Abstract
BACKGROUND: Post-stroke epilepsy (PSE) is a severe complication characterized by significant heterogeneity. Traditional anatomical models often fail to identify patients with high metabolic risk but minor structural injury. Based on the concept that systemic metabolic and nutritional disturbances exacerbate neuronal excitability, we proposed a "Two-Hit" hypothesis: an acute immune-inflammatory hit combined with a hypoxia-metabolic hit acts upon nutritionally compromised brain tissue to drive epileptogenesis. This study aims to evaluate the synergistic value of the Immuno-Nutritional Index (C-reactive protein to Albumin Ratio, CAR) and Hypoxia-Nutritional Index (Lactate to Albumin Ratio, LAR) in predicting PSE. METHODS: We conducted a multi-center retrospective cohort study involving 21,459 acute ischemic stroke patients. CAR and LAR were calculated from admission biomarkers to quantify immuno-nutritional and hypoxia-metabolic status. Restricted cubic splines (RCS) were used to model non-linear dose-response relationships. A "Two-Hit" multivariate prediction model was constructed, and its incremental value over baseline clinical features was assessed using the Net Reclassification Improvement (NRI) and Integrated Discrimination Improvement (IDI). A web-based risk calculator was developed for clinical translation. RESULTS: During a one-year follow-up, 936 patients (4.36%) developed PSE. CAR exhibited a J-shaped relationship with epilepsy risk, reflecting an inflammatory threshold, while LAR showed a bell-shaped association, indicating a "metabolic hyper-excitatory state". A significant synergistic effect was observed: patients with concurrent elevations in both indices ("Double High") had a 13.5% incidence rate compared to 2.4% in the "Double Low" group. The "Two-Hit" model achieved an AUC of 0.888, significantly outperforming single-marker and baseline models (NRI 0.820, p < 0.001). Importantly, these nutritional indices maintained predictive value even in patients with minor stroke severity (NIHSS < 4). CONCLUSION: The CAR and LAR are potent synergistic predictors of PSE, supporting a "Two-Hit" mechanism involving immuno-metabolic disturbances. The developed web-based calculator serves as a valuable preliminary screening tool to identify metabolically high-risk patients. While the model demonstrates robust internal validity, external validation is warranted before widespread clinical adoption. These findings also suggest that optimizing immuno-nutritional management may act as a novel neuroprotective strategy.