Diagnostic accuracy of magnetic resonance imaging in patients with suspected pulmonary embolism: A bivariate meta-analysis

Background Computed tomographic pulmonary angiography (CTPA) has simplified the diagnostic approach to patients with clinically suspected pulmonary embolism (PE), but alternative imaging tests are still advocated. We aimed to systematically assess the diagnostic accuracy of magnetic resonance imaging (MRI) for PE diagnosis. Methods Studies evaluating the diagnostic accuracy of MRI for the diagnosis of PE were systematically searched in the PubMed and EMBASE databases (up to May 2016). QUADAS - 2 tool was used for the quality assessment of the primary studies. A bivariate random-effects regression approach was used for summary estimates of both sensitivity and specificity. Results Thirteen studies, for a total of 1170 patients, were included. Weighted mean prevalence of PE was 37% at random-effect model. Weighted mean inconclusive MRI results were 19% at random-effect model. After exclusion of technical inadequate results, MRI bivariate weighted mean sensitivity was 80.9% (95% confidence interval [CI] 68.2, 89.4%), with a bivariate weighted mean specificity of 96.4% (95% CI 92.4, 98.3%), Conclusions MRI has high specificity but limited sensitivity for the diagnosis of PE. Inconclusive results are a major limitation to the practical application of MRI. Management studies are needed to more precisely define the role of MRI in the diagnostic workup of patients with suspected PE.