Schlattmann, Cox, Vavere, Dewey, Kendziora, Arai, Laule, George, Rief, Chen, Lima, Plotkin, Bandettini, Di Carli, Zimmermann, Miller
Purpose To compare the diagnostic performance of stress myocardial computed tomography (CT) perfusion with that of stress myocardial magnetic resonance (MR) perfusion imaging in the detection of coronary artery disease (CAD). Materials and Methods All patients gave written informed consent prior to inclusion in this institutional review board-approved study. This two-center substudy of the prospective Combined Noninvasive Coronary Angiography and Myocardial Perfusion Imaging Using 320-Detector Row Computed Tomography (CORE320) multicenter trial included 92 patients (mean age, 63.1 years ± 8.1 [standard deviation]; 73% male). All patients underwent perfusion CT and perfusion MR imaging with either adenosine or regadenoson stress. The predefined reference standards were combined quantitative coronary angiography (QCA) and single-photon emission CT (SPECT) or QCA alone. Results from coronary CT angiography were not included, and diagnostic performance was evaluated with the Mantel-Haenszel test stratified by disease status. Results The prevalence of CAD was 39% (36 of 92) according to QCA and SPECT and 64% (59 of 92) according to QCA alone. When compared with QCA and SPECT, per-patient diagnostic accuracy of perfusion CT and perfusion MR imaging was 63% (58 of 92) and 75% (69 of 92), respectively (P = .11); sensitivity was 92% (33 of 36) and 83% (30 of 36), respectively (P = .45); and specificity was 45% (25 of 56) and 70% (39 of 56), respectively (P < .01). When compared with QCA alone, diagnostic accuracy of CT perfusion and MR perfusion imaging was 82% (75 of 92) and 74% (68 of 92), respectively (P = .27); sensitivity was 90% (53 of 59) and 69% (41 of 59), respectively (P < .01); and specificity was 67% (22 of 33) and 82% (27 of 33), respectively (P = .27). Conclusion This multicenter study shows that the diagnostic performance of perfusion CT is similar to that of perfusion MR imaging in the detection of CAD. (©) RSNA, 2017 Online supplemental material is available for this article.