Characterization of Chemoresistant Cell Populations Improves Risk Stratification and Therapy Prediction in Pediatric AML

Most pediatric acute myeloid leukemia (pAML) patients achieve complete remission after chemotherapy, yet relapse is common, and nearly 40% ultimately die of the disease. Prognosis is currently assessed using cytogenetic biomarkers and measurable residual disease (MRD) after one chemotherapy cycle, with the highest-risk patients referred for stem cell transplantation (SCT) in first remission. Because aggressive therapies such as SCT are highly toxic, yet cures after relapse are rare, accurate early risk prediction is essential for improving outcomes. To address this need, we analyzed paired diagnosis-relapse samples from 33 pAML patients at single-cell resolution and identified chemoresistant populations whose abundance at diagnosis significantly improved risk prediction. Incorporating these populations into a risk model revealed a previously unrecognized subgroup, representing 20% of pAML cases and accounting for half of the deaths among patients who were not prescribed SCT in first remission, with a 5-year event-free survival (EFS) below 40%. Moreover, molecular characterization of these chemoresistant cell populations uncovered potential therapeutic targets and candidate interventions relevant to most high-risk patients, paving the way for more effective targeted treatments for high-risk pAML patients.