The only curative treatment approach for patients with myelodysplastic syndromes (MDS) is allogeneic hematopoietic stem cell transplantation (HSCT), but disease relapse after transplantation is a major concern. Predictors for disease outcome after HSCT are limited. However, unfavorable cytogenetic abnormalities have been shown to serve as predictors for MDS-relapse after transplantation. Similar to the data available in MDS-patients not undergoing HSCT (Schanz et al. J Clin Oncol 2012), there is evidence that the novel 5-group cytogenetic classification has a better predictive value for outcome after HSCT than standard IPSS cytogenetics (Deeg et al. Blood 2012).

The aim of this large multicentric, international study was to retrospectively determine the impact of the new 5-group cytogenetic MDS classification on outcome after HSCT.

Patients and Methods

Patients were selected from the EBMT database who had received HSCT for the treatment of MDS between 1982 and 2010 and for whom sufficient cytogenetic information was available. In total, 903 patients were included into the study. At time of HSCT, 97 (10.7%) patients had untreated MDS, 218 (24.1%) patients had advanced MDS or AML evolving from MDS in complete remission, and 227 (25.1%) patients were not in remission after treatment (in 12.3% information on stage of the disease was not available). Median time between diagnosis and transplant was 6.6 months (range 0.2-359.3). Matched related donor HSCT was performed in 574 patients (63.6%), and matched unrelated donor HSCT in 329 patients (36.4%). Bone marrow (35.4%) or peripheral blood (64.6%) served as stem cell graft. Myeloablative preparative regimens were used in 582 patients (64.5%), and a non-myeloablative regimen was given to 320 patients (35.4%).

Impact of cytogenetic classification was analyzed in uni- and multivariate models regarding overall survival (OS) and relapse free survival (RFS) after HSCT. Predictive performance of the 2 classifications was compared by means of the cross-validated log partial likelihood.


Estimated 5-year RFS and OS were 32% and 36% respectively. According to the 5-group cytogenetic classification 19 (2.1%) patients had very good risk cytogenetics, 204 (22.6%) normal risk cytogenetics, 438 (48.5%) intermediate risk cytogenetics, 178 (19.7%) poor risk cytogenetics, and 64 (7.1%) very poor risk cytogenetics. Good, intermediate, and poor risk cytogenetics according to IPSS were found in 192 (38.0%), 500 (40.2%), and 211 (23.7%) patients, respectively.

In univariate analysis 5-group cytogenetic information was found to be strongly associated with OS and RFS (OS: log-rank test P<.01, RFS: P<.01) (Figure 1).

Further clinicopathologic factors showed a significant impact on impaired OS and RFS: Disease status at HSCT (RA/RARS no pretreatment; RAEB(t)/sAML in CR; RAEB(t)/sAML not in CR, RAEB(t)/sAML untreated) (OS: P<.01, RFS: P<.01) and IPSS cytogenetics (good; intermediate; poor) (OS: P<.01, RFS: P<.01). Patient age showed an impact for RFS (P=.05), but not for OS (P=.09).

In multivariate analysis, statistically significant predictors for RFS and OS at HSCT were 5-group cytogenetics, IPSS-cytogenetics, disease status and patient's age. Using 5-group cytogenetics classification, patients with poor risk [(RFS: P=.001, HR=1.40 (95% CI: 1.15-1.71); OS: P=.003, HR=1.38 (95% CI: 1.12-1.70)] or very poor risk cytogenetics [(RFS: P<.001, HR=2.14 (95% CI: 1.6-2.9); OS: P<.001, HR=2.14 (95% CI: 1.59-2.87)] had worse RFS and OS than patients in the other 3 risk groups. Patients with very poor risk cytogenetics had worse RFS and OS compared to patients with poor risk cytogenetics [(RFS: P<.01, HR=1.53 (95-% CI: 1.11-2.11), OS: P<.01, HR=1.55 (95-% CI: 1.11-2.15)].

When comparing the predictive performance of a series of 3 models both for OS and for RFS – (1) with only classical risk factors, (2) these extended with IPSS cytogenetics, (3) extended with 5-group classification instead-, the model with 5-group cytogenetics performed best.


In this international, multicentric analysis we confirm that MDS patients with poor and very poor risk cytogenetics had significantly worse RFS and OS after HSCT than patients in the other risk groups of the 5-group cytogenetic classifier. New therapeutic strategies to prevent relapse after HSCT in patients with poor or very poor cytogenetics are urgently needed.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

CK and GG contributed equally