Summary:To examine whether graft-versus-host-disease (GVHD) is associated with a graft-versus-leukaemia (GVL) effect that also influences the outcome of allogeneic stem cell transplantation (SCT) in childhood acute leukaemia, we evaluated all consecutive (n ¼ 169) children who had undergone SCT for ALL and AML at our centre. Median followup was 7 years. The 5-year probability of chronic GVHD was 34%. Median time to relapse was 24 months in children with chronic GVHD and 6 months in those without. The corresponding 5-year probabilities of relapse were 30 and 45% (P ¼ 0.01). The 5-year probability of survival was 54%. Patients with chronic GVHD had a significantly better survival, 77 vs 51% (P ¼ 0.01). In a Cox regression model, chronic GVHD independently decreased the risk of relapse (RR 0.44) and further predicted an increased chance of relapse-free survival (RR 1.7) and survival (RR 2.6). The impact of chronic GVHD on survival was most apparent in late-stage disease and in ALL. Acute GVHD was not an independent predictor for relapse or death in this study. This study is in support of a GVL effect in childhood leukaemia related to chronic GVHD, reducing the risk of relapse and improving survival. Bone Marrow Transplantation (2003) 31, 175-181. doi:10.1038/sj.bmt.1703808 Keywords: bone marrow transplantation; children graft-versus-host disease; graft-versus-leukaemia; relapse survivalThe major cause of failure after haematopoietic stem cell transplantation (SCT) in childhood leukaemia is relapse of disease, followed by transplant-related causes such as graftversus-host disease (GVHD), treatment toxicity and infections. 1,2 The original concept of SCT was to enable complete eradication of malignant cells by a maximum dose of chemoradiotherapy, bypassing the otherwise dose-limiting bone marrow aplasia with a salvaging autologous or allogeneic graft. In addition, now a well-established belief in the immunotherapeuthic potential of the allogeneic graft has evolved, usually described as the graft-versus-leukaemia (GVL) effect. This antileukaemia effect was first postulated in the 1950s by Barnes 3 and later reported in humans by Weiden et al 4 in 1979. Acute and chronic GVHD were presented as favourable in decreasing the risk of relapse.This was later confirmed in an IBMTR study in 1990, 5 which further supported the existence of a GVL effect by reporting a higher probability of relapse in patients receiving T-cell-depleted (TCD) or syngeneic grafts. Several recent studies, including those on the effect of donor lymphocyte infusions (DLI), have strengthened this concept. [6][7][8][9] However, treatment of relapses post-SCT with DLI has been considerably less successful for acute leukaemias than in CML. 9 In a previous study, comparing the outcomes of MUD and sibling graft recipients, we found a significant difference in occurrence of relapse between children who did or did not develop chronic GVHD. 2 We decided to further investigate this issue in the entire group of children transplanted for acute leukaemias at our c...