MR angiography as a screening tool for intracranial aneurysms: feasibility, test characteristics, and interobserver agreement.

Abstract

MR angiography may be an appropriate tool to screen for unruptured intracranial aneurysms. Feasibility, test characteristics, and interobserver agreement in evaluation of MR angiograms were assessed by members of the MARS (Magnetic resonance Angiography in Relatives of patients with Subarachnoid hemorrhage) Study Group. We screened 626 first-degree relatives of a consecutive series of 193 patients with subarachnoid hemorrhage examined at two institutions. We used MR imaging and MR angiography (three-dimensional time-of-flight imaging at both institutions and additional three-dimensional phase-contrast imaging at one institution). Three observers independently assessed the MR angiograms. Conventional angiography was performed in relatives with possible or definite aneurysms on MR angiography and was considered the standard of reference. Thirty-three aneurysms were found in 25 (4%; 95% confidence interval [CI], 3-6%) of 626 relatives. Thirteen (8%) of 169 relatives who refused screening had MR-related reasons; an additional six persons could not be screened because of contraindications for MR imaging (pregnancy, n = 1; claustrophobia, n = 5). The positive predictive value of MR angiography was 100% (95% CI, 79-100%) for "definite" aneurysms and 58% (95% CI, 28-85%) for "possible" aneurysms. Sensitivity of MR angiography was estimated at 83% (95% CI, 65-94%) and specificity at 97% (95% CI, 94-98%). Interobserver agreement in the evaluation of MR angiograms was poor (kappa < .30), probably because different diagnostic strategies used by individual observers resulted in different use of the assessment category "possible aneurysm." MR angiography is a feasible screening tool for detection of intracranial aneurysms. Positive predictive value, sensitivity, and specificity are acceptable when at least two neuroradiologists independently assess MR angiograms.