Background: The prognostic value of the NIH consensus criteria for graft-versus-host disease (GVHD) is not well defined yet.
Patients and methods: We analyzed NIH-defined GVHD in 147 acute lymphoblastic leukemia (ALL) patients.
Results: The cumulative incidence of classic acute GVHD (aGVHD), late aGVHD and chronic GVHD (cGVHD) was 63%, 12% and 41%, respectively. cGVHD was subclassified as classic versus overlap syndrome in 40% versus 60% of cases. In multivariate Cox regression analysis with GVHD as time-dependent covariate, classic aGVHD grade III/IV had a negative impact on overall survival (OS) due to higher non-relapse mortality. cGVHD of any grade was associated with superior OS, which was due to lower relapse incidence. Classic cGVHD versus overlap syndrome had no differential impact. In 44 patients without GVHD after transplant who received donor lymphocyte infusions (DLI), the cumulative incidence of classic aGVHD, late aGVHD or cGVHD was 60%, 5% and 57%. Occurrence of cGVHD after DLI was associated with improved OS due to lower relapse incidence.
Conclusions: The NIH consensus criteria for GVHD clearly define prognostic subgroups in patients transplanted for ALL. The improved OS in patients developing cGVHD after transplant or DLI gives clear evidence for a potent graft-versus-leukemia effect in this indication.