Reproducibility of coronary artery plaque volume and composition quantification by 64-detector row coronary computed tomographic angiography: An intraobserver, interobserver, and interscan variability study

BackgroundInterscan variability of coronary arterial plaque volume and composition quantification with coronary computed tomographic angiography (CCTA), an important attribute when considering CCTA as a serial modality, has not been examined.ObjectiveWe sought to systematically determine intraobserver- and interobserver-interscan reproducibility of these measures.MethodsTwo blinded, experienced readers independently evaluated proximal coronary segments on CCTAs from 30 patients who underwent 2 scans within 200 days (median, 124 days; interquartile range, 49–155 days) without experiencing an interim acute coronary event. Readers recorded number of plaques and, in plaques that met a preset minimal length criterion, quantified total, calcified plaque (CP), and noncalcified plaque (NCP) volumes and percentage of total plaque volume occupied by NCP.ResultsOf 89 total segments studied, 36 contained detectable plaque, and 26 met criterion for quantification. Intraobserver, interobserver, and interscan agreements for normal segments were 100%. Intraobserver-interscan correlations of total, CP, and NCP volumes and percentage of NCP were excellent (r = 0.93–0.97, P values < 0.001). Interobserver-interscan correlations for all measures were also very good (r = 0.81–0.96, P values < 0.001). Variability in plaque volume quantification was significant, exceeding 60% of the averaged paired plaque volumes in the best-case scenario (interobserver-interscan CP volume). Quantification of percentage of NCP composition by volume was more consistent, with <24% variation in the worst-case scenario (interobserver-interscan).ConclusionCCTA shows promise for quantifying serial coronary plaque change. Currently, the most robust measure seems to be percentage of plaque composition, rather than plaque volume. For smaller plaques, volume quantification remains challenging.