Abstract WP146: White Matter Disease and Recurrent Stroke in ESUS: A Propensity-Weighted Analysis from the CASPR Registry

Background: Embolic stroke of undetermined source (ESUS) comprises infarct mechanisms presumed to be embolic and non-lacunar. One explanation for the lack of superiority of anticoagulation over aspirin is the inclusion of subcortical infarcts with white matter disease (WMD), reflecting small vessel pathology. We hypothesized that WMD is associated with recurrent ischemic stroke (RIS) in ESUS and this risk is not lessened by anticoagulation. Methods: We conducted a secondary analysis of the multicenter Cardiac Abnormalities in Stroke Prevention and Recurrence (CASPR) registry, including ESUS patients from 27 U.S. centers (2015–2024). WMD severity was graded using the Fazekas scale from MRI and dichotomized as mild (0–1) vs. moderate-to-severe (2–3). The primary outcome was RIS. Secondary outcomes were major bleeding and all-cause mortality. Inverse probability of treatment weighting (IPTW) adjusted for baseline differences. Cox proportional hazards models assessed associations between WMD and outcomes. Subgroup analyses explored effect modification by age, risk factors, and treatment type. Results: Among 1,890 ESUS patients with MRI data, 684 (36.2%) had Fazekas 2–3. These patients were older with more vascular risk factors, but IPTW achieved covariate balance. In the weighted cohort, RIS occurred more frequently in Fazekas 2–3 vs. 0–1 (7.1% vs. 4.6%/year; HR 1.56, 95% CI 1.10–2.22, p = 0.01). Major bleeding and mortality were similar (bleeding: 1.7% vs. 1.8%/yr, HR 0.96, p = 0.92; death: 7.0% vs. 6.7%/yr, HR 1.04, p = 0.81). Subgroup analyses showed no interaction between WMD and age, vascular risk factors, or treatment type (antiplatelet vs. anticoagulation), suggesting consistent RIS risk in those with higher WMD. Discussion: Moderate-to-severe WMD identifies ESUS patients at higher risk of ischemic stroke recurrence, with no mitigation by anticoagulation. These findings suggest that WMD may indicate non-embolic mechanisms and support the use of imaging-based risk stratification in ESUS.