Leukemia (2013) 27, 879–888 & 2013 Macmillan Publishers Limited All rights reserved 0887-6924/13 www.nature.com/leu ORIGINAL ARTICLE Monosomal karyotype predicts poor survival after allogeneic stem cell transplantation in chromosome 7 abnormal myelodysplastic syndrome and secondary acute myeloid leukemia M van Gelder1, LC de Wreede2, J Schetelig3, A van Biezen2, L Volin4, J Maertens5, M Robin6, E Petersen7, T de Witte8 and N Kro¨ ger9 on behalf of the EBMT Chronic Malignancies Working Party Treatment algorithms for poor cytogenetic-risk myelodysplastic syndrome (MDS), defined by chromosome 7 abnormalities or complex karyotype (CK), include allogeneic stem cell transplantation (alloSCT). We studied outcome of alloSCT in chromosome 7 abnormal MDS patients as this data are scarce in literature. We specifically focused on the impact of the extra presence of CK and monosomal karyotype (MK). The European Group for Blood and Marrow Transplantation database contained data on 277 adult MDS patients with a chromosome 7 abnormality treated with alloSCT. Median age at alloSCT was 45 years. Median follow-up of patients alive was 5 years. Five-year progression-free survival (PFS) and overall survival (OS) were 22% and 28%, respectively. In multivariate analysis, statistically significant predictors for worse PFS were higher MDS stages treated, but not in complete remission (CR) (hazards ratio (HR) 1.7), and the presence of CK (HR 1.5) or MK (HR 1.8). Negative predictive factors for OS were higher MDS stages treated, but not in CR (HR 1.8), and the presence of CK (HR 1.6) or MK (HR 1.7). By means of the cross-validated log partial likelihood, MK showed to have a better predictive value than CK. The results are relevant when considering alloSCT for higher-stage MDS patients having MK including a chromosome 7 abnormality. Leukemia (2013) 27, 879–888; doi:10.1038/leu.2012.297 Keywords: chromosome 7 abnormality; myelodysplastic syndrome; secondary acute myeloid leukemia; allogeneic stem cell transplantation; complex karyotype; monosomal karyotype INTRODUCTION patients with poor karyotypic risk in these studies is very low, Myelodysplastic syndrome (MDS) results from ineffective hema- which prevents to draw firm conclusions. Only one prospective topoiesis causing one or more cytopenias. The International trial addressed the question whether outcome of poor-risk MDS Prognostic Scoring System consists of several factors that together patients treated with intensive chemotherapy followed by alloSCT predict the speed of transformation to secondary acute myeloid was superior to consolidation with autologous SCT in a donor leukemia (sAML) and death. The International Prognostic Scoring versus no donor comparison.6,23 No difference in OS was found in System factors include the percentage of blasts in bone marrow the first analysis of 117 evaluable patients,6 but at longer follow- and blood, the number of cytopenias and specific karyotypic up, alloSCT predicted better progression-free survival (PFS) and abnormalities.1 Poor-risk karyotypic abnormalities are defined OS.23 Again, in this study, the number of patients with a poor-risk either as complex abnormalities (that is, X3 chromosomal karyotype was low (n ¼ 29). abnormalities, complex karyotype (CK)) or as chromosome 7 Other factors that predicted a poor outcome of MDS/sAML abnormalities, both of which carry a poor prognosis in AML as patients after alloSCT were age, time from diagnosis to alloSCT, well.2,3 Recently, monosomal karyotype (MK) was shown to be a blast percentage or remission state at alloSCT15,16,21,22,24–27 and more accurate predictor of poor outcome in AML.4 prior remission-induction therapy,28 although contradictory results The results of AML-like remission induction and consolidation have been reported.13,14,29 therapies in MDS patients with advanced stages and poor-risk To better estimate outcome of patients with cytogenetically karyotypes are disappointing. Complete remission (CR) rates are high-risk MDS/sAML, we performed a retrospective analysis on low, and because of relapses, survival (OS) is almost 0% at longer patients with a chromosome 7 abnormality treated with alloSCT follow-up.5–11 Given this poor outcome, many patients are offered using the European Group for Blood and Marrow Transplantation an allogeneic stem cell transplantation (alloSCT) because it is (EBMT) database. In addition to the role of well-known risk considered the only curative treatment. Unfortunately, many factors,30 we studied the impact of extra chromosomal instability studies show that the presence of poor-risk karyotypes adversely defined as either CK or MK in addition to the presence of a affects outcome after alloSCT.6,12–22 However, the number of chromosomal 7 abnormality. 1Division of Hematology, Department of Internal Medicine, University Hospital Maastricht, MUMC, Maastricht, The Netherlands; 2Department of Medical Statistics, Leiden University Medical Center, Leiden, The Netherlands; 3Medical Department I, University Hospital Carl Gustav Carus, Dresden, Germany; 4Department of Medicine, Helsinki University Central Hospital, Helsinki, Finland; 5Department of Hematology, University Hospital Gasthuisberg, Leuven, Belgium; 6Department of Hematology-BMT, Hopital St Louis, Paris, France; 7Department of Hematology, University Medical Centre Utrecht, Utrecht, The Netherlands; 8Department of Tumor Immunology, Radboud University Medical Center, Nijmegen, The Netherlands and 9Clinic for Stem Cell Transplantation, University Medical Center Hamburg-Eppendorf, Hamburg, Germany. Correspondence: Dr M van Gelder, Division of Hematology, Department of Internal Medicine, University Hospital Maastricht, MUMC, Postbus 5800, Maastricht 6202 AZ, The Netherlands. E-mail: [email protected] Received 6 June 2012; revised 2 October 2012; accepted 3 October 2012; accepted article preview online 16 October 2012; advance online publication, 20 November 2012 Transplant outcome in chromosome 7 abnormal MDS M van Gelder et al 880 PATIENTS AND METHODS information on all variables introduced in the Cox model (n ¼ 243 for Patient population Cox models 1 and 2, and n ¼ 247 for all other Cox models). Baseline variables that are known to have predictive power on outcome after We interrogated the EBMT registry for MDS/sAML patients with chromo- alloSCT in a large cohort of patients,30 that is, age, remission status at some 7 abnormalities and who had received an alloSCT for the first time either from related or unrelated donors. We selected only patients with alloSCT, donor type, sex match and cytomegalovirus match, were also known cytogenetics within 1 year before alloSCT, in an attempt to reduce introduced into our Cox models. We did not introduce time from diagnosis the likelihood of clonal evolution. Chromosome 7 abnormalities were to transplant in the Cox model, first because this variable was found not to detected by metaphase karyotyping, FISH or both techniques. De novo be of prognostic significance in a large EBMT cohort of MDS patients containing all cytogenetic-risk groups.30 Moreover, in MDS, this variable is AML was excluded from the analysis. Two-hundred and seventy-seven evidently correlated with a variety of other covariates like stage at patients that fulfilled these criteria were extracted from the EBMT database. diagnosis, the feasibility of adequate supportive care, age, performance and comorbidity and is therefore not informative by itself. Intensity of conditioning was not used in the Cox models because it bears no relationship to the impact of additional cytogenetic abnormalities on Definitions 34,35 31 outcome. It was also not a predictor in earlier EBMT MDS studies, and We defined disease status at alloSCT according to FAB and remission its correlation with other (partly unknown) covariates like age, status. Patients designated as RA/RARS never had a higher stage before performance, comorbidity, MDS stage and application of pretreatment alloSCT. Patients with higher MDS stages, that is, (refractory anemia makes an estimation of its net effect impossible. We omitted year of with excess blasts) RAEB(t)/sAML, at any time point before alloSCT, were transplantation30 as this variable did not reveal any significance for PFS classified into three groups according to treatment with remission- and OS in a preliminary Cox analysis with our data set, and because we induction chemotherapy and its effect: CR, not in CR, and not pre- wanted to focus on variables that can be used in clinical decision making treated. sAML is defined as AML arisen from lower-stage MDS or after for future patients. Variables indicating CK or MK, the presence of a single previous cytotoxic therapy for unrelated diseases (that is, malignant or chromosome 7 abnormality or the accompaniment of extra chromosomal autoimmune diseases) Patients were categorized as having either a abnormalities were entered as shown in the tables. We validated the monosomy 7 or another chromosome 7 abnormality. CK was defined as proportionality assumption of the model we used, to illustrate clinical X 1 the presence of 3 chromosomal abnormalities, and MK as the presence relevance using a test based on scaled Schoenfeld residuals, and found X of any autosomal monosomy accompanied by either 1 additional that it did not have to be rejected for any of the covariates. For the X 4 autosomal monosomies or 1 structural chromosomal abnormalities. purpose of plotting model-based PFS curves, a web-based OS visualization We defined the conditioning intensity as myeloablative or reduced tool