imatinib occurs in a substantial number of patients. Several molecular mechanisms, including BCRABL1 mutations, may contribute to TKI resistance in CML. Indeed, BCR-ABL1 mutations are identified in more

than 50% of all resistant patients.7,8 For these patients, 2ndand 3rd-generation TKI, including nilotinib, dasatinib, bosutinib, and ponatinib, are available and have shown beneficial effects.9-12 Using these drugs, it is now possible to cover most of the known BCR-ABL1 mutations detected in TKI-resistant CML. Ponatinib, a 3rd-generation BCRABL1 TKI, induces growth-inhibitory effects in TKI-resistant patients even if T315I is expressed.12 However, not all mutant forms of BCR-ABL1 are responsive to ponatinib. Moreover, it has been described that additional (multiple) mutations in BCR-ABL1, especially T315I-including compound mutations, confer resistance against ponatinib.13 Furthermore, resistance against TKI may occur independ-

Table 1. Patients’ characteristics and response of leukemic cells to CDDO-Me.

Patient number

WBC % BCR-ABL1/ BCR-ABL1 (x109/L) ABL mutations

Age (years)

Sex (m/f)

Sokal Score (diagnosis)

CML phase

#1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12 #13

65 64 41 46 45 62 84 31 34 19 54 23 45

m m f m f m m m f m f m f

0.89 0.77 1.65 0.77 0.7 0.84 1.05 0.98 0.79 0.61 1.53 0.76 1.29

CP CP CP CP CP CP CP CP CP CP CP CP CP

112 38 12 54 184 46 46 190 425 175 541 128 13

35.2 49.1 45.3 14.7 43.1 41.1 64.7 26.7 43.84 78.47 34.71 56.9 44.3

n.t. n.t. n.t. n.t. n.t. n.t. n.t. n.t n.t. n.t. n.t. n.d. n.d.

#14

75

m

unknown

CP

14

44.3

F359V

#15

63

f

1.6

CP

37

57.2

V379I

#16

39

m

1.07

CP

7

16.2

T315I

#17 #18

62 48

m m

unknown unknown

BP my/ly BP my

24 3.5

38.71 25.07

n.d. G250V

#19 #20

59 78

f m

15.41 n.t.

BP ly ALL

94 29

39.3 47.2

n.t. T315I/E255K

Therapy before cell sampling

CDDO-Me IC50

none none none none none none none none none none none imatinib (dis) imatinib (int) cytarabine dasatinib (res) nilotinib (res) imatinib (res) dasatinib (int) nilotinib (res) imatinib (res) dasatinib (res) nilotinib (res) cytarabine mitoxantron imatinib (res) dasatinib (res) imatinib (dis) imatinib (res) dasatinib (res) bosutinib (res) none polychemotherapy imatinib (res) ponatinib (res)

350 – 500 nM < 100 nM 100 – 250 nM < 100 nM 300–500 nM < 100 nM 300 – 500 nM 250 – 350 nM n.t. n.t. 100 – 250 nM 100 – 500 nM < 100 nM

350 – 500 nM

250 – 350 nM

100 – 250 nM 250 – 350 nM 100 – 250 nM 100 – 250 nM 250 – 500 nM

WBC: white blood count; m: male; f: female; CP: chronic phase; BP: blast phase; my: myeloid; ly: lymphatic; ALL: acute lymphatic leukemia; n.d.: not detected; n.t.: not tested; none: no therapy (diagnostic sample); dis: discontinued; int: intolerant; res: resistant; nM: nanomolar. Patient age, phase of disease, WBC and % of BCR-ABL1 relative to ABL (International Scale) were reported at the time of sampling. Responses of cells to CDDO-Me were assessed by 3H-thymidine uptake.

1520

haematologica | 2017; 102(9)


Combined targeting of STAT3 and STAT5 in CML

ent of BCR-ABL1 mutations. In such cases, overexpression of BCR-ABL1 and/or hyper-activation of additional prooncogenic signaling networks and molecules, such as AKT, mTOR, MEK, STAT3, STAT5, JAK2, or SRC kinases, have been described.14-18 These molecules and pathways are often spared by the TKI used and can, therefore, contribute to drug resistance.14-20 Recently, several targeting approaches have been proposed with the aim of overcoming TKI resistance in advanced CML. One option may be to apply combinations of targeted drugs in order to cover a larger spectrum of relevant targets in TKI-resistant cells. CDDO-Me (bardoxolone methyl) is an oleanane triterpenoid that has been described as inducing ROS-generation and to suppress a number of survival-related molecules, including AKT, mTOR, MAPK and STAT3, in malignant cells.21-26 It has also been reported that CDDO-Me promotes apoptosis in malignant cells in various neoplasms, including CML.21-26 Currently, CDDO-Me is tested in clinical trials in patients with diabetic nephropathy, a condition that may improve with CDDO-induced upreg-

ulation of the Nrf2-pathway.27,28 In addition, CDDO-Me is currently tested in clinical trials in cancer patients.29 With regard to CML, it has been reported that CDDO-Me counteracts the proliferation of BCR-ABL1+ cell lines by altering mitochondrial function and by inducing autophagy and apoptosis, regardless of the mutation status of BCR-ABL1.30 So far, the co-operative effects of BCRABL1 TKI and CDDO-Me on CML cells have not been analyzed. We hypothesized that, due to the large number of relevant targets blocked by CDDO-Me, this drug would be an optimal combination partner for BCR-ABL1 TKI. For example, CDDO-Me is a potent inhibitor of STAT3, a transcription factor that may play an important role in TKI-resistant CML cells, and is not blocked by BCR-ABL1 TKI.14,15,31-33 Indeed, recent data suggest that combined targeting of BCR-ABL1 and STAT3 exerts strong anti-leukemic effects in CML cells.34 Such combined targeting may be achieved by co-applying CDDO-Me and a BCR-ABL1 TKI. However, CDDO-Me is not able to suppress all pathways activated in CML cells. Notably, recent

A

B

C

D

Figure 1. CDDO-Me inhibits growth and viability of Philadelphia-positive (Ph+) cell lines. (A-C) Human chronic myeloid leukemia (CML) cell lines (A) and BCR-ABL1expressing Ba/F3 cells (B and C) were exposed to control medium (Co) or various concentrations of CDDO-Me for 48 hours (h). In case of K562-R, imatinib was removed prior to CDDO-Me exposure. Then, proliferation was measured by assessing 3H-thymidine uptake. Results are expressed in % of control and represent the meanÂąStandard Deviation (S.D.) of 3 independent experiments. (D) CML cell lines (K562, KU812, KCL22) were incubated in control medium or in CDDO-Me (0.5 mM) for 48 h. Thereafter, the percentage of apoptotic cells was determined by combined Annexin V/PI staining. The figure shows dot blots from one representative experiment. Almost identical results were obtained in two other experiments.

haematologica | 2017; 102(9)

1521


K.V. Gleixner et al.

data suggest that exposure of leukemic cells to CDDO-Me results in a substantial increase in heme-oxygenase-1 (HO1),30 a major survival factor that has been implicated in drug resistance of CML cells.35 As a result, HO-1 is considered an additional attractive target in TKI-resistant CML cells, and we hypothesized that combining CDDO-Me with an HO-1-inhibitor can enhance drug efficiency.

SMA-ZnPP is an HO-1 inhibitor that counteracts growth and survival in CML cells and synergizes with imatinib in producing growth inhibition.36 In the current study, we evaluated drug interactions between CDDO-Me, 2nd- and 3rd-generation TKI, and SMA-ZnPP with the aim of enhancing anti-leukemic drug effects in TKI-resistant CML.

A

B

C

D

Figure 2. CDDO-Me counteracts the proliferation of primary chronic myeloid leukemia (CML) cells. (A) Primary CML cells were isolated from the peripheral blood (PB) of 6 patients [chronic phase (CP) n=5; primary blast phase (BP) n=1]. In 3 patients, cells were obtained at diagnosis (“new”). In all 3 TKI-resistant patients, BCR-ABL1 mutations were detected as indicated. Isolated cells were incubated in control medium (Co) or various concentrations of CDDO-Me as indicated at 37°C for 48 hours (h). Then, proliferation was measured by assessing 3H-thymidine incorporation. Results are expressed in % of control and represent the mean±Standard Deviation (S.D.) of triplicates. Patients’ numbers refer to Table 1. (B) Highly purified CD34+/CD38– stem cells (black bars) and CD34+/CD38+ precursor cells (gray bars) were sorted from peripheral blood (PB) leukocytes of 3 patients (#9, #11 and #17) and were kept in control medium (Co) or various concentrations of CDDOMe as indicated at 37°C for 48 h. Then, proliferation was measured by assessing 3H-thymidine incorporation. Results are expressed as % of control and represent the mean±S.D. of 3 patients. *P<0.05 compared to control (Co). (C) Primary PB mononuclear cells (MNC) were isolated from 3 patients (#9, #11, #17) and kept in control medium (Co) or various concentrations of CDDO-Me at 37°C for 48 h. Thereafter, cells were subjected to flow cytometry to determine the % of apoptotic (Annexin V+) CD34+/CD38– (stem) cells (black bars) and CD34+/CD38+ (progenitor) cells (gray bars). Results represent the mean±S.D. of 3 patients. (D) PB MNC from 3 patients (#9, #10, #11) were cultured in methylcellulose with cytokines in the absence (Co) or presence of various concentrations of CDDO-Me as indicated for 14 days. Then, the numbers of granulocyte/macrophage (GM) colonies (white bars) and red cell-containing (burst-forming plus erythroid) colonies (gray bars) were counted under an inverted microscope. Results are expressed in % of control (100% = red colonies + GM colonies in the absence of CDDO-Me) and represent the mean±S.D. of 3 patients.

1522

haematologica | 2017; 102(9)


Combined targeting of STAT3 and STAT5 in CML

Methods Reagents Reagents used in this study are described in the Online Supplementary Appendix.

Isolation and culture of primary CML cells Primary leukemic cells were obtained from the peripheral blood (PB) of 16 patients with CP CML, one patient with myeloid BP, one with mixed myeloid/lymphoid BP, one with lymphoid BP, and one with relapsed BCR-ABL1+ acute lymphoblastic leukemia (ALL). Patients’ characteristics are shown in Table 1. Mononuclear cells (MNC) were isolated and kept in RPMI 1640 medium with 10% fetal calf serum (FCS) and antibiotics (without cytokines) as described.36 In all samples, cell viability was more than 90%. In 3 donors (CP, n=2; mixed BP, n=1), CD34+/CD38− leukemic stem cells (LSC) and CD34+/CD38+ progenitor cells were highly purified from MNC by high-speed sorting as reported37 using PE-conjugated monoclonal antibody (mAb) 581 against CD34, APC-conjugat-

ed mAb HIT2 against CD38, and a BD FACSAria (Becton Dickinson, San Jose, CA, USA). Normal bone marrow (BM) MNC were isolated from 7 donors (remission of acute leukemia or lymphoma patients without BM-involvement) after informed consent. Normal CD34+ BM cells were purchased from Lonza (Basel, Switzerland) and maintained in StemSpan Medium (Stemcell Technologies, Vancouver, Canada).

Cell lines A characterization of cell lines used in this study is shown in Online Supplementary Table S1.

Evaluation of growth and survival of leukemic cells Drug-exposed cells (cell lines, primary cells, LSC) were analyzed for proliferation, colony-formation, and survival. The bioassays employed are described in the Online Supplementary Methods.

Western blotting

Cell lines were incubated with CDDO-Me (0.1–1 mM) or pona-

A

B

C

D

Figure 3. CDDO-Me synergizes with BCR-ABL1 TKI in producing growth-inhibition in Philadelphia-positive (Ph+) cells. (A and B) Human chronic myeloid leukemia (CML) cell lines (A) or Ba/F3 cells harboring various BCR-ABL1 mutants (B) were incubated in control medium (0) or in various concentrations of CDDO-Me (●-●), BCR-ABL1-targeting TKI as indicated (■-■), or combinations of drugs at a fixed ratio of drug-concentrations as indicated (▲−▲) for 48 hours (h). Thereafter, 3Hthymidine incorporation was measured. Results are expressed in % of control and represent the mean± Standard Deviation (S.D.) of triplicates. (C) Primary neoplastic cells isolated from patient #1, #11, #17 and #20 as well as normal bone marrow (BM) mononuclear cells (MNC) obtained from 3 donors were incubated in various concentrations of CDDO-Me or ponatinib (as single drugs or in combination) as indicated. Results in the upper panels and the lower left panel are expressed in % of control and represent the mean±S.D. of triplicates. Results in the lower right panel show the mean±S.D. of 3 normal donors (gray bars) or 3 CML patients (#1, #11 and #17) (black bars). (D) Primary peripheral blood (PB) MNC were isolated from 3 patients (#9, #11, #17) and kept in control medium (0) or various concentrations of CDDO-Me or ponatinib as indicated at 37°C for 48 h. Thereafter, cells were subjected to flow cytometry to determine the % of apoptotic (Annexin V+) CD34+/CD38– (stem) cells (black bars) and CD34+/CD38+ (progenitor) cells (gray bars). Results represent the mean±S.D. of 3 patients.

haematologica | 2017; 102(9)

1523


K.V. Gleixner et al. tinib (1 mM) alone or in combination for 4 hours (h) or for 24 h. Thereafter, the expression of STAT5, phosphorylated (p) STAT5 (p-STAT5), STAT3, p-STAT3, CRKL, p-CRKL, p-JAK2, JAK2, ERK, p-ERK, S6, p-S6, HO-1, or Actin (loading control) was analyzed by Western blotting as described previously.36,38 Antibodies used are described in Online Supplementary Table S2.

shRNA-based knockdown experiments For knockdown of STAT3 or STAT5, shRNA constructs were transduced in K562 and KCL22 using VSV-G pseudotyped lentiviruses, as described previously.39 The methodology and the shRNA-constructs are described in detail in the Online Supplementary Methods.

Overexpression of STAT3 in K562 cells pMOWS retroviral vectors containing the coding sequence of GFP and wild-type (wt) STAT3 or an oncogenic mutant-form of STAT3, D661V, were kindly provided by Jürgen Scheller.40 Production of recombinant VSV-G pseudo-typed retroviruses and transduction of K562 cells were performed as described.39 After transduction, transfected cells were purified by puromycin-selection (20 ng/mL for 48 h).

Statistical analysis To determine the significance levels in differences seen between drug-exposed and untreated cells, the Student t-test for dependent samples was applied. P<0.05 was considered statistically signifi-

A

B

C

D

Figure 4. Combined inhibition of STAT3 and STAT5 with shRNA and targeted drugs results in synergistic growth inhibition in chronic myeloid leukemia (CML) cells. (A) KCL22 and K562 cells were incubated with CDDO-Me (1 mM), ponatinib (1 mM) or a combination of both drugs for 4 hours (h). Thereafter, cells were subjected to Western blot analysis using antibodies against p-STAT5, STAT5, p-STAT3, STAT3, p-CRKL, or CRKL, as indicated. (B-D) K562 and KCL22 cells were transfected with control shRNA, with an shRNA-construct directed against STAT5, or shRNA-constructs directed against STAT3 (#1 = #V3LHS_376016; #2 = #V3LHS_641818) as indicated. Protein knockdown was confirmed by western blotting using antibodies against STAT3 or STAT5. β-Actin served as loading control (B). (C) K562 cells (upper panels) and KCL22 cells (lower panels) were treated with control-shRNA (black bars) or with shRNA directed against STAT3 (construct #2, gray bars) and were then incubated with control medium (control), with BCR-ABL1 TKI (as indicated), or with the STAT5 inhibitor AC-3-019 for 48 h. Results are expressed in % of control and represent the mean±Standard Deviation (S.D.) of triplicates. *P<0.05. (D) K562 cells (left panel) and KCL22 cells (right panel) were treated with control-shRNA (black bars) or shRNA against STAT5 (gray bars) and were then incubated in control medium or in CDDO-Me for 48 h. Results are expressed in % of control and represent the mean±S.D. of triplicates. *P<0.05.

1524

haematologica | 2017; 102(9)


Combined targeting of STAT3 and STAT5 in CML

cant. Drug combination effects (additive vs. synergistic) were determined by calculating combination index (CI) values using Calcusyn software (Calcusyn; Biosoft, Ferguson, MO, USA).41 Approval was obtained from the Institutional Review Board (Department of Internal Medicine I, Division of Hematology and Hemostaseology, Medical University of Vienna, Austria) and from the Ethics Committee of the Medical University of Vienna for all series of experiments of this study.

Results CDDO-Me inhibits proliferation and viability in TKI-sensitive and TKI-resistant BCR-ABL1+ cell lines CDDO-Me was found to inhibit the proliferation of all four human CML cell lines tested, with IC50 values ranging between 0.1 and 0.5 mM (Figure 1A). A summary of growth-inhibitory effects of CDDO-Me on CML cells lines and a comparison with the effects elicited by BCRABL1 TKI are shown in Online Supplementary Table S3. CDDO-Me was also found to counteract growth of Ba/F3 cells expressing various imatinib-resistant forms of BCRABL1 (Figure 1B). In addition, CDDO-Me was found to suppress proliferation of Ba/F3 cells harboring T315Iincluding BCR-ABL1 compound mutations mediating resistance against all currently available TKI, including ponatinib, with IC50 values ranging between 0.1 and 0.35 µM (Figure 1C and Online Supplementary Table S4). Interestingly, in the human and Ba/F3 cell lines examined, no differences in IC50 values were seen when comparing TKI-sensitive and TKI-resistant clones, suggesting that BCR-ABL1 mutations did not influence responsiveness against CDDO-Me (Figure 1A-C). Growth-inhibition was accompanied by induction of apoptosis in all cell lines examined as demonstrated by flow cytometry (Figure 1D). In non-transfected, IL-3-dependent Ba/F3 cells, CDDO-Me also exerted growth-inhibitory effects, but the concentrations required to fully block cell growth were higher compared to that producing growth inhibition in BCR-ABL1+ Ba/F3 cells (Online Supplementary Figure S1).

CDDO-Me counteracts proliferation in primary CML cells Primary PB MNC of 10 patients with newly diagnosed CML (CP: n=9; BP: n=1), 5 CML patients who had developed resistance against one or more TKI (CP: n=4; BP: n=1), 2 CML patients who had discontinued imatinib and relapsed (CP: n=1; BP: n=1), and one patient suffering from ponatinib-resistant BCR-ABL1+ ALL were examined. CDDO-Me was found to suppress the proliferation of primary BCR-ABL1+ cells in all samples tested, with IC50 -values ranging between 0.1 and 0.5 mM (Table 1 and Figure 2A). Drug effects were seen in samples isolated from heavily pre-treated patients with CML, in blast cells obtained from patients with BP (including one with mixed lymphoid/myeloid BP), and in primary ALL blasts harboring BCR-ABL1T315I/E255K (Table 1 and Figure 2A). No differences in IC50 values were observed when comparing newly diagnosed and relapsed patients or cells expressing or lacking BCR-ABL1 mutations (including BCRABL1T315I/E255K), suggesting that acquired resistance against TKI does not lead to resistance against CDDO-Me (Table 1 and Figure 2A). In control BM cells, CDDO-Me also exerted growth-inhibitory effects, but the concentrations required to fully block cell growth were higher compared haematologica | 2017; 102(9)

to that producing growth inhibition in CML cells (Online Supplementary Figure S1).

CDDO-Me inhibits the proliferation of primary LSC and progenitor cells obtained from patients with CML Leukemic stem cells exhibit multiple forms of drug resistance.42,43 Therefore, we were interested to learn whether CDDO-Me would also block the proliferation of CML LSC. Indeed, we found that CDDO-Me dose-dependently inhibits the proliferation of highly enriched CD34+/CD38− stem cells and CD34+/CD38+ progenitor cells obtained from patients with CML (Figure 2B). In normal CD34+ stem- and progenitor cells, CDDO-Me produced only weak effects on proliferation (Online Supplementary Figure S1). CDDO-Me was also found to induce apoptosis in CML stem- and progenitor cells (Figure 2C). Stromal components may protect CML LSC against the effects of CDDO-Me.20,44 To address this point, the pro-apoptotic effects of CDDO-Me were analyzed in co-cultures prepared with primary CML cells and murine feeder cells. In these experiments, CDDO-Me was again found to induce apoptosis in the total CML cell compartment as well as in the CD34+/CD38− and CD34+/CD38+ cell fractions (Online Supplementary Figure S2). Finally, CDDO-Me was found to inhibit colony-formation of primary CML cells in vitro (Figure 2D).

CDDO-Me synergizes with BCR-ABL1-targeting TKI in inhibiting the proliferation of CML cells In relapsed or TKI-resistant CML, combinations of two or more substances may be required to block both BCRABL1-dependent and BCR-ABL1-independent pathways and to achieve long-lasting and stable complete responses in all patients. In this study, we combined CDDO-Me with imatinib, nilotinib, dasatinib, or ponatinib at suboptimal concentrations. These combinations induced synergistic growth-inhibitory effects in all human CML cell lines tested (Figure 3A and Online Supplementary Figure S3). Synergistic effects of these drug combinations were also observed in Ba/F3 cells expressing various BCR-ABL1 mutations, including T315I and T315I-including compound mutations (Figure 3B), whereas no significant effects were seen in untransfected Ba/F3 cells when the same drug concentrations were applied (Online Supplementary Figure S3). Synergistic drug effects were confirmed by calculating CI values using calcusyn software (Online Supplementary Figure S4). Synergistic effects of CDDO-Me and ponatinib were also found in primary BCR-ABL1+ cells, including cells isolated from a patient with CML BP and one with BCR-ABL1T315I/E255K-positive relapsed ALL (Figure 3C). Synergistic drug interactions were found in all donors and cell samples, independent of the presence of BCR-ABL1 mutations, and (in case of primary cells) independent of previous treatment. By contrast, only weak effects of this drug combination (CDDOMe+ponatinib) were observed in normal BM cells (Figure 3C). In CD34+/CD38− stem cells and CD34+/CD38+ progenitor cells obtained from 3 patients, the combination “CDDO-Me+ponatinib” was found to produce clear cooperative apoptosis-inducing effects (Figure 3D). Co-operative inhibitory effects of CDDO-Me and ponatinib were also seen in a clonogenic assay performed with primary samples obtained from 2 patients with CP-CML (Online Supplementary Figure S3). Together, these results suggest that CDDO-Me augments the anti-neoplastic effects of BCR-ABL1 TKI in CML cells, including LSC. 1525


K.V. Gleixner et al.

CDDO-Me plus BCR-ABL1-targeting TKI induce simultaneous inhibition of STAT3 and STAT5 in CML cells We next asked for mechanisms underlying the synergistic effect of the drug combinations (CDDO-Me plus BCRABL1-targeting TKI) applied. In particular, we examined

A

B

potentially involved BCR-ABL1-dependent targets, including STAT5, ERK, the S6- ribosomal protein and JAK2, and the BCR-ABL1-independent target STAT3. As expected, ponatinib was found to modulate p-CRKL, p-ERK, pJAK2, p-S6 and p-STAT5 expression in CML cells, but did

C

D

E

Figure 5. SMA-ZnPP sensitizes BCR-ABL1+ cells against CDDO-Me and against the combination “CDDO-Me+BCR-ABL1 TKI”. (A) KCL22, KU812 and BCR-ABL1+ Ba/F3 cells were incubated in control medium or in 0.1 or 0.3 mM CDDO-Me for 24 hours (h). Then, cells were subjected to Western blot analysis using antibodies against HO-1 or either β-Actin or Actin (loading control) as indicated. (B-D) Cell lines (B and C) and primary chronic myeloid leukemia (CML) cells (D) were incubated in control medium or in various concentrations of CDDO-Me (●-●), SMA-ZnPP (■-■), or a combination of both drugs (at fixed ratio of drug concentrations) (▲−▲) for 48 hours (h). Then, 3H-thymidine incorporation was measured. Results are expressed in % of control and represent the mean±Standard Deviation (S.D.) of triplicates. Patient numbers in (D) refer to Table 1. (E) KU812 and Ba/F3p210T315I/F311L cells were exposed to low doses of CDDO-Me (●-●), a BCR-ABL1 TKI (dasatinib or ponatinib) (■-■), and SMA-ZnPP (▲−▲), either as single agents (blue graphs), in 2-drug combinations (green graphs; CDDO-Me+TKI: ▼-▼; CDDO-Me+SMA-ZnPP: ♦-♦; TKI+SMA-ZnPP: ○-○), or in a 3-drug combination (red graph; □-□) for 48 h. Then, 3H-thymidine uptake was measured. Results are expressed in % of control and represent the mean±S.D. of triplicates.

1526

haematologica | 2017; 102(9)


Combined targeting of STAT3 and STAT5 in CML

only partially inhibit expression of p-STAT3 (Figure 4A and Online Supplementary Figure S5). By contrast, CDDOMe was found to inhibit p-STAT3, p-JAK2 and p-S6, but did not block phosphorylation of CRKL, and produced only a slight effect on p-STAT5 expression (Figure 4A and Online Supplementary Figure S5). Only the combination of both compounds resulted in complete de-phosphorylation of both STAT3 and STAT5. This may explain the synergy between CDDO-Me and BCR-ABL1-targeting TKI. To further validate this hypothesis, shRNAs against STAT3 and against STAT5 were employed. shRNA inducedknockdown of STAT3- and STAT5-protein expression was confirmed by Western blotting (Figure 4B). shRNA directed against STAT3 was found to sensitize CML cell lines against nilotinib, dasatinib, and ponatinib, as well as against the more specific STAT5-inhibitor AC-3-019 (Figure 4C). Moreover, we were able to show that shRNAmediated STAT5 knockdown enhances the effects of CDDO-Me, although the combination effect was additive rather than synergistic (Figure 4D). Together, these data suggest that simultaneous inhibition of STAT3 and STAT5 is a potent approach to block growth and survival of (TKIresistant) CML cells. In order to confirm the role of STAT3 as a target of CDDO-Me, the effects of STAT3 overexpression in K562 cells was examined. We found that overexpression of STAT3 or STAT3 D661V results in relative resistance against CDDO-Me, confirming that STAT3 serves as a functionally relevant target in leukemic cells (Online Supplementary Figure S5).

The HO-1 inhibitor SMA-ZnPP sensitizes CML cells against CDDO-Me and TKI Recent data suggest that CDDO-Me does not suppress all survival factors in CML cells.30 Indeed, HO-1, a survival factor with anti-apoptotic activity in CML cells, has been described to be up-regulated by CDDO-Me,30 a fact that can best be interpreted as a “tumor-escape mechanism”.36,45 CDDO-Me led to an upregulation of HO-1 protein expression in KU812, KCL22, and BCR-ABL1+ Ba/F3 cells, confirming previous observations (Figure 5A).30 We therefore combined CDDO-Me with the HO-1 inhibitor SMA-ZnPP. Indeed, SMA-ZnPP was found to sensitize CML cell lines against CDDO-Me as evidenced by 3Hthymidine-uptake experiments (Figure 5B). Synergistic effects were also observed in Ba/F3 cells expressing BCRABL1 mutants and in primary CML cells (Figure 5C and D). No growth inhibitory effects of the drug combination on normal BM cells were seen (Online Supplementary Figure S6). Finally, we asked whether a triple-combination, including CDDO-Me, a BCR-ABL1 TKI (dasatinib or ponatinib), and SMA-ZnPP, would further enhance drug effects. In these experiments, we incubated KU812 and Ba/F3p210T315I/F311L cells with low doses of 3 compounds, either as single agents or in combination. The anti-proliferative effect of the triple-combination was superior to single drug- or 2-drug combinations (Figure 5E). These observations suggest that the triple combination "CDDOMe+TKI+SMA-ZnPP" exerts strong anti-neoplastic effects in CML cells.

Discussion Despite the availability of novel potent BCR-ABL1 TKI, drug resistance develops frequently and remains a major haematologica | 2017; 102(9)

challenge in the treatment of CML.7,8,13 Important mechanisms underlying drug resistance are intrinsic stem cell resistance,42,43 BCR-ABL1 mutations, including compound mutations,7,8,13 and additional, BCR-ABL1-independent, signaling pathways and molecules. Among these are the JAK-STAT pathway, the RAS-RAF-MEK pathway, and several different survival-related molecules.16-20 We report here that BCR-ABL1 TKI co-operate with the STAT3inhibitor CDDO-Me in producing growth inhibition and apoptosis in drug-resistant CML cells. This combination blocked most of the relevant signaling and survival molecules in CML cells, except HO-1, a survival molecule that is even up-regulated upon exposure to CDDO-Me in leukemic cells. In line with this observation, the HO-1 blocker SMA-ZnPP further augmented the anti-leukemic effects produced by the combination CDDO-Me+TKI in CML cells. Recent data suggest that CDDO-Me-targets play a major role in the survival and growth of neoplastic cells in solid tumors and leukemias.21-26,30,46 In the present study, CDDO-Me was found to inhibit the proliferation of TKIsensitive and TKI-resistant CML cells and of Ba/F3 cells exhibiting various BCR-ABL1 mutations, including compound mutations involving T315I. This finding is of interest since such T315I+ compound mutations of BCR-ABL1 often confer resistance against all available TKI, including ponatinib. IC50-values were low (<0.5 mM) in all cell lines tested, without major differences between TKI-sensitive and TKI-resistant cells. Similar observations were made using primary CML cells obtained from newly diagnosed or heavily pre-treated patients with BCR-ABL1+ CML. Finally, we were able to show that CDDO-Me inhibits the proliferation of CD34+/CD38− CML stem cells and CD34+/CD38+ CML progenitor cells at pharmacologically meaningful concentrations. All these data suggest that CDDO-Me may be an interesting drug for patients in whom neoplastic cells have lost sensitivity to 2nd- or 3rdline TKI. The fact that CDDO-Me was found to inhibit the proliferation of TKI-resistant cells in all cell lines and all patients tested, suggests that no “cross-resistance” between TKI and CDDO-Me occurs. This was not surprising since CDDO-Me does not recognize BCR-ABL1, but inhibits cell proliferation by blocking other signaling pathways and molecules, such as STAT3. Previous studies have shown that CDDO-Me inhibits STAT3 activation in various neoplastic cells.23,24 In the present study, we were able to confirm that CDDO-Me blocks STAT3 activity in CML cells. By contrast, CDDO-Me did not suppress STAT5 activation. Recent data suggest that STAT3 plays an essential role in the survival of CML cells,32-34 an observation that is of particular interest in light of the fact that ponatinib exerted only weak effects on STAT3 activity in the CML cell lines K562 and KCL22 in this study. We therefore asked whether a combination consisting of CDDO-Me and a BCR-ABL1-targeting TKI would result in synergistic growth inhibition. Indeed, we were able to demonstrate that CDDO-Me synergizes with imatinib and with the 2nd- and 3rd-generation BCRABL1 TKI applied in inhibiting growth of CML cells. These effects were seen in TKI-sensitive as well as TKIresistant CML cells, suggesting that these combinations may be an attractive therapeutic approach in advanced CML. In this regard it is worthy to note that the drug combinations applied were also found to exert strong co-operative inhibitory effects on growth of primary blast cells 1527


K.V. Gleixner et al.

obtained from a patient with mixed myeloid/lymphoid BP and in one with Ph+ ALL. Since STAT3 and STAT5 may differentially contribute to the pathogenesis of myeloid and lymphoid BP in CML, it may be of importance to block both pathways to keep all PB-associated (lymphoid and myeloid determined) sub-clones under control in these patients. Recent data have shown that CDDO-Me enhances the expression of HO-1 in various neoplastic cells, including CML cells.30 HO-1 is a stress molecule that contributes to drug-resistance and “tumor escape mechanisms” in various malignancies, including CML.35,36,45 We therefore hypothesized that pharmacological suppression of HO-1 could potentially enhance the effects of CDDO-Me and also augment the effects of the combination “CDDOMe+TKI” on malignant proliferation in CML cells. We therefore combined the HO-1 inhibitor SMA-ZnPP with CDDO-Me or with the 2-drug combination “CDDOMe+TKI”. Addition of SAM-ZnPP was indeed found to augment CDDO-Me effects, and the “3-drug-combination” applied was found to exert superior anti-neoplastic effects compared to single agents or the “2-drug-combination”. This finding can best be explained by the fact that up-regulated HO-1 protects CML cells against druginduced growth inhibition and apoptosis, and that this protective effect is eliminated by exposure to SMAZnPP.36,45 This concept is also in line with our previous observations.36,45 Recently, CDDO-Me has been introduced in clinical trials in patients with solid tumors and lymphomas, as well as in patients with diabetes mellitus and renal failure. A limited number of trials have been completed so far, and only very few results are available.27-29,47 In patients with malignant diseases, CDDO-Me was well tolerated.29 However, other trials have shown that CDDO-Me can cause severe cardiotoxic side effects, especially in patients with diabetes mellitus.47 These observations have to be

References 1. Nowell PC, Hungerford DA. A minute chromosome in human granulocytic leukemia. Science. 1960;132:1497. 2. Rowley JD. A new consistent chromosomal abnormality in chronic myelogenous leukaemia identified by quinacrine fluorescence and Giemsa staining. Nature. 1973;243(5405):290-293. 3. Bedi A, Zehnbauer BA, Barber JP, Sharkis SJ, Jones RJ. Inhibition of apoptosis by BCR-ABL in chronic myeloid leukemia. Blood. 1994;83(8):2038-2044. 4. Druker BJ, Talpaz M, Resta DJ, et al. Efficacy and safety of a specific inhibitor of the BCR-ABL tyrosine kinase in chronic myeloid leukemia. N Engl J Med. 2001;344(14):1031-1037. 5. Kantarjian H, Sawyers C, Hochhaus A, et al. Hematologic and cytogenetic responses to imatinib mesylate in chronic myelogenous leukemia. N Engl J Med. 2002;346(9):645-652. 6. Druker BJ, Guilhot F, O'Brien SG, et al. Five-year follow-up of patients receiving imatinib for chronic myeloid leukemia. N Engl J Med. 2006;355(23):2408-2417. 7. Gorre ME, Mohammed M, Ellwood K, et

1528

8.

9.

10.

11.

12.

13.

taken into account before CDDO-Me can be developed further in patients with CML, especially when combined with nilotinib or ponatinib, both of which can also produce cardiovascular side effects.48-50 An additional concern may be drug-induced cytopenia. In fact, higher concentrations of CDDO-Me did not only inhibit growth of CML cells but also growth of normal BM cells. On the other hand, we expect that the application of such drug combinations will permit significant dose reductions of individual agents, thereby decreasing or even eliminating the risk of development of severe side effects. Indeed, we found that the drug combinations CDDO-Me+ponatinib, CDDO-Me+SMA-ZnPP and CDDO-Me+SMAZnPP+ponatinib did not significantly inhibit the proliferation of normal BM cells. Together, we demonstrate that drug combinations consisting of BCR-ABL1 TKI, CDDO-Me, and SMA-ZnPP, are most effective in inhibiting the proliferation and survival of TKI-resistant CML cells. The superior effects of these drug combinations are best explained by complete suppression of multiple triggers of proliferation and survival, including STAT3, STAT5, and HO-1. Whether this concept can be applied in patients with advanced CML remains to be determined in clinical trials. Acknowledgments We like to thank Gabriela Stefanzl for excellent technical assistance and Jisung Park and Gary Tin for assisting in synthesizing AC-3-019 for this study. Funding This study was supported by Austrian Science Fund (FWF), grants F4701-B20 and F4704-B20 (to PV), F4705-B20 (to TL) and F4707-B20 (to RM) and by a Grant-in-Aid from the Ministry of Welfare, Health and Labor of Japan, (201220042), and A-STEP for cancer grant (AS242Z01542Q) from the Japan Science and Technology Agency (to HM).

al. Clinical resistance to STI-571 cancer therapy caused by BCR-ABL gene mutation or amplification. Science. 2001; 293(5531):876-880. Shah NP, Nicoll JM, Nagar B, et al. Multiple BCR-ABL kinase domain mutations confer polyclonal resistance to the tyrosine kinase inhibitor imatinib (STI571) in chronic phase and blast crisis chronic myeloid leukemia. Cancer Cell. 2002;2(2):117-125. Weisberg E, Manley PW, Breitenstein W, et al. Characterization of AMN107, a selective inhibitor of native and mutant Bcr-Abl. Cancer Cell. 2005;7(2):129-141. Shah NP, Tran C, Lee FY, Chen P, Norris D, Sawyers CL. Overriding imatinib resistance with a novel ABL kinase inhibitor. Science. 2004;305(5682):399-401. Puttini M, Coluccia AM, Boschelli F, et al. In vitro and in vivo activity of SKI-606, a novel Src-Abl inhibitor, against imatinibresistant Bcr-Abl+ neoplastic cells. Cancer Res. 2006;66(23):11314-11322. O'Hare T, Shakespeare WC, Zhu X, et al. AP24534, a pan-BCR-ABL inhibitor for chronic myeloid leukemia, potently inhibits the T315I mutant and overcomes mutation-based resistance. Cancer Cell. 2009;16(5):401-412. Zabriskie MS, Eide CA, Tantravahi SK, et

14.

15.

16.

17.

18.

al. BCR-ABL1 compound mutations combining key kinase domain positions confer clinical resistance to ponatinib in Ph chromosome-positive leukemia. Cancer Cell. 2014;26(3):428-442. Coppo P, Flamant S, De Mas V, et al. BCRABL activates STAT3 via JAK and MEK pathways in human cells. Br J Haematol. 2006;134(2):171-179. Stella S, Tirrò E, Conte E, et al. Suppression of survivin induced by a BCRABL/JAK2/STAT3 pathway sensitizes imatinib-resistant CML cells to different cytotoxic drugs. Mol Cancer Ther. 2013;12(6):1085-1098. Traer E, MacKenzie R, Snead J, et al. Blockade of JAK2-mediated extrinsic survival signals restores sensitivity of CML cells to ABL inhibitors. Leukemia. 2012;26(5):1140-1143. Warsch W, Walz C, Sexl V. JAK of all trades: JAK2-STAT5 as novel therapeutic targets in BCR-ABL1+ chronic myeloid leukemia. Blood. 2013;122(13):2167-2175. Zimmerman EI, Dollins CM, Crawford M, et al. Lyn kinase-dependent regulation of miR181 and myeloid cell leukemia-1 expression: implications for drug resistance in myelogenous leukemia. Mol Pharmacol. 2010;78(5):811-817.

haematologica | 2017; 102(9)


Combined targeting of STAT3 and STAT5 in CML

19. Hayette S, Chabane K, Michallet M, et al. Longitudinal studies of SRC family kinases in imatinib- and dasatinib-resistant chronic myelogenous leukemia patients. Leuk Res. 2011;35(1):38-43. 20. Tabe Y, Jin L, Iwabuchi K, et al. Role of stromal microenvironment in nonpharmacological resistance of CML to imatinib through Lyn/CXCR4 interactions in lipid rafts. Leukemia. 2012;26(5):883-892. 21. Konopleva M, Tsao T, Ruvolo P, et al. Novel triterpenoid CDDO-Me is a potent inducer of apoptosis and differentiation in acute myelogenous leukemia. Blood. 2002; 99(1):326-335. 22. Konopleva M, Contractor R, Kurinna SM, Chen W, Andreeff M, Ruvolo PP. The novel triterpenoid CDDO-Me suppresses MAPK pathways and promotes p38 activation in acute myeloid leukemia cells. Leukemia. 2005;19(8):1350-1354. 23. Ling X, Konopleva M, Zeng Z, et al. The novel triterpenoid C-28 methyl ester of 2cyano-3, 12-dioxoolen-1, 9-dien-28-oic acid inhibits metastatic murine breast tumor growth through inactivation of STAT3 signaling. Cancer Res. 2007;67(9):4210-4218. 24. Ahmad R, Raina D, Meyer C, Kufe D. Triterpenoid CDDO-methyl ester inhibits the Janus-activated kinase-1 (JAK1)-->signal transducer and activator of transcription-3 (STAT3) pathway by direct inhibition of JAK1 and STAT3. Cancer Res. 2008;68(8):2920-2926. 25. Shishodia S, Sethi G, Konopleva M, Andreeff M, Aggarwal BB. A synthetic triterpenoid, CDDO-Me, inhibits IkappaBalpha kinase and enhances apoptosis induced by TNF and chemotherapeutic agents through down-regulation of expression of nuclear factor kappaB-regulated gene products in human leukemic cells. Clin Cancer Res. 2006;12(6):1828-1838. 26. Deeb D, Gao X, Jiang H, et al. Oleanane triterpenoid CDDO-Me inhibits growth and induces apoptosis in prostate cancer cells through a ROS-dependent mechanism. Biochem Pharmacol. 2010;79(3):350360. 27. de Zeeuw D, Akizawa T, Agarwal R, al. Rationale and trial design of Bardoxolone Methyl Evaluation in Patients with Chronic Kidney Disease and Type 2 Diabetes: the Occurrence of Renal Events (BEACON). Am J Nephrol. 2013;37(3):212-222. 28. Warnock DG, Hebbar S, Bargman J, et al.

haematologica | 2017; 102(9)

29.

30.

31.

32. 33.

34.

35.

36.

37.

38.

Prospective safety study of bardoxolone methyl in patients with type 2 diabetes mellitus, end-stage renal disease and peritoneal dialysis. Contrib Nephrol. 2012;178:157-163. Hong DS, Kurzrock R, Supko JG, et al. A phase I first-in-human trial of bardoxolone methyl in patients with advanced solid tumors and lymphomas. Clin Cancer Res. 2012;18(12):3396-3406. Samudio I, Kurinna S, Ruvolo P, et al. Inhibition of mitochondrial metabolism by methyl-2-cyano-3,12-dioxooleana-1,9diene-28-oate induces apoptotic or autophagic cell death in chronic myeloid leukemia cells. Mol Cancer Ther. 2008;7(5):1130-1139. Coppo P, Dusanter-Fourt I, Millot G, et al. Constitutive and specific activation of STAT3 by BCR-ABL in embryonic stem cells. Oncogene. 2003;22(26):4102-4110. Nair RR, Tolentino JH, Hazlehurst LA. Role of STAT3 in Transformation and Drug Resistance in CML. Front Oncol. 2012;2:30. Eiring AM, Kraft IL, Page BD, O'Hare T, Gunning PT, Deininger MW. STAT3 as a mediator of BCR-ABL1-independent resistance in chronic myeloid leukemia. Leuk Suppl. 2014;3(Suppl 1):S5-6. Eiring AM, Page BD, Kraft IL, et al. Combined STAT3 and BCR-ABL1 inhibition induces synthetic lethality in therapyresistant chronic myeloid leukemia. Leukemia. 2015;29(3):586-597. Mayerhofer M, Florian S, Krauth MT, et al. Identification of heme oxygenase-1 as a novel BCR/ABL-dependent survival factor in chronic myeloid leukemia. Cancer Res. 2004;64(9):3148-3154. Mayerhofer M, Gleixner KV, Mayerhofer J, et al. Targeting of heat shock protein 32 (Hsp32)/heme oxygenase-1 (HO-1) in leukemic cells in chronic myeloid leukemia: a novel approach to overcome resistance against imatinib. Blood. 2008;111(4):22002210. Herrmann H, Sadovnik I, Cerny-Reiterer S, et al. Dipeptidylpeptidase IV (CD26) defines leukemic stem cells (LSC) in chronic myeloid leukemia. Blood. 2014; 123(25):3951-3962. Gleixner KV, Ferenc V, Peter B, et al. Pololike kinase 1 (Plk1) as a novel drug target in chronic myeloid leukemia: overriding imatinib resistance with the Plk1 inhibitor BI 2536. Cancer Res. 2010;70(4):1513-1523.

39. Hoermann G, Cerny-Reiterer S, Herrmann H, et al. Identification of oncostatin M as a JAK2 V617F-dependent amplifier of cytokine production and bone marrow remodeling in myeloproliferative neoplasms. FASEB J. 2012;26(2):894-906. 40. Aparicio-Siegmund S, Sommer J, Monhasery N, et al. Inhibition of protein kinase II (CK2) prevents induced signal transducer and activator of transcription (STAT)1/3 and constitutive STAT3 activation. Oncotarget. 2014;5(8):2131-2148. 41. Chou TC, Talalay P. Quantitative analysis of dose-effect relationships: the combined effects of multiple drugs or enzyme inhibitors. Adv Enzyme Regul. 1984;22:2755. 42. Jiang X, Zhao Y, Smith C, et al. Chronic myeloid leukemia stem cells possess multiple unique features of resistance to BCRABL targeted therapies. Leukemia. 2007; 21(5):926-935. 43. Valent P. Emerging stem cell concepts for imatinib-resistant chronic myeloid leukaemia: implications for the biology, management, and therapy of the disease. Br J Haematol. 2008;142(3):361-378. 44. Nair RR, Tolentino J, Hazlehurst LA. The bone marrow microenvironment as a sanctuary for minimal residual disease in CML. Biochem Pharmacol. 2010;80(5):602-612. 45. Gleixner KV, Mayerhofer M, Vales A, et al. Targeting of Hsp32 in solid tumors and leukemias: a novel approach to optimize anticancer therapy. Curr Cancer Drug Targets. 2009;9(5):675-689. 46. Qin D, Wang W, Lei H, et al. CDDO-Me reveals USP7 as a novel target in ovarian cancer cells. Oncotarget. 2016; 7(47):7709677109. 47. de Zeeuw D, Akizawa T, Audhya P, et al. Bardoxolone Methyl in type 2 diabetes and stage 4 chronic kidney disease. N Engl J Med 2013;369(26):2492-2503. 48. Dalzell MD. Ponatinib pulled off market over safety issues. Manag Care. 2013; 22(12):42-43. 49. Le Coutre P, Rea D, Abruzzese E, et al. Severe peripheral arterial disease during nilotinib therapy. J Natl Cancer Inst. 2001;103(17):1347-1348. 50. Jain P, Kantarjian H, Jabbour E, et al. Ponatinib as first-line treatment for patients with chronic myeloid leukaemia in chronic phase: a phase 2 study. Lancet Haematol. 2015;2(9):e376-383.

1529


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Chronic Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2017 Volume 102(9):1530-1536

Incidence of second primary malignancies and related mortality in patients with imatinib-treated chronic myeloid leukemia

Gabriele Gugliotta,1 Fausto Castagnetti,1 Massimo Breccia,2 Francesco Albano,3 Alessandra Iurlo,4 Tamara Intermesoli,5 Elisabetta Abruzzese,6 Luciano Levato,7 Mariella D’Adda,8 Patrizia Pregno,9 Francesco Cavazzini,10 Fabio Stagno,11 Bruno Martino,12 Gaetano La Barba,13 Federica Sorà,14 Mario Tiribelli,15 Catia Bigazzi,16 Gianni Binotto,17 Massimiliano Bonifacio,18 Clementina Caracciolo,19 Simona Soverini,1 Robin Foà,2 Michele Cavo,1 Giovanni Martinelli,1 Fabrizio Pane,20 Giuseppe Saglio,21 Michele Baccarani22 and Gianantonio Rosti;1 Gruppo Italiano Malattie Ematologiche dell’Adulto - Chronic Myeloid Leukemia Working Party.

Institute of Hematology “L. and A. Seràgnoli”, “S. Orsola-Malpighi” University Hospital, Department of Experimental, Diagnostic and Specialty Medicine, University of Bologna; 2 Division of Cellular Biotechnologies and Hematology, University Sapienza, Rome; 3Chair of Hematology, University of Bari; 4Oncohematology Division, IRCCS Ca' Granda Maggiore Policlinico Hospital Foundation, Milan; 5Hematology Unit, Azienda Ospedaliera “Papa Giovanni XXIII”, Bergamo; 6Hematology Unit, “S. Eugenio” Hospital, Rome; 7 Hematology Unit, “Pugliese-Ciaccio” Hospital, Catanzaro; 8Hematology Unit, Azienda Ospedaliera “Spedali Civili”, Brescia; 9Hematology Unit, Azienda Ospedaliero Universitaria “Città della Salute e della Scienza”, Torino; 10Chair of Hematology, Azienda Ospedaliero Universitaria Arcispedale “S. Anna”, University of Ferrara; 11Chair and Division of Hematology, Azienda Ospedaliero Universitaria Policlinico – V. Emanuele, University of Catania; 12Hematology Unit, Azienda Ospedaliera “Bianchi-MelacrinoMorelli”, Reggio Calabria; 13Department of Hematology, "Spirito Santo" Hospital, Pescara; 14 Chair of Hematology, “Cattolica del Sacro Cuore” University, Fondazione Policlinico Universitario Agostino Gemelli, Rome; 15Division of Hematology and Bone Marrow Transplantation, Azienda Sanitaria Universitaria Integrata di Udine; 16Hematology Unit, "C. e G. Mazzoni" Hospital, Ascoli Piceno; 17Hematology Unit, Azienda Ospedaliera di Padova, University of Padova; 18Department of Medicine, Section of Hematology, University of Verona; 19Hematology Unit, "P. Giaccone" Hospital, Palermo; 20Department of Biochemistry and Medical Biotechnologies, “Federico II” University, Napoli; 21Chair of Hematology, Department of Clinical and Biological Sciences, “S Luigi Gonzaga” University Hospital, University of Torino, Orbassano and 22Department of Hematology and Oncology “L. and A. Seràgnoli”, University of Bologna, Italy 1

Correspondence: gabriele.gugliotta@unibo.it

Received: March 31, 2017. Accepted: May 31, 2017. Pre-published: June 1, 2017. doi:10.3324/haematol.2017.169532 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1530 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1530

ABSTRACT

T

he majority of patients with chronic myeloid leukemia are successfully managed with life-long treatment with tyrosine kinase inhibitors. In patients in chronic phase, other malignancies are among the most common causes of death, raising concerns on the relationship between these deaths and the off-target effects of tyrosine kinase inhibitors. We analyzed the incidence of second primary malignancies, and related mortality, in 514 chronic myeloid leukemia patients enrolled in clinical trials in which imatinib was given as first-line treatment. We then compared the observed incidence and mortality with those expected in the age- and sex-matched Italian general population, calculating standardized incidence and standardized mortality ratios. After a median follow-up of 74 months, 5.8% patients developed second primary malignancies. The median time from chronic myeloid leukemia to diagnosis of the second primary malignancies was 34 months. We did not find a higher incidence of second primary malignancies compared to that in the ageand sex-matched Italian general population, with standardized incidence ratios of 1.06 (95% CI: 0.57–1.54) and 1.61 (95% CI: 0.92–2.31) in males and females, respectively. Overall, 3.1% patients died of second primary malignancies. The death rate in patients with second primary malignancies was 53% (median overall survival: 18 months). Among females, the observed cancer-related mortality was superior to that expected in the haematologica | 2017; 102(9)


Outcome of second primary malignancies in CML

age- and sex-matched Italian population, with a standardized mortality ratio of 2.41 (95% CI: 1.26 – 3.56). In conclusion, our analysis of patients with imatinib-treated chronic myeloid leukemia did not reveal a higher incidence of second primary malignancies; however, the outcome of second primary malignancies in such patients was worse than expected. Clinicaltrials.gov: NCT00514488, NCT00510926.

Introduction The availability and the extensive use of tyrosine kinase inhibitors targeting the BCR-ABL protein in patients with chronic myeloid leukemia (CML) has reduced the rate of progression from chronic phase to advanced phase.1 As a consequence, at least 50% of deaths occur in patients in chronic phase, or in remission,2 raising concerns on the relationship of such deaths with the off-target effects of tyrosine kinase inhibitors.3 Although most of the attention is focused on cardiovascular adverse events,4 other malignancies are the most common cause of death in patients in chronic phase or in remission.2 Imatinib was the first tyrosine kinase inhibitor developed for the treatment of CML and is the most extensively studied. However, it is still unclear whether its immunomodulatory properties5-11 may affect anti-cancer

immune surveillance in the long-term, or whether its offtarget activity may influence oncosuppressive pathways. Of note, regardless of the underlying mechanisms, neoplastic alterations have been described in multiple tissues of rats exposed to imatinib.12 Several studies, mainly referring to imatinib-treated patients, have investigated the risk of second primary malignancy (SPM) in CML,13-20 with sometimes contrasting results. Indeed, in comparisons with the general population, some epidemiological studies of unselected CML patients reported higher incidences of SPM18,21 while similar incidences were found in three large analyses of patients enrolled in clinical trials.14,17,20 Moreover, it is still debated whether CML patients per se, regardless of the treatment used, might be at higher risk of SPM,21-24 a condition that might be now unveiled by the increased survival of patients. For all these reasons, it is important to retrieve addition-

Table 1. Characteristics of the patients at diagnosis of chronic myeloid leukemia. Patients, N (%) Males, N (%) Females, N (%) Age, years, median (range) Sokal score, N (%) Low Intermediate High EUTOS score, N (%) Low High ELTS score, N (%) Low Intermediate High ECOG performance status ≥ 1, N (%) Transcript type, N (%) b2a2 b3a3 b2a2/b3a2 other Clonal cytogenetic abnormalities, N (%)* Variant translocations, N (%)* Deletion 9q, N (%)* Prior hydroxyurea treatment, N (%) Imatinib high doses, N (%) Prior malignancies, N (%) Follow-up, months, median, (range)

All patients

Patients with SPM

Patients without SPM

P

514 309 (60) 205 (40) 52 (18 – 84)

30 (5.8) 17 (57) 13 (43) 60 (35 - 77)

484 (94.2) 292 (60) 192 (40) 51 (18 – 84)

0.002

200 (39) 204 (40) 110 (21)

7 (23) 20 (67) 3 (10)

193 (40) 184 (38) 107 (22)

0.008

476 (93) 38 (7)

30 (100) 0

446 (92) 38 (8)

0.15

282 (55) 160 (31) 72 (14) 108 (21)

15 (50) 13 (43) 2 (7) 10 (33)

267 (55) 147 (30) 70 (14) 98 (20)

0.23 0.088

185 (36) 267 (52) 56 (11) 6 (1) 17 (5) 26 (5.1) 55 (11) 238 (46) 126 (25) 25 (4.9) 74 (3-99)

11 (37) 16 (53) 3 (10) 0 1 (4.8) 3 (10) 6 (20) 9 (30) 9 (30) -

174 (36) 251 (52) 53 (11) 6 (1) 16 (5) 23 (4.7) 49 (11) 229 (47) 117 (24) -

0.98 1 0.18 0.13 0.088 0.51 -

*Of evaluable patients. The statistically significant difference observed for Sokal score probably reflects the higher median age of patients with SPM, resulting in a higher proportion of intermediate Sokal risk patients in this group. In addition, no significant statistical difference between patients with or without SPM was observed regarding values of hemoglobin, white blood cells, eosinophils, basophils, blasts, platelets, and spleen (data not shown). SPM: second primary malignancies; CML: chronic myeloid leukemia; EUTOS: EUropean Treatment Outcome Study; ELTS: EUTOS Long-Term Survival; ECOG: Eastern Cooperative Study Group.

haematologica | 2017; 102(9)

1531


G. Gugliotta et al.

al data on the potential carcinogenic role of tyrosine kinase inhibitors,12 on the incidence of other malignancies, and on their outcome.13-20 We report here on the malignancies observed in a cohort of 514 evaluable patients with newly diagnosed, chronic phase CML, treated first-line with imatinib in three multicenter national studies.

Methods We performed a retrospective analysis of 559 patients enrolled in three prospective clinical trials with imatinib front-line in 62 Italian institutions of the Gruppo Italiano Malattie Ematologiche dell’Adulto (GIMEMA) CML Working Party. Detailed inclusion criteria have been published previously.25-28 Briefly, patients were at least 18 years old, with a diagnosis of Philadelphia chromosome/BCR-ABL-positive CML in early chronic phase (6 months or less from diagnosis to starting imatinib; only hydroxyurea allowed). All the patients provided written informed consent before enrollment. The studies were reviewed and approved by the Internal Review Board of all the participating Institutions, and performed in accordance with the Declaration of Helsinki. For the purpose of the present analysis, a specific survey was conducted in all Centers with a request to review the clinical records of all the enrolled patients. Overall, 52/62 (84%) Institutions replied to the survey, and each Center reported on all its patients; overall, data were collected on 514/559 (92%) patients, 309 (60%) males and 205 (40%) females. Detailed data on all malignancies, prior to and after the diagnosis of CML, were collected, including: site, histology type, date of diagnosis, therapy (surgery, chemotherapy, radiotherapy, other), and outcome. Cancers were classified according to the International Classification of Diseases, version 10 (ICD-10). We excluded non-melanoma skin cancers (ICD-10: C-44) from the analysis, because of the possible under-reporting of such neoplasms, and acute leukemias/myelodysplastic syndromes, considering their possible relationship with CML. Prior malignancies were defined as malignancies diagnosed before CML, other malignancies denote all malignancies, including relapses of prior malignancies, diagnosed after CML. SPM, the focus of this analysis, are de novo malignancies diagnosed after CML (thus excluding relapses of prior malignancies). Descriptive statistics were used for SPM incidence and mortality. Means were compared with the t-test and frequencies with the χ2 test or Fisher exact test, as appropriate. Cumulative incidences and survival curves were estimated according to the Kaplan-Meier method. For comparison with the general population, we calculated the standardized incidence ratio (SIR) and standardized mortality ratio (SMR), which are based on the ratio between observed cases and expected cases in the general reference population in the same period, matched by sex and age (5-year age classes were considered). We reported the overall ratios (for subjects aged 2084 years), rather than those for specific age subgroups, to avoid selection biases. Data on cancer incidence and mortality in the general Italian population were taken in March 2016 from the AIRTUM (Associazione Italiana Registri TUMori) database ITACAN (AIRTUM ITACAN: Tumori in Italia, Versione 2.0; http://www.registritumori.it), which covers 51% of the Italian general population, and reports cancer incidence and mortality rates derived from real, observed cancer cases. Times to events (patient-years) were calculated from the date of diagnosis of CML to the date of diagnosis of the SPM, death, or last contact, whichever came first, for incidence; and to the date of death or last contact, for mortality.

1532

Results Patients Data from 514 patients, 309 (60%) males and 205 (40%) females, were analyzed. The median age at CML diagnosis was 52 (range, 18-84) years. The median follow-up from diagnosis of CML to death, or last contact, whichever came first, was 74 (range, 3-99) months. The total patient-years for the incidence calculation were 3011.1 (males and females: 1806.3 and 1204.8 patient-years, respectively). The total patient-years for the mortality calculation were 3077.7 (males and females: 1849.4 and 1228.3 patient-years, respectively). The characteristics of the whole cohort of patients and of patients with or without SPM are summarized in Table 1; age at CML diagnosis was significantly higher in patients with SPM than in patients without SPM (60 and 51 years, respectively; P=0.002).

Malignancies in the follow-up Overall, other malignancies were observed in 35/514 (6.8%) patients (Tables 2 and 3). Four patients had a relapse of a malignancy diagnosed before CML (2 bladder cancers, 1 renal cancer, and 1 breast cancer), and another patient developed multiple myeloma from a pre-existing monoclonal gammopathy of undetermined significance. Table 2. Malignancies observed in the follow-up and related mortality.

Malignancy type

Colon Prostate Breast Central nervous system Pancreas Bladder Liver Non-Hodgkin lymphoma Thyroid Bile duct Esophagus Lung Kidney Soft tissues Urethra Bowel Endometrium Stomach Multiple myeloma Ovary Rectum Testis TOTAL, n % of all patients (N=514)

All malignancies observed/deaths, N

Second primary malignancies observed/deaths, N

4/4 3/0 3/0 2/2 2/2 2/1 2 / 1§ 2/1 2/0 1/1 1/1 1/1 1/1 1/1 1/1 1/0 1/0 1/0 1/1 1/1 1/0 1/0 35 / 19 6.8 / 3.7

4/4 3/0 2/0 2/2 2/2 2 / 1§ 2/1 2/0 1/1 1/1 1/1 1/1 1/1 1/0 1/0 1/0 1/1 1/0 1/0 30 / 16 5.8 / 3.1

The other patient died due to progression of CML to blast phase. Treatment for other malignancies included chemotherapy in 14 patients, radiotherapy in nine patients, surgery in 19 patients, and hormonal therapy in one patient

§

haematologica | 2017; 102(9)


Outcome of second primary malignancies in CML

SPM were, therefore, diagnosed in 30/514 (5.8%) patients (17/309 males, 5.5%; 13/205, 6.2%, females). The estimated 7-year cumulative incidence of SPM was 6.3% and 8.5% in males and females, respectively (Figure 1A). In these patients, the median time from CML diagnosis to SPM diagnosis was 34 (range, 3-80) months, and the median age at SPM diagnosis was 65 (range, 38-79) years. The most frequent SPM were colon cancers (n=4), prostate cancers (n=3), breast cancers (n=2), central nervous system cancers (n=2), pancreatic cancers (n=2), liver cancers (n=2),

non-Hodgkin lymphomas (n=2), and thyroid cancers (n=2). No difference in the incidence of SPM was observed between patients initially treated with high-dose imatinib (800 mg) versus standard-dose imatinib (400 mg): 9/126 (7.1%) versus 21/388 (5.4%), respectively (P=0.51); moreover, no patient with SPM received treatment with second-generation tyrosine kinase inhibitors or underwent allogeneic stem cell transplantation. At the time of SPM diagnosis, all patients were in complete hematologic remission, 28/30 were in complete cytogenetic remission

Table 3. Details of malignancies observed during the follow-up of patients with chronic myeloid leukemia.

ID Age at Sex CML Dx (years)

Prior HU

OM

Age at Time OM Dx CML Dx (years) OM Dx (months)

Time IM OM Dx (months)

IM Dose (mg)

CCyR

OM therapy

OM Follow-up PM outcome CML Dx â&#x20AC;&#x201C; Last contact/death (months)

1

53

F

No

Bile duct cancer

56

37

36

400

Yes

S

Death

44

2

35

F

No

Breast cancer

38

25

25

400

Yes

S, CHT

CR

73

3

50

F

Yes

Breast cancer

54

51

51

800

Yes

S, CHT

CR

84

4

53

F

No

Colon cancer

53

3

3

400

No

No/palliation

Death

4

5

60

F

Yes

Colon cancer

60

3

3

800

No

S, CHT

Death

27

6

63

F

Yes

Colon cancer

65

22

21

800

Yes

S

Death

27

7

64

M

Yes

Colon cancer

68

44

43

400

Yes

No/palliation

Death

50

8

77

M

No

DLBCL

79

10

9

400

Yes

S, RT, CHT

CR

74

9

54

F

No

Endometrial cancer

61

78

76

400

Yes

S

CR

78

10

56

M

No

Esophageal cancer

58

23

23

400

Yes

RT, CHT

Death

37

11

61

M

Yes

Gastric cancer

65

48

44

400

Yes

S

CR

91

12

77

F

No

Hepatocarcinoma

79

13

13

400

Yes

13

69

M

No

Hepatocarcinoma

72

34

34

400

Yes

CHT

Death

14

60

M

No

Glioblastoma

63

26

26

400

Yes

S

Death

28

15

61

M

No

High-grade glioma

64

33

33

400

Yes

S, RT, CHT

Death

39

No/palliation Stable *

Age at pM Dx (years)

Therapy of pM

Status Time of pM pM Dx â&#x20AC;&#x201C; at CML Dx relapse (months)

73 41

16

64

F

No

Leiomyosarcoma

65

7

7

800

Yes

No/palliation

Death

8

17

57

M

No

Lung cancer

60

36

36

400

Yes

RT, CHT

Death

41

18

64

M

Yes Mantle cell lymphoma 65

6

5

400

Yes

CHT

Death

11

19

65

F

Yes

63

61

800

Yes

S, CHT

Death

85

20

70

M

No

Pancreatic cancer

72

11

10

400

Yes

No/palliation

Death

15

21

60

F

No

Pancreatic cancer

67

80

80

800

Yes

No/palliation

Death

82

22

75

M

No

Prostatic cancer

77

10

9

400

Yes

S

CR

72

23

69

M

No

Prostatic cancer

74

42

42

800

Yes

S

CR

87

24

54

M

No

Prostatic cancer

60

62

61

400

Yes

S

CR

87

25

52

F

No

Rectal cancer

58

71

70

400

Yes

S

CR

84

26

58

M

No

Small bowel cancer

65

79

78

400

Yes

S

CR

79

27

35

M

Yes

Testis cancer

38

35

34

800

Yes

S

CR

84

28

39

F

Yes

Thyroid cancer

42

30

29

400

Yes

S, RT

CR

80

29

50

M

No

Thyroid cancer

52

25

25

800

Yes

S, RT

CR

79

30

72

M

No

Urethral cancer

78

61

61

400

Yes

RT, CHT

Death

65

31

57

M

Yes

Bladder cancer

59

27

25

400

Yes

S

CR

87 Bladder cancer

51

S

CR

105

32

72

M

No

Bladder cancer

79

83

82

800

Yes

No/palliation

Death

86 Bladder cancer

70

CHT, RT

CR

117

33

54

F

Yes

Breast cancer

59

59

58

800

Yes

S, CHT

CR

82 Breast cancer

46

S, RT, CHT, H CR

156

34

79

M

Yes

Multiple myeloma

81

18

15

400

Yes

CHT

Death

46

78

No

Stable

36

35

74

M

Yes

Renal cancer

77

25

24

400

Yes

RT, CHT

Death

48 Renal cancer

74

S

CR

28

Ovarian cancer

71

MGUS

* The patient subsequently died from progression of CML to blast phase. ID: identification; CML: chronic myeloid leukemia; DX: diagnosis; OM: other malignancy; pM: prior malignancy; DLBCL: diffuse large B-cell lymphoma; MCL: mantle cell lymphoma; MGUS: monoclonal gammopathy of undetermined significance; HU: hydroxyurea; IM: imatinib; S: surgery; CHT: chemotherapy; RT: radiotherapy; H: hormone therapy; CR: complete remission; CCyR: complete cytogenetic response.

haematologica | 2017; 102(9)

1533


G. Gugliotta et al.

and 27/30 had a major molecular response. Other malignancies were the second cause of death in this cohort (19/514; 3.7%), while death from progression of CML to advanced phase was the first cause (25/514; 4.8%). As expected, patients who developed a SPM had a significantly lower overall survival rate compared to patients without a SPM (7-year overall survival: 43.6% versus 89.9%; P<0.001; Figure 1B). In detail, considering only the patients with SPM, 16/30 (53%) died because of the SPM. The median overall survival after diagnosis of the SPM was 18 months (Figure 1C) and the median age at death of patients with these malignancies was 66 (range, 53-84) years. All four patients with colon cancer died within 2 years of diagnosis (after 1, 5, 6, and 24 months).

these analyses found higher SIR for some cancers types: non-Hodgkin lymphomas in the German CML study IV;20 melanoma, kidney and endocrine tumors in the MD Anderson Cancer Center analysis.17 Unfortunately, in our study, the relatively low number and the heterogeneity of the malignancies observed precluded a proper evaluation of the SIR for different types of cancer. In contrast to these data, analysis of the epidemiological CML Swedish Registry on 868 unselected CML patients treated with tyrosine kinase inhibitors18 showed an overall increased SIR for other malignancies (1.52; 95% CI: 1.13 – 1.99); the SIR maintained a statistical significance in females (1.81; 95% CI: 1.18 – 2.66). It is worth noting that

Comparison with the general population

A

We then compared the incidence of SPM, and the related mortality, with that reported in the Italian general population, matched by sex and age (Table 4). In Italy, the standardized incidence of malignancy between 20 and 84 years of age is 7.6/1.000 and 5.2/1.000 person-years in males and females, respectively. In males, we observed 17 SPM, and the SIR was 1.06 [95% confidence interval (CI): 0.57 – 1.54]. In females, we observed 13 SPM, which resulted in a SIR of 1.61 (95% CI: 0.92 – 2.31). In Italy, the standardized mortality for malignancy between 20 and 84 years of age is 3.5/1.000 and 1.9/1.000 person-years in males and females, respectively. In our cohort, 9/309 (2.9%) males died of a SPM, and the SMR was 1.26 (95% CI: 0.53 – 1.99); 7/205 (3.4%) females died of a SPM, resulting in a statistically significantly higher SMR (2.41; 95% CI: 1.26 – 3.56).

B

Discussion The assessment of SPM risk is particularly complex: large cohorts of patients, long follow-up, accurate and comprehensive data collection, and a proper reference population are required for a good estimation of the risk. In this context, the analysis of both clinical trials and epidemiological registries, although with distinct drawbacks, provides essential information. In our cohort of 514 patients with chronic phase CML treated in clinical trials with front-line imatinib, with a median follow-up of 74 months, 30 (5.8%) patients had a SPM. A higher age at CML diagnosis was the only baseline factor significantly associated with SPM; reasonably, it justifies the higher proportion of intermediate-Sokal risk patients observed in this group. Of note, no CML-related characteristic was linked to the occurrence of SPM. The incidence of SPM among patients with chronic phase CML was not significantly increased compared to that in the age- and sex-matched Italian population. These data confirm the main findings of three large analyses of CML patients treated with tyrosine kinase inhibitors in clinical trials (German CML study IV,20 MD Anderson Cancer Center trial,17 and Novartis global database14) in which the overall incidence of SPM was similar to that of the general population. Of note, this conclusion was consistent despite the important differences of these studies regarding patients’ characteristics (including epidemiological aspects), treatments received, and follow-up. Two of 1534

C

Figure 1. Cumulative incidence of second primary malignancies and overall survival of patients with or without second primary malignancies. (A) Cumulative incidence of SPM in 514 patients (309 males; 205 females); the estimated 7-year cumulative incidence was 6.3% and 8.5% in males and females, respectively; (B) Overall survival (OS) from CML diagnosis: 7-year OS was 43.6% in patients with SPM (n=30) and 89.9% in patients without SPM (n=484); P<0.001. (C) OS from SPM diagnosis (n=30): the median OS was 18 months, and the estimated 4-year OS was 42.3%

haematologica | 2017; 102(9)


Outcome of second primary malignancies in CML

we observed a trend for an increased SIR in females (1.6; lower 95% CI: 0.92). In our study, progression of CML to advanced phase and other malignancies were the most common causes of death (4.8 and 3.7%, respectively). The death rate in patients with SPM was particularly high (53%), with a relatively short median overall survival (18 months from diagnosis of the SPM). Of note, in females, the mortality from SPM was significantly superior to that expected in the Italian age-matched female population, with a SMR of 2.41 (95% CI: 1.26 â&#x20AC;&#x201C; 3.56). Despite the limitations due to the small number of patients, some factors favoring the high mortality observed for SPM can be hypothesized. For example, the therapeutic approach to SPM in patients with CML may be less intensive (6/30 patients received only palliative care) or the biological behavior of the SPM may be more aggressive, as a consequence of CML itself, or because of imatinib. With regards to the latter point, it should be remembered that in a breast cancer model in mice, treatment with imatinib was associated with an increased malignant behavior compared to control conditons.29 Moreover, tyrosine kinase inhibitors could enhance, or facilitate, the progression of SPM through the inhibition of ABL, which is a downstream effector of the epinephrine receptors that might have a tumor-suppressor role in breast, prostate, and colorectal cancers30-32 (interestingly, all our patients with a colon cancer died) or through impairment of the immune system,6-11 potentially affecting antitumor surveillance. An in-depth evaluation of immunological mechanisms might be particularly intriguing in view of the potential use of new molecules enhancing the immune system against cancer. In conclusion, the prevalence of CML is increasing steadily and this, together with the aging of patients, means that the number of subjects at risk of developing SPM is increasing. However, the fact that an increased incidence was not detectable in the majority of the cohorts analyzed so far does not support the fear that

References 1. Apperley JF. Chronic myeloid leukaemia. Lancet. 2015;385(9976):1447-1459. 2. Pfirrmann M, Baccarani M, Saussele S, et al. Prognosis of long-term survival considering disease-specific death in patients with chronic myeloid leukemia. Leukemia. 2016;30(1):48-56. 3. Steegmann JL, Cervantes F, le Coutre P, Porkka K, Saglio G. Off-target effects of BCR-ABL1 inhibitors and their potential long-term implications in patients with chronic myeloid leukemia. Leuk Lymphoma. 2012;53(12):2351-2361. 4. Moslehi JJ, Deininger M. Tyrosine kinase Inhibitor-associated cardiovascular toxicity in chronic myeloid leukemia. J Clin Oncol. 2015;33(35):4210-4218. 5. Steegmann JL, Baccarani M, Breccia M, et al. European LeukemiaNet recommendations for the management and avoidance of adverse events of treatment in chronic myeloid leukaemia. Leukemia. 2016;30(8):1648-1671.

haematologica | 2017; 102(9)

Table 4. Second primary malignancies: comparison of incidence and mortality with those in the Italian general population, matched by sex and age*.

Sex M F

M F 3.56)

Patientyears

SPM (N)

Expected SPM

SIR

95% CI

1806.3 1204.8

17 13

16.1 8.1

1.06 1.61

(0.57; 1.54) (0.92; 2.31)

Deaths (N)

Expected deaths

SMR

95% CI

9 7

7.2 2.9

1849.4 1228.3

1.26 (0.53; 1.99) 2.41 (1.26;

* Five-year age classes were considered. We report the overall (20-84 years) ratios, rather than those for specific age subgroups, to avoid selection biases. SPM: second primary malignancies; SIR: standardized incidence ratio; SMR: standardized mortality ratio.

chronic treatment with tyrosine kinase inhibitors, in particular imatinib, cause more SPM compared to those occurring in the general population. Despite these reassuring results, large studies with a long follow-up (e.g. using data from CML registries) are warranted to properly investigate the incidence of specific types of SPM, and to fully address mortality due to SPM. This could help to improve patientsâ&#x20AC;&#x2122; management through early diagnosis of SPM and treatment optimization in conjunction with oncologists. Furthermore, the comparison of the incidence of SPM (and their outcomes) in patients treated with different tyrosine kinase inhibitors may provide important clues on the potential role of each inhibitor. Acknowledgments This study was supported by GIMEMA Onlus, BolognAIL and European LeukemiaNet (LSHC-CT-2004-503216). We thank Michela Apolinari and Miriam Fogli for the data management.

6. Cwynarski K, Laylor R, Macchiarulo E, et al. Imatinib inhibits the activation and proliferation of normal T lymphocytes in vitro. Leukemia. 2004;18(8):1332-1339. 7. Dietz AB, Souan L, Knutson GJ, Bulur PA, Litzow MR, Vuk-Pavlovic S. Imatinib mesylate inhibits T-cell proliferation in vitro and delayed-type hypersensitivity in vivo. Blood. 2004;104(4):1094-1099. 8. Gao H, Lee BN, Talpaz M, et al. Imatinib mesylate suppresses cytokine synthesis by activated CD4 T cells of patients with chronic myelogenous leukemia. Leukemia. 2005;19(11):1905-1911. 9. Steegmann JL, Moreno G, Alaez C, et al. Chronic myeloid leukemia patients resistant to or intolerant of interferon alpha and subsequently treated with imatinib show reduced immunoglobulin levels and hypogammaglobulinemia. Haematologica. 2003;88(7):762-768. 10. Appel S, Balabanov S, Brummendorf TH, Brossart P. Effects of imatinib on normal hematopoiesis and immune activation. Stem Cells. 2005;23(8):1082-1088. 11. Leder C, Ortler S, Seggewiss R, Einsele H,

12. 13.

14.

15.

16.

Wiendl H. Modulation of T-effector function by imatinib at the level of cytokine secretion. Exp Hematol. 2007;35(8):12661271. EMA. Glivec: EPAR - product information. 2015. Roy L, Guilhot J, Martineau G, Larchee R, Guilhot F. Unexpected occurrence of second malignancies in patients treated with interferon followed by imatinib mesylate for chronic myelogenous leukemia. Leukemia. 2005;19(9):1689-1692. Pilot PR, Sablinska K, Owen S, Hatfield A. Epidemiological analysis of second primary malignancies in more than 9500 patients treated with imatinib. Leukemia. 2006; 20(1):148. Roy L, Guilhot J, Martineau G, Guilhot F. Reply to 'Epidemiological analysis of second primary malignancies in more than 9500 patients treated with imatinib' by Pilot et al. Leukemia. 2006;20(1):149. Voglova J, Muzik J, Faber E, et al. Incidence of second malignancies during treatment of chronic myeloid leukemia with tyrosine kinase inhibitors in the Czech Republic and

1535


G. Gugliotta et al. Slovakia. Neoplasma. 2011;58(3):256-262. 17. Verma D, Kantarjian H, Strom SS, et al. Malignancies occurring during therapy with tyrosine kinase inhibitors (TKIs) for chronic myeloid leukemia (CML) and other hematologic malignancies. Blood. 2011;118(16):4353-4358. 18. Gunnarsson N, Stenke L, Hoglund M, et al. Second malignancies following treatment of chronic myeloid leukaemia in the tyrosine kinase inhibitor era. Br J Haematol. 2015;169(5):683-688. 19. 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. 20. Miranda MB, Lauseker M, Kraus MP, et al. Secondary malignancies in chronic myeloid leukemia patients after imatinib-based treatment: long-term observation in CML Study IV. Leukemia. 2016;30(6):1255-1262. 21. Shah BK, Ghimire KB. Second primary malignancies in chronic myeloid leukemia. Indian J Hematol Blood Transfus. 2014;30(4):236-240. 22. Curtis RE, Freedman DM, Ron E, et al. New malignancies among cancer survivors:

1536

23.

24.

25.

26.

27.

SEER cancer registries, 1973-2000. Bethesda, MD: National Cancer Institute, 2006. Rebora P, Czene K, Antolini L, Gambacorti Passerini C, Reilly M, Valsecchi MG. Are chronic myeloid leukemia patients more at risk for second malignancies? A population-based study. Am J Epidemiol. 2010;172(9):1028-1033. Frederiksen H, Farkas DK, Christiansen CF, Hasselbalch HC, Sorensen HT. Chronic myeloproliferative neoplasms and subsequent cancer risk: a Danish populationbased cohort study. Blood. 2011;118(25): 6515-6520. Gugliotta G, Castagnetti F, Palandri F, et al. Frontline imatinib treatment of chronic myeloid leukemia: no impact of age on outcome, a survey by the GIMEMA CML Working Party. Blood. 2011;117(21):55915599. Castagnetti F, Gugliotta G, Breccia M, et al. Long-term outcome of chronic myeloid leukemia patients treated frontline with imatinib. Leukemia. 2015;29(9):1823-1831. Castagnetti F, Palandri F, Amabile M, et al. Results of high-dose imatinib mesylate in intermediate Sokal risk chronic myeloid

28.

29.

30.

31.

32.

leukemia patients in early chronic phase: a phase 2 trial of the GIMEMA CML Working Party. Blood. 2009;113(15):34283434. Baccarani M, Rosti G, Castagnetti F, et al. Comparison of imatinib 400 mg and 800 mg daily in the front-line treatment of highrisk, Philadelphia-positive chronic myeloid leukemia: a European LeukemiaNet Study. Blood. 2009;113(19):4497-4504. Rappa G, Anzanello F, Lorico A. Imatinib mesylate enhances the malignant behavior of human breast carcinoma cells. Cancer Chemother Pharmacol. 2011;67(4):919926. Noren NK, Foos G, Hauser CA, Pasquale EB. The EphB4 receptor suppresses breast cancer cell tumorigenicity through an AblCrk pathway. Nat Cell Biol. 2006;8(8):815825. Dopeso H, Mateo-Lozano S, Mazzolini R, et al. The receptor tyrosine kinase EPHB4 has tumor suppressor activities in intestinal tumorigenesis. Cancer Res. 2009;69(18): 7430-7438. Pasquale EB. Eph receptors and ephrins in cancer: bidirectional signalling and beyond. Nat Rev Cancer. 2010;10(3):165-180.

haematologica | 2017; 102(9)


ARTICLE

Acute Myeloid Leukemia

High-throughput profiling of signaling networks identifies mechanism-based combination therapy to eliminate microenvironmental resistance in acute myeloid leukemia

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Zhihong Zeng,1* Wenbin Liu,2* Twee Tsao,1 YiHua Qiu,1 Yang Zhao,2 Ismael Samudio,3 Dos D. Sarbassov,4 Steven M. Kornblau,1,5 Keith A. Baggerly,2 Hagop M. Kantarjian,1 Marina Konopleva1,5 and Michael Andreeff1,5 *ZZ and WL contributed equally to this work

Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 2Department of Bioinformatics and Computational Biology, The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 3Terry Fox Laboratory, British Columbia Cancer Agency, Vancouver, BC, Canada; 4Department of Molecular and Cellular Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, USA and 5Department of Stem Cell Transplantation and Cellular Therapy, The University of Texas MD Anderson Cancer Center, Houston, TX, USA 1

Haematologica 2017 Volume 102(9):1537-1548

ABSTRACT

T

he bone marrow microenvironment is known to provide a survival advantage to residual acute myeloid leukemia cells, possibly contributing to disease recurrence. The mechanisms by which stroma in the microenvironment regulates leukemia survival remain largely unknown. Using reverse-phase protein array technology, we profiled 53 key protein molecules in 11 signaling pathways in 20 primary acute myeloid leukemia samples and two cell lines, aiming to understand stroma-mediated signaling modulation in response to the targeted agents temsirolimus (MTOR), ABT737 (BCL2/BCL-XL), and Nutlin-3a (MDM2), and to identify the effective combination therapy targeting acute myeloid leukemia in the context of the leukemia microenvironment. Stroma reprogrammed signaling networks and modified the sensitivity of acute myeloid leukemia samples to all three targeted inhibitors. Stroma activated AKT at Ser473 in the majority of samples treated with single-agent ABT737 or Nutlin-3a. This survival mechanism was partially abrogated by concomitant treatment with temsirolimus plus ABT737 or Nutlin-3a. Mapping the signaling networks revealed that combinations of two inhibitors increased the number of affected proteins in the targeted pathways and in multiple parallel signaling, translating into facilitated cell death. These results demonstrated that a mechanism-based selection of combined inhibitors can be used to guide clinical drug selection and tailor treatment regimens to eliminate microenvironment-mediated resistance in acute myeloid leukemia.

Correspondence: mkonople@mdanderson.org or mandreef@mdanderson.org Received: December 15, 2016. Accepted: June 27, 2017 Pre-published: June 28, 2017 doi:10.3324/haematol.2016.162230 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1537 Š2017 Ferrata Storti Foundation

Introduction Acute myeloid leukemia (AML) has a high initial treatment response rate, associated with the elimination of bulk leukemic cells, and an almost inevitable high relapse rate.1,2 Recent studies indicate that stroma in the bone marrow (BM) microenvironment protects resident leukemic cells and plays a key role in AML relapse.3-7 Activation of the PI3K/AKT/MTOR pathway, upregulation of the antiapoptotic BCL2 family and MDM2/P53 signaling have been identified in patients with disease recurrence8-13 and associated with stroma-mediated AML survival.14-18 Strategies for targeting the key molecules in these pathways have been developed to improve therapeutic efficacy in patients with AML.19 Temsirolimus, ABT737, and Nutlin-3a are selective small-molecule inhibitors that haematologica | 2017; 102(9)

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.

1537


Z. Zeng et al.

affect MTOR, BCL2/BCL-XL and MDM2/P53 signaling, respectively. Temsirolimus, a rapamycin analog and cytostatic inhibitor, prevents leukemic cell proliferation by blocking the formation of MTOR complex 1 (MTORC1) and MTOR complex 2 (MTORC2) and sequentially inactivating AKT/MTOR downstream signaling.20,21 ABT737, a selective small-molecule BCL2/BCL-XL antagonist, exerts its proapoptotic function by preventing BCL2 family proteins from sequestering to activate BH3-only proteins.22,23 Nutlin-3a, a small-molecule MDM2 inhibitor, binds to MDM2 in the P53-binding pocket and activates P53-mediated apoptosis.24,25 The efficacy of these inhibitors, both as single agents and in combination, has been evaluated in preclinical studies of hematological malignancy.23,26-29 Although high potency was reported in these studies, only a modest therapeutic response was observed in clinical trials.30-32 This inconsistency between preclinical results and clinical outcomes is attributable to two factors. First, most of the preclinical studies were performed under in vitro monolayer conditions that did not account for the possible influence of the microenvironment on the effectiveness of the targeted inhibitors. Second, the on-target effects of temsirolimus, ABT737, and Nutlin-3a were frequently examined only for their target-specific pathways PI3K/AKT/MTOR, BCL2/BCL-XL, and MDM2/P53 without considering parallel signaling. This focus precluded assessment of survival mechanisms mediated by compensatory signaling networks. Thus, the microenvironment-modulated signaling networks of single and combined targeted inhibitors require further investigation. Results of such studies will contribute to the development of effective treatments to target microenvironment-mediated AML survival. Reverse-phase protein array (RPPA), a high-throughput functional proteomic technology, facilitates broad and simultaneous profiling of therapeutically relevant signaling networks. This technique has been successfully used to identify signaling pathway abnormalities, pharmacodynamic markers, and proteins associated with therapeutic resistance in various cancers, including leukemia.33 In the study herein, using RPPA technology, we profiled 53 key molecules in 11 signaling pathways in 20 primary AML samples and two AML cell lines. Our goals were to understand the role of microenvironment-mediated signaling in AML survival by comparing the signaling network alterations triggered by temsirolimus, ABT737, and Nutlin-3a in samples cultured alone and co-cultured with stroma, a condition mimicking the BM microenvironment, and to identify effective combination strategies targeting stromaregulated AML. Our results indicate that stroma-mediated signaling is specific to each targeted inhibitor. By mapping the network alterations triggered by the combination of temsirolimus plus ABT737 or Nutlin-3a, we revealed the mechanisms by which combinatorial treatment abrogated stroma-mediated survival and facilitated leukemic cell death. Our findings provide a clinically relevant approach for selecting mechanism-based therapy to effectively eliminate microenvironment-protected AML.

Methods Materials, cell lines, and patient samples Information about the materials and cell lines used in this study is provided in the Online Supplementary Materials and 1538

Methods. Clinical information about the primary AML samples is provided in the Online Supplementary Table S1. All samples were collected during routine diagnostic procedures in accordance with protocols approved by the Institutional Review Board of The University of Texas MD Anderson Cancer Center. Informed consent was obtained in accordance with the Declaration of Helsinki.

Co-culture of leukemic cells and stromal cells MS-5 is an established murine stromal cell line that has been routinely used for long-term support of primary hematopoietic cells and demonstrated high stability.14,34 Unlike human BMderived mesenchymal stromal cells (MSCs), the function of MS-5 is less affected by cell passages. We chose MS-5 to construct a model of the BM microenvironment for the RPPA experiment and used normal human MSCs to verify RPPA analyses. Stromal cells were plated in 10% fetal bovine serum (FBS) αminimum essential medium and cultured overnight. The culture medium was then removed, and leukemic cells were seeded on top of stromal cells at a ratio of 10:1 (leukemic:stromal cells) in 10% FBS Roswell Park Memorial Institute (RPMI) 1640 medium. Leukemic and stromal cells were co-cultured for three hours before being exposed to single or combined agents at the concentrations specified below.

Cell treatment Leukemic cells alone or in co-culture with stromal cells were treated with the following concentrations of a single agent or a combination of two agents: temsirolimus (2.42 mM),21 ABT737 (50 nM, except for OCI-AML3 cells treated at 0.25 mM in RPPA, 0.1 or 0.25 mM in immunoblotting),23 and Nutlin-3a (5 mM).25 RPPA and immunoblotting analyses were performed on cells that had been treated for 24 hours; apoptosis induction was measured on cells treated for 72 hours.

Apoptosis assay Cell apoptosis was analyzed by flow cytometry of annexin V (Roche Diagnostics, Indianapolis, IN, USA) and propidium iodide (PI) (Sigma Chemical, St. Louis, MO, USA) positivity on a gated CD45+ population (anti-human CD45, BD Pharmingen, San Diego, CA, USA). To examine and compare inhibitor-specific apoptosis, we calculated % of specific apoptosis as previously described: % specific apoptosis = (tested – control) / (100 – control) × 100, where “tested” is the percentage of annexin V/PIpositivity in treated cells, and “control” is the percentage of annexin V/PI-positivity in untreated cells (spontaneous apoptosis).25

Immunoblotting Protein expression in treated and untreated cells was determined by immunoblotting. Protein signals were detected using an Odyssey Infrared Imaging System and quantified using Odyssey software version 3.0 (LI-COR Biosciences, Lincoln, NE, USA).

RPPA RPPA was performed on two AML cell lines, OCI-AML3 and U937, and 20 primary AML samples with high blast counts (Online Supplementary Table S1). Cells cultured alone or co-cultured with MS-5 were harvested after 24 hours of treatment. Co-cultured leukemic cells were collected according to a previously published protocol.35 To exclude the possibility of the leukemic cells being contaminated by MS-5, flow cytometry was used to verify the absence of a CD90+ (anti-mouse CD90, BD Pharmingen) population among the leukemic cells. The colhaematologica | 2017; 102(9)


Targeting microenvironment-protected AML

lected cells were then lysed and subjected to RPPA using previously described and validated methods.33,36 Raw signal intensities obtained from RPPA were processed with SuperCurve to determine relative protein concentrations, and the results were further normalized to adjust for loading bias by median-centering each marker and each sample.37,38

Statistical analysis We used a two-sided â&#x20AC;&#x153;fold-change-filteredâ&#x20AC;? binomial test and a two-tailed Student's t-test to identify the distinct protein and pathway alterations and apoptosis induction triggered by treatment and mediated by stroma. The detailed statistical analyses are described in the Online Supplementary Materials and Methods.

A

B

C

D

Figure 1. Stroma altered target proteins and the sensitivity of AML to temsirolimus. The pathways affected by temsirolimus treatment are circled in the pie charts on the left. The bar graphs display proteins whose expression was significantly altered in (A) monocultured samples treated with temsirolimus versus untreated control (n = 22) and (B) co-cultured samples treated with temsirolimus versus untreated control (n = 22). The numbers on the left are the numbers of the pathways (as indicated on the pie chart and in the Online Supplementary Figure S1). The black area indicates the percentage of affected AML samples ([number of affected samples / total number of samples] Ă&#x2014; 100) that exhibited downregulation of protein expression. The white area indicates the percentage of samples with upregulation of protein expression. The gray area represents the percentage of samples having no significant changes in protein expression. (C) Bar graph displays temsirolimusinduced specific apoptosis of samples cultured alone and co-cultured with stroma. The sample groupings and the statistical calculation of apoptosis for each defined group are described in the Online Supplementary Materials and Methods. (D) Box and whisker plots display differences in protein expression between untreated samples in groups I and III with and without stroma (top panel) and protein expression at the baseline level of untreated samples in groups I and III in co-culture (bottom panel). Whiskers indicate the range from minimum to maximum values. The line in the middle of the box is plotted at the median.

haematologica | 2017; 102(9)

1539


Z. Zeng et al.

Results RPPA profiling of key molecules in signaling networks To investigate treatment-mediated signaling networks, we profiled 53 key proteins in 11 signaling pathways: (1)

PI3K/AKT/MTOR signaling, (2) AKT/MTOR major downstream signaling, (3) MEK/ERK signaling, (4) STAT3 signaling, (5) BCL2 protein family, (6) WNT CATENIN signaling, (7) P53 family, (8) IAP family, (9) cell cycle regulation, (10) PP2A phosphatase family, and (11) MYC signal-

A

B

D C

Figure 2. Stroma altered target proteins and the sensitivity of AML to ABT737. The pathways affected by ABT737 treatment are circled in the pie charts on the left. The bar graphs display proteins whose expression was significantly altered in (A) monocultured samples treated with ABT737 versus untreated control (n = 20) and (B) co-cultured samples treated with ABT737 versus untreated control (n = 20). The numbers on the left are the numbers of the pathways (as indicated on the pie chart and in the Online Supplementary Figure S1). The calculation of the black, white, and gray area and the color key on the bar graphs are described in the legend for Figure 1. (C) Bar graph displays ABT737-induced specific apoptosis for samples cultured alone and co-cultured with stroma. The sample groupings and the statistical calculation of apoptosis are described in the Online Supplementary Materials and Methods. (D) Box and whisker plots display differences in protein expression between untreated samples in groups I and III with and without stroma (top panel) and protein expression at the baseline level of untreated samples in groups I and III in co-culture (bottom panel). Whiskers indicate the range from minimum to maximum values. The line in the middle of the box is plotted at the median.

1540

haematologica | 2017; 102(9)


Targeting microenvironment-protected AML

ing (Online Supplementary Figure S1; Online Supplementary Table S2), in two AML cell lines, OCI-AML3 and U937, and 20 primary AML samples. Treatment-induced apoptosis with or without stroma was measured for 15 of the 20 primary samples.

Stroma altered target proteins and the sensitivity of AML to temsirolimus In samples cultured alone, the MTOR inhibitor temsirolimus significantly affected five proteins in two signaling pathways; four in the PI3K/AKT/MTOR signaling and

A

B

C

D

Figure 3. Stroma altered target proteins and the sensitivity of AML to Nutlin-3a. The pathways affected by Nutlin-3a treatment are circled in the pie charts on the left. The bar graphs display proteins whose expression was significantly altered in (A) monocultured samples treated with Nutlin-3a versus untreated control (n = 22) and (B) co-cultured samples treated with Nutlin-3a versus untreated control (n = 22). The numbers on the left are the numbers of the pathways (as indicated on the pie chart and in the Online Supplementary Figure S1). The calculation of the black, white, and gray area and the color key on the bar graphs are described in the legend for Figure 1. (C) Bar graph displays Nutlin-3a-induced specific apoptosis for samples cultured alone and co-cultured with stroma. The sample groupings and the statistical calculation of apoptosis are described in the Online Supplementary Materials and Methods. (D) Box and whisker plots display differences in protein expression between untreated samples in groups I and III with and without stroma. Whiskers indicate the range from minimum to maximum values. The line in the middle of the box is plotted at the median.

haematologica | 2017; 102(9)

1541


Z. Zeng et al.

one in the STAT3 pathway (Figure 1A; Online Supplementary Figure S2A). Temsirolimus treatment inhibited p-AKT (Ser473) and p-S6 (Ser240/244), upregulated AKT3, and decreased the expression of total AKT and STAT3. In co-cultured samples, temsirolimus treatment affected five proteins in four pathways: two in the PI3K/AKT/MTOR pathway, one in the AKT/MTOR major downstream signaling, one in the P53 family, and one in the STAT3 signaling (Figure 1B; Online Supplementary Figure S2B). Temsirolimus treatment inhibited p-S6 at Ser235/236 and Ser240/244 in the PI3K/AKT/MTOR pathway, upregulated p-BAD (Ser136) in the AKT/MTOR major downstream signaling, and suppressed STAT3 expression. Interestingly, under stromal co-culture, temsirolimus treatment triggered upregulation of MDM2, a ubiquitin ligase responsible for P53 degradation.

In stromal co-culture, temsirolimus-induced apoptosis was decreased in four samples, increased in six samples, and unchanged in five samples (Figure 1C). RPPA analysis of untreated samples showed that the stroma-mediated upregulation of p-S6 (Ser240/244) and downregulation of PRAS40 was significantly greater in temsirolimus-insensitive samples than in temsirolimus-sensitive samples; in addition, temsirolimus-insensitive samples had a lower baseline expression of PTEN and BCL2 than did the temsirolimus-sensitive samples in co-culture (Figure 1D). Together, these results suggest that modulation of the sensitivity of AML cells to temsirolimus was the result of stroma-regulated alterations in signaling networks.

Stroma altered target proteins and the sensitivity of AML to ABT737 In samples cultured alone, treatment with the BCL2/BCL-XL inhibitor ABT737 significantly affected

A

B

Figure 4. Combined blockade of PI3K/AKT/MTOR and BCL2 signaling with temsirolimus and ABT737 demonstrated higher efficacy in AML under co-culture. (A) The pathways affected by combination treatment with temsirolimus and ABT737 are circled in the pie chart on the left. The bar graph displays proteins whose expression was significantly altered in samples treated with the combination of temsirolimus and ABT737 (n = 18) versus the matched samples without treatment in coculture. The numbers on the left are the numbers of the pathways (as indicated on the pie chart and in the Online Supplementary Figure S1). The calculation of the black, white, and gray area and the color key on the bar graph are described in the legend for Figure 1. (B) The bar graph on the left displays the specific apoptosis for each co-cultured sample treated with temsirolimus, ABT737, and the combination of temsirolimus and ABT737. The sample groupings and the statistical calculation of apoptosis are described in the Online Supplementary Materials and Methods. The bar graph on the right displays the specific apoptosis in each group. Values are presented as mean Âą standard error of the mean.

1542

haematologica | 2017; 102(9)


Targeting microenvironment-protected AML

three proteins in two pathways; two in the PI3K/AKT/MTOR pathway and one in the AKT/MTOR major downstream signaling (Figure 2A; Online Supplementary Figure S3A). ABT737 treatment suppressed the expression of AKT and 4EBP1 in the PI3K/AKT/MTOR pathway and decreased BAD expression in the AKT/MTOR major downstream signaling. In stromal co-culture, ABT737 treatment significantly affected three proteins in two pathways; one in the PI3K/AKT/MTOR pathway and two in the AKT/MTOR major downstream signaling (Figure 2B; Online Supplementary Figure S3B). ABT737 treatment, contrary to its effects without stroma, upregulated p-AKT (Ser473) in the PI3K/AKT/MTOR pathway. It also increased the expression of FOXO3A and inhibited BAD in the AKT/MTOR major downstream signaling. ABT737-induced apoptosis was decreased in eight and increased in six samples under co-culture (Figure 2C). RPPA analysis of untreated samples indicated that stromamediated upregulation of p-4EBP1 (Thr37/46) and PP2A was significantly greater in the eight insensitive samples than in the ABT737-sensitive samples; in addition, the insensitive samples had higher baseline expression of GSK3, STAT3, XIAP, CIAP, and PP2A and lower level of pBAD (Ser136) than did the ABT737-sensitive samples in co-culture (Figure 2D). These results suggest that stroma reprogrammed signaling networks, thus altering the sensitivity of AML cells to ABT737.

Stroma altered target proteins and the sensitivity of AML to Nutlin-3a In samples cultured alone, treatment with the MDM2 inhibitor Nutlin-3a significantly inhibited p-4EBP1 (Thr37/46) in the PI3K/AKT/MTOR signaling (Figure 3A; Online Supplementary Figure S4A). In samples co-cultured with stroma, Nutlin-3a significantly affected two proteins in two signaling pathways; treatment upregulated p-AKT (Ser473) in the PI3K/AKT/MTOR signaling and increased MDM2 expression in the P53 family (Figure 3B; Online Supplementary Figure S4B). Nutlin-3a-induced apoptosis was decreased in eight and increased in five samples under co-culture (Figure 3C). RPPA analysis of untreated samples demonstrated that the stroma-mediated downregulation of TSC2 was greater in the eight samples that were insensitive to Nutlin-3a than in the sensitive samples (Figure 3D). However, baseline protein expression between co-cultured Nutlin-3a-sensitive and insensitive samples were not significantly different.

Combined blockade of PI3K/AKT/MTOR and BCL2 signaling with temsirolimus and ABT737 demonstrated higher efficacy in AML under co-culture Stroma-mediated AKT activation in samples treated with ABT737 indicates that PI3K/AKT/MTOR signaling is the survival mechanism triggered by BCL2/BCL-XL inhibition. This finding prompted us to evaluate the therapeutic effect of combined inhibition of PI3K/AKT/MTOR and BCL2 in AML samples co-cultured with stroma. In co-culture, treatment with temsirolimus and ABT737 targeted more proteins than did treatment with temsirolimus or ABT737 alone. The two-inhibitor combination significantly affected 11 proteins in eight signaling pathways: three proteins in PI3K/AKT/MTOR signaling, haematologica | 2017; 102(9)

two in the AKT/MTOR major downstream pathway, one in the IAP family, one in the BCL2 family, one in the P53 family, one in the WNT CATENIN pathway, one in cell cycle regulation, and one in STAT3 signaling (Figure 4A; Online Supplementary Figure S5A). In the PI3K/AKT/MTOR pathway, combination treatment suppressed the expression of PDK1 and 4EBP1 and upregulated p-PRAS40 (Thr246). In the AKT/MTOR major downstream signaling, this combination upregulated FOXO3A and inhibited BAD expression. ABT737 plus temsirolimus decreased the expression of STAT3 and cell cycle protein P27 and increased the expression of P53 and BCL2 family protein BIM. This co-treatment also upregulated CTNNB1 in the WNT CATENIN signaling and increased SURVIVIN expression in the IAP pathway. The combination treatment significantly restored drug sensitivity and facilitated cell death in 12 of 15 co-cultured samples (Figure 4B). Among the 12 samples, combination increased the sensitivity of three samples that were resistant to temsirolimus (samples 9, 13, 17) and two samples that were insensitive to ABT737 (10, 14). Mechanistically, the co-treatment repressed the ABT737-upregulated pAKT (Ser473) in four samples (4, 7, 8, 11) and decreased pAKT expression that had been unaffected by ABT737 in two samples (2, 9). The combination did not further enhance apoptosis in three samples (1, 3, 5) that had various levels of ABT737-induced p-AKT expression and were highly sensitive to ABT737 single-agent treatment (Online Supplementary Figures S3B and S5B). Taken together, these results indicate that co-inhibition of PI3K/AKT/MTOR and BCL2/BCL-XL overcomes stroma-mediated AML survival, resulting in increased therapeutic efficacy against most AMLs examined.

Combined blockade of PI3K/AKT/MTOR and MDM2 signaling with temsirolimus and Nutlin-3a demonstrated higher efficacy in AML under co-culture Stroma-mediated AKT activation in samples treated with Nutlin-3a indicates that PI3K/AKT/MTOR signaling is the survival mechanism triggered by MDM2 inhibition. This finding prompted us to evaluate the therapeutic effect of combined inhibition of PI3K/AKT/MTOR and MDM2 in AML samples co-cultured with stroma. Treatment with temsirolimus and Nutlin-3a affected 11 proteins in five pathways in co-cultured samples: six in the PI3K/AKT/MTOR signaling, one in the AKT/MTOR major downstream pathway, two in the P53 family, one in the WNT CATENIN pathway, and one in cell cycle regulation (Figure 5A; Online Supplementary Figure S6A). In the PI3K/AKT/MTOR pathway, the two-inhibitor combination downregulated p-S6 at Ser235/236 and Ser240/244, decreased the expression of AKT, MTOR, and 4EBP1, and upregulated p-PRAS40 (Thr246). This combination also decreased BAD expression in the AKT/MTOR major downstream pathway and upregulated the cell cycle protein P21. Furthermore, it increased the expression of MDM2 and P53 in the P53 family. Finally, this co-treatment resulted in upregulation of p-CTNNB1 (Ser33/37/Thr41) in the WNT CATENIN signaling. The co-treatment of temsirolimus and Nutlin-3a significantly facilitated cell death in seven of 15 co-cultured samples (Figure 5B). It restored the sensitivity of three samples that were insensitive to temsirolimus (samples 2, 5, 12) and four samples that were either resistant (5, 11) or insensitive to Nutlin-3a (12, 14). Among these seven samples, p1543


Z. Zeng et al.

AKT upregulation triggered by Nutlin-3a and inhibited by the co-treatment of temsirolimus and Nutlin-3a was observed in four samples (5, 7, 11, 12). The combination had no effect or a reduced effect in eight samples. Nutlin3a-upregulated p-AKT was found in three of the eight samples (1, 8, 17), suggesting that PI3K/AKT/MTOR-independent survival signaling played a role in treatment sensitivity in the remaining five samples (Online Supplementary Figures S4B and S6B). Together, our results suggest that combined inhibition of PI3K/AKT/MTOR and MDM2 with temsirolimus and Nutlin-3a can restore drug sensitivity in some co-cultured AML samples. However, the overall efficacy of this combination was inferior to that of co-inhibition of PI3K/AKT/MTOR and BCL2/BCL-XL with temsirolimus and ABT737.

RPPA validation To validate the RPPA analysis, we performed separate experiments treating OCI-AML3 and U937 cells co-cul-

tured with a healthy BM-derived MSC. Treated cells were examined via conventional immunoblotting (Figure 6A). Results demonstrated that temsirolimus inhibited p-AKT (Ser473) and p-S6 (Ser240/244) in co-cultured OCI-AML3 and U937 cells. Temsirolimus treatment also inhibited p4EBP1 (Thr37/46) in OCI-AML3 cells but upregulated it in U937 cells. Nutlin-3a treatment resulted in MDM2 upregulation and p-S6 (Ser240/244) inhibition in OCI-AML3 cells, p-AKT (Ser473) upregulation in U937 cells and p4EBP1 (Thr37/46) inhibition in both OCI-AML3 and U937 cells. ABT737 treatment inhibited p-S6 (Ser240/244) in OCI-AML3 cells and upregulated p-AKT (Ser473) in U937 cells. It modestly affected MDM2 in OCI-AML3, and 4EBP1 in both OCI-AML3 and U937 cells. Treatment with a combination of temsirolimus and ABT737 or Nutlin-3a partially repressed MDM2 upregulation in OCI-AML3, abrogated p-AKT upregulation triggered by ABT737 and Nutlin-3a in U937, and synergistically inhibited p-AKT (Ser473) and p-S6 (Ser240/244) in both cell lines. The immunoblotting findings that single-agent treatment

A

B

Figure 5. Combined blockade of PI3K/AKT/MTOR and MDM2 signaling with temsirolimus and Nutlin-3a demonstrated higher efficacy in AML under co-culture. (A) The pathways affected by combination treatment with temsirolimus and Nutlin-3a are circled in the pie chart. The bar graph displays proteins whose expression was significantly altered in samples treated with the combination of temsirolimus and Nutlin-3a (n = 21) versus the matched samples without treatment in co-culture. The numbers on the left are the numbers of the pathways (as indicated on the pie chart on the left and in the Online Supplementary Figure S1). The calculation of the black, white, and gray area and the color key on the bar graph are described in the legend for Figure 1. (B) The bar graph on the left displays the specific apoptosis for each co-cultured sample treated with temsirolimus, Nutlin-3a, and the combination of temsirolimus and Nutlin-3a. The sample groupings and the statistical calculation of apoptosis are described in the Online Supplementary Materials and Methods. The bar graph on the right displays the specific apoptosis in each group. Values are presented as mean Âą standard error of the mean.

1544

haematologica | 2017; 102(9)


Targeting microenvironment-protected AML

induced up- or downregulation of p-AKT, p-S6, p-4EBP1, and MDM2 were generally consistent with the data obtained by RPPA in both cell lines co-cultured with MS5 (Online Supplementary Figure S7A). Co-treatment enhanced cell death, and growth inhibition was confirmed in the co-cultured OCI-AML3 cells (Figure 6B). In addition, we conducted studies juxtaposing the effect of treatment on OCI-AML3 co-cultured with MS-5 and different healthy BM MSCs. The level of protection against treatment-induced cell death provided by various MSCs and MS-5 in OCI-AML3 cells was very similar (Online Supplementary Figure S7B). These results indicate that combination treatment with temsirolimus plus ABT737 or Nutlin-3a was more effective than single-agent treatment under conditions mimicking the BM microenvironment. To confirm the RPPA results in primary samples, we compared p-AKT (Ser473) detected via immunoblotting and RPPA in the same cell lysates harvested from primary samples treated with single-agent temsirolimus, ABT737, or Nutlin-3a in co-culture. Similar results of inhibitor-triggered p-AKT up- or downregulation were observed in all seven samples (Figure 6C and Online Supplementary Figure S7C). Taken together, the overall similarities in the results obtained using RPPA and conventional immunoblotting confirmed the reliability of RPPA technology and provided a second line of evidence supporting the conclusions of the RPPA analysis.

AML heterogeneity and responses under stromal co-culture Given the well-known heterogeneity of AML, we analyzed the responses to the targeted inhibitors and proteomic profiles with respect to disease status at the time of sampling (newly diagnosed vs. relapsed) and AML characteristics (molecular or cytogenetic subtypes). We found no statistically significant correlations with disease status or cytogenetics. However, four AML samples harboring FLT3 mutations were universally less sensitive to dual BCL2/BCL-XL inhibitor ABT737 and were instead responsive to the combination of temsirolimus plus ABT737 in stromal co-culture (Online Supplementary Figure S8A). RPPA analysis revealed stroma-dependent protein alterations in multiple pathways, including the PI3K/AKT/MTOR and BCL2 signaling, which could be selectively associated with FLT3-mutated AML (Online Supplementary Figure S8B).

Discussion In this study, we used RPPA technology to examine cellular signaling alterations triggered by the blockade of MTOR, BCL2/BCL-XL, and MDM2 in AML cells under monoculture and stromal co-culture conditions. We identified the mechanisms by which stroma reprograms signaling networks in response to temsirolimus, ABT737, and Nutlin-3a and demonstrated that stroma-altered drug sensitivity is a result of signaling modulation. We further demonstrated that the PI3K/AKT/MTOR pathway is one of the major stroma-regulated survival pathways triggered by BCL2/BCL-XL and MDM2 inhibition. We identified the differences in the stroma-mediated networks affected by the combinations of temsirolimus plus ABT737 and temsirolimus plus Nutlin-3a and demonstrated that the simultaneous blockade of PI3K/AKT/MTOR and BCL2 is an effective strategy to combat stroma-mediated AML haematologica | 2017; 102(9)

survival. These findings, which we confirmed via conventional immunoblotting, provide a clinically relevant and mechanism-based tool for selecting effective combination therapies to treat microenvironment-protected AML. RPPA profiling of the signaling networks revealed that targeted inhibition affects proteins in non-targeted pathways. Our findings that STAT3 was inhibited by temsirolimus and that 4EBP1 expression was affected by Nutlin-3a are consistent with those of published reports;27,39 however, the finding that BAD expression was downregulated by ABT737 is documented here for the first time. Whether this effect is a result of blocking the downstream signaling cascade, of pathway crosstalk, or an off-target effect remains to be addressed in a future study. Our results showing signaling network differences in AML with and without stroma confirm the physiological role of stroma in AML and support the need to study target-selective inhibitors in the context of the BM microenvironment. Our results provide key evidence that PI3K/AKT/MTOR signaling is a stroma-regulated survival mechanism for AML. We are the first to report that the stroma-mediated mutual response of AML to ABT737 and Nutlin-3a activates AKT. These findings suggest that signaling crosstalk between BCL2 and MDM2 may take place in the setting of stroma-leukemia interaction. Indeed, Matter and Ruoslahti40 reported that extracellular integrin α5β1-αVβ3-mediated BCL2 transcription occurs through the activation of AKT in the PI3K/AKT/MTOR signaling; similarly, Du et al.41 showed that insulin-like growth factor 1 (IGF-1) upregulates MDM2 expression through the PI3K/AKT/MTOR pathway. Thus, the direct blockade of BCL2 or MDM2 may trigger a feedback response that promotes the stroma-leukemia interaction via the surface molecules α5β1-αVβ3 or by enhancing stromal IGF-1 secretion, either of which would lead to AKT activation. Directly targeting the stroma-leukemia interaction to prevent these feedback responses might synergize the therapeutic effect of BCL2 and MDM2 inhibition. Such a therapeutic strategy is currently under investigation. The synergistic effect of PI3K/AKT/MTOR and BCL2 blockade was demonstrated in a recent in vivo study in AML42 and in an in vitro study of solid tumor cells that formed spheroids, three-dimensional structures mimicking tumor cell growth in the microenvironment.43 Importantly, these findings provide a basis for investigating potential synergy at reduced doses, whereby combination treatment increases the therapeutic index without affecting overall therapeutic efficacy. Such an adjustment could be clinically valuable for reducing on-target toxicities associated with high doses of BCL2 or BCL-XL inhibitors, such as tumor lysis syndrome with ABT19932 and thrombocytopenia with ABT236.44 The combination of temsirolimus plus Nutlin-3a at 5 mM25 was less effective than the combination of temsirolimus plus ABT737. Recently, Lee et al.45 showed that the efficacy of combination treatment is affected by changes in the order and duration of drug exposure. Therefore, optimizing the administration sequence of temsirolimus and Nutlin-3a or the length of exposure to the single and combined treatments may increase target availability and improve therapeutic outcomes. Further studies should investigate these options. In addition, the toxicity and tolerance of this combination should be evaluated in an in vivo study. 1545


Z. Zeng et al.

Our findings demonstrated that the response of AML to targeted inhibition of PI3K/AKT/MTOR, BCL2/BCL-XL, and MDM2 is heterogeneous and complex and support the feasibility of personalized AML treatments. AMLs that are highly sensitive to ABT737 or Nutlin-3a alone may not require combination treatment. AMLs which are less sensitive to these targeted inhibitors and/or combinations in co-culture may require the use of novel agents with differ-

ent target specificity. The newly developed MTOR inhibitors, such as MLN0128 (formerly called INK128) and Torin1, target the ATP binding site of MTOR. These inhibitors effectively suppress MTORC1-dependent 4EBP1 phosphorylation and MTORC2-dependent AKT activation; therefore, they block the functions of both MTOR complexes that are resistant to rapamycin and rapamycin analogs.46,47 We and others reported that these

A

B

C Figure 6. Immunoblotting analysis validating RPPA results in AML cell lines and primary AML samples co-cultured with stroma. (A) Immunoblots of co-cultured OCI-AML3 and U937 cells treated with single inhibitor or dual inhibitors at the same concentration used in RPPA for 24 hours. The expression level of phosphorylated proteins (p-protein), total proteins, MDM2, and GAPDH was quantified. The ratios of the density of p-protein to that of total protein or GAPDH and of MDM2 to GAPDH were calculated and then divided by the density ratios of the same proteins in untreated cells. (B) Bar graphs display growth inhibition and apoptosis induction in co-cultured OCI-AML3 cells under the indicated treatments for 72 hours. Values are presented as mean Âą standard deviation of the mean. (C) Lysates of primary AML samples analyzed using RPPA were reanalyzed using immunoblotting to detect the expression of p-AKT (Ser473) and AKT in the treated and untreated cells under co-culture. The bands were quantified, and the density ratios were calculated as described above.

1546

haematologica | 2017; 102(9)


Targeting microenvironment-protected AML

inhibitors are more potent than rapamycin and its analogs.35,47,48 Rahmani et al.42 recently reported that the combination of INK128 and ABT737 synergistically inhibited the growth of U937 cells in vitro and prolonged survival of leukemic mice. This study demonstrated an alternative strategy of replacing temsirolimus with these MTOR inhibitors in order to increase therapeutic efficacy, and warrant an investigation into the co-treatment of MTORC1/C2 inhibitor plus ABT737 or Nutlin-3a in primary samples under stromal co-culture. Finally, our data support the notion of AML disease heterogeneity.49 Despite the variability in responses to the targeted inhibitors and their combinations among the 20 primary AML samples studied, distinct proteins/pathway alterations and apoptosis induction could correlate with unique disease characteristics. Likely due to the small sample size, results were largely negative except for associations with FLT3 mutation, one of the most common mutations in AML associated with poor prognosis. All four FLT3-mutated AML samples were less sensitive to ABT737 and instead responsive to the combination of temsirolimus plus ABT737 in stromal co-culture. Of note, we recently reported reduced sensitivity of FLT3-mutated AML to the selective BCL2 inhibitor ABT199 in a phase II AML clinical trial,32 possibly related to the reported upregulation of MCL1 protein conferring resistance to BCL2/BCL-XL inhibitors.50 In turn, MTOR

References 1. Estey E, Dohner H. Acute myeloid leukaemia. Lancet. 2006;368:1894-1907. 2. Dombret H, Gardin C. An update of current treatments for adult acute myeloid leukemia. Blood. 2016;127(1):53-61. 3. Shaffer BC, Gillet JP, Patel C, Baer MR, Bates SE, Gottesman MM. Drug resistance: still a daunting challenge to the successful treatment of AML. Drug Resist Updat. 2012;15(1-2):62-69. 4. Patel C, Stenke L, Varma S, et al. Multidrug resistance in relapsed acute myeloid leukemia: evidence of biological heterogeneity. Cancer. 2013;119(16):3076-3083. 5. Konopleva MY, Jordan CT. Leukemia stem cells and microenvironment: biology and therapeutic targeting. J Clin Oncol. 2011; 29(5):591-599. 6. Correia AL, Bissell MJ. The tumor microenvironment is a dominant force in multidrug resistance. Drug Resist Updat. 2012;15(12):39-49. 7. Becker PS. Dependence of acute myeloid leukemia on adhesion within the bone marrow microenvironment. ScientificWorldJournal. 2012;2012:856467. 8. Min YH, Eom JI, Cheong JW, et al. Constitutive phosphorylation of Akt/PKB protein in acute myeloid leukemia: its significance as a prognostic variable. Leukemia. 2003;17(8):995-997. 9. Xu Q, Simpson SE, Scialla TJ, Bagg A, Carroll M. Survival of acute myeloid leukemia cells requires PI3 kinase activation. Blood. 2003;102(3):972-980. 10. Martelli AM, Evangelisti C, Chiarini F, McCubrey JA. The phosphatidylinositol 3kinase/Akt/mTOR signaling network as a

haematologica | 2017; 102(9)

11.

12.

13.

14.

15.

16.

17.

18.

inhibitors are capable of downregulating MCL1, especially under the condition mimicking BM microenvironment.42 A further validation study using a larger dataset of FLT3-mutated AML is needed to support these preliminary findings. In summary, utilizing RPPA technology, we profiled AML signaling networks that are triggered by the targetselective inhibitors temsirolimus, ABT737, and Nutlin-3a and modified by stroma. We demonstrated that stromaaltered drug sensitivity is the result of signaling modifications and showed that exploring drug combinations in a co-culture model identifies the mechanism-based effective therapies to combat stroma-mediated AML survival. Altogether, our findings in this study provide important guidance for developing tailored treatments to eliminate microenvironment-protected AML. Funding This work was supported in part by National Institutes of Health/National Cancer Institute Leukemia SPORE Career Development Award P50 CA100632 (to ZZ); NIH/NCI grant 5 R01 CA155056-04 and Leukemia and Lymphoma Society grant 6427-13 (to MK); and NIH grants P01 CA55164, P30 CA016672, P50 CA100632, and R01 CA163481, a Cancer Prevention and Research Institute of Texas Shared Instrument Award, and the Paul and Mary Haas Chair in Genetics (all to MA).

therapeutic target in acute myelogenous leukemia patients. Oncotarget. 2010; 1(2):89-103. Konopleva M, Zhao S, Hu W, et al. The anti-apoptotic genes Bcl-X(L) and Bcl-2 are over-expressed and contribute to chemoresistance of non-proliferating leukaemic CD34+ cells. Br J Haematol. 2002; 118(2):521-534. Bueso-Ramos CE, Yang Y, deLeon E, McCown P, Stass SA, Albitar M. The human MDM-2 oncogene is overexpressed in leukemias. Blood. 1993;82(9):2617-2623. Filipits M, Stranzl T, Pohl G, et al. Drug resistance factors in acute myeloid leukemia: a comparative analysis. Leukemia. 2000;14(1):68-76. Konopleva M, Konoplev S, Hu W, Zaritskey AY, Afanasiev BV, Andreeff M. Stromal cells prevent apoptosis of AML cells by up-regulation of anti-apoptotic proteins. Leukemia. 2002;16(9):1713-1724. Matsunaga T, Takemoto N, Sato T, et al. Interaction between leukemic-cell VLA-4 and stromal fibronectin is a decisive factor for minimal residual disease of acute myelogenous leukemia. Nat Med. 2003; 9(9):1158-1165. Hazlehurst LA, Argilagos RF, Dalton WS. Beta1 integrin mediated adhesion increases Bim protein degradation and contributes to drug resistance in leukaemia cells. Br J Haematol. 2007;136(2):269-275. Tabe Y, Jin L, Tsutsumi-Ishii Y, et al. Activation of integrin-linked kinase is a critical prosurvival pathway induced in leukemic cells by bone marrow-derived stromal cells. Cancer Res. 2007;67(2):684694. Kojima K, McQueen T, Chen Y, et al. p53

19. 20. 21.

22.

23.

24.

25.

26.

27.

activation of mesenchymal stromal cells partially abrogates microenvironmentmediated resistance to FLT3 inhibition in AML through HIF-1alpha-mediated downregulation of CXCL12. Blood. 2011; 118(16):4431-4439. Shafer D, Grant S. Update on rational targeted therapy in AML. Blood Rev. 2016; 30(4):275-283. Ma WW, Jimeno A. Temsirolimus. Drugs Today (Barc). 2007;43(10):659-669. Zeng Z, Sarbassov dos D, Samudio IJ, et al. Rapamycin derivatives reduce mTORC2 signaling and inhibit AKT activation in AML. Blood. 2007;109(8):3509-3512. Oltersdorf T, Elmore SW, Shoemaker AR, et al. An inhibitor of Bcl-2 family proteins induces regression of solid tumours. Nature. 2005;435(7042):677-681. Konopleva M, Contractor R, Tsao T, et al. Mechanisms of apoptosis sensitivity and resistance to the BH3 mimetic ABT-737 in acute myeloid leukemia. Cancer Cell. 2006; 10(5):375-388. Vassilev LT, Vu BT, Graves B, et al. In vivo activation of the p53 pathway by smallmolecule antagonists of MDM2. Science. 2004;303(5659):844-848. Kojima K, Konopleva M, Samudio IJ, et al. MDM2 antagonists induce p53-dependent apoptosis in AML: implications for leukemia therapy. Blood. 2005;106(9):31503159. Teachey DT, Obzut DA, Cooperman J, et al. The mTOR inhibitor CCI-779 induces apoptosis and inhibits growth in preclinical models of primary adult human ALL. Blood. 2006;107(3):1149-1155. Kojima K, Shimanuki M, Shikami M, et al. The dual PI3 kinase/mTOR inhibitor PI-103

1547


Z. Zeng et al.

28.

29.

30.

31.

32.

33.

34.

35.

1548

prevents p53 induction by Mdm2 inhibition but enhances p53-mediated mitochondrial apoptosis in p53 wild-type AML. Leukemia. 2008;22(9):1728-1736. Kline MP, Rajkumar SV, Timm MM, et al. ABT-737, an inhibitor of Bcl-2 family proteins, is a potent inducer of apoptosis in multiple myeloma cells. Leukemia. 2007; 21(7):1549-1560. Zhang W, Ruvolo VR, Gao C, et al. Evaluation of apoptosis induction by concomitant inhibition of MEK, mTOR, and Bcl-2 in human acute myelogenous leukemia cells. Mol Cancer Ther. 2014; 13(7):1848-1859. Yee KWL, Garcia-Manero G, Thomas D, et al. A phase II study of temsirolimus (CCI779) in patients with advanced leukemias. Blood. 2004;104(11):4523. Andreeff M, Kelly KR, Yee K, et al. Results of the phase I trial of RG7112, a small-molecule MDM2 antagonist in leukemia. Clin Cancer Res. 2016;22(4):868-876. Konopleva M, Pollyea DA, Potluri J, et al. Efficacy and biological correlates of response in a phase II study of venetoclax monotherapy in patients with acute myelogenous leukemia. Cancer Discov. 2016; 6(10):1106-1117. Kornblau SM, Tibes R, Qiu YH, et al. Functional proteomic profiling of AML predicts response and survival. Blood. 2009; 113(1):154-164. Itoh K, Tezuka H, Sakoda H, et al. Reproducible establishment of hemopoietic supportive stromal cell lines from murine bone marrow. Exp Hematol. 1989; 17(2):145-153. Zeng ZH, Shi YX, Tsao T, et al. Targeting of

36.

37.

38.

39.

40.

41.

42.

mTORC1/2 by the mTOR kinase inhibitor PP242 induces apoptosis in AML cells under conditions mimicking the bone marrow microenvironment. Blood. 2012; 120(13):2679-2689. Tibes R, Qiu Y, Lu Y, et al. Reverse phase protein array: validation of a novel proteomic technology and utility for analysis of primary leukemia specimens and hematopoietic stem cells. Mol Cancer Ther. 2006;5(10):2512-2521. Hu J, He X, Baggerly KA, Coombes KR, Hennessy BTJ, Mills GB. Non-parametric quantification of protein lysate arrays. Bioinformatics. 2007;23(15):1986-1994. Neeley ES, Baggerly KA, Kornblau SM. Surface adjustment of reverse phase protein arrays using positive control spots. Cancer Inform. 2012;11:77-86. Kim SH, Zukowski K, Novak RF. Rapamycin effects on mTOR signaling in benign, premalignant and malignant human breast epithelial cells. Anticancer Res. 2009;29(4):1143-1150. Matter ML, Ruoslahti E. A signaling pathway from the alpha5beta1 and alpha(v)beta3 integrins that elevates bcl-2 transcription. J Biol Chem. 2001; 276(30):27757-2763. Du W, Yi Y, Zhang H, et al. Rapamycin inhibits IGF-1-mediated up-regulation of MDM2 and sensitizes cancer cells to chemotherapy. Plos One. 2013;8(4):e63179. Rahmani M, Aust MM, Hawkins E, et al. Co-administration of the mTORC1/TORC2 inhibitor INK128 and the Bcl-2/Bcl-xL antagonist ABT-737 kills human myeloid leukemia cells through Mcl-1 down-regulation and AKT inactiva-

43.

44.

45.

46.

47.

48.

49. 50.

tion. Haematologica. 2015;100(12):15531563. Muranen T, Selfors LM, Worster DT, et al. Inhibition of PI3K/mTOR leads to adaptive resistance in matrix-attached cancer cells. Cancer Cell. 2012;21(2):227-239. Cang S, Iragavarapu C, Savooji J, Song Y, Liu D. ABT-199 (venetoclax) and BCL-2 inhibitors in clinical development. J Hematol Oncol. 2015;8(1):129. Lee MJ, Ye AS, Gardino AK, et al. Sequential application of anticancer drugs enhances cell death by rewiring apoptotic signaling networks. Cell. 2012;149(4):780794. Thoreen CC, Kang SA, Chang JW, et al. An ATP-competitive mammalian target of rapamycin inhibitor reveals rapamycinresistant functions of mTORC1. J Biol Chem. 2009;284(12):8023-8032. Janes MR, Vu C, Mallya S, et al. Efficacy of the investigational mTOR kinase inhibitor MLN0128/INK128 in models of B-cell acute lymphoblastic leukemia. Leukemia. 2013; 27(3):586-594. Zeng Z, Wang RY, Qiu YH, et al. MLN0128, a novel mTOR kinase inhibitor, disrupts survival signaling and triggers apoptosis in AML and AML stem/ progenitor cells. Oncotarget. 2016;7(34):55083-55097. Dohner H, Weisdorf DJ, Bloomfield CD. Acute myeloid leukemia. N Engl J Med 2015;373(12):1136-1152. 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.

haematologica | 2017; 102(9)


ARTICLE

Acute Myeloid Leukemia

Ultrasensitive detection of acute myeloid leukemia minimal residual disease using single molecule molecular inversion probes Adam Waalkes, Kelsi Penewit, Brent L. Wood, David Wu and Stephen J. Salipante

Department of Laboratory Medicine, University of Washington, Seattle, WA, USA

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(9):1549-1557

ABSTRACT

T

he identification of minimal residual disease is the primary diagnostic finding which predicts relapse in patients treated for acute myeloid leukemia. Ultrasensitive detection of minimal residual disease would enable better patient risk stratification and could open opportunities for early therapeutic intervention. Herein we apply single molecule molecular inversion probe capture, a technology combining multiplexed targeted sequencing with error correction schemes based on molecular barcoding, in order to detect mutations identifying minimal residual disease with ultrasensitive and quantitative precision. We designed a single molecule molecular inversion probe capture panel spanning >50 kb and targeting 32 factors relevant to acute myeloid leukemia pathogenesis. We demonstrate linearity and quantitative precision over 100-fold relative abundance of mutant cells (1 in 100 to 1 in 1,500), with estimated error rates approaching 1 in 1,200 base pairs sequenced and maximum theoretical limits of detection exceeding 1 in 60,000 mutant alleles. In 3 of 4 longitudinally collected specimens from patients with acute myeloid leukemia, we find that single molecule molecular inversion probe capture detects somatic mutations identifying minimal residual disease at substantially earlier time points and with greater sensitivity than clinical diagnostic approaches used as current standard of care (flow cytometry and conventional molecular diagnosis), and identifies persisting neoplastic cells during clinical remission. In 2 patients, single molecule molecular inversion probe capture detected heterogeneous, subclonal acute myeloid leukemia populations carrying distinct mutational signatures. Single molecule molecular inversion probe technology uniquely couples scalable target enrichment with sequence read error correction, providing an integrated, ultrasensitive approach for detecting minimal residual disease identifying mutations. Intrtoduction Acute myeloid leukemia (AML) is a highly aggressive, immature hematopoietic neoplasm with significant mortality and morbidity. Despite sustained improvements in our understanding of the mechanisms underlying this disease, persistent challenges regarding improving patient outcomes remain. Individuals with AML who can tolerate hematopoietic stem cell transplantation have enjoyed substantial improvements in disease-free and overall survival, but even so, clinical relapse is a significant challenge for many patients, particularly for those who receive only conventional chemotherapy treatments. Several independent studies have shown that the detection of minimal residual disease (MRD), defined as small numbers of neoplastic cells which persist after cancer therapy, is a key prognostic variable in predicting disease relapse in the post-treatment setting of AML.1â&#x20AC;&#x201C;5 In current practice, MRD may be assessed by a variety of disparate molecular or phenotypic assessment techniques including detection of abnormal immunophenotype by flow cytometry,6,7 fluorescent in situ hybridization (FISH)/cytogenetics,8 and real time polymerase chain reaction (RT-PCR) for individual AML-specific genetic haematologica | 2017; 102(9)

Correspondence: stevesal@uw.edu

Received: March 21, 2017. Accepted: May 31, 2017. Pre-published: June 1, 2017. doi:10.3324/haematol.2017.169136 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1549 Š2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1549


A. Waalkes et al. lesions.2,9,10 These methods have variable performance characteristics, and no single phenotypic or molecular signature of disease likely exists, ultimately making current patient stratification achievable with variable levels of success. Consequently, robust and standardized methods for identifying MRD are critically needed.11 For many hematologic disorders, next-generation sequencing has emerged as an appealing strategy for detecting and monitoring MRD.12-14 In the case of AML, whole genome and exome sequencing15-18 has revealed that the number of mutations occurring in AML is limited, with estimates of the mutational burden of any given malignancy being in the order of only 20 to 30 events.15 This presents a special challenge in detecting meaningful genomic markers of AML MRD, particularly in sub-types that lack recurrent genomic fusions such as acute promyelocytic leukemia with PML-RARA. Successful strategies to detect AML MRD using high-throughput sequencing have consequently focused on 1 of 2 strategies. The first of these is focused deep sequence analysis of recurrent mutational hotspots, such as those occurring in NPM1,19 RUNX1,20 or FLT3.21 Although potentially powerful molec-

ular markers of malignancy, these variants are not necessarily present nor informative in all patients. Restricting analysis to single genes or sites also limits the ability to detect and monitor clonal heterogeneity, which is thought to be an important feature of AML pathogenesis.22,23 The second strategy is to perform whole genome or whole exome sequencing of fulminant malignancy in order to catalog informative, patient-specific mutations which can be later targeted using personalized assays and followed over time.24-26 Although more comprehensive, this latter strategy is both more costly and more time consuming,27 and in practice also requires the subsequent validation of patient-specific assays for MRD detection. To improve upon existing paradigms for conventional and next-generation sequencing-based AML MRD detection, herein we adapt single molecule molecular inversion probes (smMIPs)28,29 in order to interrogate common genetic lesions in AML with ultrasensitive limits of detection. smMIP technology unites multiplexed targeted sequencing with an error correction strategy based on unique molecular identifiers (UMIDs),30 degenerate oligonucleotide barcodes that mark sequence reads

Table 1. Patient specimens and clinical testing results.

Patient

Day from initial diagnosis

Days post-transplant

Molecular results

Flow cytometry results

83 196 329 412 531 0

N/A** N/A N/A N/A N/A -97

FLT3 and NPM1 negative FLT3 and NPM1 negative FLT3 Positive, 35%. NPM1 Positive. FLT3 Positive, 15%. NPM1 Positive. FLT3 Positive, 97%. NPM1 Positive. FLT3 Negative. NPM1 Positive.

2

69 125 393

28 84 352

NP NP FLT3 and NPM1 negative

3

397 461 579 746 0 94 143 269 346 437 542 584 599 269 1083 1214 1556

356 420 538 705 -115 -21 28 154 231 322 427 469 484 -917 -103 28 370

NP FLT3 negative NP FLT3 Positive, 10.46% FLT3 and NPM1 negative FLT3 and NPM1 negative FLT3 and NPM1 negative FLT3 and NPM1 negative FLT3 and NPM1 negative FLT3 and NPM1 negative FLT3 and NPM1 negative FLT3 and NPM1 negative JAK2 mutation negative NPM1 Negative. NPM1 Negative. NPM1 Negative. NPM1 Negative.

NP* NP NP NP 98% abnormal cells 0.98% abnormal cells (regenerating blasts vs. MRD) Negative for abnormal cells Negative for abnormal cells 0.14% abnormal cells (suspicious but not diagnostic for MRD) NP Negative for abnormal cells Negative for abnormal cells NP NP Negative for abnormal cells Negative for abnormal cells Negative for abnormal cells Negative for abnormal cells Negative for abnormal cells Negative for abnormal cells Negative for abnormal cells NP NP Negative for abnormal cells Negative for abnormal cells Negative for abnormal cells

1

4

*Not performed. **Not Applicable. MRD: minimal residual disease.

1550

haematologica | 2017; 102(9)


Ultrasensitive leukemia monitoring

derived from a common progenitor molecule and which allow for random sequencing errors to be eliminated through oversampling and consensus calling (Figure 1). When combined, these features enable simultaneous enrichment, detection, and quantitation of single nucleotide polymorphisms and small insertions and deletions in targeted genomic regions with a sensitivity greatly exceeding27,28 the inherent error rate of next-generation sequencing (~2% per nucleotide).31 These pilot studies define the performance characteristics of smMIP capture as a molecular diagnostic, and demonstrate the utility of the approach when applied in clinical practice.

Methods

A

B

C

Samples and Cell Lines Residual, clinical samples were obtained and de-identified according to the University of Washington Institutional Review Board guidelines. This project was approved by the University of Washington Human Subjects Division and was conducted in accordance with the Declaration of Helsinki. A total of 25 samples, derived from prior, routine sampling from 4 patients were used in this pilot study (Table 1, Online Supplementary Table S1). All patients had a confirmed histologic or flow cytometry diagnosis of acute myeloid leukemia and were selected for further study if sufficient residual DNA (at least 500 ng) was available for analysis. Staging marrows from lymphoma patients under 40 years of age for whom myeloid flow cytometry was pre-screened as negative were used as normal bone marrow controls. For linearity and sensitivity studies, suspensions of cell lines and normal human bone marrow were quantified using flow cytometry, combined in defined proportions, and incremental serial dilutions were prepared. Cell lines were obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen (cell lines KMH2, OCI-AML3, and L1236, Braunschweig, Germany) or ATCC (cell line Raji, Manassas, VA, USA) and were cultured in accordance with supplier specifications. NA12878 genomic DNA was obtained from Coriell Biorepository (Camden, NJ, USA).

D

E

F

G

smMIP design, capture, and sequencing A smMIP capture panel was designed against AML-relevant targets (Table 2, Online Supplementary Table S2), identifying polymorphisms carried in cell lines, and other clinically relevant genes (ABL1, ALK, JAK2, NT5C2, and ROS1) using the program Molecular Interaction Potential Generator (MIPgen).32 As detailed in the Online Supplementary Methods, 500 ng genomic DNA was hybridized with the panel, exonuclease-treated, and PCR-amplified to generate sequencing libraries. Single libraries were prepared from all specimens in this study. Sequencing was performed using the 300 cycle Illumina NextSeq 500/550 High Output v2 kit (Illumina, San Diego, CA, USA).

Data analysis Pipeline Sequencing data were analyzed as described in full in the Online Supplementary Methods. Briefly, reads were demultiplexed, regions corresponding to the smMIP backbone were removed, and read pairs self-assembled. After mapping to the human genome (hg37), reads corresponding to each smMIP were grouped based on whether their UMID was contained in 2, or more than 2 independent reads, with UMIDs represented in only 1 sequence read discarded. Single nucleotide polymorphisms and indels were called on these read groups using a â&#x20AC;&#x153;majority ruleâ&#x20AC;? approach. We applied an empiric site- and mutation-specific error model in order to assess the significance of the observed variation at each site in haematologica | 2017; 102(9)

H

Figure 1. Overview of smMIP capture. (A) smMIPs are single-stranded (ss)DNA oligonucleotides with domains at both ends that are complimentary to genomic targets of interest (gray), each flanked by a 4bp fully degenerate UMID sequence (red), totaling an 8bp molecular tag, and linked by a backbone sequence common to all probes (black). After hybridization of probes to DNA, (B) DNA polymerase copies targeted genomic DNA by extension of the free probe arm and (C) the smMIP is joined into a covalently sealed circle by the action of DNA ligase. After exonuclease digestion of unbound probes and free genomic DNA, (D) PCR primed against the defined smMIP backbone is performed to amplify successful capture events and their identifying UMIDs. (E) Paired-end, high-throughput sequencing is performed. (F) Derived paired-end reads are merged into a single, contiguous read, providing inherent intra-read error correction. (G) A consensus sequence is generated out of all reads sharing a common UMID. Low prevalence, random errors are canceled out. (H) Extremely low variant allele frequencies can be accurately quantified by assaying the fraction of error-corrected reads bearing a mutation of interest out of the larger population of read groups derived from unique UMID-tagged smMIPs.

1551


A. Waalkes et al.

the smMIP panel, and sites with P>0.005 were excluded as potential artefacts. In summary, the analysis pipeline expresses variant calls in terms of the number of unique smMIP capture events which are consistent with a given variant over the total number of smMIP capture events overlying that site, and assesses the statistical significance of individual variants.

Data availability Sequence data generated for this study are available from the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) under study accession number SRP097634.

Results smMIP panel design and performance We designed a smMIP panel targeting the most recurrently mutated genes in AML15,33 and other high-yield or actionable mutational targets, featuring a combination of full gene tiling and focused hotspot interrogation as appropriate to the selected targets (Table 2). After rebalancing the relative concentration of individual probes in the capture pool in order to promote evenness of performance, and subsequently removing smMIPs with persistently low performance (n=83), the final capture design included 511 probes which spanned a total of 50,176 base pairs (bp) of

A

genomic DNA (Online Supplementary Table S2). With an allotment of 1.05X106 ± 5.9X106 (average ± standard deviation) reads per specimen, an average of 17,763 unique UMIDs were obtained per smMIP (range of 973 to 52,437), with an average of 11.2 sequence read pairs per UMID (Figure 2A, Online Supplementary Figure S1). Given our criteria for data filtering, these performance characteristics correspond to an average theoretical sensitivity of ~1 in 9,000 mutant alleles. Minor variations were seen among replicates, possibly due to differences in DNA quality.

Error reduction and error profile In order to initially evaluate the performance of smMIPmediated error correction compared to conventional sequencing we analyzed 2 specimens, reference cell line NA12878 and a bone marrow sample derived from a healthy donor, which were subjected to smMIP capture and sequencing. After masking sites of variation that were consistent with heterozygous or homozygous germline polymorphisms, we quantified the number of variant calls derived both from raw sequencing reads and after applying smMIP-mediated error correction (Table 3). Applying standard cutoffs for accepting variants with equal to or greater than 2% variant allele frequency31 (Online Supplementary Methods), conventional deep sequencing registered an average of 1,047 ± 17 (average ± standard deviation) artefactual single nucleotide variant

B

Figure 2. Performance characteristics of single molecule molecular inversion probes (smMIPs). (A) Number of unique capture events recovered per smMIP (arbitrarily sorted along X-axis) for 3 different control specimens. (B) Linearity and reproducibility of the smMIP panel for 13 identifying loci in immortalized cell lines spiked into normal human bone marrow cells at different levels of relative abundance. Error bars indicate standard deviation.

1552

haematologica | 2017; 102(9)


Ultrasensitive leukemia monitoring

calls and 10 ± 2 indels, for a cumulative error frequency of 1 artefact reaching detection thresholds per 50 nucleotides sequenced. Using parameters enabled to detect low prevalence variation (Online Supplementary Methods), the conventional deep sequencing analysis approach increased errors considerably, resulting in 65,127 ± 1464 (average ± standard deviation) artefactual single nucleotide variants and 6,471 ± 334 (average ± standard deviation) artefactual indels per specimen, equating to an average of 1.42 different low prevalence variants detected per nucleotide sequenced. By contrast, using smMIP capture and associated analytic error modeling, we identified a total of 1,025±664 (average ± standard deviation) single nucleotide variants and 220±2 (average ± standard deviation) indels per specimen occurring at any level of detectable abundance. This equates to 1 error per 40 bp sequenced, comparable to error rates seen using deep sequencing with standard variant calling, and representing a 60-fold reduction in error rate over ultrasensitive mutation detection by conventional sequencing. However, smMIP capture resulted in a non-uniform distribution of errors across the landscape of potential nucleotide substitutions (Table 3), for which there was a predominance of transversions involving cytosine and guanine. This spectrum of changes is consistent with a low frequency of oxidative DNA damage (G>T and C>A)25,34 and spontaneous deamination of 5-methylcytosine (C>T and G>A),28 which exist in template molecules prior to sequencing. Interestingly, an elevation of these error types was not observed for either of the conventional sequencing analysis approaches, presumably because DNA damage events are too low frequency to be detected, or are masked by random, higher frequency sequencing errors overlapping these sites. In light of these observations, we conclude that much of the low level variation detected by smMIP capture represents pre-analytic DNA damage, rather than sequencing errors per se. Removing these state changes from consideration, the error rate of smMIP capture is estimated to approach 1 error per 1,200 bp sequenced, an order of magnitude less than that seen with standard variant calling and more than 2 orders of magnitude less than seen for ultrasensitive calls made without error correction.

Linearity and quantitative precision In order to assess the effectiveness with which smMIP capture was able to recover known mutations at various levels of relative abundance, we constructed a series of synthetic specimens where calibrated numbers of cells from normal human bone marrow were combined with lesser proportions of different immortalized human cell lines marked by identifying SNPs and indels.35 We evaluated 2-fold serial dilutions, ranging from 1% to 0.0625% relative abundance, of 4 different cell lines representing a total of 14 identifying variants (although 2 of these variants, MYC c.G62C and c.G162C, were targeted by a single smMIP). We were able to recover all but 1 of the polymorphisms occurring at the lowest dilution examined (Figure 2B, Online Supplementary Table S3). Linearity and quantitation of variant allele frequency was achieved over 2 orders of magnitude, with consistency across samples and the individual mutations typed (coefficient of variation range of 0.41 to 0.74, depending on dilution). These data demonstrate that smMIP capture has both sensitivity and quantitative precision to at least 1 in 1,500 mutant alleles. haematologica | 2017; 102(9)

Detection of MRD in longitudinal patient samples using smMIP capture To investigate the potential of smMIPs to identify MRD from patient samples, we examined a cohort of 4 patients for whom MRD had been evaluated clinically over multiple time points (Table 1). Two of the patients in our cohort relapsed during the period of sample collection, whereas the other 2 had negative MRD detection results at all time points. The AML at initial diagnosis from each of these patients was positive for NPM1 mutation by conventional testing, providing a known molecular marker against which to benchmark performance. We subjected each specimen to smMIP capture and compared the ability of that approach to detect MRD with the results of conventional diagnostics that had been applied during the course of patient care (Table 1, Figure 3, Online Supplementary Table S4).

Table 2. smMIP panel design.

Gene

Capture Design

Nucleotides sequenced (bp)

Number smMIPs

ABL1 ALK BRAF CEBPA DNMT3A EZH2 FAM5C FLT3 HNRNPK IDH1 IDH2 JAK2 KIT KRAS NPM1 NRAS NT5C2 PHF6 PIK3CA PPM1D PTPN11 RAD21 RET ROS1 RUNX1 SMC1A SMC3 STAG2 TET2 TP53 U2AF1 WT1 Cell line variants Total

Hotspot Hotspot Hotspot Full Gene Hotspot Full Gene Full Gene Hotspot Full Gene Hotspot Full Gene Hotspot Hotspot Hotspot Hotspot Hotspot Hotspot Full Gene Hotspot Full Gene Full Gene Full Gene Hotspot Hotspot Full Gene Full Gene Full Gene Full Gene Full Gene Full Gene Hotspot Full Gene Hotspot N/A

817 333 222 203 111 3117 2583 873 2281 111 205 120 1041 333 89 222 444 927 333 1971 2418 2410 111 333 1894 5060 5147 5163 6326 2069 222 1688 999 50176

8 3 2 2 1 32 28 8 24 1 2 1 10 3 1 2 4 9 3 22 24 25 1 3 19 53 52 53 66 22 2 16 9 511

N/A: not applicable; smMIPs: single molecule molecular inversion probes.

1553


A. Waalkes et al.

In 1 patient (patient 4), no low prevalence somatic mutations were identified in any specimen, consistent with the clinical diagnosis of the patient being free of detectable MRD. By contrast, and remarkably, for the remaining 3 patients multiple somatic mutations identifying the presence of MRD were observed at every time point interrogated, regardless of clinical diagnostic status. Between 4 and 7 variants were identified in these cases, representing mutations in coding, intronic, and UTR regions, to a minimum variant allele frequency of 0.02% (Online Supplementary Table S4). Despite their low prevalence, all reported variants occurred at levels significantly higher (P<0.005) than predicted under our empiric error models, and can therefore be ascribed to a biological, rather than artefactual, source. In order to provide orthologous validation of MRD in specimens where abnormal cells were not identified by standard clinical diagnostics, we performed ultrasensitive detection of NPM1 mutations using a previously described, targeted next-generation sequencing assay19 (Figure 3, Online Supplementary Table S4). Although sensitivity of the NPM1 assay is validated only to a variant allele frequency of 0.03%, quantitation of NPM1-mutated cells by this independent assay closely mirrored the results obtained using smMIP capture, although at several time points smMIP capture identified the presence of MRD at levels occurring below the limits of detection of the targeted NPM1 assay. These results provide support for the clin-

A

ical validity and quantitative measurement of low prevalence mutations detected by smMIP capture.

MRD heterogeneity in patient samples AML is considered an oligoclonal disease, marked by the emergence of sublineages which evolve over time and which may exhibit different functional properties from one another,36-38 and we therefore assessed our ability to identify genetically distinct subclones in our patient cohort. In 2 patient specimens we observed dynamic changes in the mutations seen over time, which is consistent with the recovery of discrete sublineages (Figure 3, Online Supplementary Table S4). In patient 2, we detected 2 low frequency variants in TP53, which initially emerged on day 579 following initial diagnosis and persisted through the next and final timepoint on day 746. However, these 2 mutations did not increase in abundance at the time that AML relapse was clinically diagnosed; by contrast, the 5 mutations which were originally identified in the patientâ&#x20AC;&#x2122;s neoplasm markedly increased in prevalence. These finding suggest both that a discrete subclonal lineage of TP53-mutated hematogenic stem and progenitor cells39 evolved in the patient over time, and that cells most closely related to the patientâ&#x20AC;&#x2122;s original neoplasm, rather than this subclone, were responsible for the patientâ&#x20AC;&#x2122;s relapse. In patient 3, a different pattern was seen. One distinction is that the variant allele frequency of somatic muta-

B

C

Figure 3. Detection of MRD and clonal heterogeneity in patient specimens using smMIP capture. Results are shown for patient 1 (A), patient 2 (B), and patient 3 (C). The variant allele frequency of the individual somatic mutations identified in each case are plotted as a function of time (in days) the first sampling in the time series and, where available, days status posttransplant. Days marked in red indicate time points at which neoplastic cells were detected by conventional clinical assays. Correlation of results with alternative testing for NPM1 mutation are indicated for each case (NPM1NGS, double thick line).

1554

haematologica | 2017; 102(9)


Ultrasensitive leukemia monitoring

tions identified in this background was considerably lower than for the other 2 patients, potentially reflecting that this patient did not undergo relapse during the period of study. Additionally, the relative abundance of individual mutations did not correlate in any obvious pattern, and instead fluctuated independently. These observations are consistent with the presence of multiple subclones carrying distinct identifying mutations, without any single lineage coming to dominate the environment.

Discussion For next-generation sequencing applications intended to detect low prevalence variation in heterogeneous samples, inherent error rates of sequencing platforms rapidly become limiting by conventional workflows; even with the highest fidelity chemistries, single nucleotide variants occurring below ~2% relative abundance cannot be reliably distinguished from artefacts.31 Various strategies for circumventing this problem have been demonstrated,30,40-44 each with their own comparative advantages and disadvantages. smMIP technology uniquely couples scalable target enrichment with sequence read error correction, providing an integrated approach which is both facile and quantitative.28 Moreover, due to the modular nature of smMIP assays, additional targets can be incorporated into an existing panel without necessitating assay redesign or resynthesis of existing probes, allowing for ready expansion. In order to detect low abundance mutations which identify AML MRD, we have designed a smMIP panel targeted against coding genes and mutational hotspots relevant to

AML pathogenesis, and have developed optimized protocols and analytic techniques to maximize sensitivity. Using cell line dilutions, we demonstrated that smMIP capture is able to interrogate relevant SNPs and indels occurring in AML with an average sensitivity of at least 1:1,500 mutant alleles. This sensitivity for low prevalence variation both exceeds that of conventional deep sequencing by 3 orders of magnitude,27 and surpasses the limits of detection previously achieved using smMIP technology by 1 order of magnitude.28,29 Nevertheless, the UMID counts obtained per smMIP enable theoretical limits of detection exceeding even these figures, in the order of 1 in 9,000 mutant alleles for a probe with average performance and up to 1 in 61,000 for the best performers. It is noteworthy that the levels of sensitivity achieved using smMIP capture approach those achievable by error corrected deep sequencing of single mutational targets in AML,19,25 but the technology is distinguished from those methods in that ultrasensitive performance is obtained over a far greater breadth of coverage; in this application, greater than 50 kb of genomic sequence were interrogated. However, the sensitivity of smMIP capture scales linearly with reaction size; given sufficient quantities of DNA template and allocated sequencing coverage, the approach should be able to achieve detection of low prevalence mutations exceeding the limits practically demonstrated in the study herein. A measurable amount of low prevalence genetic variation identified in our test materials corresponded to artefacts resulting from DNA deamination and oxidative damage.25,28,34 In practice, these artefacts can be identified and distinguished from true mutations by capturing both strands of DNA independently and evaluating concur-

Table 3. Single nucleotide variant error rates before and after single molecule molecular inversion probe (smMIP)-mediated error correction.

Error Type

Conventional sequencing with 2% variant allele frequency calling Average errors Standard per base deviation

A>N 8.334x10-3 C>N 3.537x10-2 G>N 3.005x10-2 T>N 1.452x10-2 A>C 2.847x10-3 A>G 2.881x10-3 A>T 2.607x10-3 C>A 2.708x10-2 C>G 5.498x10-3 C>T 2.795x10-3 G>A 5.751x10-3 G>C 9.451x10-3 G>T 1.484x10-2 T>A 8.869x10-3 T>C 2.503x10-3 T>G 3.147x10-3 Total average errors per base 2.067x10-2 Excluding C>A, C>T, G>A, G>T9.677x10-3 haematologica | 2017; 102(9)

2.401x10-4 2.199x10-3 2.853x10-3 2.575x10-3 3.430x10-5 6.860x10-5 1.372x10-4 2.703x10-3 4.582x10-4 4.582x10-5 4.012x10-4 1.783x10-4 2.274x10-3 2.790x10-3 0 2.146x10-4 3.255x10-4 8.385x10-4

Conventional sequencing with ultrasensitive variant Average errors Standard per base deviation 1.182 1.371 1.419 1.218 0.281 0.450 0.451 0.489 0.447 0.435 0.461 0.474 0.484 0.475 0.444 0.299 1.285 0.877

4.791x10-2 1.150x10-2 9.317x10-3 5.325x10-2 3.471x10-2 7.683x10-3 5.522x10-3 5.452x10-3 1.026x10-2 6.689x10-3 9.807x10-4 2.675x10-3 5.662x10-3 8.583x10-4 7.832x10-3 4.456x10-2 2.888x10-2 3.01x10-2

smMIP-mediated ultrasensitive variant calling calling Average errors Standard per base deviation 1.303x10-3 5.205x10-2 3.540x10-2 8.941x10-4 3.087x10-4 6.517x10-4 3.430x10-4 3.775x10-2 3.665x10-4 1.393x10-2 1.092x10-2 7.133x10-4 2.376x10-2 1.073x10-4 5.722x10-4 2.146x10-4 1.966x10-2 8.582x10-4

2.058x10-4 2.712x10-2 1.810x10-2 1.073x10-4 3.430x10-5 1.715x10-4 6.860x10-5 2.190x10-2 9.163x10-5 5.315x10-3 4.413x10-3 4.458x10-4 1.413x10-2 3.576x10-5 1.431x10-4 7.153x10-5 9.815x10-3 1.480x10-4 1555


A. Waalkes et al. rence between strands at sites of variation.29 However, this measure would require twice as many smMIPs and subsequent sequence coverage, with a negative practical impact on sensitivity. We have found that the incidence of these artefacts is non-limiting in our application, and have therefore elected for a minimally redundant capture design in order to maximize the detection of low prevalence alleles. In comparison with existing, standard methods of clinical MRD detection, smMIP capture was able to identify mutations indicative of MRD with greater sensitivity and earlier in the course of patient care, up to 677 days before abnormal cells were detectable using standard of care diagnostics. In 3 of 4 patients with longitudinal specimens available, smMIP capture identified mutations consistent with MRD at all time points. Although the finding of low level neoplastic cells during remission is somewhat surprising, it is compatible with prior work by our group using ultrasensitive MRD detection methods targeting the NPM1 locus in AML patients,19 and also with the results of other groups using ultrasensitive sequencing approaches.38 It has been argued that persistent, low level MRD indicates successful immune surveillance and suppression of abnormal cells rather than necessarily predicting the early stages of relapse.11 This view is supported in our current study by the long periods of time over which background levels of MRD were identified without a marked increase in the size of the abnormal cell population. As such, a more informative biomarker for AML relapse may prove to be the growth kinetics of low prevalence abnormal cell populations,19 a strategy which would be facilitated by the quantitative nature of smMIP capture. smMIP capture has proven robust and has a workflow which is compatible with clinical implementation and timescales (2 days for library preparation and 1 to 2 days for sequencing, dependent on read requirements, and scalable computation time), and is expected to become more rapid with continued improvements to sequencing speed and throughput. However, there are limitations to the approach. One consideration applicable to all ultrasensitive approaches is the amount of sequencing power required to interrogate enough individual molecules to enable the detection of low prevalence mutations. In our studies, approximately 80 to 100 million sequence reads per specimen were needed to recover most unique capture events with a minimal redundancy of 2 more reads each. This currently incurs practical limits to the assay in terms of sequencing costs and also the computational time and resources required to analyze the data. Read requirements and computation time should scale linearly with the size of the capture design, so these limitations could be bypassed in exchange for a decreased breadth of the assay. It should also be noted that the number of reads required to detect a mutation by smMIP capture is inversely proportional to its variant allele frequency; although an aver-

References 1. Walter RB, Gooley TA, Wood BL, et al. Impact of pretransplantation minimal residual disease, as detected by multiparametric flow cytometry, on outcome of myeloablative hematopoietic cell transplantation for acute myeloid leukemia. J Clin Oncol. 2011;29(9):1190-1197.

1556

age of 80 million reads were needed to identify mutations at a 0.0625% variant allele frequency, an average of only 6 million reads could reliably detect minor alleles at 1% relative abundance (Online Supplementary Figure S2). Detection of mutations occurring above the limits of detection by smMIP capture can consequently be achieved with more restricted read requirements. Separately, given the limited number of nucleotides which can be efficiently captured by an individual smMIP (~160 bp), it is also an inherent limitation of this approach that large-scale indels and more complex rearrangements, such as sizable internal tandem duplications, inversions, and translocations, cannot be recovered by the technology. Additionally, some regions of the genome are less amenable to smMIP capture due to relative guanine-cytosine (GC) content and other factors,32 a bias which can be improved but not entirely corrected by empiric rebalancing, resulting in uneven performance for a minority of probes. Lastly, although our capture design is relatively large and is directed against high-yield targets, it is not comprehensive and it is likely that mutations identifying MRD will not be determined in all cases of AML. Based on published AML exome sequence data,15 our capture panel is predicted to identify cancer-associated mutations in ~80% of AML specimens. Despite these limitations, we have shown in this pilot study that smMIP capture is in principle well suited to the practical application of monitoring MRD in AML patients. Aside from providing ultrasensitive and quantitative detection of mutations, smMIPs can simultaneously interrogate a large number of high-yield mutational targets for MRDidentifying mutations. This combination of features enables screening for MRD without developing panels specific to individual patients,24-26 and provides functional redundancy and added robustness, both in being able to identify multiple mutations which may serve as markers of disease, and in being able to monitor the emergence of genetically distinct AML subclones which may have different functional properties than the initial malignancy.36,37 The ability to detect MRD-associated mutations with levels of sensitivity far exceeding those achievable by current approaches will require new criteria for assessing the significance of positive findings. Future work will incorporate larger numbers of patients and a greater variety of AML subtypes in order to correlate clinical outcomes with sequencing results, and to begin addressing this outstanding question. Acknowledgments We thank A. Thomas and UW Hematopathology staff for technical help, and C. Pritchard for helpful discussions. Funding This work was supported by grant CA192980 from the National Cancer Institute (to SJS).

2. Schnittger S, Kern W, Tschulik C, et al. Minimal residual disease levels assessed by NPM1 mutation-specific RQ-PCR provide important prognostic information in AML. Blood. 2009;114(11):2220-2231. 3. Walter RB, Buckley SA, Pagel JM, et al. Significance of minimal residual disease before myeloablative allogeneic hematopoietic cell transplantation for AML in first and second complete remission.

Blood. 2013;122(10):1813-1821. 4. Shayegi N, Kramer M, Bornhäuser M, et al. The level of residual disease based on mutant NPM1 is an independent prognostic factor for relapse and survival in AML. Blood. 2013;122(1):83-92. 5. Pastore F, Levine RL. Next-generation sequencing and detection of minimal residual disease in acute myeloid leukemia: ready for clinical practice? JAMA. 2015;

haematologica | 2017; 102(9)


Ultrasensitive leukemia monitoring

314(8):778-780. 6. Miyazaki T, Fujita H, Fujimaki K, et al. Clinical significance of minimal residual disease detected by multidimensional flow cytometry: serial monitoring after allogeneic stem cell transplantation for acute leukemia. Leuk Res. 2012;36(8):998-1003. 7. Hourigan CS, Karp JE. Minimal residual disease in acute myeloid leukaemia. Nat Rev Clin Oncol. 2013;10(8):460-471. 8. Gallo JH, Robson LG, Watson NW, Sharma P, Smith A. Comparison of metaphase and interphase FISH monitoring of minimal residual disease with MLL gene probe: case study of AML with t(9;11). Ann Genet. 1999;42(2):109-112. 9. Ivey A, Hills RK, Simpson MA, et al. Assessment of Minimal Residual Disease in Standard-Risk AML. N Engl J Med. 2016; 374(5):422-433. 10. Schnittger S, Weisser M, Schoch C, Hiddemann W, Haferlach T, Kern W. New score predicting for prognosis in PMLRARA+, AML1-ETO+, or CBFBMYH11+ acute myeloid leukemia based on quantification of fusion transcripts. Blood. 2003;102(8):2746-2755. 11. Paietta E. Minimal residual disease in acute myeloid leukemia: coming of age. Hematol Am Soc Hematol Educ Program. 2012; 2012:35-42. 12. Ladetto M, Brüggemann M, Monitillo L, et al. Next-generation sequencing and realtime quantitative PCR for minimal residual disease detection in B-cell disorders. Leukemia. 2014;28(6):1299-1307. 13. Wu D, Sherwood A, Fromm JR, et al. Highthroughput sequencing detects minimal residual disease in acute T lymphoblastic leukemia. Sci Transl Med. 2012; 4(134):134ra63. 14. Kotrova M, Muzikova K, Mejstrikova E, et al. The predictive strength of next-generation sequencing MRD detection for relapse compared with current methods in childhood ALL. Blood. 2015;126(8):1045-1047. 15. Cancer Genome Atlas Research Network. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):2059-2074. 16. Ley TJ, Mardis ER, Ding L, et al. DNA sequencing of a cytogenetically normal acute myeloid leukaemia genome. Nature. 2008;456(7218):66-72. 17. Mardis ER, Ding L, Dooling DJ, et al. Recurring mutations found by sequencing an acute myeloid leukemia genome. N Engl J Med. 2009;361(11):1058-1066. 18. Ley TJ, Minx PJ, Walter MJ, et al. A pilot study of high-throughput, sequence-based mutational profiling of primary human acute myeloid leukemia cell genomes. Proc Natl Acad Sci USA. 2003; 100(24):1427514280.

haematologica | 2017; 102(9)

19. Salipante SJ, Fromm JR, Shendure J, Wood BL, Wu D. Detection of minimal residual disease in NPM1-mutated acute myeloid leukemia by next-generation sequencing. Mod Pathol. 2014;27(11):1438-1446. 20. Kohlmann A, Nadarajah N, Alpermann T, et al. Monitoring of residual disease by next-generation deep-sequencing of RUNX1 mutations can identify acute myeloid leukemia patients with resistant disease. Leukemia. 2014;28(1):129-137. 21. Zuffa E, Franchini E, Papayannidis C, et al. Revealing very small FLT3 ITD mutated clones by ultra-deep sequencing analysis has important clinical implications in AML patients. Oncotarget. 2015; 6(31):3128431294. 22. Walter MJ, Shen D, Ding L, et al. Clonal architecture of secondary acute myeloid leukemia. N Engl J Med. 2012; 366(12):1090-1098. 23. Hughes AEO, Magrini V, Demeter R, et al. Clonal architecture of secondary acute myeloid leukemia defined by single-cell sequencing. PLoS Genet. 2014; 10(7):e1004462. 24. Klco JM, Miller CA, Griffith M, et al. Association between mutation clearance after induction therapy and outcomes in acute myeloid leukemia. JAMA. 2015; 314(8):811-822. 25. Young AL, Wong TN, Hughes AEO, et al. Quantifying ultra-rare pre-leukemic clones via targeted error-corrected sequencing. Leukemia. 2015;29(7):1608-1611. 26. Malmberg EB, Ståhlman S, Rehammar A, et al. Patient-tailored analysis of minimal residual disease in acute myeloid leukemia using next generation sequencing. Eur J Haematol. 2016;98(1):26-37 27. Duncavage EJ, Tandon B. The utility of next-generation sequencing in diagnosis and monitoring of acute myeloid leukemia and myelodysplastic syndromes. Int J Lab Hematol. 2015;37 Suppl 1:115-121. 28. Hiatt JB, Pritchard CC, Salipante SJ, O’Roak BJ, Shendure J. Single molecule molecular inversion probes for targeted, high-accuracy detection of low-frequency variation. Genome Res. 2013;23(5):843-854. 29. Eijkelenboom A, Kamping E, Kastner-van Raaij A, et al. Reliable next-generation sequencing of formalin-fixed, paraffinembedded tissue using single molecule tags. J Mol Diagn. 2016;18(6):851-863 30. Kinde I, Wu J, Papadopoulos N, Kinzler KW, Vogelstein B. Detection and quantification of rare mutations with massively parallel sequencing. Proc Natl Acad Sci USA. 2011;108(23):9530-9535. 31. Spencer DH, Tyagi M, Vallania F, et al. Performance of common analysis methods for detecting low-frequency single nucleotide variants in targeted next-genera-

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

tion sequence data. J Mol Diagn. 2014; 16(1):75-88. Boyle EA, O’Roak BJ, Martin BK, Kumar A, Shendure J. MIPgen: optimized modeling and design of molecular inversion probes for targeted resequencing. Bioinformatics. 2014;30(18):2670-2672. Ohgami RS, Ma L, Merker JD, et al. Nextgeneration sequencing of acute myeloid leukemia identifies the significance of TP53, U2AF1, ASXL1, and TET2 mutations. Mod Pathol. 2015;28(5):706-714. Costello M, Pugh TJ, Fennell TJ, et al. Discovery and characterization of artifactual mutations in deep coverage targeted capture sequencing data due to oxidative DNA damage during sample preparation. Nucleic Acids Res. 2013;41(6):e67. Forbes SA, Tang G, Bindal N, et al. COSMIC (the Catalogue of Somatic Mutations in Cancer): a resource to investigate acquired mutations in human cancer. Nucleic Acids Res. 2010;38(Database issue):D652-657. Ding L, Ley TJ, Larson DE, et al. Clonal evolution in relapsed acute myeloid leukaemia revealed by whole-genome sequencing. Nature. 2012; 481(7382):506510. Klco JM, Spencer DH, Miller CA, et al. Functional heterogeneity of genetically defined subclones in acute myeloid leukemia. Cancer Cell. 2014;25(3):379-392. Uy GL, Duncavage EJ, Chang GS, et al. Dynamic changes in the clonal structure of MDS and AML in response to epigenetic therapy. Leukemia. 2017;31(4):872-881 Wong TN, Ramsingh G, Young AL, et al. Role of TP53 mutations in the origin and evolution of therapy-related acute myeloid leukaemia. Nature. 2015; 518(7540): 552-555. Yost SE, Alakus H, Matsui H, et al. Mutascope: sensitive detection of somatic mutations from deep amplicon sequencing. Bioinformatics. 2013;29(15):1908-1909. Schmitt MW, Kennedy SR, Salk JJ, Fox EJ, Hiatt JB, Loeb LA. Detection of ultra-rare mutations by next-generation sequencing. Proc Natl Acad Sci USA. 2012; 109(36):14508-14513. Kennedy SR, Schmitt MW, Fox EJ, et al. Detecting ultralow-frequency mutations by Duplex Sequencing. Nat Protoc. 2014; 9(11):2586–2606. Wang K, Ma Q, Jiang L, et al. Ultra-precise detection of mutations by droplet-based amplification of circularized DNA. BMC Genomics. 2016;17:214. Ståhlberg A, Krzyzanowski PM, Jackson JB, Egyud M, Stein L, Godfrey TE. Simple, multiplexed, PCR-based barcoding of DNA enables sensitive mutation detection in liquid biopsies using sequencing. Nucleic Acids Res. 2016;44(11):e105.

1557


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Acute Lymphoblastic Leukemia

Ferrata Storti Foundation

CRISPR-Cas9-induced t(11;19)/MLL-ENL translocations initiate leukemia in human hematopoietic progenitor cells in vivo

Jana Reimer,1 Sabine Knöß,1 Maurice Labuhn,1 Emmanuelle M. Charpentier,2,3 Gudrun Göhring,4 Brigitte Schlegelberger,4 Jan-Henning Klusmann1 and Dirk Heckl1

Haematologica 2017 Volume 102(9):1558-1566

1 Pediatric Hematology & Oncology, Hannover Medical School, Germany; 2Max Planck Institute for Infection Biology, Berlin, Germany; 3The Laboratory for Molecular Infection Medicine Sweden, Umeå University, Sweden and 4Human Genetics, Hannover Medical School, Germany

JR and SK contributed equally to the manuscript J-HK and DH contributed equally to the manuscript

ABSTRACT

C

doi:10.3324/haematol.2017.164046

hromosomal translocations that generate oncogenic fusion proteins are causative for most pediatric leukemias and frequently affect the MLL/KMT2A gene. In vivo modeling of bona fide chromosomal translocations in human hematopoietic stem and progenitor cells is challenging but essential to determine their actual leukemogenic potential. We therefore developed an advanced lentiviral CRISPR-Cas9 vector that efficiently transduced human CD34+ hematopoietic stem and progenitor cells and induced the t(11;19)/MLL-ENL translocation. Leveraging this system, we could demonstrate that hematopoietic stem and progenitor cells harboring the translocation showed only a transient clonal growth advantage in vitro. In contrast, t(11;19)/MLL-ENL-harboring CD34+ hematopoietic stem and progenitor cells not only showed longterm engraftment in primary immunodeficient recipients, but t(11;19)/MLL-ENL also served as a first hit to initiate a monocytic leukemia-like disease. Interestingly, secondary recipients developed acute lymphoblastic leukemia with incomplete penetrance. These findings indicate that environmental cues not only contribute to the disease phenotype, but also to t(11;19)/MLL-ENL-mediated oncogenic transformation itself. Thus, by investigating the true chromosomal t(11;19) rearrangement in its natural genomic context, our study emphasizes the importance of environmental cues for the pathogenesis of pediatric leukemias, opening an avenue for novel treatment options.

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1558

Introduction

Correspondence: Heckl.Dirk@mh-hannover.de or Klusmann.Jan-Henning@mh-hannover Received: January 7, 2017. Accepted: May 31, 2017. Pre-published: June 1, 2017.

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

1558

Reciprocal chromosomal translocations are the causative genetic aberration in almost 60% of cases of pediatric acute myeloid leukemia (AML).1 Among these, rearrangements of the MLL1/KMT2A gene located on chromosome 11q23 are most frequent, accounting for almost 25% of pediatric and 50% of infant AML cases.1 The oncogenicity of MLL fusions has been investigated in various mouse models, including human CD34+ xenografts, which provided strong evidence that MLL fusion oncogenes are sufficient to transform human hematopoietic stem and progenitor cells (HSPC).2-4 These studies also highlighted the environmental influence on disease phenotype and the role of the MLL fusion partner on the overall oncogenicity. However, the results were mainly obtained using retroviral expression systems, which invariably express one fusion protein at non-physiological levels, and neglect the loss of the wild-type alleles of both involved genes and a potential contribution of the reciprocal product of the translocation.5,6 Investigation of MLL rearrangements via knock-in in human HSPC supported their oncogenicity at haematologica | 2017; 102(9)


CRISPR-Cas9-induced t(11;19)/MLL-ENL translocations

endogenous expression levels, however, the loss of one wild-type allele of each fusion partner, and potential involvement of a reciprocal fusion product, still remain elusive with this approach.4 It is therefore desirable to investigate the oncogenicity of bona fide MLL rearrangements in primary human HSPC both in vitro and in vivo, which can be facilitated with genome editing technologies.7-10 The CRISPR-Cas9 system has only recently been successfully established to modify functional human HSPC.11,12 It was shown to allow the generation of chromosomal inversions in vitro9,10,13 and in murine in vivo models,14 and may thus overcome efficacy hurdles that may have caused the failure of models based on transcription activator-like effector nucleases (TALEN).15 While attempts to recapitulate transformation of primary human HSPC by chromosomal rearrangements in vitro failed,15 and thereby raised the question of whether the efficacy of the technology or an inherent protection of HSPC against chromosomal rearrangements caused the failure,15,16 in vivo experiments with CRISPR-Cas9-induced chromosomal translocations in human HSPC may have the power to answer these open questions, but have not yet been accomplished. Here we successfully generated chromosomal rearrangements (t[11;19]/MLL-ENL) in CD34+ HSPC, resulting in clonal outgrowth in vitro and monocytic and Blineage leukemia in vivo, which we accomplished with an improved lentiviral CRISPR-Cas9 system to generate chromosomal rearrangements based on our former work.17 Furthermore, our study highlights the impact of non-cell-autonomous signals influencing not only the phenotype but the overall transformation of HSPC by MLL rearrangements. Thus, our study presents the first human de novo leukemia model with CRISPR-Cas9-engineered chromosomal translocations and highlights the power of this advanced approach.

Methods Plasmids and viral particle production

Xenotransplantation All animal experiments were approved by the local authorities. For transplantation, 4x105 CD34+ cells were prestimulated and transduced as described above. Twenty-four hours after transduction, the HSPC were transplanted intravenously into sublethally irradiated (2.5 Gy) male or female NSGS mice20 (age: 8-12 weeks).

Flow cytometry Cells were stained with antibodies specific for human CD45 (APC, V500), CD33 (PE), CD14 (APC-Cy7), CD11b (APC), CD64 (PE), CD117 (PE-Cy7), CD19 (PE-Cy7), CD3 (APC-Cy7), CD10 (BV605), CD20 (V450), CD22 (APC), and CD34 (PerCP-Cy5.5) (all from BD Biosciences). Dead cells were excluded using 4',6diamidin-2-phenylindol (DAPI) counterstaining, where applicable. Cells were analyzed on a FACS Canto flow cytometer (BD Biosciences), and data were analyzed with FlowJo V10 software.

RNA extraction and polymerase chain reaction Total RNA was purified from cells with the RNeasy Micro or Mini Kit (Qiagen), complementary DNA (cDNA) was synthesized using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems), according to the manufacturer’s instructions. The primers are summarized in Online Supplementary Table S3.

Genomic DNA extraction and polymerase chain reaction Genomic DNA was extracted using the QIAmp DNA Mini or Micro Kit (Qiagen) according to the manufacturer’s protocols. Detection of the genomic breakpoint polymerase chain reaction was performed with Extensor 2x Master Mix (Thermo Scientific). Primers are summarized in Online Supplementary Table S3.

T7 endonuclease I based surveyor assay Polymerase chain reaction primers containing the on-target and off-target of MLL and ENL are listed in Online Supplementary Table S3. The T7 endonuclease I (T7E-I) assay was performed according to the manufacturer’s protocols (New England Biolabs). Digested fragments were separated by DNA gel electrophoresis and imaged with a BioRad GelDoc™ XR+ imaging system. Absolute quantification of DNA fragments was done with the Image Lab 3.0 software (BioRad). The band intensity of the DNA marker (Thermo Scientific) was correlated to its specific amount of input, and subsequently, the absolute quantities of all DNA fragments were defined accordingly and the ratios of cut and uncut fragments were calculated.

A detailed description of the vector construction is provided in the Online Supplementary Methods. CRISPR-Cas9 target sites were selected using the CCTop selection tool.19 Lentiviral vectors were provided via Addgene (# 69146-69148, 69212, 69215, 8939289395). Lentiviral particles were produced as described before.17

Statistical analysis

Reporter assay-based testing of the spacers

Statistical analysis was performed with GraphPad Prism software (Graphpad Software).

Reporter-based single-guide RNA (sgRNA) efficacy testing was performed as described elsewhere.17

Results Cell culture Cell lines were maintained as described in the Online Supplementary Methods. Human CD34+ cells were isolated from cord blood using the CD34 MicroBead Kit (Miltenyi Biotec) according to the manufacturer’s instructions. The cord blood was provided with the parents’ consent by the Department of Gynecology and Obstetrics, Hannover Medical School, and experiments were approved by the local ethics committee. Details on CD34+ cell maintenance and transduction are described in the Online Supplementary Methods section. Colony-forming unit assays were performed according to manufacturer’s instructions. Upon initial plating, 1x104 cells were seeded. For replating a total of 3x104 cells per plate were seeded. haematologica | 2017; 102(9)

Advanced lentiviral dual single-guide RNA CRISPR-Cas9 vectors for the generation of chromosomal rearrangements With the aim of elucidating the transformative nature of endogenous chromosomal rearrangements in primary human HSPC, we developed an advanced, all-in-one lentiviral CRISPR-Cas9 system with two sgRNA expression cassettes (L-CRISPR-CTN=CRISPR-TranslocationsmNeon) (Figure 1A). We introduced an enhanced sgRNA backbone18 and enhancer elements to boost genomic RNA production,21 resulting in significantly higher titers as tested 1559


J. Reimer et al.

on multiple hematopoietic cell lines (Figure 1B; Online Supplementary Figure S1). Utilizing fluorescence reporter-based spacer testing (Online Supplementary Figure S2), we established highly efficient sgRNAs (>80% cleavage activity) targeting MLL/KMT2A and ENL intronic sequences to generate the t(11;19)/MLL-ENL translocation (Figure 1C; Online

Supplementary Table S1). To prospectively achieve CRISPRCas9-induced chromosomal rearrangement from a single vector (Figure 1A), the knock-out efficacy of the various selected sgRNAs expressed from a H1 promoter was tested. Only minor, non-significant differences in the cleavage efficacy were detectable with sgRNA expression either from a H1 or U6 promoter (Figure 1D). Importantly, the activity of

A

B

C

D

E

Figure 1. An improved lentiviral vector system for generation of CRISPR-Cas9-induced chromosomal rearrangements. (A) Schematic presentation of lentiviral vector architecture including genomic RNA-generating promoter assembly: published architecture (L-CRISPR) (i), improved architecture with cytomegalovirus enhancer (CMV) and simian virus 40 enhancer (SV40) and exchangeability of the hU6 promoter for a H1 promoter (L40C) (ii), and lentiviral vector for dual sgRNA delivery (LCRISPR-CTN (iii). (B) Analysis of viral titers in three independent cell lines with two different sgRNAs and three replicates each. (C) Knock-out efficacies (fluorescence reporter assay) of sgRNAs targeting intronic sequences of ENL and MLL. Selected sgRNAs are marked. (D) Knock-out efficacies of selected sgRNAs expressed from a human U6 or H1 promoter, as indicated. Knock-out efficacies of selected sgRNAs in L-CRISPR-CTN dual sgRNA configuration (DV) (neg ctrl= anti-luciferase (Luc) sgRNA, pos ctrl= Tet2 sgRNA). (E) T7-endonuclease-I assay for on-target sites and the top five predicted off-target sites in HEL cells (OT-1-OT-5). Indel frequencies at endogenous loci are indicated below. Analysis in cells transduced with targeting (+) and Luc (-) sgRNAs: MLL-I9-#1(i) and ENL-I1-#4 (ii).

1560

haematologica | 2017; 102(9)


CRISPR-Cas9-induced t(11;19)/MLL-ENL translocations

the sgRNA was retained in the dual sgRNA vector configuration (Figure 1D). No recombination of the integrated provirus was observed (Online Supplementary Figure S3), warranting high confidence delivery of both sgRNAs. Off-target cleavage activity is a major concern with the application of genome editing, but is mostly abrogated by three or more mismatches between the spacer and protospacer.22,23 T7E-I assays for the top five off-target sites and the on-target sites of our pre-selected sgRNAs verified high on-target and no detectable off-target activity at the endogenous loci (Figure 1E, Online Supplementary Table S2). Based on these results, we tested the generation of chromosomal rearrangements [t[11;19]; L-CRISPR-CTN(11;19)] in hematopoietic cells (Figure 2A). In L-CRISPRCTN(11;19)-transduced K562 cells, we could readily detect expression of the MLL-ENL and reciprocal ENL-MLL transcript in the bulk population (Figure 2B,C). Amplification of the genomic breakpoint and sequence analysis further verified polyclonal t(11;19) induction at the targeted genomic sequences (Online Supplementary Figure S4). We could thereby establish an all-in-one lentiviral CRISPR-Cas9 system for the efficient induction of chromosomal translocations (LCRISPR-CTN).

L-CRISPR-CTN-induced t(11;19) translocations increase the re-plating efficiency of primary human hematopoietic stem and progenitor cells in vitro To determine the impact of endogenous t(11;19) on HSPC, we transduced cord blood-derived CD34+ HSPC (16.6Âą5.2%, n=10). FACS-sorted cells grown in methylcellulose were tested for MLL-ENL expression at the first replating and transcript identity was validated via sequencing (Online Supplementary Figure S5). Of note, vector expression was not monitored over time since genome editing does not rely on continuous expression of the Cas9 or the sgRNA. In three independent experiments, MLLENL messenger RNA (mRNA) was detectable, resulting in a rearrangement efficacy of at least 1.6x10-3 (Âą0.26x10-3). Upon serial re-plating we detected the extended plating capacity of t(11;19)-containing cells (Figure 2D), accompanied by robust expression of the MLL-ENL transcript (Figure 2E, Online Supplementary Figure S6) and detection of the genomic breakpoint (Figure 2F, Online Supplementary Figure S6). The latter was not detectable at the first re-plating, likely due to lower sensitivity at the genomic level compared to mRNA expression. Our experiments thus provide evidence that CRISPR-Cas9-induced t(11;19)

B

A

C

D

E

G

F

Figure 2. CRISPR-Cas9-induced MLL-ENL rearrangements cause clonal expansion of human CD34+ hematopoietic stem and progenitor cells. (A) Schematic depiction of CRISPR-Cas9-induced chromosomal rearrangements at the MLL and ENL loci. (B) Reverse transcriptase polymerase chain reaction-based detection of MLLENL transcript in K562 cells. Ctrl = MLL-I9-#1 + Luc sgRNAs. (C) RT-PCR-based detection of reciprocal ENL-MLL transcript in K562 cells. Ctrl = MLL-I9-#1 + Luc sgRNA. (D) Serial plating of CD34+ HSPC after transduction with L-CRISPR-CTN. (E) Detection of MLL-ENL expression in CD34+ HSPCs at fifth plating. Control (MLLI9-#1 + Luc sgRNAs) template from third plating. (F) Detection of the genomic MLL-ENL breakpoint in CD34+ HSPCs at fifth plating (control (MLL-I9-#1 + Luc sgRNAs) template from second plating) compared to the first plating. (G) Analysis of MLL target genes in MLL-ENL-expressing cells (fifth plating) compared to controls (third plating). Differential regulation marked above. Ctrl = MLL-I9-#1 + Luc sgRNAs.

haematologica | 2017; 102(9)

1561


J. Reimer et al.

translocations can provide self-renewal capacity to human HSPC. However, the cord blood CD34+ cells with t(11;19) formed normal hematopoietic colonies and eventually ceased proliferating. Results in methylcellulose were further supported by the transient clonal outgrowth of MLLENL-expressing cells in one out of four experiments performed in liquid culture (Online Supplementary Figure S7). Long-term tracking of MLL-ENL-expressing cells in other experiments without growth advantage indicates the need for additional stimuli to induce transformation (Online Supplementary Figure S8A,B). When clonal outgrowth was observed it was accompanied by a robust up-regulation of known downstream effectors of MLL-ENL, such as HOXA9, HOXA10, MEIS1, and PBX3 (Figure 2G, Online Supplementary Figure S7D),24 which was absent in samples without clonal outgrowth (Online Supplementary Figure S8C). In contrast to former studies, genes associated with a leukemic stem cell phenotype in mice (CBX5, HMGB3, MYBL2)24 were not upregulated in samples with clonal outgrowth (Online Supplementary Figure S7D). Overall, these experiments underline both the potential and insufficiencies of endogenous t(11;19) to modulate selfrenewal and growth of human HSPC in vitro. This is in line with a recent report on TALEN-induced MLL rearrangements in CD34+ HSPC in vitro.15 This cumulative evidence points towards a non-cell-autonomous component in cellular transformation by MLL rearrangements beyond phenotypic manifestation.

In vivo environment affects the oncogenic transformation of primary human hematopoietic stem and progenitor cells by t(11;19) To further test this hypothesis, we performed transplantation of freshly-transduced, non-sorted L-CRISPRCTN(11;19) and L-CRISPR-CTN(ctrl) CD34+ cells into immunodeficient mice (Figure 3A, Online Supplementary Figure S9). At weeks 25, 26, and 27 after transplantation, three moribund mice were analyzed (Figure 3B). One of the mice presented with intermediate human engraftment, mixed lineage reconstitution and only moderate expression of the vector backbone, excluding leukemia as the cause of sickness (Online Supplementary Figure S10AC). Two of the mice presented with a hCD45+CD33+ monocytic phenotype (Figure 3C, Online Supplementary Figure S10B), robust vector expression (Figure 3C, Online Supplementary Figure S10B), and the MLL rearrangement was confirmed by fluorescence in situ hybridization in cultured cells from one mouse (Figure 3E, MLL-split probe 7% positivity). An excess of monocytic cells with immature features in the bone marrow was verified by nuclear staining (>80%, Figure 3D). Severe infiltration of myeloid cells in the liver further supported hematologic disease (Figure 3F). Sequence analysis of the genomic MLL-ENL breakpoints detected in all analyzed mice (diseased and healthy, Sanger sequencing) indicated a clonal origin and outgrowth of t(11;19)-containing cells (Figure 3G). MLLENL and ENL-MLL breakpoints (Figure 3H), expression and identity of MLL-ENL and ENL-MLL transcripts (Figure 3I, Online Supplementary Figure S11), and presence of human cells expressing the introduced vector were confirmed (Figure 3C, Online Supplementary Figures S10 and S12A). Notably, while no correlation between MLLENL expression and human cell content was observed, MLL-ENL expression significantly correlated with the 1562

percentage of vector-expressing cells, further supporting our approach (Online Supplementary Figure S12B,C). To validate a malignant phenotype, secondary transplantation of the monocytic samples was performed, assuming that healthy committed myelo-monocytic progenitors would fail to engraft, which was indeed the case for two control mice transplanted with L-CRISPRCTN(ctrl)-transduced CD34+ cells (Figure 4A). Surprisingly, five out of six recipients from both donors with a monocytic phenotype presented with disease symptoms 15-16 weeks after transplantation (Figure 4A). The immature monocytic phenotype was confirmed, with CD64 expression, absence of CD14 and CD15 expression, and partial CD117 expression by flow cytometric analysis, as well as by morphological analyses of hepatic infiltration (Figure 4B,C, Online Supplementary Figure S13). Of note, while disease latency was slightly shortened (25-26 weeks versus 16-17 weeks), the absence of blast morphology and CD34 expression indicated no progression of the monocytic leukemia-like disease to a more immature AML phenotype, despite the transplantation-induced stress. Next, we asked whether EVI1 expression – an indicator of stem cell origin in MLL-rearranged leukemia25-27 – and known downstream MLL targets were deregulated in the mice. While EVI1 was not detectable in our samples or MLL-rearranged control cell lines, we found upregulation of PBX3 in the majority of mice. Some samples also showed upregulation of HOXA10, MEIS1, and MYB (Online Supplementary Figure S14). Our experiments thereby provide compelling evidence that endogenous t(11;19) can initiate a monocytic leukemialike phenotype in our xenotransplantation setting, but lacks the capacity to initiate an immature AML. The detection of t(11;19) in non-diseased mice also strongly indicates that additional events are either required for the development of frank leukemia or influence the latency of disease appearance. To test this hypothesis of inherent stress and environmental cues as contributing factors in disease manifestation caused by t(11;19), we harvested bone marrow from primary recipients freshly-transplanted with L-CRISPRCTN(11;19)- and L-CRISPR-CTN(ctrl)-transduced CD34+ cells 8-10 weeks after transplantation – before the onset of leukemia – and transplanted the bone marrow into secondary recipients. Supposedly-clonal genomic MLL-ENL breakpoints were detectable in the bone marrow of two out of three mice at the time of transplantation (Online Supplementary Figure S15). Indeed, we monitored human cell engraftment in the secondary animals and observed rapid expansion indicating leukemic transformation (Figure 4A). Detailed analysis of the cells revealed the development of B-cell acute lymphoblastic leukemia (B-ALL) with expression of CD19, CD22, CD33 and partial expression of CD20 and CD34 (Figure 4D). Of note, the expression of CD33 is a common feature of B-ALL cells.28,29 The occurrence of B-ALL was further confirmed by morphological analysis of bone marrow cells showing an excess of blasts (Figure 4E). The presence of the MLL rearrangement was additionally verified by fluorescence in situ hybridization (Figure 4F, MLL-split probe 67% positivity). Interestingly, the genomic MLL-ENL breakpoint revealed that the leukemic clone developed independently from the dominant clone in the primary recipients, further supporthaematologica | 2017; 102(9)


CRISPR-Cas9-induced t(11;19)/MLL-ENL translocations

B

A

C

D

E

F

G

H

I

Figure 3. CRISPR-Cas9-induced MLL-ENL rearrangements are leukemogenic in a CD34+ hematopoietic stem and progenitor cell xenotransplantation model. (A) Schematic depiction of the xenotransplantation model. (B) Survival of mice transplanted with L-CRISPR-CTN-transduced CD34+ HSPC. Mice that succumbed to nonhematopoietic disease were censored and are indicated (ticked). (C) Flow cytometric analysis of one mouse with hematopoietic disease compared to a control, with markers as indicated (ctrl = MLL-I9-#1 + Luc sgRNAs). (D) Bone marrow (BM) cytospin analysis of a diseased mouse (MGG, 1000X). (E) Detection of an MLL translocation with fluorescence in situ hybridization on interphase nuclei (Vysis LSI MLL probe; Abbott Laboratories) in BM cells from one diseased mouse. (F) Histopathological analysis of liver tissue from a healthy control mouse (ctrl = MLL-I9-#1 + Luc sgRNA) (top) and a mouse with monocytic leukemia-like disease transplanted with L-CRISPR-CTN(11;19)-containing CD34+ HSPC (bottom) (HE, 100x). (G) Alignment of Sanger sequencing-derived genomic t(11;19)/MLL-ENL breakpoints of mice with a detectable MLL-ENL breakpoint. (H) MLL-ENL and ENL-MLL genomic breakpoints detected in the BM of two mice with a monocytic leukemia-like disease. (I) Expression of the MLL-ENL (left) and reciprocal ENL-MLL (right) fusion genes, measured by quantitative polymerase chian reaction from the BM of mice with a detectable MLL-ENL breakpoint compared to control mice (MLL-I9-#1 + Luc sgRNAs).

haematologica | 2017; 102(9)

1563


J. Reimer et al.

A

B

C

D

F

E

G

Figure 4. In vivo environment affects the oncogenic transformation of primary human hematopoietic stem and progenitor cells by t(11;19) (A) Survival of serially transplanted mice with human cell engraftment. Donor: primary recipients with [L-CRISPR-CTN(11;19)] a monocytic leukemia-like disease (Mono #1/#2), a healthy mouse with a detectable MLL-ENL breakpoint but no disease in the primary recipient, and two control donors (MLL-I9-#1 + Luc sgRNAs) (n = 3 per donor). (B) Analysis of bone marrow (BM) cell morphology (left: MGG, 1000x) and liver histopathology (right: HE, 100X) showing severe infiltration of a secondary recipient transplanted with monocytic leukemia-like disease cells. (C) Flow cytometry analysis of monocytic leukemia cells for monocytic and progenitor cell surface marker expression. (D) Flow cytometry analysis of B-ALL cells. (E) Analysis of BM cell morphology (MGG, 1000x) of a secondary recipient with B-ALL. (F) Detection of an MLL translocation with fluorescence in situ hybridization on interphase nuclei of B-ALL cells (Vysis LSI MLL probe; Abbott Laboratories). (G) Alignment of Sanger sequencingderived genomic t(11;19)/MLL-ENL breakpoints of mice with B-ALL (Secondary) compared to the healthy primary (Primary) mouse.

1564

haematologica | 2017; 102(9)


CRISPR-Cas9-induced t(11;19)/MLL-ENL translocations

ing our hypothesis that environmental signals may alter the transformation of MLL-ENL-harboring cells (Figure 4G). Utilizing serial transplantation and tracking of NHEJ-scarring at genomic breakpoints, as with our system, potentially provides a tool to investigate the clonal evolution of leukemia with potentially variable stress conditions, such as drug treatments, in vivo.

Discussion By leveraging CRISPR-Cas9 genome editing, we induced bona fide chromosomal rearrangements in primary human HSPC in vivo, resulting in a de novo leukemia-like disease reflecting all the oncogenic properties of MLL-rearrangements: namely the combined, endogenous expression of both resulting fusion mRNA and concurrent loss of one wild-type allele of each fusion partner. To achieve efficient genome editing in CD34+ HSPC we improved our formerly constructed lentiviral CRISPR-Cas9 delivery system17 for the generation of chromosomal translocations and transduction of cord blood-derived CD34+ HSPC. Successful transduction of human HSPC with our system reproducibly resulted in both MLL-ENL and ENL-MLL mRNA expression, providing strong evidence for the formation of true t(11;19)/MLL-ENL translocations. Intriguingly, functional studies revealed only a transient and variable growth advantage affecting the minority of cultures. No full transformation was observed in liquid cultures or methylcellulose-based in vitro assays using CD34+ HSPC, which is in line with a recent study utilizing TALEN to induce MLL translocations or knock-in of AF9/ENL cDNA into the MLL locus in CD34+ HSPC.4,15 With MLL-ENL causing cellular senescence32 and an intrinsic protection of early HSPC against MLL-ENL transformation,16 MLL rearrangements may only serve as a first hit insufficient to cause full transformation without additional oncogenic events. Isolation of clonal lines and testing of additional cytokine combinations may help to understand the lack of in vitro transformation in consecutive studies. Leukemic transformation in patients lacking additional known driver mutations34 prompted us to hypothesize that environmental cues may not only affect the phenotype of MLL rearrangement-induced leukemia,3 but may also determine the overall transformation capacity. Of note, MLL rearrangements are particularly frequent in pediatric and infant AML,1 in which both the environment and the cell of origin differ from those in adults.35,36 The latter was already shown to affect the transformation capacity of retrovirally expressed MLL-AF9.36 Supporting the idea that yet unknown external stimuli of the in vivo environment alter transformation permissiveness, we noted successful engraftment and persistence of t(11;19)-harboring cells after long-term observation, exceeding the time of the in vitro experiments. This engraftment culminated in the development of a monocytic leukemia-like disease in primary recipients, contrasting with the lack of transformation in vitro, despite similar input cell numbers. Taking into account that leukemic transformation exceeded culture time about 3-fold, the lack of supportive signals for maintaining a preleukemic cell in vitro, or a prolonged time-frame for the acquisition of additional genetic lesions in vivo could explain the observed phenomenon, although similar conditions have been used to transform human HSPC with retroviral overexpression of MLL fusion oncogenes.2,3 The development of B-ALL in secondary recipients of a healthy primary haematologica | 2017; 102(9)

donor may either indicate a disturbance of cellular homeostasis contributing to leukemic outgrowth, or result from B-ALL having a higher proportion of leukemia-initiating cells.39 The detection of MLL rearrangements in mice without leukemia-like disease in the same time-frame strongly indicates a progressive acquisition of cooperating mutations needed for full leukemic transformation. Sequencing of NRAS and KRAS hotspots, which are frequently mutated in MLL-rearranged leukemia,34 did not reveal any mutations (data not shown). More comprehensive sequencing to detect additional oncogenic hits, tracking of clonal t(11;19) isolates in serial transplantations, and a larger cohort are required to identify cooperating events in the future. Former studies have made major contributions to the understanding of leukemogenesis induced by MLL fusion oncogenes.2,3,24 However, several aspects may have limited the grasp on MLL rearrangement-guided leukemogenicity and could hamper the development of novel treatment options. First, retroviral expression of MLL fusion oncogenes poses the risk of overexpression-induced phenotypes, which may be amplified by the use of murine models over human xenotransplantation models.2,3 To overcome these limitations of retroviral studies, another recent study has utilized TALEN-induced DNA-DSB-supported knock-in of AF9 and ENL cDNA into the MLL locus in HSPC, thereby recapitulating endogenous expression of the fusion oncogene.4 Yet, lack of expression of the reciprocal fusion gene, in some cases shown to be essential for the phenotype,5 loss of one wild-type allele of the fusion partner, potential effects of the translocation of enhancer elements,40 and microRNA binding sites in the untranslated region of the fusion partner can only be covered by the induction of true chromosomal translocations, as shown in our study. Furthermore, an excess of input cells insufficiently reflecting the clonal disease development and homeostatic mechanisms regulating clonal outgrowth in patients could have influenced the outcomes of former studies, which may be indicated by our detection of chromosomal translocations in mice not showing disease. Notably, CRISPR-Cas9induced NHEJ-scaring of the chromosomal breakpoints allowed tracking of clonal fluctuation and revealed a clonal shift during leukemic outgrowth in secondary recipients. Our study reporting a humanized, CRISPR-Cas9-induced cancer model with in vivo transformation uncovers new aspects of the oncogenic potency and limitations of endogenous MLL rearrangements in human HSPC, and advances disease modeling closer to patient-specific disease progression than before. Leveraging our approach to generate precise cancer models with the CRISPR-Cas9 system will allow more detailed analysis of potential homeostatic mechanisms and oncogenic cooperations during leukemic transformation, as well as the development of targeted therapies and investigation of drug resistance mechanisms. Acknowledgments We thank Dr. Dr. A. Schambach and Michelle Ng for discussions and D. Trono of EPFL, Lausanne, Switzerland, for kindly providing both pMD2.G (Addgene plasmid 12259) and psPAX2 (Addgene plasmid 12260). DH is supported by the German Cancer Aid (111743). JHK is a fellow of the Emmy NoetherProgramme from the DFG (KL-2374/2-1). This work was supported by grants to DH and JHK from the DFG (HE-7482/1-1, KL-2374/2-1; KL-2374/1-3) and to GG and BS for the Cluster of Excellence REBIRTH (DFG, EXC 62/3). JR and ML were supported by the Hannover Biomedical Research School. 1565


J. Reimer et al.

References 1. Creutzig U, van den Heuvel-Eibrink MM, Gibson B, et al. Diagnosis and management of acute myeloid leukemia in children and adolescents: recommendations from an international expert panel. Blood. 2012;120 (16):3187-3205. 2. Barabe F, Kennedy JA, Hope KJ, Dick JE. Modeling the initiation and progression of human acute leukemia in mice. Science. 2007;316(5824):600-604. 3. Wei J, Wunderlich M, Fox C, et al. Microenvironment determines lineage fate in a human model of MLL-AF9 leukemia. Cancer Cell. 2008;13(6):483-495. 4. Buechele C, Breese EH, Schneidawind D, et al. MLL leukemia induction by genome editing of human CD34+ hematopoietic cells. Blood. 2015;126(14):1683-1694. 5. Bursen A, Schwabe K, Ruster B, et al. The AF4.MLL fusion protein is capable of inducing ALL in mice without requirement of MLL.AF4. Blood. 2010;115(17):3570-3579. 6. Chen W, Kumar AR, Hudson WA, et al. Malignant transformation initiated by MllAF9: gene dosage and critical target cells. Cancer Cell. 2008;13(5):432-440. 7. Jinek M, Chylinski K, Fonfara I, Hauer M, Doudna JA, Charpentier E. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science. 2012;337(6096):816-821. 8. Piganeau M, Ghezraoui H, De Cian A, et al. Cancer translocations in human cells induced by zinc finger and TALE nucleases. Genome Res. 2013;23(7):1182-1193. 9. Torres R, Martin MC, Garcia A, Cigudosa JC, Ramirez JC, Rodriguez-Perales S. Engineering human tumour-associated chromosomal translocations with the RNA-guided CRISPR-Cas9 system. Nat Commun. 2014;5:3964. 10. Choi PS, Meyerson M. Targeted genomic rearrangements using CRISPR/Cas technology. Nat Commun. 2014;5:3728. 11. Mandal PK, Ferreira LM, Collins R, et al. Efficient ablation of genes in human hematopoietic stem and effector cells using CRISPR/Cas9. Cell Stem Cell. 2014;15(5): 643-652. 12. Gundry MC, Brunetti L, Lin A, et al. Highly Efficient genome editing of murine and

1566

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

human hematopoietic progenitor cells by CRISPR/Cas9. Cell Rep. 2016;17(5):14531461. Vanden Bempt M, Demeyer S, Mentens N, et al. Generation of the Fip1l1-Pdgfra fusion gene using CRISPR/Cas genome editing. Leukemia. 2016;30(9):1913-1916. Blasco RB, Karaca E, Ambrogio C, et al. Simple and rapid in vivo generation of chromosomal rearrangements using CRISPR/Cas9 technology. Cell Rep. 2014;9(4):1219-1227. Breese EH, Buechele C, Dawson C, Cleary ML, Porteus MH. Use of genome engineering to create patient specific MLL translocations in primary human hematopoietic stem and progenitor cells. PLoS One. 2015;10(9): e0136644. Ugale A, Norddahl GL, Wahlestedt M, et al. Hematopoietic stem cells are intrinsically protected against MLL-ENL-mediated transformation. Cell Rep. 2014;9(4):1246-1255. Heckl D, Kowalczyk MS, Yudovich D, et al. Generation of mouse models of myeloid malignancy with combinatorial genetic lesions using CRISPR-Cas9 genome editing. Nat Biotechnol. 2014;32(9):941-946. Chen B, Gilbert LA, Cimini BA, et al. Dynamic imaging of genomic loci in living human cells by an optimized CRISPR/Cas system. Cell. 2013;155(7):1479-1491. Stemmer M, Thumberger T, Del Sol Keyer M, Wittbrodt J, Mateo JL. CCTop: an intuitive, flexible and reliable CRISPR/Cas9 target prediction tool. PLoS One. 2015;10(4): e0124633. Wunderlich M, Chou FS, Link KA, et al. AML xenograft efficiency is significantly improved in NOD/SCID-IL2RG mice constitutively expressing human SCF, GM-CSF and IL-3. Leukemia. 2010;24(10):1785-1788. Schambach A, Mueller D, Galla M, et al. Overcoming promoter competition in packaging cells improves production of self-inactivating retroviral vectors. Gene Ther. 2006;13(21):1524-1533. Fu Y, Foden JA, Khayter C, et al. High-frequency off-target mutagenesis induced by CRISPR-Cas nucleases in human cells. Nat Biotechnol. 2013;31(9):822-826. Hsu PD, Scott DA, Weinstein JA, et al. DNA targeting specificity of RNA-guided Cas9 nucleases. Nat Biotechnol. 2013;31(9):827832.

24. Somervaille TC, Matheny CJ, Spencer GJ, et al. Hierarchical maintenance of MLL myeloid leukemia stem cells employs a transcriptional program shared with embryonic rather than adult stem cells. Cell Stem Cell. 2009;4(2):129-140. 25. Arai S, Yoshimi A, Shimabe M, et al. Evi-1 is a transcriptional target of mixed-lineage leukemia oncoproteins in hematopoietic stem cells. Blood. 2011;117(23):6304-6314. 26. Bindels EM, Havermans M, Lugthart S, et al. EVI1 is critical for the pathogenesis of a subset of MLL-AF9-rearranged AMLs. Blood. 2012;119(24):5838-5849. 27. Stavropoulou V, Kaspar S, Brault L, et al. MLL-AF9 expression in hematopoietic stem cells drives a highly invasive AML expressing EMT-related genes linked to poor outcome. Cancer Cell. 2016;30(1):43-58. 28. Mejstrikova E, Kalina T, Trka J, Stary J, Hrusak O. Correlation of CD33 with poorer prognosis in childhood ALL implicates a potential of anti-CD33 frontline therapy. Leukemia. 2005;19(6):1092-1094. 29. Suggs JL, Cruse JM, Lewis RE. Aberrant myeloid marker expression in precursor Bcell and T-cell leukemias. Exp Mol Pathol. 2007;83(3):471-473. 30. Takacova S, Slany R, Bartkova J, et al. DNA damage response and inflammatory signaling limit the MLL-ENL-induced leukemogenesis in vivo. Cancer Cell. 2012;21(4):517-531. 31. Andersson AK, Ma J, Wang J, et al. The landscape of somatic mutations in infant MLLrearranged acute lymphoblastic leukemias. Nat Genet. 2015;47(4):330-337. 32. Orkin SH, Zon LI. Hematopoiesis: an evolving paradigm for stem cell biology. Cell. 2008;132(4):631-644. 33. Horton SJ, Jaques J, Woolthuis C, et al. MLLAF9-mediated immortalization of human hematopoietic cells along different lineages changes during ontogeny. Leukemia. 2013;27(5):1116-1126. 34. Rehe K, Wilson K, Bomken S, et al. Acute B lymphoblastic leukaemia-propagating cells are present at high frequency in diverse lymphoblast populations. EMBO Mol Med. 2013;5(1):38-51. 35. Groschel S, Sanders MA, Hoogenboezem R, et al. A single oncogenic enhancer rearrangement causes concomitant EVI1 and GATA2 deregulation in leukemia. Cell. 2014;157(2): 369-381.

haematologica | 2017; 102(9)


ARTICLE

Acute Myeloid Leukemia

Reduced hematopoietic stem cell frequency predicts outcome in acute myeloid leukemia

Wenwen Wang,1,8 Thomas Stiehl,2 Simon Raffel,1,3,4 Van T. Hoang,1,9 Isabel Hoffmann,1 Laura Poisa-Beiro,1 Borhan R. Saeed,1 Rachel Blume,1 Linda Manta,1 Volker Eckstein,1 Tilmann Bochtler,1,5 Patrick Wuchter,1,10 Marieke Essers,3,4 Anna Jauch,6 Andreas Trumpp,3,4,7 Anna Marciniak-Czochra,2 Anthony D. Ho1 and Christoph Lutz1,7

Department of Medicine V, Heidelberg University, Germany; 2Institute of Applied Mathematics, Interdisciplinary Center for Scientific Computing (IWR), BIOQUANT, Heidelberg University, Germany; 3Division of Stem Cells and Cancer, Deutsches Krebsforschungszentrum (DKFZ), Heidelberg, Germany; 4Heidelberg Institute for Stem Cell Technology and Experimental Medicine (HI-STEM gGmbH), Germany; 5Clinical Cooperation Unit Molecular Hematology/Oncology, Deutsches Krebsforschungszentrum (DKFZ), Heidelberg, Germany; 6Institute of Human Genetics, Heidelberg University, Germany; 7German Cancer Consortium (DKTK), Heidelberg, Germany; 8Present Address: Department of Oncology, Nanjing Medical University Affiliated Wuxi Second Hospital, Jiangsu, China; 9Present Address: Georg-Speyer-Haus, Institute for Tumor Biology and Experimental Therapy, Frankfurt am Main, Germany and 10Present Address: Institute of Transfusion Medicine and Immunology, Medical Faculty Mannheim, Heidelberg University; German Red Cross Blood Service Baden-Württemberg – Hessen, Germany 1

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(9):1567-1577

ABSTRACT

I

n patients with acute myeloid leukemia and low percentages of aldehyde-dehydrogenase-positive cells, non-leukemic hematopoietic stem cells can be separated from leukemic cells. By relating hematopoietic stem cell frequencies to outcome we detected poor overall- and disease-free survival of patients with low hematopoietic stem cell frequencies. Serial analysis of matched diagnostic and follow-up samples further demonstrated that hematopoietic stem cells increased after chemotherapy in patients who achieved durable remissions. However, in patients who eventually relapsed, hematopoietic stem cell numbers decreased dramatically at the time of molecular relapse demonstrating that hematopoietic stem cell levels represent an indirect marker of minimal residual disease, which heralds leukemic relapse. Upon transplantation in immune-deficient mice cases with low percentages of hematopoietic stem cells of our cohort gave rise to leukemic or no engraftment, whereas cases with normal hematopoietic stem cell levels mostly resulted in multi-lineage engraftment. Based on our experimental data, we propose that leukemic stem cells have increased niche affinity in cases with low percentages of hematopoietic stem cells. To validate this hypothesis, we developed new mathematical models describing the dynamics of healthy and leukemic cells under different regulatory scenarios. These models suggest that the mechanism leading to decreases in hematopoietic stem cell frequencies before leukemic relapse must be based on expansion of leukemic stem cells with high niche affinity and the ability to dislodge hematopoietic stem cells. Thus, our data suggest that decreasing numbers of hematopoietic stem cells indicate leukemic stem cell persistence and the emergence of leukemic relapse.

Introduction Acute myeloid leukemia (AML) is a malignant disease and affected people have poor overall survival (OS) rates.1-3 Although most AML patients achieve complete remissions after standard chemotherapy, the majority subsequently relapse with haematologica | 2017; 102(9)

Correspondence: christoph.lutz@med.uni-heidelberg.de

Received: January 5, 2017. Accepted: May 17, 2017. Pre-published: May 26, 2017. doi:10.3324/haematol.2016.163584 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1567 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1567


W. Wang et al.

more aggressive and resistant disease demonstrating the necessity to improve therapeutic strategies.4-6 As relapse is the main cause of death, it is important to effectively stratify AML patients according to their individual risk of relapse and to identify patients who require more aggressive treatment protocols, such as allogeneic hematopoietic stem cell transplantation (HSCT). Cytogenetic analysis is so far the most important and reliable risk stratification tool with specific abnormalities and the degree of mutations indicating cases with good and poor prognosis. However, up to 50% of AML patients have normal karyotypes and can be classified in the cytogenetic intermediate-risk group that combines patients with highly variable clinical outcomes.1,5 The systematic analysis and identification of AML-specific mutations, such as Fms-related tyrosine kinase 3 (FLT3)-internal tandem duplication (ITD), CCAAT/enhancer-binding protein-α (CEBPA) and nucleophosmin 1 (NPM1), further improved AML risk stratification, increasing the number of patients who can be stratified.7,8 Another important improvement in AML management and risk stratification is the ability to track minimal residual disease (MRD) in follow-up samples of patients undergoing therapy; MRD can be assessed by flow cytometry or molecular analysis of existing mutations, such as NPM1.9 So far, tracking of molecular MRD by polymerase chain reaction analysis represents the most sensitive approach, with detection levels ranging from 0.1-0.001%. This approach is, however, limited by the fact that only 6070% of patients have suitable markers that can be monitored on a molecular basis.10,11 In contrast flow cytometrybased MRD analysis is applicable in over 90% of patients but its level of sensitivity can be low (0.1-0.01%) and optimal cut off values are debatable.11-13 A completely different approach for AML risk stratification has been suggested by our recent study in which we have shown that functionally normal and non-leukemic hematopoietic stem cells (HSC) can be isolated from a subgroup of patients using the CD34+(CD38-)ALDH+ (aldehyde dehydrogenase) phenotype.14 Importantly, patients suitable for this distinction can be prospectively identified by their low frequency of total ALDH+ cells [<1.9% of total mononuclear cells (MNC)] - named ALDH-rare AML.6,14,15 Patients in whom non-leukemic HSC (nl-HSC) can be separated represent a cohort with favorable outcome compared to the rest of AML patients.14,16,17 However, within ALDH-rare AML, nl-HSC frequencies vary significantly and even patients without any detectable ALDH+ cells exist (unpublished observation).18 Studies analyzing HSC in AML have mostly focused on the cells’ functional and genetic properties. In contrast to previous studies, we have analyzed how nl-HSC relate to survival and behave during the course of disease in relation to disease burden. We have also analyzed the functional properties of nl-HSC in vitro and in vivo in correlation to their frequency within the bone marrow compartment which may serve as an indicator of niche changes or competition both facilitated by leukemic infiltration. Our experimental results show that nl-HSC frequencies predict outcome and correlate to MRD in follow-up samples. We hypothesized that the correlation of nl-HSC frequencies, MRD status and patients’ survival can be explained by an ongoing competition between leukemic stem cells (LSC) and nl-HSC in the bone marrow niche. To support this hypothesis, we proposed a novel mathemati1568

cal model. Mathematical modeling has been shown to be a useful tool, improving our understanding of the hematopoietic system and its diseases, as it allows the study of processes that cannot be observed in conventional experiments. In this context, our model enables linkage of clinical data to unobservable dynamic processes in the human stem cell niche.19-23 Based on our model simulations we conclude that cell competition within the niche is required to explain the decline of nl-HSC before overt relapse.

Methods Sample collection Bone marrow aspirates derived from 61 AML patients and 11 healthy donors were collected after informed consent between October 2011 and August 2015. All experiments were approved by the Ethics Committee of the Medical Faculty of Heidelberg University. The patients’ characteristics are shown in Online Supplementary Table S1. The exact time-points of follow-up sample collection are indicated in the captions of the relevant figures.

Preparation of cells MNC were isolated with Biocoll separation solution (Biochrom, Berlin, Germany) by gradient centrifugation and used fresh or frozen in liquid nitrogen with fetal calf serum/12.5% dimethyl sulfoxide for future studies. Numbers of nl-HSC were calculated as described in the Online Supplement. Mesenchymal stromal cells from healthy donors were isolated and cultured as described previously.24

Flow cytometry and fluorescence-activated cell sorting MNC were stained with Aldefluor reagent (Stem Cell Technologies, Vancouver, BC, Canada), CD34-APC, CD45-APCH7, propidium iodide (PI) or 7-AAD, CD38-PE or CD38-PE-Cy7 (clone HB7) (BD Bioscience, Heidelberg, Germany) as described previously.14 Cells were analyzed using a BD LSR II flow cytometer and subpopulations were sorted using a FACSAria II sorter (BD Bioscience). Guided by the percentages of HSC detected in the bone marrow of healthy donors, 0.01% (CD34+CD38ALDH+/MNC) was chosen as a cut-off value to distinguish between nl-HSC-/low and nl-HSC+ AML (Figure 2C). This corresponds to the 5% percentile of HSC frequencies in healthy individuals.

In vitro colony assays To compare the long-term colony forming abilities of different subpopulations, we performed the limiting dilution long-term culture-initiating cell (LTC-IC) assay with HSC-CFU lite with Epo (Miltenyi Biotec, Bergisch Gladbach, Germany) as described previously.14 To compare progenitor potential of different fractions, short-term colony-forming cell (CFC) assays were performed using HSC-CFU complete with Epo (Miltenyi Biotec) according to the manufacturer’s instructions. LTC-IC frequencies were determined by L-Calc Limiting Dilution Software (Stem Cell Technologies, Vancouver, BC, Canada).

Mathematical modeling To uncover mechanisms leading to the dynamics observed in the experimental data, we used computer simulations of mathematical models reflecting different plausible interactions of healthy and leukemic cells. In particular, we developed a novel mathematical model describing dynamics in the bone marrow niche. The model is an extension of our previously published haematologica | 2017; 102(9)


Predictive value of non-leukemic HSC in AML

model on LSC dynamics in acute leukemias.23 The new model includes competition of healthy and leukemic cells for niche spaces and dislodgement of healthy cells from the niche by leukemic cells. The model is based on a system of non-linear ordinary differential equations, describing proliferation, self-renewal, differentiation, death and various possible interactions of healthy and leukemic cells. For additional details of the study methods see the Online Supplementary Methods section.

Results CD34+CD38-ALDH+ cells are enriched for hematopoietic stem cell potential Our previous analyses using in vivo and in vitro assays showed that nl-HSC can be isolated from leukemic cells

in a subgroup of AML patients using the CD34+ALDH+ marker combination (Figure 1A).14 In an attempt to further enrich nl-HSC within this compartment we analyzed CD34+CD38-ALDH+ cells for their in vitro colonyforming potential in comparison to that of other subcompartments including CD34+CD38+ALDH+, CD34ALDH+, ALDH- and CD34+CD38-ALDH- cells. Limiting dilution LTC-IC assays quantifying the long-term colony-forming ability of tested subpopulations were performed for 24 diagnostic ALDH-rare AML cases. LTC-IC frequencies were highest for CD34+CD38ALDH+ cells (1 in 2.5 cells) confirming that these cells are enriched for HSC potential (Figure 1B). Normal bone marrow samples served as healthy controls and showed comparable LTC-IC frequencies for CD34+CD38-ALDH+ cells (Figure 1C). Progenitor potential was also determined by plating the same subpopulations in methylcellulose. The results obtained confirmed the LTC-IC data

A

B

C

Figure 1. CD34+CD38-ALDH+ cells are enriched for hematopoietic stem cell potential. (A) Example of sorting and analysis strategy of CD34+ALDH+ cells. (B) LTC-IC frequencies of sorted subpopulations of 24 ALDHrare AML samples. Results are shown in LTCIC per 100 plated cells. (C) LTC-IC frequencies of sorted subpopulations of four normal bone marrow samples. Results are shown in LTC-IC per 100 plated cells. Data are shown as mean Âą SEM.

haematologica | 2017; 102(9)

1569


W. Wang et al.

with comparable CFC potential for CD34+CD38-ALDH+ cells from leukemic and normal bone marrow samples (Online Supplementary Figure S1A,B). These colonies also contained erythroid colonies which have been shown to indicate the non-leukemic origin of the tested cells.25 Fluorescence in situ hybridization analyses of CD34+ALDH+ cells were negative for leukemia-specific mutations (Online Supplementary Figure S2A,B). These data are in line with results from our previous study, in which we showed, on a larger scale by polymerase chain reaction analysis and fluorescence in situ hybridization, that CD34+ALDH+ cells are functionally normal, nonleukemic and capable of multi-lineage engraftment in NSG mice.14

Hematopoietic stem cell frequencies vary in the cohort of ALDH-rare acute myeloid leukemias We next systematically determined nl-HSC frequencies of 61 ALDH-rare AML diagnostic samples and found highly variable percentages (range, 0% to 0.564%) (Figure 2A). These patients also included two cases without any detectable CD34+CD38-ALDH+ cells. These two patients had extremely poor survival and both died of early relapse

(Figure 2A and Online Supplementary Figure S3). Guided by these observations and the percentages of HSC detected in the bone marrow of healthy donors, 0.01% (CD34+CD38ALDH+/MNC) was chosen as a cut-off value to stratify ALDH-rare AML patients in order to assess whether nlHSC frequencies correlate with disease outcome (Figure 2B). The majority of all 61 studied patients (45 out of 61) had relatively high HSC proportions (nl-HSC+ AML; ≥0.01% of total MNC) while patients with nl-HSC-/low AML (<0.01% of total MNC) were less frequent. Since nlHSC proportions do not necessarily reflect absolute nlHSC numbers, we calculated absolute numbers according to bone marrow sample volumes, which were available for 47 AML patients. Results of nl-HSC proportions and nl-HSC numbers (cells/mL) were consistent and showed low absolute numbers for patients with nl-HSC-/low AML compared to healthy subjects and patients with nl-HSC+ AML (Figure 2C).

Non-leukemic hematopoietic stem cell frequency in patients’ bone-marrow predicts acute myeloid leukemia outcome Survival analyses of 58 ALDH-rare AML patients for

A

B

C

Figure 2. Frequencies of non-leukemic hematopoietic stem cells vary in ALDH-rare acute myeloid leukemia. (A) nl-HSC frequencies of all 61 studied patients [HSC% of total bone marrow (BM) MNC]. (B) ALDH-rare AML patients were stratified according to the frequencies of CD34+CD38-ALDH+ cells into nl-HSC+ AML (≥0.01% of total) and nl-HSC-/low AML (<0.01% of total). Representative FACS plots of each subgroup are shown. (C) CD34+CD38-ALDH+ cells/mL frequencies displayed as percentages of total MNC and cell numbers as cells/mL of 45 nl-HSC+ AML (median: 0.037%), 16 nl-HSC-/low AML (median: 0.0026%) and 11 healthy bone marrow controls (median: 0.064%). nl-HSC numbers of nl-HSC+ AML (median: 10623) were significantly higher compared to nl-HSC-/low AML (median: 240) (P<0.05). * Only patients with available nl-HSC numbers were included [35 nl-HSC+ AML, 12 nl-HSC-/low AML and 6 healthy bone marrow controls (median: 3176)].

1570

haematologica | 2017; 102(9)


Predictive value of non-leukemic HSC in AML

whom we had survival data showed a correlation between nl-HSC frequencies at the time of diagnosis and survival. HSC-/low AML patients had extremely poor OS with a median OS of 383 days and median disease-free survival (DFS) of 197 days, while the median OS and DFS have not been reached for patients with nl-HSC+ AML. Mean values for OS and DFS were substantially longer for nl-HSC+AML patients compared to nl-HSC-/low AML patients (mean OS: 1177 days versus 414 days, mean DFS: 1214 days versus 284 days) (Figure 3A). The mutation status of patients with nl-HSC-/low and nl-HSC+ AML did not explain the observed differences in survival (Online Supplementary Table S1). However, most patients could be stratified into the intermediate-risk category in which there are huge variations in patientsâ&#x20AC;&#x2122; outcomes.

We therefore analyzed the outcome of 39 patients for whom we had survival data categorized in the cytogenetic intermediate-risk group, comparing nl-HSC-/low and nl-HSC+ AML. This analysis also identified the nl-HSC/low AML cohort as a subpopulation with higher risk of relapse, with a median OS of 361 days (mean OS: 376 days) and median DFS of 150 days (mean DFS: 272 days), compared to nl-HSC+ AML for which the respective median OS has not been reached and the median DFS is 1248 days. Again, mean values for OS and DFS were substantially longer for nl-HSC+ AML than for nl-HSC-/low AML (mean OS: 1218 days versus 376 days, mean DFS: 1115 days versus 272 days) (Figure 3B). Comparative survival analysis of patients with nl-HSC+ and nl-HSC-/low AML who underwent allogeneic HSCT revealed that

A

B

Figure 3. Patients with nl-HSC-/low acute myeloid leukemia have extremely poor survival. (A) Survival analysis revealed significantly shorter overall and diseasefree survival for patients with nl-HSC-/low AML compared to those with nl-HSC+ AML [OS: nl-HSC+ AML (n=43) nl-HSC-/low AML (n=15); DFS nl-HSC+ AML (n=36) nl-HSC/low AML (n=13)]. (B) Within the cytogenetic intermediate-risk group, nl-HSC-/low AML also represent a poor prognosis group [OS: nl-HSC+ AML (n=26) nl-HSC-/low AML (n=13); DFS nl-HSC+ AML (n=24) nl-HSC/low AML (n=11)]. (C) For patients undergoing allogeneic HSCT stratification into nlHSC-/low and nl-HSC+ AML identifies nl-HSC/low AML as a cohort with poor therapy response (P<0.05) [OS: nl-HSC+ AML (n=18) nl-HSC-/low AML (n=5); DFS nl-HSC+ AML (n=18) nl-HSC-/low AML (n=5)]. *Patients who never achieved a complete remission (CR) were excluded from the DFS analysis.

C

haematologica | 2017; 102(9)

1571


W. Wang et al.

even transplantation in complete remission was not sufficient to prevent relapse in most nl-HSC-/low cases. For patients with nl-HSC-/low AML, the median OS was 446 days (mean OS: 455 days) and the median DFS 242 days (mean DFS: 310 days), while median OS and DFS have not been reached for patients with nl-HSC+ AML (mean OS: 1356 days, mean DFS: 1302 days) (Figure 3C).

Non-leukemic hematopoietic stem cell frequency predicts leukemic versus multi-lineage human engraftment in NSG mice and correlates with non-leukemic hematopoietic stem cell in vitro function To test whether nl-HSC frequencies are correlated with leukemia-initiating potential we transplanted 39 diagnostic AML cases from our cohort of ALDH-rare AML (28 nlHSC+ AML and 11 nl-HSC-/low AML) in immune-deficient NSG mice (Figure 4A). Most nl-HSC+ AML (24/28 cases) gave rise to multi-lineage engraftment and only rarely resulted in AML (3/28 cases) or non-engraftment (1/28 case). In contrast, nl-HSC/low AML always gave rise to abnormal engraftment with 5/11 cases leading to AML and 6/11 cases not engrafting at all (Figure 4A). To test whether nl-HSC derived from nl-

HSC-/low AML are functionally impaired we analyzed and compared the in vitro potential of CD34+CD38-ALDH+ cells derived from nl-HSC-/low and nl-HSC+ AML. Results from LTC-IC assays showed that LTC-IC frequencies of nlHSC+ AML (1 in 2 cells) were comparable to those of normal bone marrow and higher than those of nl-HSC-/low AML (1 in 7 cells) (Figure 4B). In line with these results, the colony-forming potential of CD34+CD38-ALDH+ cells was also higher compared to that of cells derived from nl-HSC/low AML suggesting a functional impairment of nl-HSC in nl-HSC-/low AML (Figure 4C).

Non-leukemic hematopoietic stem cell frequency in patientsâ&#x20AC;&#x2122; bone marrow correlates with acute myeloid leukemia disease status To understand the behavior of nl-HSC during therapy, matched sample series from diagnosis and various remission time points were analyzed and the results correlated with those of routine bone marrow and MRD analysis, if available (for detailed characteristics of the patients, see Online Supplementary Table S2). Initially, we focused on four patients who achieved durable remissions after chemotherapy, which were followed by allogeneic HSCT

A

B

C

Figure 4. Frequencies of non-leukemic hematopoietic stem cells predict leukemic versus non-leukemic engraftment and negatively correlate with in vitro hematopoietic stem cell function. (A) Mouse transplantation strategy with examples of AML engraftment (case 1), multi-lineage engraftment (case 2) or no engraftment (case 3). nl-HSC+ AML mostly gave rise to multi-lineage engraftment (24/28) and rarely resulted in AML (3/28) or non-engraftment (1/28), whereas nl-HSC-/low AML only gave rise to abnormal engraftment with 5/11 cases leading to AML and 6/11 cases not engrafting at all. (B) LTC-IC frequencies of sorted CD34+CD38-ALDH+ cells derived from nl-HSC+ AML (n=13) and nl-HSC-/low AML (n=9) revealed impaired in vitro function of CD34+CD38-ALDH+ cells from nl-HSC-/low AML. Results are shown in LTC-IC per 100 plated cells. (C) Comparison of CFC frequencies of CD34+CD38-ALDH+ cells derived from nl-HSC+ AML (n=12) and nl-HSC-/low AML (n=10). Data are shown as mean Âą SEM.

1572

haematologica | 2017; 102(9)


Predictive value of non-leukemic HSC in AML

in three cases. We found a consistent pattern in all cases with nl-HSC proportions increasing over time (Figure 5). As nl-HSC frequencies were measured by proportions of total MNC, which can be misleading for time points after chemotherapy with aplastic marrow conditions or diagnostic time-points with hyperplastic bone marrow, we also calculated nl-HSC numbers on the basis of available volume data (cells/mL). This analysis showed an increase of nl-HSC numbers in patients who achieved complete remissions (Figure 5A-D). MRD data were available for one patient and confirmed the observations that nl-HSC recover when disease burden goes down (Figure 5A). Some patients had persistent disease after the first course of induction chemotherapy which correlates with poor prognosis.5 In these cases we found that nl-HSC did not recover (Figure 5B and Online Supplementary Figure S4). In one patient for whom we had many follow-up samples available we found that nl-HSC remained rare at time points of persistent disease and only recovered when a durable remission was reached after allogeneic HSCT (Figure 5B).

points and the time of relapse. In all analyzed samples nlHSC frequencies and numbers decreased at the time of relapse (Figure 6). In this context, nl-HSC numbers (cells/mL) represent the more accurate measurement of nlHSC compared to nl-HSC percentages as proportions can be skewed by large total cell numbers. Our analysis showed that nl-HSC numbers decreased during therapy and never recovered in patients who finally relapsed. For two sample series MRD data were available with molecular relapse preceding frank leukemic relapse. In both cases nl-HSC numbers and frequencies already decreased to low levels at the time of molecular relapse suggesting that reduced nl-HSC can predict leukemic relapse (Figure 6A,B). For the remaining two cases no MRD data were available. However, both cases finally relapsed with nlHSC numbers and frequencies decreasing before frank relapses were detected (Figure 6C,D). For detailed nl-HSC numbers and frequencies during therapy see Online Supplementary Table S3.

Decline of non-leukemic hematopoietic stem cell frequency predicts acute myeloid leukemia relapse

Mathematical modeling suggests that competition of non-leukemic and leukemic stem cells for niche spaces is responsible for the decline of non-leukemic hematopoietic stem cells before relapse

We next sought to examine the behavior of nl-HSC in patients with recurrent disease by analyzing five matched sample series from diagnosis, various remission time

According to our observations a decrease in nl-HSC numbers is associated with the presence of MRD and always precedes a full-blown AML relapse. Based on this

A

B

C

D Figure 5. Frequencies of non-leukemic hematopoietic stem cells recover in patients who achieved complete remissions. Blast percentages and nl-HSC percentages at diagnosis and various followup time-points are shown with the percentage contribution of these populations in total bone marrow MNC of patients 3-6. nl-HSC numbers (if available) are shown as cells/mL and MRD data of patient 3 are shown as CBFbMYH11/ABL ratio. Treatment times, time points of persistence and the event of allogeneic HSCT are indicated on the respective time line. Detailed characteristics on this and other AML patients are described in Online Supplementary Table S2. CR (complete remission): blast% <5%; PR (partial remission): blast% 525%; relapse: loss of CR with blasts â&#x2030;Ľ5%

haematologica | 2017; 102(9)

1573


W. Wang et al.

observation, we hypothesized that this phenomenon reflects competition of nl-HSC and LSC for a limited number of bone marrow niche spaces. Part of this hypothesis is that LSC have a higher affinity than nl-HSC for such niches. This increased affinity enables LSC to dislodge nl-HSC from the protective niche and to occupy their spaces. Due to a lack of direct experimental confirmation, we developed a novel mathematical model to validate this hypothesis. The presented model can simulate different possible scenarios of interaction between leukemic and hematopoietic cells. These scenarios include: (i) competition of nl-HSC and LSC for a joint niche and dislodgement of nl-HSC by LSC; (ii) separate niches for LSC and nl-HSC, impairment of nl-HSC selfrenewal and/or progenitor cell expansion by the leukemic cell bulk; (iii) competition of healthy and leukemic cells for systemic factors such as cytokines, including cytokine dependence of leukemic cells; and (iv) increased death rates of hematopoietic cells in the presence of leukemic cells. Notably, the early decline of nl-HSC before overt relapse was only reproduced by model simulations in the presence of a niche with dislodgement of nl-HSC by LSC (Figure 7A,B). In all other considered versions of the model, the leukemic blast count increases before the decline of nl-HSC counts, which is in contradiction to the

A

C

1574

clinical observations (Figure 7C,D, Online Supplementary Figures S7 and S8). This modeling result, together with the clinical observations, supports the hypothesis that expansion of LSC in the bone marrow niche and dislodgement of nl-HSC are responsible for the decline of nlHSC before frank relapse. This mechanism also explains the observed correlation of high MRD levels and reduced nl-HSC frequencies. Furthermore the model helps to understand possible functional differences of LSC in nlHSC+ and nl-HSC-/low AML. Simulations indicate that the niche affinity of LSC in nl-HSC-/low AML is higher than in nl-HSC+AML. This may explain the poor prognosis of nlHSC-/low AML and its resistance to treatment (for details see the Online Supplement).

Discussion Previous studies have shown that ALDH activity alone or in combination with other markers can be used to identify nl-HSC in a subgroup of AML patients.6,18,26 In this context, we previously established a prospective stratification method based on the level of ALDH+ cells separating ALDH-numerous (â&#x2030;Ľ1.9%) and ALDH-rare (<1.9%) AML. We showed that in ALDH-rare AML, nl-HSC can be separated from leukemic cells by using the CD34+ALDH+

B

D

Figure 6. Frequencies of nonleukemic hematopoietic stem cells correlate with disease status and predict relapse. Blast percentages and nl-HSC percentages at diagnosis and various follow-up time-points are shown with the percentage contribution of these populations in total bone marrow MNC of patients 7, 8, 9, 10. nl-HSC numbers (if available) are shown as cells/mL and MRD data for patients 7 and 8 are shown as NPM1/ABL ratio. Treatment times, time-points of persistence/relapse and the event of allogeneic HSCT are indicated on the respective time line. Detailed characteristics of this and other AML patients are described in Online Supplementary Table S2. CR (complete remission): blast% <5%; PR (partial remission): blast% 5-25%; Persistent: blast% >25%; relapse: loss of CR with blasts â&#x2030;Ľ5%. * Patient for whom only peripheral blood was available at diagnosis.

haematologica | 2017; 102(9)


Predictive value of non-leukemic HSC in AML

phenotype.14 The present study confirms these data and demonstrates that the CD34+CD38-ALDH+ phenotype further enriches for HSC potential in ALDH-rare AML. However, despite these functional results it cannot be excluded that these nl-HSC already carry some leukemiaspecific mutations. These so-called pre-leukemic HSC have been shown to be functionally normal with the ability to give rise to multi-lineage engraftment.27,28 In our detailed analyses of 61 patients with ALDH-rare AML we found highly variable nl-HSC frequencies including two cases with no detectable nl-HSC. These patients had extremely poor survival leading to the hypothesis that nl-HSC correlate with outcome in AML. We, therefore, stratified all the patients with ALDH-rare AML, based on their nl-HSC frequencies, into those with nl-HSC-/low (<0.01%) or nl-HSC+ (â&#x2030;Ľ0.01%) AML and found that this stratification was correlated with disease outcome. In this analysis, patients with nl-HSC-/low AML had poor OS and DFS which were significantly worse than those of the group of nl-HSC+ AML patients. Risk stratification is particularly important for the identification of patients who require more intensive treatment strategies, such as allogeneic HSCT. Classical risk stratification relies on cytogenetic markers that reliably distinguish patients with a good from those with a poor prognosis. However, some patients do not display either good- or poor-risk cytogenetics resulting in a large cohort of cytogenetic intermediate-risk patients who have variable outcomes. The uncertainty in risk prediction for so many patients highlights the need to improve existing risk stratification approaches.1 We, therefore, also analyzed the cytogenetic intermediate-risk group according to our nl-HSC-/low and nl-HSC+ AML criteria and detected a clear separation in outcomes with extremely poor survival for nl-HSC-/low and very good outcome for nl-HSC+ AML. We also found that nl-HSC-/low AML do not just have poor survival but are also at such a high risk of relapse that allogeneic HSCT in complete remission appears to have virtually no effect on DFS. These results suggest distinct biological properties of these

subgroups of patients from the cohort of ALDH-rare AML and may explain the differences observed in our xenotransplantation assays. More importantly and also clinically relevant is the conclusion that these patients seem to require new treatment strategies in order to avoid leukemic relapse. There are a number of possible explanations for the poor survival of patients with nl-HSC-/low AML but the most plausible is competition for identical bone marrow niches between nl-HSC and LSC which has already been suggested for acute lymphobastic leukemia and AML.29-31 In our ALDH-rare AML representative xenotransplantation assays we found that mice transplanted with cells from nl-HSC-/low AML either failed to engraft or gave rise to AML which corresponded to their low nl-HSC frequencies. In contrast, nl-HSC+ AML almost exclusively gave rise to normal multi-lineage engraftment as might be expected from samples with high nl-HSC frequencies. However, this difference was most likely not just a result of increased nl-HSC frequencies but also due to impaired HSC functionality of nl-HSC-/low AML which we demonstrated in vitro. The observed in vitro functional impairment might be the result of pre-leukemic mutations which have been shown to have variable penetrance within the nlHSC compartment affecting up to all tested nl-HSC.27 For DNMT3A-mutated nl-HSC a competitive growth advantage has been shown in xentotransplantation assays but for most pre-leukemic mutations the exact functional impact remains unclear.28 It is, therefore, possible that in nl-HSC-/low AML the burden of pre-leukemic mutations is higher with resulting functional impairment. In order to clarify this observation more detailed functional studies at the single cell level combined with genetic analysis are needed. As transplantation of leukemic bulk mimics competition of nl-HSC and LSC for niche spaces in vivo, we can infer from our data that in nl-HSC-/low AML LSC do indeed outcompete nl-HSC. However, this conclusion is limited by the fact that nl-HSC numbers are reduced and we do not

A

B

C

D

Figure 7. Mathematical modeling suggests that leukemic stem cell niche affinity is responsible for relapse and early decrease of non-leukemic hematopoietic stem cell numbers. (A) Simulation of nl-HSC counts and blast fractions reproduce the early decrease of HSC before overt relapse. (B) The simulations are based on the assumption that LSC and nl-HSC share identical stem cell niche spaces. Daughter LSC emerging from divisions can dislodge nl-HSC from the niche and occupy their spaces. The dislodged nl-HSC differentiate and lose HSC potential. (C) Scenarios without direct niche competition and without nl-HSC dislodgement cannot reproduce the early decrease of nl-HSC counts. (D) The simulation depicted in (C) is based on the assumption that nl-HSC and LSC reside in different niches and that leukemic cells inhibit nl-HSC self-renewal.

haematologica | 2017; 102(9)

1575


W. Wang et al.

know the exact LSC number for the cases we transplanted. A better and less artificial way to approach competition between nl-HSC and LSC is the analysis of serially collected samples from patients undergoing chemotherapy. In this analysis we found clear patterns for patients who achieved long-term remissions and those suffering from relapse. In cases of long-term remission nl-HSC frequencies and numbers recovered, even when strongly suppressed initially. In contrast, in patients who relapsed after a short period of remission nl-HSC frequencies and numbers decreased when leukemic activity was detectable, even at time points when only MRD could be measured. Suppressed nl-HSC numbers were always followed by leukemic relapse with some cases in which nl-HSC never recovered and other cases in which nl-HSC decreased after a short time of recovery. Comparison of matched diagnostic and relapse samples showed that nl-HSC numbers (cells/mL) were always massively reduced at frank relapse compared to diagnostic time-points, which may be the result of clonal selection of more aggressive subclones (Online Supplementary Table S3). To support our hypothesis of nl-HSC and LSC competition for identical bone marrow niches we developed a mathematical model that enables clinical data to be linked to processes in the stem cell niche that cannot be observed directly. We used computer simulations of models describing different mechanisms to conclude that niche competition with active dislodgement of nl-HSC (independent of HSC cell divisions) is required for the early decline of nlHSC as observed in this study. Scenarios lacking niche competition (e.g. separate niches to support nl-HSC and LSC) do not reproduce the experimental observation. Importantly, this result is independent of cytokine-dependence of leukemic cells. The dislodgement of nl-HSC from the niche explains the correlation of reduced nl-HSC counts and MRD. Model simulations also suggest that the niche affinity of leukemic cells is higher in nl-HSC-/low AML than in nl-HSC+ AML. Translated into our setting, HSC-negativity would represent cases in which LSC have occupied bona fide HSC bone marrow niches, which have been described to be of limited number, thereby squeezing out nl-HSC.32-34 Together with clinical data the modeling results lead to the conclusion that dislodgement of nl-HSC from the niche is a relevant mechanism in nl-HSC-/low AML Supporting our hypothesis, real-time imaging studies performed in xenotransplantation assays demonstrated that acute lymphoblastic leukemia cells can displace nlHSC from their marrow niche by inducing changes to the microenvironment.30 It remains unclear whether the effect of nl-HSC suppression is a distinct biological property or just a sign of disease stage. It might be that uncontrolled AML growth ultimately dislodges all HSC with the differences we observe just reflecting disease-stage at the time of diagnosis. However, judging from our data this seems to be unlikely as (i) some nl-HSC+ AML display extremely high white blood counts and (ii) nl-HSC suppression can be observed in follow-up and early relapse samples from time points when disease burden is still moderate. Our data show that at the time of hematologic relapse nl-HSC numbers decrease massively compared to the time of diagnosis with no nl-HSC-recovery between these time points. This suggests that in these cases LSC survived chemotherapy by occupying the protective bone marrow 1576

niches. Importantly, even allogeneic HSCT did not improve patientsâ&#x20AC;&#x2122; overall poor outcome which was potentially caused by a niche occupation of therapy-resistant LSC. Supporting this hypothesis is the observation that engraftment of nl-HSC-/low AML patients who relapsed after allogeneic HSCT was delayed in comparison to that of nl-HSC+ AML patients who remained in remission after allogeneic HSCT. For this analysis we compared three of four possible nl-HSC-/low AML patients who relapsed after allogeneic HSCT. One patient was excluded because this patient only received a reduced conditioning regime with 2 Gy total body irradiation/fludarabine and never became aplastic. These patients were compared to 11 of 12 nlHSC+ AML patients who remained in remission after allogeneic HSCT for whom engraftment data were available. For detailed information see Online Supplementary Table S5 and Online Supplementary Figure S9. Recently a study has been published analyzing the effects of human AML xenotransplantation on murine HSC and progenitors.35 The authors observed that after xenotransplantation of human AML in immune-deficient mice, murine HSC were not depleted but a differentiation block at the progenitor stage occurred. These results were confirmed by analyzing a cohort of primary human AML samples. There are important differences between our study and the above-described results. We do not consider that a xenotransplantation model examining murine hematopoiesis can be used to draw conclusions on human hematopoiesis in the human bone marrow. Another difference between the studies is that they only used a certain subgroup of AML with very low CD34 expression for their analysis of human bone marrow. In these leukemias, which are enriched for good prognosis nucleophosminmutant AML, non-leukemic cells can be isolated by sorting for CD34+ cells. In contrast, in our study we analyzed all subsets of AML and did not limit the analysis to such a rare and good-risk enriched subtype. Especially in the poor prognosis nl-HSC-/low AML cohort we found a different scenario with decreased nl-HSC numbers. We therefore think that their conclusions cannot be generalized to all AML cases, and in particular to high-risk cases. Collectively these data suggest that AML show different behaviors in influencing non-leukemic hematopoiesis which seem to vary according to their aggressiveness and niche affinity. One limitation of our study is that the results presented are only representative of the group of ALDH-rare AML in which HSC isolation is possible and which accounts for about 77% of all AML cases, whereas in ALDH-numerous cases this isolation is not feasible.14 The reason for this is most likely that in these cases normal HSC and leukemic cells both exhibit high ALDHactivity. In order to use our approach for outcome prediction in all AML cases a different HSC-isolation strategy is necessary. Various approaches for HSC isolation from AML samples have been described.27,28 However, these studies only analyzed small series and ALDH status was not reported. Future studies should therefore compare these approaches in order to test whether nl-HSC can be isolated from all AML and whether the observed behavior of nl-HSC is universal or specific to the subgroup of ALDH-rare AML. Taken together our results suggest that nl-HSC can serve as a risk stratification tool and as qualitative marker of MRD in a subgroup of AML patients. These cases may haematologica | 2017; 102(9)


Predictive value of non-leukemic HSC in AML

also provide a starting point to understand the interactions of LSC and the protective bone marrow niche. Acknowledgments This work was supported by research funding from the German Research Foundation DFG (SFB 873; subprojects A13 to CL, B04 to AT, B07 to ADH, B08 to AMC and Z02 to VE), the SyTASC consortium funded by the Deutsche Krebshilfe and the Dietmar Hopp Foundation (both to AT).

References 1. Grimwade D, Walker H, Oliver F, et al. The importance of diagnostic cytogenetics on outcome in AML: analysis of 1,612 patients entered into the MRC AML 10 trial. The Medical Research Council Adult and Children's Leukaemia Working Parties. Blood. 1998;92(7):2322-2333. 2. Lowenberg B, Downing JR, Burnett A. Acute myeloid leukemia. N Engl J Med. 1999;341(14):1051-1062. 3. Estey E, Dohner H. Acute myeloid leukaemia. Lancet. 2006;368(9550):18941907. 4. Cheson BD, Bennett JM, Kopecky KJ, et al. Revised recommendations of the International Working Group for Diagnosis, Standardization of Response Criteria, Treatment Outcomes, and Reporting Standards for Therapeutic Trials in Acute Myeloid Leukemia. J Clin Oncol. 2003;21(24):4642-4649. 5. Dohner H, Estey EH, Amadori S, et al. Diagnosis and management of acute myeloid leukemia in adults: recommendations from an international expert panel, on behalf of the European LeukemiaNet. Blood. 2010;115(3):453-474. 6. Gerber JM, Smith BD, Ngwang B, et al. A clinically relevant population of leukemic CD34(+)CD38(-) cells in acute myeloid leukemia. Blood. 2012;119(15):3571-3577. 7. Papadaki C, Dufour A, Seibl M, et al. Monitoring minimal residual disease in acute myeloid leukaemia with NPM1 mutations by quantitative PCR: clonal evolution is a limiting factor. Br J Haematol. 2009;144(4):517-523. 8. Yanada M, Matsuo K, Suzuki T, Kiyoi H, Naoe T. Prognostic significance of FLT3 internal tandem duplication and tyrosine kinase domain mutations for acute myeloid leukemia: a meta-analysis. Leukemia. 2005;19(8):1345-1349. 9. Gruszka AM, Lavorgna S, Consalvo MI, et al. A monoclonal antibody against mutated nucleophosmin 1 for the molecular diagnosis of acute myeloid leukemias. Blood. 2010;116(12):2096-2102. 10. Gabert J, Beillard E, van der Velden VH, et al. Standardization and quality control studies of 'real-time' quantitative reverse transcriptase polymerase chain reaction of fusion gene transcripts for residual disease detection in leukemia - a Europe Against Cancer program. Leukemia. 2003;17(12): 2318-2357. 11. Buccisano F, Maurillo L, Del Principe MI, et al. Prognostic and therapeutic implications of minimal residual disease detection in acute myeloid leukemia. Blood. 2012;119

haematologica | 2017; 102(9)

The authors thank Karina Borowski for her technical assistance (FACS), Michaela Brough for her performance of FISH, Sandra Blaszkiewicz for her help in mouse experiments, Katrin Miesala for the isolation of primary cells and help in tissue culture and Anke Diehlmann for the isolation and culture of MSC. We are grateful to all clinicians in the Hematology Department of Heidelberg University Hospital for their contribution to collection of bone marrow material and patients and healthy donors who participated in this study.

(2):332-341. 12. Al-Mawali A, Gillis D, Hissaria P, Lewis I. Incidence, sensitivity, and specificity of leukemia-associated phenotypes in acute myeloid leukemia using specific five-color multiparameter flow cytometry. Am J Clin Pathol. 2008;129(6):934-945. 13. Inaba H, Coustan-Smith E, Cao X, et al. Comparative analysis of different approaches to measure treatment response in acute myeloid leukemia. J Clin Oncol. 2012;30(29):3625-3632. 14. Hoang VT, Buss EC, Wang W, et al. The rarity of ALDH(+) cells is the key to separation of normal versus leukemia stem cells by ALDH activity in AML patients. Int J Cancer. 2015;137(3):525-536. 15. Schuurhuis GJ, Meel MH, Wouters F, et al. Normal hematopoietic stem cells within the AML bone marrow have a distinct and higher ALDH activity level than co-existing leukemic stem cells. PLoS One. 2013;8(11):e78897. 16. Ran D, Schubert M, Pietsch L, et al. Aldehyde dehydrogenase activity among primary leukemia cells is associated with stem cell features and correlates with adverse clinical outcomes. Exp Hematol. 2009;37(12):1423-1434. 17. Ran D, Schubert M, Taubert I, et al. Heterogeneity of leukemia stem cell candidates at diagnosis of acute myeloid leukemia and their clinical significance. Exp Hematol. 2012;40(2):155-165.e151. 18. Pearce DJ, Taussig D, Simpson C, et al. Characterization of cells with a high aldehyde dehydrogenase activity from cord blood and acute myeloid leukemia samples. Stem Cells. 2005;23(6):752-760. 19. Whichard ZL, Sarkar CA, Kimmel M, Corey SJ. Hematopoiesis and its disorders: a systems biology approach. Blood. 2010;115(12):2339-2347. 20. Catlin SN, Busque L, Gale RE, Guttorp P, Abkowitz JL. The replication rate of human hematopoietic stem cells in vivo. Blood. 2011;117(17):4460-4466. 21. Wodarz D, Garg N, Komarova NL, et al. Kinetics of CLL cells in tissues and blood during therapy with the BTK inhibitor ibrutinib. Blood. 2014;123(26):4132-4135. 22. Stiehl T, Ho AD, Marciniak-Czochra A. The impact of CD34+ cell dose on engraftment after SCTs: personalized estimates based on mathematical modeling. Bone Marrow Transplant. 2014;49(1):30-37. 23. Stiehl T, Baran N, Ho AD, MarciniakCzochra A. Cell division patterns in acute myeloid leukemia stem-like cells determine clinical course: a model to predict patient survival. Cancer Res. 2015;75(6):940-949.

24. Wagner W, Horn P, Castoldi M, et al. Replicative senescence of mesenchymal stem cells: a continuous and organized process. PLoS One. 2008;3(5):e2213. 25. Taussig DC, Vargaftig J, Miraki-Moud F, et al. Leukemia-initiating cells from some acute myeloid leukemia patients with mutated nucleophosmin reside in the CD34(-) fraction. Blood. 2010;115(10): 1976-1984. 26. Lioznov MV, Freiberger P, Kroger N, Zander AR, Fehse B. Aldehyde dehydrogenase activity as a marker for the quality of hematopoietic stem cell transplants. Bone Marrow Transplant. 2005;35(9):909-914. 27. Jan M, Snyder TM, Corces-Zimmerman MR, et al. Clonal evolution of preleukemic hematopoietic stem cells precedes human acute myeloid leukemia. Sci Transl Med. 2012;4(149):149ra118. 28. Shlush LI, Zandi S, Mitchell A, et al. Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia. Nature. 2014;506(7488):328333. 29. Boyd AL, Campbell CJ, Hopkins CI, et al. Niche displacement of human leukemic stem cells uniquely allows their competitive replacement with healthy HSPCs. J Exp Med. 2014;211(10):1925-1935. 30. Colmone A, Amorim M, Pontier AL, Wang S, Jablonski E, Sipkins DA. Leukemic cells create bone marrow niches that disrupt the behavior of normal hematopoietic progenitor cells. Science. 2008;322(5909):1861-1865. 31. Sipkins DA, Wei X, Wu JW, et al. In vivo imaging of specialized bone marrow endothelial microdomains for tumour engraftment. Nature. 2005;435(7044):969973. 32. Colvin GA, Lambert JF, Abedi M, et al. Murine marrow cellularity and the concept of stem cell competition: geographic and quantitative determinants in stem cell biology. Leukemia. 2004;18(3):575-583. 33. Czechowicz A, Kraft D, Weissman IL, Bhattacharya D. Efficient transplantation via antibody-based clearance of hematopoietic stem cell niches. Science. 2007;318(5854):1296-1299. 34. Ishikawa F, Yoshida S, Saito Y, et al. Chemotherapy-resistant human AML stem cells home to and engraft within the bonemarrow endosteal region. Nat Biotechnol. 2007;25(11):1315-1321. 35. Miraki-Moud F, Anjos-Afonso F, Hodby KA, et al. Acute myeloid leukemia does not deplete normal hematopoietic stem cells but induces cytopenias by impeding their differentiation. Proc Natl Acad Sci USA. 2013;110(33):13576-13581.

1577


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Acute Lymphoblastic Leukemia

Ferrata Storti Foundation

HLA-DRB1*07:01–HLA-DQA1*02:01–HLADQB1*02:02 haplotype is associated with a high risk of asparaginase hypersensitivity in acute lymphoblastic leukemia Nóra Kutszegi,1,2 Xiaoqing Yang,3 András Gézsi,1 Géza Schermann,1 Dániel J. Erdélyi,2 Ágnes F. Semsei,1 Krisztina M. Gábor,4 Judit C. Sági,1 Gábor T. Kovács,2 András Falus,1 Hongyun Zhang5 and Csaba Szalai1,6

Haematologica 2017 Volume 102(9):1578-1586

Department of Genetics, Cell- and Immunobiology, Semmelweis University, Budapest, Hungary; 22nd Department of Paediatrics, Semmelweis University, Budapest, Hungary; BGI-Shenzhen, China; 4Department of Pediatrics and Pediatric Health Care Center, Faculty of Medicine, University of Szeged, Szeged, Hungary; 5BGI Clinical Laboratory, Shenzhen, China and 6Central Laboratory, Heim Pal Children Hospital, Budapest, Hungary 1 3

NK and XY contributed equally to this work.

ABSTRACT

H

Correspondence: szalaics@gmail.com or zhanghy@genomics.cn

Received: March 10, 2017. Accepted: May 30, 2017. Pre-published: June 8, 2017. doi:10.3324/haematol.2017.168211 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1578 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1578

ypersensitivity reactions are the most frequent dose-limiting adverse reactions to Escherichia coli-derived asparaginase in pediatric acute lymphoblastic leukemia (ALL) patients. The aim of the present study was to identify associations between sequence-based Human Leukocyte Antigen Class II region alleles and asparaginase hypersensitivity in a Hungarian ALL population. Four-digit typing of HLA-DRB1 and HLA-DQB1 loci was performed in 359 pediatric ALL patients by using next-generation sequencing method. Based on genotypic data of the two loci, haplotype reconstruction was carried out. In order to investigate the possible role of the HLA-DQ complex, the HLADQA1 alleles were also inferred. Multivariate logistic regression analysis and a Bayesian network-based approach were applied to identify relevant genetic risk factors of asparaginase hypersensitivity. Patients with HLA-DRB1*07:01 and HLA-DQB1*02:02 alleles had significantly higher risk of developing asparaginase hypersensitivity compared to non-carriers [P=4.56x10-5; OR=2.86 (1.73-4.75) and P=1.85x10-4; OR=2.99 (1.685.31); n=359, respectively]. After haplotype reconstruction, the HLADRB1*07:01-HLA-DQB1*02:02 haplotype was associated with an increased risk. After inferring the HLA-DQA1 alleles the HLA-DRB1*07:01–HLA-DQA1*02:01–HLA-DQB1*02:02 haplotype was associated with the highest risk of asparaginase hypersensitivity [P=1.22x10-5; OR=5.00 (2.43-10.29); n=257]. Significantly fewer T-cell ALL patients carried the HLA-DQB1*02:02 allele and the associated haplotype than did pre-B-cell ALL patients (6.5%; vs. 19.2%, respectively; P=0.047). In conclusion, we identified a haplotype in the Human Leukocyte Antigen Class II region associated with a higher risk of asparaginase hypersensitivity. Our results confirm that variations in HLA-D region might influence the development of asparaginase hypersensitivity. Introduction Asparaginase is a pivotal component of pediatric acute lymphoblastic leukemia (ALL) treatment. It converts asparagine to aspartic acid and ammonia. Lymphoblast cells lose the ability to produce asparagines; hence the loss of exogenous asparagine leads to cell death. However, the use of asparaginase can be challenging as either hypersensitivity reactions or neutralizing antibody formation can occur against this heterologous enzyme. Both can consequently lead to lower exposure to asparaginase which may result in suboptimal treatment response.1-3 In haematologica | 2017; 102(9)


HLA haplotype may predict asparaginase allergy

addition, hypersensitivity reactions to asparaginase range in severity from mild, transient (flushing or rash) to generalized anaphylaxis, which can be potentially life-threatening.4 Few studies have investigated the genetic basis of asparaginase hypersensitivity (AH). In a genome-wide association study (GWAS) of 485 pediatric ALL patients of European ancestry, Chen et al. found that rs4958351 of the glutamate receptor gene GRIA1 located at 5q33 associated with AH.5 Recently, in a GWAS of a multiethnic cohort of 3308 patients, Fernandez et al. replicated the finding between the rs4958351 variant and AH and found that the association was stronger in patients treated with native Escherichia coli-derived asparaginase (E. coli asparaginase) than in those receiving pegylated asparaginase.6 However, the significance of the association (P=0.03) did not reach the genome-wide threshold (P=5 x 10-8). This association has been replicated in a relatively small Slovenian population of 146 patients.7 In our previous candidate-gene study, we found that although genetic variants of GRIA1 influenced the risk to AH, significant differences were observed according to sex and patient subgroup (T-cell or pre-B-cell ALL).8 The association of Human Leukocyte Antigen (HLA)DRB1*07:01 allele with an increased risk of AH was revealed in a candidate-gene study of Fernandez et al. of 1870 pediatric ALL patients of European ancestry.9 Later, in the subsequent GWAS of the same group on AH, a SNP linked to HLA-DRB1*07:01 also acted as a risk allele in patients of diverse ancestry. In this study, the rs6021191 variant in NFATC2 was also associated with a higher risk of AH at the genome-wide significance threshold (P=4.1 x 10-8, OR=3.11).6 The minor allele frequency (MAF) of rs6021191 was only 0.001 among patients of European descent; therefore, this result has more relevance in patients of non-European ancestry. HLA class II alleles are involved in presentation of peptides derived from extracellular proteins to T cells and subsequent activation of the immune response. They are located in the Major Histocompatibility Complex region on chromosome 6 and are expressed in antigen presenting cells. This genetic region is highly polymorphic with considerable linkage disequilibrium. The primary goal of our study was to test the associations of HLA class II alleles with E. coli AH in a Hungarian population of 359 pediatric ALL patients using next-generation sequencing (NGS)-based HLA typing of HLA-DRB1 and HLA-DQB1 alleles. In addition, we aimed to evaluate the possible role of the HLA-DRB1–HLA-DQA1–HLADQB1 haplotypes and the polymorphic amino acid positions located in the peptide-binding groove of the HLADQ complex in the mechanism of AH.

Table 1. Patients’ characteristics. Sex (%) Male Female Age at diagnosis, years Mean (± SD) Median (range) Risk category (%) Low-risk Medium-risk High-risk Immunophenotype (%) Pre-B ALL T-ALL Unknown

200 (55.7) 159 (44.3) 6.3 (±4.2) 4.8 (1-18) 102 (28.4) 213 (59.3) 44 (12.3) 287 (79.9) 46 (12.8) 26 (7.2)

SD: Standard Deviation: pre-B ALL: pre-B-cell acute lymphoblastic leukemia (ALL); TALL: T-cell ALL.

asparaginase have been described in detail previously.8 Written informed consent was obtained from the study participants or from the next of kin, carers, or guardians on behalf of the minors/children who took part in the study. The study was conducted according to the Declaration of Helsinki and approved by the Hungarian Scientific and Research Ethics Committee of the Medical Research Council (ETT TUKEB; case n.: 8-374/20091018EKU 914/PI/08). The patients had a median age of 4.8 years at diagnosis (range 1-18 years) (Table 1). The overall incidence of E. coli AH was 39.0%. Data collection was carried out retrospectively from medical records. The National Cancer Institute Common Toxicity Criteria (CTC) system v.3.0 was used to assess the grade of hypersensitivity. We regarded a case as AH when signs of allergic reactions or anaphylactic reactions CTC grade 1 and above were noted, as described earlier.8

Sequence-based typing of HLA-DRB1 and DQB1 For sequencing, DNA samples from pediatric ALL patients with peripheral blood or bone marrow origin were available.8 Only those patients whose DNA sample met the quality control criteria of NGS-based typing method were enrolled in this study. All of the 359 patients were typed at high-resolution level for HLADRB1 and HLA-DQB1, in the Beijing Genomics Institute (BGI) using an NGS-based genotyping method that applies a massively parallel paired-end sequencing on Illumina MiSeq, effective variant phasing and haploid sequence assembling pipeline, as previously described.10 Based on this, exon 2 was sequenced in HLA-DRB1, and exons 2 and 3 were sequenced in HLA-DQB1 genes.

Estimation of HLA-DRB1–HLA-DQA1–HLA-DQB1 haplotypes Methods Patients DNA samples from 359 pediatric ALL patients were available to investigate the association of HLA class II alleles with AH. The patients were treated with protocols from the Berlin-FrankfurtMünster Study Group (accrued patients from 1990 to 2011): ALLBFM 90 (n=72), ALL-BFM 95 (n=165), ALL IC-BFM 2002 (n=117), and ALL IC-BFM 2009 (n=5). The combined chemotherapy regimens contained E. coli asparaginase (Kidrolase or Asparaginase medac) as first-line treatment. The dosing schedules of E. coli haematologica | 2017; 102(9)

HLA-DRB1–HLA-DQB1 haplotypes were estimated based on HLA-DRB1 and HLA-DQB1 4-digit typing data of ALL patients by using the PHASE software (v.2.1.1). In addition, the HLA-DRB1 and HLA-DQB1 genotypic data was used to search for three-gene HLA-DRB1–HLA-DQA1–HLA-DQB1 haplotypes in the Allele Frequency Net Database-HLA Haplotype Frequency Search (http://www.allelefrequencies.net/hla6003a.asp).13 Using this approach, it was possible to infer independently the most likely haplotypes (including information about the HLA-DQA1 locus) using haplotype data pertaining to Caucasoid ethnic origin as reference panel (Online Supplementary Text S1). 1579


N. Kutszegi et al.

Inference of polymorphic amino acid positions in HLA class II alleles

BN-BMLA method: we also applied Bayesian network based Bayesian multilevel analysis of relevance (BN-BMLA) to extend our genetic association study by estimating a posteriori probabilities of strong relevance (posteriors).14-17 The BN-BMLA method was explained in detail in our previous studies.18-20 Briefly, it computes the a posteriori probability of the strongly relevant variable sets with respect to a target variable (e.g. a dichotomous variable (yes/no) describing the AH status of the patients). The strongly relevant variables have a direct influence on the target. Values for posterior (P) range from 0 to 1, where P=1 and P=0 mean that the probability of the target being dependent on a predictor (e.g. HLA allele) is 100% and 0%, respectively. There were two types of strong relevance: direct relevance (e.g. casual genetic variant) and pure interaction (e.g. epistasis). For analyses, HLA-DRB1, HLADQB1 and HLA-DQA1 alleles, HLA-DRB1–HLA-DQA1–HLADQB1 haplotypes, sex, ALL immunophenotype, age at diagnosis, risk group, and treatment protocol were included in the model as predictors.

Amino acid sequences encoded by the HLA-DRB1, HLA-DQA1 and HLA-DQB1 exon 2 were inferred using the four-digit sequencing results of HLA-DRB1 and HLA-DQB1 as well as the inference results for HLA-DQA1. The Alignment Viewer of Database of Major Histocompatibility Complex (dbMHC; https://www.ncbi.nlm.nih.gov/gv/mhc/align.fcgi?cmd=aligndisplay&user _id=0&probe_id=0&source_id=0&locus_id=0&locus_group=1&proto _id=0&kit_id=0&banner=1) was used to assess the polymorphic amino acid positions in HLA-DRB1, HLA-DQB1 and HLA-DQA1 chains.

Statistical analysis Frequentist methods: multivariate logistic regression was used to test the associations of HLA class II alleles, HLA-DRB1–HLADQA1–HLA-DQB1 haplotypes and polymorphic amino acid positions with E. coli AH. Sex, ALL immunophenotype (pre-B or TALL), age at diagnosis (≤10 or >10 years), risk group (standard- or medium- or high-risk), and treatment protocol were included in the model as categorical covariates. Assuming an additive genetic model, odds ratios (ORs) and 95% confidence intervals (CIs) were obtained to estimate risks for each variable to AH. To account for multiple testing Bonferroni correction was used (P≤2.66 x 10-4, based on a total of 66 HLA class II alleles, 38 haplotypes and 84 polymorphic amino acid positions tested). R statistical software (3.1.2) and IBM SPSS Statistic (v.20) software were used for analyses.

A

Results Association of HLA-DRB1 and HLA-DQB1 alleles with asparaginase hypersensitivity We investigated the associations of HLA class II alleles (HLA-DRB1, HLA-DQB1 and HLA-DQA1) and haplotypes with hypersensitivity reactions to E. coli asparaginase in

B

C

Figure 1. Frequencies of HLA-DRB1, HLA-DQB1 and HLADQA1 alleles in pediatric patients with acute lymphoblastic leukemia. High-resolution sequence-based typing of HLA class II alleles resulted in 35 unique HLA-DRB1 (A) and 19 unique HLA-DRB1 (B) alleles in 359 patients. HLA-DQA1 alleles were inferred in 257 patients and resulted in 7 unique alleles (C).

1580

haematologica | 2017; 102(9)


HLA haplotype may predict asparaginase allergy Table 2. Association of HLA-DRB1–HLA-DQB1 haplotypes with E. coli asparaginase hypersensitivity in acute lymphoblastic leukemia patients (n=359).

Haplotype DRB1*01:01–DQB1*05:01 DRB1*01:01–DQB1*06:03 DRB1*01:02–DQB1*05:01 DRB1*03:01–DQB1*02:01 DRB1*04:01–DQB1*03:01 DRB1*04:01–DQB1*03:02 DRB1*04:01–DQB1*03:05 DRB1*04:02–DQB1*03:02 DRB1*04:03–DQB1*03:02 DRB1*04:03–DQB1*03:04 DRB1*04:03–DQB1*03:05 DRB1*04:04–DQB1*03:02 DRB1*04:04–DQB1*04:02 DRB1*04:05–DQB1*03:02 DRB1*04:07–DQB1*03:01 DRB1*04:08–DQB1*03:04 DRB1*04:10–DQB1*04:02 DRB1*07:01–DQB1*02:02 DRB1*07:01–DQB1*03:03 DRB1*08:01–DQB1*03:01 DRB1*08:01–DQB1*03:02 DRB1*08:01–DQB1*04:02 DRB1*08:04–DQB1*04:02 DRB1*09:01–DQB1*03:03 DRB1*10:01–DQB1*05:01 DRB1*11:01–DQB1*03:01 DRB1*11:03–DQB1*03:01 DRB1*11:04–DQB1*03:01 DRB1*11:15–DQB1*03:01 DRB1*11:43–DQB1*03:01 DRB1*12:01–DQB1*03:01 DRB1*12:01–DQB1*03:12 DRB1*12:02–DQB1*03:01 DRB1*13:01–DQB1*06:02 DRB1*13:01–DQB1*06:03 DRB1*13:02–DQB1*06:03 DRB1*13:02–DQB1*06:04 DRB1*13:02–DQB1*06:09 DRB1*13:03–DQB1*03:01 DRB1*13:05–DQB1*03:01 DRB1*13:15–DQB1*03:01 DRB1*13:18–DQB1*06:03 DRB1*14:04–DQB1*05:03 DRB1*14:54–DQB1*05:02 DRB1*14:54–DQB1*05:03 DRB1*15:01–DQB1*05:02 DRB1*15:01–DQB1*06:01 DRB1*15:01–DQB1*06:02 DRB1*15:01–DQB1*06:03 DRB1*15:01–DQB1*06:39 DRB1*15:02–DQB1*05:03 DRB1*15:02–DQB1*06:01 DRB1*16:01–DQB1*05:02 DRB1*16:01–DQB1*05:05 DRB1*16:01–DQB1*06:02 DRB1*16:02–DQB1*05:02

Frequency

OR1

95% CI2

P1

0.081 0.001 0.015 0.097 0.014 0.015 0.001 0.025 0.008 0.003 0.003 0.017 0.001 0.004 0.004 0.007 0.001 0.091 0.038 0.001 0.001 0.036 0.001 0.008 0.017 0.070 0.010 0.060 0.001 0.001 0.013 0.003 0.001 0.001 0.071 0.001 0.019 0.004 0.019 0.001 0.001 0.001 0.017 0.004 0.028 0.006 0.001 0.070 0.001 0.001 0.008 0.017 0.064 0.001 0.001 0.007

0.71 NA3 1.39 1.05 0.37 1.30 NA3 0.86 2.25 NA3 NA3 4.74 NA3 NA3 1.35 2.47 NA3 2.99 1.71 NA3 NA3 0.48 NA3 1.63 0.12 0.40 1.13 1.30 NA3 NA3 2.34 NA3 NA3 NA3 0.51 NA3 0.90 0.81 0.61 NA3 NA3 NA3 1.23 1.23 0.72 0.49 NA3 1.32 NA3 NA3 NA3 0.22 1.25 NA3 NA3 0.45

0.38-1.31 NA3 0.39-4.94 0.60-1.82 0.08-1.80 0.38-4.47 NA3 0.31-2.35 0.37-13.73 NA3 NA3 1.35-16.61 NA3 NA3 0.12-15.78 0.39-15.55 NA3 1.68-5.31 0.76-3.84 NA3 NA3 0.18-1.26 NA3 0.31-8.64 0.02-0.98 0.19-0.82 0.24-5.32 0.67-2.54 NA3 NA3 0.60-9.12 NA3 NA3 NA3 0.26-1.00 NA3 0.28-2.90 0.07-9.13 0.18-2.06 NA3 NA3 NA3 0.38-4.01 0.37-48.73 0.26-1.97 0.05-5.17 NA3 0.70-2.47 NA3 NA3 NA3 0.05-1.06 0.66-2.38 NA3 NA3 0.05-4.10

0.27 0.98 0.61 0.87 0.22 0.68 0.98 0.77 0.38 0.98 0.98 0.02 0.98 0.99 0.81 0.34 0.98 1.85 x 10-4 0.20 0.98 0.98 0.13 0.98 0.57 0.05 0.01 0.88 0.44 0.98 0.98 0.22 0.98 0.98 0.98 0.05 0.98 0.85 0.86 0.43 0.98 0.98 0.98 0.73 0.25 0.52 0.55 0.98 0.39 0.98 0.98 0.98 0.06 0.50 0.98 0.98 0.48

Results that reached the significance threshold (P≤2.66 x 10-4) are in bold. NA: not available. 1For the association between the allele and asparaginase hypersensitivity. 295% confidence interval for the estimated odds ratio. 3The odds ratios and confidence intervals were not estimated due to low allele frequency..

haematologica | 2017; 102(9)

1581


N. Kutszegi et al.

Figure 2. Association of HLA-DQB1 and HLA-DQB1 alleles with E. coli asparaginase hypersensitivity. Multivariate logistic regression was applied to assess the association of HLA class II alleles with asparaginase hypersensitivity. Sex, age, treatment protocol, risk group, acute lymphoblastic leukemia (ALL) immunophenotype were included in the analysis as categorical covariates. The dashed horizontal line indicates the Bonferroni-corrected statistical significance threshold. HLA-DRB1*07:01 and HLA-DQB1*02:02 alleles were significantly associated with asparaginase hypersensitivity (n=359; P=4.56x10-5 and 1.85x10-4, respectively).

pediatric ALL patients. Among 359 patients, the high-resolution sequence-based typing resulted in 35 unique HLADRB1 and 19 unique HLA-DQB1 alleles (Figure 1A and B). Univariate and multivariate logistic regression was performed to test the potential risk factors for AH (Online Supplementary Table S1). The incidence of AH varied by risk group (P=5.7 x 10-5), ranging from 31% for medium-risk group to 77% for the high-risk group and treatment protocol (P=3.36 x 10-3), ranging from 26% for ALL-BFM 90 to 60% for ALL IC-BFM 2002 experimental protocol (Online Supplementary Table S1). Using multivariate analysis risk group and treatment protocol also showed significant association with AH (Online Supplementary Table S1). For analyses with genetic data, we included all risk factors in the model as categorical covariates. Applying Bonferroni correction (P≤2.66x10-4) in multivariate logistic regression analyses HLA-DRB1*07:01 and HLA-DQB1*02:02 alleles showed significant associations with E. coli AH in 359 pediatric ALL patients (Figure 2). None of the patients was homozygous for either allele. Patients with HLA-DRB1*07:01 allele had significantly higher risk of developing AH compared to non-carriers [P=4.56x10-5; OR=2.86 (1.73-4.75)] (Figure 3). Similarly, patients harboring HLA-DQB1*02:02 were at greater risk of having AH [P=1.85x10-4; OR=2.99 (1.68-5.31)] (Figure 3). Allelic odds ratios and the AH rate of patients depending on the carrier state are shown in Online Supplementary Table S2. There was a significant difference in the proportion of HLA-DQB1*02:02 carriers between patients with pre-Bcell and T-cell ALL (19.2% vs. 6.5%; in patients with preB-cell and T-cell ALL, respectively; P=0.047). No such difference could be detected in the case of HLA-DRB1*07:01 carriers. The BN-BMLA approach was used to investigate the different dependency relations between HLA alleles, age, ALL immunophenotype, sex, treatment protocol, risk 1582

group, and AH. The analysis revealed that if both HLADRB1 and HLA-DQB1 alleles were included in the Bayesian Network, among genetic factors only HLADQB1*02:02 allele had direct strong relevance to AH (P=0.95) beside the influence of the risk group status (P=1.00) (Figure 4). There was no interaction or redundancy between predictors.

Haplotype estimation and the inference of HLA-DQA1 alleles In order to further investigate the relationship between HLA-DRB1*07:01 and HLA-DQB1*02:02 alleles associated with AH, we estimated haplotypes using the PHASE software (v.2.1.1). Using this method, the haplotype reconstruction resulted in 56 different haplotypes (Table 2). Two haplotypes containing HLA-DRB1*07:01 allele were estimated among patients: HLA-DRB1*07:01–HLADQB1*02:02 and HLA-DRB1*07:01–HLA-DQB1*03:03. Out of these, only HLA-DRB1*07:01–HLA-DQB1*02:02 showed a positive association with AH [P=1.85 x 10-4; OR=2.99 (1.68-5.31)] (Table 2). According to this, the HLA-DRB1*07:01 and HLA-DQB1*02:02 are risk alleles and in contrast to the HLA-DRB1*07:01–HLADQB1*03:03 haplotype, HLA-DRB1*07:01–HLADQB1*02:02 reached statistical significance. This is in agreement with the result of the BN-BMLA, i.e. in respect of AH, the presence of HLA-DQB1*02:02 seems to be important. Next, the HLA-DRB1-HLA-DQA1-HLA-DQB1 haplotypes were independently inferred by using reference data from The Allele Frequency Net Database. The aim of inferring HLA-DQA1 alleles was to investigate the possible role of the structure of the HLA-DQ complex in the pathomechanism of AH. In contrast to the HLA-DR complex, the alpha chain is also polymorphic in the HLA-DQ heterodimer, therefore polymorphic amino acid positions in HLA-DQA1 may also influence the peptide preference of the HLA-DQ complexes and consequently the develophaematologica | 2017; 102(9)


HLA haplotype may predict asparaginase allergy

Figure 3. Patients carrying the HLADRB1*07:01 or HLA-DQB1*02:02 alleles had strong association with asparaginase hypersensitivity. The incidence of hypersensitivity reactions was 57% (52 of 92) and 33% (88 of 267) for patients carrying HLA-DRB1*07:01 allele and for noncarriers, respectively. In the case of patients with HLA-DQB1*02:02 allele, the incidence of asparaginase hypersensitivity was 60% (39 of 65), while it was 34% (101 of 294) for patients who did not have the allele. Odds ratios (OR) and 95% confidence intervals were estimated by using multivariate logistic regression analysis.

ment of a specific immune response against asparaginase. Using data for haplotype frequencies in Caucasoid populations, and applying the aforementioned method, the estimation of extended haplotypes was possible in 72% of the patients (n=257; Online Supplementary Table S3). We compared the haplotype estimation results obtained by using the PHASE software to those extended haplotype data inferred by using The Allele Frequency Net Database. Results were in 100% agreement. Finally, among these 257 patients, 7 unique HLA-DQA1 alleles were inferred (Figure 1C). Multivariate logistic regression analysis showed that HLA-DQA1*02:01 allele and HLA-DRB1*07:01–HLADQA1*02:01–HLA-DQB1*02:02 haplotype were positively and significantly associated with AH [P=3.45x10-5; OR=3.69 (1.99-6.84) and P=1.22x10-5; OR=5.00 (2.4310.29), respectively] (Online Supplementary Table S3). These statistical associations were equivalent to the associations of HLA-DRB1*07:01 and HLA-DQB1*02:02 alleles, respectively; hence HLA-DQA1*02:01 exclusively and in all cases occurred together with HLA-DRB1*07:01.

with leukemia, a total of 27 positions associated significantly (Table 3). A valine at position 78 in HLA-DRB1 showed the strongest association with AH [P=8.16x10-6; OR=4.01 (2.18-7.37)] (Table 3). Out of the 27 risk amino acids, 7 and 3 were uniquely coded by HLA-DRB1*07:01 and HLADQA1*02:01, respectively [P=3.45x10-5; OR=3.69 (1.996.84)] (Table 3). All positions with risk amino acids in HLA-DQB1 belonged to at least two alleles and were, to a lesser extent, associated with hypersensitivity [P=1.69x10-4; OR=2.87 (1.66-4.97)] (Table 3). Interestingly, while all amino acids conferring risk to AH were present in HLA-DQB1*02:02; the other HLA-DQB1 allele linked to HLA-DRB1*07:01, HLA-DQB1*03:03 possessed no risk amino acid at any risk position. These results confirm the important role of the HLADRB1*07:01–HLA-DQA1*02:01–HLA-DQB1*02:02 haplotype in the increased risk of AH.

Inference of polymorphic amino acid positions in HLA class II alleles

The aim of our study was to investigate the association of the variations in the HLA class II region with E. coli AH in pediatric ALL patients. By analyzing 359 Hungarian patients, we found that HLA-DRB1*07:01 and HLADQB1*02:02 were associated with an increased risk of developing AH. Haplotype reconstruction also revealed that the HLA-DRB1*07:01–HLA-DQA1*02:01–HLADQB1*02:02 haplotype positively and significantly associated with an increased risk of AH. Previously, in 2014, a study by Fernandez et al. of 1870

A total of 84 polymorphic amino acid positions were identified from 4-digit typing results of HLA-DRB1 and HLA-DQB1 genes and from inferred HLA-DQA1 allelic data using dbMHC database. In order to further investigate the possible sequence features that confer a greater risk to AH, multivariate logistic regression was performed to test for associations between polymorphic amino acid positions in HLA class II alleles and AH. In our patients haematologica | 2017; 102(9)

Discussion

1583


N. Kutszegi et al. Table 3. Polymorphic amino acid positions in HLA-DRB1, HLA-DQB1 and HLA-DQA1 associated with E. coli asparaginase hypersensitivity in acute lymphoblastic leukemia patients (n=257).

HLA class II AA Risk locus position1 residue2

OR3

95% CI4

P3

Other possible amino acids5

HLA-DRB1

HLA-DQA1

HLA-DQB1

10

Q

2.22

1.45-3.40

2.30×10-4

E,Y

11 13 14 25 30 37 57 60 73 74 78 47 50 52 53 54 28 30 37 46 47 52 55 57 71 74

G Y K Q L F V S G Q V K L H R L S S I E F L L A K A

3.69 3.69 3.69 3.69 3.69 3.69 3.52 3.52 2.96 3.69 4.01 3.69 3.03 3.69 3.03 3.69 2.87 2.87 2.87 2.87 2.87 2.87 2.87 2.60 2.87 2.87

1.99-6.84 1.99-6.84 1.99-6.84 1.99-6.84 1.99-6.84 1.99-6.84 2.00-6.19 2.00-6.19 1.71-5.12 1.99-6.84 2.18-7.37 1.99-6.84 1.78-5.14 1.99-6.84 1.78-5.14 1.99-6.84 1.66-4.97 1.66-4.97 1.66-4.97 1.66-4.97 1.66-4.97 1.66-4.97 1.66-4.97 1.58-4.26 1.66-4.97 1.66-4.97

3.45×10-5 3.45×10-5 3.45×10-5 3.45×10-5 3.45×10-5 3.45×10-5 1.31×10-5 1.31×10-5 1.10×10-4 3.45×10-5 8.16×10-6 3.45×10-5 4.09×10-5 3.45×10-5 4.09×10-5 3.45×10-5 1.69×10-4 1.69×10-4 1.69×10-4 1.69×10-4 1.69×10-4 1.69×10-4 1.69×10-4 1.61×10-4 1.69×10-4 1.69×10-4

D,L,P,S,V F,G,H,R,S E R C,G,H,R,Y L,N,S,Y D,S Y A A,E,L,R Y C,Q,R E,V R,S K,Q F T H,Y D,Y V Y P P,R D,S,V A,D,T E,S

HLA class II alleles with risk residue *01:01, *01:02, *04:01, *04:02, *04:03, *04:04, *04:05, *04:07, *07:01, *09:01, *15:01, *15:02,*16:01, *16:02 *07:01 *07:01 *07:01 *07:01 *07:01 *07:01 *07:01, *09:01, *12:01 *07:01, *09:01, *12:01 *07:01, *03:01 *07:01 *07:01, *09:01 *02:01 *02:01, *03:01 *02:01 *02:01, *03:01 *02:01 *02:01, *02:02 *02:01, *02:02 *02:01, *02:02 *02:01, *02:02 *02:01, *02:02 *02:01, *02:02 *02:01, *02:02 *02:01, *02:02, *03:02, *03:04, *03:05 *02:01, *02:02 *02:01, *02:02

HLA class II alleles significantly associated with asparaginase hypersensitivity are in bold. 1Amino acid (AA) position of the HLA class II protein sequence without the leader signal sequence. 2Amino acid associated with asparaginase hypersensitivity. 3For the association between the allele and asparaginase hypersensitivity. 495% confidence interval for the estimated odds ratio. 5Other possible amino acids refer to other amino acids within that amino acid position observed in our patients (n=257).

pediatric ALL patients of European ancestry revealed the association of HLA-DRB1*07:01 allele with AH.9 Later, the same group confirmed the role of this allele in an ethnically diverse pediatric population. In our study, we also confirmed these findings, but after haplotype reconstruction we also found that only the HLA-DRB1*07:01–HLADQB1*02:02 haplotype associated significantly, and the HLA-DRB1*07:01–HLA-DQB1*03:03 haplotype did not reach statistical significance, possibly because of the lower allele frequency (0.038 vs. 0.091 for the HLA-DRB1*07:01– HLA-DQB1*02:02 haplotype). However, further investigations are warranted to elucidate the actual causality. The importance of the HLA-DQB1*02:02 allele in the hypersensitivity was also shown by the BN-BMLA method, which showed that if both HLA-DRB1 and HLA-DQB1 alleles were included in the Bayesian Network only HLADQB1*02:02 allele had direct strong relevance to AH with a high posterior probability (0.95). The situation is the same with the HLA-DQA1*02:01 allele as with the HLADRB1*07:01 allele, because they are in complete linkage disequilibrium with each other. This may suggest that, in 1584

our population, the HLA-DQB1*02:02 allele also seems to be an important genetic risk factor for AH in this region. However, when the amino acid results are evaluated, there is some controversy in this statement. It is well known that proteins coded by the HLA-D, or also known as the MHC class II genes play a critical role in the immune response to extracellular antigens. Antigen presenting cells present processed antigens, or epitopes bound in a pocket formed by the MHC II chains to T-cell receptors on the surface of helper T cells, resulting in cytokine secretion by T cells leading to, among others, differentiation of B cells into antibody-secreting plasma cells. According to the theory, the stronger the MHC II protein binds an epitope, the stronger the immune response is. The strength of the binding is determined by the 3D structure of the MHC II protein which depends on its amino acid composition. Table 2 shows the amino acids associated with the AH and the different HLA class II alleles coding for these variants. Regarding E. coli asparaginase epitopes, Fernandez et al. predicted with bioinformatic methods that the HLA-DRB1*07:01 is a high binding allele haematologica | 2017; 102(9)


HLA haplotype may predict asparaginase allergy

Figure 4. A posteriori strong relevance of predictors to asparaginase hypersensitivity. The BN-BMLA approach was used to investigate the different dependency relations between HLA alleles, age, acute lymphoblastic leukemia (ALL) immunophenotype, sex, treatment protocol, risk group and asparaginase hypersensitivity. HLA-DQB1*02:02 allele and risk group had direct strong relevance to asparaginase hypersensitivity (P=0.95 and 1.00, respectively).

which could be the reason for its association with AH. In our results, several amino acids within the HLA-DQB1 protein were also associated with increased AH, and all of them were present in HLA-DQB1*02:02. But they were also present in HLA-DQB1*02:01 which was not associated with AH. It means that from these amino acid results it cannot be predicted why amino acids in HLADQB1*02:02 were associated with AH. This may mean that although the presence of HLA-DQB1*02:02 allele may be decisive in the development of AH, the presence of the HLA-DRB1*07:01 and/or HLA-DQA1*02:01 alleles may also be necessary for the increased risk, raising the possibility that it is the extended haplotype rather than the individual alleles that is important in this respect. There was another interesting finding in the present study. A significantly smaller proportion of T-cell ALL patients carried the HLA-DQB1*02:02 allele than did preB-cell ALL patients (6.5%; vs. 19.2% of patients with Tcell and pre-B-cell ALL, respectively). This may suggest two interpretations. Either the allele increases the risk to pre-B-cell ALL or decreases the risk to T-cell ALL. Earlier the HLA-DQB1*02:02 allele was found to be associated with increased risk of celiac disease indicating its autoimmunogenic potential.21 Although we did not determine its allele frequency in a healthy control population, its frequency data in The Allele Frequency Net Database in Caucasian populations were around the value found in the pre-B-cell ALL patients in our study. Thus, in contrast its role in celiac disease, the HLA-DQB1*02:02 allele might decrease the risk of the development of T-cell ALL. haematologica | 2017; 102(9)

Furthermore, because this allele is always on the HLADRB1*07:01–HLA-DQA1*02:01–HLA-DQB1*02:02 haplotype, this could mean that this haplotype might provide some protection against development of T-cell ALL but not against pre-B-cell ALL. Naturally, subjects from the same population must be compared to test this assumption. This study has some limitations. First, patients who died during the chemotherapy due to therapy-resistant progressive disease, or due to infections or toxicities of therapy are under-represented in our ALL population. Furthermore, data about hypersensitivity reactions to E. coli asparaginase was collected retrospectively from the patients' files. This does not allow for meticulous documentation or fine grading of hypersensitivity reactions. In some cases the inference of HLA-DQA1 alleles could not be precisely carried out. For further analyses with HLA-DQA1 alleles, data from undetermined haplotypes were merged into one of the possible HLA-DQA1 allele categories (Online Supplementary Table S4). In most cases, exon 2 of the involved alleles was identical at the amino acid sequence level and no haplotype was found with the other HLA-DQA1 allele in The Allele Frequency Net Database either. In the case of HLA-DRB1*13:01–HLADQA1*01:02/*01:03–HLA-DQB1*06:03 haplotype, the HLA-DQA1 allele was regarded as HLA-DQA1*01:03 because it was more likely based on the overall Caucasian haplotype frequency data. To increase the reliability of our inference results, a strict algorithm was applied, which led to the over-representation of common haplotypes. 1585


N. Kutszegi et al.

There are, however, several strengths to this study. First, we studied a relatively large and homogenous European population (Hungarian). Second, we used NGS technology to determine the HLA-DQB1 and HLA-DQB1 HLA types, which is more accurate than the SNP-based imputation used in other studies. In addition, the nucleotide sequence-based HLA typing, in contrast to the costly and laborious direct experimental HLA typing, allows polymorphic HLA amino acid positions from the HLA alleles to be inferred. Third, ours is the first study to evaluate the

References 1. Silverman LB, Gelber RD, Dalton VK, et al. Improved outcome for children with acute lymphoblastic leukemia: results of DanaFarber Consortium Protocol 91-01. Blood. 2001;97(5):1211-1218. 2. Rizzari C, Conter V, Stary J, et al. Optimizing asparaginase therapy for acute lymphoblastic leukemia. Curr Opin Oncol. 2013;25 Suppl 1:S1-9. 3. Pieters R, Hunger SP, Boos J, et al. Lasparaginase treatment in acute lymphoblastic leukemia: a focus on Erwinia asparaginase. Cancer. 2011;117(2):238-249. 4. Schmiegelow K, Attarbaschi A, Barzilai S, et al. Consensus definitions of 14 severe acute toxic effects for childhood lymphoblastic leukaemia treatment: a Delphi consensus. Lancet Oncol. 2016;17(6):e231-239. 5. Chen SH, Pei D, Yang W, et al. Genetic variations in GRIA1 on chromosome 5q33 related to asparaginase hypersensitivity. Clin Pharmacol Ther. 2010;88(2):191-196. 6. Fernandez CA, Smith C, Yang W, et al. Genome-wide analysis links NFATC2 with asparaginase hypersensitivity. Blood. 2015;126(1):69-75. 7. Rajic V, Debeljak M, Goricar K, Jazbec J. Polymorphisms in GRIA1 gene are a risk factor for asparaginase hypersensitivity during the treatment of childhood acute lymphoblastic leukemia. Leuk Lymphoma. 2015:1-6. 8. Kutszegi N, Semsei AF, Gezsi A, et al. Subgroups of Paediatric Acute Lymphoblastic Leukaemia Might Differ Significantly in Genetic Predisposition to Asparaginase Hypersensitivity. PLoS One. 2015;10(10):e0140136.

1586

role of HLA haplotypes in AH, and our results indicated that the HLA haplotype-based risk prediction might be more precise than the single allele-based risk calculation. In conclusion, in our study, we found that the HLA-DRB1*07:01–HLA-DQA1*02:01–HLA-DQB1*02:02 haplotype was associated with a high risk of E. coli AH in pediatric ALL patients. Based on our results, besides the previously described HLA-DRB1*07:01 allele, the HLA-DQB1*02:02 allele might also be important in the development of AH. This needs further investigation.

9. Fernandez CA, Smith C, Yang W, et al. HLA-DRB1*07:01 is associated with a higher risk of asparaginase allergies. Blood. 2014;124(8):1266-1276. 10. Cao H, Wang Y, Zhang W, et al. A shortread multiplex sequencing method for reliable, cost-effective and high-throughput genotyping in large-scale studies. Hum Mutat. 2013;34(12):1715-1720. 11. Stephens M, Scheet P. Accounting for decay of linkage disequilibrium in haplotype inference and missing-data imputation. Am J Hum Genet. 2005;76(3):449-462. 12. Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data. Am J Hum Genet. 2001;68(4):978-989. 13. Gonzalez-Galarza FF, Takeshita LY, Santos EJ, et al. Allele frequency net 2015 update: new features for HLA epitopes, KIR and disease and HLA adverse drug reaction associations. Nucleic Acids Res. 2015;43(Database issue):D784-788. 14. Hullám G, Antal P, Szalai C, Falus A. Evaluation of a Bayesian model-based approach in GA studies. In: Sašo D, Pierre G, Juho R, eds. Proceedings of the third International Workshop on Machine Learning in Systems Biology. Proceedings of Machine Learning Research: PMLR, 2009:30-43. 15. Antal P, Hullam G, Gézsi A, Millinghoffer A. Learning complex bayesian network features for classification. In: Studený M, Vomlel J, editors. Proceedings of the Third European Workshop on Probabilistic Graphical Models; 2006 September 12−15; Prague: Czech Republic: Action M Agency; 2006. p. 9-16. 16. Antal P, Millinghoffer A, Hullám G, et al.

17.

18.

19.

20.

21.

Bayesian, systems-based, multilevel analysis of associations for complex phenotypes: from interpretation to decisions. In: Sinoquet C, Mourad R, eds. Probabilistic Graphical Models for Genetics, Genomics and Postgenomics. Oxford: Oxford University Press, 2014:318-352. Antal P, Millinghoffer A, Hullám G, Szalai C, Falus A. A Bayesian View of Challenges in Feature Selection: Feature Aggregation, Multiple Targets, Redundancy and Interaction. In: Yvan S, Huan L, Iñaki I, Louis W, Yves Van de P, eds. Proceedings of the Workshop on New Challenges for Feature Selection in Data Mining and Knowledge Discovery at ECML/PKDD 2008. Proceedings of Machine Learning Research: PMLR, 2008:74-89. Ungvari I, Hullam G, Antal P, et al. Evaluation of a partial genome screening of two asthma susceptibility regions using bayesian network based bayesian multilevel analysis of relevance. PLoS One. 2012;7(3):e33573. Lautner-Csorba O, Gezsi A, Erdelyi DJ, et al. Roles of genetic polymorphisms in the folate pathway in childhood acute lymphoblastic leukemia evaluated by Bayesian relevance and effect size analysis. PLoS One. 2013;8(8):e69843. Lautner-Csorba O, Gezsi A, Semsei AF, et al. Candidate gene association study in pediatric acute lymphoblastic leukemia evaluated by Bayesian network based Bayesian multilevel analysis of relevance. BMC Med Genomics. 2012;5:42. Dieli-Crimi R, Cenit MC, Nunez C. The genetics of celiac disease: A comprehensive review of clinical implications. J Autoimmun. 2015;64:26-41.

haematologica | 2017; 102(9)


ARTICLE

Chronic Lymphocytic Leukemia

The prohibitin-binding compound fluorizoline induces apoptosis in chronic lymphocytic leukemia cells through the upregulation of NOXA and synergizes with ibrutinib, 5-aminoimidazole-4-carboxamide riboside or venetoclax

Ana M. Cosialls,1,* Helena Pomares,1,2,* Daniel Iglesias-Serret,1 José Saura-Esteller,1 Sonia Núñez-Vázquez,1 Diana M. González-Gironès,1 Esmeralda de la Banda,3 Sara Preciado,4 Fernando Albericio,4,5,6 Rodolfo Lavilla,5,7 Gabriel Pons,1 Eva M. González-Barca2 and Joan Gil1

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(9):1587-1593

Departament de Ciències Fisiològiques, Facultat de Medicina i Ciències de la Salut, Universitat de Barcelona–IDIBELL (Institut d’Investigació Biomèdica de Bellvitge), L'Hospitalet de Llobregat, Barcelona, Spain; 2Servei d’Hematologia Clínica, Institut Català d’Oncologia (ICO)-IDIBELL, L'Hospitalet de Llobregat, Barcelona, Spain; 3Unitat de Citohematologia, Servei d’Anatomia Patològica, Hospital Universitari de Bellvitge (HUB)-IDIBELL, L'Hospitalet de Llobregat, Barcelona, Spain; 4Department of Organic Chemistry, University of Barcelona, Spain; 5CIBER-BBN, Networking Centre on Bioengineering, Biomaterials and Nanomedicine, Barcelona Science Park, Spain; 6 School of Chemistry & Physics, University of KwaZulu-Natal, Durban, South Africa and 7 Laboratory of Organic Chemistry, Faculty of Pharmacy, University of Barcelona, Spain 1

*

AMC and HP contributed equally to this work.

ABSTRACT

F

luorizoline is a new synthetic molecule that induces apoptosis by selectively targeting prohibitins. In the study herein, the pro-apoptotic effect of fluorizoline was assessed in 34 primary samples from patients with chronic lymphocytic leukemia. Fluorizoline induced apoptosis in chronic lymphocytic leukemia cells at concentrations in the low micromolar range. All primary samples were sensitive to fluorizoline irrespective of patients’ clinical or genetic features, whereas normal T lymphocytes were less sensitive. Fluorizoline increased the protein levels of the pro-apoptotic B-cell lymphoma 2 family member NOXA in chronic lymphocytic leukemia cells. Furthermore, fluorizoline synergized with ibrutinib, 5-aminoimidazole-4-carboxamide riboside or venetoclax to induce apoptosis. These results suggest that targeting prohibitins could be a new therapeutic strategy for chronic lymphocytic leukemia.

Introduction Chronic lymphocytic leukemia (CLL) is a malignant lymphoproliferative disorder of monoclonal B lymphocytes that accumulate in the blood, bone marrow, lymph nodes and other lymphoid tissues.1,2 It represents the most common adult leukemia in the western world, mainly affecting elderly individuals. Although the progression-free survival (PFS) and overall survival (OS) of CLL patients have increased with the introduction of first-line therapy, there is no cure for CLL and all patients will ultimately relapse. The standard of treatment for physically fit patients is chemoimmunotherapy with fludarabine, cyclophosphamide and rituximab (FCR),3 and for older patients bendamustine plus rituximab may be a better option.4 Relapsed patients or those with altered TP53 can be treated with the bruton tyrosine kinase (BTK) inhibitor ibrutinib, and also with the phosphoinositide 3-kinase (PI3K) inhibitor idelalisib or the B-cell lymphoma 2 (BCL-2) inhibitor venetoclax (ABT-199).5 Recently, ibrutinib has been approved to treat CLL patients in first-line therapy.5 Nevertheless, a percentage of patients are resistant to ibrutinib or do not haematologica | 2017; 102(9)

Correspondence: jgil@ub.edu

Received: December 22, 2016. Accepted: June 8, 2017. Pre-published: June 15, 2017. doi:10.3324/haematol.2016.162958 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1587 ©2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1587


Ana M. Cosialls et al. tolerate the drug.6-9 Therefore, it is necessary to identify new agents with selective toxicity for malignant B cells and to develop therapeutic strategies that can overcome cellular resistance mechanisms to current therapies, that can overcome cellular resistance mechanisms to current therapies. Hence, the nucleoside analogue 5-aminoimidazole-4-carboxamide riboside (AICAR) induces apoptosis in CLL cells independently of p53 status.10 Recently, our group has described novel pro-apoptotic small molecules with fluorinated thiazole scaffolds.11 The diaryl trifluorothiazoline compound 1a, also termed fluorizoline (Figure 1A), was selected as the best apoptosis inductor in a wide range of cancer cell lines from different tissue origin, including hematopoietic cell lines, and different p53 status, proving that fluorizoline exerts its antitumor action in a p53-independent manner. Fluorizoline selectively binds to prohibitin (PHB) 1 and 211 and, strik-

1588

A

B

C

D

E

F

ingly, these proteins are necessary for apoptosis induction by this compound.12 Fluorizoline treatment induces mitochondrial-mediated apoptosis, a pathway that is controlled by the BCL-2 family proteins. In this regard, induction of the pro-apoptotic protein NOXA is required for fluorizoline-induced apoptosis, and BIM is also involved depending on the cellular context.12 Prohibitins are ubiquitous, evolutionarily conserved scaffold proteins mainly localized in mitochondria and implicated in many cellular processes, including mitochondrial biogenesis, differentiation, cell survival and apoptosis. Two highly homologous proteins, PHB1 and PHB2/REA, have been described.13,14 Increasing evidence links PHBs and tumorigenesis.15,16 In B lymphocytes, PHBs were identified as proteins associated with the immunoglobulin M (IgM) B cell receptor (BCR).17 More recently, PHBs were described as pro-

Figure 1. Cytotoxicity of fluorizoline in CLL cells ex vivo. (A) Chemical structure of fluorizoline. (B) Dose response of fluorizoline on primary CLL cells. PBMNC from 5 representative untreated CLL patients (#6, 9, 12, 14 and 29) out of 34 were incubated for 24 h with increasing doses of fluorizoline ranging from 1.25 to 10 µM. (C) PBMNC from 34 CLL patients were incubated for 24 h with or without 10 µM fluorizoline. (D) Time course of fluorizoline-induced apoptosis in CLL cells. Cells from 5 patients were untreated or incubated for different times ranging from 2 to 24 h with 10 mM fluorizoline. (E) Dose response of the cytotoxic effect of fluorizoline on B and T cells from CLL patients. PBMNC from CLL patients were incubated for 24 h with increasing doses of fluorizoline ranging from 1.25 to 10 mM. Viability was measured on CD3+ (T cells, n=15) and CD19+ (B cells, n=34) populations. (F) Dose response of the cytotoxic effect of fluorizoline on normal B and T cells. PBMNC from 12 healthy donors were incubated for 24 h with increasing doses of fluorizoline ranging from 1.25 to 10 mM. Viability was measured on CD3+ (T cells) and CD19+ (B cells) populations. (B, C, D, E and F) Viability (annexin V negative) was measured by analysis of phosphatidylserine exposure in total population or in CD19+ and CD3+ populations and is expressed (B and C) as the percentage of non-apoptotic cells or (D, E and F) as the percentage of the viability of untreated cells. (D, E and F) Data are shown as the mean±SEM. (D) Two-tailed paired Student’s t-test was performed. (E and F) Two-tailed unpaired Student’s t-test was performed. Significant P values are indicated: *P<0.05; **P<0.01; ***P<0.001 treated versus untreated cells or CD19+ versus CD3+ cells.

haematologica | 2017; 102(9)


Targeting PHBs induces apoptosis in CLL cells

teins associated with phosphorylated protein tyrosine kinase Syk18 and the receptor CD8619 at the inner plasma membranes of B lymphocytes, thus likely having a role in signal transduction after receptor engagement. In CLL cells, PHB is increased after phorbol ester-induced maturation,20 and nuclear PHB is upregulated after in vitro treatment with fludarabine.21 These data strongly suggest that targeting PHBs would be a candidate approach for the treatment of B-cell neoplasias, and PHB-binding compounds, such as fluorizoline, emerge as interesting new pro-apoptotic agents. Preliminary data in a small number of CLL samples showed that fluorizoline induces apoptosis in these cells.11 The objective of the work herein was to investigate the mechanism of induction of apoptosis by fluorizoline in CLL cells and the effect of its combination with ibrutinib, AICAR or venetoclax.

Fluorizoline (a diaryl trifluorothiazoline; see molecular structure in Figure 1A) was synthesized as previously described.11 Other reagents used in this study are detailed in the Online Supplementary Data.

Analysis of cell purity and cell viability by flow cytometry Cell viability was assessed by phosphatidylserine exposure and measured as the percentage of annexin V negative cell population. Cells were acquired using the FACSCantoTM II flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA) and the data of total cells or CD19+- or CD3+-gated cells were analyzed using FACSDivaTM software (Becton Dickinson). A detailed protocol of cell staining can be found in the Online Supplementary Data.

Reverse transcriptase multiplex ligation-dependent probe amplification (RT-MLPA)

Methods Primary samples and cell isolation Peripheral blood (PB) samples from 34 untreated patients with CLL and 12 healthy donors were included. All patients and healthy controls signed an informed consent form approved by the Institutional Review Boards according to the Declaration of Helsinki. The patients’ characteristics are shown in Online Supplementary Table S1. Briefly, PB mononuclear cells (PBMNC) were obtained by centrifugation on a Biocoll gradient. This fraction included normal B and T cells from healthy donors or mainly B-CLL cells from patients. To ensure high B cell purity (≥80%), an initial isolation step was performed by negative selection. The complete sample handling protocol is available in the Online Supplementary Data.

A

Reagents

Ribonucleic acid (RNA) isolated from cells was analyzed by RTMLPA using SALSA MLPA KIT R011-C1 Apoptosis messenger (m)RNA from MRC-Holland (Amsterdam, The Netherlands) for the simultaneous detection of 40 mRNA molecules, including apoptosis-related genes.12,22 The protocol is accurately described in the Online Supplementary Data.

Western blot The antibodies used in this study and the western blot protocol are described in the Online Supplementary Data.

Statistical analysis Results are shown as the mean ± standard error of the mean (SEM) of values obtained in 3 or more independent experiments as

B

C

D

Figure 2. Induction of NOXA protein by fluorizoline in primary CLL cells. (A) PBMNC from patient #11 were incubated for different times ranging from 2 to 24 h with 10 mM fluorizoline. After the times stated on the figure, cells were collected. (B) PBMNC from patients #5, 12, 23 and 29 were incubated with 10 mM fluorizoline for 24 h. (C) PBMNC from patient #5 were pre-incubated with 20 mM caspase inhibitor Q-VD-OPh for 30 min and then treated with 10 mM fluorizoline for 24 h. (D) PBMNC from patient #8 with 17p deletion were untreated (U) or treated with 5 and 10 mM fluorizoline (F) for 48 h. (A, B, C and D) Cells were lysed and analyzed by western blot. BCL-2 was used for loading normalization. Viability was measured by analysis of phosphatidylserine exposure and is expressed as the percentage of non-apoptotic (annexin V negative) cells. These are representative patients of at least 3 analyzed (n=5 for A; n=8 for B; n=4 for C; n=1 for D).

haematologica | 2017; 102(9)

1589


Ana M. Cosialls et al.

indicated in each figure legend. Data were analyzed using SPSS® Statistics v22.0 software package (IBM®, Armonk, NY, USA). Two-tailed paired or unpaired Student’s t-test with normal-based 95% confidence interval was used to compare the differences between samples, as required. Differences were considered statistically significant at P values below 0.05. In two-drug combination studies, the combination index (CI) was calculated according to the Chou-Talalay method23 by using CalcuSyn software version 2.11 (Biosoft, Cambridge, UK). A CI of below 1 indicates a synergistic effect.

the percentage of viable normal B and T cells (48.6±6.8% and 82.8±6.3% of viable cells at 24 hours of treatment with 10 mM fluorizoline in normal CD19+ and CD3+ populations, respectively) (Figure 1F) with a mean EC50 value of 10.9±0.8 mM and 19.1±2.2 mM at 24 hours for normal B and T cells, respectively. Thus, CLL cells are slightly more sensitive to fluorizoline than normal B lymphocytes.

Results

To analyze the mechanism of apoptosis induction upon fluorizoline treatment in B lymphocytes of CLL samples we sought to examine the changes in the protein levels of NOXA and BIM, 2 members of the BCL-2 family that have been involved previously in fluorizoline-induced apoptosis.12 We observed a time-dependent upregulation of NOXA that was detected after 8 hours of incubation, occurring simultaneously with the decrease of cell viability (Figure 2A), and was also clearly found upregulated after 24 hours in all samples analyzed (Figure 2B). The protein levels of MCL-1, the anti-apoptotic counterpart of NOXA, were slightly upregulated during the first hours of incubation with fluorizoline and were not altered after 24 hours. PUMA and BIM, as well as PHBs protein levels, were not modified upon fluorizoline treatment (Figure 2A). In addition, fluorizoline clearly induced poly(ADP-ribose) polymerase (PARP) cleavage (Figure 2A), thus confirming an apoptotic mechanism. The induction of NOXA preceded caspase activation, as pre-incubation with the pan-caspase inhibitor Q-VD-OPh did not block its upregulation (Figure 2C). As expected, BIM protein expression was not modified upon caspase inhibition, whereas MCL-1 protein levels were increased at 24 hours after caspase activity arrest, both in the absence and the presence of fluorizoline, indicating a late caspase-dependent degradation of this protein (Figure 2C). Finally, the increase of NOXA protein expression was also detected in CLL samples from patients with chromosomal alterations that cause loss of p53 expression (Figure 2D), thus corroborating the fact that fluorizoline-induced NOXA upregulation occurs in a p53-independent manner. Altogether, these results indicate that fluorizoline causes an increase of NOXA protein levels prior to caspase activation and these modulations could explain the apoptotic outcome observed in primary CLL cells. The induction of NOXA protein by fluorizoline could be due to the modulation of the corresponding mRNA levels. To that purpose, we analyzed the changes in the overall apoptosis mRNA expression profile by RT-MLPA. NOXA levels were not modified after fluorizoline treatment, neither at the initial stages nor at 24 hours of incubation of CLL cells (Online Supplementary Figure S1). Among all apoptosis-related genes analyzed, only the pro-apoptotic BCL-2 family member MOAP1 and the anti-apoptotic HIAP1 and HIAP2 were weakly upregulated and downregulated upon fluorizoline treatment, respectively. This result indicates that fluorizoline-induced NOXA protein upregulation does not result from mRNA modulation in CLL cells.

Fluorizoline induces apoptosis in primary CLL cells ex vivo The cytotoxicity of fluorizoline (Figure 1A) was evaluated in samples obtained from patients with CLL prior to any treatment (see Online Supplementary Table S1 for details of patient samples). PBMNC from 34 different patients were exposed ex vivo to a range of fluorizoline concentrations (from 1.25 to 20 mM). Incubation with fluorizoline strongly reduced cell viability in a dose-dependent manner (Figure 1B). All CLL samples were sensitive to fluorizoline, and cell viability decreased from 70.0±1.9% to 28.1±2.6% (n=34) after incubation with 10 mM fluorizoline for 24 hours (Figure 1C), with half-maximal effective concentration (EC50) values ranging from 2.5 to 20 mM (mean 8.1±0.6 mM; Online Supplementary Table S1). Longer exposition to fluorizoline for 48 hours slightly reduced the mean EC50 value to 5.5±0.6 mM (n=25; Online Supplementary Table S1). Treatment with 10 mM fluorizoline induced a time-dependent decrease of cell viability that was detected after the first 8 hours of incubation (Figure 1D). Some patients included in this study harbored alterations of the TP53 and ATM genes detected by fluorescent in situ hybridization (FISH; Online Supplementary Table S1), which are associated with poor response to chemotherapy and chemoimmunotherapy and a worse prognosis.1,2 Interestingly, samples from these patients had similar sensitivity to ex vivo treatment with fluorizoline (mean EC50 value of 8.3±0,5 mM at 24 hours for samples from patients with 17p or 11q deletion, n=5) compared to samples from patients without these alterations (mean EC50 value of 8,1±0,8 mM at 24 hours, n=29). Similarly, ex vivo cytotoxicity of fluorizoline was similar in cells from CLL patients with unmutated and mutated immunoglobulin heavy chain variable region (IGHV) genes (mean EC50 values at 24 hours of 10,0±1,6 mM, n=5; and 10,3±3,5 mM, n=4; respectively) (Online Supplementary Table S1). To examine the effects on the non-leukemic T lymphocytes of CLL patients, apoptosis induction was assayed in the CD3+ population of 15 CLL samples. As depicted in Figure 1E, the reduction in cell viability in the presence of fluorizoline was higher within the leukemic CD19+ population (35.3±34.9% of viable cells at 24 hours treatment with 10 mM fluorizoline) compared to the normal CD3+ population (83.8±7.5% of viable cells at 24 hours treatment with 10 mM fluorizoline, with EC50 values higher than 20 mM in 9 samples out of 15), demonstrating that fluorizoline preferentially induces apoptosis in malignant B lymphocytes. Additionally, to evaluate the cytotoxicity of fluorizoline in non-malignant B cells, the effect of fluorizoline on normal PBMNC from healthy donors was assessed. Incubation with increasing doses of fluorizoline reduced 1590

NOXA is upregulated by fluorizoline in primary CLL cells

Fluorizoline synergizes with ibrutinib, AICAR or venetoclax to induce apoptosis in CLL cells Finally, we sought to analyze the effect of the combination of fluorizoline with other drugs. For that purpose we chose the BTK irreversible inhibitor ibrutinib, the nucleohaematologica | 2017; 102(9)


Targeting PHBs induces apoptosis in CLL cells

side analogue AICAR, which has demonstrated selective anti-tumor activity in CLL ex vivo10,24 and was tested in a phase I/II clinical trial for relapsed/refractory CLL,25 and the BCL-2 inhibitor venetoclax (ABT-199). Ibrutinib therapy in vivo causes an intracellular MCL-1 protein decrease,26 and partially downregulates MCL-1 protein levels in vitro in some CLL patient samples.27 Interestingly, the combination of fluorizoline with ibrutinib was more effective than single drug treatment in all patients analyzed. Thus, the addition of ibrutinib enhanced fluorizoline cytotoxic effect (CI values ranging from 0.192 to 0.797, indicating a synergistic effect; the lowest CI values correspond to the combination of fluorizoline with 10 mM ibrutinib) (Figure 3A). Similarly, the combination of fluorizoline with AICAR increased cell death compared to each drug alone (CI values ranging from 0.643 to 0.991; the lowest CI values correspond to the combination of AICAR with 10 mM fluorizoline) (Figure 3B). Finally, the combination of fluorizoline and venetoclax showed a synergic effect in all conditions analyzed (CI values ranging from 0.492 to 0.824; the lowest CI values correspond to the combination of venetoclax with 10 mM fluorizoline) (Figure 3C). Thus, these results show a synergistic interaction between fluorizoline and ibrutinib, AICAR or venetoclax in CLL cells.

A

B

Discussion In the study herein we describe the mechanism of apoptosis induction by the prohibitin-binding compound fluorizoline in CLL cells. It was previously described that fluorizoline directly binds to PHB1 and PHB211 and reduces cell viability through the upregulation of NOXA and BIM.12 Although PHBs are necessary for apoptosis induction by fluorizoline,12 we cannot discard that it could interact with other proteins. Expression analysis revealed a consistent upregulation of the BH3-only protein NOXA in CLL cells upon fluorizoline treatment. Fluorizoline induced increases in NOXA protein levels prior to caspase activation, which could explain the apoptotic outcome. Formerly, the effects of fluorizoline in primary cancer cells have been analyzed in cells from patients with chronic myeloid leukemia in blast crisis, mantle cell lymphoma, B cell chronic lymphoproliferative syndrome, adult T-cell leukemia/lymphoma,12 and in acute myeloid leukemia cells.28 Interestingly, treatment with fluorizoline resulted in a decrease in viability and an increase in NOXA protein levels, whereas BIM protein levels were not modified. Hence, fluorizoline seems to mainly increase NOXA protein levels in leukemia cells. NOXA has been described as a particularly relevant proapoptotic BCL-2 family member in CLL cells. NOXA protein is induced by histone deacetylase (HDAC) inhibitors,29,30 proteasome inhibitors,31 bendamustine,32 Akt inhibitors,33 AICAR,10 cyclin-dependent kinase inhibitors34 and microtubule disrupting agents.35 Furthermore, CLL development is accelerated in mice with a deficiency of NOXA.36 NOXA is a pro-apoptotic BH3-only member that has been classified as a “sensitizer” because it was considered as an inhibitor of the anti-apoptotic MCL-1 and A1 proteins.37 However, recent data indicate that NOXA is also an “activator” of the BAX and BAK multidomain pro-apoptotic BCL-2 family members.38 Related to our study with a PHB-binding compound in CLL, rocaglamide silvestrol induces apoptosis of CLL haematologica | 2017; 102(9)

C

Figure 3. Combination of fluorizoline and other drugs in CLL cells ex vivo. (A) PBMNC from 5-9 untreated CLL patients (#11, 12, 15, 16, 17, 18, 23, 28 and 30) were incubated for 24 h with increasing doses of fluorizoline ranging from 1.25 to 10 mM and combined with increasing doses of ibrutinib ranging from 1 to 10 mM. (B) PBMNC from 4-5 untreated CLL patients (#5, 11, 17, 18 and 23) were incubated for 24 h with increasing doses of fluorizoline ranging from 1.25 to 10 mM and combined with increasing doses of 5-aminoimidazole-4-carboxamide riboside (AICAR) ranging from 0.125 to 0.5 mM. (C) PBMNC from 5-6 untreated CLL patients (#10, 13, 14, 16, 23 and 25) were incubated for 24 h with increasing doses of fluorizoline ranging from 1.25 to 10 mM and combined with increasing doses of venetoclax (ABT-199) ranging from 0.1 to 1 nM. Viability was measured by analysis of phosphatidylserine exposure and is expressed as the percentage of the viability (annexin V negative) of untreated cells. Data are shown as the mean±SEM.

1591


Ana M. Cosialls et al. cells.39 Interestingly, some rocaglamides bind to PHB and it has been described that this binding mediates its anti-proliferative effects through inhibition of the Raf-MEK-ERK pathway.40 However, whether or not silvestrol binds to PHB is presently unknown. Our study has shown that T cells from CLL patients are less sensitive to fluorizoline-induced apoptosis. Thus, the differential effect of fluorizoline in B and T lymphocytes is of great interest and may be useful in the therapy of CLL, since immunosuppression caused by classic chemotherapy could be avoided or reduced. In vivo experiments could be necessary in the future and it is possible that the effective concentration of fluorizoline would be higher than in the in vitro conditions. Thus, further testing the effect of fluorizoline in CLL cells co-cultured with bone marrow stromal cells would be interesting in order to better reproduce the CLL microenvironment conditions.41 Finally, our results show that the combination of fluorizoline with ibrutinib, AICAR or venetoclax has synergistic effects in the induction of apoptosis in CLL cells. Likely the induction of NOXA by fluorizoline is involved in these synergistic effects and could overcome resistance to BCL-2 inhibitors that do not inhibit MCL-1.

References 1. Zenz T, Mertens D, Kuppers R, Dohner H, Stilgenbauer S. From pathogenesis to treatment of chronic lymphocytic leukaemia. Nat Rev Cancer. 2010;10(1):37-50. 2. Hallek M. Chronic lymphocytic leukemia: 2013 update on diagnosis, risk stratification and treatment. Am J Hematol. 2013; 88(9):803-816. 3. Hallek M, Fischer K, Fingerle-Rowson G, et al. Addition of rituximab to fludarabine and cyclophosphamide in patients with chronic lymphocytic leukaemia: A randomised, open-label, phase 3 trial. Lancet. 2010;376(9747):1164-1174. 4. Jain N, O’Brien S. Initial treatment of CLL: Integrating biology and functional status. Blood. 2015;126(4):463-470. 5. Robak T, Stilgenbauer S, Tedeschi A. Frontline treatment of CLL in the era of novel agents. Cancer Treat Rev. 2017;53:70-78. 6. Woyach JA, Furman RR, Liu T-M, et al. Resistance mechanisms for the bruton’s tyrosine kinase inhibitor ibrutinib. New Engl J Med. 2014;370(24):2286-2294. 7. Furman RR, Cheng S, Lu P, et al. Ibrutinib resistance in chronic lymphocytic leukemia. N Engl J Med. 2014; 370(24):2352-2354. 8. Mato AR, Nabhan C, Barr PM, Ujjani CS, Hill BT, Lamanna N, et al. Outcomes of CLL patients treated with sequential kinase inhibitor therapy: a real world experience. Blood. 2016;128(18):2199–2205. 9. UK CLL Forum. Ibrutinib for relapsed/refractory chronic lymphocytic leukemia: a UK and Ireland analysis of outcomes in 315 patients. Haematologica. 2016;101(12):1563-1572. 10. Santidrián AF, González-Gironès DM, Iglesias-Serret D, et al. AICAR induces apoptosis independently of AMPK and p53 through up-regulation of the BH3-only proteins BIM and NOXA in chronic lym-

1592

11.

12.

13. 14.

15.

16.

17.

18.

19.

20.

Altogether, our results suggest that fluorizoline could be an alternative therapy for resistant/refractory patients to classic chemotherapy or novel drugs such as kinase inhibitors or BCL-2 antagonists approved for the treatment of CLL. Acknowledgments The authors would like to thank the Scientific and Technological Centers of the Bellvitge Campus at the University of Barcelona (CCiTUB) for helpful discussions and suggestions. Moreover, we thank the Genomics Facility from the CCiTUB for their technical support. Funding This study was supported by grants from the Ministerio de Economía y Competitividad (SAF2013-41611-R to JG and BQU-CTQ2015-67870-P to RL), the Instituto de Salud Carlos III (RTICC RD12/0036/0029 to JG), Fundació Bosch i Gimpera (AVCRI-PPV022-08 to JG) and the AGAURGeneralitat de Catalunya (2014SGR935 to JG and 2014SGR137 to FA). JS-E and SN-V are recipients of research fellowships from the Ministerio de Economía y Competitividad and Universitat de Barcelona, respectively.

phocytic leukemia cells. Blood. 2010; 116(16):3023-3032. Pérez-Perarnau A, Preciado S, Palmeri CM, et al. A trifluorinated thiazoline scaffold leading to pro-apoptotic agents targeting prohibitins. Angew Chemie Int Ed. 2014; 53(38):10150-10154. Moncunill-Massaguer C, Saura-Esteller J, Pérez-Perarnau A, et al. A novel prohibitinbinding compound induces the mitochondrial apoptotic pathway through NOXA and BIM upregulation. Oncotarget. 2015;6(39):41750-41765. Artal-Sanz M, Tavernarakis N. Prohibitin and mitochondrial biology. Trends Endocrinol Metab. 2009;20(8):394-401. Osman C, Merkwirth C, Langer T. Prohibitins and the functional compartmentalization of mitochondrial membranes. J Cell Sci. 2009;122(Pt 21):38233830. Theiss AL, Sitaraman SV. The role and therapeutic potential of prohibitin in disease. Biochim Biophys Acta. 2011; 1813(6):1137-1143. Thuaud F, Ribeiro N, Nebigil CG, Désaubry L. Prohibitin ligands in cell death and survival: Mode of action and therapeutic potential. Chem Biol. 2013; 20(3):316-331. Terashima M, Kim KM, Adachi T, et al. The IgM antigen receptor of B lymphocytes is associated with prohibitin and a prohibitin-related protein. EMBO J. 1994;13(1):3782-3792. Paris LL, Hu J, Galan J, et al. Regulation of Syk by phosphorylation on serine in the linker insert. J Biol Chem. 2010; 285(51):39844-39854. Lucas CR, Cordero-Nieves HM, Erbe RS, et al. Prohibitins and the cytoplasmic domain of CD86 cooperate to mediate CD86 signaling in B lymphocytes. J Immunol. 2013; 190(2):723-736. Woodlock TJ, Bethlendy G, Segel GB. Prohibitin expression is increased in phor-

21.

22.

23.

24.

25.

26.

27.

28.

bol ester-treated chronic leukemic B-lymphocytes. Blood Cells Mol Dis. 2001;27(1):27–34. Henrich S, Mactier S, Best G, Mulligan SP, Crossett B, Christopherson RI. Fludarabine nucleoside modulates nuclear “survival and death” proteins in resistant chronic lymphocytic leukemia cells. Nucleosides Nucleotides Nucleic Acids. 2011;30(12): 1181–1189. Eldering E, Spek CA, Aberson HL, et al. Expression profiling via novel multiplex assay allows rapid assessment of gene regulation in defined signalling pathways. Nucleic Acids Res. 2003;31(23):e153. Chou T-C, Talalay P. Analysis of combined drug effects: a new look at a very old problem. Trends Pharmacol Sci. 1983;4:450454. Campàs C, López JM, Santidrián AF, Bellosillo B, Colomer D, Gil J. Acadesine activates AMPK and induces apoptosis in B-cell chronic lymphocytic leukemia cells but not in T lymphocytes. Blood. 2003; 101(9):3674-3680. Van den Neste E, Cazin B, Janssens A, et al. Acadesine for patients with relapsed/refractory chronic lymphocytic leukemia (CLL): A multicenter phase I/II study. Cancer Chemother Pharmacol. 2013;71(3):581-591. Cervantes-Gómez F, Lamothe B, Woyach JA, et al. Pharmacological and protein profiling suggests venetoclax (ABT-199) as optimal partner with ibrutinib in chronic lymphocytic leukemia. Clin Cancer Res. 2015;21(16):3705-3715. Bojarczuk K, Sasi BK, Gobessi S, et al. BCR signaling inhibitors differ in their ability to overcome Mcl-1-mediated resistance of CLL B cells to ABT-199. Blood. 2016; 127(25):3192-3201. Pomares H, Palmeri CM, Iglesias-Serret D, et al. Targeting prohibitins induces apoptosis in acute myeloid leukemia cells.

haematologica | 2017; 102(9)


Targeting PHBs induces apoptosis in CLL cells

Oncotarget. 2016;7(40):64987-65000. 29. Inoue S, Riley J, Gant TW, Dyer MJS, Cohen GM. Apoptosis induced by histone deacetylase inhibitors in leukemic cells is mediated by Bim and Noxa. Leukemia. 2007;21(8):1773-1782. 30. Pérez-Perarnau A, Coll-Mulet L, RubioPatiño C, et al. Analysis of apoptosis regulatory genes altered by histone deacetylase inhibitors in chronic lymphocytic leukemia cells. Epigenetics. 2011;6(10):1228-1235. 31. Smit LA, Hallaert DYH, Spijker R, et al. Differential Noxa / Mcl-1 balance in peripheral versus lymph node chronic lymphocytic leukemia cells correlates with survival capacity. Blood. 2007;109(4):16601668. 32. Roué G, López-Guerra M, Milpied P, et al. Bendamustine is effective in p53-deficient B-cell neoplasms and requires oxidative stress and caspase-independent signaling. Clin Cancer Res. 2008;14(21):6907–6915.

haematologica | 2017; 102(9)

33. De Frias M, Iglesias-Serret D, Cosialls AM, et al. Akt inhibitors induce apoptosis in chronic lymphocytic leukemia cells. Haematologica. 2009;94(12):1698-1707. 34. Paiva C, Godbersen JC, Soderquist RS, et al. Cyclin-dependent kinase inhibitor P1446A induces apoptosis in a JNK/p38 MAPK-dependent manner in chronic lymphocytic leukemia B-cells. PLoS One. 2015;10(11):1-16. 35. Bates D, Feris EJ, Danilov A V., Eastman A. Rapid induction of apoptosis in chronic lymphocytic leukemia cells by the microtubule disrupting agent BNC105. Cancer Biol Ther. 2016;17(3):291-299. 36. Slinger E, Wensveen F, Guikema J, Kater A, Eldering E. Chronic lymphocytic leukemia development is accelerated in mice with deficiency of the pro-apoptotic regulator NOXA. Haematologica. 2016;101(9):e374377. 37. Delbridge AR, Strasser A. The BCL-2 pro-

38.

39.

40.

41.

tein family, BH3-mimetics and cancer therapy. Cell Death Differ. 2015;22(7):10711080. Chen H-C, Kanai M, Inoue-Yamauchi A, et al. An interconnected hierarchical model of cell death regulation by the BCL-2 family. Nat Cell Biol. 2015;17(10):1270–1281. Lucas DM, Edwards RB, Lozanski G, et al. The novel plant-derived agent silvestrol has B-cell selective activity in chronic lymphocytic leukemia and acute lymphoblastic leukemia in vitro and in vivo. Blood. 2009; 113(19):4656–4666. Polier G, Neumann J, Thuaud F, et al. The natural anticancer compounds rocaglamides inhibit the Raf-MEK-ERK pathway by targeting prohibitin 1 and 2. Chem Biol. 2012;19(9):1093-1104. Fabbri G, Dalla-Favera R. The molecular pathogenesis of chronic lymphocytic leukaemia. Nat Rev Cancer. 2016;16(3): 145-162.

1593


ARTICLE EUROPEAN HEMATOLOGY ASSOCIATION

Chronic Lymphocytic Leukemia

Ferrata Storti Foundation

Haematologica 2017 Volume 102(9):1594-1604

Extracellular vesicles of bone marrow stromal cells rescue chronic lymphocytic leukemia B cells from apoptosis, enhance their migration and induce gene expression modifications Emerence Crompot,1 Michael Van Damme,1 Karlien Pieters,1 Marjorie Vermeersch,2 David Perez-Morga,2 Philippe Mineur,3 Marie Maerevoet,4 Nathalie Meuleman,4 Dominique Bron,4 Laurence Lagneaux1* and Basile Stamatopoulos1*

Laboratory of Clinical Cell Therapy, Université Libre de Bruxelles (ULB), Jules Bordet Institute; 2Center for Microscopy and Molecular Imaging (CMMI), Université Libre de Bruxelles (ULB), Gosselies; 3Department of Hemato-Oncology, Grand Hôpital de Charleroi, Gilly and 4Hematology Department, Jules Bordet Institute, Brussels, Belgium

1

*

LL and BS contributed equally to this work

ABSTRACT

I Correspondence: emerence.crompot@ulb.ac.be or bstamato@ulb.ac.be Received: December 27, 2016. Accepted: June 5, 2017. Pre-published: June 8, 2017. doi:10.3324/haematol.2016.163337 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1594

nteractions between chronic lymphocytic leukemia (CLL) B cells and the bone marrow (BM) microenvironment play a major function in the physiopathology of CLL. Extracellular vesicles (EVs), which are composed of exosomes and microparticles, play an important role in cell communication. However, little is known about their role in CLL / microenvironment interactions. In the present study, EVs purified by ultracentrifugation from BM mesenchymal stromal cell (BM-MSC) cultures were added to CLL B cells. After their integration into CLL B cells, we observed a decrease of leukemic cell spontaneous apoptosis and an increase in their chemoresistance to several drugs, including fludarabine, ibrutinib, idelalisib and venetoclax after 24 hours. Spontaneous (P=0.0078) and stromal cell-derived factor 1α-induced migration capacities of CLL B cells were also enhanced (P=0.0020). A microarray study highlighted 805 differentially expressed genes between leukemic cells cultured with or without EVs. Of these, genes involved in the B-cell receptor pathway such as CCL3/4, EGR1/2/3, and MYC were increased. Interestingly, this signature presents important overlaps with other microenvironment stimuli such as B-cell receptor stimulation, CLL/nurse-like cells co-culture or those provided by a lymph node microenvironment. Finally, we showed that EVs from MSCs of leukemic patients also rescue leukemic cells from spontaneous or drug-induced apoptosis. However, they induce a higher migration and also a stronger gene modification compared to EVs of healthy MSCs. In conclusion, we show that EVs play a crucial role in CLL B cells/BM microenvironment communication.

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

1594

Introduction Chronic lymphocytic leukemia (CLL) is the most prevalent leukemia in the Western world and is characterized by the progressive accumulation of monoclonal, mature CD5+/CD19+/CD23+ B cells in the blood, bone marrow, lymph nodes and spleen. The interactions between malignant B cells and the microenvironment (ME) play a major role in the physiopathology of CLL.1 These interactions drive CLL B cells into the ME, where they receive advantageous signals from different populations of accessory cells mediated by diverse receptors such as CD40, Tolllike receptors (TLR), C-X-C chemokine receptor type 4 (CXCR4) and the B-cell receptor (BCR). CLL B cells present a defect of apoptosis in vivo; however, they raphaematologica | 2017; 102(9)


Functional effects of EVs on CLL B-cells

idly undergo spontaneous apoptosis when they are cultured in vitro, suggesting the importance of the ME in their survival. Interactions with the ME, such as bone marrow mesenchymal stromal cells (BM-MSCs) or nurse-like cells (NLC), protect CLL B cells from spontaneous and druginduced apoptosis.2 However, the complex cellular and molecular mechanisms underlying this protection are not fully understood. Previously, extracellular vesicles (EVs) were described as cellular debris, but today, EVs are known to play an important role in cell communication. EVs are plasma membrane released from the endosomal compartment (exosomes) or derived directly from the membrane (microparticles) of various cell types. Microparticles (MPs) are submicron vesicles from 100 nm to 1 mm, whereas exosomes have a smaller size from 30 nm to 100 nm.3 To avoid confusion, ‘‘EVs’’ will be used to describe vesicles including exosomes and MPs.4 These vesicles have an impact on several physio-pathological processes including immune responses, tissue regeneration, and blood coagulation. They contain selective patterns of microRNAs, proteins or DNA, which may be transferred into the target cells (reviewed by Abels and Breakefield5). Here, we investigated the role of EVs derived from BMMSCs in different biological processes critical for CLL cell

A

B

survival. In addition, we broadened our understanding of the EV effect by studying the gene expression profile induced by EVs in leukemic cells.

Methods Biological samples and cell culture This study was approved by the Jules Bordet Institute ethics committee (Belgium). Peripheral blood mononuclear cells (PBMC) from CLL samples were obtained from patients after written informed consent. Additional details about sample isolation, CLL cell culture, patients’ characteristics, BM-MSC isolation and characterization are available in the Online Supplementary Table S1 and Online Supplementary Figure S1.

Extracellular vesicle isolation and characterization To obtain EVs, BM-MSCs were cultured with serum deprivation for 24 hours (h) to avoid contaminations by Fetal Bovine Serum (FBS)-derived vesicles. We also compared these conditions with exosome-depleted FBS and we did not observe any difference; serum deprivation over 24 h did not induce senescence and morphology modifications of BM-MSCs. We used 500 mL of BMMSC supernatant to produce the EV pellets. Cell-free supernatants were obtained by 2 successive centrifugations at 300xg for 10 min-

C

Figure 1. Characterization of bone marrow (BM) extracellular vesicles (EVs) and uptake by chronic lymphocytic leukemia (CLL) B cells. Bone marrow mesenchymal stromal cells (BM-MSCs) were characterized by electron microscopy (A) and flow cytometry (B) using the latex bead technique. EVs showed a classical spherical appearance by transmission electron microscopy (TEM). EVs expressed CD63 (EV marker) and CD73 (MSC marker) and were negative for CD45 (hematopoietic marker). (C) CLL B cells were incubated with BM-MSC EVs, previously labeled with PKH67, for 10 minutes (min), 1 hour (h), 3 h and 24 h. Flow cytometry showed a rapid increase of the mean of fluorescence intensity (MFI), depending on the incubation time. The uptake of BM-MSC EVs was a fast process; after 10 min, more than 60% of CLL B cells had integrated fluorescent vesicles. PBS: phosphate-buffered saline.

haematologica | 2017; 102(9)

1595


E. Comprot et al.

utes. The supernatants of BM-MSC cultures were then concentrated with a 3K Macrosep advance centrifugal device (Pall Life Science, New York, USA). The supernatants were then subjected to 150,000xg centrifugation for 1 h at 4°C (Ultracentrifuge MX 120+, Swinging Bucket rotor S50-ST, k-factor 77, Thermo Scientific, Waltham, MA, USA). EV pellets were washed using 0.2mm-filtered phosphate buffered saline (PBS) by centrifugation for 1 h at 150,000xg. The pellets were finally reconstituted in 100 mL of PBS and stored at -80°C until use. The amount and size of BMMSC EVs were determined using Nanosight technology (Nanosight Ltd., Minton Park, UK) and the concentration was determined using the Nanodrop (Thermo Scientific, Nanodrop

A

B

2000c) and BCA kit assay (Thermo Scientific). Details of EV characterization6 and monitoring of BM-MSC EV uptake by CLL B cells are available in the Online Supplementary Appendix.

Apoptosis, cell viability, migration and chemosensitivity assay Chronic lymphocytic leukemia B cells were cultured with or without 2 mg of EVs in 24-well plates (4x106 cells/well) in 10% FBS RPMI-1640. Apoptosis was determined using annexin V/7Aminoactinomycin D (7AAD) staining and viability by 3,3′dihexyloxacarbocyanine iodide (DiOC6)/propidium iodide (PI) staining as previously described.7 CLL B cells cultured without EVs

C

D

E

F

Figure 2. Bone marrow (BM) extracellular vesicles (EVs) protect chronic lymphocytic leukemia (CLL) B cells from spontaneous apoptosis. CLL B cells were incubated with bone marrow mesenchymal stromal cell (BM-MSC) EVs, and after 24 hours (h), cells were stained with annexin V-FITC/7AAD. We observed a decrease in spontaneous apoptosis (A) with a significant effect on early (B) and late (C) apoptosis. (D) A representative flow cytometry analysis of CLL B cells cultured with EVs for 24 h. (E) Expression of the anti-apoptotic protein Mcl-1 (myeloid cell leukemia 1). Red line: CLL B cells without EVs; blue line: CLL B cells cultured with EVs. The expression of secondary antibody was used as a negative control (n=10, P=0.0006). Results were also confirmed by qPCR (n=25, P=0.0001) (F).

1596

haematologica | 2017; 102(9)


Functional effects of EVs on CLL B-cells

were used as a negative control. The migration assay was performed as previously described.7 Additional details can be found in the Online Supplementary Appendix. Cells were incubated with bortezomib, cladribine, fludarabine, flavopiridol or methylprednisolone at the half maximal (50%) inhibitory concentration (IC50) as previously established.7,8 Ibrutinib, idelalisib and venetoclax were also used with specific concentration, as described in several studies.9-11 Additional details are provided in the Online Supplementary Appendix and Online Supplementary Table S2.

Gene expression profile and bioinformatic analysis Gene expression profiles comparing CLL cells cultured with or without 5 mg of EVs were performed as previously described.12 Additional details and bioinformatics analysis can be found in the Online Supplementary Appendix.

EVs effect on CLL B-cell BCR activation Measurement of intracellular calcium flux, mitogen-activated protein kinase 1 (ERK) and v-akt murine thymoma viral oncogene homolog 1 (AKT) phosphorylation were investigated by flow cytometry and are detailed in the Online Supplementary Appendix.

Statistical analyses The Wilcoxon matched pairs test was used to analyze the statistical significance of experimental results. P<0.05 was considered statistically significant. Data were analyzed and graphics were constructed using GraphPad Prism v.5.0 (GraphPad Software, San Diego, CA, USA).

BM-MSC EVs protect CLL B cells from spontaneous apoptosis The addition of EVs resulted in a decrease of spontaneous apoptosis after 24 h of culture (P<0.0001) (Figure 2A). The integration of EVs also induced a decrease of both early and late apoptosis (Figure 2B and C), and this protection against apoptosis was dose-dependent (Online Supplementary Figure S2). Figure 2D shows a representative flow cytometry analysis of the EV effect on CLL B cells after 24 h. We observed a significant decrease of annexin V positive cells. We also studied the expression of Mcl-1, an anti-apoptotic protein (see the Online Supplementary Appendix for details). The addition of EVs increased Mcl-1 expression as measured by flow cytometry (n=10, P=0.0006) (Figure 2E) and qPCR (n=25, P<0.0001) (Figure 2F). We also observed an increase in the expression of B-cell lymphoma extra-large (BCLXL) (n=15, P=0.0067) and X-linked inhibitor apoptosis protein (XIAP) (n=10, P=0.0020), two other anti-apoptotic proteins. However, no difference was observed for B-cell CLL/lymphoma 2 (BCL2) and BCL2-associated X protein (BAX) expression (Online Supplementary Figure S3). Finally, viability was studied by DiOC6-PI labeling; EV addition significantly increased CLL cell viability (n=15, P<0.0001/n=10, P<0.01) (Online Supplementary Figure S4). These different observations demonstrated the capacity of BM-MSC EVs to protect CLL B cells from spontaneous apoptosis, thereby increasing their viability.

BM-MSC EVs increase CLL B-cell migration capacity Results Characterization of BM-MSC EVs and uptake by CLL B cells We first verified the presence of BM-MSC EVs by characterization using transmission electron microscopy, and they presented classical aspects (Figure 1A). EVs were also characterized by latex bead flow cytometry technique, as previously described,6 and we observed weak expression of CD73 (marker commonly found on MSCs) and high expression of CD63, a classical marker of EVs.13 CD45 was used as a negative control (Figure 1B). EV uptake by CLL cells was verified and monitored by PKH67 fluorescent labeling (Figure 1C and the Online Supplementary Appendix).

A

B

When unstimulated, CLL B cells were plated in the upper chamber of a Transwell, low spontaneous migration capacity was observed. The pre-incubation of CLL cells with EVs for 4 h significantly increased the level of migrating cells (P=0.0078) (Figure 3A). When CLL B cells were pre-incubated with BM-MSC EVs for 4 h, a significant increase of migration was observed in the presence of the SDF1α (n=10, P=0.0020) (Figure 3B). These results demonstrate that EVs from BM-MSCs can enhance the spontaneous and SDF-1-induced migration of CLL B cells. Interestingly, we observed no effect of CXCR4 receptor antagonist AMD3100 (P=0.9143) and the addition of pertussis toxin (P=0.0634) did not significantly decrease the number of migrating cells (n=25) (Figure 3C).

C

Figure 3. Bone marrow (BM) extracellular vesicles (EVs) increase chronic lymphocytic leukemia (CLL) B-cell migration capacity. CLL B cells were incubated with bone marrow mesenchymal stromal cell (BM-MSC) EVs in a Transwell assay. EV addition increased the spontaneous migration (A). The migration index was also calculated as the number of cells transmigrating in the presence of SDF-1α divided by the number of cells transmigrating in the absence of SDF1α. A significant increase in migration was observed in the presence of the chemoattractant SDF-1α (B) (n=10, P=0.0020). (C) Migration of CLL B cells in the presence of AMD3100 and pertussis toxin (P. tox). We did not observe a decrease in the migrating cells in the presence of AMD3100 or pertussis toxin (P. tox).

haematologica | 2017; 102(9)

1597


E. Comprot et al.

We determined whether EVs are implicated in the drug resistance of CLL B cells. Eight different drugs were tested (Figure 4). The cytotoxicity of these drugs was studied after 24 h of incubation on CLL B cells cultured alone (as a negative control) or pre-treated with EVs for 4 h. CLL B cells presented significant resistance to these 8 drugs in the presence of BM-MSC EVs; they significantly decreased the number of apoptotic CLL B cells after 24 h (Figure 4). Interestingly, for cladribine and bortezomib, EVs completely suppressed drug-induced apoptosis.

Impact of BM-MSC EVs on CLL B-cell gene expression profile The transcriptomic profiles of CLL B cells from 3 different patients were determined to obtain a global view of gene expression differences in leukemic cells cultured with or without BM-MSC EVs. A P value less than 0.05 allowed us to identify 805 genes differentially expressed between leukemic cells cultured with or without EVs. Among them, 152 (19%) were up-regulated and 653 (81%) were down-regulated in CLL B cells cultured with BM-MSC EVs (Figure 5A). Gene set enrichment analysis investigating gene GO categories demonstrated that up-regulated

genes in CLL B cells after EV treatment were highly represented in the categories of cell-cell signaling (GO:0007267), actin cytoskeleton organization (GO:0015629, GO:0007010, GO:0030036), receptor binding (GO:0005102), and positive regulation of transcription (GO:0001228); all significant categories with the LS-permutation P-value selection are listed in the Online Supplementary Appendix Table S3. To validate the robustness of our microarray analysis, a panel of 7 genes implicated in CLL physiopathology was selected for qPCR validation on an extended cohort (n=18) (Figure 5B). All tested genes were confirmed as differentially expressed in CLL B cells after BM-MSC EV addition, highlighting the robustness of our microarray data.

Comparison of microarray signatures with other studies We compared our microarray results with 2 previously published studies of CLL B cells co-cultured with NLC culture14 or stimulated by anti-IgM stimulation.15 The intersection of differentially expressed genes after these different ME stimulations showed important overlaps. In total, 177 (22% of our study) and 226 (28%) genes were shared between our study and that of Burger et al. and Guarini et

A

E

B

F

C

G

D

H

Drug-induced apoptosis

BM-MSC EVs increase chemoresistance of CLL B cells

Figure 4. Bone marrow (BM) extracellular vesicles (EVs) increase chemoresistance of chronic lymphocytic leukemia (CLL) B cells. CLL B cells were incubated with bone marrow mesenchymal stromal cell (BM-MSC) EVs for 4 hours (h) and then treated with drugs for 24 h: bortezomib (A), cladribine (B), fludarabine (C), flavopiridol (D), methylprednisolone (E), ibrutinib (F), idelalisib (G) and venetoclax (H). Cells were then stained with annexin V-FITC/7AAD. The addition of BM-MSC EVs in CLL B-cell culture could protect them from drug-induced apoptosis; we observed a decrease in annexin-positive cells in the presence of each of the 8 drugs. The results were normalized by the subtraction of spontaneous apoptosis (cells without drug and EVs).

1598

haematologica | 2017; 102(9)


Functional effects of EVs on CLL B-cells

al., respectively. A total of 69 genes (8.5%) were common in the 3 studies. Using a “super exact test”,16 we obtained a P value of 7.04x10-65 indicating that the 69 genes in common between the 3 studies were not due to hazard (Figure 5C). Among them, CCL4/3, early growth response (EGR) family, TLR10, IL21R, and HDAC9 were all up-regulated after the different ME stimuli. The complete list of common genes is provided in Online Supplementary Table S4. The percentages of similarities between studies are shown in Online Supplementary Table S5, and summarized in Figure 5C. We observed 20% and 50% of common genes in the 20 most decreased genes in CLL B cells after BCR stimulation and NLC co-culture, respectively. A total of 70% of genes were common within the 20 most increased genes from both signatures (Online Supplementary Tables S6 and S7). We conclude that EVs alone can induce a significant part of the gene expression modifications induced by NLC culture or BCR stimulation. We finally compared

our microarray results with the study of Herishanu et al.17 who analyzed the differential gene expression of CLL cells obtained from different compartments: peripheral blood (PB), bone marrow (BM) and lymph node (LN). We observed significant overlap between an EV treatment and a stimulation provided by an LN microenvironment (Figure 5D). In addition, 34 genes were in common between our study and CLL cells receiving stimuli from BM and LN (P=8.89x10-15) (Online Supplementary Tables S6S8). Of these, we again noticed genes linked to BCR pathway. Interestingly, EV gene signature shared 3 times more genes with LN than BM signature.

EV effects on CLL B-cell BCR activation Regarding the significant overlap between IgM stimulation and EV treatment, we investigated the potential activation of the BCR pathway by studying Ca2+ flux using FLUO8, the phosphorylation of ERK and AKT by flow

A

B

C

D

Figure 5. Impact of bone marrow (BM) extracellular vesicles (EVs) on chronic lymphocytic leukemia (CLL) B-cell gene expression profile and comparison with other microarray studies. The genetic profiles of CLL B cells from 3 different patients were determined to obtain a global view of gene expression differences in leukemic cells with or without bone marrow mesenchymal stromal cell (BM-MSC) EV treatment. (A) A total of 805 genes were differentially expressed between leukemic cells cultured with or without EVs (P<0.05). (B) We selected 7 differentially expressed genes in our microarray analysis and confirmed their expression by real-time PCR. All genes were validated, confirming our microarray results. (C) We compared our microarray results with 2 other published studies on CLL B cells activated by NLC culture14 or IgM stimulation.15 We obtained a substantial overlap between the differentially expressed genes; 177 and 226 genes were shared between our study and those of Burger et al.14 and Guarini et al.,15 respectively, and 69 genes were common among all 3 studies. (D) Our microarray results were also compared with the study of Herishanu et al.17 who analyzed the differential gene expression of CLL cells obtained from different compartments: peripheral blood (PB), bone marrow (BM) and lymph node (LN). A significant overlap was observed between an EV treatment and a stimulation provided by an LN microenvironment. In addition, 34 genes were in common between our study and CLL cells receiving stimuli from BM and LN (P=8.89x10-15). The “SuperExactTest’’ was applied to evaluate the statistical value of the intersections between 3 studies as indicated by the expected overlap and the P values. BCR: B-cell receptor.

haematologica | 2017; 102(9)

1599


E. Comprot et al.

cytometry. In order to work with â&#x20AC;&#x2DC;respondingâ&#x20AC;&#x2122; cells, we performed these experiments with the Raji cell line known to be an excellent responder to IgM stimulation. Despite a BCR-like signature, we did not observe calcium flux in Raji cells (Figure 6A). While an IgM stimulation increases the ERK and AKT phosphorylation, EV treatment did not, neither in Raji cells (Figure 6B and C), nor in primary CLL cells (Figure 6D and E). We also studied other targets known to be regulated after BCR activation: CCL4, lipoprotein lipase (LPL) and MYC expression by qPCR and CXCR4 (CD184), and CD69 expression, by flow cytometry. After 24 h of an EV treatment, we observed in all cases a modulation of these targets in the same way as a BCR activation (increase in CD69, LPL, MYC, CCL4 and a decrease in CXCR4) (Figure 6F-H). Interestingly, while we did not observe a calcium flux and ERK/AKT phosphorylation, inhibition of BCR pathway using ibrutinib or ide-

A

G

B

H

C

I

D

E

J

F

1600

lalisib abolished EV-induced migration (Figure 6I). In addition, ibrutinib treatment also abolished gene modifications induced by EVs (Figure 6 J).

EV effects are due to de novo cell regulation rather than RNA transfer In order to investigate if the EV increases of gene expression are due to RNA transfer or de novo transcription, we treated CLL cells with actinomycin D (5 mg/mL) prior to EV incubation. After 4 h of EV integration, we did not observe any significant change in gene expression of some representative targets [CCL4 (Figure 6 H), FCRL5 and TLR10 (data not shown)] indicating that the increase of these genes was not due to an RNA transfer. After 24 h, we confirmed the increase in these genes but this increase was abolished by an actinomycin D treatment. In addi-

Figure 6. Extracellular vesicles (EV) effects on chronic lymphocytic leukemia (CLL) B-cell receptor (BCR) activation. (A) Calcium flux in Raji cell line (n=3) after anti-IgM or EV treatment measured using FLUO8 by flow cytometry. The mean of fluorescence intensity (MFI) was normalized by the MFI of untreated cells and plotted every ten seconds. The phosphorylation of ERK (B and D) and AKT (C and E) was evaluated by flow cytometry. Data were presented as the mean fluorescence intensity (MFI) ratio of IgM stimulation/unstimulated cells and EVs treatment/untreated cells. Downstream targets of BCR activation in CLL B cells after EV integration were analyzed by real-time PCR (MYC, LPL, CCL4) (F-H) and flow cytometry (CD69, CXCR4 or CD184) (G). (H) Four targets of BCR activation (CCL4, miR-229c, miR-150, miR21) after 4 hours (h) or 24 h of EV integration treated or not with actinomycin D (5 mg/mL). Genes were normalized using cyclophilin A (PPIA) as endogenous control while microRNAs were normalized using RNU48. (I) CLL B cells were incubated with BM-MSC EVs in a Transwell assay. Cells were treated with or without ibrutinib (ibrut.: 10 mM) or idelalisib (idelal.: 10 mM), with or without EVs. The migration ratio represents the number of migrating cells in each condition divided by the spontaneous migrating cells (in absence of EVs or drugs). (J) Cells were treated with or without ibrutinib (10 mM) and with or without EVs. Gene expression for a selection of genes was evaluated by real-time PCR. Data were normalized with the expression of untreated cells (without EVs and without ibrutinib).

haematologica | 2017; 102(9)


Functional effects of EVs on CLL B-cells

tion, we tested, in the same conditions, 3 microRNAs previously investigated in CLL (miR-29c, miR-150, miR-16)1820 and known to be decreased after BCR stimulation.21,22 Again, we did not observe expression modification at 4 h, but after 24 h, we observed that EVs are able to induce an miR-29c, miR-150 and miR-16 decrease, as is the case after BCR stimulation (Figure 6H). This decrease was abolished by actinomycin D, demonstrating that this decrease was due to de novo regulation.

Quantitative and qualitative comparison between healthy and CLL-derived EVs In this study, we used EVs from MSC culture established from healthy donors. In order to complete this

work, we performed the same experiments with EVs produced by BM-MSCs obtained from CLL patients. Nanoparticle tracking analysis (NTA) was used to evaluate size distribution and the concentration of EVs. Despite the difficulty to maintain them in ‘long-term culture’, CLL BM-MSCs seem to be a ‘higher producer’ of EVs. After concentration of the collected medium, we indeed obtained a mean of 955±172 EVs/MSC/day (n=17) while this number reached 1634±387/MSC/day for CLL MSC (n=5) (P=0.0417) without any significant change in their size (Figure 7A-D). Furthermore, CLL BM-MSC EVs induced similar effects on apoptosis (Figure 7E) and viability (Figure 7F) of CLL B cells. In addition, we also compared the protective effect of both EV types after ibrutinib,

A

B

C

D

E

F

G

H

I

J

K

Figure 7. Quantitative and qualitative comparison between healthy and chronic lymphocytic leukemia (CLL)-derived extracellular vesicles (EVs). (A) Measure of mean size of EVs from healthy and CLL MSC. (B and C) A representative size distribution of healthy and CLL mesenchymal stromal cell (MSCs), respectively. CLL B cells were incubated with healthy or CLL-MSC EVs for 24 hours (h) and cells were stained with annexin V-FITC/7AAD to measure apoptosis (E) or DioC6/propidium iodide to measure cell viability (F). We observed a decrease of spontaneous apoptosis. CLL B cells were incubated with healthy or CLL-MSC EVs for 4 h and then treated for 24 h with ibrutinib (G), idelalisib (H) or venetoclax (I). Cells were then stained with annexin V-FITC/7AAD. The results were normalized by the subtraction of spontaneous apoptosis (cells without drug and EVs). (J) CLL B cells were incubated with healthy or CLL MSC EVs in a Transwell assay. The migration ratio represents the number of migrating cells in each condition divided by the spontaneous migrating cells (in absence of EVs). (K) Cells were incubated with healthy or CLL MSC EVs for 24 h. Gene expression for a selection of genes was evaluated by real-time PCR. Data were normalized with the expression of untreated cells (without EVs).

haematologica | 2017; 102(9)

1601


E. Comprot et al.

idelalisib and venetoclax treatment, and we did not observe any difference (Figure 7G-I). However, spontaneous migration of CLL B cells after addition of BM-MSC EVs from CLL patients is higher compared to healthy EVs (P=0.0049) (Figure 7J). Interestingly, when we investigated EV effect on gene expression, both EV types are able to induce gene modifications described above, but CLL MSC-derived EVs display a stronger action as is shown for MYC, CCL4, CD69 increase or HRK, LY9 decrease (Figure 7K).

Discussion In hematologic malignancies, the microenvironment interacts with tumor cells principally through cell-cell contact and soluble factor production to support leukemic development.23 Recently, EVs were proposed as a new mechanism of cross-talk and cellular communication.24 PB, BM, and LN-derived B cells present different RNA signatures, confirming that the ME gives signals activating specific pathways.17 The production of soluble factors by MSCs can attract CLL B cells in the ME and protect them from spontaneous apoptosis,2 but it is now widely accepted that MSCs can also communicate by EVs.25 MSCs show bidirectional cross-talk with normal B cells,26 neoplastic B cells27 or other malignant cells, such as multiple myeloma and chronic myelogenous leukemia cells.28,29 Roccaro et al. demonstrated that BM-MSC EVs contribute to disease progression in multiple myeloma.28 Because the exact role of BM-MSC EVs remains unknown in CLL, we investigated modifications induced by EVs in CLL B cells using microarray analyses and determined their impact on CLL B-cell survival, migration and chemoresistance. We studied EVs (microparticles and exosomes together) because this is more similar to physiological conditions. To maintain EVs as close as possible to their native state, we did not use any activator to increase EV production, and serum deprivation was applied on BM-MSC cultures to avoid any fetal bovine serum vesicle contamination. Numerous authors used a concentration of EVs between 30 and 50 mg/mL.29-31 In the present study, we used 10 times lower concentrations (between 2 and 5 mg/mL) to be closer to the physiological condition, and observed significant effects. Furthermore, the addition of conditioned medium versus EV-depleted conditioned medium from BM-MSC culture already induces a protective effect, illustrating the implication of EVs in cell functions in more physiological conditions (Online Supplementary Figure S5). We showed that CLL B cells can rapidly integrate EVs from BM-MSCs. This active uptake of the BM-MSC EVs by target cells was shown in other studies, indicating the in vivo relevance of EV transfer.32 BM-MSCs increase the migration capacity,33 decrease apoptosis,2 and increase chemoresistance7 of CLL B cells after direct contact. Here, we report for the first time that EVs alone can induce similar effects as their cell counterparts in CLL. Indeed, EVs protect CLL cells from spontaneous apoptosis similar to a stromal layer.2 In addition, it is now well known that CLL B cells can escape from chemotherapy by migrating into stromal niches.34 Here, we observed that EVs increase the migratory capacity of CLL cells and, subsequently, could play a role in in vivo survival. Interestingly, the effect of EVs on CLL-cell migration is independent of CXCR4 1602

receptor or other G protein-coupled receptors. Currently, CLL remains incurable with classical treatments, and relapse remains the major cause of death. In this study, we demonstrated that the addition of BM-MSC EVs to CLL Bcell cultures increases their resistance against 8 different drugs, including purine nucleoside analogs, corticosteroids, ibrutinib, idelalisib or venetoclax. A study showed the same effect against bortezomib in multiple myeloma cells after the integration of exosomes from BM-MSCs.35 These important data highlight the crucial role of EVs in the context of chemoresistance. Our microarray study revealed 805 genes differentially expressed between CLL B cells treated with or without EVs. Lymphocyte antigen 9 (LY9), which presents decreased expression (-2.2-fold) after EVs integration, is a tumor antigen targeted by T cells in CLL B cells.36 Moreover, phosphodiesterase 8A (PDE8A), which presents decreased expression in CLL B cells treated with EVs, is a predictor of treatment-free survival (TFS) and overall survival (OS).12 By contrast, some mRNAs were increased. We observed the doubled expression of CD83 (+1.9-fold), which is associated with a shorter TFS.37 Furthermore, HDAC9, which was significantly increased in our microarray data (+1.5-fold), is correlated with ZAP70-positive CLL patients.38 Based on these different observations, we suggest that EVs from the microenvironment can promote the development of a poor prognosis profile and influence the clinical outcome of CLL patients. Interestingly, we also observed that BM-MSC EVs increase the CD69 mRNA expression (+1.7-fold). This result was also confirmed by flow cytometry, as previously reported.17 This antigen on activated B cells is well described in the CLL literature. In 2001, Dâ&#x20AC;&#x2122;Arena et al. hypothesized that CD69 is a prognostic factor in CLL disease due to more genetic abnormalities in CD69 positive patients, and thereafter, Del Poeta showed its independent prognostic power.39 Importantly, other authors had previously investigated the impact of other stimuli from the microenvironment on CLL B-cell gene expression. Burger et al. studied CLL B cells activated by NLC,14 whereas Guarini and colleagues analyzed the effect of BCR stimulation using anti-IgM.15 In total, 226 (28% of the genes of our study) and 177 (22%) differentially expressed genes in our study were described as differentially expressed in the studies by Burger et al. and Guarini et al., respectively. When we compared our microarray results with the gene modifications induced in CLL cells obtained from different compartments (BM and LN),17 we observed that these different stimuli could induce similar gene expression modifications compared to EV treatment. These results indicate that a non-negligible part of the ME effects are also mediated through EVs. Among the common increased genes, we identified CCL4 (+4.6-fold), CCL3 (+3.8-fold), EGR family members (+1.8-fold/+3.8fold), FCRL 5 (+2.2-fold), and MYC (+2.1-fold), which are genes that are increased after BCR activation.40,41 We hypothesize that BM-MSC EVs induce BCR-like activation, leading to cell survival and drug resistance. Complementary to our study, other authors showed that BCR activation induces CLL EV overproduction and leads to a worse prognosis.42 A possible activation loop mediated by EVs could potentially lead to BCR activation, survival factor production and association with a poor prognosis in CLL. Furthermore, Yeh et al. reported that the number of exosomes is not directly correlated with the haematologica | 2017; 102(9)


Functional effects of EVs on CLL B-cells

number of B cells, but with BCR activation, contrary to the findings of Ghosh et al.,42,43 another finding that supports our hypothesis. To further investigate the hypothesis of BCR activation, we studied CCL4, CD69, CXCR4 (CD184), lipoprotein lipase (LPL) and MYC expression or protein level: EVs increase the expression of CCL4, CD69, LPL and MYC, and decrease the level of CXCR4 similar to BCR activation, as previously described.14,22 Similarly, miR29c, miR-150 or miR-16 are decreased after EV treatment as is also the case after a BCR activation.21 To evaluate the potential BCR activation, we measured intracellular calcium flux, AKT and ERK phosphorylation by flow cytometry, but we did not observe phosphorylation of these targets or calcium flux with EV treatment after 24 h (nor in a shorter period of time). Moreover, several authors observed a BCR signature but none of them demonstrated a direct BCR activation using calcium flux or ERK phosphorylation.14,17 Surprisingly, inhibition of BCR pathway is able to decrease/abolish the gene modifications induced by EVs as well as the EV-related migration, indicating that BCR pathway is linked to EVs. Based on these observations, we could hypothesize that EVs are not able to activate BCR as is classically shown in the literature (no calcium flux, no AKT or ERK phosphorylation). However, we observed a BCR-like gene signature, microRNA BCRrelated modifications, inhibition of gene signature modifications, and migration by BCR pathway inhibitor suggesting that the BCR pathway could be activated by EVs in an intracellular manner. Finally, we also observed that BM-MSCs of CLL patients produce a higher amount of EVs, suggesting that EVs could even be a more important actor in leukemic cells/microenvironment interaction in a pathological context. Similarly to BM-MSC EVs from healthy donors, we

References 7. 1. Ten Hacken E, Burger JA. Microenvironment dependency in Chronic Lymphocytic Leukemia: The basis for new targeted therapies. Pharmacol Ther. 2014;144(3):338-348. 2. Lagneaux L, Delforge A, Bron D, De Bruyn C, Stryckmans P. Chronic lymphocytic leukemic B cells but not normal B cells are rescued from apoptosis by contact with normal bone marrow stromal cells. Blood. 1998;91(7):2387-2396. 3. Raposo G, Stoorvogel W. Extracellular vesicles: exosomes, microvesicles, and friends. J Cell Biol. 2013;200(4):373-383. 4. Lotvall J, Hill AF, Hochberg F, et al. Minimal experimental requirements for definition of extracellular vesicles and their functions: a position statement from the International Society for Extracellular Vesicles. J Extracell Vesicles. 2014;3:26913. 5. Abels ER, Breakefield XO. Introduction to Extracellular Vesicles: Biogenesis, RNA Cargo Selection, Content, Release, and Uptake. Cell Mol Neurobiol. 2016; 36(3):301-312. 6. Crompot E, Van Damme M, Duvillier H, et al. Avoiding false positive antigen detection by flow cytometry on blood cell derived microparticles: the importance of an appro-

haematologica | 2017; 102(9)

8.

9.

10.

11.

12.

demonstrated that BM-MSC EVs of CLL patients also rescue CLL B cells from spontaneous apoptosis. However, CLL MSC-derived EVs induce a higher migration compared to healthy MSC-derived EVs, and also a stronger gene modification. All these data taken together indicated that EVs from CLL MSC could even play a more important role in a CLL in vivo context. In conclusion, we report for the first time that EVs from ME can increase viability, rescue CLL B cells from spontaneous and drug-induced apoptosis, and enhance their migratory capacities in a contact-independent manner. In addition, EVs induced gene expression profile modifications of CLL cells that substantially overlap with transcriptomic signatures obtained after BCR stimulation, NLC coculture or CLL in a lymph node ME. In this work, we focused on EVs from bone marrow ME, specifically from MSC, but vesicles from other cells (T cells, NK cells, NLC, etc.) may also interfere with the communication between CLL B cells and the ME. Collectively, we provide evidence that EVs alone can modulate several key cellular processes, suggesting that interference with EV production/internalization may be a new target for therapeutic strategies. Funding This work was supported by a Télévie grant, “projet de recherche” (PDR) both provided by the F.R.S.-FNRS (Fonds National de la Recherche Scientifique) of Belgium, the ‘’David and Alice Van Buuren Fund’’, the ‘’Fonds IRIS-Recherche’’, the “Les Amis de l’Institut Jules Bordet”, FER (Fonds d’Encouragement à la Recherche) of the Faculty of Medicine, Université Libre de Bruxelles and the foundation LambeauMarteaux. The CMMI is supported by the European Regional Development Fund and Wallonia. We finally thank Dr Nicolas Rosewick for his precious help with statistical analysis.

priate negative control. PLoS One. 2015; 10(5):e0127209. Stamatopoulos B, Meuleman N, De Bruyn C, et al. AMD3100 disrupts the cross-talk between chronic lymphocytic leukemia cells and a mesenchymal stromal or nurselike cell-based microenvironment: pre-clinical evidence for its association with chronic lymphocytic leukemia treatments. Haematologica. 2012;97(4):608-615. Stamatopoulos B, Meuleman N, De Bruyn C, et al. Antileukemic activity of valproic acid in chronic lymphocytic leukemia B cells defined by microarray analysis. Leukemia. 2009;23(12):2281-2289. Amin NA, Balasubramanian S, Saiya-Cork K, et al. Cell-Intrinsic Determinants of Ibrutinib-Induced Apoptosis in Chronic Lymphocytic Leukemia. Clin Cancer Res. 2017;23(4):1049-1059. Hoellenriegel J, Meadows SA, Sivina M, et al. The phosphoinositide 3'-kinase delta inhibitor, CAL-101, inhibits B-cell receptor signaling and chemokine networks in chronic lymphocytic leukemia. Blood. 2011;118(13):3603-3612. Oppermann S, Ylanko J, Shi Y, et al. Highcontent screening identifies kinase inhibitors that overcome venetoclax resistance in activated CLL cells. Blood. 2016; 128(7):934-947. Stamatopoulos B, Haibe-Kains B, Equeter

13.

14.

15.

16. 17.

C, et al. Gene expression profiling reveals differences in microenvironment interaction between patients with chronic lymphocytic leukemia expressing high versus low ZAP70 mRNA. Haematologica. 2009; 94(6):790-799. Kowal J, Arras G, Colombo M, et al. Proteomic comparison defines novel markers to characterize heterogeneous populations of extracellular vesicle subtypes. Proc Natl Acad Sci USA. 2016;113(8):E968-E977. Burger JA, Quiroga MP, Hartmann E, et al. High-level expression of the T-cell chemokines CCL3 and CCL4 by chronic lymphocytic leukemia B cells in nurselike cell cocultures and after BCR stimulation. Blood. 2009;113(13):3050-3058. Guarini A, Chiaretti S, Tavolaro S, et al. BCR ligation induced by IgM stimulation results in gene expression and functional changes only in IgV H unmutated chronic lymphocytic leukemia (CLL) cells. Blood. 2008;112(3):782-792. Wang M, Zhao Y, Zhang B. Efficient test and visualization of multi-set intersections. Sci Rep. 2015; 5:16923. Herishanu Y, Perez-Galan P, Liu D, et al. The lymph node microenvironment promotes B-cell receptor signaling, NF-kappaB activation, and tumor proliferation in chronic lymphocytic leukemia. Blood. 2011;117(2):563-574.

1603


E. Comprot et al. 18. Stamatopoulos B, Meuleman N, HaibeKains B, et al. microRNA-29c and microRNA-223 down-regulation has in vivo significance in chronic lymphocytic leukemia and improves disease risk stratification. Blood. 2009;113(21):5237-5245. 19. Stamatopoulos B, Van Damme M, Crompot E, et al. Opposite Prognostic Significance of Cellular and Serum Circulating MicroRNA-150 in Patients with Chronic Lymphocytic Leukemia. Mol Med. 2015;21:123-133. 20. Cimmino A, Calin GA, Fabbri M, et al. miR-15 and miR-16 induce apoptosis by targeting BCL2. Proc Natl Acad Sci USA. 2005;102(39):13944-13949. 21. Mraz M, Kipps TJ. MicroRNAs and B cell receptor signaling in chronic lymphocytic leukemia. Leuk Lymphoma. 2013; 54(8):1836-1839. 22. Pede V, Rombout A, Vermeire J, et al. CLL cells respond to B-Cell receptor stimulation with a microRNA/mRNA signature associated with MYC activation and cell cycle progression. PLoS One. 2013;8(4):e60275. 23. Hanahan D, Coussens LM. Accessories to the crime: functions of cells recruited to the tumor microenvironment. Cancer Cell. 2012;21(3):309-322. 24. Camussi G, Deregibus MC, Bruno S, Cantaluppi V, Biancone L. Exosomes/ microvesicles as a mechanism of cell-to-cell communication. Kidney Int. 2010; 78(9):838-848. 25. Lopatina T, Gai C, Deregibus MC, Kholia S, Camussi G. Cross Talk between Cancer and Mesenchymal Stem Cells through Extracellular Vesicles Carrying Nucleic Acids. Front Oncol. 2016;6:125. 26. Budoni M, Fierabracci A, Luciano R, et al. The immunosuppressive effect of mesenchymal stromal cells on B lymphocytes is mediated by membrane vesicles. Cell Transplant. 2013;22(2):369-379. 27. Ding W, Nowakowski GS, Knox TR, et al.

1604

28.

29.

30.

31.

32.

33.

34.

35.

Bi-directional activation between mesenchymal stem cells and CLL B-cells: implication for CLL disease progression. Br J Haematol. 2009;147(4):471-483. Roccaro AM, Sacco A, Maiso P, et al. BM mesenchymal stromal cell-derived exosomes facilitate multiple myeloma progression. J Clin Invest 2013;123(4):1542-1555. Corrado C, Raimondo S, Saieva L, et al. Exosome-mediated crosstalk between chronic myelogenous leukemia cells and human bone marrow stromal cells triggers an interleukin 8-dependent survival of leukemia cells. Cancer Lett. 2014;348(1-2):71-76. Paggetti J, Haderk F, Seiffert M, et al. Exosomes released by chronic lymphocytic leukemia cells induce the transition of stromal cells into cancer-associated fibroblasts. Blood. 2015;126(9):1106-1117. Greening DW, Nguyen HPT, Elgass K, Simpson RJ, Salamonsen LA. Human Endometrial Exosomes Contain HormoneSpecific Cargo Modulating Trophoblast Adhesive Capacity: Insights into Endometrial-Embryo Interactions. Biol Reprod. 2016;94(2):38. Skog J, Wurdinger T, van Rijn S, et al. Glioblastoma microvesicles transport RNA and proteins that promote tumour growth and provide diagnostic biomarkers. Nat Cell Biol. 2008;10(12):1470-1476. Trimarco V, Ave E, Facco M, et al. Crosstalk between chronic lymphocytic leukemia (CLL) tumor B cells and mesenchymal stromal cells (MSCs): implications for neoplastic cell survival. Oncotarget. 2015;6(39):42130-42149. Burger JA, Burger M, Kipps TJ. Chronic lymphocytic leukemia B cells express functional CXCR4 chemokine receptors that mediate spontaneous migration beneath bone marrow stromal cells. Blood. 1999; 94(11):3658-3667. Wang J, De Veirman K, De Beule N. et al. The bone marrow microenvironment

36.

37.

38.

39.

40.

41.

42.

43.

enhances multiple myeloma progression by exosome-mediated activation of myeloidderived suppressor cells. Oncotarget. 2015; 6(41):43992-44004. Bund D, Mayr C, Kofler DM, Hallek M, Wendtner CM. Human Ly9 (CD229) as novel tumor-associated antigen (TAA) in chronic lymphocytic leukemia (B-CLL) recognized by autologous CD8+ T cells. Exp Hematol. 2006;34(7):860-869. Hock BD, Fernyhough LJ, Gough SM, et al. Release and clinical significance of soluble CD83 in chronic lymphocytic leukemia. Leuk Res. 2009;33(8):1089-1095. Wang JC, Kafeel MI, Avezbakiyev B, et al. betHistone deacetylase in chronic lymphocytic leukemia. Oncology. 2011;81(56):325-329. Del Poeta G, Del Principe MI, Zucchetto A, et al. CD69 is independently prognostic in chronic lymphocytic leukemia: a comprehensive clinical and biological profiling study. Haematologica. 2012;97(2):279-287. Quiroga MP, Balakrishnan K, Kurtova AV, et al. B-cell antigen receptor signaling enhances chronic lymphocytic leukemia cell migration and survival: specific targeting with a novel spleen tyrosine kinase inhibitor, R406. Blood. 2009;114(5):10291037. Krysov S, Dias S, Paterson A, et al. Surface IgM stimulation induces MEK1/2-dependent MYC expression in chronic lymphocytic leukemia cells. Blood. 2012;119(1):170-179. Yeh YY, Ozer HG, Lehman AM, et al. Characterization of CLL exosomes reveals a distinct microRNA signature and enhanced secretion by activation of BCR signaling. Blood. 2015;125(21):3297-3305. Ghosh AK, Secreto CR, Knox TR, et al. Circulating microvesicles in B-cell chronic lymphocytic leukemia can stimulate marrow stromal cells: implications for disease progression. Blood. 2010;115(9):1755-1764.

haematologica | 2017; 102(9)


ARTICLE

Non-Hodgkin Lymphoma

Anaplastic lymphoma kinase-positive anaplastic large cell lymphoma with the variant RNF213-, ATIC- and TPM3-ALK fusions is characterized by copy number gain of the rearranged ALK gene Jo-Anne van der Krogt,1,* Marlies Vanden Bempt,1,2,* Julio Finalet Ferreiro,1 Nicole Mentens,1,2 Kris Jacobs,1,2 Ursula Pluys,1 Kathleen Doms,1 Ellen Geerdens,1,2 Anne Uyttebroeck,3 Pascal Pierre,4 Lucienne Michaux,1 Timothy Devos,5 Peter Vandenberghe,1,5 Thomas Tousseyn,6,7 Jan Cools1,2 and Iwona Wlodarska1 *JAvdK and MVB contributed equally to this work

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(9):1605-1616

Center for Human Genetics, KU Leuven; 2Center for Cancer Biology, VIB, Leuven; Department of Pediatrics, University Hospitals Leuven; 4Department of Hematology, Cliniques Sud Luxembourg, Arlon; 5Department of Hematology, University Hospitals Leuven; 6Translational Cell and Tissue Research KU Leuven and 7Department of Pathology, University Hospitals Leuven, Belgium 1 3

ABSTRACT

A

naplastic lymphoma kinase (ALK)-positive anaplastic large cell lymphoma is characterized by 2p23/ALK aberrations, including the classic t(2;5)(p23;q35)/NPM1-ALK rearrangement present in ~80% of cases and several variant t(2p23/ALK) occurring in the remaining cases. The ALK fusion partners play a key role in the constitutive activation of the chimeric protein and its subcellular localization. Using various molecular technologies, we have characterized ALK fusions in eight recently diagnosed anaplastic large cell lymphoma cases with cytoplasmic-only ALK expression. The identified partner genes included EEF1G (one case), RNF213/ALO17 (one case), ATIC (four cases) and TPM3 (two cases). Notably, all cases showed copy number gain of the rearranged ALK gene, which is never observed in NPM1-ALK-positive lymphomas. We hypothesized that this could be due to lower expression levels and/or lower oncogenic potential of the variant ALK fusions. Indeed, all partner genes, except EEF1G, showed lower expression in normal and malignant T cells, in comparison with NPM1. In addition, we investigated the transformation potential of endogenous Npm1-Alk and Atic-Alk fusions generated by clustered regularly interspaced short palindromic repeats/Cas9 genome editing in Ba/F3 cells. We found that Npm1-Alk has a stronger transformation potential than Atic-Alk, and we observed a subclonal gain of Atic-Alk after a longer culture period, which was not observed for Npm1-Alk. Taken together, our data illustrate that lymphomas driven by the variant ATIC-ALK fusion (and likely by RNF213ALK and TPM3-ALK), but not the classic NPM1-ALK, require an increased dosage of the ALK hybrid gene to compensate for the relatively low and insufficient expression and signaling properties of the chimeric gene. Introduction Anaplastic large cell lymphoma (ALCL) expressing ALK (ALK+ ALCL) is a rare but well-defined subtype of peripheral T-cell lymphoma (PTCL).1 It accounts for approximately 3% of all non-Hodgkin lymphomas (NHL) in adults, 10-15% of pediatric lymphomas and 60-80% of all ALCLs. ALK+ ALCL is hallmarked by various 2p23/ALK chromosomal rearrangements leading to an aberrant expression and constitutive activation of the ALK tyrosine kinase. The most prevalent lesion occurring in more than 80% of ALK+ ALCL is t(2;5)(p23;q35) involving ALK and NPM1 haematologica | 2017; 102(9)

Correspondence: iwona.wlodarska@uzleuven.be

Received: March 24, 2016. Accepted: June 26, 2017. Pre-published: June 28, 2017. doi:10.3324/haematol.2016.146571 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1605 Š2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1605


J-A. van der Krogt et al. (nucleophosmin) genes, respectively.2 The translocation generates a chimeric protein comprising the N-terminal oligomerization domain of NPM1 and the C-terminal region of ALK, including its intracellular tyrosine kinase domain.2 The fusion acts as an oncogene and its transforming potential was proven in a number of in vitro and in vivo studies.3-5 The remaining ALK+ ALCL cases harbor variant 2p23/ALK rearrangements affecting at least nine partner genes: TPM3 (1q21.3; previously assigned to 1q25),6 ATIC (2q35), TFG (3q12.2), TRAF1 (9q33.2), CLTC (17q23.1), ALO17/RNF213 (17q25.3), TPM4 (19p13.12), MYH9 (22q12.3) and MSN (Xq12).7 These partners play a key role in the constitutive activation of the chimeric protein by mediating its oligomerization and determining the subcellular localization of ALK fusion (cytoplasmic and nuclear in NPM1-ALK-positive cases, and cytoplasmiconly or membranous in variant fusions).8,9 In addition, they impact a range of biological activities of ALK chimeras, including proliferation, transformation and metastatic capacities.10,11 Comparative analysis of the biological properties of ALK oncoproteins, however, is hampered by the relative low-frequency of particular variant ALK fusions. Interestingly, ALK rearrangements have also been detected in large B-cell lymphoma and various tumors of mesenchymal and epithelial origin, including inflammatory myofibroblastic tumors, lung cancer, esophageal squamous cell carcinoma and others.7 Notably, the same ALK fusions, such as TPM3-ALK, TPM4-ALK, TFG-ALK, SEC31A-ALK, ATIC-ALK, CLTC-ALK and EML4-ALK occur in ALK+ malignancies of different cells of origin, highlighting the crucial role of ALK in tumorigenesis.7,12 These findings prompted the development of new therapeutic strategies regarding ALK+ tumors. Of particular importance is the recent discovery of specific ALK tyrosine inhibitors,13,14 one of which, crizotinib, has proven to have clinical efficacy in the treatment of ALK+ non-small cell lung cancer and ALCL.15,16 Herein, we report the molecular characterization of ALK fusions in eight ALK+ ALCL cases exhibiting a cytoplasmic-only ALK staining pattern recently diagnosed in our institution. Intriguingly, all cases analyzed by fluorescence in situ hybridization (FISH) at the time of diagnosis revealed copy number gain of the rearranged ALK gene.

Methods Case selection Eight cases of ALK+ ALCL with a cytoplasmic-only expression of ALK and available bioarchived material were selected from the database of the Center for Human Genetics and Department of Pathology, University Hospital Leuven, Belgium. Morphology, immunophenotype and clinical records of the reported cases were reviewed. The institutional review board “Commissie Medische Ethiek” of the University Hospital approved this retrospective study and renounced the need for written informed consent from the participants (S56799, ML10896: 12/08/2014).

Cytogenetics and FISH Conventional Giemsa (G)-banding chromosomal analysis and FISH analysis followed routine protocols. The probes applied on patient material and Ba/F3 cell lines are listed in the Online Supplementary Table S1. Non-commercial probes were directly labeled with SpectrumOrange- and SpectrumGreen-dUTP (Abbott Molecular, Ottignies, Belgium) by random primed label1606

ing. FISH images were acquired with a fluorescence microscope equipped with an Axiophot 2 camera (Carl Zeiss Microscopy, Jena, Germany) and a MetaSystems Isis imaging system (MetaSystems, Altlussheim, Germany).

Array-based genomic hybridization (aCGH) Total genomic DNA was isolated from frozen lymphoma samples using standard procedures. High-resolution aCGH was performed using Affymetrix Cytoscan HD, following the manufacturer protocols. Downstream data analysis of copy number alterations was conducted using the software Array Studio, version 6.2.

Cell culture and growth curves Ba/F3 cells constitutively expressing Cas9 (Ba/F3 Cas9) were generated using a retroviral expression vector containing a Cas9 expression cassette (Online Supplementary Figure S1). Ba/F3 Cas9 cells were cultured with interleukin 3 (IL3) in Roswell Park Memorial Institute (RPMI) medium + 10% fetal bovine serum (FBS) before and after electroporation. For growth curves, viability and proliferation were measured on a Guava flow cytometer (Merck Millipore, Darmstadt, Germany) for several consecutive days.

Quantitative real time polymerase chain reaction (QRT-PCR) QRT-PCR with the GoTaq qPCR Master Mix (Promega, Madison, WI, USA) was used to analyze relative messenger ribonucleic acid (mRNA) expression of two Alk partner genes (Npm1 and Atic) in Ba/F3 Cas9 cells. Mouse Hprt1 and Rpl4 were used as an internal control. Primers are listed in the Online Supplementary Table S2. All samples were run in triplicate and data were analyzed with the comparative Ct (DDCt) method.

CRISPR/Cas genome editing Guide (g)RNAs were designed using the Zhang lab's clustered regularly interspaced short palindromic repeats (CRISPR) design tool (Broad Institute of MIT and Harvard, Cambridge, MA, USA) (Online Supplementary Table S3). gRNAs were cloned into an expression plasmid (pX321, derived from pX330, Zhang lab; Online Supplementary Figure S2). Electroporations of Ba/F3 Cas9 cells were carried out using a Gene Pulser Xcell electroporation system (Bio-Rad Laboratories, Hercules, CA, USA). After electroporation, cells were kept in RPMI medium + 10% FBS + IL3 for three days before IL3 depletion was carried out. Other applied techniques, including the 5’ Rapid Amplification of Complementary (c)DNA ends (RACE) PCR, Low Coverage Full Genome Sequencing (LCFGS), Targeted Locus Amplification (TLA) and Nested RT-PCR are briefly described in the Online Supplementary Methods.

Results Clinical and pathological features Relevant clinical and pathological features of the reported cases of ALK+ ALCL are shown in Table 1. There were two children (a 5-year-old boy and a 13-year-old girl) and six adults (two female and four males) ranging from 49 to 78 years of age (mean 64.8 years). All patients presented with lymph node involvement and one displayed additional skin lesions (stages I [cases 3 and 4], II [cases 5 and 8], III [cases 2 and 7] and IV [cases 1 and 6]). Five adult patients (cases 2-6) were treated with the chemotherapy regimen cyclophosphamide, adriamycin, vincristine and haematologica | 2017; 102(9)


Recurrent gain of ALK in ALK+ ALCL

prednisone (CHOP), reached complete remission (CR) and are still alive. The sixth adult patient (case 7), initially diagnosed with classical Hodgkin lymphoma, was treated with doxorubicin, bleomycin, vinblastinea and dacarbazine (ABVD) and radiotherapy, and also reached CR. He relapsed very recently (after 165 months) and died due to disease-related complications after the third course of CHOP. Two other patients also experienced a more aggressive disease course: one (case 2) achieved a second CR following treatment with dexamethasone, cytarabine and cisplatin (DHAP), and the first pediatric patient (case

1) initially treated according to the ALCL99 protocol, experienced three relapses. The latter achieved CR after treatment with crizotinib and allogenic stem cell transplantation, but died 72 months after diagnosis due to severe graft-versus-host disease (GvHD) and respiratory failure. The second pediatric patient (case 8) received six cycles of polychemotherapy (ALCL99) and remains in CR. Histopathology showed proliferation of anaplastic lymphoid cells and the presence of hallmark doughnut cells in all cases. The immunophenotype of the individual tumors is shown in Table 1 and illustrated in the Online

Table 1. Relevant clinical and pathological data. Case

Sex/ Sites of Age involvement, at Ann Arbor diagnosis stage

Immunophenotype

Histology

Treatment

Outcome

Survival (months)

Status (A/D)

1

M/5

CD3-, CD20-, CD30+, CD4 partial, CD8-, ALK(cyto)+ ,TIA1+, perforin +, granzyme B focal +

lymphohistiocytic subtype

Sequential polychemotherapy (ALCL99 protocol, high risk); 1st R: ALCL relapse protocol, BEAM + auto SCT; 2nd R: weekly vinblastine until 1.5 year; 3rd R: daily crizotinib + weekly vinblastine followed by MUD allo SCT

1st R 8 mo AD; 2nd R: 15 mo AD; 61 mo AD

72

D in CR (toxic death: severe GvHD,

2

M/55

Skin, supra and infradiaphragmatic lymph nodes. Stage IV.

respiratory failure)

Lymph nodes

CD3-, CD20-,

classical

8 CHOP; 1st R: 4 DHAP;

CR after CHOP;

(neck).

CD30+, CD5-, CD10-,

morphology

BEAM + auto SCT

R: CR after DHAP

prekeratin-, EMA-,

of ALCL

Stage III.

63

A

CD2-, CD4+, CD7-, CD8-, TIA1-, granzymB-, perforin+, ALK(cyto)+, EBV-, bcl2 weak+, CD433

F/49

Lymph nodes (axillary). Stage I.

CD2-, CD3-, CD20-, PAX5-, CD15-, CD30+, CD4+, CD8-, CD5-, MUM1+, ALK(cyto)+, TIA1-, granzymeB-, perforin+, EBV-, EMA+

intrasinusoidal growth, numerous plasma cells and eosinophils in the background

N CHOP

CR

20

A

4

M/77

Lymph nodes

CD2-, CD30+, CD4-,

classical

6 CHOP

CR

24

A

(neck).

CD8-, ALK(cyto)+,

morphology

Stage I.

TIA1-, granzymeB-,

of ALCL,

perforin+, EMA+,

foci of necrosis;

weak CD138, PAX5-,

histiocytes

IgA-, HHV8-, CD7-,

in background large multinuclear 8 CHOP cells with nucleoli; polymorphic stromal reactions with numerous plasma cells, neutrophils and eosinophils

CR

45

A

CR

14

A

5

M/52

Lymph nodes (mesenteric). Stage IIA.

CD45-, CD79aCD30+, CD4+, CD8-, ALK(cyto)+, TIA1+, perforin+, granzymeB+, EBV-,EMA+

6

F/78

Bone, supra

CD30+, ALK (cyto)+,

classical

and infra

CD45+, CD4+,

morphology

diaphragmatic

CD20 weak+,

of ALCL

lymph nodes.

Pax5 weak+,

Stage IVB

CD3-, CD5-, CD2-, CD8-

8 CHOP

Perforin+, granzyme+, EMA+, TIA1haematologica | 2017; 102(9)

1607


J-A. van der Krogt et al. Table 1: continuation 7

8

M/65

F/13

Lymph nodes (neck, axillary, retoroperitoneal, iliacal). Stage IIIa.

Lymph nodes

CD30+, CD15 false +, CD20-, CD3-, misdiagnosed as classical Hodgkin. Performed in 2016 after recurrence: PAX5-, CD2-, CD4-, CD8-, ALK(cyto)+, TIA1+ (partial), perforin+, granzymeB+ (partial), EBV-

Nodular sheet-like proliferation of immunoblastic to anaplastic cells; prominent intrasinusoidal localization; similar at recurrence, plus prominent histiocytic response with erythrophagocytosis and emperipolesis

2002: (initially misdiagnosed as classical Hodgkin): 6 ABVD + RT at left cervical and axillar region (30 Gy).

CD30+, ALK

Diffuse parenchymal

ALCL 99

and pseudo-cohesive

protocol: 6 courses

retroperitoneal,

intrasinusoidal

of polychemotherapy

iliacal).

+, EMA+, TIA1+,

spread of

Stage IIA.

perforin+, GranzymeB+,

characteristic

168

D

CR

14

A

Relapse in February 2016: stadium IVB (neck, supra- and infradiaphragmatic lymph nodes, lung (subpleural lesions)). 8 CHOP (3 received).

(infradiaphragmatic: (cytoplasmic, membraneous)

R: 166 mo AD

CD4+ (weak), CD20-, CD3- hallmark cells. No prominent stromal reaction. M: male; F: female; R: relapse; SCT: stem cell transplantation; ABVD: doxorubicin, bleomycin, vinblastine, dacarbazine; CHOP: cyclophosphamide, adriamycin, vincristine, prednisone; DHAP: dexamethasone, cytarabine, cisplatin; BEAM: carmustine, etoposide, cytarabine, melphalan; MUD: matched unrelated donor; RT: radiotherapy; N: number of cycles unknown; CR: complete response; AD: after diagnosis; D: died; A: alive; GvHD, graft-versus-host disease; mo: months; cyto: cyoplasmic; ALCL: anaplastic large cell lymphoma.

Supplementary Figure S3. All cases were CD30 positive, expressed a CD4+ or null cell phenotype in conjunction with variable cytotoxic markers, and overexpressed ALK1 in a strictly cytoplasmic pattern. The stromal infiltrate was prominent in all cases with a variable amount of histiocytes and/or neutrophils.

Cytogenetic and molecular studies Case 1. Cytogenetic analysis showed a complex karyotype with del(2)(p23) (Table 2). FISH with LSI ALK demonstrated a break apart (BA) pattern with the green (5’ALK) signal on del(2)(p23) and two red (3’ALK) signals on a marker chromosome of postulated chromosome 11 origin (Figure 1A). Whole chromosome painting proved that der(11) indeed harbors the duplicated fragment of chromosome 2 inserted at 11q. The t(2p23/ALK) rearrangement was further investigated using the 5’ RACE PCR approach. The analysis identified an in-frame fusion transcript in which exon 8 of EEF1G, the gene located at 11q12.3, was fused to exon 20 of ALK (Figure 1B). More precisely, the fusion joined nucleotide 1161 of EEF1G (The National Center for Biotechnology Information (NCBI) ref: NM-001404), corresponding to an intronic region between exons 8 and 9, with nucleotide 4226 of ALK (NM-004304) situated in exon 20. The EEF1G-ALK fusion was subsequently confirmed by Sanger sequencing of the fragment obtained by nested RT-PCR (Figure 1B) and by FISH using probes for 5’EEF1G and 3’ALK (data not shown). Array Comparative Genomic Hybridization (CGH) analysis demonstrated that 3’ALK is involved in the gain of 2p23.2p25.3 and that the gain of 11q11q13.4 affects EEF1G (Figure 1C,D). In addition, the gain of five other regions, loss of three large regions, including that of 9p21.3p24.3/CDKN2A/B, and three microdeletions were detected (Figure 1C; Table 1). Interestingly, the focal deletion at 8q24.21 covered ~280 kbp sequences flanking the centromeric border of MYC. The loss was confirmed by 1608

FISH with LSI MYC; one of two apparently normal looking chromosomes 8 [der(8)] revealed a significantly diminished red signal (Figure 1A). Case 2. Cytogenetic analysis was unsuccessful. Interphase FISH with LSI ALK showed a BA pattern associated with the gain of up to seven red (3’ALK) signals (Figure 2A). The FISH pattern suggested a diploid and tetraploid status of abnormal cells. The ALK rearrangement was further investigated using LCFGS. The analysis identified the RNF213-ALK fusion (supported by at least eight read pairs) whereby exon 8 of RNF213 is fused to exon 20 of ALK, resulting in an in-frame fusion transcript equal to the previously described cases (Figure 2B).6,17,18 The fusion and its gain were confirmed by FISH (Figure 2A). The performed aCGH analysis demonstrated the ALK- and RNF213-associated gain of 2p23.2p25.3 and 17q23.3q25.3, respectively (Figure 2C,D). In addition, aCGH detected the gain of six other regions and loss of three regions, all listed in Table 2. Notably, breakpoints of the amplified 4p12p16 and 5q33.3q35.3 regions affected the RNF212/4p16 and ITK/5q33.3 genes, respectively. FISH with the respective BA probes (see Online Supplementary Table S1) confirmed the unbalanced rearrangement and partial gain of both genes. Furthermore, FISH with the probes for 5’RNF212 and 3’ITK showed multiple, but not colocalized, signals, precluding the RNF212-ITK hybrid gene (data not shown). Unfortunately, LCFGS did not identify any fusion of RNF212 and ITK. Cases 3-6. The cases were not subjected to cytogenetic analysis. FISH with LSI ALK applied on formalin-fixed paraffin-embedded (FFPE) material confirmed ALK rearrangement and displayed the presence of three to seven red (3’ALK) signals in all cases (Table 2, Figure 3A). Although a number of fused (F), red (R) and green (G) signals in analyzed cells varied, all cases showed a clear predominance of the red signals and variations of the 1F1G3R haematologica | 2017; 102(9)


Recurrent gain of ALK in ALK+ ALCL

pattern. As a similar pattern was observed in ATIC-ALKpositive lymphoma,19 we initially examined the status of this gene. FISH with the ATIC BA assay revealed a BA pattern and the presence of extra 5’ATIC signals in all four cases (Figure 3B). Subsequent FISH with the 5’ATIC and 3’ALK probes demonstrated three to five fused signals per

cell, confirming the ATIC-ALK rearrangement and gain of the chimeric gene in all four cases (Figure 3C). Case 7. The biopsy taken at time of relapse (2016) showed a complex diploid karyotype with a presumed der(1)ins(1;?)(q21;?) and ider(1)(q10) containing a duplicated 1q arm of der(1)ins(1;?)(q21;?) and del(2)(p23) (Table 2,

A

B

C

D

Figure 1. Cytogenetic and molecular analysis of case 1.(A) Examples of FISH experiments with LSI ALK and LSI MYC BA probes. Metaphase FISH demonstrated BA LSI ALK pattern associated with a duplication of the red/3’ALK signal on der(11) (left image). Paints for chromosomes 2 (red) and 11 (green) confirmed insertion of 2p23p25 at 11q12 (inset). The same aberrant LSI ALK pattern (one colocalized-one green-two red signals) was observed in interphase cells (middle image). Metaphase FISH with LSI MYC showed a diminished red signal on der(8), confirming a focal deletion at 8q24.21 (right image). (B) Schematic representation of the EEF1G, ALK and EEF1G-ALK protein structures (upper panel). Sequencing of the fragment amplified by EEF1G-ALK nested RT-PCR identified an in-frame fusion between exon 8 of EEF1G (breakpoint between exon 8 and 9) and exon 20 of ALK (breakpoint in the middle of exon 20) as shown in the electropherogram (lower panel). (C) Array CGH profile of case 1 showing several unbalanced regions, including gain of 2p23pter and 11q11q13.4 (marked). (D) The selected 2pter (upper panel) and 11q (lower panel) regions. Note the 2p23pter gain-associated break within the ALK gene (gain of 3’ALK) and localization of EEF1G in the middle of gained 11q11q13.4 region.

haematologica | 2017; 102(9)

1609


J-A. van der Krogt et al.

Figure 3D). Interphase FISH with LSI ALK revealed the 1F1G3R pattern (Figure 3E), as found in ATIC-ALK-positive cases. Considering that the 1q21 breakpoint merges with the localization of TPM3, the known partner of ALK,6 a possible involvement of this gene was investigated by FISH with the designed TPM3 BA assay (Online Supplementary Table S1). Neoplastic cells revealed the 1F3(RsepG) pattern, thus postulating the rearrangement of

TPM3 due to ins(1;2)(q21.3;p23p25) (Figure 3F). The TPM3-ALK fusion was subsequently confirmed by FISH using the 5’TPM3/3’ALK probes. In the proceeding step, we revised the diagnostic sample (the case was initially misdiagnosed as classical Hodgkin lymphoma), whereby cytogenetic analysis identified rare hypertetraploid cells with i(1)(q10) and del(2)(p23) (Table 2). FISH with LSI ALK (Figure 3G), TPM3 BA (Figure 3H) and the 5’TPM3/3’ALK

A

B

C

D

1610

Figure 2. Cytogenetic and molecular analysis of case 2. (A) Examples of interphase FISH experiments. Note a BA pattern of LSI ALK and RNF213 BA probes associated with gain of six 3’ALK (red) signals (left image) and six 5’RNF213 (red) signals (middle image). Several colocalized 5’RNF213 (red) and the 3’ALK (green) signals in interphase cells confirm presence and gain of the RNF213-ALK hybrid gene in this case (left image). (B) Schematic representation of the RNF213, ALK and RNF213-ALK protein structures (upper panel). LCFGS resulted in at least eight single read pairs covering the in-frame fusion between exon 8 of RNF213 and exon 20 of ALK (lower panel). (C) Array CGH profile of case 2 showing several unbalanced regions, including gains of 2p23pter and 17q23qter (marked). (D) The selected 2p23pter (upper panel) and 17q23qter (lower panel) regions evidencing the gainassociated breaks within the ALK and RNF213 genes, respectively.

haematologica | 2017; 102(9)


Recurrent gain of ALK in ALK+ ALCL

probes (Figure 3I) demonstrated the presence, at the time of diagnosis (2002), of the TPM3-ALK rearrangement and copy number gain of the ALK fusion. Case 8. Cytogenetic analysis identified only one abnormal (near tetraploid) metaphase cell harboring five copies of chromosome 2 (Table 2). LSI ALK applied on this

metaphase cell hybridized with all chromosomes 2: one showed a fused signal and the other four were marked by red signals (Figure 3J). The abnormal FISH pattern (1-2F4R) was detected in 11% of interphase cells. Loss of the 5’ALK sequences suggested a focal interstitial del(2)(p23p23) leading to a novel fusion of ALK with a gene located at the centromeric breakpoint of the deletion. To identify a

Table 2. Results of cytogenetic and molecular studies.

Case Sample/ Cytogenetics status

aCGH profile

1a

LN/P

gains: 1q21.3q44, 2p23.2p25.3, 5q15q35.3, 6p22.2p25,11q11q13.4, 22q13.1q13.33 20q13.1q13.33, losses: 8q24q24,9p21.3p24.3, 10q25.3q25, 11q22.3q25

2a

LN/D

44-47,XY,-Y,del(2) (p23),-5[3], der(6)t(5;6) (q22;p25),add(8)(p23)[3],-9, del(10)(q25),der(11)(?::11pter-> q12.3::2p23-> 2pter::2pter-> p23::11q12.3->q11::q13.5-> 23.1::?), add(12)(q24)[3],+16[2],+1-4mar[4] NM

FISH LSI ALK

ish: 2p(F),der(2)(G), WCP2:2+,del(2p)+, der(11)( Rx2); der(11)+; WCP11:11+, nuc ish:1F,1G,2R (85%) der11+;5’EEF1G-SO/ 3’ALK-SG:2F1R1G; LSI MYC: 8F,8GRdim

gains: 1q21.3q44, 2p23.2p25.3, 4p12p16, 5q33.3q35.3, 6, nuc ish:1-3F,1-3G, 12q15q23.2, 16p11.2p13.3, 1-7R (19%) 17q23.3q25.3, 21q11.2q22.3 losses: 1p36.2p36.3, 5q15q21.2, 18q11.2q23

3b

LN/D

ND

ND

nuc ish:1-2F,1-2G,5-7R (16%)

4b

LN/D

ND

ND

nuc ish: 1F,1G,5R (15%)

5b

LN/D

ND

ND

nuc ish:1-3F,1-2G,3-5R (21%)

6a

LN/D

NM

ND

nuc ish:1-2F,1-3G,3-5R (9%)

7a

LN/D

88-89,XXY,ND Y,i(1)(q10),+i(1)(q10),del(2)(p11), +del(2)(p11),+del(2)(p23)x2, -4,-5,-8,-8,-11,-15,-15,-18,-18,+20[2], +20[2],-21,-22[2],+2-6mar[cp3] 47,XY,ins(1;2)(q21.3;p23p25), ND +ider(1)(q10)ins(1;2)(q21.3;p23p25), del(2)(p23)[2] 88,XX,-X,-X,-1,add(1)(q44),+2, ND -3,add(4)(p16),+5,+7,-11,+12,-13,-15,16,+i(17)(q10),-19,-20[1]

nuc ish:2-5F,1-2G, 5-6R (15%)

LN/P

8

LN/D

Other probes

nuc ish:1F,1G,3R (24%) 2F,4xdel(2)(p23p?) Nuc ish:1F4R (11%)

ALK fusion partner gene (localization)

Detected by

EEF1G (11q12.3) 5'RACE PCR

RFN213 BA:3F3-7R3-7G; RFN213 (17q25.3) LCFGS 5’RFN213-SO/3’ALK-SG: 3-7F2-3R2-3G; RNF212 BA: 5-7GITK BA: 5-7R5’RNF SO/3’ITK-SG: 5-7R5-7G ATIC BA: variable pattern ATIC (2q35) FISH with gain of red signals; 5'ATIC-SO/3'ALK- SG: 2-5F and variable R/G signals ATIC BA: ATIC (2q35) FISH variable pattern with gain of red signals; 5'ATIC-SO/3'ALK- SG: 2-5F and variable R/G signals ATIC BA: ATIC (2q35) FISH variable pattern with gain of red signals; 5'ATIC-SO/3'ALK- SG: 2-5F and variable R/G signals ATIC BA: ATIC (2q35) FISH (9%) variable pattern with gain of red signals; 5'ATIC-SO/3'ALK- SG: 2-5F and variable R/G signals TPM3 BA: TPM3 (1q21.3) FISH 1-2F5(RsepG); 5'TPM3-S0/3'ALK-SG: 1-2R1-3G5F TPM3 BA: 1F3(RsepG); 5'TPM3-S0/3'ALK-SG: 1R1G3F TPM3 BA: 2F4R; 5'TPM3-S0/ 3'ALK-SG:2R1G4F

TPM3 (1q21.3)

TLA

a FISH performed on cytogenetic specimen; bFISH performed on FFPE material. FISH: fluorescence in situ hybridization; aCGH: array-based comparative genomic hybridization; LN: lymph node; D: diagnosis; P: progression; NM: no mitosis, ND: not done; ish: in situ hybridization; nuc ish: interphase in situ hybridization; F: fused ; G: green; R: red; BA: break apart; RACE PCR: rapid amplification of cDNA ends polymerase chain reaction; LCFGS: low coverage full genome sequencing; TLA: targeted locus amplification technology.

haematologica | 2017; 102(9)

1611


J-A. van der Krogt et al.

potential partner gene, we used the TLA approach. Unexpectedly, the analysis identified the TPM3-ALK fusion with the breakpoints at the intron 7 of TPM3 and the intron 19 of ALK (Online Supplementary Figure S4). FISH confirmed the TPM3 rearrangement, which was associated with loss of the 3’TPM3 sequences, and showed that four out of five chromosomes 2 harbor one copy of TPM3-ALK (Figure 3K,L). Altogether, we identified a cryptic insertion of the 5’TPM3 into the rearranged ALK locus, loss of the 5’ALK and 3’TPM3 sequences, and gain of two copies of TPM3-ALK.

Functional studies Gain of the EEF1G-, RFN213-, ATIC- and TPM3-ALK hybrid genes found in the present cases contrasts with a constant occurrence of a single copy of NPM1-ALK in t(2;5)-positive ALCL.20-22 We hypothesized that ALCLs harboring the variant rearrangements compensate an insufficient expression of the ALK fusions (driven by relatively weak promoters of the partner genes) via an increased dosage of the chimeric gene. To validate this concept, we analyzed expression levels of NPM1, EEF1G, RNF213, ATIC and TPM3 in various normal lymphoid

Figure 3. Cytogenetic and FISH analysis of cases 3, 7 and 8. (A-C) Note a BA pattern of LSI ALK and ATIC BA probes associated with a gain of four 3’ALK (red) and four 5’ATIC (red) signals, respectively, in case 3. Several colocalized 5’ATIC/3’ALK signals in interphase cells demonstrate gain of the ATICALK hybrid gene in this sample. Similar FISH results were obtained in cases 4 and 5. Case 7: (D) Partial karyotype (at time of relapse) illustrating insertion of 2p23p25 at 1q21.3 (arrowhead), isochromosome 1q containing duplicated long arm of ins(1;2)(q21.3;p23q25) (two arrowheads) and del(2)(p23) (arrow). (E) Note rearrangement of ALK and gain of two extra 3’ALK (red) signals, and (F) three separated red and green TPM3 BA probes, likely marking ins(1) and ider(1). (G-I) Similar FISH patterns were observed in the diagnostic sample. Case 8: (J) Note four copies of der(2) marked by red signals of LSI ALK, (K) unbalanced rearrangement of TPM3 and (L) loss of TPM3 from one chromosome 1 and four copies of der(2)/TPM3-ALK.

1612

haematologica | 2017; 102(9)


Recurrent gain of ALK in ALK+ ALCL

cells and T-cell malignancies using previously generated RNA sequencing (Seq) data.23,24 The analysis revealed a significantly lower (P-value <0.001) expression of RNF213, ATIC and TPM3 in all samples when compared to NPM1

(Figure 4A). In contrast, expression of EEF1G was similar to that of NPM1. Based on these findings, we hypothesized that oncogenic transforming properties of the three variant ALK fusions, RFN213-ALK, ATIC-ALK and

B

A

C

D

F

E

G

H

Figure 4. Functional analysis of the NPM1-, EEF1G-, RNF213-, TPM3- and ATIC-ALK fusions. (A) Expression analysis of the five ALK partner genes using previously generated RNA-Seq data.23,24 In contrast to EEF1G, expression of TPM3, RNF213, and ATIC is significantly lower (P-value <0.001) when compared to NPM1 in different malignant and non-malignant cell types (HSTL: hepatosplenic T-cell lymphoma [n=4]; T-ALL: T-cell acute lymphoblastic leukemia [n=5]; PTCL: peripheral T-cell lymphoma [n=2]; Spleen [n=1]; Thymus [n=1]; LN: lymph nodes [n=3]; Th1: T helper 1 cells [n=5]; Th2: T helper 2 cells [n=5]; Treg [n=5], and CD4 naĂŻve T-cells [n=4]). Error bars represent the standard deviation (SD). (B) QRT-PCR on Ba/F3 Cas9 cells showing the expression levels of Npm1 and Atic. The expression of Atic is significantly lower than Npm1 (P<0.001). Error bars represent the SD. (C) Representation of the Alk, Npm1, and Atic mouse genes. Exons are indicated by vertical bars. Red arrows indicate the location of the gRNA target sites. (D) Growth curve showing the transforming capacities of Ba/F3 Cas9 cells harboring an endogenous Npm1Alk or Atic-Alk fusion. Error bars represent the SD. (E) Growth curve showing the growth rate of Ba/F3 Cas9 cells after transformation by the endogenous Alk fusion. Error bars represent the SD. (F-H) Examples of metaphase and interphase FISH results showing the endogenous Alk fusions in Ba/F3 Cas9 cells. Arrows indicate colocalized signals/chimeric genes. Note a constant presence of a single copy of Npm1-Alk and gain of Atic-Alk in late cultures. w/o: without; IL3: interleukin 3.

haematologica | 2017; 102(9)

1613


J-A. van der Krogt et al.

TPM3-ALK, could be lower than the strong NPM1-ALK kinase. To validate our hypothesis, we attempted to generate the Npm-Alk and Atic-Alk fusions in the genome of Ba/F3 cells using CRISPR/Cas9 genome editing. The expression level of Npm1 is around 50-fold higher than the expression level of Atic in this cell line, making it a suitable model to test our hypothesis (Figure 4B). We designed a gRNA targeting exon 20 of Alk and gRNAs targeting Npm1 and Atic in regions corresponding to the breakpoints observed in our patient samples (Figure 4C). gRNAs targeting Alk and a fusion partner were simultaneously introduced in Ba/F3 Cas9 cells by electroporation. Upon IL3 depletion, both endogenous Alk fusions were able to transform the Ba/F3 Cas9 cells, although cells harboring an endogenous Atic-Alk fusion needed more time to recover from this depletion (Figure 4D). In addition to the slower transformation rate, cells harboring an endogenous Atic-Alk fusion presented with a lower growth rate after transformation than cells with an endogenous Npm1-Alk fusion (Figure 4E). FISH demonstrated the presence of one copy of the Npm1-Alk and Atic-Alk chimeric genes per cell in the respective cell lines (Figure 4F,G). After keeping the cells in culture for three months, we observed gain of one to three copies of the Atic-Alk fusion gene in approximately 20% of these cells. In contrast, the Npm1-Alk cell line kept a single copy of the chimeric gene in all cells (Figure 4H). Altogether, these data demonstrate that the Npm1-Alk fusion is a more potent driver of proliferation than Atic-Alk in Ba/F3 cells, and that the expression level of the fusion partner is a key factor in the transformation potential of the oncogenic fusions.

Discussion Our study provided evidence that ALCL driven by at least three variant ALK fusions, RNF213-ALK, ATIC-ALK and TPM3-ALK, is characterized by a copy number gain of the ALK hybrid gene. Gain of RNF213-ALK was already observed by us in the previously reported case of ALK+ ALCL, harboring two copies of der(17) Genetic t(2;17)(p23.2;q25.3)/RNF213(ALO17)-ALK.17 mechanisms underlying the amplification of RNF213-ALK in case 2, however, remain unknown. Notably, the four ATIC-ALK-positive cases presented herein, as well as all reported ATIC-ALK cases documented by FISH,19,25-27 showed extra copies of the chimeric gene. We previously documented that the ATIC-ALK fusion is generated by a pericentric inv(2)(p23q35) and is constantly accompanied by isochromosome 2q [derivative of inv(2)] comprising two extra copies of ATIC-ALK.19,25 Based on these data, we presume that the cases reported herein also carry inv(2)(p23q35) and one to three copies of ider(2q). Intriguingly, a similar mechanism of gain of ALK hybrid emerged to underpin the TPM3-ALK rearrangement detected in case 7. The fusion was likely created by the insertion of 3’ALK/2p23p25 at 1q21.3/TPM3 and gained by a subsequent formation of ider(1)(q10). In the second TPM3-ALK-positive case, however, the chimeric gene was produced by a cryptic insertion of 5’TPM3 into the rearranged ALK locus and gained by a duplication of the der(2), as in the case reported by Siebert et al.28 Notably, one more informative case with TPM3-ALK also presented with two to three copies of the 3’ALK.29 Lamant et al.,6 who originally described the TPM3-ALK fusion in ALK+ ALCL, linked it to t(1;2)(q25;p23). Considering, however, 1614

an opposite transcriptional orientation of both genes, generation of the in-frame TPM3-ALK fusion requires more complicated events, as illustrated in the present and previously reported cases.28 Altogether, our genetic findings, supported by the published data,17,19,26-29 highlight a strong association of RNF213-ALK, ATIC-ALK and TPM3-ALK with a copy number gain of the chimeric gene. A spectrum of such ALK fusions is probably broader, but a frequent lack of cytogenetic and/or FISH data hampers their identification. These observations contrast with ALCL driven by NPM1ALK20-22 and at least three variant fusions, CLTC-ALK,17 TPM4-ALK30,31 and TRAF1-ALK,27,32 shown to carry a single copy of the ALK hybrid gene. We initially included the novel EEF1G-ALK variant found to be duplicated on der(11) in the former category. Given, however, that two recently published ALCL cases with EEF1G-ALK did not show copy number gain of the ALK fusion gene,33 duplication of EEF1G-ALK in our case was likely associated with progression of the disease, similar to a case of leukemic ALK+ ALCL with an extra copy of NPM1-ALK.34 This conclusion is additionally supported by our other data illustrating a strong expression of EEF1G in normal and malignant T cells, similar to that of NPM1. Thus, there is no reason why the EEF1G-ALK fusion gene would require an increased copy number. In contrast, expression of RNF213, ATIC and TPM3 in lymphoid cells was significantly lower than NPM1 and EEF1G. These findings support the concept that ALCL driven by the RNF213-ALK, ATIC-ALK and TPM3ALK fusions might require an increased dosage of the ALK chimeric gene to compensate an insufficient expression of ALK in neoplastic cells. To test if the best documented ATIC-ALK fusion is indeed less transforming than the NPM1-ALK fusion, we generated the Npm1-Alk and Atic-Alk fusion genes by inducing Cas9 mediated chromosomal rearrangements in Ba/F3 cells. Growth properties of these engineered cells showed that the transformation potential of Atic-Alk is significantly lower when compared to Npm1-Alk, likely due to a lower expression level of Atic. Interestingly, the tendency of the Atic-Alk-positive Ba/F3 cells to gain extra copies of the Atic-Alk chimeric gene recapitulates the events observed in the original tumors and confirms the selective pressure of the cells to acquire additional copies of the Atic-Alk fusion. We also attempted to generate an endogenous Rnf213-Alk fusion, but since Rnf213 is very lowly expressed in Ba/F3 cells, this fusion could not transform the cells. Tpm3 was not included in the functional studies, since it was only recently added to the study. Of note, Giuriato et al.35 previously showed that in conditional knock-in mice both TPM3-ALK and NPM1ALK could induce B-cell lymphoma/leukemia with a similar disease latency. In these transgenic mice, however, the expression levels of both ALK fusions were similar, since their expression was driven by an external promoter. This could explain the observed equal tumorigenic potential of both ALK fusions. The novel EEF1G-ALK fusion which we identified in a child with ALK+ ALCL extends the already long list of ALK fusion partners comprising approximately 20 genes.7 The very recent report of two pediatric patients with EEF1G-ALK33 indicates that the fusion is recurrent and strongly associated with pediatric ALK+ ALCL. EEF1G, located at 11q12.3, has 10 exons encoding for an eukaryotic translation elongation factor 1γ. Together with α, β haematologica | 2017; 102(9)


Recurrent gain of ALK in ALK+ ALCL and δ subunits, it forms the eukaryotic elongation factor complex, which is predominantly involved in protein biosynthesis with an elongation of the polypeptide chains.36,37 EEF1G comprises a glutathione-S-transferase (GST)-like N-terminal domain and a C-terminal (CT) domain.37 Although all four subunits of the elongation factor complex are highly expressed in most eukaryotic cells, the role of human EEF1G is poorly understood. Notably, an increased expression of EEF1G has been detected in various human carcinomas.37-40 It has been suggested that overexpression of EEF1G stimulates the overgrowth of neoplastic cells.41 In the present case of ALK+ ALCL, exons 1 to 8 of EEF1G are fused to exon 20-29 of ALK, resulting in a chimeric protein with the complete GST-like N-terminal domain and part of the CT domain of EEF1G, fused to the complete intracellular protein tyrosine kinase (PTK) domain of ALK (Figure 2B). Even though the complete PTK domain is involved, the ALK fragment contains the final 513 amino acids, which is shorter in comparison to the fragment involved in most other ALK fusions, containing the final 563 amino acids.42 Of note, both cases reported by Palacios et al.33 revealed breakpoints within intron 7 of EEF1G and intron 20 of ALK. As the fusions were identified by RNA-Seq, genetic mechanisms underlying these rearrangements are unknown. The postulated t(2;11)(p23;q12.3), however, is unlikely due to an opposite transcriptional orientation of both genes, similar to TPM3 and ALK. Functional studies performed by the authors provided evidence of the dimerization properties of the EEF1G-ALK fusion, constitutive activation of ALK kinase and its strong oncogenic potential similar to that of NMP1-ALK. Our patient with EEF1G-ALK experienced an unfavorable and fatal clinical course. Although ALK+ ALCL has a relatively good prognosis, chemoresistant and aggressive forms of this lymphoma, recurrently featured by MYC aberrations, have been recurrently reported.32,43-45 Notably, our case displayed a microdeletion neighboring MYC and loss of CDKN2A/B at 9p21, a known poor prognostic factor in tumors.46 We presume that a focal del(8)(q24q24) could activate MYC by loss of negative regulatory elements upstream of the gene, for example. Although clini-

References 6. 1. Swerdlow SH, Campo E, Harris NL, et al. WHO classification of tumours of haematopoietic and lymphoid tissues. 4th ed. Lyon, France: International Agency for Research on Cancer: 2008. 2. Morris SW, Kirstein MN, Valentine MB, et al. Fusion of a kinase gene, ALK, to a nucleolar protein gene, NPM, in non-Hodgkin's lymphoma. Science. 1994;263(5151):12811284. 3. Chiarle R, Gong JZ, Guasparri I, et al. NPM-ALK transgenic mice spontaneously develop T-cell lymphomas and plasma cell tumors. Blood. 2003;101(5):1919-1927. 4. Kuefer MU, Look aT, Pulford K, et al. Retrovirus-mediated gene transfer of NPM-ALK causes lymphoid malignancy in mice. Blood. 1997;90(8):2901-2910. 5. Lange K, Uckert W, Blankenstein T, et al. Overexpression of NPM-ALK induces different types of malignant lymphomas in

haematologica | 2017; 102(9)

7.

8.

9.

10.

cal data of the EEF1G-ALK-positive cases reported by Palacios et al.33 are not available, we believe that an aggressive clinical course and unfavorable prognosis of certain ALK+ ALCL is not determined by the type of ALK fusion, but rather by secondary hits affecting potent oncogenes and tumor suppressor genes (e.g., MYC, TP53, PRDM1 and CDKNA2/B).32,44,45,47 In summary, we investigated eight recently diagnosed cases of ALK+ ALCL with a cytoplasmic-only expression of ALK and identified one novel EEF1G-ALK rearrangement and three already known fusions, RNF213-ALK, TPM3-ALK and ATIC-ALK. Occurrence of the latter rearrangement in four out of eight (50%) of the cases studied confirms that ATIC-ALK is the most prevalent variant fusion in ALK+ ALCL. Importantly, RNF213ALK, TPM3-ALK and ATIC-ALK fusions were recurrently present in two or more copies, contrasting with the NPM1-ALK chimeric gene which constantly occurs in one copy. Our functional studies show that ALK+ ALCL driven by ATIC-ALK compensates the weak expression and possibly weak oncogenic properties of this variant ALK fusion by an increased gene dosage. We propose a similar explanation for the copy number gain of RNF213-ALK and TPM3-ALK. Altogether, our findings support the hypothesis that the transforming capacities of ALK fusions depend on the biological features of the partner genes.48 Acknowledgments The authors would like to thank Vanessa Vanspauwen and Emilie Bittoun for their excellent technical assistance, and Rita Logist for the editorial help. Funding This study was supported by the concerted action grant from the KU Leuven no. 3M040406 (JAvdK, PV, TT, JC and IW), research grants from the FWO Vlaanderen (G081411N to TT) and “Stichting tegen Kanker” (PV). MVDB holds a SB Fellowship of the Research Foundation-Flanders. PV is a senior clinical investigator of the FWO-Vlaanderen. TT holds a Mandate for Fundamental and Translational Research from the “Stichting tegen Kanker” (2014-083).

IL-9 transgenic mice. Oncogene. 2003; 22(4):517-527. Lamant L, Dastugue N, Pulford K, Delsol G, Mariame B. A new fusion gene TPM3ALK in anaplastic large cell lymphoma created by a (1;2)(q25;p23) translocation. Blood. 1999;93(9):3088-3095. H Hallberg B, Palmer RH. Mechanistic insight into ALK receptor tyrosine kinase in human cancer biology. Nat Rev Cancer. 2013;13(10):685-700. Falini B, Pileri S, Zinzani PL, et al. ALK+ lymphoma: clinico-pathological findings and outcome. Blood. 1999;93(8):26972706. Boi M, Zucca E, Inghirami G, Bertoni F. Advances in understanding the pathogenesis of systemic anaplastic large cell lymphomas. Br J Haematol. 2015;168(6):771-783. Drexler HG, Gignac SM, von WR, Werner M, Dirks WG. Pathobiology of NPM-ALK and variant fusion genes in anaplastic large cell lymphoma and other lymphomas. Leukemia. 2000;14(9):1533-1559.

11. Stein H, Foss Hd, Du H, et al. CD30+ anaplastic large cell lymphoma : a review of its histopathologic, genetic, and clinical features. Blood. 2000;96(12):3681-3695. 12. Sakamoto K, Nakasone H, Togashi Y, et al. ALK-positive large B-cell lymphoma: identification of EML4-ALK and a review of the literature focusing on the ALK immunohistochemical staining pattern. Int J Hematol. 2016;103(4):399-408. 13. Mologni L. Inhibitors of the anaplastic lymphoma kinase. Expert Opin Investig Drugs. 2012;21(7):985-994. 14. Pall G. The next-generation ALK inhibitors. Curr Opin Oncol. 2015;27(2):118-124. 15. Kwak EL, Bang YJ, Camidge DR, et al. Anaplastic lymphoma kinase inhibition in non-small-cell lung cancer. N Engl J Med. 2010;363(18):1693-1703. 16. Gambacorti PC, Farina F, Stasia A, et al. Crizotinib in advanced, chemoresistant anaplastic lymphoma kinase-positive lymphoma patients. J Natl Cancer Inst. 2014; 106(2):djt378.

1615


J-A. van der Krogt et al. 17. Cools J, Wlodarska I, Somers R, et al. Identification of novel fusion partners of ALK, the anaplastic lymphoma kinase, in anaplastic large-cell lymphoma and inflammatory myofibroblastic tumor. Genes Chromosomes Cancer. 2002;34(4):354-362. 18. Hernandez L, Bea S, Bellosillo B, et al. Diversity of genomic breakpoints in TFGALK translocations in anaplastic large cell lymphomas: identification of a new TFGALK(XL) chimeric gene with transforming activity. Am J Pathol. 2002;160(4):14871494. 19. Wlodarska I, De Wolf-Peeters C, Falini B, et al. The cryptic inv(2)(p23q35) defines a new molecular genetic subtype of ALKpositive anaplastic large-cell lymphoma. Blood. 1998;92(8):2688-2695. 20. Mathew P, Sanger WG, Weisenburger DD, et al. Detection of the t(2;5)(p23;q35) and NPM-ALK fusion in non-Hodgkin's lymphoma by two-color fluorescence in situ hybridization. Blood. 1997;89(5):16781685. 21. Perkins SL, Pickering D, Lowe EJ, et al. Childhood anaplastic large cell lymphoma has a high incidence of ALK gene rearrangement as determined by immunohistochemical staining and fluorescent in situ hybridisation: a genetic and pathological correlation. Br J Haematol. 2005; 131(5):624-627. 22. 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. 23. Finalet Ferreiro J, Rouhigharabaei L, Urbankova H, et al. Integrative genomic and transcriptomic analysis identified candidate genes implicated in the pathogenesis of hepatosplenic T-cell lymphoma. PloS One. 2014;9(7):e102977. 24. Atak ZK, Gianfelici V, Hulselmans G, et al. Comprehensive analysis of transcriptome variation uncovers known and novel driver events in T-cell acute lymphoblastic leukemia. PLoS Genet. 2013; 9(12):e1003997. 25. Ma Z, Cools J, Marynen P, et al. Inv(2)(p23q35) in anaplastic large-cell lymphoma induces constitutive anaplastic lymphoma kinase (ALK) tyrosine kinase activation by fusion to ATIC, an enzyme involved in purine nucleotide biosynthesis. Blood. 2000;95(6):2144-2149. 26. Colleoni GW, Bridge JA, Garicochea B, Liu

1616

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

J, Filippa DA, Ladanyi M. ATIC-ALK: A novel variant ALK gene fusion in anaplastic large cell lymphoma resulting from the recurrent cryptic chromosomal inversion, inv(2)(p23q35). Am J Pathol. 2000; 156(3):781-789. Takeoka K, Okumura A, Honjo G, Ohno H. Variant translocation partners of the anaplastic lymphoma kinase (ALK) gene in two cases of anaplastic large cell lymphoma, identified by inverse cDNA polymerase chain reaction. J Clin Exp Hematop. 2014;54(3):225-235. Siebert R, Gesk S, Harder L, et al. Complex variant translocation t(1;2) with TPM3ALK fusion due to cryptic ALK gene rearrangement in anaplastic large-cell lymphoma. Blood. 1999;94(10):3614-3617. Hoshino A, Nomura K, Hamashima T, et al. Aggressive transformation of anaplastic large cell lymphoma with increased number of ALK-translocated chromosomes. Int J Hematol. 2015;101(2):198-202. Meech SJ, McGavran L, Odom LF, et al. Unusual childhood extramedullary hematologic malignancy with natural killer cell properties that contains tropomyosin 4-anaplastic lymphoma kinase gene fusion. Blood. 2001;98(4):1209-1216. Liang X, Meech SJ, Odom LF, et al. Assessment of t(2;5)(p23;q35) translocation and variants in pediatric ALK+ anaplastic large cell lymphoma. Am J Clin Pathol. 2004;121(4):496-506. Abate F, Todaro M, van der Krogt J, et al. A novel patient derived tumorgraft model with TRAF1-ALK anaplastic large cell lymphoma translocation. Leukemia. 2014;(6):132. Palacios G, Shaw TI, Li Y, et al. Novel ALK fusion in anaplastic large cell lymphoma involving EEF1G, a subunit of the eukaryotic elongation factor-1 complex. Leukemia. 2017;31(3):743-747. Ries S, Rnjak L, Mitrovic Z, Kuvezdic KG, Nola M, Sucic M. CD13+ anaplastic large cell lymphoma with leukemic presentation and additional chromosomal abnormality. Diagn Cytopathol. 2010;38(2):141-146. Giuriato S, Foisseau M, Dejean E, et al. Conditional TPM3-ALK and NPM-ALK transgenic mice develop reversible ALKpositive early B-cell lymphoma/leukemia. Blood. 2010;115(20):4061-4070. Kumabe T, Sohma Y, Yamamoto T. Human cDNAs encoding elongation factor 1 gamma and the ribosomal protein L19. Nucleic Acids Res. 1992;20(10):2598.

37. Achilonu I, Siganunu TP, Dirr HW. Purification and characterisation of recombinant human eukaryotic elongation factor 1 gamma. Protein Expr Purif. 2014;99:7077. 38. Chi K, Jones DV, Frazier ML. Expression of an elongation factor 1 gamma-related sequence in adenocarcinomas of the colon. Gastroenterology. 1992;103(1):98-102. 39. Ender B, Lynch P, Kim YH, Inamdar NV, Cleary KR, Frazier ML. Overexpression of an elongation factor-1 gamma-hybridizing RNA in colorectal adenomas. Mol Carcinog. 1993;7:18-20. 40. Lew Y, Jones DV, Mars WM, Evans D, Byrd D, Frazier ML. Expression of elongation factor-1 gamma-related sequence in human pancreatic cancer. Pancreas. 1992;7:144152. 41. Mimori K, Mori M, Inoue H, et al. Elongation factor 1 gamma mRNA expression in oesophageal carcinoma. Gut. 1996;38(1):66-70. 42. Van Roosbroeck K, Wlodarska I. Oncogenic anaplastic lymphoma kinase rearrangements in lymphoma. European Haematology. 2009;3:50-56. 43. Moritake H, Shimonodan H, Marutsuka K, Kamimura S, Kojima H, Nunoi H. C-MYC rearrangement may induce an aggressive phenotype in anaplastic lymphoma kinase positive anaplastic large cell lymphoma: Identification of a novel fusion gene ALO17/C-MYC. Am J Hematol. 2011; 86(1):75-78. 44. Liang X, Branchford B, Greffe B, et al. Dual ALK and MYC rearrangements leading to an aggressive variant of anaplastic large cell lymphoma. J Pediatr Hematol Oncol. 2013; 35(5):e209-e213. 45. Monaco S, Tsao L, Murty VV, et al. Pediatric ALK+ anaplastic large cell lymphoma with t(3;8)(q26.2;q24) translocation and c-myc rearrangement terminating in a leukemic phase. Am J Hematol. 2007; 82(1):59-64. 46. LaPak KM, Burd CE. The molecular balancing act of p16(INK4a) in cancer and aging. Mol Cancer Res. 2014;12(2):167-183. 47. Boi M, Rinaldi A, Kwee I, et al. PRDM1/BLIMP1 is commonly inactivated in anaplastic large T-cell lymphoma. Blood. 2013;122(15):2683-2693. 48. Armstrong F, Duplantier MMl, Trempat P, et al. Differential effects of X-ALK fusion proteins on proliferation, transformation, and invasion properties of NIH3T3 cells. Oncogene. 2004;23(36):6071-6082.

haematologica | 2017; 102(9)


ARTICLE

Plasma Cell Disorders

The spectrum of somatic mutations in monoclonal gammopathy of undetermined significance indicates a less complex genomic landscape than that in multiple myeloma Aneta Mikulasova,1,2,3,4 Christopher P. Wardell,1 Alexander Murison,5 Eileen M. Boyle,5 Graham H. Jackson,6 Jan Smetana,2,3 Zuzana Kufova,7,8 Ludek Pour,9 Viera Sandecka,9 Martina Almasi,10 Pavla Vsianska,10 Evzen Gregora,11 Petr Kuglik,2,3 Roman Hajek,7,8 Faith E. Davies,1 Gareth J. Morgan1 and Brian A. Walker1

EUROPEAN HEMATOLOGY ASSOCIATION

Ferrata Storti Foundation

Haematologica 2017 Volume 102(9):1617-1625

Myeloma Institute, University of Arkansas for Medical Sciences, Little Rock, AR, USA; Department of Experimental Biology, Faculty of Science, Masaryk University, Brno, Czech Republic; 3Department of Medical Genetics, University Hospital Brno, Czech Republic; 4Department of Pathological Physiology, Faculty of Medicine, Masaryk University, Brno, Czech Republic; 5Center for Myeloma Research, Division of Molecular Pathology, Institute of Cancer Research, London, UK; 6Northern Institute for Cancer Research, Newcastle University, Newcastle upon Tyne, UK; 7Faculty of Medicine, University of Ostrava, Czech Republic; 8Department of Hematooncology, University Hospital Ostrava, Czech Republic; 9Department of Internal Medicine, Hematology and Oncology, University Hospital Brno, Czech Republic; 10Department of Clinical Hematology, University Hospital Brno, Czech Republic and 11Department of Internal Medicine and Hematology, University Hospital Kralovske Vinohrady, Prague, Czech Republic 1 2

ABSTRACT

M

onoclonal gammopathy of undetermined significance is a premalignant precursor of multiple myeloma with a 1% risk of progression per year. Although targeted analyses have shown the presence of specific genetic abnormalities such as IGH translocations, RB1 deletion, 1q gain, hyperdiploidy or RAS gene mutations, little is known about the molecular mechanism of malignant transformation. We performed whole exome sequencing together with comparative genomic hybridization plus single nucleotide polymorphism array analysis in 33 flow-cytometry-separated abnormal plasma cell samples from patients with monoclonal gammopathy of undetermined significance to describe somatic gene mutations and chromosome changes at the genome-wide level. Non-synonymous mutations and copy-number alterations were present in 97.0% and in 60.6% of cases, respectively. Importantly, the number of somatic mutations was significantly lower in monoclonal gammopathy of undetermined significance than in myeloma (P<10-4) and we identified six genes that were significantly mutated in myeloma (KRAS, NRAS, DIS3, HIST1H1E, EGR1 and LTB) within the monoclonal gammopathy of undetermined significance dataset. We also found a positive correlation with increasing chromosome changes and somatic gene mutations. IGH translocations, comprising t(4;14), t(11;14), t(14;16) and t(14;20), were present in 27.3% of cases and in a similar frequency to myeloma, consistent with the primary lesion hypothesis. MYC translocations and TP53 deletions or mutations were not detected in samples from patients with monoclonal gammopathy of undetermined significance, indicating that they may be drivers of progression to myeloma. Data from this study show that monoclonal gammopathy of undetermined significance is genetically similar to myeloma, however overall genetic abnormalities are present at significantly lower levels in monoclonal gammopathy of undetermined significant than in myeloma.

haematologica | 2017; 102(9)

Correspondence: bwalker2@uams.edu

Received: January 10, 2017. Accepted: May 16, 2017. Pre-published: May 26, 2017. doi:10.3324/haematol.2017.163766 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/102/9/1617 Š2017 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

1617


A. Mikulasova et al.

Introduction Monoclonal gammopathy of undetermined significance (MGUS) is one of the most common pre-malignant conditions and affects 3.2% of people over 50 years old, 5.3% over 70 and 7.5% over the age of 85 years.1,2 MGUS is characterized by a serum monoclonal protein <30 g/L, <10% plasma cells in the bone marrow, and the absence of end-organ damage (CRAB: hypercalcemia, renal insufficiency, anemia, or bone lesions). The MGUS progresses to multiple myeloma (MM) in approximately 1% of patients per year.3 Risk stratification models have been proposed to assess risk of transformation using flowcytometry and serum free light chain.4-6 Advances in molecular genetics have opened up the possibility of identifying genetic events involved in malignant transformation. Previous studies in MGUS have shown that interphase fluorescence in situ hybridization can detect known myeloma-specific chromosomal abnormalities in MGUS patients. These chromosomal abnormalities include IGH (14q32) translocations, RB1 (13q14) deletion, 1q gain and hyperdiploidy. These abnormalities are present at lower frequencies in MGUS than in myeloma.7-9 The potential prognostic significance of these abnormalities in relation to the progression of MGUS has not been specified.10 It has been shown that the incidence of these variants increases from MGUS through smoldering MM (SMM) to MM.11 Single nucleotide polymorphism (SNP) arrays have also been used to detect copy-number alterations (CNA) and these also increase in frequency from MGUS (5/patient) through SMM (7.5/patient) to MM (12/patient).12 Activation of proto-oncogenes, such as activation of KRAS, NRAS, MYC and BRAF, has been less frequently described in MGUS than in myeloma.13-15 In a previous study,16 we described the exome mutation profile of four MGUS patients which suggested that genomic complexity increased from MGUS, through SMM, MM and plasma cell leukemia. To understand the molecular pathogenesis of MGUS and the role of genetic events in relation to malignant transformation, more genome-wide studies in MGUS datasets are required and in this study we performed a comprehensive analysis of flow-sorted abnormal plasma cells from 33 MGUS patients using whole exome sequencing together with comparative genomic hybridization (CGH)+SNP arrays.

Methods Patients’ samples Overall, 33 MGUS patients from centers in the Czech Republic (Brno, Prague and Ostrava) were included in this study, which was approved by the University Hospital Brno Ethical Committee, after giving informed consent (Online Supplementary Table S1). Bone marrow plasma cells were isolated from the mononuclear cell fraction with a FACSAria (BD Biosciences, San Jose, CA, USA) using CD138-PE, CD19-APC and CD56-FITC antibodies (Beckman Coulter, Brea, CA, USA or Exbio, Prague, Czech Republic) to obtain a phenotypically abnormal plasma cell population (CD138+CD19-CD56+/-)17 with a median purity of 99.0% (range, 93.6–99.9%). The flow-cytometry data before and after plasma cell sorting are presented in Online Supplementary Figures S1 and S2. The median number of sorted cells was 57×103 (range, 15×103 – 480×103). Tumor DNA was isolated using a Gentra 1618

Puregene Kit, amplified using the REPLI-g Midi Kit and purified using the QIAamp DNA Mini Kit (all from Qiagen, Hilden, Germany). Previous studies had demonstrated the suitability of whole-genome-amplified DNA for array-CGH18-20 as well as nextgeneration sequencing21-23 analysis. Control DNA was obtained from peripheral white blood cells using a MagNA Pure System (Hoffmann-La Roche, Basel, Switzerland). The quality and quantity of DNA were measured by Qubit Fluorometer, Pico-green (both from Thermo Fisher Scientific, Waltham, MA, USA) and/or 2200 Tapestation (Agilent Technologies, Santa Clara, CA, USA).

Comparative genomic hybridization and single nucleotide polymorphism arrays As previously described, 2–3 µg of whole-genome amplified tumor DNA and Agilent Euro Male/Female (Agilent Technologies) as control DNA were fragmented by AluI and RsaI (both from Promega, Madison, WI, USA) restriction enzymes and fluorescently labeled with the BioPrime Total for Agilent aCGH Kit (Thermo Fisher Scientific) or treated with the SureTag Complete DNA Labeling Kit (Agilent Technologies).24 After purification of labeled DNA, tumor and control DNA samples were combined with COT Human DNA (Hoffmann-La Roche) and hybridization mix (Oligo aCGH Hybridization Kit, Agilent Technologies), and co-hybridized to SurePrint G3 CGH+SNP, 4x180K (Agilent Technologies) arrays. After hybridization and washing, DNA microarrays were scanned using a Microarray Scanner (Agilent Technologies) with 3 mm resolution. Feature Extraction Software 12.0.2.2 (Agilent Technologies) was used for data extraction and quality control evaluation. Genomic Workbench 7.0.4.0 (Agilent Technologies) was used for CNA calling by the ADM-2 algorithm with the following settings: ≥100 kb size, ≥0.2 fold change of log2 ratio, ≥5 consecutive probes. CNA were manually curated and the default Database of Genomic Variants (http://www.openhelix.com) for hg19 was used to eliminate common human population copy-number variants. The array data supporting the results of this article are available at Gene Expression Omnibus (GEO), National Center for Biotechnology Information (NCBI) under the accession number GSE77979.

Exome sequencing A previously published protocol was used for exome sequencing.25 A total of 200 ng DNA from peripheral blood and 3 mg whole-genome amplified tumor DNA were fragmented by the Covaris E-Series. Fragmented DNA was end-repaired, A-tailed and adaptors ligated by the NEBNext DNA library prep master mix set for Illumina (New England Biolabs, Ipswich, MA, USA). Modified DNA was amplified by NEBNext High-fidelity polymerase chain reaction (PCR) master mix using either eight or four PCR cycles in the case of control and tumor DNA, respectively. A total of 750 ng amplified DNA was hybridized to customdesigned RNA baits overnight (SureSelect Human All Exon V5, Agilent Technologies; enriched for IGH, IGK, IGL and MYC region capture). Captured DNA was indexed and amplified by Herculase II fusion DNA polymerase (Agilent Technologies) for eight PCR cycles. Samples were sequenced using a HiSeq 2000 (Illumina, San Diego, CA, USA) using four pooled samples per lane and 76-bp paired-ends reads. Additional information about data quality metrics and processing, somatic mutation calling and non-negative matrix factorization is given in the Online Supplementary Methods. Sequence read data for this study have been submitted to the European Genome-Phenome Archive (EGA) under accession number EGAS00001001658. The findings from the 33 MGUS patients in this study were compared to data from a cohort of 463 newly diagnosed MM (NDMM) patients from a previous study (EGAS00001001147).26 haematologica | 2017; 102(9)


Somatic mutations in monoclonal gammopathies

A

C

B

D

Figure 1. Basic sequencing characteristics of the study. (A) Number of specific variants by nucleotide changes. (B) Comparison of specific variants by proportion of nucleotide changes in MGUS and NDMM. (C) Number of variants by their effect on transcription. (D) Comparison of proportion of variants by type in MGUS and NDMM.

Basic statistical analysis Data were analyzed using Statistica 12 software (StatSoft, Prague, Czech Republic) and MedCalc 14.8.1 software (MedCalc Software, Ostend, Belgium). Statistical tests were used as follows: the Fisher exact test for categorical data, the Mann-Whitney U test for continuous variables and Pearson correlation. P values ≤0.05 were considered statistically significant.

Results Fewer copy-number changes are found in monoclonal gammopathy of undetermined significance than in multiple myeloma Using high-density oligonucleotide CGH+SNP arrays, CNA were detected in 60.6% (20/33) of MGUS patients in comparison to 100% of MM patients, which were described in our previous study.27 A summary plot of CNA in the 33 MGUS cases is given in Online Supplementary Figure S3 and frequencies of CNA at the chromosome-arm level are listed in Online Supplementary Table S2. We found 123 CNA (42 losses and 81 gains). Although CGH+SNP arrays with higher resolution were used in this study, the median number of CNA per patient was only two (range, 0–15), fewer than the 16 (range, 1–52) found in the MM dataset.27 Numerical CNA were present in 54.5% (18/33) of cases; whole chromosome losses and whole chromosome gains were found in 39.4% (13/33) and 30.3% (10/33) of cases, respectively. Analogous to MM, we identified two distinct subgroups within MGUS: non-hyperdiploid and hyperhaematologica | 2017; 102(9)

diploid. Non-hyperdiploidy was present in 72.7% (24/33) of patients and we distinguished subtypes within this subgroup as hypodiploid, pseudodiploid and diploid in 18.2% (6/33), 9.1% (3/33) and 45.5% (15/33) of cases, respectively. The most frequently lost chromosomes were 13 (27.3%, 9/33), X (18.2%, 6/33) and Y (12.1%, 4/33). Hyperdiploidy was detected in 27.3% (9/33) of cases and the most frequently gained chromosomes were 9 (27.3%, 9/33), 19 (27.3%, 9/33) and 3 (18.2%, 6/33). The median number of chromosomes in hyperdiploid patients was 52 (range, 48–55). Interestingly, 88.9% (8/9) of hyperdiploid patients also carried structural abnormalities in comparison to 29.2% (7/24) of non-hyperdiploid patients (P=4.39×10-3). On the other hand, 54.2% (13/24) of nonhyperdiploid patients had no CNA detected by CGH+SNP arrays. Structural abnormalities were seen in 45.5% (15/33) of MGUS samples. These related to changes in complete chromosome arms in 30.3% (10/33) of patients, mostly 1q gain (27.3%, 9/33) and 16q loss (6.1%, 2/33). Smaller interstitial changes were seen in 30.3% (10/33) of patients with a median of 0 (range, 0–4) changes per patient with a median size of 6.6 Mb (range, 0.1–88.8 Mb). We distinguished both interstitial and telomeric changes in 27.3% (9/33) and 15.2% (5/33) of patients, respectively. Recurrent deletions were detected at 1p and 6q, both in 6.1% (2/33) of samples, and at 14q in 12.1% (4/33) of patients. Only one case of homozygous deletion was found at 21q22.13, which did not include any known tumor-associated genes (Online Supplementary Figure S4). No deletions of 17p, a poor prognostic marker in MM and the location of TP53, were detected. 1619


A. Mikulasova et al.

MYC translocations are not detected in monoclonal gammopathy of undetermined significance

Table 1. Frequency of IGH translocations in monoclonal gammopathy of undetermined significance compared to newly diagnosed multiple myeloma.

The exome capture was enriched for the IGH (14q32), IGK (2p12), IGL (22q11.2) and MYC (8q24.21) loci, enabling analysis of the most frequent chromosomal translocations in MM. We identified IGH translocations in 27.3% (9/33), consisting of t(11;14) in 12.1% (4/33), t(4;14) in 9.1% (3/33), t(14;16) in 3.0% (1/33) and t(14;20) in 3.0% (1/33) (Table 1). We defined the chromosome breakpoints on chromosomes 4, 11, 14, 16 and 20 (Online Supplementary Figure S5), and the findings did not differ from those in MM.28 All nine cases with an IGH translocation were non-hyperdiploid. Two males with a t(11;14) as well as one male with a t(14;20) did not have either numerical or structural CNA and were diploid. Similarly, two females with a t(11;14) only had loss of the X chromosome and were, therefore, pseudodiploid. All three MGUS cases with a t(4;14) had similar profiles with loss of chromosome 13 and loss of one gonosome (2 males with loss of Y and 1 female with loss of X) and were thus also hypodiploid (43, 44 and 44 chromosomes in a total). Furthermore, 66.7% (2/3) of those with t(4;14) also had 1q gain. The patient with a t(14;16) showed similarity to three cases of t(4;14) with 1q gain and 13q loss. In this cohort of 33 MGUS patients we did not find any translocations involving MYC, even though they were detected in 18.4% of NDMM using the same assay.29

Translocation

There are fewer mutations in monoclonal gammopathy of undetermined significance than in multiple myeloma Acquired single nucleotide variants (SNV) were present in all (33/33) MGUS patients with a median of 89 (range, 9–315) per patient, most frequently as transition rather than transversion mutations, similarly to NDMM (Figure 1A,B). Exonic mutations and indels were found in a total of 857 genes, with 70.4% (603/857) of these being nonsynonymous SNV (NS-SNV) (Figure 1C,D). These mutations were present at a significantly lower level in MGUS than in NDMM (P<10-4) (Figure 2; Online Supplementary Table S3). However, 73.5% (443/603) of genes affected by NS-SNV intersected with genes mutated in NDMM. NSSNV were present in 97.0% (32/33) of cases with a median of 19 (range, 0–70) NS-SNV per patient. We did not find any significantly mutated genes, but overall 35 genes were recurrently mutated and only three genes were mutated in more than two cases: KLHL6 (missense mutations p.L90V p.L71Q and a c.-20T>A mutation in the translation start site, a gene mutated in 13 patients in NDMM), NPIPL2 (3 cases with missense mutation p.H211R) and AKAP9 (missense mutations p.S3313N, p.N2792S and p.R1973T; mutated in 6 NDMM). In five MGUS cases we identified SNV in six genes which were found to be significantly mutated in NDMM, including KRAS (n=2), HIST1H1E (n=2) and NRAS, DIS3, EGR1, LTB (all n=1) (Online Supplementary Table S4). When a mutation was present in one of these genes the variant allele frequency was not significantly different in MGUS compared to NDMM (Online Supplementary Figure S6), but was often lower. The only example of variant allele frequency being equivalent in MGUS and NDMM was for HIST1H1E, which was clonal, which may indicate that it is a key driver. 1620

t(4;14) t(6;14) t(11;14) t(14;16) t(14;20)

MGUS (n = 33) Cases (%)

NDMM (n = 463) Cases (%)

P

3 (9.1%) 0 (0.0%) 4 (12.1%) 1 (3.0%) 1 (3.0%)

59 (12.7%) 5 (1.1%) 86 (18.6%) 17 (3.7%) 4 (0.9%)

0.79 1.00 0.48 1.00 0.29

We found one t(11;14) MGUS patient with two mutations in CCND1 (p.K50T, p.E51D), which are associated with a negative impact on survival in patients with MM. We did not find any mutations in TP53, ATM, ATR or ZFHX4, which have been identified as unfavorable factors for patients’ survival and are involved in the DNA repair pathway. We tested for the presence of specific mutational signatures,30 which we have previously shown to be related to the pathological activity of specific cytidine deaminases of the APOBEC family.29 The APOBEC mutational signature was not found in this cohort of 33 MGUS patients, even among those with a t(14;16) or t(14;20), possibly suggesting that APOBEC activity does not drive disease progression in the MAF subgroup at the MGUS stage and that it is, instead, acquired later in the development of MM (Online Supplementary Figure S7). With regards to the APOBEC signature, the frequency of mutations was higher in t(14;16) MM than in other subgroups.29 However, as we had only one t(14;16) case in this series we could not conclusively show a higher mutation rate in the MGUS disease stage (Online Supplementary Figure S8).

Copy-number alterations are associated with increased mutation rate Some associations between mutations and structural changes in NDMM have been described.26 In MGUS we identified a patient with t(11;14) with two CCND1 (p.K50T, p.E51D) mutations, a case with a DIS3 (p.D488N) mutation and 13q loss and a case with an EGR1 mutation (p.M29L) with hyperdiploidy. Although, DIS3 mutations are associated with t(4;14) or t(14;16) in NDMM, the MGUS case with a DIS3 mutation did not have an IGH translocation. An association between KRAS mutations and t(11;14) has previously been documented, but neither of the two MGUS patients with KRAS (p.Q61L and p.A146T) mutations had a t(11;14); one had no IGH translocation and the other had t(14;20). We also identified a patient with both a KRAS (p.Q61L) and an NRAS (p.G13R) mutation which, although not mutually exclusive, are negatively correlated in NDMM. This patient was also hyperdiploid, which has a positive correlation with NRAS mutations in NDMM, and did not have deletion of 13q, which is negatively correlated with NRAS mutations in NDMM. The presence of more than one Ras pathway mutation in MM is associated with intraclonal heterogeneity, where the Ras mutations are present in different subclones. Here, the presence of two Ras pathway mutations indicates that heterogeneity can occur early in the disease process. Associations of spehaematologica | 2017; 102(9)


Somatic mutations in monoclonal gammopathies

cific SNV, CNA and clinical parameters are shown in Online Supplementary Figure S9. Interestingly, MGUS patients with CNA and/or IGH translocations (n=23) had significantly higher numbers of total SNV (P=8.17×10-5), exonic SNV (P=1.43×10-4), NS-SNV (P=1.82×10-3) and synonymous SNV (S-SNV) (P=3.75×104 ) in comparison to MGUS patients without any of these changes (n=10) (Table 2). We also found a positive correlation of increasing number of SNV and chromosomal abnormalities (Figure 3).

Risk stratification of patients with monoclonal gammopathy of undetermined significance Using a risk-stratification model,4 we divided 32 MGUS patients into low risk (n=14), intermediate-low risk (n=9) and intermediate-high risk (n=9). We found that the median number of CNA and/or IGH translocations increased from low to intermediate-low and intermediate-high risk groups: 0 (range, 0–10), 4 (range, 0–15) and 6 (range, 0–10), respectively (Table 3). Gain of 1q [present in 7.1% (1/14), 11.1% (1/9) and 66.7% (6/9) patients, in the three risk groups], as well as frequency of patients with at least one structural CNA [14.3% (2/14), 55.6% (5/9) and 77.8% (7/9), respectively], also increased with risk group. We did not find a clear increase of SNV across the risk groups (Table 3).

Presence of clonal abnormalities is associated with higher risk of progression MGUS cases were divided into six groups based on the structure of intratumor heterogeneity (Figure 4): 86.7%

(13/15) of cases with at least one clonal CNA and NS-SNV showed intermediate-low/high risk, while other groups had small proportions of cases with higher risk of progression (29.4%, 5/17, P=0.002). This fact was caused by the association of clonal alterations with non-IgG variant (50.0%, 8/16; others: 11.8%, 2/17; P=0.03) and abnormal serum kappa/lambda free light chain ratio (73.3%, 11/15; others: 23.5%, 4/17; P=0.02). Chromosome abnormalities were preceded by gene mutations as a total of 63.6% (21/33) of cases showed at least one NS-SNV with a 10% or higher proportion than any CNA present. There were no examples with CNA but without NS-SNV, and no cases with a CNA at a frequency of 10% or greater than that of any NS-SNV.

Discussion MGUS is considered a relatively benign disease, being present in 3% of the population >50 years old but without evidence of end-organ damage. However, recent evidence indicates that nearly all cases of MM are preceded by an MGUS phase.31,32 Analysis of the genomes of MGUS samples has revealed that the genetic composition in this disorder is strikingly similar to that in MM, with the presence of IGH translocations, hyperdiploidy, gain 1q, and deletion 1p. However, these abnormalities are, in general, present at lower frequencies in the MGUS population.12 These abnormalities have been characterized in MGUS using classical cytogenetics and fluorescence in situ hybridization, as well as mapping arrays to detect changes

Figure 2. Number of single nucleotide variants in 33 patients with monoclonal gammopathy of undetermined significance compared to 463 patients with newly diagnose multiple myeloma.

Table 2. Relationship between the number of single nucleotide variants and the presence of chromosomal abnormalities.

SNV category

At least one chromosomal abnormality (n = 23) Median (range)

No chromosomal abnormality (n = 10) Median (range)

P

102 (32–315) 30 (5–111) 23 (4–70) 11 (1–42)

29 (9–92) 11 (2–23) 9 (0–24) 3 (1–6)

8.17×10-5 1.43×10-4 1.82×10-3 3.75×10-4

Total SNV Exonic SNV NS-SNV S-SNV

Chromosomal abnormalities include CNA tested by CGH+SNP arrays and IGH translocations defined by exome sequencing.

haematologica | 2017; 102(9)

1621


A. Mikulasova et al.

at a higher resolution.11,12,33,34 However, exome sequencing in MGUS has only been performed in a handful of patients16 so the dataset has been too small to make meaningful conclusions. Here, we report the first comprehensive analysis of genome-wide genetic changes in 33 MGUS samples in which flow-cytometry-separated phenotypically abnormal plasma cells were analyzed, to exclude contamination by phenotypically normal plasma cells, followed by array CGH and exome sequencing. We found that the frequency of chromosomal gains and losses, including hyperdiploidy, gain 1q, and del(13q), is lower in MGUS than in MM. Hyperdiploidy is considered as a primary myeloma lesion; however it has prognostic potential in asymptomatic stages as has been shown in SMM.35 The frequency of MGUS samples with CNA is

60.6%. the minimum size of the alterations is 100 kb and there are a median of two CNA per case. These numbers are lower compared to those for cases of MM which we previously analyzed.27 Homozygous deletion affecting genes such as FAF1/CDKN2C, BIRC2/BIRC3, RB1, TRAF3/AMN and CYLD are common in MM,27,36 but they were not present in MGUS. The number of SNV in the samples was also significantly lower in MGUS than in NDMM; exonic, non-synonymous, synonymous and total SNV were all found at a lower frequency. Of the variants, there were no significantly mutated genes. However, there were variants present which were significant in our previous NDMM dataset, including KRAS, NRAS, HIST1H1E, DIS3, EGR1 and LTB.

Table 3. Number of chromosomal abnormalities and single nucleotide variants per case across the monoclonal gammopathy of undetermined significance risk groups.

Chromosomal abnormalities Total SNV Exonic SNV NS-SNV S-SNV

Low risk (n = 14) Median (range)

Int.-low risk (n = 9) Median (range)

Int.-high risk (n = 9) Median (range)

P for low vs. int.-low

P for low vs. int-high

P for int.low vs. int.-high

0 (0–10)

4 (0–15)

6 (0–10)

1.90×10-2

3.44×10-2

0.86

58.5 (13–128) 16.5 (2–32) 13 (0–29) 3.5 (1–12)

112 (44–315) 33 (5–111) 24 (4–70) 11 (1–42)

89 (9–146) 29 (3–38) 21 (1–30) 9 (2–14)

8.12×10-3 1.17×10-2 2.12×10-2 5.31×10-3

0.15 1.96×10-2 0.11 7.05×10-3

0.12 0.31 0.20 0.23

Chromosomal abnormalities include CNA tested by CGH+SNP arrays and IGH translocations defined by exome sequencing. MGUS groups are defined by a model based on risk factors as follows: non-IgG isotype of serum monoclonal protein, ≥15 g/L of serum monoclonal protein and abnormal serum kappa/lambda free light chain ratio (<0.26 or >1.65).

A

B

C

D

Figure 3. Correlation analysis of increasing number of chromosome abnormalities and single nucleotide variants per sample in 33 patients with monoclonal gammopathy of undetermined significance. Chromosome abnormalities (CHA) include CNA tested by CGH+SNP arrays and IGH translocations defined by exome sequencing. (A) CHA and total SNV. (B) CHA and exonic SNV. (C) CHA and NS-SNV. (D) CHA and S-SNV.

1622

haematologica | 2017; 102(9)


Somatic mutations in monoclonal gammopathies

A

E

C

B

D

F

Figure 4. Intratumor heterogeneity in monoclonal gammopathy of undetermined significance. Each column shows the proportion (vertical axis) of NS-SNV (red points) and CNA (blue points) in each MGUS patient. The black line represents a frequency distribution of somatic changes. The presence of myeloma-significantly mutated genes, 1q gain and chromosome 13 loss is highlighted by dark colors. Red stars mark cases with at least one NS-SNV with 10% or higher proportion than any CNA present. Patients are divided into groups by clonal features of somatic alterations: (A) Subclonal NS-SNV and no CNA. (B) Clonal/subclonal NS-SNV and no CNA. (C) Subclonal NS-SNV and subclonal CNA. (D) Clonal/subclonal NS-SNV and subclonal CNA. (E) Clonal/sublonal NS-SNV and clonal/subclonal CNA. (F) Clonal/subclonal NS-SNV and clonal CNA. Only NS-SNV with a minimum 10% proportion are displayed. One of 33 patients is not shown as no NS-SNV and CNA were detected.

Insights into the molecular timing of genetic events can be gained by analyses of known MM-specific events in MGUS cases. NRAS mutations have been detected in MGUS previously, but KRAS mutations have not previously been found in MGUS and were implicated in the transition from MGUS to MM.15 Previously, in a limited number of MGUS samples (n=20), one NRAS mutation and no KRAS mutations were detected, whereas in our 33 MGUS cases we found one patient with a clonal KRAS mutation and one with both subclonal KRAS and NRAS mutations. Neither of these two patients with RAS mutations had progressed to MM after a follow-up of 50 and 67 months. Our results confirmed that the frequency of RAS mutations in MGUS is significantly lower than in MM, but also showed that activation of this pathway does not necessarily mark the onset of disease progression. Interestingly, the presence of both a KRAS and NRAS mutation in one patient indicates early diversification and heterogeneity in this pre-malignant condition. Another candidate for association with disease progression is del(17p) and/or mutation of the TP53 gene.37 We detected neither in this MGUS dataset, but they account for up to 11% of mutations in NDMM. Mutations in ATM and ATR were also not detected in MGUS but their prevalence in NDMM is low (<2% each) so would not have been expected in this dataset. MYC translocations were not detected in MGUS by our targeted capture of 2 Mb surrounding MYC, which has detected translocations in haematologica | 2017; 102(9)

18% of NDMM.26 This change at the MYC locus is consistent with data from gene expression analyses showing that MYC over-expression occurs in two-thirds of MM cases, but uncommonly in MGUS.38-41 RAS mutations, TP53 alterations and MYC translocations are all known oncogenic drivers and are likely candidates for being involved in disease progression but not initiation of the myeloma propagating cell. Moreover, del(17p)35,42 and MYC rearrangement43 have been observed as risk progression factors in SMM. We have previously identified mutations in CCND1, associated with the t(11;14), which have a negative impact on survival. In our MGUS dataset there were four t(11;14) samples, of which one had two missense mutations in CCND1. This is consistent with NDMM, given the low numbers of t(11;14) samples, perhaps implying that there is not a link with disease progression. The frequency of IGH translocations detected by exome sequencing did not differ significantly from that found in previous fluorescence in situ hybridization-based studies, however a significantly lower t(4;14) and higher t(14;20) frequency in MGUS compared to MM was not seen due to the limited number of MGUS cases.44,45 In conclusion, MGUS is better defined by the genomic abnormalities that are absent than by the ones that are present. We show that some MM-specific structural and SNV are present in MGUS, but the overall prevalence of such lesions is significantly lower than in MM. Structural 1623


A. Mikulasova et al.

changes such as gain 1q and del(1p) are present in MGUS, at relatively high frequencies, implying they may lay the ground for progression but are not key drivers of progression. Key oncogenic drivers, namely TP53 deletion and/or mutation and MYC translocations, are noticeably absent from this MGUS dataset and are, therefore, better candidates for the onset of disease progression. Here, we show that samples with clonal copy-number changes and mutations are associated with non-IgG heavy chain isotype and abnormal light chain ratio, which are known markers of high-risk MGUS. The clonal complexity of these samples (Figure 4E,F) is in stark contrast to the relative simplicity of those associated with a low risk of disease progression (Figure 4A-D). Follow up of these cases over time and analysis of progressive samples from MGUS, through SMM to MM will provide interesting insights into the evolutionary mechanisms of selection for aggressive malignant clones containing key driver events. These observations are consistent with a Darwinian model of myeloma evolution whereby genomic abnor-

References 1. Kyle RA, Rajkumar SV. Monoclonal gammopathy of undetermined significance and smouldering multiple myeloma: emphasis on risk factors for progression. Br J Haematol. 2007;139(5):730-743. 2. Kyle RA, Therneau TM, Rajkumar SV, et al. Prevalence of monoclonal gammopathy of undetermined significance. N Engl J Med. 2006;354(13):1362-1369. 3. Kyle RA, Therneau TM, Rajkumar SV, et al. A long-term study of prognosis in monoclonal gammopathy of undetermined significance. N Engl J Med. 2002;346(8):564569. 4. Rajkumar SV, Kyle RA, Therneau TM, et al. Serum free light chain ratio is an independent risk factor for progression in monoclonal gammopathy of undetermined significance. Blood. 2005;106(3):812-817. 5. Perez-Persona E, Mateo G, Garcia-Sanz R, et al. Risk of progression in smouldering myeloma and monoclonal gammopathies of unknown significance: comparative analysis of the evolution of monoclonal component and multiparameter flow cytometry of bone marrow plasma cells. Br J Haematol. 2010;148(1):110-114. 6. Perez-Persona E, Vidriales MB, Mateo G, et al. New criteria to identify risk of progression in monoclonal gammopathy of uncertain significance and smoldering multiple myeloma based on multiparameter flow cytometry analysis of bone marrow plasma cells. Blood. 2007;110(7):2586-2592. 7. Avet-Loiseau H, Facon T, Daviet A, et al. 14q32 translocations and monosomy 13 observed in monoclonal gammopathy of undetermined significance delineate a multistep process for the oncogenesis of multiple myeloma. Intergroupe Francophone du Myelome. Cancer Res. 1999;59(18):45464550. 8. Avet-Loiseau H, Li JY, Morineau N, et al. Monosomy 13 is associated with the transition of monoclonal gammopathy of undetermined significance to multiple myeloma. Intergroupe Francophone du Myelome. Blood. 1999;94(8):2583-2589.

1624

malities are accumulated as disease progresses. We hypothesize that transformation from MGUS to MM is due to the acquisition of critical driver gene variants which alters the behavior of a myeloma propagating cell, which is accompanied by a wave of clonal expansion and clonal dominance resulting in crucial differences in clinical behavior. Funding This work was supported by a Myeloma UK program grant, Cancer Research UK CTAAC sample collection grants (A12136 and A17761) and a Cancer Research UK Biomarkers and Imaging Discovery and Development grant (A14261) as well as funds from the National Institute of Health Biomedical Research Centre at the Royal Marsden Hospital. This study was partly funded by a P01 grant from the National Institutes of Health grant number CA055819. The study was also supported by the Ministry of Health of the Czech Republic (NT13492, 1529667A) and the Ministry of Education, Youth and Sports of the Czech Republic (IRP201550).

9. Chng WJ, Van Wier SA, Ahmann GJ, et al. A validated FISH trisomy index demonstrates the hyperdiploid and nonhyperdiploid dichotomy in MGUS. Blood. 2005;106(6):2156-2161. 10. van de Donk NW, Palumbo A, Johnsen HE, et al. The clinical relevance and management of monoclonal gammopathy of undetermined significance and related disorders: recommendations from the European Myeloma Network. Haematologica. 2014; 99(6):984-996. 11. Lopez-Corral L, Gutierrez NC, Vidriales MB, et al. The progression from MGUS to smoldering myeloma and eventually to multiple myeloma involves a clonal expansion of genetically abnormal plasma cells. Clin Cancer Res. 2011;17(7):1692-1700. 12. Lopez-Corral L, Sarasquete ME, Bea S, et al. SNP-based mapping arrays reveal high genomic complexity in monoclonal gammopathies, from MGUS to myeloma status. Leukemia. 2012;26(12):2521-2529. 13. Chesi M, Robbiani DF, Sebag M, et al. AIDdependent activation of a MYC transgene induces multiple myeloma in a conditional mouse model of post-germinal center malignancies. Cancer Cell. 2008;13(2):167-180. 14. Chng WJ, Gonzalez-Paz N, Price-Troska T, et al. Clinical and biological significance of RAS mutations in multiple myeloma. Leukemia. 2008;22(12):2280-2284. 15. Rasmussen T, Kuehl M, Lodahl M, Johnsen HE, Dahl IM. Possible roles for activating RAS mutations in the MGUS to MM transition and in the intramedullary to extramedullary transition in some plasma cell tumors. Blood. 2005;105(1):317-323. 16. Walker BA, Wardell CP, Melchor L, et al. Intraclonal heterogeneity is a critical early event in the development of myeloma and precedes the development of clinical symptoms. Leukemia. 2014;28(2):384-390. 17. Bataille R, Jego G, Robillard N, et al. The phenotype of normal, reactive and malignant plasma cells. Identification of "many and multiple myelomas" and of new targets for myeloma therapy. Haematologica. 2006;91(9):1234-1240. 18. Lage JM, Leamon JH, Pejovic T, et al.

19.

20.

21.

22.

23.

24.

25.

26.

27.

Whole genome analysis of genetic alterations in small DNA samples using hyperbranched strand displacement amplification and array-CGH. Genome Res. 2003;13(2):294-307. Corneveaux JJ, Kruer MC, Hu-Lince D, et al. SNP-based chromosomal copy number ascertainment following multiple displacement whole-genome amplification. Biotechniques. 2007;42(1):77-83. Pugh TJ, Delaney AD, Farnoud N, et al. Impact of whole genome amplification on analysis of copy number variants. Nucleic Acids Res. 2008;36(13):e80. Pinard R, de Winter A, Sarkis GJ, et al. Assessment of whole genome amplification-induced bias through high-throughput, massively parallel whole genome sequencing. BMC Genomics. 2006;7:216. Rykalina VN, Shadrin AA, Amstislavskiy VS, Rogaev EI, Lehrach H, Borodina TA. Exome sequencing from nanogram amounts of starting DNA: comparing three approaches. PLoS One. 2014;9(7):e101154. Huang L, Ma F, Chapman A, Lu S, Xie XS. Single-cell whole-genome amplification and sequencing: methodology and applications. Annu Rev Genomics Hum Genet. 2015;16:79-102. Mikulasova A, Smetana J, Wayhelova M, et al. Genome-wide profiling of copy-number alteration in monoclonal gammopathy of undetermined significance. Eur J Haematol. 2016;97(6):568-575. Kozarewa I, Rosa-Rosa JM, Wardell CP, et al. A modified method for whole exome resequencing from minimal amounts of starting DNA. PLoS One. 2012;7(3):e32617. Walker BA, Boyle EM, Wardell CP, et al. Mutational spectrum, copy number changes, and outcome: results of a sequencing study of patients with newly diagnosed myeloma. J Clin Oncol. 2015;33(33):3911-3920. Smetana J, Frohlich J, Zaoralova R, et al. Genome-wide screening of cytogenetic abnormalities in multiple myeloma patients using array-CGH technique: a Czech multicenter experience. Biomed Res Int. 2014;2014:209670.

haematologica | 2017; 102(9)


Somatic mutations in monoclonal gammopathies

28. Walker BA, Wardell CP, Johnson DC, et al. Characterization of IGH locus breakpoints in multiple myeloma indicates a subset of translocations appear to occur in pregerminal center B cells. Blood. 2013;121(17): 3413-3419. 29. Walker BA, Wardell CP, Murison A, et al. APOBEC family mutational signatures are associated with poor prognosis translocations in multiple myeloma. Nat Commun. 2015;6:6997. 30. Alexandrov LB, Nik-Zainal S, Wedge DC, et al. Signatures of mutational processes in human cancer. Nature. 2013;500(7463):415421. 31. Landgren O, Kyle RA, Pfeiffer RM, et al. Monoclonal gammopathy of undetermined significance (MGUS) consistently precedes multiple myeloma: a prospective study. Blood. 2009;113(22):5412-5417. 32. Weiss BM, Abadie J, Verma P, Howard RS, Kuehl WM. A monoclonal gammopathy precedes multiple myeloma in most patients. Blood. 2009;113(22):5418-5422. 33. Fonseca R, Bailey RJ, Ahmann GJ, et al. Genomic abnormalities in monoclonal gammopathy of undetermined significance. Blood. 2002;100(4):1417-1424. 34. Nilsson T, Lenhoff S, Rylander L, et al. High frequencies of chromosomal aberrations in multiple myeloma and monoclonal gam-

haematologica | 2017; 102(9)

35.

36.

37.

38.

39.

40.

mopathy of undetermined significance in direct chromosome preparation. Br J Haematol. 2004;126(4):487-494. Neben K, Jauch A, Hielscher T, et al. Progression in smoldering myeloma is independently determined by the chromosomal abnormalities del(17p), t(4;14), gain 1q, hyperdiploidy, and tumor load. J Clin Oncol. 2013;31(34):4325-4332. Walker BA, Leone PE, Chiecchio L, et al. A compendium of myeloma-associated chromosomal copy number abnormalities and their prognostic value. Blood. 2010;116 (15):e56-65. Ackermann J, Meidlinger P, Zojer N, et al. Absence of p53 deletions in bone marrow plasma cells of patients with monoclonal gammopathy of undetermined significance. Br J Haematol. 1998;103(4):1161-1163. Chng WJ, Huang GF, Chung TH, et al. Clinical and biological implications of MYC activation: a common difference between MGUS and newly diagnosed multiple myeloma. Leukemia. 2011;25(6):1026-1035. Shou Y, Martelli ML, Gabrea A, et al. Diverse karyotypic abnormalities of the cmyc locus associated with c-myc dysregulation and tumor progression in multiple myeloma. Proc Natl Acad Sci USA. 2000;97(1):228-233. Avet-Loiseau H, Gerson F, Magrangeas F, et

41.

42.

43.

44.

45.

al. Rearrangements of the c-myc oncogene are present in 15% of primary human multiple myeloma tumors. Blood. 2001;98(10): 3082-3086. Xiao R, Cerny J, Devitt K, et al. MYC protein expression is detected in plasma cell myeloma but not in monoclonal gammopathy of undetermined significance (MGUS). Am J Surg Pathol. 2014;38(6):776-783. Rajkumar SV, Gupta V, Fonseca R, et al. Impact of primary molecular cytogenetic abnormalities and risk of progression in smoldering multiple myeloma. Leukemia. 2013;27(8):1738-1744. Chiecchio L, Dagrada GP, Protheroe RK, et al. Loss of 1p and rearrangement of MYC are associated with progression of smouldering myeloma to myeloma: sequential analysis of a single case. Haematologica. 2009;94(7):1024-1028. Ross FM, Chiecchio L, Dagrada G, et al. The t(14;20) is a poor prognostic factor in myeloma but is associated with long-term stable disease in monoclonal gammopathies of undetermined significance. Haematologica. 2010;95(7):1221-1225. Chiecchio L, Dagrada GP, Ibrahim AH, et al. Timing of acquisition of deletion 13 in plasma cell dyscrasias is dependent on genetic context. Haematologica. 2009;94 (12):1708-1713.

1625


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

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

Ancient Greek

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

Scientific Latin

haematologicus (adjective) = related to blood

Scientific Latin

haematologica (adjective, plural and neuter, used as a noun) = hematological subjects

Modern English

The oldest hematology journal, publishing the newest research results. 2016 JCR impact factor = 7.702

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, Volume 102, issue 9  
Read more
Read more
Similar to
Popular now
Just for you