Association and predictive value of blood branched-chain and excitatory amino acids with autism spectrum disorder severity

BACKGROUND: Autism spectrum disorder (ASD) is a neurodevelopmental disorder with heterogeneous severity. Abnormal amino acid metabolism may contribute to ASD pathogenesis, but the relationship between amino acid levels and severity remains unclear. This study investigated the association and predictive value of blood amino acid profiles for ASD severity. METHODS: We enrolled 474 children with ASD, divided into mild-to-moderate (CARS 30-36, n = 123) and severe groups (CARS 37-60, n = 351). Tandem mass spectrometry measured blood branched-chain amino acids (BCAAs) and excitatory amino acids (EAAs). Multivariate logistic and restricted cubic spline regressions analyzed associations and dose-response relationships. Model performance was evaluated by ROC, calibration, and decision curve analysis. RESULTS: The severe group had higher total BCAAs (307.50 ± 79.94 vs. 284.74 ± 70.51 µmol/L, P = 0.005) and EAAs (245.18 ± 72.02 vs. 227.83 ± 67.32 µmol/L, P = 0.020). After full adjustment, leucine/isoleucine showed the strongest association with severe ASD (OR = 1.012, 95% CI: 1.006-1.018, P < 0.001), followed by glutamate (OR = 1.005, 95% CI: 1.002-1.009, P = 0.006). BCAAs, leucine/isoleucine, and EAAs exhibited linear positive correlations with severe ASD risk; glutamate showed nonlinear association. Leucine/isoleucine had optimal predictive performance (AUC = 0.807, 95% CI: 0.765-0.849; sensitivity 69.80%, specificity 76.42% at 125.595 µmol/L). Subgroup analysis indicated that the associations between amino acids and ASD severity were stronger in children ≤ 3 years old and those with low BMI, with valine showing an interaction with sex (P = 0.045). CONCLUSIONS: Blood BCAAs and EAAs were closely associated with ASD severity, with leucine/isoleucine as the strongest predictor. Amino acid profile-based models provide biomarker evidence for identifying severe ASD and precision intervention.