ECG-gated CT in Aortic Perivalvular Abscess: Comparison with Transesophageal Echocardiography and Intraoperative Findings.

Background The 2015 European Society of Cardiology guidelines acknowledged similar diagnostic performance of electrocardiography (ECG)-gated CT on perivalvular abscesses compared with transesophageal echocardiography (TEE), but data on ECG-gated CT remain insufficient. Purpose To determine the diagnostic performance of ECG-gated CT for assessing aortic root perivalvular abscesses and to compare it with TEE. Materials and Methods Between January 2008 and June 2019, the imaging records of surgically confirmed infective endocarditis were retrospectively reviewed for presence of aortic perivalvular abscesses, their extension, fistulization, vegetations, and valvular destruction. The diagnostic performance of ECG-gated CT was analyzed in all patients (part A) and in an noninferiority analysis (part B; δ = -10%) in patients undergoing TEE. Results A total of 178 patients (median age, 54 years [interquartile range, 15 years]; 147 men) were evaluated (CT, n = 178; TEE, n = 35). In part A, the sensitivity and specificity of CT were 70 of 71 (99% [95% confidence interval (CI): 96%, 100%]) and 102 of 107 (95% [95% CI: 91%, 99%]) for abscess; 65 of 68 (96% [95% CI: 91%, 100%]) and 107 of 110 (97% [95% CI: 94%, 100%]) for extension, 36 of 36 (100% [95% CI: 100%, 100%]) and 139 of 142 (98% [95% CI: 96%, 100%]) for fistulization, 153 of 160 (96% [95% CI: 93%, 99%]) and five of 18 (28% [95% CI: 7%, 49%]) for vegetations, and 90 of 90 (100% [95% CI: 100%, 100%]) and 24 of 88 (27% [95% CI: 18%, 37%]) for valvular destruction. In part B, ECG-gated CT had noninferior sensitivity compared with TEE for detecting abscess (difference, 14 percentage points [lower one-sided 95% CI: -4 percentage points]), extension (difference, 0 percentage points [lower one-sided 95% CI: 0 percentage points]), fistulization (difference, 0 percentage points [lower one-sided 95% CI: 0 percentage points]), and valvular destruction (difference, 5 percentage points [lower one-sided 95% CI: -4 percentage points]). Specificity of CT was inferior for demonstrating perivalvular abscess (difference, 5 percentage points [lower one-sided 95% CI: -11 percentage points]) and valvular destruction (difference, -62 percentage points [lower one-sided 95% CI: -92 percentage points]). ECG-gated CT had inferior sensitivity in detecting vegetations (difference, -6 percentage points [lower one-sided 95% CI: -14 percentage points]). Conclusion Electrocardiography-gated CT had noninferior sensitivity compared with transesophageal echocardiography for identification of aortic perivalvular abscesses, extension of these abscesses, fistulization, and valvular destruction but had inferior sensitivity in detection of vegetations. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by Sakuma in this issue.