Large vessel cardioembolic stroke and embolic stroke of undetermined source share a common profile of matrix metalloproteinase‐9 level and susceptibility vessel sign length

BACKGROUND Embolic stroke of undetermined source (ESUS) account for up to 25% of ischemic strokes. Identification of biomarkers that could improve the prediction of stroke subtype and subsequently of stroke prevention still remains a major issue. METHODS HIBISCUS-STROKE cohort includes ischemic stroke patients with large vessel occlusion treated with mechanical thrombectomy following admission magnetic resonance imaging. Presence and length of SVS were assessed on gradient recalled echo T2*-weighted imaging. MMP-9 was measured on sera collected at admission. A multiple logistic regression model was performed to detect independent markers to distinguish cardioembolic (CE) from large artery atherosclerosis (LAA) subtype. RESULTS A total of 147 patients were included, of them the etiology was distributed as follows: 86 (58.5%) CE, 26 (17.7%) LAA and 35 (23.8%) ESUS. The optimal cutoff for differentiating CE from LAA subtype was 14.5 mm for SVS length (sensitivity, 79.7%; specificity, 72.7%) and 1110 ng/mL for admission MMP-9 level (sensitivity, 85.3%; specificity, 52.2%). Multivariate analysis revealed that current smoking (odds ratio (OR) 0.07, 95% confidence interval (CI) 0.01-0.93), tandem occlusion (OR 0.01, 95% CI 0.01-0.21), SVS length (OR 0.78, 95% CI 0.63-0.97) and admission MMP-9 level (OR 0.99, 95% CI 0.99-1.00) were inversely associated with CE subtype. SVS length and MMP-9 level did not differ between ESUS and CE subtypes. CONCLUSIONS SVS length and admission MMP-9 level may improve the prediction of CE subtype whose profile is close to ESUS, thus suggesting a common cardiac embolic source.