RESUMEN
BACKGROUND AND PURPOSE: Etiology of acute ischemic stroke remains undetermined (cryptogenic) in about 25% of patients after state-of-the-art diagnostic work-up. METHODS: One-hundred and three patients with magnetic resonance imaging (MRI)-proven acute ischemic stroke of undetermined origin were prospectively enrolled and underwent 3-T cardiac MRI and magnetic resonance angiography of the aortic arch in addition to state-of-the-art diagnostic work-up, including transesophageal echocardiography (TEE). We analyzed the feasibility, diagnostic accuracy, and added value of cardiovascular MRI (cvMRI) compared with TEE for detecting sources of stroke. RESULTS: Overall, 102 (99.0%) ischemic stroke patients (median 63 years [interquartile range, 53-72], 24% female, median NIHSS (National Institutes of Health Stroke Scale) score on admission 2 [interquartile range, 1-4]) underwent cvMRI and TEE in hospital; 89 (86.4%) patients completed the cvMRI examination. In 93 cryptogenic stroke patients, a high-risk embolic source was found in 9 (8.7%) patients by cvMRI and in 11 (11.8%) patients by echocardiography, respectively. cvMRI and echocardiography findings were consistent in 80 (86.0%) patients, resulting in a degree of agreement of κ=0.24. In 82 patients with cryptogenic stroke according to routine work-up, including TEE, cvMRI identified stroke etiology in additional 5 (6.1%) patients. Late gadolinium enhancement consistent with previous myocardial infarction was found in 13 (14.6%) out of 89 stroke patients completing cvMRI. Only 2 of these 13 patients had known coronary artery disease. CONCLUSIONS: Our study demonstrated that cvMRI was feasible in the vast majority of included patients with acute ischemic stroke. The diagnostic information of cvMRI seems to be complementary to TEE but is not replacing echocardiography after acute ischemic stroke. CLINICAL TRIAL REGISTRATION: URL: http://www.clinicaltrials.gov. Unique identifier: NCT01917955.