Atherosclerotic Plaque Composition and Classification Identified by Coronary Computed Tomography

BACKGROUND: Computed tomography (CT) is used routinely for coronary angiography, and higher-risk features of plaques can also be identified. However, the ability of CT to discriminate individual plaque components and classify plaques according to accepted histological definitions is unknown. METHODS AND RESULTS: We first determined CT attenuation ranges for individual plaque components using combined in vivo CT coregistered with virtual histology intravascular ultrasound (VH-IVUS) in 108 plaques from 57 patients. Comparison with contrast attenuation created plaque/contrast attenuation ratios that were significantly different for each component. In a separate validation cohort of 47 patients, these Plaque Maps correlated significantly with VH-IVUS-determined plaque component volumes (necrotic core: r=0.41, P=0.002; fibrous plaque: r=0.54, P<0.001; calcified plaque: r=0.59, P<0.001; total plaque: r=0.62, P<0.001). We also assessed VH-IVUS and CT Plaque Maps against coregistered histology in 72 (VH-IVUS) and 87 (CT) segments from 8 postmortem coronary arteries. The diagnostic accuracy of CT to detect calcified plaque (83% versus 92%), necrotic core (80% versus 65%), and fibroatheroma (80% versus 79%) was comparable with VH-IVUS. However, although VH-IVUS could identify thin-cap fibroatheromas (TCFA) with a diagnostic accuracy of between 74% and 82% (depending on the TCFA definition used), the spatial resolution of CT prevented direct identification of TCFA. CONCLUSIONS: CT-derived Plaque Maps based on contrast-adjusted attenuation ranges can define individual plaque components with a similar accuracy to VH-IVUS ex vivo. However, coronary CT Plaque Maps could not reliably classify plaques and identify TCFA, such that high-risk plaques may be misclassified or overlooked.