With the ever-rising number of patients undergoing allogeneic stem cell transplantation (allo-SCT), the number of those requiring admission to the intensive care unit (ICU) has also increased over the last two decades. Indeed, allo-SCT is associated with potentially life-threatening complications such as conditioning-regimen toxicity, graft-versus-host disease (GVHD) and relapse.

Several studies suggest that ICU admission may fail to provide benefits with a mortality rate attaining 95% in a subset of allo-SCT patients. Therefore, ICU triage policies have been set in order to identify patients who have the best chances of recovering from life-threatening complications. Not only were those policies based on studies that were restricted to patients admitted to ICU, but they also included very heterogeneous cohorts of patients in terms of underlying disease, patients characteristics and transplantations modalities. In addition, interpretation of the data is confounded by the inclusion of both autologous and allogeneic SCT patients and the absence of a clear distinction between acute, chronic and refractory GVHD.

With the purpose of identifying factors that affect overall survival (OS), we conducted a retrospective multicenter study on a homogenous cohort. The aim of this study was to establish a new prognostic score usable in screening patients requiring ICU and bases on early post-transplant complications.

Patients and methods

Transplantation modalities were made as homogeneous as possible using the following inclusion/exclusion criteria: (i) Patients with myelodysplastic syndrome aged more than 18 years referred for first allo-SCT following RIC with at least one year of follow-up (ii) source of stem cell was marrow or blood from either a sibling or an HLA-identical unrelated donor at allelic level (so called 10/10).

Patients who received allo-SCT from an HLA-mismatched donor, cord blood or T-cell-depleted graft were excluded as well as patients with CMML. Consequently, 275 consecutive patients with MDS who underwent allo-SCT between January 1999 and December 2009 in 24 French and Belgian centers were identified.

Patient and donor characteristics, transplantation modalities and complications occurred within 100 days after transplant were considered for univariate analysis. Multivariate Cox proportional hazards regression was used to investigate prognostic factors by backward stepwise analysis of the factors considered significant in univariate analysis. A prognostic score for 3-year OS was designed using the variables from the multivariate model and according to the rounded HRs values. Risk categories derived from this score were compared using log-rank test.


In multivariate analysis, 3-year OS was influenced by grade III/IV acute GVHD [HR=2.8; CI95=1.8-4.3; p<0.001], relapse before d+100 after transplant [HR=3; CI95=1.9-4.7; p<0.001), and the lack of platelet recovery (defined as the first occurrence of a stable platelet count ≥ 20G/L before d+100, persisting for more than 7 days without transfusion) [HR=6.6; CI95=4.1-10.6; p<0.001]. Therefore, a prognostic score was based on grade III/IV acute GVHD (1 point), relapse before d+100 (1point), lack of platelet recovery (2 points). Thus, this score discriminated three risk groups: good (score=0; n=191, 3-year OS=63%), intermediate (score=1; n=50; 3-year OS=20%) and poor (score≥2; n=34; 3-year OS=3%) p<0.001.


Our new score based on early post-transplant complications appears to be simple and helpful in screening RIC allo-SCT patients who require ICU. This prognostic score should be validated in larger cohorts.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.