Brain-age defines a structural signature of outcomes after vagus nerve stimulation in children

Introduction: Outcomes following vagus nerve stimulation (VNS) are difficult to predict prior to surgery in pediatric drug-resistant epilepsy (DRE). We investigated whether structural brain differences among children may explain variability in VNS response. Methods: A pediatric-specific brain-age model was developed using structural MRI from 2623 healthy individuals aged 1.9 to 90 years, capturing nonlinear neurodevelopmental trajectories. Model accuracy was evaluated using cross-validation and independent validation in a pediatric cohort. The model was then applied to a multicenter cohort of 126 children with DRE treated with VNS. Results: The brain-age model demonstrated high accuracy (R² = 0.93; mean absolute error = 3.5 years) and generalized well to an independent pediatric cohort (mean absolute error = 2.16 years). Applied to the VNS cohort, it revealed significantly elevated brain-age gaps (BrainAGEgap) compared to age- and sex-matched controls (t = 6.27, p < 0.0001), indicating cumulative structural divergence from age-referenced norms. BrainAGEgap correlated with baseline seizure burden and postoperative changes in seizure frequency. Lower pre-surgical BrainAGEgap predicted a significantly higher likelihood of becoming a clinical responder to VNS (p = 0.01). Local age within the cingulate cortex, thalamus, nucleus accumbens, and prefrontal cortex contributed disproportionately to BrainAGEgap differences, aligning with VNS-related circuitry. Conclusion: Developmental structural differences can be summarized using a robust, single-metric biomarker that reflects both disease burden and treatment responsiveness. BrainAGEgap may support individualized prognostic assessment and improve clinical decision-making in pediatric DRE undergoing neuromodulation.