Gut dysbiosis and bile acid metabolic perturbations in post-stroke cognitive impairment: a multi-omics study

Background Post-stroke cognitive impairment (PSCI) is a major contributor to post-stroke disability worldwide, yet its underlying mechanisms remain incompletely understood. This study aims to characterize the gut microbial and metabolic signatures of PSCI and explore their potential diagnostic value. Methods This prospective cohort included 94 patients with acute ischemic stroke. Cognition was assessed 3 months after stroke using the Montreal Cognitive Assessment; PSCI was defined as a score ≤24. Fecal metagenomics, targeted plasma metabolomics, and multiplex cytokine assays were performed. Machine-learning (ML) models with nested cross-validation evaluated multimodal discrimination of PSCI. Results Compared to non-PSCI controls, PSCI patients exhibited a distinct gut ecosystem characterized by an enrichment of opportunistic pathogens (e.g., Clostridioides difficile ) and a depletion of beneficial commensals (e.g., Phocaeicola vulgatus ). Plasma metabolomics revealed elevated conjugated bile acids (glycochenodeoxycholic acid (GCDCA) and glycodeoxycholic acid (GDCA)), microbial-derived toxins (p-cresyl sulfate), and neuroendocrine markers, alongside nominal elevated IL-22. An exploratory mediation analysis suggested a nominal indirect association of P. vulgatus with PSCI via conjugated bile acids. A multimodal ML model (SVM-RBF) achieved an AUC of 0.817 (95% CI 0.729–0.897) with favorable calibration. Conclusion PSCI is associated with a multimodal gut–brain axis signature involving gut dysbiosis, conjugated bile acid elevation, and systemic inflammatory changes. These descriptive findings provide a basis for future targeted studies in larger cohorts.