PS1339 UNSUPERVISED HIERARCHICAL CLUSTERING OF PRE‐TREATMENT IMMUNOPHENOTYPIC FEATURES IDENTIFIES MDS PATIENTS AT HIGH RISK OF DISEASE PROGRESSION AFTER ALLOGENEIC TRANSPLANTATION

Abstract

Background:Myelodysplastic syndrome (MDS) and myelodysplatic/myeloproliferative neoplasm (MDS/MPN) belong to a heterogeneous group of bone marrow (BM) diseases. Immunophenotyping is part of the basic diagnostic workup, but the biological implications and clinical consequences of distinct surface antigen expression patterns in MDS have not been reported.Aims:To assess pre‐treatment immunophenotypic features in MDS patients (pts) and to identify associated disease characteristics and outcome.Methods:We analyzed 175 pts with MDS (n = 155) or MDS/MPN (n = 20) who received an allogeneic cell transplantation (HCT, median age 59.2 [range 18.5–74.4] years) using peripheral blood (n = 174) or cord blood (n = 1) stem cells after myeloablative (n = 45), reduced‐intensity (n = 58) or non‐myeloablative (n = 72) conditioning. Donors were matched related (n = 31) or unrelated (n = 96) or had at least one allelic mismatch (n = 48). Prior to HCT, 53% of pts received cytoreductive therapy (24% azacitidine, 21% intensive chemotherapy, 8% both). Pts were grouped according to IPSS‐R (0% very good, 8% good, 31% intermediate, 25% poor, 37% very poor). Flow cytometric analysis was performed centrally on pre‐treatment bone marrow mononuclear cells. Median follow up after HCT was 3.9 years.Results:We identified four pts subgroups with distinct pre‐treatment immunophenotypic features by unsupervised hierarchical clustering (Figure 1A). For pts in the first cluster (n = 12) a particularly adverse outcome after HCT was observed. Compared to all other clusters, this cluster was characterized by a higher expression of immature surface antigens (CD34, P < .001; CD117, P < .001) and a higher expression of the leukemic stem cell population CD34+/CD38‐ (P = .001), of myeloid antigens (CD11b, P = .05; CD13 P < .01; CD33, P < .01; CD15, P < .01; CD65, P = .05 and CD64, P < .01), of pan‐leukocyte antigens (CD45, P = .03) and a lower expression of T‐cell antigens (CD2, P < .01 & CD7, P < .01). Pts in cluster 1 were more likely to present with an excess of blasts (P = .03) and to harbor a complex karyotype (KT, P = .02). Although pts in cluster 1 also more often received cytoreductive treatment prior to HCT (P = .04) they had a higher cumulative incidence of relapse/progression (CIR, P < .01, Figure 1B) including a higher incidence of secondary acute myeloid leukemia (CIsAML, P < .01, Figure 1C). This adverse prognostic impact was also seen in multivariate analyses for CIR after adjustment for age at HCT and IPSS‐R (Hazard ratio [HR] 3.2, 95% Confidence interval [CI] 1.1–9.6, P = .04) and for CIsAML after adjustment for excess of blasts and blast count prior to HCT (HR 3.2, 95% CI 1.0–9.6, P = .04). Cluster 1 immunophenotypic features also remained predictive for higher CIR and CIsAML in bivariate analyses including the presence of a monosomal (HR 2.9, P = .04 & HR 3.4, P = .02, respectively) or complex KT (HR 2.6, P = .08 & HR 3.0, P = .04, respectively).Summary/Conclusion:Assessment of pre‐treatment surface antigen expression patterns allows for identifying a subgroup of MDS pts with adverse clinical outcome after HCT irrespective of clinical prognosticators or sensitivity to pre‐HCT treatment approaches. This subgroup of pts requires post‐HCT strategies to maintain disease control.image