Serum sCD163 as a potential biomarker for predicting poor functional outcome at 3 months in acute ischemic stroke without reperfusion: a prospective study.

Introduction: Neuroinflammation plays a critical role in the prognosis of acute ischemic stroke (AIS). Soluble CD163 (sCD163), a marker of monocyte/macrophage activation, has been linked to stroke severity, but its independent prognostic value and clinical utility remain uncertain. This study aimed to evaluate whether serum sCD163 measured within 24 hours of admission predicts 3_month poor functional outcome in AIS patients without reperfusion therapy, and to construct and externally validate a prognostic model incorporating sCD163. Methods: This prospective, dual_cohort study enrolled 664 first_ever AIS patients (training cohort, 2021 -2023) and 300 patients from an independent center (external validation cohort, 2024). All patients did not receive intravenous thrombolysis or mechanical thrombectomy. Serum sCD163 levels were measured at admission by ELISA. The primary outcome was poor functional outcome at 3 months, defined as modified Rankin Scale (mRS) > 2. Multivariable logistic regression with LASSO variable selection was used to identify independent predictors. Model performance was assessed by discrimination (AUC), calibration (Hosmer_Lemeshow test), and decision curve analysis (DCA), with internal bootstrap validation and external validation. Results: Poor outcome occurred in 34.1% (226/664) of the training cohort and 34.0% (102/300) of the validation cohort. Serum sCD163 levels were significantly higher in poor_outcome patients (820 ± 110 ng/mL vs. 650 ± 80 ng/mL, P < 0.001). The optimal cutoff was 742.5 ng/mL. After multivariable adjustment, high sCD163 (≥ 742.5 ng/mL) was an independent predictor of poor outcome (adjusted OR = 7.52, 95% CI: 4.42 -12.78, P < 0.001), along with baseline NIHSS score, DWI infarct volume, and atrial fibrillation. The combined model achieved an internal bootstrap_validated AUC of 0.877 (95% CI: 0.829 -0.926) and an external validation AUC of 0.867 (95% CI: 0.815 -0.921), with good calibration (Hosmer_Lemeshow P = 0.868 and 0.680, respectively) and superior net clinical benefit on DCA. Adding sCD163 significantly improved model performance over clinical predictors alone (ΔAUC = 0.118, P < 0.001). Discussion: Serum sCD163 within 24 hours of admission is a robust, independent biomarker for predicting 3_month poor functional outcome in AIS patients without reperfusion. The prognostic model combining sCD163 with traditional clinical and imaging variables demonstrates excellent discrimination, calibration, and clinical utility in an independent external cohort. Our findings support the early measurement of sCD163 for individualized risk stratification, though further mechanistic and multi_center studies are needed to confirm generalizability and causality.