A holistic prognostic model for leukemia-free survival after allogeneic transplantation in acute leukemia

Allogeneic hematopoietic cell transplantation (allo-HCT) can cure acute leukemia, but relapse and non-relapse mortality restrict its success. Existing models focus on single prognostic areas and are not designed to predict leukemia-free survival (LFS), the outcome that best reflects allo-HCT's curative intent. In this large retrospective cohort of allografted AL patients (N = 24,317), we combined key pre-transplant patient, disease, and treatment factors into a practical holistic H-score to predict post-allo-HCT LFS. Component weights were derived from Cox-regression hazard ratios in a training cohort (N = 19,029). The model was validated in a geographically split testing cohort (N = 4760), with outcomes assessed using Kaplan-Meier analysis and multivariate Cox models. In the overall cohort (N = 24,317), 2-year overall survival and LFS were 64% and 56%, respectively. The H-score stratified patients into four risk groups, with 2-year LFS ranging from 66.2% in the low-risk group to 32.0% in the very high-risk group. The H-score was the strongest independent predictor of LFS (p < 0.0001), outperforming individual indices and offering more refined risk stratification. The H-score was also an independent predictor of overall survival, relapse, and non-relapse mortality. While individual prediction is limited, the H-score aids patient counseling and provides a useful baseline for comparing new transplant treatments.