Background: Diagnostic strategies for pulmonary embolism are complex and consist of non-invasive diagnostic tests done to avoid conventional pulmonary angiography as much as possible. We aimed to assess the diagnostic accuracy of magnetic resonance angiography (MRA) for the diagnosis of pulmonary embolism, using conventional pulmonary angiography as a reference method.
Methods: In a prospective study, we enrolled 141 patients with suspected pulmonary embolism and an abnormal perfusion scan. Patients underwent MRA before conventional pulmonary angiography. Two reviewers, masked with respect to the results of conventional pulmonary angiography, assessed MRA images independently. Statistical analyses used chi(2) and 95% CI.
Findings: MRA was contraindicated in 13 patients (9%), and images were not interpretable in eight (6%). MRA was done in two patients in whom conventional pulmonary angiography was contraindicated. Thus, MRA and conventional pulmonary angiography results were available in 118 patients (84%). Prevalence of pulmonary embolism was 30%. Images were read independently in 115 patients, and agreement obtained in 105 (91%), kappa=0.75. MRA identified 27 of 35 patients with proven pulmonary embolism (sensitivity 77%, 95% CI 61-90). Sensitivity of MRA for isolated subsegmental, segmental, and central or lobar pulmonary embolism was 40%, 84%, and 100%, respectively (p<0.01 for isolated subsegmental vs segmental or larger pulmonary embolism). However, subgroups contained small numbers. MRA identified pulmonary embolism in two patients with normal angiogram (98%, 92-100).
Interpretation: MRA is sensitive and specific for segmental or larger pulmonary embolism. Results are similar to those obtained with helical computed tomography, but MRA has safer contrast agents and does not involve ionising radiation. MRA could become part of the diagnostic strategy for pulmonary embolism.