Diagnostic value of gut microbiota profiling and circulating biomarkers to predict post-stroke infection in acute ischemic stroke

Background Post-stroke infection (PSI), particularly pneumonia and urinary tract infection, is a common and serious complication after acute ischemic stroke (AIS). Current diagnostic biomarkers provide limited accuracy when used in isolation. This study aimed to evaluate the diagnostic value of circulating biomarkers and gut microbiota profiling, both individually and in combination, to predict PSI in AIS patients. Methods We conducted a prospective observational study at Prof. Dr. dr. Mahar Mardjono National Brain Center Hospital, Jakarta. A total of 80 AIS patients admitted within 24 h of onset were enrolled and followed for 7 days to assess PSI. Blood samples were analyzed for NMDAR, butyrate, TMAO, RANKL, iFABP, and LPS. Stool samples were collected for 16S rRNA sequencing. Diagnostic performance was evaluated using ROC curves, with AUC, sensitivity, and specificity calculated. Multivariate logistic regression models were constructed to assess independent predictors and combined diagnostic accuracy. Results PSI occurred in 37/80 patients (46.3%). NMDAR showed the highest diagnostic performance (AUC 0.911; sensitivity 86.5%; specificity 90.7), followed by iFABP (AUC 0.894), LPS (AUC 0.896), RANKL (AUC 0.881), butyrate (AUC 0.866), and TMAO (AUC 0.865). Gut microbiota analysis revealed reduced evenness and dominance imbalance in infected patients, with enrichment of pathogenic taxa ( Escherichia coli, Salmonella enterica ) and depletion of SCFA-producing commensals ( Faecalibacterium prausnitzii, Roseburia intestinalis ). Multivariate models integrating microbiota features and biomarkers improved predictive accuracy compared with single-domain approaches. Conclusion Integrating circulating biomarkers with gut microbiota profiling significantly enhances early prediction of PSI in AIS. These findings highlight the role of the gut–brain–immune axis in post-stroke complications and support combined biomarker–microbiota models for risk stratification and preventive strategies.