Longitudinal serum metabolomics predicts therapeutic outcome in acute myeloid leukemia

Acute myeloid leukemia (AML) is the most common acute leukemia in adults, with approximately 50 % of patients failing to achieve remission during initial treatment and progressing to refractory AML. Metabolomics, a technology directly linked to clinical phenotypes, offers more precise susceptibility biomarkers compared to genomics and epigenetics. Our study compares metabolic samples collected at multiple time points pre- and after-chemotherapy, performs longitudinal integrated analysis to characterize dynamic alterations, and assesses the temporal impacts of therapeutic responses. Four metabolites-pseudouridine, O-phospho-L‑serine, l-aspartate, and 2-deoxy-d-ribose 1-phosphate-were significantly elevated in AML patients, mechanistically linking dysregulated nucleotide biosynthesis and adaptive amino acid metabolic reprogramming to leukemogenic proliferation.Longitudinal sampling during AML treatment revealed temporal metabolic changes, identifying key metabolites and pathways associated with therapeutic responses.By integrating pre- and after-treatment metabolic index with clinical indicators, we developed predictive models for treatment outcomes. The pre-treatment metabolic model (AUC=0.9143, 95 % CI 0.816-1) and the after-treatment metabolic index (AUC=0.9136, 95 % CI 0.83-1) both demonstrated excellent predictive performance for AML therapeutic outcomes. In conclusion, our findings underscore the potential of targeting glycolipid synthesis and amino acid metabolism to improve clinical outcomes. The dynamic metabolic reprogramming landscape serves as a robust indicator of AML treatment efficacy, offering novel directions for precision therapy in AML.