A win ratio approach for comparing graft-versus-host disease-free, relapse-free survival among alternative donors

The optimal alternative donor type for patients lacking human leucocyte antigen (HLA)-matched related or unrelated donors remains unclear. In comparative studies evaluating donor types, graft-versus-host disease (GVHD)-free, relapse-free survival (GRFS) represents a well-established end-point but has limitations. The win ratio approach addresses these limitations by analysing multiple end-points with varying severities to account for the relative component priorities. We compared HLA-mismatched unrelated donors, haploidentical donors and cord blood using both the hazard ratio (HR) of GRFS and the win ratio related to GRFS. The haploidentical donor group had a similar HR of GRFS (HR: 1.01, 95% confidence interval [CI]: 0.85-1.19, p = 0.916) and win ratio (win ratio: 0.86, 95% CI: 0.72-1.02, p = 0.081) to HLA-mismatched unrelated donors. Cord blood transplantation showed similar GRFS compared to HLA-mismatched unrelated donors in the Cox proportional model (HR: 1.14, 95% CI: 0.98-1.32, p = 0.085), significantly lower win ratio (win ratio: 0.80, 95% CI: 0.68-0.93, p = 0.004) and similar outcomes to haploidentical donors. HLA-mismatched unrelated donor transplantation showed comparable to superior outcomes among alternative donor types. Our results indicate the need to present the win ratio alongside conventional methods to assess the end-point robustly.