Predictive value of lipopolysaccharide for cerebral-cardiac syndrome, stroke-associated pneumonia, and prognosis in acute ischemic stroke.

backgroundLipopolysaccharide (LPS), a key mediator of systemic inflammation via the brain-gut axis, is elevated in acute ischemic stroke (AIS). We aimed to investigate the association of plasma LPS levels with cerebral-cardiac syndrome (CCS), stroke-associated pneumonia (SAP), and poor functional outcome in AIS patients receiving intravenous thrombolysis (IVT).

methodsIn this prospective cohort of 120 AIS patients receiving IVT, plasma LPS levels were assessed at baseline (designated LPS-T1) and 24 h after IVT (designated LPS-T2). Outcomes included CCS, SAP, and poor 3-month outcome (modified Rankin Scale [mRS] score 3–6). Variable selection was performed via least absolute shrinkage and selection operator regression followed by backward stepwise elimination based on the Akaike Information Criterion. Final multivariable logistic models underwent internal validation via 1000 bootstrap resamples, with performance assessed using area under the receiver operating characteristic curve (AUC), calibration plots, and decision curve analysis (DCA).

resultsAmong 120 AIS patients receiving IVT, 21.67% developed CCS, 40.83% developed SAP, and 28.33% had poor functional outcome. Patients who developed CCS or SAP had significantly higher plasma LPS levels than those without these complications at both baseline and 24 h after IVT (all p < 0.05). Additionally, LPS-T2 levels were higher in those with poor functional outcome compared to those with good outcome (p = 0.018). The final prediction model for CCS incorporated age, C-reactive protein, LPS‑T2, lymphocyte count, and triglyceride level, while the SAP model included age, admission mRS score, platelet-to-neutrophil ratio, LPS‑T2, thrombin time, and protein S activity. The poor‑outcome model comprised admission National Institutes of Health Stroke Scale score, neutrophil-to-lymphocyte ratio, protein S, and LPS‑T2. All three models demonstrated excellent discrimination, with AUC values of 0.866 (95% confidence interval: 0.778–0.954) for CCS, 0.877 (0.816–0.939) for SAP, and 0.856 (0.779–0.933) for poor outcome. Bootstrap internal validation (1000 replicates) yielded optimism‑corrected AUCs of 0.844, 0.849, and 0.837, respectively. The models showed good calibration and DCA indicated a positive net clinical benefit.

conclusionsPlasma LPS level at 24 h after IVT serves as a potential biomarker for predicting CCS, SAP, and poor outcome in AIS patients, with the prediction models incorporating LPS‑T2 showing robust predictive performance.