Accuracy of quantitative coronary angiography with computed tomography and its dependency on plaque composition

Objective: To determine the impact of plaque composition on accuracy of quantitative 64-slice computed tomography coronary angiography (CTCA). Methods: The institutional review board approved this study; written informed consent was obtained from all patients. One hundred consecutive patients (42 women, mean age 64.6 ± 9.4 years, age range 39–87 years) underwent CTCA and invasive quantitative coronary angiography (QCA) to determine (a) the diagnostic accuracy of CTCA for the detection of significant stenosis (diameter reduction of ≥50%), and (b) the accuracy of stenosis grading. In CTCA stenosis severity was graded in 10% steps and evaluated separately for calcified and non-calcified coronary lesions using Pearson-linear-regression analysis, Bland/Altman-analysis (BA), and Mann-Whitney-U-test. Results: In 60/100 patients 139 significant coronary artery stenoses were identified with QCA. On a per-segment analysis, sensitivity of CTCA was 75.5%, and specificity was 96.6% (positive predictive value: 72.9%, negative predictive value: 97.0%). Quantification of stenosis grading correlated moderately between methods (r = 0.60; P < 0.001), with an overestimation by CTCA of 5.5% (BA limits-of-agreement −29 to 39%). BA limits-of-agreement were greater in calcified lesions (−29.2 to 45.6%; mean error 8.2%) than in non-calcified lesions (−25.9 to 30.2%; mean error 2.2%) and differed significantly (P < 0.05). Conclusions: Diagnostic accuracy of CTCA is high, however agreement for quantitative lesion severity assessment between CTCA and QCA is moderate for calcified but superior for non-calcified lesions.