Cardiac MRI improves identification of etiology of acute ischemic stroke.

BACKGROUND: An accurate subtype classification of acute ischemic stroke is important in clinical practice as it can greatly influence patient care in terms of acute management and devising secondary stroke prevention strategies. Approximately, one third of ischemic strokes are cryptogenic despite a comprehensive workup. Diagnostic workup for detecting cardioaortic sources of cerebral embolism commonly includes transthoracic echocardiography (TTE). However, TTE has a limited diagnostic power to detect some of the cardioaortic abnormalities and additional imaging modalities are often needed to accurately assess such abnormalities.
PURPOSE: We evaluated the feasibility of cardiovascular magnetic resonance (CMR) imaging to detect the cardioaortic sources of ischemic stroke.
METHODS: A total of 106 patients were included, of which 85 had an ischemic stroke and 21 had a transient ischemic attack (TIA). Routine diagnostic workup (RDW) included brain diffusion-weighted image MRI, telemetry, magnetic resonance angiography/CT angiography of head and neck, carotid duplex ultrasonography, laboratory studies and TTE. Patients additionally underwent CMR. Subtype assignment was performed in accordance with the Stop Stroke Study of the Trial of Org 10172 in Acute Stroke Treatment classification system by a stroke neurologist after reviewing the admission notes and diagnostic test results. A second subtype classification was assigned with an additional criterion defined based on delayed enhancement (DE)-CMR findings. Additionally, the presence of non-coronary artery disease (CAD) scarring was assessed in ischemic stroke patients and compared with the TIA patients as the control group.
RESULTS: RDW detected cardioaortic embolism (CAE) stroke in 32 (37.6%) patients and cryptogenic stroke in 23 patients (27.1%). Addition of CMR resulted in a 26.1% reduction in the rate of cryptogenic strokes (6 patients). Furthermore, DE-CMR findings allowed for reclassification of three additional cryptogenic subtypes, resulting in a 39.1% reduction of cryptogenic stroke rate. Non-CAD scarring was detected in 13 (15.3%) stroke patients as opposed to only 1 (4.8%) TIA patient.
CONCLUSIONS: CMR is a valuable tool for the detection of CAE sources in patients with cryptogenic ischemic stroke and provides clinicians with a unique set of information that may substantially change the long-term management of these patients. DE-CMR also detects non-CAD scarring, which may indicate a predisposition to ischemic stroke. Further studies with larger samples and long-term follow-up are needed to further evaluate the clinical significance of our findings.