A Comprehensive Prognostic Model for Predicting Leukemia-Free Survival Following Allogeneic Hematopoietic Cell Transplantation in Acute Leukemia

Highlight

Allogeneic hematopoietic cell transplantation (allo-HCT) serves as a potential cure for acute leukemia (AL); however, relapse and non-relapse mortality remain significant obstacles. Current predictive tools are limited to isolated prognostic aspects and do not effectively forecast leukemia-free survival (LFS), the most relevant metric reflecting the curative intent of allo-HCT. The newly developed holistic H-score integrates multiple pre-transplant factors, enabling four-tier risk stratification with superior predictive capacity for LFS and overall survival (OS). This model may enhance clinical decision-making and patient counseling, and establish a baseline for evaluating novel transplant interventions.

Study Background

Acute leukemia represents a heterogeneous group of aggressive hematologic malignancies characterized by rapid proliferation of immature myeloid or lymphoid cells. Allogeneic hematopoietic cell transplantation (allo-HCT) remains a cornerstone curative therapy for eligible patients. Despite advances in transplant techniques, conditioning regimens, and supportive care, outcomes are limited primarily by disease relapse and non-relapse mortality, including graft-versus-host disease and infections.

Historically, prognostic models have targeted singular domains such as disease biology, patient fitness, or transplant-related factors but fail to jointly predict the key clinical endpoint of leukemia-free survival (LFS), which encompasses both relapse and mortality events. Therefore, there is an unmet need for a comprehensive, practical prognostic tool synthesizing multiple dimensions of patient and treatment characteristics to accurately estimate individualized LFS after allo-HCT for acute leukemia.

Study Design

This investigation is a retrospective cohort study encompassing a large, international dataset of 24,317 acute leukemia patients who underwent allo-HCT. The dataset was divided into a training cohort (N=19,029) and a geographically distinct validation cohort (N=4,760). The study incorporated key pre-transplant variables, including patient demographics, disease features, and treatment details.

The primary endpoint was leukemia-free survival (LFS) post-transplant, with secondary endpoints including overall survival (OS), relapse incidence, and non-relapse mortality. Prognostic weights for each component variable were calculated using Cox proportional hazards regression models in the training cohort. These weights were combined into a composite H-score, which stratified patients into four risk groups: low, intermediate, high, and very high risk.

Key Findings

In the overall cohort, 2-year overall survival was 64%, and 2-year leukemia-free survival was 56%. Application of the H-score stratified patients robustly:

• Low-risk group exhibited a 2-year LFS of 66.2%
• Very high-risk group showed a markedly reduced 2-year LFS of 32.0%

The H-score was identified as the strongest independent predictor of LFS (p < 0.0001), outperforming existing individual indices that focus on single prognostic domains. Kaplan-Meier survival curves demonstrated clear separation across risk strata, reaffirming the model’s discriminative power.

Moreover, the H-score independently predicted overall survival, relapse, and non-relapse mortality, underscoring its comprehensive prognostic utility. Its multidimensional integration allows clinicians to appraise both disease- and transplant-related risks concurrently, enhancing personalized risk assessment.

Expert Commentary

This study represents a significant advance in prognostic modeling for allo-HCT in acute leukemia by incorporating a holistic approach that captures the complex interplay of patient, disease, and treatment variables into one actionable score. Such integrative tools align with precision medicine goals, facilitating informed clinical decision-making, including donor selection, conditioning intensity, and post-transplant interventions.

While individual prediction remains inherently limited by biological variability and unmeasured confounders, the H-score offers critical insight beyond existing models. Its validation in a geographically independent cohort increases generalizability across transplant centers and patient populations.

Future studies should explore prospective validation and assess the score’s utility in guiding therapeutic intensification strategies or novel cellular therapies. Integration with emerging molecular and genomic biomarkers might further refine prognostication.

Conclusion

The holistic H-score provides a practical, evidence-based prognostic tool for leukemia-free survival following allogeneic hematopoietic cell transplantation in acute leukemia. By synthesizing multiple prognostic domains, it enhances risk stratification beyond current models and supports precision medicine approaches. This tool can assist clinicians in patient counseling and serve as a benchmark for comparative effectiveness studies of new transplant technologies and treatments.

Ongoing research to prospectively validate and integrate molecular data may enhance its predictive accuracy and clinical utility, ultimately aiming to improve long-term outcomes for acute leukemia patients undergoing allo-HCT.