Interobserver Variability of 64-Slice Computed Tomography for the Quantification of Non-Calcified Coronary Atherosclerotic Plaque

Abstract

Multidetector CT (MDCT) is a reliable tool to assess and quantify calcified plaque in coronary arteries. Only very limited information is available concerning the accuracy of MDCT for evaluating non-calcified plaque. We determined the interobserver variability for measuring non-calcified plaque volumes in the three main coronary arteries using 64-slice computed tomography. We retrospectively evaluated data sets of 41 patients who received a 64-slice CT scan (Sensation 64, Siemens Forchheim, Germany, 330 msec rotation, 0.6 mm collimation, 60 ml contrast agent i. v. at 5 ml/sec) due to suspected stable coronary artery disease. The patients showed presence of non-calcified plaque in the proximal part of at least one main coronary artery. The image quality was defined on the basis of a 4-point rating scale. Two independent and blinded investigators measured the plaque volume of the non-calcified plaque by manually tracing plaque areas in contiguous cross-sectional reconstructions rendered along the vessel centerline using a slice thickness of 1 mm and an increment of 0.5 mm. The interobserver variability was evaluated and the influence of plaque volume and image quality on interobserver variability was determined. The mean volume of non-calcified plaque was 157 +/- 85 mm (3), 76 +/- 43 mm (3) and 133 +/- 80 mm (3) for the LAD, LCX and RCA, respectively (LAD vs. LCX: p < 0.01; LAD vs. RCA: p = 0.33; LCX vs. RCA: p < 0.01). There was a mean absolute difference in plaque volume between the two observers of 23 +/- 15 mm (3), of 20 +/- 9 mm (3) and of 38 +/- 21 mm (3), which corresponds to a mean interobserver variability of 17 +/- 10 %, 29 +/- 13 % and 32 +/- 13 % for the LAD, LCX and RCA, respectively (LAD vs. LCX: p < 0.01; LAD vs. RCA: p < 0.01; LCX vs. RCA: p = 0.87). A significant inverse correlation between interobserver variability and the extent of the plaque volume (r = - 0.48; p = 0.01) was found. Interobserver variability was dependent on image quality: The highest image quality was observed in the LAD (2.4 +/- 0.5), while the image quality in the LCX (2.1 +/- 0.7) and the RCA (2.0 +/- 0.6) was lower. Interobserver variability for the quantification of non-calcified plaque volumes in 64-slice MDCT is substantial. Interobserver variability in the LAD was significantly lower than in the LCX and the RCA. This might be due to a larger mean plaque volume and better image quality in the LAD than in other coronary arteries.