Establishment and validation of a non-relapse mortality risk prediction model for high-risk and refractory acute leukemia patients undergoing allogeneic hematopoietic stem cell

This study aims to identify risk factors for non-relapse mortality following allogeneic hematopoietic stem cell transplantation (HSCT) in patients with high-risk refractory acute leukemia and to develop a predictive model. Clinical data and laboratory findings of 966 patients who underwent HSCT at the Aerospace Central Hospital from January 2015 to December 2021 were collected retrospectively, with 595 patients selected as study subjects. Univariate analysis using competing risk models identified potential risk factors, and cumulative incidence curves were plotted. The data were divided into a training set (n = 417) and a validation set (n = 178) at a 7:3 ratio. In the training set, the least absolute shrinkage and selection operator (Lasso) regression was used to identify predictors of non-relapse mortality, and a predictive model was constructed using competing risk analysis. The model was internally validated in the validation set. A nomogram was developed for visualization, and its performance was assessed. Predictors included age (< 18, 18–45, ≥ 45 years), pre-transplant status (CR/Non CR), extramedullary infiltration (No/Yes), HCT score (0, 1–2, ≥ 3), neutrophil engraftment time (< 16, ≥ 16 days), platelet engraftment time (< 16, ≥ 16 days), and aGVHD (0-I degree/II-IV degree). The nomogram accurately predicted 1-, 2-, and 3-year non-relapse mortality with AUCs of 0.839 (95% CI: 0.793–0.884), 0.802 (95% CI: 0.75–0.854), and 0.753 (95% CI: 0.691–0.815), respectively. Time-dependent ROC, time-dependent AUC, and calibration curves confirmed the model's discriminative power and precision. This nomogram may assist clinicians in assessing patient prognosis and devising personalized treatment plans.