Background: To compare the diagnostic yield of echocardiography and cardiovascular MRI (CMR) to detect structural sources of embolism, in patients with ischemic stroke with a secondary analysis of non-stroke populations.
Methods and Results: We searched MEDLINE/Embase (from 01.01.2000 to 24.04.2021) for studies including CMR to assess prespecified sources of embolism. Comparison included transthoracic and/or transesophageal echocardiography. Two authors independently screened studies, extracted data and assessed bias using the QUADAS-2 tool. Estimates of diagnostic yield were reported and pooled. Twenty-seven studies with 2,525 patients were included in a study-level analysis. Most studies had moderate to high risk of bias. Persistent foramen ovale, complex aortic plaques, left ventricular and left atrial thrombus were the most common pathologies. There was no difference in the yield of left ventricular thrombus detection between both modalities for stroke populations (4 studies), but an increased yield of CMR in non-stroke populations (28.1 vs. 16.0%, P < 0.001, 10 studies). The diagnostic yield in stroke patients for detection of persistent foramen ovale was lower in CMR compared to transoesophageal echocardiography (29.3 vs. 53.7%, P < 0.001, 5 studies). For both echocardiography and CMR the clinical impact of the management consequences derived from many of the diagnostic findings remained undetermined in the identified studies.
Conclusions: Echocardiography and CMR seem to have similar diagnostic yield for most cardioaortic sources of embolism except persistent foramen ovale and left ventricular thrombus. Randomized controlled diagnostic trials are necessary to understand the impact on the management and potential clinical benefits of the assessment of structural cardioaortic stroke sources.
Registration: PROSPERO: CRD42020158787.