ABSTRACT NUMBER: ESOC2026A5 DUAL ANTIPLATELET THERAPY AND VERY EARLY NEUROLOGICAL DETERIORATION IN PATIENTS WITH NON-CARDIOEMBOLIC STROKE

Abstract Background and aims Although guidelines recommend initiating dual antiplatelet therapy (DAPT) within 24 hours after minor non-cardioembolic stroke, the prevalence, predictors, and prevention of very early neurological deterioration (vEND) in this period remain unlcear. This study examined (i) vEND prevalence, (ii) patient characteristics with vs. without vEND, and (iii) the effect of DAPT (versus single antiplatelet therapy) on vEND, stratified by stroke subtype and severity. Methods Retrospective single-center cohort study of adults with non-cardioembolic ischemic stroke treated at a Swiss tertiary hospital (2014-2023). vEND was defined as a > 2-point NIHSS increase within 24 hours. Patients were stratified by stroke etiology (large artery atherosclerosis [LAA] vs. non-LAA) and POINT-eligibility. Adjusted risk differences (RD) were calculated using logistic regression, with interaction and sensitivity analyses assessing effect modification and result robustness. Results Among 2,308 patients, vEND occurred in 5.8%, more frequently in LAA-patients than in non-LAA-patients (7.0% vs. 5.5%, P = 0.18). Patients with vEND had more severe deficits at admission, greater vascular risk, and worse in-hospital outcomes. In LAA-patients, DAPT was associated with lower vEND risk vs. SAPT (RD -5.4%, 95%CI [-10.2%, -0.6%], P = 0.027). No significant difference was seen in non-LAA patients (RD 3.6%, 95%CI [-2.0%, 9.2%], P = 0.210). Interaction analyses indicated a difference in the DAPT effect between groups (bootstrapped pinteraction = 0.010). POINT-eligibility did not modify the effect. Conclusions Our findings suggest DAPT may lower vEND risk in LAA-patients, although confirmation in larger prospective studies is needed. No such effect was observed in non-LAA-patients. If confirmed, these results support stroke subtype-specific antiplatelet strategies. Conflict of interest All authors declare no conflicts of interest relevant to this study. Figure 1 - belongs to Results