ABSTRACT NUMBER: ESOC2026A57 INCREMENTAL VALUE OF INFLAMMATORY BIOMARKERS AS COMPLEMENT TO ICH SCORE IN PREDICTING INTRACEREBRAL HEMORRHAGE PROGNOSIS

Abstract Background and aims The ICH Scale (ICHs) is the established predictor of prognosis in intracerebral hemorrhage (ICH). Simple inflammatory biomarkers might improve risk stratification, but their incremental value is unknown. To evaluate whether inflammatory biomarkers (neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, systemic inflammation index, and systemic inflammation response index) add discriminative capacity when combined with the ICHs. Methods Retrospective cohort of 95 patients with primary ICH. The ICHs was constructed using five components: age, GCS, infratentorial location, volume, and intraventricular extension. Three logistic regression models were developed: Model1 used ICHs alone, Model2 used biomarkers alone (NLR, PLR, SII, SIRI), and Model3 combined ICHs with biomarkers. Area under the curve (AUC) was calculated with bootstrap validation using 1000 replications. Two outcomes were evaluated at 90-day mortality and functional dependence. Results For 90-day mortality (n=21/95, 22%), ICHs alone achieved AUC:0.775 (95%CI: 0.672-0.881), biomarkers alone AUC:0.590 (95%CI: 0.466-0.715), and the combined model AUC:0.813 (95%CI: 0.707-0.908), representing +0.038 point improvement (+4.9%). For functional dependence (n=66/95, 70%), ICHs alone achieved AUC 0.684 (95%CI: 0.581-0.786), biomarkers alone AUC 0.624 (95%CI: 0.511-0.733), and the combined model AUC 0.744 (95%CI: 0.638-0.842), representing +0.060 point improvement (+8.8%). In multivariable logistic regression, ICHs remained an independent significant predictor for both outcomes (p<0.01). Biomarkers were not independently significant. Conclusions Biomarkers alone have poor discriminatory power, but when integrated with the validated ICHs they significantly improve prognostic capacity. This suggests value as a complement to validated clinical scales for improving early risk stratification in ICH. Conflict of interest Nothing to disclose Table 1 - belongs to Results