ESMO Congress 2024 | OncologyPRO

Abstract 830P

Background

Current pediatric acute myeloid leukemia (pAML) risk stratification heavily relies on cytogenetic findings. However, despite the generally low-risk feature associated with RUNX1-RUNX1T1 fusion gene, a substantial proportion of patients still experience unfavorable overall survival (OS) and/or event-free survival (EFS). Hence, a specific clinical decision support tool is crucial for the precise evaluation of survival in RUNX1-RUNX1T1+ pAML patients, aiming for enhanced outcomes.

Methods

The study included 2009 pAML patients, of whom 284 carried the RUNX1-RUNX1T1 fusion gene (RR-pAML). The patient data was randomly divided into training and validation subsets; the primary endpoints were OS and EFS. The prognostic model was constructed using univariate and multivariate Cox analyses. Model performance was evaluated using C-index and AUC values.

Results

The RR-pAML model was constructed based on two clinical factors. This model effectively categorized patients into low- and high-risk groups, which exhibited distinct clinical characteristics, response rates, relapse risk and mortality. The 5-year OS rates for the low- and high-risk groups were 91.0% and 73.0%, respectively (p=0.024, AUC 0.69). For EFS, the 5-year rates were 76.8% and 50.2%, respectively (p<0.001, AUC 0.70). Compared to previous prognostic models, the new model demonstrated superior performance in C-index and AUCs. It reclassified 31.7% of patients into the high-risk category and predicted relapse risk.

Conclusions

The new model is a straightforward yet effective clinical stratification tool for pAML patients carrying the RUNX1-RUNX1T1 fusion gene; it enhances risk assessment and facilitates more informed decision-making in the management of RUNX1-RUNX1T1+ patients.