Oral and gut microbiota dysbiosis with strengthened oral-gut connectivity in post-stroke cognitive impairment.

BACKGROUND

Microbiome studies in post-stroke cognitive impairment (PSCI) have focused on the gut, while upstream oral dysbiosis and oral-gut axis signatures remain undercharacterized. We profiled paired oral and gut microbiota in PSCI to assess coordinated alterations and cross-site associations.

METHODS

This single-center cross-sectional study enrolled 133 post-stroke participants (64 PSCI). Paired tongue-coating and fecal samples underwent 16S rRNA gene sequencing. We compared α/β diversity, taxonomic composition, and predicted functional pathways at both sites, quantified within-individual oral-gut dissimilarity and the gut fraction of oral-gut shared microbiota, and adapted an oral enrichment score (OES) to index gut "oralization". Differential microbiota, predicted pathways, and MMSE/MoCA scores were integrated into a multi-layer association framework. Machine-learning models were built using oral features, gut features, and combined oral-gut features, with SHAP for interpretability.

RESULTS

PSCI showed reduced oral richness (ACE/Chao1) and reduced gut diversity/evenness (Shannon/Simpson), with significant β-diversity differences at both sites. Oral commensals (Leptotrichia, Neisseria) were depleted, whereas opportunistic taxa (Pseudomonas, Alloprevotella, Streptococcus) were enriched. In the gut, SCFA-associated microbiota (Coprococcus, Faecalibacterium, Ruminococcus) decreased, while Gram-negative potential pathogens (Enterobacter, Pseudomonas, Klebsiella) increased. Tax4Fun2-based inference suggested predicted gut functional alterations involving lipopolysaccharide biosynthesis potential and tryptophan metabolism-related pathways. Oral-gut metrics supported stronger oral-gut association in PSCI, including lower paired dissimilarity, a higher shared-genera fraction, and elevated OES (median 0.0368 vs 0.0142), mainly driven by oral-dominant microbiota (Streptococcus, Fusobacterium, Veillonella and Haemophilus). Combined oral-gut features showed the most favorable exploratory test-set performance (XGBoost AUC 0.945; average precision 0.943).

CONCLUSIONS

PSCI was associated with coordinated oral and gut dysbiosis, characterized by loss of commensals, enrichment of opportunistic or Gram-negative taxa, and a stronger gut oralization signal reflected by reduced oral-gut dissimilarity and elevated OES. Tax4Fun2 suggested predicted functional potential related to lipopolysaccharide biosynthesis and tryptophan metabolism, while combined oral-gut features showed favorable exploratory internal discrimination of PSCI. These findings require validation in larger longitudinal multi-omics cohorts.