Embolic infarct topology differs between atrial fibrillation subtypes and embolic stroke of undetermined source

The lack of superiority of anticoagulation over antiplatelet therapy in embolic stroke of undetermined source (ESUS) may be in part due to the misclassification of radiographic ESUS patterns as cardioembolic. In this imaging analysis, we sought to differentiate clinical and radiographic patterns of ESUS patients from patterns in patients with a highly probable cardioembolic source.

A prospective registry of consecutive adults with acute infarction on diffusion-weighted magnetic resonance imaging was queried. Patients with infarctions due to small vessel disease, large vessel disease, and other causes were excluded. Multivariable logistic regression was used to identify independent predictors of two potentially embolic patterns: (1) multifocal and (2) cortical lesions, comparing patients with ESUS against those with atrial fibrillation (AF).

Among 1243 screened patients, 343 (27.6%) experienced strokes due to ESUS or AF. Prior to the index stroke, patients with AF as compared to ESUS were older (median 75 vs. 65, p<0.01) and had more heart failure (25.9% vs. 8.4%, p<0.01). The odds of multifocal infarction were the same between patients with ESUS and both AF subtypes (p>0.05), however, cortical involvement was more associated with both AF versus ESUS (77.7% vs. 65.7%, P=0.02). A higher Fazekas grade of white matter disease was inversely associated with cortical infarction among included patients (aOR 0.77, 95% CI 0.62-0.96).

Cortical infarctions were twice as common among patients with AF versus ESUS. Subcortical infarct topography was strongly associated with chronic microvascular ischemic changes and therefore may not represent embolic phenomena. Larger-scale investigations are warranted to discern whether large or multifocal subcortical infarcts ought to be excluded from the ESUS designation.