Plasma KCC2, NKCC1, and GABA as peripheral biomarkers in autism spectrum disorder: a combined ROC analysis.

Autism spectrum disorder (ASD) and epilepsy frequently co-occur, yet clinically actionable markers to stratify seizure risk and anticipate drug resistance remain limited. We evaluated whether plasma γ-aminobutyric acid (GABA) and the chloride co-transporters KCC2 and NKCC1, singly and in combination, provide a discriminative signal for ASD status and severity using receiver operating characteristic (ROC) methodology. Forty-six males with ASD and twenty-six age-matched neurotypical controls were phenotyped with the Childhood Autism Rating Scale and Social Responsiveness Scale. Plasma GABA, KCC2, and NKCC1 were quantified by ELISA. Nonparametric tests, logistic regression, and combined ROC analyses were applied. ASD was associated with reduced GABA (0.06 ± 0.04 vs. 0.12 ± 0.05 ng/mL), KCC2 (1.19 ± 1.01 vs. 4.92 ± 3.27 ng/mL), and NKCC1 (8.07 ± 7.08 vs. 10.96 ± 6.72 ng/mL) relative to controls (all p ≤ 0.035), alongside a lower KCC2/NKCC1 ratio (0.193 ± 0.172 vs. 0.525 ± 0.365; p = 0.001). KCC2 and NKCC1 were positively correlated in controls (R = 0.634; p = 0.001), whereas in severe ASD, GABA correlated negatively with KCC2 (R = - 0.638; p = 0.004), consistent with altered chloride homeostasis and GABAergic signaling. Discrimination was highest for KCC2 overall (AUC = 0.931) and in severe ASD (AUC = 0.987); GABA showed good discrimination (AUC = 0.827), and NKCC1 was modest (AUC = 0.664). Marker combinations improved classification: KCC2 + GABA achieved AUC = 0.939 overall and 0.922 in mild-moderate ASD, while GABA + NKCC1 reached AUC = 0.885 in severe ASD. Logistic models yielded odds ratios < 1 across strata, aligning with the observed decrements in ASD. These data indicate that combined ROC analysis of peripheral measures indexing neuronal chloride transport and inhibition provides a robust discriminative signal for ASD stratification and may inform future, mechanism-guided studies of seizure liability and pharmacoresistance.